Merge pull request #7 from python-packaging/homepage-project-urls

Parse metadata fields from sdist artifacts

Author: amjith

metadata_please/sdist.py

@@ -30,12 +30,23 @@ def from_zip_sdist(zf: ZipFile) -> bytes:
 def basic_metadata_from_zip_sdist(zf: ZipFile) -> BasicMetadata:
     requires = [f for f in zf.namelist() if f.endswith("/requires.txt")]
     requires.sort(key=len)
-    if not requires:
-        return BasicMetadata((), frozenset(), "-")
+    if requires:
+        requires_data = zf.read(requires[0])
+        assert requires_data is not None
+    else:
+        requires_data = b""
 
-    data = zf.read(requires[0])
-    assert data is not None
-    return BasicMetadata.from_sdist_pkg_info_and_requires(b"", data)
+    # Find the PKG-INFO file with the shortest path. This is to avoid picking up
+    # a PKG-INFO file from a nested test directory.
+    pkg_info = sorted(
+        (f for f in zf.namelist() if f == "PKG-INFO" or f.endswith("/PKG-INFO")),
+        key=len,
+    )[0]
+
+    pkg_info_data = zf.read(pkg_info)
+    assert pkg_info_data is not None
+
+    return BasicMetadata.from_sdist_pkg_info_and_requires(pkg_info_data, requires_data)
 
 
 def from_tar_sdist(tf: TarFile) -> bytes:
@@ -58,17 +69,31 @@ def from_tar_sdist(tf: TarFile) -> bytes:
         buf.append(f"Requires-Dist: {req}\n")
     for extra in sorted(extras):
         buf.append(f"Provides-Extra: {extra}\n")
+
     return ("".join(buf)).encode("utf-8")
 
 
 def basic_metadata_from_tar_sdist(tf: TarFile) -> BasicMetadata:
     # XXX Why do ZipFile and TarFile not have a common interface ?!
     requires = [f for f in tf.getnames() if f.endswith("/requires.txt")]
     requires.sort(key=len)
-    if not requires:
-        return BasicMetadata((), frozenset())
-
-    fo = tf.extractfile(requires[0])
-    assert fo is not None
-
-    return BasicMetadata.from_sdist_pkg_info_and_requires(b"", fo.read())
+    if requires:
+        requires_fo = tf.extractfile(requires[0])
+        assert requires_fo is not None
+        requires_data = requires_fo.read()
+    else:
+        requires_data = b""
+
+    # Find the PKG-INFO file with the shortest path. This is to avoid picking up
+    # a PKG-INFO file from a nested test directory.
+    pkg_info = sorted(
+        (f for f in tf.getnames() if f == "PKG-INFO" or f.endswith("/PKG-INFO")),
+        key=len,
+    )[0]
+
+    pkg_info_fo = tf.extractfile(pkg_info)
+    assert pkg_info_fo is not None
+
+    return BasicMetadata.from_sdist_pkg_info_and_requires(
+        pkg_info_fo.read(), requires_data
+    )

metadata_please/tests/metadata_contents.py

@@ -0,0 +1,15 @@
+METADATA_CONTENTS = b"""\
+Requires-Dist: foo
+Version: 1.2.58
+Summary: Some Summary
+Home-page: http://example.com
+Author: Chicken
+Author-email: [email protected]
+Keywords: farm,animals
+Requires-Python: >=3.6
+Description-Content-Type: text/markdown
+
+# Foo
+
+A very important package.
+"""

metadata_please/tests/sdist.py

@@ -8,13 +8,15 @@
 )
 from ._tar import MemoryTarFile
 from ._zip import MemoryZipFile
+from .metadata_contents import METADATA_CONTENTS
 
 
 class ZipSdistTest(unittest.TestCase):
     def test_requires_as_expected(self) -> None:
         z = MemoryZipFile(
             {
                 "foo/__init__.py": b"",
+                "foo.egg-info/PKG-INFO": b"\n",
                 "foo.egg-info/requires.txt": b"""\
 a
 [e]
@@ -36,6 +38,7 @@ def test_basic_metadata(self) -> None:
         z = MemoryZipFile(
             {
                 "foo/__init__.py": b"",
+                "foo.egg-info/PKG-INFO": b"\n",
                 "foo.egg-info/requires.txt": b"""\
 a
 [e]
@@ -68,6 +71,7 @@ def test_basic_metadata_absl_py_09(self) -> None:
         z = MemoryZipFile(
             {
                 "foo/__init__.py": b"",
+                "foo.egg-info/PKG-INFO": b"\n",
                 "foo.egg-info/requires.txt": b"""\
 six
 
