refactor(pypi): parse entry_points without Python (#3429)

This is a small utility function to get us Python free when wheels are
extracted in the repository phase.

Next is to extract the wheel using `repository_ctx.extract` (#3430).

Whereas patching the wheel after extracting is more involved to be done
without Python because we need to rezip the wheel and that has to be
done
with Python for a few reasons (to stay sane). If we want to remove this,
then we would have to create a `whl` file in the build phase, which
could
work, but will need to be an exercise for the reader.

Nevertheless, this moves us towards removing any side-effects from
Python
interpreter, so changing the default interpreter would not cause us to
refetch everything.

---------

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

Author: aignas

python/private/pypi/whl_installer/wheel_installer.py

@@ -96,7 +96,17 @@ def _extract_wheel(
     whl = wheel.Wheel(wheel_file)
     whl.unzip(installation_dir)
 
+    if enable_pipstar:
+        return
+
+    extras_requested = extras[whl.name] if whl.name in extras else set()
+    dependencies = whl.dependencies(extras_requested, platforms)
+
     metadata = {
+        "name": whl.name,
+        "version": whl.version,
+        "deps": dependencies.deps,
+        "deps_by_platform": dependencies.deps_select,
         "entry_points": [
             {
                 "name": name,
@@ -106,18 +116,6 @@ def _extract_wheel(
             for name, (module, attribute) in sorted(whl.entry_points().items())
         ],
     }
-    if not enable_pipstar:
-        extras_requested = extras[whl.name] if whl.name in extras else set()
-        dependencies = whl.dependencies(extras_requested, platforms)
-
-        metadata.update(
-            {
-                "name": whl.name,
-                "version": whl.version,
-                "deps": dependencies.deps,
-                "deps_by_platform": dependencies.deps_select,
-            }
-        )
 
     with open(os.path.join(installation_dir, "metadata.json"), "w") as f:
         json.dump(metadata, f)

python/private/pypi/whl_library.bzl

@@ -391,18 +391,21 @@ def _whl_library_impl(rctx):
             logger = logger,
         )
 
-        metadata = json.decode(rctx.read("metadata.json"))
-        rctx.delete("metadata.json")
+        metadata = whl_metadata(
+            install_dir = whl_path.dirname.get_child("site-packages"),
+            read_fn = rctx.read,
+            logger = logger,
+        )
 
         # NOTE @aignas 2024-06-22: this has to live on until we stop supporting
         # passing `twine` as a `:pkg` library via the `WORKSPACE` builds.
         #
         # See ../../packaging.bzl line 190
         entry_points = {}
-        for item in metadata["entry_points"]:
-            name = item["name"]
-            module = item["module"]
-            attribute = item["attribute"]
+        for item in metadata.entry_points:
+            name = item.name
+            module = item.module
+            attribute = item.attribute
 
             # There is an extreme edge-case with entry_points that end with `.py`
             # See: https://github.com/bazelbuild/bazel/blob/09c621e4cf5b968f4c6cdf905ab142d5961f9ddc/src/test/java/com/google/devtools/build/lib/rules/python/PyBinaryConfiguredTargetTest.java#L174
@@ -418,12 +421,6 @@ def _whl_library_impl(rctx):
             )
             entry_points[entry_point_without_py] = entry_point_script_name
 
-        metadata = whl_metadata(
-            install_dir = whl_path.dirname.get_child("site-packages"),
-            read_fn = rctx.read,
-            logger = logger,
-        )
-
         build_file_contents = generate_whl_library_build_bazel(
             name = whl_path.basename,
             sdist_filename = sdist_filename,

python/private/pypi/whl_metadata.bzl

@@ -4,6 +4,7 @@ _NAME = "Name: "
 _PROVIDES_EXTRA = "Provides-Extra: "
 _REQUIRES_DIST = "Requires-Dist: "
 _VERSION = "Version: "
+_CONSOLE_SCRIPTS = "[console_scripts]"
 
 def whl_metadata(*, install_dir, read_fn, logger):
     """Find and parse the METADATA file in the extracted whl contents dir.
@@ -23,7 +24,13 @@ def whl_metadata(*, install_dir, read_fn, logger):
     """
     metadata_file = find_whl_metadata(install_dir = install_dir, logger = logger)
     contents = read_fn(metadata_file)
-    result = parse_whl_metadata(contents)
+    entry_points_file = metadata_file.dirname.get_child("entry_points.txt")
+    if entry_points_file.exists:
+        entry_points_contents = read_fn(entry_points_file)
+    else:
+        entry_points_contents = ""
+
+    result = parse_whl_metadata(contents, entry_points_contents)
 
     if not (result.name and result.version):
         logger.fail("Failed to parse the wheel METADATA file:\n{}\n{}\n{}".format(
@@ -35,11 +42,12 @@ def whl_metadata(*, install_dir, read_fn, logger):
 
     return result
 
-def parse_whl_metadata(contents):
+def parse_whl_metadata(contents, entry_points_contents = ""):
     """Parse .whl METADATA file
 
