fix: match directory/tutorial keywords against the bare repo name too

full_name is "owner/repo", and "/" isn't a [-_\s] word boundary, so
DIRECTORY_WORDS and LEARNING_WORDS could never match a trigger word
sitting at the start of the repo name itself (e.g. "awesome-dsh-plugins"
never matched "awesome" via full_name). This was previously masked
because "directory"/"curated" in the description usually covered for
it; removing those words in the prior commit exposed real awesome-list
repos slipping through. Search the bare repo name in addition to
full_name and description.

Author: aiwarelife

scripts/sync_topic_plugins.py

@@ -604,9 +604,13 @@ def metadata_exclusion(repo: dict[str, Any]) -> tuple[str, list[str]] | None:
         return "excluded", ["inactive_repository"]
     if int(repo.get("size") or 0) == 0:
         return "excluded", ["empty_repository"]
-    if DIRECTORY_WORDS.search(full_name + " " + description):
+    # full_name is "owner/repo"; the slash isn't a [-_\s] boundary, so a
+    # repo named e.g. "awesome-dsh-plugins" would never match against
+    # full_name alone. Search the bare repo name too.
+    evidence_text = " ".join((repo.get("name", ""), full_name, description))
+    if DIRECTORY_WORDS.search(evidence_text):
         return "excluded", ["directory_or_collection"]
-    if LEARNING_WORDS.search(full_name + " " + description):
+    if LEARNING_WORDS.search(evidence_text):
         return "excluded", ["tutorial_or_handbook"]
     return None
 

tests/test_topic_sync.py

@@ -62,6 +62,20 @@ def test_directory_is_excluded(self):
         self.assertEqual(result["status"], "excluded")
         self.assertIn("directory_or_collection", result["reasons"])
 
+    def test_directory_words_match_the_bare_repo_name(self):
+        result = topic_sync.classify(
+            repository(
+                name="awesome-dsh-plugins",
+                full_name="coolbat/awesome-dsh-plugins",
+                description="Browse every DeepSeek Harness plugin on GitHub in one place.",
+            ),
+            "# Awesome DSH plugins",
+            ["README.md"],
+            None,
+        )
+        self.assertEqual(result["status"], "excluded")
+        self.assertIn("directory_or_collection", result["reasons"])
+
     def test_directory_words_do_not_match_unrelated_technical_usage(self):
         result = topic_sync.classify(
             repository(