support 'diff-only' hooks by passing diff via stdin (#1279)

pre_commit/clientlib.py

@@ -67,6 +67,7 @@ MANIFEST_HOOK_DICT = cfgv.Map(
     cfgv.Optional('args', cfgv.check_array(cfgv.check_string), []),
     cfgv.Optional('always_run', cfgv.check_bool, False),
     cfgv.Optional('pass_filenames', cfgv.check_bool, True),
+    cfgv.Optional('pass_diff', cfgv.check_bool, False),
     cfgv.Optional('description', cfgv.check_string, ''),
     cfgv.Optional('language_version', cfgv.check_string, C.DEFAULT),
     cfgv.Optional('log_file', cfgv.check_string, ''),

pre_commit/commands/run.py

@@ -139,10 +139,14 @@ def _run_single_hook(
 
         diff_cmd = ('git', 'diff', '--no-ext-diff')
         diff_before = cmd_output_b(*diff_cmd, retcode=None)
-        if not hook.pass_filenames:
-            filenames = ()
+        if hook.pass_diff:
+            inputs = (diff_before[1],)
+        elif hook.pass_filenames:
+            inputs = filenames
+        else:
+            inputs = ()
         time_before = time.time()
-        retcode, out = hook.run(filenames, use_color)
+        retcode, out = hook.run(inputs, use_color)
         duration = round(time.time() - time_before, 2) or 0
         diff_after = cmd_output_b(*diff_cmd, retcode=None)
 

pre_commit/languages/helpers.py

@@ -106,4 +106,8 @@ def run_xargs(
     # but do it deterministically in case a hook cares about ordering.
     file_args = _shuffled(file_args)
     kwargs['target_concurrency'] = target_concurrency(hook)
+    if hook.pass_diff:
+        input_data, = file_args
+        kwargs['input_data'] = input_data
+        file_args = ()
     return xargs(cmd, file_args, **kwargs)

pre_commit/util.py

@@ -138,6 +138,7 @@ def cmd_output_b(
         **kwargs: Any,
 ) -> Tuple[int, bytes, Optional[bytes]]:
     retcode = kwargs.pop('retcode', 0)
+    input_data = kwargs.pop('input_data', None)
     cmd, kwargs = _cmd_kwargs(*cmd, **kwargs)
 
     try:
@@ -146,7 +147,7 @@ def cmd_output_b(
         returncode, stdout_b, stderr_b = e.to_output()
     else:
         proc = subprocess.Popen(cmd, **kwargs)
-        stdout_b, stderr_b = proc.communicate()
+        stdout_b, stderr_b = proc.communicate(input_data)
         returncode = proc.returncode
 
     if retcode is not None and retcode != returncode:
@@ -207,6 +208,7 @@ if os.name != 'nt':  # pragma: windows no cover
     ) -> Tuple[int, bytes, Optional[bytes]]:
         assert kwargs.pop('retcode') is None
         assert kwargs['stderr'] == subprocess.STDOUT, kwargs['stderr']
+        input_data = kwargs.pop('input_data', None)
         cmd, kwargs = _cmd_kwargs(*cmd, **kwargs)
 
         try:
@@ -216,8 +218,12 @@ if os.name != 'nt':  # pragma: windows no cover
 
         with open(os.devnull) as devnull, Pty() as pty:
             assert pty.r is not None
-            kwargs.update({'stdin': devnull, 'stdout': pty.w, 'stderr': pty.w})
+            stdin = subprocess.PIPE if input_data else devnull
+            kwargs.update({'stdin': stdin, 'stdout': pty.w, 'stderr': pty.w})
             proc = subprocess.Popen(cmd, **kwargs)
+            if input_data:
+                proc.stdin.write(input_data)
+                proc.stdin.close()
             pty.close_w()
 
             buf = b''