Preserve extension on cached refs

Rather than pure MD5, capture the extension used. This allows for
`.json5` and `.yaml` files, which indicate filetype to parsing.

Author: sirosen

src/check_jsonschema/schema_loader/resolver.py

@@ -12,6 +12,21 @@
 from ..utils import filename2path
 
 
+def ref_url_to_cache_filename(ref_url: str) -> str:
+    """
+    Given a $ref URL, convert it to the filename in the refs/ cache dir.
+    Rules are as follows:
+    - the base filename is an md5 hash of the URL
+    - if the filename ends in an extension (.json, .yaml, etc) that extension
+      is appended to the hash
+    """
+    filename = hashlib.md5(ref_url.encode()).hexdigest()
+    if "." in (last_part := ref_url.rpartition("/")[-1]):
+        _, _, extension = last_part.rpartition(".")
+        filename = f"{filename}.{extension}"
+    return filename
+
+
 def make_reference_registry(
     parsers: ParserSet, retrieval_uri: str | None, schema: dict, disable_cache: bool
 ) -> referencing.Registry:
@@ -75,7 +90,7 @@ def validation_callback(content: bytes) -> None:
 
             bound_downloader = downloader.bind(
                 full_uri,
-                hashlib.md5(full_uri.encode()).hexdigest(),
+                ref_url_to_cache_filename(full_uri),
                 validation_callback,
             )
             with bound_downloader.open() as fp:

tests/acceptance/test_remote_ref_resolution.py

@@ -1,9 +1,10 @@
-import hashlib
 import json
 
 import pytest
 import responses
 
+from check_jsonschema.schema_loader.resolver import ref_url_to_cache_filename
+
 CASES = {
     "case1": {
         "main_schema": {
@@ -36,10 +37,6 @@
 }
 
 
-def _md5(s):
-    return hashlib.md5(s.encode()).hexdigest()
-
-
 @pytest.mark.parametrize("check_passes", (True, False))
 @pytest.mark.parametrize("casename", ("case1", "case2"))
 def test_remote_ref_resolution_simple_case(run_line, check_passes, casename, tmp_path):
@@ -95,7 +92,9 @@ def test_remote_ref_resolution_cache_control(
 
     cache_locs = []
     for ref_loc in ref_locs:
-        cache_locs.append(cache_dir / "check_jsonschema" / "refs" / _md5(ref_loc))
+        cache_locs.append(
+            cache_dir / "check_jsonschema" / "refs" / ref_url_to_cache_filename(ref_loc)
+        )
     assert cache_locs  # sanity check
     if disable_cache:
         for loc in cache_locs:
@@ -126,7 +125,7 @@ def test_remote_ref_resolution_loads_from_cache(
         ref_locs.append(other_schema_loc)
 
         # but populate the cache with "good data"
-        cache_loc = ref_cache_dir / _md5(other_schema_loc)
+        cache_loc = ref_cache_dir / ref_url_to_cache_filename(other_schema_loc)
         cache_locs.append(cache_loc)
         cache_loc.write_text(json.dumps(subschema))