Add FileIO to make it possible to cache e.g. files from zip files

parso/cache.py

@@ -78,21 +78,26 @@ class _NodeCacheItem(object):
         self.change_time = change_time
 
 
-def load_module(hashed_grammar, path, cache_path=None):
+def load_module(hashed_grammar, file_io, cache_path=None):
     """
     Returns a module or None, if it fails.
     """
     try:
-        p_time = os.path.getmtime(path)
+        p_time = file_io.get_last_modified()
     except FileNotFoundError:
         return None
 
     try:
-        module_cache_item = parser_cache[hashed_grammar][path]
+        module_cache_item = parser_cache[hashed_grammar][file_io.path]
         if p_time <= module_cache_item.change_time:
             return module_cache_item.node
     except KeyError:
-        return _load_from_file_system(hashed_grammar, path, p_time, cache_path=cache_path)
+        return _load_from_file_system(
+            hashed_grammar,
+            file_io.path,
+            p_time,
+            cache_path=cache_path
+        )
 
 
 def _load_from_file_system(hashed_grammar, path, p_time, cache_path=None):
@@ -123,9 +128,10 @@ def _load_from_file_system(hashed_grammar, path, p_time, cache_path=None):
         return module_cache_item.node
 
 
-def save_module(hashed_grammar, path, module, lines, pickling=True, cache_path=None):
+def save_module(hashed_grammar, file_io, module, lines, pickling=True, cache_path=None):
+    path = file_io.path
     try:
-        p_time = None if path is None else os.path.getmtime(path)
+        p_time = None if path is None else file_io.get_last_modified()
     except OSError:
         p_time = None
         pickling = False

parso/file_io.py

@@ -0,0 +1,26 @@
+import os
+
+
+class FileIO:
+    def __init__(self, path):
+        self.path = path
+
+    def read(self):
+        with open(self.path) as f:
+            return f.read()
+
+    def get_last_modified(self):
+        """
+        Returns float - timestamp
+        Might raise FileNotFoundError
+        """
+        return os.path.getmtime(self.path)
+
+
+class KnownContentFileIO(FileIO):
+    def __init__(self, path, content):
+        super(KnownContentFileIO, self).__init__(path)
+        self._content = content
+
+    def read(self):
+        return self._content

parso/grammar.py

@@ -12,6 +12,7 @@ from parso.parser import BaseParser
 from parso.python.parser import Parser as PythonParser
 from parso.python.errors import ErrorFinderConfig
 from parso.python import pep8
+from parso.file_io import FileIO, KnownContentFileIO
 
 _loaded_grammars = {}
 
@@ -77,14 +78,14 @@ class Grammar(object):
 
     def _parse(self, code=None, error_recovery=True, path=None,
                start_symbol=None, cache=False, diff_cache=False,
-               cache_path=None, start_pos=(1, 0)):
+               cache_path=None, file_io=None, start_pos=(1, 0)):
         """
         Wanted python3.5 * operator and keyword only arguments. Therefore just
         wrap it all.
         start_pos here is just a parameter internally used. Might be public
         sometime in the future.
         """
-        if code is None and path is None:
+        if code is None and path is None and file_io is None:
             raise TypeError("Please provide either code or a path.")
 
         if start_symbol is None:
@@ -93,15 +94,18 @@ class Grammar(object):
         if error_recovery and start_symbol != 'file_input':
             raise NotImplementedError("This is currently not implemented.")
 
-        if cache and path is not None:
-            module_node = load_module(self._hashed, path, cache_path=cache_path)
+        if file_io is None:
+            if code is None:
+                file_io = FileIO(path)
+            else:
+                file_io = KnownContentFileIO(path, code)
+
+        if cache and file_io.path is not None:
+            module_node = load_module(self._hashed, file_io, cache_path=cache_path)
             if module_node is not None:
                 return module_node
 
-        if code is None:
-            with open(path, 'rb') as f:
-                code = f.read()
-
+        code = file_io.read()
         code = python_bytes_to_unicode(code)
 
         lines = split_lines(code, keepends=True)
@@ -110,7 +114,7 @@ class Grammar(object):
                 raise TypeError("You have to define a diff parser to be able "
                                 "to use this option.")
             try:
-                module_cache_item = parser_cache[self._hashed][path]
+                module_cache_item = parser_cache[self._hashed][file_io.path]
             except KeyError:
                 pass
             else:
@@ -125,7 +129,7 @@ class Grammar(object):
                     old_lines=old_lines,
                     new_lines=lines
                 )
-                save_module(self._hashed, path, new_node, lines,
+                save_module(self._hashed, file_io, new_node, lines,
                             # Never pickle in pypy, it's slow as hell.
                             pickling=cache and not is_pypy,
                             cache_path=cache_path)
@@ -141,7 +145,7 @@ class Grammar(object):
         root_node = p.parse(tokens=tokens)
 
         if cache or diff_cache:
-            save_module(self._hashed, path, root_node, lines,
+            save_module(self._hashed, file_io, root_node, lines,
                         # Never pickle in pypy, it's slow as hell.
                         pickling=cache and not is_pypy,
                         cache_path=cache_path)

test/test_cache.py

@@ -10,6 +10,7 @@ from parso.cache import _NodeCacheItem, save_module, load_module, \
     _get_hashed_path, parser_cache, _load_from_file_system, _save_to_file_system
 from parso import load_grammar
 from parso import cache
+from parso import file_io
 
 
 @pytest.fixture()
@@ -76,12 +77,13 @@ def test_modulepickling_simulate_deleted_cache(tmpdir):
     path = tmpdir.dirname + '/some_path'
     with open(path, 'w'):
         pass
+    io = file_io.FileIO(path)
 
-    save_module(grammar._hashed, path, module, [])
-    assert load_module(grammar._hashed, path) == module
+    save_module(grammar._hashed, io, module, lines=[])
+    assert load_module(grammar._hashed, io) == module
 
     unlink(_get_hashed_path(grammar._hashed, path))
     parser_cache.clear()
 
-    cached2 = load_module(grammar._hashed, path)
+    cached2 = load_module(grammar._hashed, io)
     assert cached2 is None