From b081d5c50e25fab8e88ba92c0e61e5ee1cdd7b6c Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 14:33:33 +0900
Subject: [PATCH 01/31] seam: add hub_builder

---
 python/private/pypi/BUILD.bazel     |  9 +++++++++
 python/private/pypi/extension.bzl   | 12 +++++++++---
 python/private/pypi/hub_builder.bzl | 16 ++++++++++++++++
 3 files changed, 34 insertions(+), 3 deletions(-)
 create mode 100644 python/private/pypi/hub_builder.bzl

diff --git a/python/private/pypi/BUILD.bazel b/python/private/pypi/BUILD.bazel
index cb3408a191..ed268b2474 100644
--- a/python/private/pypi/BUILD.bazel
+++ b/python/private/pypi/BUILD.bazel
@@ -111,6 +111,7 @@ bzl_library(
     deps = [
         ":attrs_bzl",
         ":evaluate_markers_bzl",
+        ":hub_builder_bzl",
         ":hub_repository_bzl",
         ":parse_requirements_bzl",
         ":parse_whl_name_bzl",
@@ -167,6 +168,14 @@ bzl_library(
     ],
 )
 
+bzl_library(
+    name = "hub_builder_bzl",
+    srcs = ["hub_builder.bzl"],
+    visibility = ["//:__subpackages__"],
+    deps = [
+    ],
+)
+
 bzl_library(
     name = "hub_repository_bzl",
     srcs = ["hub_repository.bzl"],
diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 03af863e1e..580df221aa 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -26,6 +26,7 @@ load("//python/private:version.bzl", "version")
 load("//python/private:version_label.bzl", "version_label")
 load(":attrs.bzl", "use_isolated")
 load(":evaluate_markers.bzl", "evaluate_markers_py", EVALUATE_MARKERS_SRCS = "SRCS", evaluate_markers_star = "evaluate_markers")
+load(":hub_builder.bzl", "hub_builder")
 load(":hub_repository.bzl", "hub_repository", "whl_config_settings_to_json")
 load(":parse_requirements.bzl", "parse_requirements")
 load(":parse_whl_name.bzl", "parse_whl_name")
@@ -658,10 +659,11 @@ You cannot use both the additive_build_content and additive_build_content_file a
         for pip_attr in mod.tags.parse:
             hub_name = pip_attr.hub_name
             if hub_name not in pip_hub_map:
-                pip_hub_map[pip_attr.hub_name] = struct(
+                builder = hub_builder(
+                    name = hub_name,
                     module_name = mod.name,
-                    python_versions = [pip_attr.python_version],
                 )
+                pip_hub_map[pip_attr.hub_name] = builder
             elif pip_hub_map[hub_name].module_name != mod.name:
                 # We cannot have two hubs with the same name in different
                 # modules.
@@ -687,7 +689,11 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     version = pip_attr.python_version,
                 ))
             else:
-                pip_hub_map[pip_attr.hub_name].python_versions.append(pip_attr.python_version)
+                builder = pip_hub_map[pip_attr.hub_name]
+
+            builder.add(
+                python_version = pip_attr.python_version,
+            )
 
             get_index_urls = None
             if pip_attr.experimental_index_url:
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
new file mode 100644
index 0000000000..c1af73b6e2
--- /dev/null
+++ b/python/private/pypi/hub_builder.bzl
@@ -0,0 +1,16 @@
+"""A hub repository builder for incrementally building the hub configuration."""
+
+def hub_builder(*, name, module_name):
+    """Return a hub builder instance"""
+    self = struct(
+        name = name,
+        module_name = module_name,
+        python_versions = [],
+        # buildifier: disable=uninitialized
+        add = lambda *args, **kwargs: _add(self, *args, **kwargs),
+        # buildifier: enable=uninitialized
+    )
+    return self
+
+def _add(self, *, python_version):
+    self.python_versions.append(python_version)

From e9e118a1f772528fe20174c895148ee870d39577 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 14:34:49 +0900
Subject: [PATCH 02/31] refactor: move error handling

---
 python/private/pypi/extension.bzl   | 10 ----------
 python/private/pypi/hub_builder.bzl | 11 +++++++++++
 2 files changed, 11 insertions(+), 10 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 580df221aa..ad38425554 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -678,16 +678,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     second_module = mod.name,
                 ))
 
-            elif pip_attr.python_version in pip_hub_map[hub_name].python_versions:
-                fail((
-                    "Duplicate pip python version '{version}' for hub " +
-                    "'{hub}' in module '{module}': the Python versions " +
-                    "used for a hub must be unique"
-                ).format(
-                    hub = hub_name,
-                    module = mod.name,
-                    version = pip_attr.python_version,
-                ))
             else:
                 builder = pip_hub_map[pip_attr.hub_name]
 
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index c1af73b6e2..3b580ebd35 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -13,4 +13,15 @@ def hub_builder(*, name, module_name):
     return self
 
 def _add(self, *, python_version):
+    if python_version in self.python_versions:
+        fail((
+            "Duplicate pip python version '{version}' for hub " +
+            "'{hub}' in module '{module}': the Python versions " +
+            "used for a hub must be unique"
+        ).format(
+            hub = self.name,
+            module = self.module_name,
+            version = python_version,
+        ))
+
     self.python_versions.append(python_version)

From a2bc6535942595117d0f81d4b7b219c2b5170be6 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 14:40:24 +0900
Subject: [PATCH 03/31] move get_index_url configuration elsewhere

---
 python/private/pypi/BUILD.bazel     |  1 +
 python/private/pypi/extension.bzl   | 34 ++-----------------
 python/private/pypi/hub_builder.bzl | 52 +++++++++++++++++++++++++++--
 3 files changed, 53 insertions(+), 34 deletions(-)

diff --git a/python/private/pypi/BUILD.bazel b/python/private/pypi/BUILD.bazel
index ed268b2474..a4d53e56b0 100644
--- a/python/private/pypi/BUILD.bazel
+++ b/python/private/pypi/BUILD.bazel
@@ -173,6 +173,7 @@ bzl_library(
     srcs = ["hub_builder.bzl"],
     visibility = ["//:__subpackages__"],
     deps = [
+        "//python/private:normalize_name_bzl",
     ],
 )
 
diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index ad38425554..82e8edf5af 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -662,6 +662,8 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 builder = hub_builder(
                     name = hub_name,
                     module_name = mod.name,
+                    simpleapi_download_fn = simpleapi_download,
+                    simpleapi_cache = simpleapi_cache,
                 )
                 pip_hub_map[pip_attr.hub_name] = builder
             elif pip_hub_map[hub_name].module_name != mod.name:
@@ -685,41 +687,11 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 python_version = pip_attr.python_version,
             )
 
-            get_index_urls = None
-            if pip_attr.experimental_index_url:
-                skip_sources = [
-                    normalize_name(s)
-                    for s in pip_attr.simpleapi_skip
-                ]
-                get_index_urls = lambda ctx, distributions: simpleapi_download(
-                    ctx,
-                    attr = struct(
-                        index_url = pip_attr.experimental_index_url,
-                        extra_index_urls = pip_attr.experimental_extra_index_urls or [],
-                        index_url_overrides = pip_attr.experimental_index_url_overrides or {},
-                        sources = [
-                            d
-                            for d in distributions
-                            if normalize_name(d) not in skip_sources
-                        ],
-                        envsubst = pip_attr.envsubst,
-                        # Auth related info
-                        netrc = pip_attr.netrc,
-                        auth_patterns = pip_attr.auth_patterns,
-                    ),
-                    cache = simpleapi_cache,
-                    parallel_download = pip_attr.parallel_download,
-                )
-            elif pip_attr.experimental_extra_index_urls:
-                fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")
-            elif pip_attr.experimental_index_url_overrides:
-                fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
-
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
             out = _create_whl_repos(
                 module_ctx,
                 pip_attr = pip_attr,
-                get_index_urls = get_index_urls,
+                get_index_urls = builder.get_index_urls(pip_attr),
                 whl_overrides = whl_overrides,
                 config = config,
                 **kwargs
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 3b580ebd35..a9e991a395 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -1,15 +1,28 @@
 """A hub repository builder for incrementally building the hub configuration."""
 
-def hub_builder(*, name, module_name):
+load("//python/private:normalize_name.bzl", "normalize_name")
+
+def hub_builder(
+        *,
+        name,
+        module_name,
+        simpleapi_download_fn,
+        simpleapi_cache = {}):
     """Return a hub builder instance"""
+
+    # buildifier: disable=uninitialized
     self = struct(
         name = name,
         module_name = module_name,
         python_versions = [],
-        # buildifier: disable=uninitialized
+        _simpleapi_download_fn = simpleapi_download_fn,
+        _simpleapi_cache = simpleapi_cache,
+        # keep sorted
         add = lambda *args, **kwargs: _add(self, *args, **kwargs),
-        # buildifier: enable=uninitialized
+        get_index_urls = lambda *args, **kwargs: _get_index_urls(self, *args, **kwargs),
     )
+
+    # buildifier: enable=uninitialized
     return self
 
 def _add(self, *, python_version):
@@ -25,3 +38,36 @@ def _add(self, *, python_version):
         ))
 
     self.python_versions.append(python_version)
+
+def _get_index_urls(self, pip_attr):
+    get_index_urls = None
+    if pip_attr.experimental_index_url:
+        skip_sources = [
+            normalize_name(s)
+            for s in pip_attr.simpleapi_skip
+        ]
+        get_index_urls = lambda ctx, distributions: self._simpleapi_download_fn(
+            ctx,
+            attr = struct(
+                index_url = pip_attr.experimental_index_url,
+                extra_index_urls = pip_attr.experimental_extra_index_urls or [],
+                index_url_overrides = pip_attr.experimental_index_url_overrides or {},
+                sources = [
+                    d
+                    for d in distributions
+                    if normalize_name(d) not in skip_sources
+                ],
+                envsubst = pip_attr.envsubst,
+                # Auth related info
+                netrc = pip_attr.netrc,
+                auth_patterns = pip_attr.auth_patterns,
+            ),
+            cache = self._simpleapi_cache,
+            parallel_download = pip_attr.parallel_download,
+        )
+    elif pip_attr.experimental_extra_index_urls:
+        fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")
+    elif pip_attr.experimental_index_url_overrides:
+        fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
+
+    return get_index_urls

From 8645443aa1655a02ed0982b62f1e5087dbd02684 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 14:53:19 +0900
Subject: [PATCH 04/31] refactor: move interpreter detection to a simple
 function

---
 python/private/pypi/extension.bzl   | 51 ++++++++---------------------
 python/private/pypi/hub_builder.bzl | 48 +++++++++++++++++++++++++--
 2 files changed, 59 insertions(+), 40 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 82e8edf5af..8dcec601e5 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -120,22 +120,17 @@ def _create_whl_repos(
         pip_attr,
         whl_overrides,
         config,
-        available_interpreters = INTERPRETER_LABELS,
+        hub,
         minor_mapping = MINOR_MAPPING,
-        evaluate_markers = None,
-        get_index_urls = None):
+        evaluate_markers = None):
     """create all of the whl repositories
 
     Args:
         module_ctx: {type}`module_ctx`.
         pip_attr: {type}`struct` - the struct that comes from the tag class iteration.
         whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
+        hub: TODO.
         config: The platform configuration.
-        get_index_urls: A function used to get the index URLs
-        available_interpreters: {type}`dict[str, Label]` The dictionary of available
-            interpreters that have been registered using the `python` bzlmod extension.
-            The keys are in the form `python_{snake_case_version}_host`. This is to be
-            used during the `repository_rule` and must be always compatible with the host.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
         evaluate_markers: the function used to evaluate the markers.
@@ -152,7 +147,8 @@ def _create_whl_repos(
             rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
-    python_interpreter_target = pip_attr.python_interpreter_target
+    get_index_urls = hub.get_index_urls(pip_attr)
+    interpreter = hub.detect_interpreter(pip_attr)
 
     # containers to aggregate outputs from this function
     whl_map = {}
@@ -162,31 +158,10 @@ def _create_whl_repos(
     }
     whl_libraries = {}
 
-    # if we do not have the python_interpreter set in the attributes
-    # we programmatically find it.
-    hub_name = pip_attr.hub_name
-    if python_interpreter_target == None and not pip_attr.python_interpreter:
-        python_name = "python_{}_host".format(
-            pip_attr.python_version.replace(".", "_"),
-        )
-        if python_name not in available_interpreters:
-            fail((
-                "Unable to find interpreter for pip hub '{hub_name}' for " +
-                "python_version={version}: Make sure a corresponding " +
-                '`python.toolchain(python_version="{version}")` call exists.' +
-                "Expected to find {python_name} among registered versions:\n  {labels}"
-            ).format(
-                hub_name = hub_name,
-                version = pip_attr.python_version,
-                python_name = python_name,
-                labels = "  \n".join(available_interpreters),
-            ))
-        python_interpreter_target = available_interpreters[python_name]
-
     # TODO @aignas 2025-06-29: we should not need the version in the pip_name if
     # we are using pipstar and we are downloading the wheel using the downloader
     pip_name = "{}_{}".format(
-        hub_name,
+        hub.name,
         version_label(pip_attr.python_version),
     )
     major_minor = _major_minor_version(pip_attr.python_version)
@@ -249,8 +224,8 @@ def _create_whl_repos(
                 }
                 for k, plats in requirements.items()
             },
-            python_interpreter = pip_attr.python_interpreter,
-            python_interpreter_target = python_interpreter_target,
+            python_interpreter = interpreter.path,
+            python_interpreter_target = interpreter.target,
             srcs = pip_attr._evaluate_markers_srcs,
             logger = logger,
         )
@@ -293,7 +268,7 @@ def _create_whl_repos(
         # Construct args separately so that the lock file can be smaller and does not include unused
         # attrs.
         whl_library_args = dict(
-            dep_template = "@{}//{{name}}:{{target}}".format(hub_name),
+            dep_template = "@{}//{{name}}:{{target}}".format(hub.name),
         )
         maybe_args = dict(
             # The following values are safe to omit if they have false like values
@@ -306,8 +281,8 @@ def _create_whl_repos(
             group_deps = group_deps,
             group_name = group_name,
             pip_data_exclude = pip_attr.pip_data_exclude,
-            python_interpreter = pip_attr.python_interpreter,
-            python_interpreter_target = python_interpreter_target,
+            python_interpreter = interpreter.path,
+            python_interpreter_target = interpreter.target,
             whl_patches = {
                 p: json.encode(args)
                 for p, args in whl_overrides.get(whl.name, {}).items()
@@ -664,6 +639,8 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     module_name = mod.name,
                     simpleapi_download_fn = simpleapi_download,
                     simpleapi_cache = simpleapi_cache,
+                    minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
+                    available_interpreters = kwargs.pop("available_interpreters", INTERPRETER_LABELS),
                 )
                 pip_hub_map[pip_attr.hub_name] = builder
             elif pip_hub_map[hub_name].module_name != mod.name:
@@ -690,8 +667,8 @@ You cannot use both the additive_build_content and additive_build_content_file a
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
             out = _create_whl_repos(
                 module_ctx,
+                hub = builder,
                 pip_attr = pip_attr,
-                get_index_urls = builder.get_index_urls(pip_attr),
                 whl_overrides = whl_overrides,
                 config = config,
                 **kwargs
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index a9e991a395..1bd67ef353 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -6,20 +6,37 @@ def hub_builder(
         *,
         name,
         module_name,
+        minor_mapping,
+        available_interpreters,
         simpleapi_download_fn,
         simpleapi_cache = {}):
-    """Return a hub builder instance"""
+    """Return a hub builder instance
+
+    Args:
+        name: TODO
+        module_name: TODO
+        minor_mapping: TODO
+        available_interpreters: {type}`dict[str, Label]` The dictionary of available
+            interpreters that have been registered using the `python` bzlmod extension.
+            The keys are in the form `python_{snake_case_version}_host`. This is to be
+            used during the `repository_rule` and must be always compatible with the host.
+        simpleapi_download_fn: TODO
+        simpleapi_cache: TODO
+    """
 
     # buildifier: disable=uninitialized
     self = struct(
         name = name,
         module_name = module_name,
         python_versions = [],
+        _minor_mapping = minor_mapping,
+        _available_interpreters = available_interpreters,
         _simpleapi_download_fn = simpleapi_download_fn,
         _simpleapi_cache = simpleapi_cache,
         # keep sorted
-        add = lambda *args, **kwargs: _add(self, *args, **kwargs),
-        get_index_urls = lambda *args, **kwargs: _get_index_urls(self, *args, **kwargs),
+        add = lambda *a, **k: _add(self, *a, **k),
+        get_index_urls = lambda *a, **k: _get_index_urls(self, *a, **k),
+        detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
@@ -71,3 +88,28 @@ def _get_index_urls(self, pip_attr):
         fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
 
     return get_index_urls
+
+def _detect_interpreter(self, pip_attr):
+    python_interpreter_target = pip_attr.python_interpreter_target
+    if python_interpreter_target == None and not pip_attr.python_interpreter:
+        python_name = "python_{}_host".format(
+            pip_attr.python_version.replace(".", "_"),
+        )
+        if python_name not in self._available_interpreters:
+            fail((
+                "Unable to find interpreter for pip hub '{hub_name}' for " +
+                "python_version={version}: Make sure a corresponding " +
+                '`python.toolchain(python_version="{version}")` call exists.' +
+                "Expected to find {python_name} among registered versions:\n  {labels}"
+            ).format(
+                hub_name = self.name,
+                version = pip_attr.python_version,
+                python_name = python_name,
+                labels = "  \n".join(self._available_interpreters),
+            ))
+        python_interpreter_target = self._available_interpreters[python_name]
+
+    return struct(
+        target = python_interpreter_target,
+        path = pip_attr.python_interpreter,
+    )

From df8d2be38401e69092a2ebddfd1c2264a77b84ad Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 14:57:35 +0900
Subject: [PATCH 05/31] minor: move declaration closer to use

---
 python/private/pypi/extension.bzl | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 8dcec601e5..d8704eda24 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -164,7 +164,6 @@ def _create_whl_repos(
         hub.name,
         version_label(pip_attr.python_version),
     )
-    major_minor = _major_minor_version(pip_attr.python_version)
 
     whl_modifications = {}
     if pip_attr.whl_modifications != None:
@@ -315,7 +314,7 @@ def _create_whl_repos(
                     get_index_urls != None,  # defaults to True if the get_index_urls is defined
                 ),
                 auth_patterns = config.auth_patterns or pip_attr.auth_patterns,
-                python_version = major_minor,
+                python_version = _major_minor_version(pip_attr.python_version),
                 is_multiple_versions = whl.is_multiple_versions,
                 enable_pipstar = config.enable_pipstar,
             )

From 863243b8b58588786e1a43a300130354bf4e0d5a Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 15:10:25 +0900
Subject: [PATCH 06/31] refactor: move platform building for the hub to the
 builder

