Merge pull request #11052 from SpecLad/fix-script-record-hash

Update the RECORD entry when rewriting the shebang line in a script

Author: sbidoul

news/10744.bugfix.rst

@@ -0,0 +1,2 @@
+When pip rewrites the shebang line in a script during wheel installation,
+update the hash and size in the corresponding ``RECORD`` file entry.

src/pip/_internal/operations/install/wheel.py

@@ -224,19 +224,16 @@ def _normalized_outrows(
     )
 
 
-def _record_to_fs_path(record_path: RecordPath) -> str:
-    return record_path
-
-
-def _fs_to_record_path(path: str, relative_to: Optional[str] = None) -> RecordPath:
-    if relative_to is not None:
-        # On Windows, do not handle relative paths if they belong to different
-        # logical disks
-        if (
-            os.path.splitdrive(path)[0].lower()
-            == os.path.splitdrive(relative_to)[0].lower()
-        ):
-            path = os.path.relpath(path, relative_to)
+def _record_to_fs_path(record_path: RecordPath, lib_dir: str) -> str:
+    return os.path.join(lib_dir, record_path)
+
+
+def _fs_to_record_path(path: str, lib_dir: str) -> RecordPath:
+    # On Windows, do not handle relative paths if they belong to different
+    # logical disks
+    if os.path.splitdrive(path)[0].lower() == os.path.splitdrive(lib_dir)[0].lower():
+        path = os.path.relpath(path, lib_dir)
+
     path = path.replace(os.path.sep, "/")
     return cast("RecordPath", path)
 
@@ -259,7 +256,7 @@ def get_csv_rows_for_installed(
         old_record_path = cast("RecordPath", row[0])
         new_record_path = installed.pop(old_record_path, old_record_path)
         if new_record_path in changed:
-            digest, length = rehash(_record_to_fs_path(new_record_path))
+            digest, length = rehash(_record_to_fs_path(new_record_path, lib_dir))
         else:
             digest = row[1] if len(row) > 1 else ""
             length = row[2] if len(row) > 2 else ""
@@ -475,7 +472,7 @@ def record_installed(
         newpath = _fs_to_record_path(destfile, lib_dir)
         installed[srcfile] = newpath
         if modified:
-            changed.add(_fs_to_record_path(destfile))
+            changed.add(newpath)
 
     def is_dir_path(path: RecordPath) -> bool:
         return path.endswith("/")

tests/functional/test_install_wheel.py

@@ -1,5 +1,7 @@
+import base64
 import csv
 import distutils
+import hashlib
 import os
 import shutil
 from pathlib import Path
@@ -364,6 +366,44 @@ def test_wheel_record_lines_have_hash_for_data_files(
     ]
 
 
+def test_wheel_record_lines_have_updated_hash_for_scripts(
+    script: PipTestEnvironment,
+) -> None:
+    """
+    pip rewrites "#!python" shebang lines in scripts when it installs them;
+    make sure it updates the RECORD file correspondingly.
+    """
+    package = make_wheel(
+        "simple",
+        "0.1.0",
+        extra_data_files={
+            "scripts/dostuff": "#!python\n",
+        },
+    ).save_to_dir(script.scratch_path)
+    script.pip("install", package)
+    record_file = script.site_packages_path / "simple-0.1.0.dist-info" / "RECORD"
+    record_text = record_file.read_text()
+    record_rows = list(csv.reader(record_text.splitlines()))
+    records = {r[0]: r[1:] for r in record_rows}
+
+    script_path = script.bin_path / "dostuff"
+    script_contents = script_path.read_bytes()
+    assert not script_contents.startswith(b"#!python\n")
+
+    script_digest = hashlib.sha256(script_contents).digest()
+    script_digest_b64 = (
+        base64.urlsafe_b64encode(script_digest).decode("US-ASCII").rstrip("=")
+    )
+
+    script_record_path = os.path.relpath(
+        script_path, script.site_packages_path
+    ).replace(os.path.sep, "/")
+    assert records[script_record_path] == [
+        f"sha256={script_digest_b64}",
+        str(len(script_contents)),
+    ]
+
+
 @pytest.mark.incompatible_with_test_venv
 @pytest.mark.usefixtures("with_wheel")
 def test_install_user_wheel(