Speed up filename filtering.

Before there was a `getcwd` syscall for every filename which was filtered.

Instead this is now cached per-run.

- When all files are identified by filename only: ~45% improvement
- When no files are identified by filename only: ~55% improvement

This makes little difference to overall execution, the bigger win is
eliminating the `memoize_by_cwd` hack.  Just removing the memoization would
have *increased* the runtime by 300-500%.
This commit is contained in:
Anthony Sottile 2019-01-23 20:42:27 -08:00
parent 38308dc02d
commit b1389603e0
6 changed files with 61 additions and 103 deletions

View file

@ -17,14 +17,47 @@ from pre_commit.repository import all_hooks
from pre_commit.repository import install_hook_envs
from pre_commit.staged_files_only import staged_files_only
from pre_commit.util import cmd_output
from pre_commit.util import memoize_by_cwd
from pre_commit.util import noop_context
logger = logging.getLogger('pre_commit')
tags_from_path = memoize_by_cwd(tags_from_path)
def filter_by_include_exclude(names, include, exclude):
include_re, exclude_re = re.compile(include), re.compile(exclude)
return [
filename for filename in names
if include_re.search(filename)
if not exclude_re.search(filename)
]
class Classifier(object):
def __init__(self, filenames):
self.filenames = [f for f in filenames if os.path.lexists(f)]
self._types_cache = {}
def _types_for_file(self, filename):
try:
return self._types_cache[filename]
except KeyError:
ret = self._types_cache[filename] = tags_from_path(filename)
return ret
def by_types(self, names, types, exclude_types):
types, exclude_types = frozenset(types), frozenset(exclude_types)
ret = []
for filename in names:
tags = self._types_for_file(filename)
if tags >= types and not tags & exclude_types:
ret.append(filename)
return ret
def filenames_for_hook(self, hook):
names = self.filenames
names = filter_by_include_exclude(names, hook.files, hook.exclude)
names = self.by_types(names, hook.types, hook.exclude_types)
return names
def _get_skips(environ):
@ -36,37 +69,12 @@ def _hook_msg_start(hook, verbose):
return '{}{}'.format('[{}] '.format(hook.id) if verbose else '', hook.name)
def _filter_by_include_exclude(filenames, include, exclude):
include_re, exclude_re = re.compile(include), re.compile(exclude)
return [
filename for filename in filenames
if (
include_re.search(filename) and
not exclude_re.search(filename) and
os.path.lexists(filename)
)
]
def _filter_by_types(filenames, types, exclude_types):
types, exclude_types = frozenset(types), frozenset(exclude_types)
ret = []
for filename in filenames:
tags = tags_from_path(filename)
if tags >= types and not tags & exclude_types:
ret.append(filename)
return tuple(ret)
SKIPPED = 'Skipped'
NO_FILES = '(no files to check)'
def _run_single_hook(filenames, hook, args, skips, cols):
include, exclude = hook.files, hook.exclude
filenames = _filter_by_include_exclude(filenames, include, exclude)
types, exclude_types = hook.types, hook.exclude_types
filenames = _filter_by_types(filenames, types, exclude_types)
def _run_single_hook(classifier, hook, args, skips, cols):
filenames = classifier.filenames_for_hook(hook)
if hook.language == 'pcre':
logger.warning(
@ -193,10 +201,11 @@ def _run_hooks(config, hooks, args, environ):
skips = _get_skips(environ)
cols = _compute_cols(hooks, args.verbose)
filenames = _all_filenames(args)
filenames = _filter_by_include_exclude(filenames, '', config['exclude'])
filenames = filter_by_include_exclude(filenames, '', config['exclude'])
classifier = Classifier(filenames)
retval = 0
for hook in hooks:
retval |= _run_single_hook(filenames, hook, args, skips, cols)
retval |= _run_single_hook(classifier, hook, args, skips, cols)
if retval and config['fail_fast']:
break
if retval and args.show_diff_on_failure and git.has_diff():