---
 python/private/pypi/BUILD.bazel     |  6 ++-
 python/private/pypi/extension.bzl   | 70 ++++-------------------------
 python/private/pypi/hub_builder.bzl | 66 ++++++++++++++++++++++++++-
 3 files changed, 78 insertions(+), 64 deletions(-)

diff --git a/python/private/pypi/BUILD.bazel b/python/private/pypi/BUILD.bazel
index a4d53e56b0..73451b9784 100644
--- a/python/private/pypi/BUILD.bazel
+++ b/python/private/pypi/BUILD.bazel
@@ -117,7 +117,6 @@ bzl_library(
         ":parse_whl_name_bzl",
         ":pep508_env_bzl",
         ":pip_repository_attrs_bzl",
-        ":python_tag_bzl",
         ":simpleapi_download_bzl",
         ":whl_config_setting_bzl",
         ":whl_library_bzl",
@@ -173,7 +172,12 @@ bzl_library(
     srcs = ["hub_builder.bzl"],
     visibility = ["//:__subpackages__"],
     deps = [
+        ":pep508_env_bzl",
+        ":pep508_evaluate_bzl",
+        ":python_tag_bzl",
+        "//python/private:full_version_bzl",
         "//python/private:normalize_name_bzl",
+        "//python/private:version_bzl",
     ],
 )
 
diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index d8704eda24..d77e1df881 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -31,9 +31,7 @@ load(":hub_repository.bzl", "hub_repository", "whl_config_settings_to_json")
 load(":parse_requirements.bzl", "parse_requirements")
 load(":parse_whl_name.bzl", "parse_whl_name")
 load(":pep508_env.bzl", "env")
-load(":pep508_evaluate.bzl", "evaluate")
 load(":pip_repository_attrs.bzl", "ATTRS")
-load(":python_tag.bzl", "python_tag")
 load(":requirements_files_by_platform.bzl", "requirements_files_by_platform")
 load(":simpleapi_download.bzl", "simpleapi_download")
 load(":whl_config_setting.bzl", "whl_config_setting")
@@ -69,57 +67,11 @@ def _whl_mods_impl(whl_mods_dict):
             whl_mods = whl_mods,
         )
 
-def _platforms(*, python_version, minor_mapping, config):
-    platforms = {}
-    python_version = version.parse(
-        full_version(
-            version = python_version,
-            minor_mapping = minor_mapping,
-        ),
-        strict = True,
-    )
-
-    for platform, values in config.platforms.items():
-        # TODO @aignas 2025-07-07: this is probably doing the parsing of the version too
-        # many times.
-        abi = "{}{}{}.{}".format(
-            python_tag(values.env["implementation_name"]),
-            python_version.release[0],
-            python_version.release[1],
-            python_version.release[2],
-        )
-        key = "{}_{}".format(abi, platform)
-
-        env_ = env(
-            env = values.env,
-            os = values.os_name,
-            arch = values.arch_name,
-            python_version = python_version.string,
-        )
-
-        if values.marker and not evaluate(values.marker, env = env_):
-            continue
-
-        platforms[key] = struct(
-            env = env_,
-            triple = "{}_{}_{}".format(abi, values.os_name, values.arch_name),
-            whl_abi_tags = [
-                v.format(
-                    major = python_version.release[0],
-                    minor = python_version.release[1],
-                )
-                for v in values.whl_abi_tags
-            ],
-            whl_platform_tags = values.whl_platform_tags,
-        )
-    return platforms
-
 def _create_whl_repos(
         module_ctx,
         *,
         pip_attr,
         whl_overrides,
-        config,
         hub,
         minor_mapping = MINOR_MAPPING,
         evaluate_markers = None):
@@ -130,7 +82,6 @@ def _create_whl_repos(
         pip_attr: {type}`struct` - the struct that comes from the tag class iteration.
         whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
         hub: TODO.
-        config: The platform configuration.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
         evaluate_markers: the function used to evaluate the markers.
@@ -185,16 +136,12 @@ def _create_whl_repos(
         whl_group_mapping = {}
         requirement_cycles = {}
 
-    platforms = _platforms(
-        python_version = pip_attr.python_version,
-        minor_mapping = minor_mapping,
-        config = config,
-    )
+    platforms = hub.platforms(pip_attr.python_version)
 
     if evaluate_markers:
         # This is most likely unit tests
         pass
-    elif config.enable_pipstar:
+    elif hub.config.enable_pipstar:
         evaluate_markers = lambda _, requirements: evaluate_markers_star(
             requirements = requirements,
             platforms = platforms,
@@ -287,7 +234,7 @@ def _create_whl_repos(
                 for p, args in whl_overrides.get(whl.name, {}).items()
             },
         )
-        if not config.enable_pipstar:
+        if not hub.config.enable_pipstar:
             maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
 
         whl_library_args.update({k: v for k, v in maybe_args.items() if v})
@@ -308,15 +255,15 @@ def _create_whl_repos(
                 src = src,
                 whl_library_args = whl_library_args,
                 download_only = pip_attr.download_only,
-                netrc = config.netrc or pip_attr.netrc,
+                netrc = hub.config.netrc or pip_attr.netrc,
                 use_downloader = use_downloader.get(
                     whl.name,
                     get_index_urls != None,  # defaults to True if the get_index_urls is defined
                 ),
-                auth_patterns = config.auth_patterns or pip_attr.auth_patterns,
+                auth_patterns = hub.config.auth_patterns or pip_attr.auth_patterns,
                 python_version = _major_minor_version(pip_attr.python_version),
                 is_multiple_versions = whl.is_multiple_versions,
-                enable_pipstar = config.enable_pipstar,
+                enable_pipstar = hub.config.enable_pipstar,
             )
             if repo == None:
                 # NOTE @aignas 2025-07-07: we guard against an edge-case where there
@@ -332,7 +279,7 @@ def _create_whl_repos(
                 ))
             whl_libraries[repo_name] = repo.args
 
-            if not config.enable_pipstar and "experimental_target_platforms" in repo.args:
+            if not hub.config.enable_pipstar and "experimental_target_platforms" in repo.args:
                 whl_libraries[repo_name] |= {
                     "experimental_target_platforms": sorted({
                         # TODO @aignas 2025-07-07: this should be solved in a better way
@@ -636,6 +583,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 builder = hub_builder(
                     name = hub_name,
                     module_name = mod.name,
+                    config = config,
                     simpleapi_download_fn = simpleapi_download,
                     simpleapi_cache = simpleapi_cache,
                     minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
@@ -669,9 +617,9 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 hub = builder,
                 pip_attr = pip_attr,
                 whl_overrides = whl_overrides,
-                config = config,
                 **kwargs
             )
+
             hub_whl_map.setdefault(hub_name, {})
             for key, settings in out.whl_map.items():
                 for setting, repo in settings.items():
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 1bd67ef353..e74e5072c0 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -1,11 +1,17 @@
 """A hub repository builder for incrementally building the hub configuration."""
 
+load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
+load("//python/private:version.bzl", "version")
+load(":pep508_env.bzl", "env")
+load(":pep508_evaluate.bzl", "evaluate")
+load(":python_tag.bzl", "python_tag")
 
 def hub_builder(
         *,
         name,
         module_name,
+        config,
         minor_mapping,
         available_interpreters,
         simpleapi_download_fn,
@@ -15,6 +21,7 @@ def hub_builder(
     Args:
         name: TODO
         module_name: TODO
+        config: The platform configuration.
         minor_mapping: TODO
         available_interpreters: {type}`dict[str, Label]` The dictionary of available
             interpreters that have been registered using the `python` bzlmod extension.
@@ -28,7 +35,12 @@ def hub_builder(
     self = struct(
         name = name,
         module_name = module_name,
-        python_versions = [],
+        python_versions = {},
+        config = config,
+        whl_map = {},
+        exposed_packages = {},
+        extra_aliases = {},
+        whl_libraries = {},
         _minor_mapping = minor_mapping,
         _available_interpreters = available_interpreters,
         _simpleapi_download_fn = simpleapi_download_fn,
@@ -37,6 +49,7 @@ def hub_builder(
         add = lambda *a, **k: _add(self, *a, **k),
         get_index_urls = lambda *a, **k: _get_index_urls(self, *a, **k),
         detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
+        platforms = lambda version: self.python_versions[version],
     )
 
     # buildifier: enable=uninitialized
@@ -54,7 +67,11 @@ def _add(self, *, python_version):
             version = python_version,
         ))
 
-    self.python_versions.append(python_version)
+    self.python_versions[python_version] = _platforms(
+        python_version = python_version,
+        minor_mapping = self._minor_mapping,
+        config = self.config,
+    )
 
 def _get_index_urls(self, pip_attr):
     get_index_urls = None
@@ -113,3 +130,48 @@ def _detect_interpreter(self, pip_attr):
         target = python_interpreter_target,
         path = pip_attr.python_interpreter,
     )
+
+def _platforms(*, python_version, minor_mapping, config):
+    platforms = {}
+    python_version = version.parse(
+        full_version(
+            version = python_version,
+            minor_mapping = minor_mapping,
+        ),
+        strict = True,
+    )
+
+    for platform, values in config.platforms.items():
+        # TODO @aignas 2025-07-07: this is probably doing the parsing of the version too
+        # many times.
+        abi = "{}{}{}.{}".format(
+            python_tag(values.env["implementation_name"]),
+            python_version.release[0],
+            python_version.release[1],
+            python_version.release[2],
+        )
+        key = "{}_{}".format(abi, platform)
+
+        env_ = env(
+            env = values.env,
+            os = values.os_name,
+            arch = values.arch_name,
+            python_version = python_version.string,
+        )
+
+        if values.marker and not evaluate(values.marker, env = env_):
+            continue
+
+        platforms[key] = struct(
+            env = env_,
+            triple = "{}_{}_{}".format(abi, values.os_name, values.arch_name),
+            whl_abi_tags = [
+                v.format(
+                    major = python_version.release[0],
+                    minor = python_version.release[1],
+                )
+                for v in values.whl_abi_tags
+            ],
+            whl_platform_tags = values.whl_platform_tags,
+        )
+    return platforms

From 7ee62e7963d1590df503b5333c090e3b25524520 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 15:14:55 +0900
Subject: [PATCH 07/31] refactor: store the whl_map on the builder itself

---
 python/private/pypi/extension.bzl | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index d77e1df881..6a77c13868 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -102,7 +102,6 @@ def _create_whl_repos(
     interpreter = hub.detect_interpreter(pip_attr)
 
     # containers to aggregate outputs from this function
-    whl_map = {}
     extra_aliases = {
         whl_name: {alias: True for alias in aliases}
         for whl_name, aliases in pip_attr.extra_hub_aliases.items()
@@ -290,7 +289,7 @@ def _create_whl_repos(
                     }),
                 }
 
-            mapping = whl_map.setdefault(whl.name, {})
+            mapping = hub.whl_map.setdefault(whl.name, {})
             if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
                 fail(
                     "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
@@ -303,7 +302,6 @@ def _create_whl_repos(
                 mapping[repo.config_setting] = repo_name
 
     return struct(
-        whl_map = whl_map,
         exposed_packages = exposed_packages,
         extra_aliases = extra_aliases,
         whl_libraries = whl_libraries,
@@ -620,10 +618,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 **kwargs
             )
 
-            hub_whl_map.setdefault(hub_name, {})
-            for key, settings in out.whl_map.items():
-                for setting, repo in settings.items():
-                    hub_whl_map[hub_name].setdefault(key, {}).setdefault(repo, []).append(setting)
             extra_aliases.setdefault(hub_name, {})
             for whl_name, aliases in out.extra_aliases.items():
                 extra_aliases[hub_name].setdefault(whl_name, {}).update(aliases)
@@ -653,6 +647,12 @@ You cannot use both the additive_build_content and additive_build_content_file a
             # using an alternative cycle resolution strategy.
             hub_group_map[hub_name] = pip_attr.experimental_requirement_cycles
 
+    for hub in pip_hub_map.values():
+        hub_whl_map.setdefault(hub.name, {})
+        for key, settings in hub.whl_map.items():
+            for setting, repo in settings.items():
+                hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
+
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
         # args are manipulated in the code going before.

From 106789f6561f0d0192efb50ac05c7918223caae1 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 15:24:34 +0900
Subject: [PATCH 08/31] refactor: store the whl_libraries for each hub in the
 builder

---
 python/private/pypi/extension.bzl   | 68 ++++++-----------------------
 python/private/pypi/hub_builder.bzl | 45 +++++++++++++++++++
 2 files changed, 58 insertions(+), 55 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 6a77c13868..20a3f2ef18 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -23,7 +23,6 @@ load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:repo_utils.bzl", "repo_utils")
 load("//python/private:version.bzl", "version")
-load("//python/private:version_label.bzl", "version_label")
 load(":attrs.bzl", "use_isolated")
 load(":evaluate_markers.bzl", "evaluate_markers_py", EVALUATE_MARKERS_SRCS = "SRCS", evaluate_markers_star = "evaluate_markers")
 load(":hub_builder.bzl", "hub_builder")
@@ -106,14 +105,6 @@ def _create_whl_repos(
         whl_name: {alias: True for alias in aliases}
         for whl_name, aliases in pip_attr.extra_hub_aliases.items()
     }
-    whl_libraries = {}
-
-    # TODO @aignas 2025-06-29: we should not need the version in the pip_name if
-    # we are using pipstar and we are downloading the wheel using the downloader
-    pip_name = "{}_{}".format(
-        hub.name,
-        version_label(pip_attr.python_version),
-    )
 
     whl_modifications = {}
     if pip_attr.whl_modifications != None:
@@ -264,47 +255,15 @@ def _create_whl_repos(
                 is_multiple_versions = whl.is_multiple_versions,
                 enable_pipstar = hub.config.enable_pipstar,
             )
-            if repo == None:
-                # NOTE @aignas 2025-07-07: we guard against an edge-case where there
-                # are more platforms defined than there are wheels for and users
-                # disallow building from sdist.
-                continue
-
-            repo_name = "{}_{}".format(pip_name, repo.repo_name)
-            if repo_name in whl_libraries:
-                fail("attempting to create a duplicate library {} for {}".format(
-                    repo_name,
-                    whl.name,
-                ))
-            whl_libraries[repo_name] = repo.args
-
-            if not hub.config.enable_pipstar and "experimental_target_platforms" in repo.args:
-                whl_libraries[repo_name] |= {
-                    "experimental_target_platforms": sorted({
-                        # TODO @aignas 2025-07-07: this should be solved in a better way
-                        platforms[candidate].triple.partition("_")[-1]: None
-                        for p in repo.args["experimental_target_platforms"]
-                        for candidate in platforms
-                        if candidate.endswith(p)
-                    }),
-                }
-
-            mapping = hub.whl_map.setdefault(whl.name, {})
-            if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
-                fail(
-                    "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
-                        mapping[repo.config_setting],
-                        repo.config_setting,
-                        repo_name,
-                    ),
-                )
-            else:
-                mapping[repo.config_setting] = repo_name
+            hub.add_whl_library(
+                python_version = pip_attr.python_version,
+                whl = whl,
+                repo = repo,
+            )
 
     return struct(
         exposed_packages = exposed_packages,
         extra_aliases = extra_aliases,
-        whl_libraries = whl_libraries,
     )
 
 def _whl_repo(
@@ -631,15 +590,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                         continue
                     intersection[pkg] = None
                 exposed_packages[hub_name] = intersection
-            whl_libraries.update(out.whl_libraries)
-            for whl_name, lib in out.whl_libraries.items():
-                if enable_pipstar:
-                    whl_libraries.setdefault(whl_name, lib)
-                elif whl_name in lib:
-                    fail("'{}' already in created".format(whl_name))
-                else:
-                    # replicate whl_libraries.update(out.whl_libraries)
-                    whl_libraries[whl_name] = lib
 
             # TODO @aignas 2024-04-05: how do we support different requirement
             # cycles for different abis/oses? For now we will need the users to
@@ -653,6 +603,14 @@ You cannot use both the additive_build_content and additive_build_content_file a
             for setting, repo in settings.items():
                 hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
 
+        whl_libraries.update(hub.whl_libraries)
+        for whl_name, lib in hub.whl_libraries.items():
+            if whl_name in lib:
+                fail("'{}' already in created".format(whl_name))
+            else:
+                # replicate whl_libraries.update(out.whl_libraries)
+                whl_libraries[whl_name] = lib
+
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
         # args are manipulated in the code going before.
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index e74e5072c0..b3b8aa06b6 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -3,6 +3,7 @@
 load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:version.bzl", "version")
+load("//python/private:version_label.bzl", "version_label")
 load(":pep508_env.bzl", "env")
 load(":pep508_evaluate.bzl", "evaluate")
 load(":python_tag.bzl", "python_tag")
@@ -50,6 +51,7 @@ def hub_builder(
         get_index_urls = lambda *a, **k: _get_index_urls(self, *a, **k),
         detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
         platforms = lambda version: self.python_versions[version],
+        add_whl_library = lambda *a, **k: _add_whl_library(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
@@ -175,3 +177,46 @@ def _platforms(*, python_version, minor_mapping, config):
             whl_platform_tags = values.whl_platform_tags,
         )
     return platforms
+
+def _add_whl_library(self, *, python_version, whl, repo):
+    if repo == None:
+        # NOTE @aignas 2025-07-07: we guard against an edge-case where there
+        # are more platforms defined than there are wheels for and users
+        # disallow building from sdist.
+        return
+
+    platforms = self.python_versions[python_version]
+
+    # TODO @aignas 2025-06-29: we should not need the version in the repo_name if
+    # we are using pipstar and we are downloading the wheel using the downloader
+    repo_name = "{}_{}_{}".format(self.name, version_label(python_version), repo.repo_name)
+
+    if repo_name in self.whl_libraries:
+        fail("attempting to create a duplicate library {} for {}".format(
+            repo_name,
+            whl.name,
+        ))
+    self.whl_libraries[repo_name] = repo.args
+
+    if not self.config.enable_pipstar and "experimental_target_platforms" in repo.args:
+        self.whl_libraries[repo_name] |= {
+            "experimental_target_platforms": sorted({
+                # TODO @aignas 2025-07-07: this should be solved in a better way
+                platforms[candidate].triple.partition("_")[-1]: None
+                for p in repo.args["experimental_target_platforms"]
+                for candidate in platforms
+                if candidate.endswith(p)
+            }),
+        }
+
+    mapping = self.whl_map.setdefault(whl.name, {})
+    if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
+        fail(
+            "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
+                mapping[repo.config_setting],
+                repo.config_setting,
+                repo_name,
+            ),
+        )
+    else:
+        mapping[repo.config_setting] = repo_name

From 76d082eb36fcd9dcf449e0af501ef6bbb86fb984 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 15:40:01 +0900
Subject: [PATCH 09/31] refactor: get_index_urls early return

