Fix zipimport invalidate caches

Author: desmondcheongzx

Lib/test/test_zipimport.py

@@ -520,10 +520,10 @@ def testInvalidateCaches(self):
                 z.writestr(zinfo, data)
 
         zi = zipimport.zipimporter(TEMP_ZIP)
-        self.assertEqual(zi._files.keys(), files.keys())
+        self.assertEqual(zi._get_files().keys(), files.keys())
         # Check that the file information remains accurate after reloading
         zi.invalidate_caches()
-        self.assertEqual(zi._files.keys(), files.keys())
+        self.assertEqual(zi._get_files().keys(), files.keys())
         # Add a new file to the ZIP archive
         newfile = {"spam2" + pyc_ext: (NOW, test_pyc)}
         files.update(newfile)
@@ -535,14 +535,14 @@ def testInvalidateCaches(self):
                 z.writestr(zinfo, data)
         # Check that we can detect the new file after invalidating the cache
         zi.invalidate_caches()
-        self.assertEqual(zi._files.keys(), files.keys())
+        self.assertEqual(zi._get_files().keys(), files.keys())
         spec = zi.find_spec('spam2')
         self.assertIsNotNone(spec)
         self.assertIsInstance(spec.loader, zipimport.zipimporter)
         # Check that the cached data is removed if the file is deleted
         os.remove(TEMP_ZIP)
         zi.invalidate_caches()
-        self.assertFalse(zi._files)
+        self.assertFalse(zi._get_files())
         self.assertIsNone(zipimport._zip_directory_cache.get(zi.archive))
         self.assertIsNone(zi.find_spec("name_does_not_matter"))
 

Lib/zipimport.py

@@ -93,7 +93,7 @@ def __init__(self, path):
         except KeyError:
             files = _read_directory(path)
             _zip_directory_cache[path] = files
-        self._files = files
+        self._cache_is_valid = True
         self.archive = path
         # a prefix directory following the ZIP file path.
         self.prefix = _bootstrap_external._path_join(*prefix[::-1])
@@ -152,7 +152,7 @@ def get_data(self, pathname):
             key = pathname[len(self.archive + path_sep):]
 
         try:
-            toc_entry = self._files[key]
+            toc_entry = self._get_files()[key]
         except KeyError:
             raise OSError(0, '', key)
         return _get_data(self.archive, toc_entry)
@@ -189,7 +189,7 @@ def get_source(self, fullname):
             fullpath = f'{path}.py'
 
         try:
-            toc_entry = self._files[fullpath]
+            toc_entry = self._get_files()[fullpath]
         except KeyError:
             # we have the module, but no source
             return None
@@ -268,14 +268,26 @@ def get_resource_reader(self, fullname):
         return ZipReader(self, fullname)
 
 
-    def invalidate_caches(self):
-        """Reload the file data of the archive path."""
+    def _get_files(self):
+        """Return the files within the archive path."""
+        if not self._cache_is_valid:
+            try:
+                _zip_directory_cache[self.archive] = _read_directory(self.archive)
+            except ZipImportError:
+                _zip_directory_cache.pop(self.archive, None)
+            self._cache_is_valid = True
+
         try:
-            self._files = _read_directory(self.archive)
-            _zip_directory_cache[self.archive] = self._files
-        except ZipImportError:
-            _zip_directory_cache.pop(self.archive, None)
-            self._files = {}
+            files = _zip_directory_cache[self.archive]
+        except KeyError:
+            files = {}
+
+        return files
+
+
+    def invalidate_caches(self):
+        """Invalidates the cache of file data of the archive path."""
+        self._cache_is_valid = False
 
 
     def __repr__(self):
@@ -305,15 +317,15 @@ def _is_dir(self, path):
     # of a namespace package. We test by seeing if the name, with an
     # appended path separator, exists.
     dirpath = path + path_sep
-    # If dirpath is present in self._files, we have a directory.
-    return dirpath in self._files
+    # If dirpath is present in self._get_files(), we have a directory.
+    return dirpath in self._get_files()
 
 # Return some information about a module.
 def _get_module_info(self, fullname):
     path = _get_module_path(self, fullname)
     for suffix, isbytecode, ispackage in _zip_searchorder:
         fullpath = path + suffix
-        if fullpath in self._files:
+        if fullpath in self._get_files():
             return ispackage
     return None
 
@@ -656,7 +668,7 @@ def _get_mtime_and_size_of_source(self, path):
         # strip 'c' or 'o' from *.py[co]
         assert path[-1:] in ('c', 'o')
         path = path[:-1]
-        toc_entry = self._files[path]
+        toc_entry = self._get_files()[path]
         # fetch the time stamp of the .py file for comparison
         # with an embedded pyc time stamp
         time = toc_entry[5]
@@ -676,7 +688,7 @@ def _get_pyc_source(self, path):
     path = path[:-1]
 
     try:
-        toc_entry = self._files[path]
+        toc_entry = self._get_files()[path]
     except KeyError:
         return None
     else:
@@ -692,7 +704,7 @@ def _get_module_code(self, fullname):
         fullpath = path + suffix
         _bootstrap._verbose_message('trying {}{}{}', self.archive, path_sep, fullpath, verbosity=2)
         try:
-            toc_entry = self._files[fullpath]
+            toc_entry = self._get_files()[fullpath]
         except KeyError:
             pass
         else: