speed up cache by approximately 42x by avoiding pathlib (#1953)

diff --git a/CHANGES.md b/CHANGES.md
index ca8a0472a3bbc6ccf2d55fe3e4851b256140b7f8..5f7ca4f4da7825f4b04794f0c32e40b6dbcae225 100644 (file)
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -26,6 +26,8 @@
 
 - use lowercase hex strings (#1692)
 
 
 - use lowercase hex strings (#1692)
 

+- speed up caching by avoiding pathlib (#1950)
+

 #### _Packaging_
 
 - Self-contained native _Black_ binaries are now provided for releases via GitHub
 #### _Packaging_
 
 - Self-contained native _Black_ binaries are now provided for releases via GitHub

diff --git a/src/black/__init__.py b/src/black/__init__.py
index 9034bf6cf77226880e7e2fc05fd8f95a3a647573..7c1a013aae3663949767567e356acdb586a04e3d 100644 (file)
--- a/src/black/__init__.py
+++ b/src/black/__init__.py
@@ -93,7 +93,7 @@ Transformer = Callable[["Line", Collection["Feature"]], Iterator["Line"]]
 Timestamp = float
 FileSize = int
 CacheInfo = Tuple[Timestamp, FileSize]
 Timestamp = float
 FileSize = int
 CacheInfo = Tuple[Timestamp, FileSize]

-Cache = Dict[Path, CacheInfo]
+Cache = Dict[str, CacheInfo]

 out = partial(click.secho, bold=True, err=True)
 err = partial(click.secho, fg="red", err=True)
 
 out = partial(click.secho, bold=True, err=True)
 err = partial(click.secho, fg="red", err=True)
 


@@ -724,7 +724,8 @@ def reformat_one(
             if write_back not in (WriteBack.DIFF, WriteBack.COLOR_DIFF):
                 cache = read_cache(mode)
                 res_src = src.resolve()
             if write_back not in (WriteBack.DIFF, WriteBack.COLOR_DIFF):
                 cache = read_cache(mode)
                 res_src = src.resolve()

-                if res_src in cache and cache[res_src] == get_cache_info(res_src):
+                res_src_s = str(res_src)
+                if res_src_s in cache and cache[res_src_s] == get_cache_info(res_src):

                     changed = Changed.CACHED
             if changed is not Changed.CACHED and format_file_in_place(
                 src, fast=fast, write_back=write_back, mode=mode
                     changed = Changed.CACHED
             if changed is not Changed.CACHED and format_file_in_place(
                 src, fast=fast, write_back=write_back, mode=mode


@@ -6781,8 +6782,8 @@ def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set
     """
     todo, done = set(), set()
     for src in sources:
     """
     todo, done = set(), set()
     for src in sources:

-        src = src.resolve()
-        if cache.get(src) != get_cache_info(src):
+        res_src = src.resolve()
+        if cache.get(str(res_src)) != get_cache_info(res_src):

             todo.add(src)
         else:
             done.add(src)
             todo.add(src)
         else:
             done.add(src)


@@ -6794,7 +6795,10 @@ def write_cache(cache: Cache, sources: Iterable[Path], mode: Mode) -> None:
     cache_file = get_cache_file(mode)
     try:
         CACHE_DIR.mkdir(parents=True, exist_ok=True)
     cache_file = get_cache_file(mode)
     try:
         CACHE_DIR.mkdir(parents=True, exist_ok=True)