---
 python/private/pypi/extension.bzl   |  6 +--
 python/private/pypi/hub_builder.bzl | 69 +++++++++++++++--------------
 2 files changed, 38 insertions(+), 37 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 20a3f2ef18..21608b5f6a 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -97,7 +97,7 @@ def _create_whl_repos(
             rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
-    get_index_urls = hub.get_index_urls(pip_attr)
+    get_index_urls = hub.get_index_urls(pip_attr.python_version)
     interpreter = hub.detect_interpreter(pip_attr)
 
     # containers to aggregate outputs from this function
@@ -564,9 +564,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
             else:
                 builder = pip_hub_map[pip_attr.hub_name]
 
-            builder.add(
-                python_version = pip_attr.python_version,
-            )
+            builder.add(pip_attr = pip_attr)
 
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
             out = _create_whl_repos(
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index b3b8aa06b6..595dc70a37 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -46,10 +46,11 @@ def hub_builder(
         _available_interpreters = available_interpreters,
         _simpleapi_download_fn = simpleapi_download_fn,
         _simpleapi_cache = simpleapi_cache,
+        _get_index_urls = {},
         # keep sorted
         add = lambda *a, **k: _add(self, *a, **k),
-        get_index_urls = lambda *a, **k: _get_index_urls(self, *a, **k),
         detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
+        get_index_urls = lambda version: self._get_index_urls[version],
         platforms = lambda version: self.python_versions[version],
         add_whl_library = lambda *a, **k: _add_whl_library(self, *a, **k),
     )
@@ -57,7 +58,8 @@ def hub_builder(
     # buildifier: enable=uninitialized
     return self
 
-def _add(self, *, python_version):
+def _add(self, *, pip_attr):
+    python_version = pip_attr.python_version
     if python_version in self.python_versions:
         fail((
             "Duplicate pip python version '{version}' for hub " +
@@ -74,39 +76,40 @@ def _add(self, *, python_version):
         minor_mapping = self._minor_mapping,
         config = self.config,
     )
+    self._get_index_urls[python_version] = _get_index_urls(self, pip_attr)
 
 def _get_index_urls(self, pip_attr):
-    get_index_urls = None
-    if pip_attr.experimental_index_url:
-        skip_sources = [
-            normalize_name(s)
-            for s in pip_attr.simpleapi_skip
-        ]
-        get_index_urls = lambda ctx, distributions: self._simpleapi_download_fn(
-            ctx,
-            attr = struct(
-                index_url = pip_attr.experimental_index_url,
-                extra_index_urls = pip_attr.experimental_extra_index_urls or [],
-                index_url_overrides = pip_attr.experimental_index_url_overrides or {},
-                sources = [
-                    d
-                    for d in distributions
-                    if normalize_name(d) not in skip_sources
-                ],
-                envsubst = pip_attr.envsubst,
-                # Auth related info
-                netrc = pip_attr.netrc,
-                auth_patterns = pip_attr.auth_patterns,
-            ),
-            cache = self._simpleapi_cache,
-            parallel_download = pip_attr.parallel_download,
-        )
-    elif pip_attr.experimental_extra_index_urls:
-        fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")
-    elif pip_attr.experimental_index_url_overrides:
-        fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
-
-    return get_index_urls
+    if not pip_attr.experimental_index_url:
+        if pip_attr.experimental_extra_index_urls:
+            fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")
+        elif pip_attr.experimental_index_url_overrides:
+            fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
+
+        return None
+
+    skip_sources = [
+        normalize_name(s)
+        for s in pip_attr.simpleapi_skip
+    ]
+    return lambda ctx, distributions: self._simpleapi_download_fn(
+        ctx,
+        attr = struct(
+            index_url = pip_attr.experimental_index_url,
+            extra_index_urls = pip_attr.experimental_extra_index_urls or [],
+            index_url_overrides = pip_attr.experimental_index_url_overrides or {},
+            sources = [
+                d
+                for d in distributions
+                if normalize_name(d) not in skip_sources
+            ],
+            envsubst = pip_attr.envsubst,
+            # Auth related info
+            netrc = pip_attr.netrc,
+            auth_patterns = pip_attr.auth_patterns,
+        ),
+        cache = self._simpleapi_cache,
+        parallel_download = pip_attr.parallel_download,
+    )
 
 def _detect_interpreter(self, pip_attr):
     python_interpreter_target = pip_attr.python_interpreter_target

From 044b004510ff31d42666ac734965d4b4ba30c5bc Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 15:51:41 +0900
Subject: [PATCH 10/31] refactor: add a method for checking if we should use
 the downloader

---
 python/private/pypi/extension.bzl   | 12 ++--------
 python/private/pypi/hub_builder.bzl | 36 +++++++++++++++++++++--------
 2 files changed, 28 insertions(+), 20 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 21608b5f6a..bd3e3903a6 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -97,7 +97,6 @@ def _create_whl_repos(
             rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
-    get_index_urls = hub.get_index_urls(pip_attr.python_version)
     interpreter = hub.detect_interpreter(pip_attr)
 
     # containers to aggregate outputs from this function
@@ -184,15 +183,11 @@ def _create_whl_repos(
         ),
         platforms = platforms,
         extra_pip_args = pip_attr.extra_pip_args,
-        get_index_urls = get_index_urls,
+        get_index_urls = hub.get_index_urls(pip_attr.python_version),
         evaluate_markers = evaluate_markers,
         logger = logger,
     )
 
-    use_downloader = {
-        normalize_name(s): False
-        for s in pip_attr.simpleapi_skip
-    }
     exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
@@ -246,10 +241,7 @@ def _create_whl_repos(
                 whl_library_args = whl_library_args,
                 download_only = pip_attr.download_only,
                 netrc = hub.config.netrc or pip_attr.netrc,
-                use_downloader = use_downloader.get(
-                    whl.name,
-                    get_index_urls != None,  # defaults to True if the get_index_urls is defined
-                ),
+                use_downloader = hub.use_downloader(pip_attr.python_version, whl.name),
                 auth_patterns = hub.config.auth_patterns or pip_attr.auth_patterns,
                 python_version = _major_minor_version(pip_attr.python_version),
                 is_multiple_versions = whl.is_multiple_versions,
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 595dc70a37..3e42b6864f 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -47,12 +47,18 @@ def hub_builder(
         _simpleapi_download_fn = simpleapi_download_fn,
         _simpleapi_cache = simpleapi_cache,
         _get_index_urls = {},
+        _pip_attrs = {},
+        _use_downloader = {},
         # keep sorted
         add = lambda *a, **k: _add(self, *a, **k),
         detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
-        get_index_urls = lambda version: self._get_index_urls[version],
+        get_index_urls = lambda version: self._get_index_urls.get(version),
         platforms = lambda version: self.python_versions[version],
         add_whl_library = lambda *a, **k: _add_whl_library(self, *a, **k),
+        use_downloader = lambda python_version, whl_name: self._use_downloader.get(python_version, {}).get(
+            normalize_name(whl_name),
+            python_version in self._get_index_urls,
+        ),
     )
 
     # buildifier: enable=uninitialized
@@ -76,22 +82,32 @@ def _add(self, *, pip_attr):
         minor_mapping = self._minor_mapping,
         config = self.config,
     )
-    self._get_index_urls[python_version] = _get_index_urls(self, pip_attr)
+    _set_index_urls(self, pip_attr)
+    self._pip_attrs[python_version] = pip_attr
 
-def _get_index_urls(self, pip_attr):
+def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:
             fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")
         elif pip_attr.experimental_index_url_overrides:
             fail("'experimental_index_url_overrides' is a no-op unless 'experimental_index_url' is set")
+        elif pip_attr.simpleapi_skip:
+            fail("'simpleapi_skip' is a no-op unless 'experimental_index_url' is set")
+        elif pip_attr.netrc:
+            fail("'netrc' is a no-op unless 'experimental_index_url' is set")
+        elif pip_attr.auth_patterns:
+            fail("'auth_patterns' is a no-op unless 'experimental_index_url' is set")
+
+        # parallel_download is set to True by default, so we are not checking/validating it
+        # here
+        return
 
-        return None
-
-    skip_sources = [
-        normalize_name(s)
+    python_version = pip_attr.python_version
+    self._use_downloader.setdefault(python_version, {}).update({
+        normalize_name(s): False
         for s in pip_attr.simpleapi_skip
-    ]
-    return lambda ctx, distributions: self._simpleapi_download_fn(
+    })
+    self._get_index_urls[python_version] = lambda ctx, distributions: self._simpleapi_download_fn(
         ctx,
         attr = struct(
             index_url = pip_attr.experimental_index_url,
@@ -100,7 +116,7 @@ def _get_index_urls(self, pip_attr):
             sources = [
                 d
                 for d in distributions
-                if normalize_name(d) not in skip_sources
+                if self.use_downloader(python_version, d)
             ],
             envsubst = pip_attr.envsubst,
             # Auth related info

From 0f83603e3ef9a393aeb784c9e3f5970f60ce7fca Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 16:05:34 +0900
Subject: [PATCH 11/31] refactor: move evaluate_markers fn creation to the
 builder

---
 python/private/pypi/BUILD.bazel     |  1 +
 python/private/pypi/extension.bzl   | 52 ++++-------------------------
 python/private/pypi/hub_builder.bzl | 49 +++++++++++++++++++++++++++
 3 files changed, 57 insertions(+), 45 deletions(-)

diff --git a/python/private/pypi/BUILD.bazel b/python/private/pypi/BUILD.bazel
index 73451b9784..ea41a3cf24 100644
--- a/python/private/pypi/BUILD.bazel
+++ b/python/private/pypi/BUILD.bazel
@@ -172,6 +172,7 @@ bzl_library(
     srcs = ["hub_builder.bzl"],
     visibility = ["//:__subpackages__"],
     deps = [
+        ":evaluate_markers_bzl",
         ":pep508_env_bzl",
         ":pep508_evaluate_bzl",
         ":python_tag_bzl",
diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index bd3e3903a6..084e009d3c 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -24,7 +24,7 @@ load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:repo_utils.bzl", "repo_utils")
 load("//python/private:version.bzl", "version")
 load(":attrs.bzl", "use_isolated")
-load(":evaluate_markers.bzl", "evaluate_markers_py", EVALUATE_MARKERS_SRCS = "SRCS", evaluate_markers_star = "evaluate_markers")
+load(":evaluate_markers.bzl", EVALUATE_MARKERS_SRCS = "SRCS")
 load(":hub_builder.bzl", "hub_builder")
 load(":hub_repository.bzl", "hub_repository", "whl_config_settings_to_json")
 load(":parse_requirements.bzl", "parse_requirements")
@@ -72,8 +72,7 @@ def _create_whl_repos(
         pip_attr,
         whl_overrides,
         hub,
-        minor_mapping = MINOR_MAPPING,
-        evaluate_markers = None):
+        minor_mapping = MINOR_MAPPING):
     """create all of the whl repositories
 
     Args:
@@ -83,7 +82,6 @@ def _create_whl_repos(
         hub: TODO.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
-        evaluate_markers: the function used to evaluate the markers.
 
     Returns a {type}`struct` with the following attributes:
         whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
@@ -127,44 +125,6 @@ def _create_whl_repos(
 
     platforms = hub.platforms(pip_attr.python_version)
 
-    if evaluate_markers:
-        # This is most likely unit tests
-        pass
-    elif hub.config.enable_pipstar:
-        evaluate_markers = lambda _, requirements: evaluate_markers_star(
-            requirements = requirements,
-            platforms = platforms,
-        )
-    else:
-        # NOTE @aignas 2024-08-02: , we will execute any interpreter that we find either
-        # in the PATH or if specified as a label. We will configure the env
-        # markers when evaluating the requirement lines based on the output
-        # from the `requirements_files_by_platform` which should have something
-        # similar to:
-        # {
-        #    "//:requirements.txt": ["cp311_linux_x86_64", ...]
-        # }
-        #
-        # We know the target python versions that we need to evaluate the
-        # markers for and thus we don't need to use multiple python interpreter
-        # instances to perform this manipulation. This function should be executed
-        # only once by the underlying code to minimize the overhead needed to
-        # spin up a Python interpreter.
-        evaluate_markers = lambda module_ctx, requirements: evaluate_markers_py(
-            module_ctx,
-            requirements = {
-                k: {
-                    p: platforms[p].triple
-                    for p in plats
-                }
-                for k, plats in requirements.items()
-            },
-            python_interpreter = interpreter.path,
-            python_interpreter_target = interpreter.target,
-            srcs = pip_attr._evaluate_markers_srcs,
-            logger = logger,
-        )
-
     requirements_by_platform = parse_requirements(
         module_ctx,
         requirements_by_platform = requirements_files_by_platform(
@@ -184,7 +144,7 @@ def _create_whl_repos(
         platforms = platforms,
         extra_pip_args = pip_attr.extra_pip_args,
         get_index_urls = hub.get_index_urls(pip_attr.python_version),
-        evaluate_markers = evaluate_markers,
+        evaluate_markers = hub.evaluate_markers(pip_attr),
         logger = logger,
     )
 
@@ -536,7 +496,9 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     simpleapi_download_fn = simpleapi_download,
                     simpleapi_cache = simpleapi_cache,
                     minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
-                    available_interpreters = kwargs.pop("available_interpreters", INTERPRETER_LABELS),
+                    evaluate_markers_fn = kwargs.get("evaluate_markers", None),
+                    available_interpreters = kwargs.get("available_interpreters", INTERPRETER_LABELS),
+                    logger = repo_utils.logger(module_ctx, "pypi:hub:" + hub_name),
                 )
                 pip_hub_map[pip_attr.hub_name] = builder
             elif pip_hub_map[hub_name].module_name != mod.name:
@@ -564,7 +526,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 hub = builder,
                 pip_attr = pip_attr,
                 whl_overrides = whl_overrides,
-                **kwargs
+                minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
             extra_aliases.setdefault(hub_name, {})
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 3e42b6864f..cf69cd5689 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -4,6 +4,7 @@ load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:version.bzl", "version")
 load("//python/private:version_label.bzl", "version_label")
+load(":evaluate_markers.bzl", "evaluate_markers_py", evaluate_markers_star = "evaluate_markers")
 load(":pep508_env.bzl", "env")
 load(":pep508_evaluate.bzl", "evaluate")
 load(":python_tag.bzl", "python_tag")
@@ -16,6 +17,8 @@ def hub_builder(
         minor_mapping,
         available_interpreters,
         simpleapi_download_fn,
+        evaluate_markers_fn,
+        logger,
         simpleapi_cache = {}):
     """Return a hub builder instance
 
@@ -24,12 +27,14 @@ def hub_builder(
         module_name: TODO
         config: The platform configuration.
         minor_mapping: TODO
+        evaluate_markers_fn: the function used to evaluate the markers.
         available_interpreters: {type}`dict[str, Label]` The dictionary of available
             interpreters that have been registered using the `python` bzlmod extension.
             The keys are in the form `python_{snake_case_version}_host`. This is to be
             used during the `repository_rule` and must be always compatible with the host.
         simpleapi_download_fn: TODO
         simpleapi_cache: TODO
+        logger: TODO
     """
 
     # buildifier: disable=uninitialized
@@ -42,6 +47,8 @@ def hub_builder(
         exposed_packages = {},
         extra_aliases = {},
         whl_libraries = {},
+        _evaluate_markers_fn = evaluate_markers_fn,
+        _logger = logger,
         _minor_mapping = minor_mapping,
         _available_interpreters = available_interpreters,
         _simpleapi_download_fn = simpleapi_download_fn,
@@ -59,6 +66,7 @@ def hub_builder(
             normalize_name(whl_name),
             python_version in self._get_index_urls,
         ),
+        evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
@@ -239,3 +247,44 @@ def _add_whl_library(self, *, python_version, whl, repo):
         )
     else:
         mapping[repo.config_setting] = repo_name
+
+def _evaluate_markers(self, pip_attr):
+    if self._evaluate_markers_fn:
+        return self._evaluate_markers_fn
+
+    if self.config.enable_pipstar:
+        return lambda _, requirements: evaluate_markers_star(
+            requirements = requirements,
+            platforms = self.python_versions[pip_attr.python_version],
+        )
+
+    interpreter = self.detect_interpreter(pip_attr)
+
+    # NOTE @aignas 2024-08-02: , we will execute any interpreter that we find either
+    # in the PATH or if specified as a label. We will configure the env
+    # markers when evaluating the requirement lines based on the output
+    # from the `requirements_files_by_platform` which should have something
+    # similar to:
+    # {
+    #    "//:requirements.txt": ["cp311_linux_x86_64", ...]
+    # }
+    #
+    # We know the target python versions that we need to evaluate the
+    # markers for and thus we don't need to use multiple python interpreter
+    # instances to perform this manipulation. This function should be executed
+    # only once by the underlying code to minimize the overhead needed to
+    # spin up a Python interpreter.
+    return lambda module_ctx, requirements: evaluate_markers_py(
+        module_ctx,
+        requirements = {
+            k: {
+                p: self.python_versions[pip_attr.python_version][p].triple
+                for p in plats
+            }
+            for k, plats in requirements.items()
+        },
+        python_interpreter = interpreter.path,
+        python_interpreter_target = interpreter.target,
+        srcs = pip_attr._evaluate_markers_srcs,
+        logger = self._logger,
+    )

From 7988d4470a71c6d6c244fc25e456b926ab8b4299 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 16:07:55 +0900
Subject: [PATCH 12/31] minor: code reshuffling

---
 python/private/pypi/extension.bzl | 41 ++++++++++++++-----------------
 1 file changed, 18 insertions(+), 23 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 084e009d3c..3851199667 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -97,34 +97,12 @@ def _create_whl_repos(
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
     interpreter = hub.detect_interpreter(pip_attr)
 
-    # containers to aggregate outputs from this function
-    extra_aliases = {
-        whl_name: {alias: True for alias in aliases}
-        for whl_name, aliases in pip_attr.extra_hub_aliases.items()
-    }
-
     whl_modifications = {}
     if pip_attr.whl_modifications != None:
         for mod, whl_name in pip_attr.whl_modifications.items():
             whl_modifications[normalize_name(whl_name)] = mod
 
-    if pip_attr.experimental_requirement_cycles:
-        requirement_cycles = {
-            name: [normalize_name(whl_name) for whl_name in whls]
-            for name, whls in pip_attr.experimental_requirement_cycles.items()
-        }
-
-        whl_group_mapping = {
-            whl_name: group_name
-            for group_name, group_whls in requirement_cycles.items()
-            for whl_name in group_whls
-        }
-    else:
-        whl_group_mapping = {}
-        requirement_cycles = {}
-
     platforms = hub.platforms(pip_attr.python_version)
-
     requirements_by_platform = parse_requirements(
         module_ctx,
         requirements_by_platform = requirements_files_by_platform(
@@ -148,6 +126,20 @@ def _create_whl_repos(
         logger = logger,
     )
 
