Add json_metadata

Author: sbidoul

src/pip/_internal/metadata/base.py

@@ -8,8 +8,10 @@
 from typing import (
     IO,
     TYPE_CHECKING,
+    Any,
     Collection,
     Container,
+    Dict,
     Iterable,
     Iterator,
     List,
@@ -35,6 +37,8 @@
 from pip._internal.utils.misc import is_local, normalize_path
 from pip._internal.utils.urls import url_to_path
 
+from .json import msg_to_json
+
 if TYPE_CHECKING:
     from typing import Protocol
 else:
@@ -359,6 +363,17 @@ def metadata(self) -> email.message.Message:
         """
         raise NotImplementedError()
 
+    @property
+    def json_metadata(self) -> Dict[str, Any]:
+        """PEP 566 compliant JSON-serializable representation of METADATA or PKG-INFO.
+
+        This should return an empty dict if the metadata file is unavailable.
+
+        :raises NoneMetadataError: If the metadata file is available, but does
+            not contain valid metadata.
+        """
+        return msg_to_json(self.metadata)
+
     @property
     def metadata_version(self) -> Optional[str]:
         """Value of "Metadata-Version:" in distribution metadata, if available."""

src/pip/_internal/metadata/json.py

@@ -0,0 +1,80 @@
+# Extracted from https://github.com/pfmoore/pkg_metadata
+
+from email.header import Header, decode_header, make_header
+from email.message import Message
+from typing import Any, Dict, List, Union
+
+METADATA_FIELDS = [
+    # Name, Multiple-Use
+    ("Metadata-Version", False),
+    ("Name", False),
+    ("Version", False),
+    ("Dynamic", True),
+    ("Platform", True),
+    ("Supported-Platform", True),
+    ("Summary", False),
+    ("Description", False),
+    ("Description-Content-Type", False),
+    ("Keywords", False),
+    ("Home-page", False),
+    ("Download-URL", False),
+    ("Author", False),
+    ("Author-email", False),
+    ("Maintainer", False),
+    ("Maintainer-email", False),
+    ("License", False),
+    ("Classifier", True),
+    ("Requires-Dist", True),
+    ("Requires-Python", False),
+    ("Requires-External", True),
+    ("Project-URL", True),
+    ("Provides-Extra", True),
+    ("Provides-Dist", True),
+    ("Obsoletes-Dist", True),
+]
+
+
+def json_name(field: str) -> str:
+    return field.lower().replace("-", "_")
+
+
+def msg_to_json(msg: Message) -> Dict[str, Any]:
+    def sanitise_header(h: Union[Header, str]) -> str:
+        if isinstance(h, Header):
+            chunks = []
+            for bytes, encoding in decode_header(h):
+                if encoding == "unknown-8bit":
+                    try:
+                        # See if UTF-8 works
+                        bytes.decode("utf-8")
+                        encoding = "utf-8"
+                    except UnicodeDecodeError:
+                        # If not, latin1 at least won't fail
+                        encoding = "latin1"
+                chunks.append((bytes, encoding))
+            return str(make_header(chunks))
+        return str(h)
+
+    result = {}
+    for field, multi in METADATA_FIELDS:
+        if field not in msg:
+            continue
+        key = json_name(field)
+        if multi:
+            value: Union[str, List[str]] = [
+                sanitise_header(v) for v in msg.get_all(field)
+            ]
+        else:
+            value = sanitise_header(msg.get(field))
+            if key == "keywords":
+                if "," in value:
+                    value = [v.strip() for v in value.split(",")]
+                else:
+                    value = value.split()
+        result[key] = value
+
+    payload = msg.get_payload()
+    if payload:
+        result["description"] = payload
+
+    return result

tests/unit/metadata/test_metadata.py

@@ -1,12 +1,15 @@
 import logging
+from pathlib import Path
 from typing import cast
 from unittest import mock
 
 import pytest
 from pip._vendor.packaging.utils import NormalizedName
 
-from pip._internal.metadata import BaseDistribution
+from pip._internal.metadata import BaseDistribution, get_wheel_distribution
+from pip._internal.metadata.base import FilesystemWheel
 from pip._internal.models.direct_url import DIRECT_URL_METADATA_NAME, ArchiveInfo
+from tests.lib.wheel import make_wheel
 
 
 @mock.patch.object(BaseDistribution, "read_text", side_effect=FileNotFoundError)
@@ -55,3 +58,17 @@ class FakeDistribution(BaseDistribution):
     mock_read_text.assert_called_once_with(DIRECT_URL_METADATA_NAME)
     assert direct_url.url == "https://e.c/p.tgz"
     assert isinstance(direct_url.info, ArchiveInfo)
+
+
+def test_json_metadata(tmpdir: Path) -> None:
+    """Basic test of BaseDistribution json_metadata.
+    
+    More tests are available in the original pkg_metadata project where this
+    function comes from, and which we may vendor in the future.
+    """
+    wheel_path = make_wheel(name="pkga", version="1.0.1").save_to_dir(tmpdir)
+    wheel = FilesystemWheel(wheel_path)
+    dist = get_wheel_distribution(wheel, "pkga")
+    json_metadata = dist.json_metadata
+    assert json_metadata["name"] == "pkga"
+    assert json_metadata["version"] == "1.0.1"