Extractor: normalize path when checking against overlay changed files

Author: d10c

python/extractor/semmle/traverser.py

@@ -40,7 +40,7 @@ def __init__(self, options, modulenames, logger):
                 with open(overlay_changes_file, 'r', encoding='utf-8') as f:
                     data = json.load(f)
                     changed_paths = data.get('changes', [])
-                    self.overlay_changes = { os.path.abspath(p) for p in changed_paths }
+                    self.overlay_changes = { os.path.normcase(os.path.normpath(os.path.abspath(p))) for p in changed_paths }
             except (IOError, ValueError) as e:
                 logger.warn("Failed to read overlay changes from '%s' (falling back to full extraction): %s", overlay_changes_file, e)
                 self.overlay_changes = None
@@ -62,18 +62,18 @@ def __iter__(self):
                 if mod is None:
                     self.logger.error("No module named '%s'.", name)
                     raise ExtractorFailure()
-                if self.overlay_changes is not None and mod.path not in self.overlay_changes:
+                if not self._changed_in_overlay(mod.path):
                     self.logger.debug("Skipping module '%s' as it was not changed in overlay extraction.", name)
                     continue
                 yield mod.get_extractable()
         for path in self.paths:
-            if self.overlay_changes is not None and path not in self.overlay_changes:
+            if not self._changed_in_overlay(path):
                 self.logger.debug("Skipping path '%s' as it was not changed in overlay extraction.", path)
                 continue
             yield Extractable.from_path(path)
         for path in self.recurse_files:
             for modpath in self._treewalk(path):
-                if self.overlay_changes is not None and modpath not in self.overlay_changes:
+                if not self._changed_in_overlay(modpath):
                     self.logger.debug("Skipping file '%s' as it was not changed in overlay extraction.", modpath)
                     continue
                 yield Extractable.from_path(modpath)
@@ -89,11 +89,15 @@ def __iter__(self):
                 self.logger.error("Package '%s' does not have a path.", name)
                 raise ExtractorFailure()
             for modpath in self._treewalk(path):
-                if self.overlay_changes is not None and modpath not in self.overlay_changes:
+                if not self._changed_in_overlay(modpath):
                     self.logger.debug("Skipping package '%s' as it was not changed in overlay extraction.", modpath)
                     continue
                 yield Extractable.from_path(modpath)
 
+    def _changed_in_overlay(self, path):
+        return self.overlay_changes is None or \
+            os.path.normcase(os.path.normpath(os.path.abspath(path))) in self.overlay_changes
+
     def _treewalk(self, path):
         '''Recursively walk the directory tree, skipping sym-links and
         hidden files and directories.'''