Use new key for core metadata reference (#28)

Implements PEP 714, swapping dist-info-metadata for core-metadata Still supports old key in source index responses, and provides old key in HTML responses Tests added for core-metadata keys/attributes
EpicWink · Jun 8, 2023 · caa09de · caa09de
1 parent d20e2e4
commit caa09de
Show file tree

Hide file tree

Showing 9 changed files with 112 additions and 7 deletions.
diff --git a/src/proxpi/_cache.py b/src/proxpi/_cache.py
@@ -101,7 +101,8 @@ def to_json_response(self) -> t.Dict[str, t.Any]:
         if self.requires_python is not None:
             data["requires-python"] = self.requires_python
         if self.dist_info_metadata is not None:
-            data["dist-info-metadata"] = self.dist_info_metadata
+            # PEP 714: only emit new key in JSON
+            data["core-metadata"] = self.dist_info_metadata
         if self.gpg_sig is not None:
             data["gpg-sig"] = self.gpg_sig
         if self.yanked is not None:
@@ -124,11 +125,20 @@ def from_html_element(
     ) -> "File":
         """Construct from HTML API response."""
         url = urllib.parse.urljoin(request_url, el.attrib["href"])
+
+        attributes = {k: v for k, v in el.attrib.items() if k != "href"}
+
+        # PEP 714: accept both core-metadata attributes, and emit both in HTML
+        if "data-core-metadata" in attributes:
+            attributes["data-dist-info-metadata"] = attributes["data-core-metadata"]
+        elif "data-dist-info-metadata" in attributes:
+            attributes["data-core-metadata"] = attributes["data-dist-info-metadata"]
+
         return cls(
             name=el.text,
             url=url,
             fragment=urllib.parse.urlsplit(url).fragment,
-            attributes={k: v for k, v in el.attrib.items() if k != "href"},
+            attributes=attributes,
         )
 
     @property
@@ -141,7 +151,7 @@ def requires_python(self):
 
     @property
     def dist_info_metadata(self):
-        metadata = self.attributes.get("data-dist-info-metadata")
+        metadata = self.attributes.get("data-core-metadata")
         if metadata is None:
             return None
         hashes = self._parse_hash(metadata)
@@ -198,7 +208,10 @@ def from_json_response(cls, data: t.Dict[str, t.Any], request_url: str) -> "File
             url=urllib.parse.urljoin(request_url, data["url"]),
             hashes=data["hashes"],
             requires_python=data.get("requires-python"),
-            dist_info_metadata=data.get("dist-info-metadata"),
+            # PEP 714: accept both core-metadata keys
+            dist_info_metadata=(
+                data.get("core-metadata") or data.get("dist-info-metadata")
+            ),
             gpg_sig=data.get("gpg-sig"),
             yanked=data.get("yanked"),
         )
@@ -218,6 +231,8 @@ def attributes(self) -> t.Dict[str, str]:
             attributes["data-dist-info-metadata"] = self._stringify_hashes(
                 self.dist_info_metadata,
             ) if isinstance(self.dist_info_metadata, dict) else ""  # fmt: skip
+            # PEP 714: emit both core-metadata attributes in HTML
+            attributes["data-core-metadata"] = attributes["data-dist-info-metadata"]
         if self.gpg_sig is not None:
             attributes["data-gpg-sig"] = "true" if self.gpg_sig else "false"
         if self.yanked:

diff --git a/tests/data/indexes/root/numpy/index.html b/tests/data/indexes/root/numpy/index.html
@@ -8,6 +8,8 @@
 </head>
 <body>
     <a href="numpy-1.23.1-cp310-cp310-manylinux_2_17_x86_64.whl#sha256=38d1e325e3d8828d3361730fc935aabb9cc5ab14f7ea99f9438f2768e2c5f272" data-requires-python="&gt;=3.8">numpy-1.23.1-cp310-cp310-manylinux_2_17_x86_64.whl</a><br />
+    <a href="numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl#sha256=6e8d4970bf49f5b1d682dd5d74e1fc383bd462ec973f0a12bfbaae0734dbf53a" data-requires-python="&gt;=3.8" data-core-metadata>numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl</a><br />
+    <a href="numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl#sha256=57fddb18d8862f6240443ea00bf413c3d4b967b50b0d416317dac8a0365138a6" data-requires-python="&gt;=3.8" data-core-metadata="sha256=9e4a576848f4ea0051b63fe503b9357eb7200f9d979eb983a7139478450d2e6a">numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl</a><br />
     <a href="numpy-1.23.1-cp310-cp310-win_amd64.whl#sha256=c82522a5e593eba845234044e3ad5c1d2a666e2be7e872fca8a65287bc72b383" data-requires-python="&gt;=3.8" data-yanked="">numpy-1.23.1-cp310-cp310-win_amd64.whl</a><br />
     <a href="numpy-1.23.1.tar.gz#sha256=0024c61ffaf1d4b424eaff554bb6cec1cfd580cc2ff64ca7858c7ced6b880429" data-requires-python="&gt;=3.8">numpy-1.23.1.tar.gz</a><br />
 </body>