-        new_cache = {**cache, **{src.resolve(): get_cache_info(src) for src in sources}}
+        new_cache = {
+            **cache,
+            **{str(src.resolve()): get_cache_info(src) for src in sources},
+        }

         with tempfile.NamedTemporaryFile(dir=str(cache_file.parent), delete=False) as f:
             pickle.dump(new_cache, f, protocol=4)
         os.replace(f.name, cache_file)
         with tempfile.NamedTemporaryFile(dir=str(cache_file.parent), delete=False) as f:
             pickle.dump(new_cache, f, protocol=4)
         os.replace(f.name, cache_file)

diff --git a/tests/test_black.py b/tests/test_black.py
index 28b75787bd6593d630c0812adfb2bbda340a6bc8..cfd3cbd91b3d59056eaa285557d0a38e9564399b 100644 (file)
--- a/tests/test_black.py
+++ b/tests/test_black.py
@@ -1003,7 +1003,7 @@ class BlackTestCase(BlackBaseTestCase):
                 fobj.write("print('hello')")
             self.invokeBlack([str(src)])
             cache = black.read_cache(mode)
                 fobj.write("print('hello')")
             self.invokeBlack([str(src)])
             cache = black.read_cache(mode)

-            self.assertIn(src, cache)
+            self.assertIn(str(src), cache)

 
     def test_cache_single_file_already_cached(self) -> None:
         mode = DEFAULT_MODE
 
     def test_cache_single_file_already_cached(self) -> None:
         mode = DEFAULT_MODE


@@ -1035,8 +1035,8 @@ class BlackTestCase(BlackBaseTestCase):
             with two.open("r") as fobj:
                 self.assertEqual(fobj.read(), 'print("hello")\n')
             cache = black.read_cache(mode)
             with two.open("r") as fobj:
                 self.assertEqual(fobj.read(), 'print("hello")\n')
             cache = black.read_cache(mode)

-            self.assertIn(one, cache)
-            self.assertIn(two, cache)
+            self.assertIn(str(one), cache)
+            self.assertIn(str(two), cache)

 
     def test_no_cache_when_writeback_diff(self) -> None:
         mode = DEFAULT_MODE
 
     def test_no_cache_when_writeback_diff(self) -> None:
         mode = DEFAULT_MODE


@@ -1116,8 +1116,8 @@ class BlackTestCase(BlackBaseTestCase):
             src.touch()
             black.write_cache({}, [src], mode)
             cache = black.read_cache(mode)
             src.touch()
             black.write_cache({}, [src], mode)
             cache = black.read_cache(mode)

-            self.assertIn(src, cache)
-            self.assertEqual(cache[src], black.get_cache_info(src))
+            self.assertIn(str(src), cache)
+            self.assertEqual(cache[str(src)], black.get_cache_info(src))

 
     def test_filter_cached(self) -> None:
         with TemporaryDirectory() as workspace:
 
     def test_filter_cached(self) -> None:
         with TemporaryDirectory() as workspace:


@@ -1128,7 +1128,10 @@ class BlackTestCase(BlackBaseTestCase):
             uncached.touch()
             cached.touch()
             cached_but_changed.touch()
             uncached.touch()
             cached.touch()
             cached_but_changed.touch()

-            cache = {cached: black.get_cache_info(cached), cached_but_changed: (0.0, 0)}
+            cache = {
+                str(cached): black.get_cache_info(cached),
+                str(cached_but_changed): (0.0, 0),
+            }

             todo, done = black.filter_cached(
                 cache, {uncached, cached, cached_but_changed}
             )
             todo, done = black.filter_cached(
                 cache, {uncached, cached, cached_but_changed}
             )


@@ -1156,8 +1159,8 @@ class BlackTestCase(BlackBaseTestCase):
                 fobj.write('print("hello")\n')
             self.invokeBlack([str(workspace)], exit_code=123)
             cache = black.read_cache(mode)
                 fobj.write('print("hello")\n')
             self.invokeBlack([str(workspace)], exit_code=123)
             cache = black.read_cache(mode)

-            self.assertNotIn(failing, cache)
-            self.assertIn(clean, cache)
+            self.assertNotIn(str(failing), cache)
+            self.assertIn(str(clean), cache)

 
     def test_write_cache_write_fail(self) -> None:
         mode = DEFAULT_MODE
 
     def test_write_cache_write_fail(self) -> None:
         mode = DEFAULT_MODE


@@ -1210,9 +1213,9 @@ class BlackTestCase(BlackBaseTestCase):
             path.touch()
             black.write_cache({}, [path], mode)
             one = black.read_cache(mode)
             path.touch()
             black.write_cache({}, [path], mode)
             one = black.read_cache(mode)