     Args:
         contents: {type}`str` the contents of the file.
+        entry_points_contents: {type}`str` the contents of the `entry_points.txt` file if it exists.
 
     Returns:
         A struct with parsed values:
@@ -48,6 +56,8 @@ def parse_whl_metadata(contents):
         * `requires_dist`: {type}`list[str]` the list of requirements.
         * `provides_extra`: {type}`list[str]` the list of extras that this package
           provides.
+        * `entry_points`: {type}`list[struct]` the list of
+            entry_point metadata.
     """
     parsed = {
         "name": "",
@@ -79,6 +89,7 @@ def parse_whl_metadata(contents):
         provides_extra = parsed["provides_extra"],
         requires_dist = parsed["requires_dist"],
         version = parsed["version"],
+        entry_points = _parse_entry_points(entry_points_contents),
     )
 
 def find_whl_metadata(*, install_dir, logger):
@@ -110,3 +121,47 @@ def find_whl_metadata(*, install_dir, logger):
     else:
         logger.fail("The '*.dist-info' directory could not be found in '{}'".format(install_dir.basename))
     return None
+
+def _parse_entry_points(contents):
+    """parse the entry_points.txt file.
+
+    Args:
+        contents: {type}`str` The contents of the file
+
+    Returns:
+        A list of console_script entry point metadata.
+    """
+    start = False
+    ret = []
+    for line in contents.split("\n"):
+        line = line.rstrip()
+
+        if line == _CONSOLE_SCRIPTS:
+            start = True
+            continue
+
+        if not start:
+            continue
+
+        if start and line.startswith("["):
+            break
+
+        line, _, _comment = line.partition("#")
+        line = line.strip()
+        if not line:
+            continue
+
+        name, _, tail = line.partition("=")
+
+        # importable.module:object.attr
+        py_import, _, extras = tail.strip().partition(" ")
+        module, _, attribute = py_import.partition(":")
+
+        ret.append(struct(
+            name = name.strip(),
+            module = module.strip(),
+            attribute = attribute.strip(),
+            extras = extras.replace(" ", ""),
+        ))
+
+    return ret

tests/pypi/whl_metadata/whl_metadata_tests.bzl

@@ -81,12 +81,14 @@ def _parse_whl_metadata(env, **kwargs):
             version = result.version,
             requires_dist = result.requires_dist,
             provides_extra = result.provides_extra,
+            entry_points = result.entry_points,
         ),
         attrs = dict(
             name = subjects.str,
             version = subjects.str,
             requires_dist = subjects.collection,
             provides_extra = subjects.collection,
+            entry_points = subjects.collection,
         ),
     )
 
@@ -171,6 +173,53 @@ Requires-Dist: this will be ignored
 
 _tests.append(_test_parse_metadata_multiline_license)
 
+def _test_parse_entry_points_txt(env):
+    got = _parse_whl_metadata(
+        env,
+        contents = """\
+Name: foo
+Version: 0.0.1
+""",
+        entry_points_contents = """\
+[something]
+interesting # with comments
+
+[console_scripts]
+foo = foomod:main
+# One which depends on extras:
+foobar = importable.foomod:main_bar [bar, baz]
+
+  # With a comment at the end
+foobarbaz = foomod:main.attr # comment
+
+[something else]
+not very much interesting
+
+""",
+    )
+    got.entry_points().contains_exactly([
+        struct(
+            attribute = "main",
+            extras = "",
+            module = "foomod",
+            name = "foo",
+        ),
+        struct(
+            attribute = "main_bar",
+            extras = "[bar,baz]",
+            module = "importable.foomod",
+            name = "foobar",
+        ),
+        struct(
+            attribute = "main.attr",
+            extras = "",
+            module = "foomod",
+            name = "foobarbaz",
+        ),
+    ])
+
+_tests.append(_test_parse_entry_points_txt)
+
 def whl_metadata_test_suite(name):  # buildifier: disable=function-docstring
     test_suite(
         name = name,