fix bug so that we only create an additional directories cache when we have additional directories

Author: kji22

src/licensedcode/cache.py

@@ -81,27 +81,9 @@ def load_or_build(
         cached_directories_file = os.path.join(idx_cache_dir, CACHED_DIRECTORIES_FILENAME)
 
         has_cache = os.path.exists(cache_file) and os.path.getsize(cache_file)
-        has_cached_directories = os.path.exists(cached_directories_file) and os.path.getsize(cached_directories_file)
-        should_rebuild_cache = False
-
-        if has_cached_directories:
-            # if we have cached additional directories of licenses, check if those licenses are equal to the additional
-            # directories passed in
-            with open(cached_directories_file, 'rb') as file:
-                cached_additional_directories = pickle.load(file)
-
-            # we need to rebuild the cache if the list of additional directories we passed in is different than
-            # the set of additional directories current included in the index cache
-            should_rebuild_cache = additional_directories is not None \
-                                   and sorted(additional_directories) != sorted(cached_additional_directories)
-        else:
-            # otherwise, we don't have a file of cached directories. If there are additional directories passed in,
-            # we know we need to make a new cache file.
-            if additional_directories:
-                should_rebuild_cache = True
 
         # bypass build if cache exists
-        if has_cache and not force and not should_rebuild_cache:
+        if has_cache and not force:
             try:
                 return load_cache_file(cache_file)
             except Exception as e:
@@ -163,7 +145,8 @@ def load_or_build(
                 with open(cache_file, 'wb') as fn:
                     pickle.dump(license_cache, fn, protocol=PICKLE_PROTOCOL)
 
-                # save the list of additional directories included in the cache
+                # save the list of additional directories included in the cache, or None if the cache does not
+                # include any additional directories
                 with open(cached_directories_file, 'wb') as file:
                     pickle.dump(additional_directories, file, protocol=PICKLE_PROTOCOL)
 
@@ -376,6 +359,8 @@ def populate_cache(force=False, index_all_languages=False, additional_directorie
     Load or build and cache a LicenseCache. Return None.
     """
     global _LICENSE_CACHE
+    if need_cache_rebuild(additional_directories):
+        force = True
     if force or not _LICENSE_CACHE:
         _LICENSE_CACHE = LicenseCache.load_or_build(
             licensedcode_cache_dir=licensedcode_cache_dir,
@@ -388,6 +373,34 @@ def populate_cache(force=False, index_all_languages=False, additional_directorie
         )
 
 
+def need_cache_rebuild(additional_directories):
+    """
+    Return true if we need to rebuild the index cache.
+    """
+    idx_cache_dir = os.path.join(licensedcode_cache_dir, LICENSE_INDEX_DIR)
+    cached_directories_file = os.path.join(idx_cache_dir, CACHED_DIRECTORIES_FILENAME)
+
+    has_cached_directories = os.path.exists(cached_directories_file)
+    should_rebuild_cache = False
+
+    if has_cached_directories:
+        # if we have cached additional directories of licenses, check if those licenses are equal to the additional
+        # directories passed in
+        with open(cached_directories_file, 'rb') as file:
+            cached_additional_directories = pickle.load(file)
+
+        # we need to rebuild the cache if the list of additional directories we passed in is not a subset of
+        # the set of additional directories currently included in the index cache
+        should_rebuild_cache = additional_directories is not None and cached_additional_directories is not None\
+                               and not set(additional_directories).issubset(set(cached_additional_directories))
+    else:
+        # otherwise, we don't have a file of cached directories. If there are additional directories passed in,
+        # we know we need to make a new cache file.
+        if additional_directories:
+            should_rebuild_cache = True
+    return should_rebuild_cache
+
+
 def load_cache_file(cache_file):
     """
     Return a LicenseCache loaded from ``cache_file``.