-            self.assertIn(path, one)
+            self.assertIn(str(path), one)

             two = black.read_cache(short_mode)
             two = black.read_cache(short_mode)

-            self.assertNotIn(path, two)
+            self.assertNotIn(str(path), two)

 
     def test_single_file_force_pyi(self) -> None:
         pyi_mode = replace(DEFAULT_MODE, is_pyi=True)
 
     def test_single_file_force_pyi(self) -> None:
         pyi_mode = replace(DEFAULT_MODE, is_pyi=True)


@@ -1226,9 +1229,9 @@ class BlackTestCase(BlackBaseTestCase):
                 actual = fh.read()
             # verify cache with --pyi is separate
             pyi_cache = black.read_cache(pyi_mode)
                 actual = fh.read()
             # verify cache with --pyi is separate
             pyi_cache = black.read_cache(pyi_mode)

-            self.assertIn(path, pyi_cache)
+            self.assertIn(str(path), pyi_cache)

             normal_cache = black.read_cache(DEFAULT_MODE)
             normal_cache = black.read_cache(DEFAULT_MODE)

-            self.assertNotIn(path, normal_cache)
+            self.assertNotIn(str(path), normal_cache)

         self.assertFormatEqual(expected, actual)
         black.assert_equivalent(contents, actual)
         black.assert_stable(contents, actual, pyi_mode)
         self.assertFormatEqual(expected, actual)
         black.assert_equivalent(contents, actual)
         black.assert_stable(contents, actual, pyi_mode)


@@ -1255,8 +1258,8 @@ class BlackTestCase(BlackBaseTestCase):
             pyi_cache = black.read_cache(pyi_mode)
             normal_cache = black.read_cache(reg_mode)
             for path in paths:
             pyi_cache = black.read_cache(pyi_mode)
             normal_cache = black.read_cache(reg_mode)
             for path in paths:

-                self.assertIn(path, pyi_cache)
-                self.assertNotIn(path, normal_cache)
+                self.assertIn(str(path), pyi_cache)
+                self.assertNotIn(str(path), normal_cache)

 
     def test_pipe_force_pyi(self) -> None:
         source, expected = read_data("force_pyi")
 
     def test_pipe_force_pyi(self) -> None:
         source, expected = read_data("force_pyi")


@@ -1280,9 +1283,9 @@ class BlackTestCase(BlackBaseTestCase):
                 actual = fh.read()
             # verify cache with --target-version is separate
             py36_cache = black.read_cache(py36_mode)
                 actual = fh.read()
             # verify cache with --target-version is separate
             py36_cache = black.read_cache(py36_mode)

-            self.assertIn(path, py36_cache)
+            self.assertIn(str(path), py36_cache)

             normal_cache = black.read_cache(reg_mode)
             normal_cache = black.read_cache(reg_mode)

-            self.assertNotIn(path, normal_cache)
+            self.assertNotIn(str(path), normal_cache)

         self.assertEqual(actual, expected)
 
     @event_loop()
         self.assertEqual(actual, expected)
 
     @event_loop()


@@ -1307,8 +1310,8 @@ class BlackTestCase(BlackBaseTestCase):
             pyi_cache = black.read_cache(py36_mode)
             normal_cache = black.read_cache(reg_mode)
             for path in paths:
             pyi_cache = black.read_cache(py36_mode)
             normal_cache = black.read_cache(reg_mode)
             for path in paths:

-                self.assertIn(path, pyi_cache)
-                self.assertNotIn(path, normal_cache)
+                self.assertIn(str(path), pyi_cache)
+                self.assertNotIn(str(path), normal_cache)

 
     def test_pipe_force_py36(self) -> None:
         source, expected = read_data("force_py36")
 
     def test_pipe_force_py36(self) -> None:
         source, expected = read_data("force_py36")