First commit of file-contents-sorter precommit hook

Author: Daniel Gallagher

.pre-commit-hooks.yaml

@@ -105,6 +105,12 @@
     entry: end-of-file-fixer
     language: python
     files: \.(asciidoc|adoc|coffee|cpp|css|c|ejs|erb|groovy|h|haml|hh|hpp|hxx|html|in|j2|jade|json|js|less|markdown|md|ml|mli|pp|py|rb|rs|R|scala|scss|sh|slim|tex|tmpl|ts|txt|yaml|yml)$
+-   id: file-contents-sorter
+    name: File Contents Sorter
+    description: Sort the lines in specified files (defaults to alphabetical). You must provide list of target files as input in your .pre-commit-config.yaml file.
+    entry: file-contents-sorter
+    language: python
+    files: ''
 -   id: fix-encoding-pragma
     name: Fix python encoding pragma
     language: python

pre_commit_hooks/file_contents_sorter.py

@@ -0,0 +1,57 @@
+"""
+A very simple pre-commit hook that, when passed one or more filenames
+as arguments, will sort the lines in those files.
+
+An example use case for this: you have a deploy-whitelist.txt file
+in a repo that contains a list of filenames that is used to specify
+files to be included in a docker container. This file has one filename
+per line. Various users are adding/removing lines from this file; using
+this hook on that file should reduce the instances of git merge
+conflicts and keep the file nicely ordered.
+"""
+from __future__ import print_function
+
+import argparse
+
+PASS = 0
+FAIL = 1
+
+
+def sort_file_contents(f):
+    before = [line for line in f]
+    after = sorted(before)
+
+    before_string = b''.join(before)
+    after_string = b''.join(after)
+
+    if before_string == after_string:
+        return PASS
+    else:
+        f.seek(0)
+        f.write(after_string)
+        f.truncate()
+        return FAIL
+
+
+def parse_commandline_input(argv):
+    parser = argparse.ArgumentParser()
+    parser.add_argument('filenames', nargs='+', help='Files to sort')
+    args = parser.parse_args(argv)
+    return args
+
+
+def main(argv=None):
+    args = parse_commandline_input(argv)
+
+    retv = PASS
+
+    for arg in args.filenames:
+        with open(arg, 'rb+') as file_obj:
+            ret_for_file = sort_file_contents(file_obj)
+
+            if ret_for_file:
+                print('Sorting {}'.format(arg))
+
+            retv |= ret_for_file
+
+    return retv

requirements-dev.txt

@@ -2,6 +2,7 @@
 
 coverage
 flake8
+ipdb
 mock
 pre-commit
 pytest

setup.py

@@ -49,6 +49,7 @@
             'detect-private-key = pre_commit_hooks.detect_private_key:detect_private_key',
             'double-quote-string-fixer = pre_commit_hooks.string_fixer:main',
             'end-of-file-fixer = pre_commit_hooks.end_of_file_fixer:end_of_file_fixer',
+            'file-contents-sorter = pre_commit_hooks.file_contents_sorter:main',
             'fix-encoding-pragma = pre_commit_hooks.fix_encoding_pragma:main',
             'forbid-new-submodules = pre_commit_hooks.forbid_new_submodules:main',
             'name-tests-test = pre_commit_hooks.tests_should_end_in_test:validate_files',

tests/file_contents_sorter_test.py

@@ -0,0 +1,54 @@
+from argparse import ArgumentError
+
+import pytest
+
+from pre_commit_hooks.file_contents_sorter import FAIL
+from pre_commit_hooks.file_contents_sorter import main
+from pre_commit_hooks.file_contents_sorter import parse_commandline_input
+from pre_commit_hooks.file_contents_sorter import PASS
+from pre_commit_hooks.file_contents_sorter import sort_file_contents
+
+
+def _n(*strs):
+    return b'\n'.join(strs) + '\n'
+
+
+# Input, expected return value, expected output
+TESTS = (
+    (b'', PASS, b''),
+    (_n('lonesome'), PASS, _n('lonesome')),
+    (b'missing_newline', PASS, b'missing_newline'),
+    (_n('alpha', 'beta'), PASS, _n('alpha', 'beta')),
+    (_n('beta', 'alpha'), FAIL, _n('alpha', 'beta')),
+    (_n('C', 'c'), PASS, _n('C', 'c')),
+    (_n('c', 'C'), FAIL, _n('C', 'c')),
+    (_n('mag ical ', ' tre vor'), FAIL, _n(' tre vor', 'mag ical ')),
+    (_n('@', '-', '_', '#'), FAIL, _n('#', '-', '@', '_')),
+)
+
+
+@pytest.mark.parametrize(('input_s', 'expected_retval', 'output'), TESTS)
+def test_integration(input_s, expected_retval, output, tmpdir):
+    path = tmpdir.join('file.txt')
+    path.write_binary(input_s)
+
+    output_retval = main([path.strpath])
+
+    assert path.read_binary() == output
+    assert output_retval == expected_retval
+
+
+def test_parse_commandline_input_errors_without_args():
+    with pytest.raises(SystemExit):
+        parse_commandline_input([])
+
+@pytest.mark.parametrize(
+    ('filename_list'), 
+    (
+        ['filename1'], 
+        ['filename1', 'filename2'],
+    )
+)
+def test_parse_commandline_input_success(filename_list):
+    args = parse_commandline_input(filename_list)
+    assert args.filenames == filename_list