diff --git a/tests/data/indexes/root/numpy/index.json b/tests/data/indexes/root/numpy/index.json
@@ -8,6 +8,26 @@
       "requires-python": ">=3.8",
       "url": "numpy-1.23.1-cp310-cp310-manylinux_2_17_x86_64.whl"
     },
+    {
+      "core-metadata": true,
+      "filename": "numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl",
+      "hashes": {
+        "sha256": "6e8d4970bf49f5b1d682dd5d74e1fc383bd462ec973f0a12bfbaae0734dbf53a"
+      },
+      "requires-python": ">=3.8",
+      "url": "numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl"
+    },
+    {
+      "core-metadata": {
+        "sha256": "9e4a576848f4ea0051b63fe503b9357eb7200f9d979eb983a7139478450d2e6a"
+      },
+      "filename": "numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl",
+      "hashes": {
+        "sha256": "57fddb18d8862f6240443ea00bf413c3d4b967b50b0d416317dac8a0365138a6"
+      },
+      "requires-python": ">=3.8",
+      "url": "numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl"
+    },
     {
       "filename": "numpy-1.23.1-cp310-cp310-win_amd64.whl",
       "hashes": {

diff --git a/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl b/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl
diff --git a/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl.metadata b/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl.metadata
@@ -0,0 +1,3 @@
+Metadata-Version: 2.3
+Name: NumPy
+Version: 1.23.1
diff --git a/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl b/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl
diff --git a/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl.metadata b/tests/data/indexes/root/numpy/numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl.metadata
@@ -0,0 +1,3 @@
+Metadata-Version: 2.3
+Name: NumPy
+Version: 1.23.1
diff --git a/tests/data/indexes/root/numpy/yanked.json b/tests/data/indexes/root/numpy/yanked.json
@@ -9,6 +9,28 @@
       "url": "numpy-1.23.1-cp310-cp310-manylinux_2_17_x86_64.whl",
       "yanked": false
     },
+    {
+      "core-metadata": true,
+      "filename": "numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl",
+      "hashes": {
+        "sha256": "6e8d4970bf49f5b1d682dd5d74e1fc383bd462ec973f0a12bfbaae0734dbf53a"
+      },
+      "requires-python": ">=3.8",
+      "url": "numpy-1.23.1-cp310-cp310-manylinux_2_24_x86_64.whl",
+      "yanked": false
+    },
+    {
+      "core-metadata": {
+        "sha256": "9e4a576848f4ea0051b63fe503b9357eb7200f9d979eb983a7139478450d2e6a"
+      },
+      "filename": "numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl",
+      "hashes": {
+        "sha256": "57fddb18d8862f6240443ea00bf413c3d4b967b50b0d416317dac8a0365138a6"
+      },
+      "requires-python": ">=3.8",
+      "url": "numpy-1.23.1-cp310-cp310-manylinux_2_28_x86_64.whl",
+      "yanked": false
+    },
     {
       "filename": "numpy-1.23.1-cp310-cp310-win_amd64.whl",
       "hashes": {

diff --git a/tests/test_integration.py b/tests/test_integration.py
@@ -241,6 +241,26 @@ def test_package(server, project, accept, index_json_response, clear_projects_ca
             else:
                 assert gpg_response.status_code == 404
 
+        if any(k == "data-dist-info-metadata" for k, _ in attributes):
+            (value,) = (v for k, v in attributes if k == "data-dist-info-metadata")
+            (expected,) = (v for k, v in attributes if k == "data-core-metadata")
+            assert value == expected
+
+        if any(k == "data-core-metadata" for k, _ in attributes):
+            (expected_core_metadata_hash,) = (
+                v for k, v in attributes if k == "data-core-metadata"
+            )
+            core_metadata_response = requests.get(urllib_parse.urljoin(
+                project_url, href_stripped + ".metadata"
+            ))
+            core_metadata_response.raise_for_status()
+            if expected_core_metadata_hash and expected_core_metadata_hash != "true":
+                hash_name, expected_hash_value = expected_core_metadata_hash.split("=")
+                core_metadata_hash_value = hashlib.new(
+                    hash_name, core_metadata_response.content
+                ).hexdigest()
+                assert core_metadata_hash_value == expected_hash_value
+
         if any(k == "data-requires-python" for k, _ in attributes):
             (python_requirement,) = (
                 v for k, v in attributes if k == "data-requires-python"
@@ -279,10 +299,10 @@ def test_package_json(
         params = {"format": accept}
     else:
         headers = {"Accept": accept}
+    project_url = f"{server}/index/{project}/"
+
     with set_mock_index_response_is_json(index_json_response):
-        response = requests.get(
-            f"{server}/index/{project}/", params=params, headers=headers
-        )
+        response = requests.get(project_url, params=params, headers=headers)
 
     assert response.status_code == 200
     assert response.headers["Content-Type"][:35] == (
@@ -301,6 +321,26 @@ def test_package_json(
         assert file["filename"] == file["url"]
         assert isinstance(file["hashes"], dict)
 
+        assert not file.get("dist-info-metadata")
+
+        url_parts: urllib_parse.SplitResult = urllib_parse.urlsplit(file["url"])
+        url_parts_stripped = url_parts._replace(fragment="")
+        url_stripped = url_parts_stripped.geturl()
+        assert url_stripped == file["filename"]
+
+        if file.get("core-metadata"):
+            core_metadata_response = requests.get(
+                urllib_parse.urljoin(project_url, url_stripped + ".metadata"),
+            )
+            core_metadata_response.raise_for_status()
+
+            if isinstance(file["core-metadata"], dict):
+                for hash_name, expected_hash_value in file["core-metadata"].items():
+                    core_metadata_hash_value = hashlib.new(
+                        hash_name, core_metadata_response.content
+                    ).hexdigest()
+                    assert core_metadata_hash_value == expected_hash_value
+
     files_by_filename = {f["filename"]: f for f in response_data["files"]}
     if project == "proxpi":
         assert not files_by_filename["proxpi-1.0.0.tar.gz"].get("requires-python")