+    if pip_attr.experimental_requirement_cycles:
+        requirement_cycles = {
+            name: [normalize_name(whl_name) for whl_name in whls]
+            for name, whls in pip_attr.experimental_requirement_cycles.items()
+        }
+
+        whl_group_mapping = {
+            whl_name: group_name
+            for group_name, group_whls in requirement_cycles.items()
+            for whl_name in group_whls
+        }
+    else:
+        whl_group_mapping = {}
+        requirement_cycles = {}
     exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
@@ -215,7 +207,10 @@ def _create_whl_repos(
 
     return struct(
         exposed_packages = exposed_packages,
-        extra_aliases = extra_aliases,
+        extra_aliases = {
+            whl_name: {alias: True for alias in aliases}
+            for whl_name, aliases in pip_attr.extra_hub_aliases.items()
+        },
     )
 
 def _whl_repo(

From 4e754f7606a053109d9df948740588ea53d7d86e Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 16:14:30 +0900
Subject: [PATCH 13/31] minor: use hub builder for all hub output storage

---
 python/private/pypi/extension.bzl   | 61 ++++++++++-------------------
 python/private/pypi/hub_builder.bzl | 23 +++++++++++
 2 files changed, 43 insertions(+), 41 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 3851199667..45968bf47c 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -66,7 +66,7 @@ def _whl_mods_impl(whl_mods_dict):
             whl_mods = whl_mods,
         )
 
-def _create_whl_repos(
+def _add_whl_repos(
         module_ctx,
         *,
         pip_attr,
@@ -82,17 +82,6 @@ def _create_whl_repos(
         hub: TODO.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
-
-    Returns a {type}`struct` with the following attributes:
-        whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
-            normalized package name and the values are the instances of the
-            {bzl:obj}`whl_config_setting` return values.
-        exposed_packages: {type}`dict[str, Any]` this is just a way to
-            represent a set of string values.
-        whl_libraries: {type}`dict[str, dict[str, Any]]` the keys are the
-            aparent repository names for the hub repo and the values are the
-            arguments that will be passed to {bzl:obj}`whl_library` repository
-            rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
     interpreter = hub.detect_interpreter(pip_attr)
@@ -126,6 +115,7 @@ def _create_whl_repos(
         logger = logger,
     )
 
+    exposed_packages = {}
     if pip_attr.experimental_requirement_cycles:
         requirement_cycles = {
             name: [normalize_name(whl_name) for whl_name in whls]
@@ -140,7 +130,6 @@ def _create_whl_repos(
     else:
         whl_group_mapping = {}
         requirement_cycles = {}
-    exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
             exposed_packages[whl.name] = None
@@ -205,13 +194,17 @@ def _create_whl_repos(
                 repo = repo,
             )
 
-    return struct(
-        exposed_packages = exposed_packages,
-        extra_aliases = {
-            whl_name: {alias: True for alias in aliases}
-            for whl_name, aliases in pip_attr.extra_hub_aliases.items()
-        },
-    )
+    if not hub.exposed_packages:
+        hub.exposed_packages.update(exposed_packages)
+    else:
+        intersection = {}
+        for pkg in exposed_packages:
+            if pkg not in hub.exposed_packages:
+                continue
+            intersection[pkg] = None
+
+        hub.exposed_packages.clear()
+        hub.exposed_packages.update(intersection)
 
 def _whl_repo(
         *,
@@ -516,7 +509,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
             builder.add(pip_attr = pip_attr)
 
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
-            out = _create_whl_repos(
+            _add_whl_repos(
                 module_ctx,
                 hub = builder,
                 pip_attr = pip_attr,
@@ -524,32 +517,15 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
-            extra_aliases.setdefault(hub_name, {})
-            for whl_name, aliases in out.extra_aliases.items():
-                extra_aliases[hub_name].setdefault(whl_name, {}).update(aliases)
-
-            if hub_name not in exposed_packages:
-                exposed_packages[hub_name] = out.exposed_packages
-            else:
-                intersection = {}
-                for pkg in out.exposed_packages:
-                    if pkg not in exposed_packages[hub_name]:
-                        continue
-                    intersection[pkg] = None
-                exposed_packages[hub_name] = intersection
-
-            # TODO @aignas 2024-04-05: how do we support different requirement
-            # cycles for different abis/oses? For now we will need the users to
-            # assume the same groups across all versions/platforms until we start
-            # using an alternative cycle resolution strategy.
-            hub_group_map[hub_name] = pip_attr.experimental_requirement_cycles
-
     for hub in pip_hub_map.values():
         hub_whl_map.setdefault(hub.name, {})
         for key, settings in hub.whl_map.items():
             for setting, repo in settings.items():
                 hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
 
+        for whl_name, aliases in hub.extra_aliases.items():
+            extra_aliases[hub.name].setdefault(whl_name, {}).update(aliases)
+
         whl_libraries.update(hub.whl_libraries)
         for whl_name, lib in hub.whl_libraries.items():
             if whl_name in lib:
@@ -558,6 +534,9 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
+        hub_group_map[hub.name] = hub.group_map
+        exposed_packages[hub.name] = hub.exposed_packages
+
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
         # args are manipulated in the code going before.
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index cf69cd5689..0dac160eb7 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -35,6 +35,18 @@ def hub_builder(
         simpleapi_download_fn: TODO
         simpleapi_cache: TODO
         logger: TODO
+
+    TODO
+    Returns a {type}`struct` with the following attributes:
+        whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
+            normalized package name and the values are the instances of the
+            {bzl:obj}`whl_config_setting` return values.
+        exposed_packages: {type}`dict[str, Any]` this is just a way to
+            represent a set of string values.
+        whl_libraries: {type}`dict[str, dict[str, Any]]` the keys are the
+            aparent repository names for the hub repo and the values are the
+            arguments that will be passed to {bzl:obj}`whl_library` repository
+            rule.
     """
 
     # buildifier: disable=uninitialized
@@ -93,6 +105,17 @@ def _add(self, *, pip_attr):
     _set_index_urls(self, pip_attr)
     self._pip_attrs[python_version] = pip_attr
 
+    self.extra_aliases.update({
+        whl_name: {alias: True for alias in aliases}
+        for whl_name, aliases in pip_attr.extra_hub_aliases.items()
+    })
+
+    # TODO @aignas 2024-04-05: how do we support different requirement
+    # cycles for different abis/oses? For now we will need the users to
+    # assume the same groups across all versions/platforms until we start
+    # using an alternative cycle resolution strategy.
+    self.group_map = pip_attr.experimental_requirement_cycles
+
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:

From 6064ab7287c5d3231655d892ae88236a8358ad30 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 16:22:11 +0900
Subject: [PATCH 14/31] wip

---
 python/private/pypi/extension.bzl   | 15 +++++++--------
 python/private/pypi/hub_builder.bzl | 14 +++++++++++++-
 2 files changed, 20 insertions(+), 9 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 45968bf47c..653673f28a 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -518,24 +518,23 @@ You cannot use both the additive_build_content and additive_build_content_file a
             )
 
     for hub in pip_hub_map.values():
+        out = hub.build()
         hub_whl_map.setdefault(hub.name, {})
-        for key, settings in hub.whl_map.items():
+        for key, settings in out.whl_map.items():
             for setting, repo in settings.items():
                 hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
 
-        for whl_name, aliases in hub.extra_aliases.items():
-            extra_aliases[hub.name].setdefault(whl_name, {}).update(aliases)
-
-        whl_libraries.update(hub.whl_libraries)
-        for whl_name, lib in hub.whl_libraries.items():
+        for whl_name, lib in out.whl_libraries.items():
             if whl_name in lib:
                 fail("'{}' already in created".format(whl_name))
             else:
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
-        hub_group_map[hub.name] = hub.group_map
-        exposed_packages[hub.name] = hub.exposed_packages
+        hub_whl_map[hub.name] = out.whl_map
+        extra_aliases[hub.name] = out.extra_aliases
+        hub_group_map[hub.name] = out.group_map
+        exposed_packages[hub.name] = out.exposed_packages
 
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 0dac160eb7..8172f2bf47 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -59,6 +59,7 @@ def hub_builder(
         exposed_packages = {},
         extra_aliases = {},
         whl_libraries = {},
+        group_map = {},
         _evaluate_markers_fn = evaluate_markers_fn,
         _logger = logger,
         _minor_mapping = minor_mapping,
@@ -79,11 +80,21 @@ def hub_builder(
             python_version in self._get_index_urls,
         ),
         evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
+        build = lambda: _build(self),
     )
 
     # buildifier: enable=uninitialized
     return self
 
+def _build(self):
+    return struct(
+        whl_map = self.whl_map,
+        exposed_packages = self.exposed_packages,
+        extra_aliases = self.extra_aliases,
+        whl_libraries = self.whl_libraries,
+        group_map = self.group_map,
+    )
+
 def _add(self, *, pip_attr):
     python_version = pip_attr.python_version
     if python_version in self.python_versions:
@@ -114,7 +125,8 @@ def _add(self, *, pip_attr):
     # cycles for different abis/oses? For now we will need the users to
     # assume the same groups across all versions/platforms until we start
     # using an alternative cycle resolution strategy.
-    self.group_map = pip_attr.experimental_requirement_cycles
+    self.group_map.clear()
+    self.group_map.update(pip_attr.experimental_requirement_cycles)
 
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:

From b9bf362ab3b43c1078371dbbee3c78bb022cd1b5 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 21:50:37 +0900
Subject: [PATCH 15/31] Revert "wip"

This reverts commit 6064ab7287c5d3231655d892ae88236a8358ad30.
---
 python/private/pypi/extension.bzl   | 15 ++++++++-------
 python/private/pypi/hub_builder.bzl | 14 +-------------
 2 files changed, 9 insertions(+), 20 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 653673f28a..45968bf47c 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -518,23 +518,24 @@ You cannot use both the additive_build_content and additive_build_content_file a
             )
 
     for hub in pip_hub_map.values():
-        out = hub.build()
         hub_whl_map.setdefault(hub.name, {})
-        for key, settings in out.whl_map.items():
+        for key, settings in hub.whl_map.items():
             for setting, repo in settings.items():
                 hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
 
-        for whl_name, lib in out.whl_libraries.items():
+        for whl_name, aliases in hub.extra_aliases.items():
+            extra_aliases[hub.name].setdefault(whl_name, {}).update(aliases)
+
+        whl_libraries.update(hub.whl_libraries)
+        for whl_name, lib in hub.whl_libraries.items():
             if whl_name in lib:
                 fail("'{}' already in created".format(whl_name))
             else:
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
-        hub_whl_map[hub.name] = out.whl_map
-        extra_aliases[hub.name] = out.extra_aliases
-        hub_group_map[hub.name] = out.group_map
-        exposed_packages[hub.name] = out.exposed_packages
+        hub_group_map[hub.name] = hub.group_map
+        exposed_packages[hub.name] = hub.exposed_packages
 
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 8172f2bf47..0dac160eb7 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -59,7 +59,6 @@ def hub_builder(
         exposed_packages = {},
         extra_aliases = {},
         whl_libraries = {},
-        group_map = {},
         _evaluate_markers_fn = evaluate_markers_fn,
         _logger = logger,
         _minor_mapping = minor_mapping,
@@ -80,21 +79,11 @@ def hub_builder(
             python_version in self._get_index_urls,
         ),
         evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
-        build = lambda: _build(self),
     )
 
     # buildifier: enable=uninitialized
     return self
 
-def _build(self):
-    return struct(
-        whl_map = self.whl_map,
-        exposed_packages = self.exposed_packages,
-        extra_aliases = self.extra_aliases,
-        whl_libraries = self.whl_libraries,
-        group_map = self.group_map,
-    )
-
 def _add(self, *, pip_attr):
     python_version = pip_attr.python_version
     if python_version in self.python_versions:
@@ -125,8 +114,7 @@ def _add(self, *, pip_attr):
     # cycles for different abis/oses? For now we will need the users to
     # assume the same groups across all versions/platforms until we start
     # using an alternative cycle resolution strategy.
-    self.group_map.clear()
-    self.group_map.update(pip_attr.experimental_requirement_cycles)
+    self.group_map = pip_attr.experimental_requirement_cycles
 
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:

From c8e5304bd6993becd59efff3aef8746cf769b634 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 21:50:48 +0900
Subject: [PATCH 16/31] Revert "minor: use hub builder for all hub output
 storage"

This reverts commit 4e754f7606a053109d9df948740588ea53d7d86e.
---
 python/private/pypi/extension.bzl   | 61 +++++++++++++++++++----------
 python/private/pypi/hub_builder.bzl | 23 -----------
 2 files changed, 41 insertions(+), 43 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 45968bf47c..3851199667 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -66,7 +66,7 @@ def _whl_mods_impl(whl_mods_dict):
             whl_mods = whl_mods,
         )
 
-def _add_whl_repos(
+def _create_whl_repos(
         module_ctx,
         *,
         pip_attr,
@@ -82,6 +82,17 @@ def _add_whl_repos(
         hub: TODO.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
+
+    Returns a {type}`struct` with the following attributes:
+        whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
+            normalized package name and the values are the instances of the
+            {bzl:obj}`whl_config_setting` return values.
+        exposed_packages: {type}`dict[str, Any]` this is just a way to
+            represent a set of string values.
+        whl_libraries: {type}`dict[str, dict[str, Any]]` the keys are the
+            aparent repository names for the hub repo and the values are the
+            arguments that will be passed to {bzl:obj}`whl_library` repository
+            rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
     interpreter = hub.detect_interpreter(pip_attr)
@@ -115,7 +126,6 @@ def _add_whl_repos(
         logger = logger,
     )
 
-    exposed_packages = {}
     if pip_attr.experimental_requirement_cycles:
         requirement_cycles = {
             name: [normalize_name(whl_name) for whl_name in whls]
@@ -130,6 +140,7 @@ def _add_whl_repos(
     else:
         whl_group_mapping = {}
         requirement_cycles = {}
+    exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
             exposed_packages[whl.name] = None
@@ -194,17 +205,13 @@ def _add_whl_repos(
                 repo = repo,
             )
 
-    if not hub.exposed_packages:
-        hub.exposed_packages.update(exposed_packages)
-    else:
-        intersection = {}
-        for pkg in exposed_packages:
-            if pkg not in hub.exposed_packages:
-                continue
-            intersection[pkg] = None
-
-        hub.exposed_packages.clear()
-        hub.exposed_packages.update(intersection)
+    return struct(
+        exposed_packages = exposed_packages,
+        extra_aliases = {
+            whl_name: {alias: True for alias in aliases}
+            for whl_name, aliases in pip_attr.extra_hub_aliases.items()
+        },
+    )
 
 def _whl_repo(
         *,
@@ -509,7 +516,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
             builder.add(pip_attr = pip_attr)
 
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
-            _add_whl_repos(
+            out = _create_whl_repos(
                 module_ctx,
                 hub = builder,
                 pip_attr = pip_attr,
@@ -517,15 +524,32 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
+            extra_aliases.setdefault(hub_name, {})
+            for whl_name, aliases in out.extra_aliases.items():
+                extra_aliases[hub_name].setdefault(whl_name, {}).update(aliases)
+
+            if hub_name not in exposed_packages:
+                exposed_packages[hub_name] = out.exposed_packages
+            else:
+                intersection = {}
+                for pkg in out.exposed_packages:
+                    if pkg not in exposed_packages[hub_name]:
+                        continue
+                    intersection[pkg] = None
+                exposed_packages[hub_name] = intersection
+
+            # TODO @aignas 2024-04-05: how do we support different requirement
+            # cycles for different abis/oses? For now we will need the users to
+            # assume the same groups across all versions/platforms until we start
+            # using an alternative cycle resolution strategy.
+            hub_group_map[hub_name] = pip_attr.experimental_requirement_cycles
+
     for hub in pip_hub_map.values():
         hub_whl_map.setdefault(hub.name, {})
         for key, settings in hub.whl_map.items():
             for setting, repo in settings.items():
                 hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
 
-        for whl_name, aliases in hub.extra_aliases.items():
-            extra_aliases[hub.name].setdefault(whl_name, {}).update(aliases)
-
         whl_libraries.update(hub.whl_libraries)
         for whl_name, lib in hub.whl_libraries.items():
             if whl_name in lib:
@@ -534,9 +558,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
-        hub_group_map[hub.name] = hub.group_map
-        exposed_packages[hub.name] = hub.exposed_packages
-
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
         # args are manipulated in the code going before.
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 0dac160eb7..cf69cd5689 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -35,18 +35,6 @@ def hub_builder(
         simpleapi_download_fn: TODO
         simpleapi_cache: TODO
         logger: TODO
-
-    TODO
-    Returns a {type}`struct` with the following attributes:
-        whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
-            normalized package name and the values are the instances of the
-            {bzl:obj}`whl_config_setting` return values.
-        exposed_packages: {type}`dict[str, Any]` this is just a way to
-            represent a set of string values.
-        whl_libraries: {type}`dict[str, dict[str, Any]]` the keys are the
-            aparent repository names for the hub repo and the values are the
-            arguments that will be passed to {bzl:obj}`whl_library` repository
-            rule.
     """
 
     # buildifier: disable=uninitialized
@@ -105,17 +93,6 @@ def _add(self, *, pip_attr):
     _set_index_urls(self, pip_attr)
     self._pip_attrs[python_version] = pip_attr
 
-    self.extra_aliases.update({
-        whl_name: {alias: True for alias in aliases}
-        for whl_name, aliases in pip_attr.extra_hub_aliases.items()
-    })
-
-    # TODO @aignas 2024-04-05: how do we support different requirement
-    # cycles for different abis/oses? For now we will need the users to
-    # assume the same groups across all versions/platforms until we start
-    # using an alternative cycle resolution strategy.
-    self.group_map = pip_attr.experimental_requirement_cycles
-
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:

From f902694e6d7d3a64114c7240a1ca97ffd47ec19d Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 21:53:02 +0900
Subject: [PATCH 17/31] refactor: move group_map to builder

---
 python/private/pypi/extension.bzl   | 8 ++------
 python/private/pypi/hub_builder.bzl | 8 ++++++++
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 3851199667..a161740cb4 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -538,12 +538,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     intersection[pkg] = None
                 exposed_packages[hub_name] = intersection
 
-            # TODO @aignas 2024-04-05: how do we support different requirement
-            # cycles for different abis/oses? For now we will need the users to
-            # assume the same groups across all versions/platforms until we start
-            # using an alternative cycle resolution strategy.
-            hub_group_map[hub_name] = pip_attr.experimental_requirement_cycles
-
     for hub in pip_hub_map.values():
         hub_whl_map.setdefault(hub.name, {})
         for key, settings in hub.whl_map.items():
@@ -558,6 +552,8 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
+        hub_group_map[hub.name] = hub.group_map
+
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
         # args are manipulated in the code going before.
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index cf69cd5689..1195f3346f 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -47,6 +47,7 @@ def hub_builder(
         exposed_packages = {},
         extra_aliases = {},
         whl_libraries = {},
+        group_map = {},
         _evaluate_markers_fn = evaluate_markers_fn,
         _logger = logger,
         _minor_mapping = minor_mapping,
@@ -93,6 +94,13 @@ def _add(self, *, pip_attr):
     _set_index_urls(self, pip_attr)
     self._pip_attrs[python_version] = pip_attr
 
