annotate doc/docchecker @ 44705:75f1197db884

manifest: fix yet another 20-byte-hash assumption Differential Revision: https://phab.mercurial-scm.org/D8371
author Augie Fackler <augie@google.com>
date Thu, 02 Apr 2020 16:01:36 -0400
parents 47ef023d0165
children c102b704edb5
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
1 #!/usr/bin/env python
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
2 #
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
3 # docchecker - look for problematic markup
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
4 #
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
5 # Copyright 2016 timeless <timeless@mozdev.org> and others
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
6 #
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
7 # This software may be used and distributed according to the terms of the
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
8 # GNU General Public License version 2 or any later version.
29168
8f2805ce93d9 py3: make doc/docchecker use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28811
diff changeset
9
29169
c9ab5a0bc7c5 py3: make doc/docchecker use print_function
Pulkit Goyal <7895pulkit@gmail.com>
parents: 29168
diff changeset
10 from __future__ import absolute_import, print_function
29168
8f2805ce93d9 py3: make doc/docchecker use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28811
diff changeset
11
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
12 import os
29168
8f2805ce93d9 py3: make doc/docchecker use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28811
diff changeset
13 import re
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
14 import sys
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
15
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
16 try:
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
17 import msvcrt
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
18
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
19 msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
20 msvcrt.setmode(sys.stderr.fileno(), os.O_BINARY)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
21 except ImportError:
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
22 pass
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
23
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
24 stdout = getattr(sys.stdout, 'buffer', sys.stdout)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
25
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
26 leadingline = re.compile(br'(^\s*)(\S.*)$')
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
27
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
28 checks = [
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
29 (
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
30 br""":hg:`[^`]*'[^`]*`""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
31 b"""warning: please avoid nesting ' in :hg:`...`""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
32 ),
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
33 (br'\w:hg:`', b'warning: please have a space before :hg:'),
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
34 (
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
35 br"""(?:[^a-z][^'.])hg ([^,;"`]*'(?!hg)){2}""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
36 b'''warning: please use " instead of ' for hg ... "..."''',
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
37 ),
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
38 ]
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
39
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
40
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
41 def check(line):
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
42 messages = []
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
43 for match, msg in checks:
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
44 if re.search(match, line):
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
45 messages.append(msg)
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
46 if messages:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
47 stdout.write(b'%s\n' % line)
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
48 for msg in messages:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
49 stdout.write(b'%s\n' % msg)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
50
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
51
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
52 def work(file):
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
53 (llead, lline) = (b'', b'')
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
54
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
55 for line in file:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
56 # this section unwraps lines
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
57 match = leadingline.match(line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
58 if not match:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
59 check(lline)
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
60 (llead, lline) = (b'', b'')
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
61 continue
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
62
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
63 lead, line = match.group(1), match.group(2)
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
64 if lead == llead:
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
65 if lline != b'':
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
66 lline += b' ' + line
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
67 else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
68 lline = line
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
69 else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
70 check(lline)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
71 (llead, lline) = (lead, line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
72 check(lline)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
73
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
74
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
75 def main():
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
76 for f in sys.argv[1:]:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
77 try:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
78 with open(f, 'rb') as file:
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
79 work(file)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
80 except BaseException as e:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
81 sys.stdout.write(r"failed to process %s: %s\n" % (f, e))
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
82
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
83
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
84 main()