fix: remove dead null-language scoring config

gittensor/validator/utils/load_weights.py

@@ -99,6 +99,22 @@ def supports_tree_sitter(self, extension: Optional[str]) -> bool:
         config = self.language_configs.get(ext)
         return config is not None and config.language is not None
 
+    def is_line_count_extension(self, extension: Optional[str]) -> bool:
+        """Check if a file extension uses line-count scoring.
+
+        Returns True when the extension is either in NON_CODE_EXTENSIONS or is
+        configured in programming_languages.json with no tree-sitter language
+        (language=None).  Unknown extensions that are absent from the config
+        entirely remain skipped-unsupported and return False.
+        """
+        if not extension:
+            return False
+        ext = extension.lstrip('.').lower()
+        if ext in NON_CODE_EXTENSIONS:
+            return True
+        config = self.language_configs.get(ext)
+        return config is not None and config.language is None
+
 
 def _get_weights_dir() -> Path:
     return Path(__file__).parent.parent / 'weights'

gittensor/validator/utils/tree_sitter_scoring.py

@@ -19,7 +19,6 @@
     INLINE_TEST_PATTERNS,
     MAX_FILE_SIZE_BYTES,
     MAX_LINES_SCORED_FOR_NON_CODE_EXT,
-    NON_CODE_EXTENSIONS,
     TEST_FILE_CONTRIBUTION_WEIGHT,
     TREE_SITTER_PARSE_TIMEOUT_MICROS,
 )
@@ -277,7 +276,7 @@ def calculate_token_score_from_file_changes(
                 is_test_file=is_test_file,
                 scoring_method='skipped',
             )
-        elif ext in NON_CODE_EXTENSIONS:
+        elif weights.is_line_count_extension(ext):
             lines_to_score = min(file.changes, MAX_LINES_SCORED_FOR_NON_CODE_EXT)
             lang_config = programming_languages.get(ext)
             lang_weight = lang_config.weight if lang_config else DEFAULT_PROGRAMMING_LANGUAGE_WEIGHT

tests/validator/test_token_scoring_integration.py

@@ -617,5 +617,93 @@ def test_added_file_null_old_content_scores_as_new_file(self, weights):
         assert file_result.nodes_scored > 0
 
 
+class TestNullLanguageLineCountScoring:
+    """Regression tests for configured null-language extensions.
+
+    Every extension in programming_languages.json that has no ``language``
+    field (no tree-sitter parser) must route to ``line-count`` scoring, not
+    ``skipped-unsupported``.  Unknown extensions that are absent from the
+    config entirely should remain ``skipped-unsupported``.
+    """
+
+    @pytest.fixture
+    def weights(self) -> TokenConfig:
+        return load_token_config()
+
+    @pytest.fixture
+    def prog_langs(self):
+        return load_programming_language_weights()
+
+    def _score_file(self, filename: str, content: str, weights: TokenConfig, prog_langs):
+        fc = FileChange(
+            pr_number=1,
+            repository_full_name='test/repo',
+            filename=filename,
+            changes=content.count('\n') or 1,
+            additions=content.count('\n') or 1,
+            deletions=0,
+            status='added',
+        )
+        result = calculate_token_score_from_file_changes(
+            [fc],
+            {filename: FileContentPair(old_content=None, new_content=content)},
+            weights,
+            prog_langs,
+        )
+        return result.file_results[0]
+
+    def test_graphql_scores_line_count(self, weights, prog_langs):
+        fr = self._score_file('schema.graphql', 'type Query { hello: String }\n', weights, prog_langs)
+        assert fr.scoring_method == 'line-count'
+        assert fr.score > 0.0
+
+    def test_gitignore_scores_line_count(self, weights, prog_langs):
+        fr = self._score_file('.gitignore', 'node_modules/\n*.pyc\n', weights, prog_langs)
+        assert fr.scoring_method == 'line-count'
+        assert fr.score > 0.0
+
+    def test_unknown_extension_stays_skipped_unsupported(self, weights, prog_langs):
+        fr = self._score_file('data.xyz123', 'some content\n', weights, prog_langs)
+        assert fr.scoring_method == 'skipped-unsupported'
+        assert fr.score == 0.0
+
+    def test_graphql_score_uses_configured_weight(self, weights, prog_langs):
+        """graphql weight=1.0; one changed line → score should equal 1.0."""
+        fc = FileChange(
+            pr_number=1,
+            repository_full_name='test/repo',
+            filename='schema.graphql',
+            changes=1,
+            additions=1,
+            deletions=0,
+            status='added',
+        )
+        result = calculate_token_score_from_file_changes(
+            [fc],
+            {'schema.graphql': FileContentPair(old_content=None, new_content='type Query { hello: String }\n')},
+            weights,
+            prog_langs,
+        )
+        fr = result.file_results[0]
+        assert fr.scoring_method == 'line-count'
+        assert fr.score == pytest.approx(1.0, abs=1e-6)
+
+    def test_all_configured_null_language_extensions_are_line_count_reachable(self, weights, prog_langs):
+        """Config-coverage guard: every programming_languages.json entry with language=None
+        must route to line-count, not skipped-unsupported.  This catches future drift
+        where a new null-language weight is added to the JSON but the scorer silently
+        ignores it."""
+        null_lang_exts = [ext for ext, cfg in prog_langs.items() if cfg.language is None]
+        assert null_lang_exts, 'Expected at least one null-language extension in programming_languages.json'
+
+        for ext in null_lang_exts:
+            filename = f'testfile.{ext}'
+            fr = self._score_file(filename, 'line one\nline two\n', weights, prog_langs)
+            assert fr.scoring_method == 'line-count', (
+                f'Extension .{ext} is configured in programming_languages.json with language=None '
+                f'but scored as {fr.scoring_method!r} instead of line-count'
+            )
+
+
 if __name__ == '__main__':
     pytest.main([__file__, '-v'])