+    # TODO @aignas 2024-04-05: how do we support different requirement
+    # cycles for different abis/oses? For now we will need the users to
+    # assume the same groups across all versions/platforms until we start
+    # using an alternative cycle resolution strategy.
+    self.group_map.clear()
+    self.group_map.update(pip_attr.experimental_requirement_cycles)
+
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:

From 7f7f635930cd064c7190e2e71a0fa8bd92f8afa5 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 21:56:00 +0900
Subject: [PATCH 18/31] refactor: move extra aliases to the hub builder

---
 python/private/pypi/extension.bzl   | 9 +--------
 python/private/pypi/hub_builder.bzl | 5 +++++
 2 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index a161740cb4..fd28eba467 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -207,10 +207,6 @@ def _create_whl_repos(
 
     return struct(
         exposed_packages = exposed_packages,
-        extra_aliases = {
-            whl_name: {alias: True for alias in aliases}
-            for whl_name, aliases in pip_attr.extra_hub_aliases.items()
-        },
     )
 
 def _whl_repo(
@@ -524,10 +520,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
-            extra_aliases.setdefault(hub_name, {})
-            for whl_name, aliases in out.extra_aliases.items():
-                extra_aliases[hub_name].setdefault(whl_name, {}).update(aliases)
-
             if hub_name not in exposed_packages:
                 exposed_packages[hub_name] = out.exposed_packages
             else:
@@ -553,6 +545,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 whl_libraries[whl_name] = lib
 
         hub_group_map[hub.name] = hub.group_map
+        extra_aliases[hub.name] = hub.extra_aliases
 
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 1195f3346f..df050250d2 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -101,6 +101,11 @@ def _add(self, *, pip_attr):
     self.group_map.clear()
     self.group_map.update(pip_attr.experimental_requirement_cycles)
 
+    for whl_name, aliases in pip_attr.extra_hub_aliases.items():
+        self.extra_aliases.setdefault(whl_name, {}).update(
+            {alias: True for alias in aliases},
+        )
+
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:

From a636ca6b9573c062959c45a40d04e2c03da07cb1 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 21:59:34 +0900
Subject: [PATCH 19/31] refactor: move exposed_packages to the hub builder

---
 python/private/pypi/extension.bzl | 37 ++++++++++---------------------
 1 file changed, 12 insertions(+), 25 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index fd28eba467..705063c34b 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -82,17 +82,6 @@ def _create_whl_repos(
         hub: TODO.
         minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
             python version used to parse package METADATA files.
-
-    Returns a {type}`struct` with the following attributes:
-        whl_map: {type}`dict[str, list[struct]]` the output is keyed by the
-            normalized package name and the values are the instances of the
-            {bzl:obj}`whl_config_setting` return values.
-        exposed_packages: {type}`dict[str, Any]` this is just a way to
-            represent a set of string values.
-        whl_libraries: {type}`dict[str, dict[str, Any]]` the keys are the
-            aparent repository names for the hub repo and the values are the
-            arguments that will be passed to {bzl:obj}`whl_library` repository
-            rule.
     """
     logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
     interpreter = hub.detect_interpreter(pip_attr)
@@ -205,9 +194,16 @@ def _create_whl_repos(
                 repo = repo,
             )
 
-    return struct(
-        exposed_packages = exposed_packages,
-    )
+    if hub.exposed_packages:
+        intersection = {}
+        for pkg in exposed_packages:
+            if pkg not in hub.exposed_packages:
+                continue
+            intersection[pkg] = None
+        hub.exposed_packages.clear()
+        exposed_packages = intersection
+
+    hub.exposed_packages.update(exposed_packages)
 
 def _whl_repo(
         *,
@@ -512,7 +508,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
             builder.add(pip_attr = pip_attr)
 
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
-            out = _create_whl_repos(
+            _create_whl_repos(
                 module_ctx,
                 hub = builder,
                 pip_attr = pip_attr,
@@ -520,16 +516,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
-            if hub_name not in exposed_packages:
-                exposed_packages[hub_name] = out.exposed_packages
-            else:
-                intersection = {}
-                for pkg in out.exposed_packages:
-                    if pkg not in exposed_packages[hub_name]:
-                        continue
-                    intersection[pkg] = None
-                exposed_packages[hub_name] = intersection
-
     for hub in pip_hub_map.values():
         hub_whl_map.setdefault(hub.name, {})
         for key, settings in hub.whl_map.items():
@@ -546,6 +532,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
 
         hub_group_map[hub.name] = hub.group_map
         extra_aliases[hub.name] = hub.extra_aliases
+        exposed_packages[hub.name] = hub.exposed_packages
 
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the

From 62dccd5dbdfb14f9dde8c9772533204059e66ad4 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:06:33 +0900
Subject: [PATCH 20/31] seam: add a build function

---
 python/private/pypi/extension.bzl   | 19 ++++++++-----------
 python/private/pypi/hub_builder.bzl | 15 +++++++++++++++
 2 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 705063c34b..d30b8f215b 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -517,22 +517,19 @@ You cannot use both the additive_build_content and additive_build_content_file a
             )
 
     for hub in pip_hub_map.values():
-        hub_whl_map.setdefault(hub.name, {})
-        for key, settings in hub.whl_map.items():
-            for setting, repo in settings.items():
-                hub_whl_map[hub.name].setdefault(key, {}).setdefault(repo, []).append(setting)
-
-        whl_libraries.update(hub.whl_libraries)
-        for whl_name, lib in hub.whl_libraries.items():
-            if whl_name in lib:
+        out = hub.build()
+
+        for whl_name, lib in out.whl_libraries.items():
+            if whl_name in whl_libraries:
                 fail("'{}' already in created".format(whl_name))
             else:
                 # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
-        hub_group_map[hub.name] = hub.group_map
-        extra_aliases[hub.name] = hub.extra_aliases
-        exposed_packages[hub.name] = hub.exposed_packages
+        hub_whl_map[hub.name] = out.whl_map
+        hub_group_map[hub.name] = out.group_map
+        extra_aliases[hub.name] = out.extra_aliases
+        exposed_packages[hub.name] = out.exposed_packages
 
     return struct(
         # We sort so that the lock-file remains the same no matter the order of how the
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index df050250d2..d3e26d850e 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -68,11 +68,26 @@ def hub_builder(
             python_version in self._get_index_urls,
         ),
         evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
+        build = lambda: _build(self),
     )
 
     # buildifier: enable=uninitialized
     return self
 
+def _build(self):
+    whl_map = {}
+    for key, settings in self.whl_map.items():
+        for setting, repo in settings.items():
+            whl_map.setdefault(key, {}).setdefault(repo, []).append(setting)
+
+    return struct(
+        whl_map = whl_map,
+        group_map = self.group_map,
+        extra_aliases = self.extra_aliases,
+        exposed_packages = self.exposed_packages,
+        whl_libraries = self.whl_libraries,
+    )
+
 def _add(self, *, pip_attr):
     python_version = pip_attr.python_version
     if python_version in self.python_versions:

From 5b39b2e11f8b4d5675bfd080d162c1e71090d4ea Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:17:57 +0900
Subject: [PATCH 21/31] move the create_whl_repos to the builder

---
 python/private/pypi/extension.bzl   | 226 +---------------------------
 python/private/pypi/hub_builder.bzl | 217 ++++++++++++++++++++++++++
 2 files changed, 219 insertions(+), 224 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index d30b8f215b..839c36b4df 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -19,27 +19,16 @@ load("@pythons_hub//:interpreters.bzl", "INTERPRETER_LABELS")
 load("@pythons_hub//:versions.bzl", "MINOR_MAPPING")
 load("@rules_python_internal//:rules_python_config.bzl", rp_config = "config")
 load("//python/private:auth.bzl", "AUTH_ATTRS")
-load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:repo_utils.bzl", "repo_utils")
-load("//python/private:version.bzl", "version")
-load(":attrs.bzl", "use_isolated")
 load(":evaluate_markers.bzl", EVALUATE_MARKERS_SRCS = "SRCS")
 load(":hub_builder.bzl", "hub_builder")
 load(":hub_repository.bzl", "hub_repository", "whl_config_settings_to_json")
-load(":parse_requirements.bzl", "parse_requirements")
 load(":parse_whl_name.bzl", "parse_whl_name")
 load(":pep508_env.bzl", "env")
 load(":pip_repository_attrs.bzl", "ATTRS")
-load(":requirements_files_by_platform.bzl", "requirements_files_by_platform")
 load(":simpleapi_download.bzl", "simpleapi_download")
-load(":whl_config_setting.bzl", "whl_config_setting")
 load(":whl_library.bzl", "whl_library")
-load(":whl_repo_name.bzl", "pypi_repo_name", "whl_repo_name")
-
-def _major_minor_version(version_str):
-    ver = version.parse(version_str)
-    return "{}.{}".format(ver.release[0], ver.release[1])
 
 def _whl_mods_impl(whl_mods_dict):
     """Implementation of the pip.whl_mods tag class.
@@ -66,216 +55,6 @@ def _whl_mods_impl(whl_mods_dict):
             whl_mods = whl_mods,
         )
 
-def _create_whl_repos(
-        module_ctx,
-        *,
-        pip_attr,
-        whl_overrides,
-        hub,
-        minor_mapping = MINOR_MAPPING):
-    """create all of the whl repositories
-
-    Args:
-        module_ctx: {type}`module_ctx`.
-        pip_attr: {type}`struct` - the struct that comes from the tag class iteration.
-        whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
-        hub: TODO.
-        minor_mapping: {type}`dict[str, str]` The dictionary needed to resolve the full
-            python version used to parse package METADATA files.
-    """
-    logger = repo_utils.logger(module_ctx, "pypi:create_whl_repos")
-    interpreter = hub.detect_interpreter(pip_attr)
-
-    whl_modifications = {}
-    if pip_attr.whl_modifications != None:
-        for mod, whl_name in pip_attr.whl_modifications.items():
-            whl_modifications[normalize_name(whl_name)] = mod
-
-    platforms = hub.platforms(pip_attr.python_version)
-    requirements_by_platform = parse_requirements(
-        module_ctx,
-        requirements_by_platform = requirements_files_by_platform(
-            requirements_by_platform = pip_attr.requirements_by_platform,
-            requirements_linux = pip_attr.requirements_linux,
-            requirements_lock = pip_attr.requirements_lock,
-            requirements_osx = pip_attr.requirements_darwin,
-            requirements_windows = pip_attr.requirements_windows,
-            extra_pip_args = pip_attr.extra_pip_args,
-            platforms = sorted(platforms),  # here we only need keys
-            python_version = full_version(
-                version = pip_attr.python_version,
-                minor_mapping = minor_mapping,
-            ),
-            logger = logger,
-        ),
-        platforms = platforms,
-        extra_pip_args = pip_attr.extra_pip_args,
-        get_index_urls = hub.get_index_urls(pip_attr.python_version),
-        evaluate_markers = hub.evaluate_markers(pip_attr),
-        logger = logger,
-    )
-
-    if pip_attr.experimental_requirement_cycles:
-        requirement_cycles = {
-            name: [normalize_name(whl_name) for whl_name in whls]
-            for name, whls in pip_attr.experimental_requirement_cycles.items()
-        }
-
-        whl_group_mapping = {
-            whl_name: group_name
-            for group_name, group_whls in requirement_cycles.items()
-            for whl_name in group_whls
-        }
-    else:
-        whl_group_mapping = {}
-        requirement_cycles = {}
-    exposed_packages = {}
-    for whl in requirements_by_platform:
-        if whl.is_exposed:
-            exposed_packages[whl.name] = None
-
-        group_name = whl_group_mapping.get(whl.name)
-        group_deps = requirement_cycles.get(group_name, [])
-
-        # Construct args separately so that the lock file can be smaller and does not include unused
-        # attrs.
-        whl_library_args = dict(
-            dep_template = "@{}//{{name}}:{{target}}".format(hub.name),
-        )
-        maybe_args = dict(
-            # The following values are safe to omit if they have false like values
-            add_libdir_to_library_search_path = pip_attr.add_libdir_to_library_search_path,
-            annotation = whl_modifications.get(whl.name),
-            download_only = pip_attr.download_only,
-            enable_implicit_namespace_pkgs = pip_attr.enable_implicit_namespace_pkgs,
-            environment = pip_attr.environment,
-            envsubst = pip_attr.envsubst,
-            group_deps = group_deps,
-            group_name = group_name,
-            pip_data_exclude = pip_attr.pip_data_exclude,
-            python_interpreter = interpreter.path,
-            python_interpreter_target = interpreter.target,
-            whl_patches = {
-                p: json.encode(args)
-                for p, args in whl_overrides.get(whl.name, {}).items()
-            },
-        )
-        if not hub.config.enable_pipstar:
-            maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
-
-        whl_library_args.update({k: v for k, v in maybe_args.items() if v})
-        maybe_args_with_default = dict(
-            # The following values have defaults next to them
-            isolated = (use_isolated(module_ctx, pip_attr), True),
-            quiet = (pip_attr.quiet, True),
-            timeout = (pip_attr.timeout, 600),
-        )
-        whl_library_args.update({
-            k: v
-            for k, (v, default) in maybe_args_with_default.items()
-            if v != default
-        })
-
-        for src in whl.srcs:
-            repo = _whl_repo(
-                src = src,
-                whl_library_args = whl_library_args,
-                download_only = pip_attr.download_only,
-                netrc = hub.config.netrc or pip_attr.netrc,
-                use_downloader = hub.use_downloader(pip_attr.python_version, whl.name),
-                auth_patterns = hub.config.auth_patterns or pip_attr.auth_patterns,
-                python_version = _major_minor_version(pip_attr.python_version),
-                is_multiple_versions = whl.is_multiple_versions,
-                enable_pipstar = hub.config.enable_pipstar,
-            )
-            hub.add_whl_library(
-                python_version = pip_attr.python_version,
-                whl = whl,
-                repo = repo,
-            )
-
-    if hub.exposed_packages:
-        intersection = {}
-        for pkg in exposed_packages:
-            if pkg not in hub.exposed_packages:
-                continue
-            intersection[pkg] = None
-        hub.exposed_packages.clear()
-        exposed_packages = intersection
-
-    hub.exposed_packages.update(exposed_packages)
-
-def _whl_repo(
-        *,
-        src,
-        whl_library_args,
-        is_multiple_versions,
-        download_only,
-        netrc,
-        auth_patterns,
-        python_version,
-        use_downloader,
-        enable_pipstar = False):
-    args = dict(whl_library_args)
-    args["requirement"] = src.requirement_line
-    is_whl = src.filename.endswith(".whl")
-
-    if src.extra_pip_args and not is_whl:
-        # pip is not used to download wheels and the python
-        # `whl_library` helpers are only extracting things, however
-        # for sdists, they will be built by `pip`, so we still
-        # need to pass the extra args there, so only pop this for whls
-        args["extra_pip_args"] = src.extra_pip_args
-
-    if not src.url or (not is_whl and download_only):
-        if download_only and use_downloader:
-            # If the user did not allow using sdists and we are using the downloader
-            # and we are not using simpleapi_skip for this
-            return None
-        else:
-            # Fallback to a pip-installed wheel
-            target_platforms = src.target_platforms if is_multiple_versions else []
-            return struct(
-                repo_name = pypi_repo_name(
-                    normalize_name(src.distribution),
-                    *target_platforms
-                ),
-                args = args,
-                config_setting = whl_config_setting(
-                    version = python_version,
-                    target_platforms = target_platforms or None,
-                ),
-            )
-
-    # This is no-op because pip is not used to download the wheel.
-    args.pop("download_only", None)
-
-    if netrc:
-        args["netrc"] = netrc
-    if auth_patterns:
-        args["auth_patterns"] = auth_patterns
-
-    args["urls"] = [src.url]
-    args["sha256"] = src.sha256
-    args["filename"] = src.filename
-    if not enable_pipstar:
-        args["experimental_target_platforms"] = [
-            # Get rid of the version for the target platforms because we are
-            # passing the interpreter any way. Ideally we should search of ways
-            # how to pass the target platforms through the hub repo.
-            p.partition("_")[2]
-            for p in src.target_platforms
-        ]
-
-    return struct(
-        repo_name = whl_repo_name(src.filename, src.sha256),
-        args = args,
-        config_setting = whl_config_setting(
-            version = python_version,
-            target_platforms = src.target_platforms,
-        ),
-    )
-
 def _plat(*, name, arch_name, os_name, config_settings = [], env = {}, marker = "", whl_abi_tags = [], whl_platform_tags = []):
     # NOTE @aignas 2025-07-08: the least preferred is the first item in the list
     if "any" not in whl_platform_tags:
@@ -482,6 +261,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     config = config,
                     simpleapi_download_fn = simpleapi_download,
                     simpleapi_cache = simpleapi_cache,
+                    # TODO @aignas 2025-09-06: do not use kwargs
                     minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
                     evaluate_markers_fn = kwargs.get("evaluate_markers", None),
                     available_interpreters = kwargs.get("available_interpreters", INTERPRETER_LABELS),
@@ -508,12 +288,10 @@ You cannot use both the additive_build_content and additive_build_content_file a
             builder.add(pip_attr = pip_attr)
 
             # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
-            _create_whl_repos(
+            builder.create_whl_repos(
                 module_ctx,
-                hub = builder,
                 pip_attr = pip_attr,
                 whl_overrides = whl_overrides,
-                minor_mapping = kwargs.get("minor_mapping", MINOR_MAPPING),
             )
 
     for hub in pip_hub_map.values():
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index d3e26d850e..4865501f9f 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -4,10 +4,19 @@ load("//python/private:full_version.bzl", "full_version")
 load("//python/private:normalize_name.bzl", "normalize_name")
 load("//python/private:version.bzl", "version")
 load("//python/private:version_label.bzl", "version_label")
+load(":attrs.bzl", "use_isolated")
 load(":evaluate_markers.bzl", "evaluate_markers_py", evaluate_markers_star = "evaluate_markers")
+load(":parse_requirements.bzl", "parse_requirements")
 load(":pep508_env.bzl", "env")
 load(":pep508_evaluate.bzl", "evaluate")
 load(":python_tag.bzl", "python_tag")
+load(":requirements_files_by_platform.bzl", "requirements_files_by_platform")
+load(":whl_config_setting.bzl", "whl_config_setting")
+load(":whl_repo_name.bzl", "pypi_repo_name", "whl_repo_name")
+
+def _major_minor_version(version_str):
+    ver = version.parse(version_str)
+    return "{}.{}".format(ver.release[0], ver.release[1])
 
 def hub_builder(
         *,
@@ -69,6 +78,7 @@ def hub_builder(
         ),
         evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
         build = lambda: _build(self),
+        create_whl_repos = lambda *a, **k: _create_whl_repos(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
@@ -316,3 +326,210 @@ def _evaluate_markers(self, pip_attr):
         srcs = pip_attr._evaluate_markers_srcs,
         logger = self._logger,
     )
+
+def _create_whl_repos(
+        self,
+        module_ctx,
+        *,
+        pip_attr,
+        whl_overrides):
+    """create all of the whl repositories
+
+    Args:
+        self: the builder.
+        module_ctx: {type}`module_ctx`.
+        pip_attr: {type}`struct` - the struct that comes from the tag class iteration.
+        whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
+    """
+    logger = self._logger
+    platforms = self.platforms(pip_attr.python_version)
+    requirements_by_platform = parse_requirements(
+        module_ctx,
+        requirements_by_platform = requirements_files_by_platform(
+            requirements_by_platform = pip_attr.requirements_by_platform,
+            requirements_linux = pip_attr.requirements_linux,
+            requirements_lock = pip_attr.requirements_lock,
+            requirements_osx = pip_attr.requirements_darwin,
+            requirements_windows = pip_attr.requirements_windows,
+            extra_pip_args = pip_attr.extra_pip_args,
+            platforms = sorted(platforms),  # here we only need keys
+            python_version = full_version(
+                version = pip_attr.python_version,
+                minor_mapping = self._minor_mapping,
+            ),
+            logger = logger,
+        ),
+        platforms = platforms,
+        extra_pip_args = pip_attr.extra_pip_args,
+        get_index_urls = self.get_index_urls(pip_attr.python_version),
+        evaluate_markers = self.evaluate_markers(pip_attr),
+        logger = logger,
+    )
+
+    whl_modifications = {}
+    if pip_attr.whl_modifications != None:
+        for mod, whl_name in pip_attr.whl_modifications.items():
+            whl_modifications[normalize_name(whl_name)] = mod
+
+    if pip_attr.experimental_requirement_cycles:
+        requirement_cycles = {
+            name: [normalize_name(whl_name) for whl_name in whls]
+            for name, whls in pip_attr.experimental_requirement_cycles.items()
+        }
+
+        whl_group_mapping = {
+            whl_name: group_name
+            for group_name, group_whls in requirement_cycles.items()
+            for whl_name in group_whls
+        }
+    else:
+        whl_group_mapping = {}
+        requirement_cycles = {}
+
+    interpreter = self.detect_interpreter(pip_attr)
+    exposed_packages = {}
+    for whl in requirements_by_platform:
+        if whl.is_exposed:
+            exposed_packages[whl.name] = None
+
+        group_name = whl_group_mapping.get(whl.name)
+        group_deps = requirement_cycles.get(group_name, [])
+
+        # Construct args separately so that the lock file can be smaller and does not include unused
+        # attrs.
+        whl_library_args = dict(
+            dep_template = "@{}//{{name}}:{{target}}".format(self.name),
+        )
+        maybe_args = dict(
+            # The following values are safe to omit if they have false like values
+            add_libdir_to_library_search_path = pip_attr.add_libdir_to_library_search_path,
+            annotation = whl_modifications.get(whl.name),
+            download_only = pip_attr.download_only,
+            enable_implicit_namespace_pkgs = pip_attr.enable_implicit_namespace_pkgs,
+            environment = pip_attr.environment,
+            envsubst = pip_attr.envsubst,
+            group_deps = group_deps,
+            group_name = group_name,
+            pip_data_exclude = pip_attr.pip_data_exclude,
+            python_interpreter = interpreter.path,
+            python_interpreter_target = interpreter.target,
+            whl_patches = {
+                p: json.encode(args)
+                for p, args in whl_overrides.get(whl.name, {}).items()
+            },
+        )
+        if not self.config.enable_pipstar:
+            maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
+
+        whl_library_args.update({k: v for k, v in maybe_args.items() if v})
+        maybe_args_with_default = dict(
+            # The following values have defaults next to them
+            isolated = (use_isolated(module_ctx, pip_attr), True),
+            quiet = (pip_attr.quiet, True),
+            timeout = (pip_attr.timeout, 600),
+        )
+        whl_library_args.update({
+            k: v
+            for k, (v, default) in maybe_args_with_default.items()
+            if v != default
+        })
+
+        for src in whl.srcs:
+            repo = _whl_repo(
+                src = src,
+                whl_library_args = whl_library_args,
+                download_only = pip_attr.download_only,
+                netrc = self.config.netrc or pip_attr.netrc,
+                use_downloader = self.use_downloader(pip_attr.python_version, whl.name),
+                auth_patterns = self.config.auth_patterns or pip_attr.auth_patterns,
+                python_version = _major_minor_version(pip_attr.python_version),
+                is_multiple_versions = whl.is_multiple_versions,
+                enable_pipstar = self.config.enable_pipstar,
+            )
+            self.add_whl_library(
+                python_version = pip_attr.python_version,
+                whl = whl,
+                repo = repo,
+            )
+
+    if self.exposed_packages:
+        intersection = {}
+        for pkg in exposed_packages:
+            if pkg not in self.exposed_packages:
+                continue
+            intersection[pkg] = None
+        self.exposed_packages.clear()
+        exposed_packages = intersection
+
+    self.exposed_packages.update(exposed_packages)
+
+def _whl_repo(
+        *,
+        src,
+        whl_library_args,
+        is_multiple_versions,
+        download_only,
+        netrc,
+        auth_patterns,
+        python_version,
+        use_downloader,
+        enable_pipstar = False):
+    args = dict(whl_library_args)
+    args["requirement"] = src.requirement_line
+    is_whl = src.filename.endswith(".whl")
+
+    if src.extra_pip_args and not is_whl:
+        # pip is not used to download wheels and the python
+        # `whl_library` helpers are only extracting things, however
+        # for sdists, they will be built by `pip`, so we still
+        # need to pass the extra args there, so only pop this for whls
+        args["extra_pip_args"] = src.extra_pip_args
+
+    if not src.url or (not is_whl and download_only):
+        if download_only and use_downloader:
+            # If the user did not allow using sdists and we are using the downloader
+            # and we are not using simpleapi_skip for this
+            return None
+        else:
+            # Fallback to a pip-installed wheel
+            target_platforms = src.target_platforms if is_multiple_versions else []
+            return struct(
+                repo_name = pypi_repo_name(
+                    normalize_name(src.distribution),
+                    *target_platforms
+                ),
+                args = args,
+                config_setting = whl_config_setting(
+                    version = python_version,
+                    target_platforms = target_platforms or None,
+                ),
+            )
+
+    # This is no-op because pip is not used to download the wheel.
+    args.pop("download_only", None)
+
+    if netrc:
+        args["netrc"] = netrc
+    if auth_patterns:
+        args["auth_patterns"] = auth_patterns
+
+    args["urls"] = [src.url]
+    args["sha256"] = src.sha256
+    args["filename"] = src.filename
+    if not enable_pipstar:
+        args["experimental_target_platforms"] = [
+            # Get rid of the version for the target platforms because we are
+            # passing the interpreter any way. Ideally we should search of ways
+            # how to pass the target platforms through the hub repo.
+            p.partition("_")[2]
+            for p in src.target_platforms
+        ]
+
+    return struct(
+        repo_name = whl_repo_name(src.filename, src.sha256),
+        args = args,
+        config_setting = whl_config_setting(
+            version = python_version,
+            target_platforms = src.target_platforms,
+        ),
+    )

From 872d0c472e74459b5a743c79f1a8fd4bd20640d5 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:20:35 +0900
Subject: [PATCH 22/31] create a pip_parse method in the builder

---
 python/private/pypi/extension.bzl   |  5 +----
 python/private/pypi/hub_builder.bzl | 10 ++++++++++
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 839c36b4df..25dba86137 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -285,10 +285,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
             else:
                 builder = pip_hub_map[pip_attr.hub_name]
 
-            builder.add(pip_attr = pip_attr)
-
-            # TODO @aignas 2025-05-19: express pip.parse as a series of configure calls
-            builder.create_whl_repos(
+            builder.pip_parse(
                 module_ctx,
                 pip_attr = pip_attr,
                 whl_overrides = whl_overrides,
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 4865501f9f..1f1dfea9a2 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -79,6 +79,7 @@ def hub_builder(
         evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
         build = lambda: _build(self),
         create_whl_repos = lambda *a, **k: _create_whl_repos(self, *a, **k),
+        pip_parse = lambda *a, **k: _pip_parse(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
@@ -533,3 +534,12 @@ def _whl_repo(
             target_platforms = src.target_platforms,
         ),
     )
+
+def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
+    self.add(pip_attr = pip_attr)
+
+    self.create_whl_repos(
+        module_ctx,
+        pip_attr = pip_attr,
+        whl_overrides = whl_overrides,
+    )

From 2bd0b42cf7da2d02015ba4c57a985d667eaa4611 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:33:13 +0900
Subject: [PATCH 23/31] simplify the code base a little

---
 python/private/pypi/extension.bzl   | 66 +++++++++--------------------
 python/private/pypi/hub_builder.bzl |  7 ++-
 2 files changed, 24 insertions(+), 49 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 25dba86137..66dfc21062 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -168,7 +168,7 @@ def parse_modules(
         enable_pipstar: {type}`bool` a flag to enable dropping Python dependency for
             evaluation of the extension.
         _fail: {type}`function` the failure function, mainly for testing.
-        **kwargs: Extra arguments passed to the layers below.
+        **kwargs: Extra arguments passed to the hub_builder.
 
     Returns:
         A struct with the following attributes:
@@ -242,15 +242,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
     pip_hub_map = {}
     simpleapi_cache = {}
 
-    # Keeps track of all the hub's whl repos across the different versions.
-    # dict[hub, dict[whl, dict[version, str pip]]]
-    # Where hub, whl, and pip are the repo names
-    hub_whl_map = {}
-    hub_group_map = {}
-    exposed_packages = {}
-    extra_aliases = {}
-    whl_libraries = {}
-
     for mod in module_ctx.modules:
         for pip_attr in mod.tags.parse:
             hub_name = pip_attr.hub_name
@@ -291,6 +282,14 @@ You cannot use both the additive_build_content and additive_build_content_file a
                 whl_overrides = whl_overrides,
             )
 
+    # Keeps track of all the hub's whl repos across the different versions.
+    # dict[hub, dict[whl, dict[version, str pip]]]
+    # Where hub, whl, and pip are the repo names
+    hub_whl_map = {}
+    hub_group_map = {}
+    exposed_packages = {}
+    extra_aliases = {}
+    whl_libraries = {}
     for hub in pip_hub_map.values():
         out = hub.build()
 
@@ -298,43 +297,21 @@ You cannot use both the additive_build_content and additive_build_content_file a
             if whl_name in whl_libraries:
                 fail("'{}' already in created".format(whl_name))
             else:
-                # replicate whl_libraries.update(out.whl_libraries)
                 whl_libraries[whl_name] = lib
 
-        hub_whl_map[hub.name] = out.whl_map
-        hub_group_map[hub.name] = out.group_map
-        extra_aliases[hub.name] = out.extra_aliases
         exposed_packages[hub.name] = out.exposed_packages
+        extra_aliases[hub.name] = out.extra_aliases
+        hub_group_map[hub.name] = out.group_map
+        hub_whl_map[hub.name] = out.whl_map
 
     return struct(
-        # We sort so that the lock-file remains the same no matter the order of how the
-        # args are manipulated in the code going before.
-        whl_mods = dict(sorted(whl_mods.items())),
-        hub_whl_map = {
-            hub_name: {
-                whl_name: dict(settings)
-                for whl_name, settings in sorted(whl_map.items())
-            }
-            for hub_name, whl_map in sorted(hub_whl_map.items())
-        },
-        hub_group_map = {
-            hub_name: {
-                key: sorted(values)
-                for key, values in sorted(group_map.items())
-            }
-            for hub_name, group_map in sorted(hub_group_map.items())
-        },
-        exposed_packages = {
-            k: sorted(v)
-            for k, v in sorted(exposed_packages.items())
-        },
-        extra_aliases = {
-            hub_name: {
-                whl_name: sorted(aliases)
-                for whl_name, aliases in extra_whl_aliases.items()
-            }
-            for hub_name, extra_whl_aliases in extra_aliases.items()
-        },
+        config = config,
+        exposed_packages = exposed_packages,
+        extra_aliases = extra_aliases,
+        hub_group_map = hub_group_map,
+        hub_whl_map = hub_whl_map,
+        whl_libraries = whl_libraries,
+        whl_mods = whl_mods,
         platform_config_settings = {
             hub_name: {
                 platform_name: sorted([str(Label(cv)) for cv in p.config_settings])
@@ -342,11 +319,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
             }
             for hub_name in hub_whl_map
         },
-        whl_libraries = {
-            k: dict(sorted(args.items()))
-            for k, args in sorted(whl_libraries.items())
-        },
-        config = config,
     )
 
 def _pip_impl(module_ctx):
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 1f1dfea9a2..ae753fd90f 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -94,8 +94,11 @@ def _build(self):
     return struct(
         whl_map = whl_map,
         group_map = self.group_map,
-        extra_aliases = self.extra_aliases,
-        exposed_packages = self.exposed_packages,
+        extra_aliases = {
+            whl: sorted(aliases)
+            for whl, aliases in self.extra_aliases.items()
+        },
+        exposed_packages = sorted(self.exposed_packages),
         whl_libraries = self.whl_libraries,
     )
 

From 462868869c51483fb98b79024085173a6839fb7d Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:44:47 +0900
Subject: [PATCH 24/31] eliminate some unnecessary methods

---
 python/private/pypi/hub_builder.bzl | 94 +++++++++++++----------------
 1 file changed, 42 insertions(+), 52 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index ae753fd90f..6fa4c2c59d 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -50,7 +50,7 @@ def hub_builder(
     self = struct(
         name = name,
         module_name = module_name,
-        python_versions = {},
+        _platforms = {},
         config = config,
         whl_map = {},
         exposed_packages = {},
@@ -64,21 +64,14 @@ def hub_builder(
         _simpleapi_download_fn = simpleapi_download_fn,
         _simpleapi_cache = simpleapi_cache,
         _get_index_urls = {},
-        _pip_attrs = {},
         _use_downloader = {},
         # keep sorted
-        add = lambda *a, **k: _add(self, *a, **k),
-        detect_interpreter = lambda *a, **k: _detect_interpreter(self, *a, **k),
         get_index_urls = lambda version: self._get_index_urls.get(version),
-        platforms = lambda version: self.python_versions[version],
-        add_whl_library = lambda *a, **k: _add_whl_library(self, *a, **k),
         use_downloader = lambda python_version, whl_name: self._use_downloader.get(python_version, {}).get(
             normalize_name(whl_name),
-            python_version in self._get_index_urls,
+            self.get_index_urls(python_version) != None,
         ),
-        evaluate_markers = lambda *a, **k: _evaluate_markers(self, *a, **k),
         build = lambda: _build(self),
-        create_whl_repos = lambda *a, **k: _create_whl_repos(self, *a, **k),
         pip_parse = lambda *a, **k: _pip_parse(self, *a, **k),
     )
 
@@ -102,39 +95,6 @@ def _build(self):
         whl_libraries = self.whl_libraries,
     )
 
-def _add(self, *, pip_attr):
-    python_version = pip_attr.python_version
-    if python_version in self.python_versions:
-        fail((
-            "Duplicate pip python version '{version}' for hub " +
-            "'{hub}' in module '{module}': the Python versions " +
-            "used for a hub must be unique"
-        ).format(
-            hub = self.name,
-            module = self.module_name,
-            version = python_version,
-        ))
-
-    self.python_versions[python_version] = _platforms(
-        python_version = python_version,
-        minor_mapping = self._minor_mapping,
-        config = self.config,
-    )
-    _set_index_urls(self, pip_attr)
-    self._pip_attrs[python_version] = pip_attr
-
-    # TODO @aignas 2024-04-05: how do we support different requirement
-    # cycles for different abis/oses? For now we will need the users to
-    # assume the same groups across all versions/platforms until we start
-    # using an alternative cycle resolution strategy.
-    self.group_map.clear()
-    self.group_map.update(pip_attr.experimental_requirement_cycles)
-
-    for whl_name, aliases in pip_attr.extra_hub_aliases.items():
-        self.extra_aliases.setdefault(whl_name, {}).update(
-            {alias: True for alias in aliases},
-        )
-
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:
@@ -254,7 +214,7 @@ def _add_whl_library(self, *, python_version, whl, repo):
         # disallow building from sdist.
         return
 
-    platforms = self.python_versions[python_version]
+    platforms = self._platforms[python_version]
 
     # TODO @aignas 2025-06-29: we should not need the version in the repo_name if
     # we are using pipstar and we are downloading the wheel using the downloader
@@ -297,10 +257,10 @@ def _evaluate_markers(self, pip_attr):
     if self.config.enable_pipstar:
         return lambda _, requirements: evaluate_markers_star(
             requirements = requirements,
-            platforms = self.python_versions[pip_attr.python_version],
+            platforms = self._platforms[pip_attr.python_version],
         )
 
-    interpreter = self.detect_interpreter(pip_attr)
+    interpreter = _detect_interpreter(self, pip_attr)
 
     # NOTE @aignas 2024-08-02: , we will execute any interpreter that we find either
     # in the PATH or if specified as a label. We will configure the env
@@ -320,7 +280,7 @@ def _evaluate_markers(self, pip_attr):
         module_ctx,
         requirements = {
             k: {
-                p: self.python_versions[pip_attr.python_version][p].triple
+                p: self._platforms[pip_attr.python_version][p].triple
                 for p in plats
             }
             for k, plats in requirements.items()
@@ -346,7 +306,7 @@ def _create_whl_repos(
         whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
     """
     logger = self._logger
-    platforms = self.platforms(pip_attr.python_version)
+    platforms = self._platforms[pip_attr.python_version]
     requirements_by_platform = parse_requirements(
         module_ctx,
         requirements_by_platform = requirements_files_by_platform(
@@ -366,7 +326,7 @@ def _create_whl_repos(
         platforms = platforms,
         extra_pip_args = pip_attr.extra_pip_args,
         get_index_urls = self.get_index_urls(pip_attr.python_version),
-        evaluate_markers = self.evaluate_markers(pip_attr),
+        evaluate_markers = _evaluate_markers(self, pip_attr),
         logger = logger,
     )
 
@@ -390,7 +350,7 @@ def _create_whl_repos(
         whl_group_mapping = {}
         requirement_cycles = {}
 
