Parse hash data before passing to MetadataFile

Author: pfmoore

src/pip/_internal/models/link.py

@@ -101,12 +101,20 @@ class MetadataFile:
 
     hashes: Optional[Dict[str, str]]
 
-    # TODO: Do we care about stripping out unsupported hash methods?
-    def __init__(self, hashes: Optional[Dict[str, str]]):
-        if hashes:
-            hashes = {n: v for n, v in hashes.items() if n in _SUPPORTED_HASHES}
-        # We need to use this as this is a frozen dataclass
-        object.__setattr__(self, "hashes", hashes)
+    def __post_init__(self) -> None:
+        if self.hashes is not None:
+            assert all(name in _SUPPORTED_HASHES for name in self.hashes)
+
+
+def supported_hashes(hashes: Optional[Dict[str, str]]) -> Optional[Dict[str, str]]:
+    # Remove any unsupported hash types from the mapping. If this leaves no
+    # supported hashes, return None
+    if hashes is None:
+        return None
+    hashes = {n: v for n, v in hashes.items() if n in _SUPPORTED_HASHES}
+    if len(hashes) > 0:
+        return hashes
+    return None
 
 
 def _clean_url_path_part(part: str) -> str:
@@ -273,7 +281,7 @@ def from_json(
         metadata_info = file_data.get("dist-info-metadata", False)
         if isinstance(metadata_info, dict):
             # The file exists, and hashes have been supplied
-            metadata_file_data = MetadataFile(metadata_info)
+            metadata_file_data = MetadataFile(supported_hashes(metadata_info))
         elif metadata_info:
             # The file exists, but there are no hashes
             metadata_file_data = MetadataFile(None)
@@ -328,7 +336,7 @@ def from_element(
             # The file exists, and hashes have been supplied
             hashname, sep, hashval = metadata_info.partition("=")
             if sep == "=":
-                metadata_file_data = MetadataFile({hashname: hashval})
+                metadata_file_data = MetadataFile(supported_hashes({hashname: hashval}))
             else:
                 # Error - data is wrong. Treat as no hashes supplied.
                 logger.debug(

tests/unit/test_collector.py

@@ -1085,10 +1085,10 @@ def test_link_hash_parsing(url: str, result: Optional[LinkHash]) -> None:
     [
         ("sha256=aa113592bbe", MetadataFile({"sha256": "aa113592bbe"})),
         ("sha256=", MetadataFile({"sha256": ""})),
-        ("sha500=aa113592bbe", MetadataFile({})),
+        ("sha500=aa113592bbe", MetadataFile(None)),
         ("true", MetadataFile(None)),
         (None, None),
-        # TODO: Are these correct?
+        # Attribute is present but invalid
         ("", MetadataFile(None)),
         ("aa113592bbe", MetadataFile(None)),
     ],
@@ -1104,4 +1104,3 @@ def test_metadata_file_info_parsing_html(
     base_url = "https://index.url/simple"
     link = Link.from_element(attribs, page_url, base_url)
     assert link is not None and link.metadata_file_data == expected
-    # TODO: Do we need to do something for the JSON data?