extract out url and filename

todo: check later if filename is necessary. Maybe not.

Author: chrisirhc

python/private/pypi/index_sources.bzl

@@ -32,6 +32,8 @@ def index_sources(line):
             * `marker` - str; the marker expression, as per PEP508 spec.
             * `requirement` - str; a requirement line without the marker. This can
                 be given to `pip` to install a package.
+            * `url` - str; URL if the requirement specifies a direct URL.
+            * `filename` - str; filename if URL is present, extracted from the URL.
     """
     line = line.replace("\\", " ")
     head, _, maybe_hashes = line.partition(";")
@@ -55,14 +57,26 @@ def index_sources(line):
         requirement,
         " ".join(["--hash=sha256:{}".format(sha) for sha in shas]),
     ).strip()
+
+    # Extract URL if present
+    url = ""
+    filename = ""
     if "@" in head:
         requirement = requirement_line
-        shas = []
+        # Extract URL from direct URL format
+        url = requirement.split("@")[1].split("#")[0].strip()
+        # Extract filename from URL
+        if url:
+            filename = url.rpartition("/")[2]
+            if not filename:
+                filename = url.rpartition("/")[0].rpartition("/")[2]
 
     return struct(
         requirement = requirement,
         requirement_line = requirement_line,
         version = version,
         shas = sorted(shas),
         marker = marker,
+        url = url,
+        filename = filename,
     )

tests/pypi/index_sources/index_sources_tests.bzl

@@ -24,27 +24,45 @@ def _test_no_simple_api_sources(env):
         "foo==0.0.1": struct(
             requirement = "foo==0.0.1",
             marker = "",
+            url = "",
+            filename = "",
         ),
         "foo==0.0.1 @ https://someurl.org": struct(
             requirement = "foo==0.0.1 @ https://someurl.org",
             marker = "",
+            url = "https://someurl.org",
+            filename = "someurl.org",
         ),
-        "foo==0.0.1 @ https://someurl.org --hash=sha256:deadbeef": struct(
-            requirement = "foo==0.0.1 @ https://someurl.org --hash=sha256:deadbeef",
+        "foo==0.0.1 @ https://someurl.org/package.whl": struct(
+            requirement = "foo==0.0.1 @ https://someurl.org/package.whl",
             marker = "",
+            url = "https://someurl.org/package.whl",
+            filename = "package.whl",
         ),
-        "foo==0.0.1 @ https://someurl.org; python_version < \"2.7\"\\    --hash=sha256:deadbeef": struct(
-            requirement = "foo==0.0.1 @ https://someurl.org --hash=sha256:deadbeef",
+        "foo==0.0.1 @ https://someurl.org/package.whl --hash=sha256:deadbeef": struct(
+            requirement = "foo==0.0.1 @ https://someurl.org/package.whl --hash=sha256:deadbeef",
+            marker = "",
+            url = "https://someurl.org/package.whl",
+            filename = "package.whl",
+            shas = ["deadbeef"],
+        ),
+        "foo==0.0.1 @ https://someurl.org/package.whl; python_version < \"2.7\"\\    --hash=sha256:deadbeef": struct(
+            requirement = "foo==0.0.1 @ https://someurl.org/package.whl --hash=sha256:deadbeef",
             marker = "python_version < \"2.7\"",
+            url = "https://someurl.org/package.whl",
+            filename = "package.whl",
+            shas = ["deadbeef"],
         ),
     }
     for input, want in inputs.items():
         got = index_sources(input)
-        env.expect.that_collection(got.shas).contains_exactly([])
+        env.expect.that_collection(got.shas).contains_exactly(want.shas if hasattr(want, "shas") else [])
         env.expect.that_str(got.version).equals("0.0.1")
         env.expect.that_str(got.requirement).equals(want.requirement)
         env.expect.that_str(got.requirement_line).equals(got.requirement)
         env.expect.that_str(got.marker).equals(want.marker)
+        env.expect.that_str(got.url).equals(want.url)
+        env.expect.that_str(got.filename).equals(want.filename)
 
 _tests.append(_test_no_simple_api_sources)