@@ -90,11 +94,32 @@ def test_basic_metadata_absl_py_09(self) -> None:
         )
         self.assertEqual({"test"}, bm.provides_extra)
 
+    def test_basic_metadata_fields(self) -> None:
+        """
+        Modern setuptools will drop a PKG-INFO file in a sdist that is very similar to the METADATA file in a wheel.
+        """
+        z = MemoryZipFile(
+            {
+                "foo/__init__.py": b"",
+                "PKG-INFO": METADATA_CONTENTS,
+            }
+        )
+        bm = basic_metadata_from_zip_sdist(z)  # type: ignore
+        self.assertEqual(["foo"], bm.reqs)
+        self.assertEqual("1.2.58", bm.version)
+        self.assertEqual("Some Summary", bm.summary)
+        self.assertEqual("http://example.com", bm.url)
+        self.assertEqual("Chicken", bm.author)
+        self.assertEqual("[email protected]", bm.author_email)
+        self.assertEqual("farm,animals", bm.keywords)
+        self.assertEqual("text/markdown", bm.long_description_content_type)
+        self.assertEqual("# Foo\n\nA very important package.\n", bm.description)
+
 
 class TarSdistTest(unittest.TestCase):
     def test_requires_as_expected(self) -> None:
         t = MemoryTarFile(
-            ["foo.egg-info/requires.txt", "foo/__init__.py"],
+            ["foo.egg-info/PKG-INFO", "foo.egg-info/requires.txt", "foo/__init__.py"],
             read_value=b"""\
 a
 [e]
@@ -113,7 +138,7 @@ def test_requires_as_expected(self) -> None:
 
     def test_basic_metadata(self) -> None:
         t = MemoryTarFile(
-            ["foo.egg-info/requires.txt", "foo/__init__.py"],
+            ["foo.egg-info/PKG-INFO", "foo.egg-info/requires.txt", "foo/__init__.py"],
             read_value=b"""\
 a
 [e]
@@ -126,3 +151,18 @@ def test_basic_metadata(self) -> None:
             bm.reqs,
         )
         self.assertEqual({"e"}, bm.provides_extra)
+
+    def test_metadata_fields_from_tar_sdist(self) -> None:
+        t = MemoryTarFile(
+            ["PKG-INFO", "foo/__init__.py"],
+            read_value=METADATA_CONTENTS,
+        )
+        bm = basic_metadata_from_tar_sdist(t)  # type: ignore
+        self.assertEqual("1.2.58", bm.version)
+        self.assertEqual("Some Summary", bm.summary)
+        self.assertEqual("http://example.com", bm.url)
+        self.assertEqual("Chicken", bm.author)
+        self.assertEqual("[email protected]", bm.author_email)
+        self.assertEqual("farm,animals", bm.keywords)
+        self.assertEqual("text/markdown", bm.long_description_content_type)
+        self.assertEqual("# Foo\n\nA very important package.\n", bm.description)

metadata_please/tests/wheel.py

@@ -2,6 +2,7 @@
 
 from ..wheel import basic_metadata_from_wheel, from_wheel, InvalidWheel
 from ._zip import MemoryZipFile
+from .metadata_contents import METADATA_CONTENTS
 
 
 class WheelTest(unittest.TestCase):
@@ -58,7 +59,7 @@ def test_basic_metadata(self) -> None:
     def test_basic_metadata_more_fields(self) -> None:
         z = MemoryZipFile(
             {
-                "foo.dist-info/METADATA": b"Requires-Dist: foo\nVersion: 1.2.58\nSummary: Some Summary\nHome-page: http://example.com\nAuthor: Chicken\nAuthor-email: [email protected]\nKeywords: farm,animals\nRequires-Python: >=3.6\nDescription-Content-Type: text/markdown",
+                "foo.dist-info/METADATA": METADATA_CONTENTS,
                 "foo/__init__.py": b"",
             }
         )
@@ -71,4 +72,4 @@ def test_basic_metadata_more_fields(self) -> None:
         self.assertEqual("[email protected]", bm.author_email)
         self.assertEqual("farm,animals", bm.keywords)
         self.assertEqual("text/markdown", bm.long_description_content_type)
-        self.assertEqual(None, bm.description)
+        self.assertEqual("# Foo\n\nA very important package.\n", bm.description)