-    interpreter = self.detect_interpreter(pip_attr)
+    interpreter = _detect_interpreter(self, pip_attr)
     exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
@@ -450,7 +410,8 @@ def _create_whl_repos(
                 is_multiple_versions = whl.is_multiple_versions,
                 enable_pipstar = self.config.enable_pipstar,
             )
-            self.add_whl_library(
+            _add_whl_library(
+                self,
                 python_version = pip_attr.python_version,
                 whl = whl,
                 repo = repo,
@@ -539,9 +500,38 @@ def _whl_repo(
     )
 
 def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
-    self.add(pip_attr = pip_attr)
+    python_version = pip_attr.python_version
+    if python_version in self._platforms:
+        fail((
+            "Duplicate pip python version '{version}' for hub " +
+            "'{hub}' in module '{module}': the Python versions " +
+            "used for a hub must be unique"
+        ).format(
+            hub = self.name,
+            module = self.module_name,
+            version = python_version,
+        ))
+
+    self._platforms[python_version] = _platforms(
+        python_version = python_version,
+        minor_mapping = self._minor_mapping,
+        config = self.config,
+    )
+    _set_index_urls(self, pip_attr)
+
+    # TODO @aignas 2024-04-05: how do we support different requirement
+    # cycles for different abis/oses? For now we will need the users to
+    # assume the same groups across all versions/platforms until we start
+    # using an alternative cycle resolution strategy.
+    self.group_map.clear()
+    self.group_map.update(pip_attr.experimental_requirement_cycles)
 
-    self.create_whl_repos(
+    for whl_name, aliases in pip_attr.extra_hub_aliases.items():
+        self.extra_aliases.setdefault(whl_name, {}).update(
+            {alias: True for alias in aliases},
+        )
+    _create_whl_repos(
+        self,
         module_ctx,
         pip_attr = pip_attr,
         whl_overrides = whl_overrides,

From c283aa3c59979030729318a851435aa471fea3cd Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:56:59 +0900
Subject: [PATCH 25/31] documnet methods on the builder

---
 python/private/pypi/hub_builder.bzl | 117 +++++++++++++++-------------
 1 file changed, 65 insertions(+), 52 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 6fa4c2c59d..d489ecaf0a 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -50,49 +50,55 @@ def hub_builder(
     self = struct(
         name = name,
         module_name = module_name,
+
+        # public methods, keep sorted and to minimum
+        build = lambda: _build(self),
+        pip_parse = lambda *a, **k: _pip_parse(self, *a, **k),
+
+        # build output
+        _exposed_packages = {},  # modified by _add_exposed_packages
+        _extra_aliases = {},  # modified by _add_extra_aliases
+        _group_map = {},  # modified by _add_group_map
+        _whl_libraries = {},  # modified by _add_whl_library
+        _whl_map = {},  # modified by _add_whl_library
+        # internal
         _platforms = {},
-        config = config,
-        whl_map = {},
-        exposed_packages = {},
-        extra_aliases = {},
-        whl_libraries = {},
-        group_map = {},
+        _get_index_urls = {},
+        _use_downloader = {},
+        _simpleapi_cache = simpleapi_cache,
+        # instance constants
+        _config = config,
         _evaluate_markers_fn = evaluate_markers_fn,
         _logger = logger,
         _minor_mapping = minor_mapping,
         _available_interpreters = available_interpreters,
         _simpleapi_download_fn = simpleapi_download_fn,
-        _simpleapi_cache = simpleapi_cache,
-        _get_index_urls = {},
-        _use_downloader = {},
-        # keep sorted
-        get_index_urls = lambda version: self._get_index_urls.get(version),
-        use_downloader = lambda python_version, whl_name: self._use_downloader.get(python_version, {}).get(
-            normalize_name(whl_name),
-            self.get_index_urls(python_version) != None,
-        ),
-        build = lambda: _build(self),
-        pip_parse = lambda *a, **k: _pip_parse(self, *a, **k),
     )
 
     # buildifier: enable=uninitialized
     return self
 
+def _use_downloader(self, python_version, whl_name):
+    return self._use_downloader.get(python_version, {}).get(
+        normalize_name(whl_name),
+        self._get_index_urls.get(python_version) != None,
+    )
+
 def _build(self):
     whl_map = {}
-    for key, settings in self.whl_map.items():
+    for key, settings in self._whl_map.items():
         for setting, repo in settings.items():
             whl_map.setdefault(key, {}).setdefault(repo, []).append(setting)
 
     return struct(
         whl_map = whl_map,
-        group_map = self.group_map,
+        group_map = self._group_map,
         extra_aliases = {
             whl: sorted(aliases)
-            for whl, aliases in self.extra_aliases.items()
+            for whl, aliases in self._extra_aliases.items()
         },
-        exposed_packages = sorted(self.exposed_packages),
-        whl_libraries = self.whl_libraries,
+        exposed_packages = sorted(self._exposed_packages),
+        whl_libraries = self._whl_libraries,
     )
 
 def _set_index_urls(self, pip_attr):
@@ -126,7 +132,7 @@ def _set_index_urls(self, pip_attr):
             sources = [
                 d
                 for d in distributions
-                if self.use_downloader(python_version, d)
+                if _use_downloader(self, python_version, d)
             ],
             envsubst = pip_attr.envsubst,
             # Auth related info
@@ -220,15 +226,15 @@ def _add_whl_library(self, *, python_version, whl, repo):
     # we are using pipstar and we are downloading the wheel using the downloader
     repo_name = "{}_{}_{}".format(self.name, version_label(python_version), repo.repo_name)
 
-    if repo_name in self.whl_libraries:
+    if repo_name in self._whl_libraries:
         fail("attempting to create a duplicate library {} for {}".format(
             repo_name,
             whl.name,
         ))
-    self.whl_libraries[repo_name] = repo.args
+    self._whl_libraries[repo_name] = repo.args
 
-    if not self.config.enable_pipstar and "experimental_target_platforms" in repo.args:
-        self.whl_libraries[repo_name] |= {
+    if not self._config.enable_pipstar and "experimental_target_platforms" in repo.args:
+        self._whl_libraries[repo_name] |= {
             "experimental_target_platforms": sorted({
                 # TODO @aignas 2025-07-07: this should be solved in a better way
                 platforms[candidate].triple.partition("_")[-1]: None
@@ -238,7 +244,7 @@ def _add_whl_library(self, *, python_version, whl, repo):
             }),
         }
 
-    mapping = self.whl_map.setdefault(whl.name, {})
+    mapping = self._whl_map.setdefault(whl.name, {})
     if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
         fail(
             "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
@@ -254,7 +260,7 @@ def _evaluate_markers(self, pip_attr):
     if self._evaluate_markers_fn:
         return self._evaluate_markers_fn
 
-    if self.config.enable_pipstar:
+    if self._config.enable_pipstar:
         return lambda _, requirements: evaluate_markers_star(
             requirements = requirements,
             platforms = self._platforms[pip_attr.python_version],
@@ -325,7 +331,7 @@ def _create_whl_repos(
         ),
         platforms = platforms,
         extra_pip_args = pip_attr.extra_pip_args,
-        get_index_urls = self.get_index_urls(pip_attr.python_version),
+        get_index_urls = self._get_index_urls.get(pip_attr.python_version),
         evaluate_markers = _evaluate_markers(self, pip_attr),
         logger = logger,
     )
@@ -382,7 +388,7 @@ def _create_whl_repos(
                 for p, args in whl_overrides.get(whl.name, {}).items()
             },
         )
-        if not self.config.enable_pipstar:
+        if not self._config.enable_pipstar:
             maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
 
         whl_library_args.update({k: v for k, v in maybe_args.items() if v})
@@ -403,12 +409,12 @@ def _create_whl_repos(
                 src = src,
                 whl_library_args = whl_library_args,
                 download_only = pip_attr.download_only,
-                netrc = self.config.netrc or pip_attr.netrc,
-                use_downloader = self.use_downloader(pip_attr.python_version, whl.name),
-                auth_patterns = self.config.auth_patterns or pip_attr.auth_patterns,
+                netrc = self._config.netrc or pip_attr.netrc,
+                use_downloader = _use_downloader(self, pip_attr.python_version, whl.name),
+                auth_patterns = self._config.auth_patterns or pip_attr.auth_patterns,
                 python_version = _major_minor_version(pip_attr.python_version),
                 is_multiple_versions = whl.is_multiple_versions,
-                enable_pipstar = self.config.enable_pipstar,
+                enable_pipstar = self._config.enable_pipstar,
             )
             _add_whl_library(
                 self,
@@ -417,16 +423,19 @@ def _create_whl_repos(
                 repo = repo,
             )
 
-    if self.exposed_packages:
+    _add_exposed_packages(self, exposed_packages)
+
+def _add_exposed_packages(self, exposed_packages):
+    if self._exposed_packages:
         intersection = {}
         for pkg in exposed_packages:
-            if pkg not in self.exposed_packages:
+            if pkg not in self._exposed_packages:
                 continue
             intersection[pkg] = None
-        self.exposed_packages.clear()
+        self._exposed_packages.clear()
         exposed_packages = intersection
 
-    self.exposed_packages.update(exposed_packages)
+    self._exposed_packages.update(exposed_packages)
 
 def _whl_repo(
         *,
@@ -499,6 +508,20 @@ def _whl_repo(
         ),
     )
 
+def _add_group_map(self, group_map):
+    # TODO @aignas 2024-04-05: how do we support different requirement
+    # cycles for different abis/oses? For now we will need the users to
+    # assume the same groups across all versions/platforms until we start
+    # using an alternative cycle resolution strategy.
+    self._group_map.clear()
+    self._group_map.update(group_map)
+
+def _add_extra_aliases(self, extra_hub_aliases):
+    for whl_name, aliases in extra_hub_aliases.items():
+        self._extra_aliases.setdefault(whl_name, {}).update(
+            {alias: True for alias in aliases},
+        )
+
 def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
     python_version = pip_attr.python_version
     if python_version in self._platforms:
@@ -515,21 +538,11 @@ def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
     self._platforms[python_version] = _platforms(
         python_version = python_version,
         minor_mapping = self._minor_mapping,
-        config = self.config,
+        config = self._config,
     )
     _set_index_urls(self, pip_attr)
-
-    # TODO @aignas 2024-04-05: how do we support different requirement
-    # cycles for different abis/oses? For now we will need the users to
-    # assume the same groups across all versions/platforms until we start
-    # using an alternative cycle resolution strategy.
-    self.group_map.clear()
-    self.group_map.update(pip_attr.experimental_requirement_cycles)
-
-    for whl_name, aliases in pip_attr.extra_hub_aliases.items():
-        self.extra_aliases.setdefault(whl_name, {}).update(
-            {alias: True for alias in aliases},
-        )
+    _add_group_map(self, pip_attr.experimental_requirement_cycles)
+    _add_extra_aliases(self, pip_attr.extra_hub_aliases)
     _create_whl_repos(
         self,
         module_ctx,

From 5764570db577704e5baf8da63cd413ac24b1c7c9 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 22:58:56 +0900
Subject: [PATCH 26/31] documnet args on the builder

---
 python/private/pypi/hub_builder.bzl | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index d489ecaf0a..43e7bce3e0 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -32,18 +32,18 @@ def hub_builder(
     """Return a hub builder instance
 
     Args:
-        name: TODO
-        module_name: TODO
+        name: {type}`str`, the name of the hub.
+        module_name: {type}`str`, the module name that has created the hub.
         config: The platform configuration.
-        minor_mapping: TODO
-        evaluate_markers_fn: the function used to evaluate the markers.
+        minor_mapping: {type}`dict[str, str]` the mapping between minor and full versions.
+        evaluate_markers_fn: the override function used to evaluate the markers.
         available_interpreters: {type}`dict[str, Label]` The dictionary of available
             interpreters that have been registered using the `python` bzlmod extension.
             The keys are in the form `python_{snake_case_version}_host`. This is to be
             used during the `repository_rule` and must be always compatible with the host.
-        simpleapi_download_fn: TODO
-        simpleapi_cache: TODO
-        logger: TODO
+        simpleapi_download_fn: the function used to download from SimpleAPI.
+        simpleapi_cache: the cache for the download results.
+        logger: the logger for this builder.
     """
 
     # buildifier: disable=uninitialized

From 46d65b9ba6a497a4e53909409a83a1c4e1a26e20 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sat, 6 Sep 2025 23:05:35 +0900
Subject: [PATCH 27/31] fix bzl_library deps

---
 python/private/pypi/BUILD.bazel     |  15 +-
 python/private/pypi/hub_builder.bzl | 207 ++++++++++++++--------------
 2 files changed, 115 insertions(+), 107 deletions(-)

diff --git a/python/private/pypi/BUILD.bazel b/python/private/pypi/BUILD.bazel
index ea41a3cf24..fd850857e9 100644
--- a/python/private/pypi/BUILD.bazel
+++ b/python/private/pypi/BUILD.bazel
@@ -109,22 +109,17 @@ bzl_library(
     name = "extension_bzl",
     srcs = ["extension.bzl"],
     deps = [
-        ":attrs_bzl",
         ":evaluate_markers_bzl",
         ":hub_builder_bzl",
         ":hub_repository_bzl",
-        ":parse_requirements_bzl",
         ":parse_whl_name_bzl",
         ":pep508_env_bzl",
         ":pip_repository_attrs_bzl",
         ":simpleapi_download_bzl",
-        ":whl_config_setting_bzl",
         ":whl_library_bzl",
-        ":whl_repo_name_bzl",
-        "//python/private:full_version_bzl",
+        "//python/private:auth_bzl",
         "//python/private:normalize_name_bzl",
-        "//python/private:version_bzl",
-        "//python/private:version_label_bzl",
+        "//python/private:repo_utils_bzl",
         "@bazel_features//:features",
         "@pythons_hub//:interpreters_bzl",
         "@pythons_hub//:versions_bzl",
@@ -172,13 +167,19 @@ bzl_library(
     srcs = ["hub_builder.bzl"],
     visibility = ["//:__subpackages__"],
     deps = [
+        ":attrs_bzl",
         ":evaluate_markers_bzl",
+        ":parse_requirements_bzl",
         ":pep508_env_bzl",
         ":pep508_evaluate_bzl",
         ":python_tag_bzl",
+        ":requirements_files_by_platform_bzl",
+        ":whl_config_setting_bzl",
+        ":whl_repo_name_bzl",
         "//python/private:full_version_bzl",
         "//python/private:normalize_name_bzl",
         "//python/private:version_bzl",
+        "//python/private:version_label_bzl",
     ],
 )
 
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 43e7bce3e0..32eeba08b2 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -78,11 +78,7 @@ def hub_builder(
     # buildifier: enable=uninitialized
     return self
 
-def _use_downloader(self, python_version, whl_name):
-    return self._use_downloader.get(python_version, {}).get(
-        normalize_name(whl_name),
-        self._get_index_urls.get(python_version) != None,
-    )
+### PUBLIC methods
 
 def _build(self):
     whl_map = {}
@@ -101,6 +97,108 @@ def _build(self):
         whl_libraries = self._whl_libraries,
     )
 
+def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
+    python_version = pip_attr.python_version
+    if python_version in self._platforms:
+        fail((
+            "Duplicate pip python version '{version}' for hub " +
+            "'{hub}' in module '{module}': the Python versions " +
+            "used for a hub must be unique"
+        ).format(
+            hub = self.name,
+            module = self.module_name,
+            version = python_version,
+        ))
+
+    self._platforms[python_version] = _platforms(
+        python_version = python_version,
+        minor_mapping = self._minor_mapping,
+        config = self._config,
+    )
+    _set_index_urls(self, pip_attr)
+    _add_group_map(self, pip_attr.experimental_requirement_cycles)
+    _add_extra_aliases(self, pip_attr.extra_hub_aliases)
+    _create_whl_repos(
+        self,
+        module_ctx,
+        pip_attr = pip_attr,
+        whl_overrides = whl_overrides,
+    )
+
+### end of PUBLIC methods
+### setters for build outputs
+
+def _add_exposed_packages(self, exposed_packages):
+    if self._exposed_packages:
+        intersection = {}
+        for pkg in exposed_packages:
+            if pkg not in self._exposed_packages:
+                continue
+            intersection[pkg] = None
+        self._exposed_packages.clear()
+        exposed_packages = intersection
+
+    self._exposed_packages.update(exposed_packages)
+
+def _add_group_map(self, group_map):
+    # TODO @aignas 2024-04-05: how do we support different requirement
+    # cycles for different abis/oses? For now we will need the users to
+    # assume the same groups across all versions/platforms until we start
+    # using an alternative cycle resolution strategy.
+    self._group_map.clear()
+    self._group_map.update(group_map)
+
+def _add_extra_aliases(self, extra_hub_aliases):
+    for whl_name, aliases in extra_hub_aliases.items():
+        self._extra_aliases.setdefault(whl_name, {}).update(
+            {alias: True for alias in aliases},
+        )
+
+def _add_whl_library(self, *, python_version, whl, repo):
+    if repo == None:
+        # NOTE @aignas 2025-07-07: we guard against an edge-case where there
+        # are more platforms defined than there are wheels for and users
+        # disallow building from sdist.
+        return
+
+    platforms = self._platforms[python_version]
+
+    # TODO @aignas 2025-06-29: we should not need the version in the repo_name if
+    # we are using pipstar and we are downloading the wheel using the downloader
+    repo_name = "{}_{}_{}".format(self.name, version_label(python_version), repo.repo_name)
+
+    if repo_name in self._whl_libraries:
+        fail("attempting to create a duplicate library {} for {}".format(
+            repo_name,
+            whl.name,
+        ))
+    self._whl_libraries[repo_name] = repo.args
+
+    if not self._config.enable_pipstar and "experimental_target_platforms" in repo.args:
+        self._whl_libraries[repo_name] |= {
+            "experimental_target_platforms": sorted({
+                # TODO @aignas 2025-07-07: this should be solved in a better way
+                platforms[candidate].triple.partition("_")[-1]: None
+                for p in repo.args["experimental_target_platforms"]
+                for candidate in platforms
+                if candidate.endswith(p)
+            }),
+        }
+
+    mapping = self._whl_map.setdefault(whl.name, {})
+    if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
+        fail(
+            "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
+                mapping[repo.config_setting],
+                repo.config_setting,
+                repo_name,
+            ),
+        )
+    else:
+        mapping[repo.config_setting] = repo_name
+
+### end of setters, below we have various functions to implement the public methods
+
 def _set_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:
@@ -213,49 +311,6 @@ def _platforms(*, python_version, minor_mapping, config):
         )
     return platforms
 
-def _add_whl_library(self, *, python_version, whl, repo):
-    if repo == None:
-        # NOTE @aignas 2025-07-07: we guard against an edge-case where there
-        # are more platforms defined than there are wheels for and users
-        # disallow building from sdist.
-        return
-
-    platforms = self._platforms[python_version]
-
-    # TODO @aignas 2025-06-29: we should not need the version in the repo_name if
-    # we are using pipstar and we are downloading the wheel using the downloader
-    repo_name = "{}_{}_{}".format(self.name, version_label(python_version), repo.repo_name)
-
-    if repo_name in self._whl_libraries:
-        fail("attempting to create a duplicate library {} for {}".format(
-            repo_name,
-            whl.name,
-        ))
-    self._whl_libraries[repo_name] = repo.args
-
-    if not self._config.enable_pipstar and "experimental_target_platforms" in repo.args:
-        self._whl_libraries[repo_name] |= {
-            "experimental_target_platforms": sorted({
-                # TODO @aignas 2025-07-07: this should be solved in a better way
-                platforms[candidate].triple.partition("_")[-1]: None
-                for p in repo.args["experimental_target_platforms"]
-                for candidate in platforms
-                if candidate.endswith(p)
-            }),
-        }
-
-    mapping = self._whl_map.setdefault(whl.name, {})
-    if repo.config_setting in mapping and mapping[repo.config_setting] != repo_name:
-        fail(
-            "attempting to override an existing repo '{}' for config setting '{}' with a new repo '{}'".format(
-                mapping[repo.config_setting],
-                repo.config_setting,
-                repo_name,
-            ),
-        )
-    else:
-        mapping[repo.config_setting] = repo_name
-
 def _evaluate_markers(self, pip_attr):
     if self._evaluate_markers_fn:
         return self._evaluate_markers_fn
@@ -425,18 +480,6 @@ def _create_whl_repos(
 
     _add_exposed_packages(self, exposed_packages)
 
-def _add_exposed_packages(self, exposed_packages):
-    if self._exposed_packages:
-        intersection = {}
-        for pkg in exposed_packages:
-            if pkg not in self._exposed_packages:
-                continue
-            intersection[pkg] = None
-        self._exposed_packages.clear()
-        exposed_packages = intersection
-
-    self._exposed_packages.update(exposed_packages)
-
 def _whl_repo(
         *,
         src,
@@ -508,44 +551,8 @@ def _whl_repo(
         ),
     )
 
-def _add_group_map(self, group_map):
-    # TODO @aignas 2024-04-05: how do we support different requirement
-    # cycles for different abis/oses? For now we will need the users to
-    # assume the same groups across all versions/platforms until we start
-    # using an alternative cycle resolution strategy.
-    self._group_map.clear()
-    self._group_map.update(group_map)
-
-def _add_extra_aliases(self, extra_hub_aliases):
-    for whl_name, aliases in extra_hub_aliases.items():
-        self._extra_aliases.setdefault(whl_name, {}).update(
-            {alias: True for alias in aliases},
-        )
-
-def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
-    python_version = pip_attr.python_version
-    if python_version in self._platforms:
-        fail((
-            "Duplicate pip python version '{version}' for hub " +
-            "'{hub}' in module '{module}': the Python versions " +
-            "used for a hub must be unique"
-        ).format(
-            hub = self.name,
-            module = self.module_name,
-            version = python_version,
-        ))
-
-    self._platforms[python_version] = _platforms(
-        python_version = python_version,
-        minor_mapping = self._minor_mapping,
-        config = self._config,
-    )
-    _set_index_urls(self, pip_attr)
-    _add_group_map(self, pip_attr.experimental_requirement_cycles)
-    _add_extra_aliases(self, pip_attr.extra_hub_aliases)
-    _create_whl_repos(
-        self,
-        module_ctx,
-        pip_attr = pip_attr,
-        whl_overrides = whl_overrides,
+def _use_downloader(self, python_version, whl_name):
+    return self._use_downloader.get(python_version, {}).get(
+        normalize_name(whl_name),
+        self._get_index_urls.get(python_version) != None,
     )

From c69f1a13e421b00876ab74c98e1f973385279ae1 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sun, 7 Sep 2025 00:20:25 +0900
Subject: [PATCH 28/31] refactor: clarify name

---
 python/private/pypi/hub_builder.bzl | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 32eeba08b2..85d172bf47 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -115,7 +115,7 @@ def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
         minor_mapping = self._minor_mapping,
         config = self._config,
     )
-    _set_index_urls(self, pip_attr)
+    _set_get_index_urls(self, pip_attr)
     _add_group_map(self, pip_attr.experimental_requirement_cycles)
     _add_extra_aliases(self, pip_attr.extra_hub_aliases)
     _create_whl_repos(
@@ -199,7 +199,7 @@ def _add_whl_library(self, *, python_version, whl, repo):
 
 ### end of setters, below we have various functions to implement the public methods
 
-def _set_index_urls(self, pip_attr):
+def _set_get_index_urls(self, pip_attr):
     if not pip_attr.experimental_index_url:
         if pip_attr.experimental_extra_index_urls:
             fail("'experimental_extra_index_urls' is a no-op unless 'experimental_index_url' is set")

From c06b99c68e8c9563ee4ebbfa2ec9749c78808c21 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sun, 7 Sep 2025 00:28:18 +0900
Subject: [PATCH 29/31] move the construction of the requirement cycles

---
 python/private/pypi/hub_builder.bzl | 31 +++++++++++++----------------
 1 file changed, 14 insertions(+), 17 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 85d172bf47..90005f9eb5 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -63,6 +63,7 @@ def hub_builder(
         _whl_map = {},  # modified by _add_whl_library
         # internal
         _platforms = {},
+        _group_name_by_whl = {},
         _get_index_urls = {},
         _use_downloader = {},
         _simpleapi_cache = simpleapi_cache,
@@ -145,8 +146,19 @@ def _add_group_map(self, group_map):
     # cycles for different abis/oses? For now we will need the users to
     # assume the same groups across all versions/platforms until we start
     # using an alternative cycle resolution strategy.
+    group_map = {
+        name: [normalize_name(whl_name) for whl_name in whls]
+        for name, whls in group_map.items()
+    }
     self._group_map.clear()
+    self._group_name_by_whl.clear()
+
     self._group_map.update(group_map)
+    self._group_name_by_whl.update({
+        whl_name: group_name
+        for group_name, group_whls in self._group_map.items()
+        for whl_name in group_whls
+    })
 
 def _add_extra_aliases(self, extra_hub_aliases):
     for whl_name, aliases in extra_hub_aliases.items():
@@ -396,29 +408,14 @@ def _create_whl_repos(
         for mod, whl_name in pip_attr.whl_modifications.items():
             whl_modifications[normalize_name(whl_name)] = mod
 
-    if pip_attr.experimental_requirement_cycles:
-        requirement_cycles = {
-            name: [normalize_name(whl_name) for whl_name in whls]
-            for name, whls in pip_attr.experimental_requirement_cycles.items()
-        }
-
-        whl_group_mapping = {
-            whl_name: group_name
-            for group_name, group_whls in requirement_cycles.items()
-            for whl_name in group_whls
-        }
-    else:
-        whl_group_mapping = {}
-        requirement_cycles = {}
-
     interpreter = _detect_interpreter(self, pip_attr)
     exposed_packages = {}
     for whl in requirements_by_platform:
         if whl.is_exposed:
             exposed_packages[whl.name] = None
 
-        group_name = whl_group_mapping.get(whl.name)
-        group_deps = requirement_cycles.get(group_name, [])
+        group_name = self._group_name_by_whl.get(whl.name)
+        group_deps = self._group_map.get(group_name, [])
 
         # Construct args separately so that the lock file can be smaller and does not include unused
         # attrs.

From b00aff0da5d7454d7ba682beb500ea9b6a83619e Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sun, 7 Sep 2025 00:52:17 +0900
Subject: [PATCH 30/31] refactor: pass whl_overrides differently

---
 python/private/pypi/extension.bzl   |   2 +-
 python/private/pypi/hub_builder.bzl | 112 +++++++++++++++-------------
 2 files changed, 62 insertions(+), 52 deletions(-)

diff --git a/python/private/pypi/extension.bzl b/python/private/pypi/extension.bzl
index 66dfc21062..c73e88ac0d 100644
--- a/python/private/pypi/extension.bzl
+++ b/python/private/pypi/extension.bzl
@@ -250,6 +250,7 @@ You cannot use both the additive_build_content and additive_build_content_file a
                     name = hub_name,
                     module_name = mod.name,
                     config = config,
+                    whl_overrides = whl_overrides,
                     simpleapi_download_fn = simpleapi_download,
                     simpleapi_cache = simpleapi_cache,
                     # TODO @aignas 2025-09-06: do not use kwargs
@@ -279,7 +280,6 @@ You cannot use both the additive_build_content and additive_build_content_file a
             builder.pip_parse(
                 module_ctx,
                 pip_attr = pip_attr,
-                whl_overrides = whl_overrides,
             )
 
     # Keeps track of all the hub's whl repos across the different versions.
diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index 90005f9eb5..b5b0de7d2d 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -23,6 +23,7 @@ def hub_builder(
         name,
         module_name,
         config,
+        whl_overrides,
         minor_mapping,
         available_interpreters,
         simpleapi_download_fn,
@@ -35,6 +36,7 @@ def hub_builder(
         name: {type}`str`, the name of the hub.
         module_name: {type}`str`, the module name that has created the hub.
         config: The platform configuration.
+        whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
         minor_mapping: {type}`dict[str, str]` the mapping between minor and full versions.
         evaluate_markers_fn: the override function used to evaluate the markers.
         available_interpreters: {type}`dict[str, Label]` The dictionary of available
@@ -69,6 +71,7 @@ def hub_builder(
         _simpleapi_cache = simpleapi_cache,
         # instance constants
         _config = config,
+        _whl_overrides = whl_overrides,
         _evaluate_markers_fn = evaluate_markers_fn,
         _logger = logger,
         _minor_mapping = minor_mapping,
@@ -98,7 +101,7 @@ def _build(self):
         whl_libraries = self._whl_libraries,
     )
 
-def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
+def _pip_parse(self, module_ctx, pip_attr):
     python_version = pip_attr.python_version
     if python_version in self._platforms:
         fail((
@@ -123,7 +126,6 @@ def _pip_parse(self, module_ctx, pip_attr, whl_overrides):
         self,
         module_ctx,
         pip_attr = pip_attr,
-        whl_overrides = whl_overrides,
     )
 
 ### end of PUBLIC methods
@@ -368,15 +370,13 @@ def _create_whl_repos(
         self,
         module_ctx,
         *,
-        pip_attr,
-        whl_overrides):
+        pip_attr):
     """create all of the whl repositories
 
     Args:
         self: the builder.
         module_ctx: {type}`module_ctx`.
         pip_attr: {type}`struct` - the struct that comes from the tag class iteration.
-        whl_overrides: {type}`dict[str, struct]` - per-wheel overrides.
     """
     logger = self._logger
     platforms = self._platforms[pip_attr.python_version]
@@ -403,58 +403,25 @@ def _create_whl_repos(
         logger = logger,
     )
 
+    _add_exposed_packages(self, {
+        whl.name: None
+        for whl in requirements_by_platform
+        if whl.is_exposed
+    })
+
     whl_modifications = {}
     if pip_attr.whl_modifications != None:
         for mod, whl_name in pip_attr.whl_modifications.items():
             whl_modifications[normalize_name(whl_name)] = mod
 
-    interpreter = _detect_interpreter(self, pip_attr)
-    exposed_packages = {}
     for whl in requirements_by_platform:
-        if whl.is_exposed:
-            exposed_packages[whl.name] = None
-
-        group_name = self._group_name_by_whl.get(whl.name)
-        group_deps = self._group_map.get(group_name, [])
-
-        # Construct args separately so that the lock file can be smaller and does not include unused
-        # attrs.
-        whl_library_args = dict(
-            dep_template = "@{}//{{name}}:{{target}}".format(self.name),
-        )
-        maybe_args = dict(
-            # The following values are safe to omit if they have false like values
-            add_libdir_to_library_search_path = pip_attr.add_libdir_to_library_search_path,
-            annotation = whl_modifications.get(whl.name),
-            download_only = pip_attr.download_only,
-            enable_implicit_namespace_pkgs = pip_attr.enable_implicit_namespace_pkgs,
-            environment = pip_attr.environment,
-            envsubst = pip_attr.envsubst,
-            group_deps = group_deps,
-            group_name = group_name,
-            pip_data_exclude = pip_attr.pip_data_exclude,
-            python_interpreter = interpreter.path,
-            python_interpreter_target = interpreter.target,
-            whl_patches = {
-                p: json.encode(args)
-                for p, args in whl_overrides.get(whl.name, {}).items()
-            },
-        )
-        if not self._config.enable_pipstar:
-            maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
-
-        whl_library_args.update({k: v for k, v in maybe_args.items() if v})
-        maybe_args_with_default = dict(
-            # The following values have defaults next to them
-            isolated = (use_isolated(module_ctx, pip_attr), True),
-            quiet = (pip_attr.quiet, True),
-            timeout = (pip_attr.timeout, 600),
+        whl_library_args = _whl_library_args(
+            self,
+            module_ctx,
+            pip_attr = pip_attr,
+            whl = whl,
+            whl_modifications = whl_modifications,
         )
-        whl_library_args.update({
-            k: v
-            for k, (v, default) in maybe_args_with_default.items()
-            if v != default
-        })
 
         for src in whl.srcs:
             repo = _whl_repo(
@@ -475,7 +442,50 @@ def _create_whl_repos(
                 repo = repo,
             )
 
-    _add_exposed_packages(self, exposed_packages)
+def _whl_library_args(self, module_ctx, *, pip_attr, whl, whl_modifications):
+    interpreter = _detect_interpreter(self, pip_attr)
+    group_name = self._group_name_by_whl.get(whl.name)
+    group_deps = self._group_map.get(group_name, [])
+
+    # Construct args separately so that the lock file can be smaller and does not include unused
+    # attrs.
+    whl_library_args = dict(
+        dep_template = "@{}//{{name}}:{{target}}".format(self.name),
+    )
+    maybe_args = dict(
+        # The following values are safe to omit if they have false like values
+        add_libdir_to_library_search_path = pip_attr.add_libdir_to_library_search_path,
+        annotation = whl_modifications.get(whl.name),
+        download_only = pip_attr.download_only,
+        enable_implicit_namespace_pkgs = pip_attr.enable_implicit_namespace_pkgs,
+        environment = pip_attr.environment,
+        envsubst = pip_attr.envsubst,
+        group_deps = group_deps,
+        group_name = group_name,
+        pip_data_exclude = pip_attr.pip_data_exclude,
+        python_interpreter = interpreter.path,
+        python_interpreter_target = interpreter.target,
+        whl_patches = {
+            p: json.encode(args)
+            for p, args in self._whl_overrides.get(whl.name, {}).items()
+        },
+    )
+    if not self._config.enable_pipstar:
+        maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
+
+    whl_library_args.update({k: v for k, v in maybe_args.items() if v})
+    maybe_args_with_default = dict(
+        # The following values have defaults next to them
+        isolated = (use_isolated(module_ctx, pip_attr), True),
+        quiet = (pip_attr.quiet, True),
+        timeout = (pip_attr.timeout, 600),
+    )
+    whl_library_args.update({
+        k: v
+        for k, (v, default) in maybe_args_with_default.items()
+        if v != default
+    })
+    return whl_library_args
 
 def _whl_repo(
         *,

From cc2b35370776c651ade768baa48a279d0adeee92 Mon Sep 17 00:00:00 2001
From: Ignas Anikevicius <240938+aignas@users.noreply.github.com>
Date: Sun, 7 Sep 2025 00:56:02 +0900
Subject: [PATCH 31/31] split the whl_library_arg creation

---
 python/private/pypi/hub_builder.bzl | 44 ++++++++++++++++++++---------
 1 file changed, 30 insertions(+), 14 deletions(-)

diff --git a/python/private/pypi/hub_builder.bzl b/python/private/pypi/hub_builder.bzl
index b5b0de7d2d..b6088e4ded 100644
--- a/python/private/pypi/hub_builder.bzl
+++ b/python/private/pypi/hub_builder.bzl
@@ -414,15 +414,17 @@ def _create_whl_repos(
         for mod, whl_name in pip_attr.whl_modifications.items():
             whl_modifications[normalize_name(whl_name)] = mod
 
+    common_args = _common_args(
+        self,
+        module_ctx,
+        pip_attr = pip_attr,
+    )
     for whl in requirements_by_platform:
-        whl_library_args = _whl_library_args(
+        whl_library_args = common_args | _whl_library_args(
             self,
-            module_ctx,
-            pip_attr = pip_attr,
             whl = whl,
             whl_modifications = whl_modifications,
         )
-
         for src in whl.srcs:
             repo = _whl_repo(
                 src = src,
@@ -442,10 +444,8 @@ def _create_whl_repos(
                 repo = repo,
             )
 
-def _whl_library_args(self, module_ctx, *, pip_attr, whl, whl_modifications):
+def _common_args(self, module_ctx, *, pip_attr):
     interpreter = _detect_interpreter(self, pip_attr)
-    group_name = self._group_name_by_whl.get(whl.name)
-    group_deps = self._group_map.get(group_name, [])
 
     # Construct args separately so that the lock file can be smaller and does not include unused
     # attrs.
@@ -455,20 +455,13 @@ def _whl_library_args(self, module_ctx, *, pip_attr, whl, whl_modifications):
     maybe_args = dict(
         # The following values are safe to omit if they have false like values
         add_libdir_to_library_search_path = pip_attr.add_libdir_to_library_search_path,
-        annotation = whl_modifications.get(whl.name),
         download_only = pip_attr.download_only,
         enable_implicit_namespace_pkgs = pip_attr.enable_implicit_namespace_pkgs,
         environment = pip_attr.environment,
         envsubst = pip_attr.envsubst,
-        group_deps = group_deps,
-        group_name = group_name,
         pip_data_exclude = pip_attr.pip_data_exclude,
         python_interpreter = interpreter.path,
         python_interpreter_target = interpreter.target,
-        whl_patches = {
-            p: json.encode(args)
-            for p, args in self._whl_overrides.get(whl.name, {}).items()
-        },
     )
     if not self._config.enable_pipstar:
         maybe_args["experimental_target_platforms"] = pip_attr.experimental_target_platforms
@@ -487,6 +480,29 @@ def _whl_library_args(self, module_ctx, *, pip_attr, whl, whl_modifications):
     })
     return whl_library_args
 
+def _whl_library_args(self, *, whl, whl_modifications):
+    group_name = self._group_name_by_whl.get(whl.name)
+    group_deps = self._group_map.get(group_name, [])
+
+    # Construct args separately so that the lock file can be smaller and does not include unused
+    # attrs.
+    whl_library_args = dict(
+        dep_template = "@{}//{{name}}:{{target}}".format(self.name),
+    )
+    maybe_args = dict(
+        # The following values are safe to omit if they have false like values
+        annotation = whl_modifications.get(whl.name),
+        group_deps = group_deps,
+        group_name = group_name,
+        whl_patches = {
+            p: json.encode(args)
+            for p, args in self._whl_overrides.get(whl.name, {}).items()
+        },
+    )
+
+    whl_library_args.update({k: v for k, v in maybe_args.items() if v})
+    return whl_library_args
+
 def _whl_repo(
         *,
         src,