doc/docchecker
author Matt Harbison <matt_harbison@yahoo.com>
Fri, 06 Jan 2023 12:20:09 -0500
changeset 49859 8b369bcbadf1
parent 48875 6000f5b25c9b
permissions -rwxr-xr-x
remotefilelog: byteify the message for a few StorageErrors Flagged by pytype locally.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
45830
c102b704edb5 global: use python3 in shebangs
Gregory Szorc <gregory.szorc@gmail.com>
parents: 43691
diff changeset
     1
#!/usr/bin/env python3
27730
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
8f2805ce93d9 py3: make doc/docchecker use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28811
diff changeset
    10
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    11
import os
29168
8f2805ce93d9 py3: make doc/docchecker use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28811
diff changeset
    12
import re
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    13
import sys
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    14
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    15
try:
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    16
    import msvcrt
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    17
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    18
    msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    19
    msvcrt.setmode(sys.stderr.fileno(), os.O_BINARY)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    20
except ImportError:
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    21
    pass
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    22
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    23
stdout = getattr(sys.stdout, 'buffer', sys.stdout)
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    24
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    25
leadingline = re.compile(br'(^\s*)(\S.*)$')
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    26
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    27
checks = [
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    28
    (
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    29
        br""":hg:`[^`]*'[^`]*`""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    30
        b"""warning: please avoid nesting ' in :hg:`...`""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    31
    ),
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    32
    (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
    33
    (
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    34
        br"""(?:[^a-z][^'.])hg ([^,;"`]*'(?!hg)){2}""",
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    35
        b'''warning: please use " instead of ' for hg ... "..."''',
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    36
    ),
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    37
]
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    38
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    39
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    40
def check(line):
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    41
    messages = []
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    42
    for match, msg in checks:
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    43
        if re.search(match, line):
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    44
            messages.append(msg)
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    45
    if messages:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    46
        stdout.write(b'%s\n' % line)
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    47
        for msg in messages:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    48
            stdout.write(b'%s\n' % msg)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    49
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    50
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    51
def work(file):
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    52
    (llead, lline) = (b'', b'')
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    53
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    54
    for line in file:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    55
        # this section unwraps lines
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    56
        match = leadingline.match(line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    57
        if not match:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    58
            check(lline)
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    59
            (llead, lline) = (b'', b'')
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    60
            continue
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    61
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    62
        lead, line = match.group(1), match.group(2)
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    63
        if lead == llead:
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    64
            if lline != b'':
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    65
                lline += b' ' + line
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    66
            else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    67
                lline = line
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    68
        else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    69
            check(lline)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    70
            (llead, lline) = (lead, line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    71
    check(lline)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    72
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    73
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    74
def main():
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    75
    for f in sys.argv[1:]:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    76
        try:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    77
            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
    78
                work(file)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    79
        except BaseException as e:
41006
9bfbb9fc5871 py3: byteify docchecker
Matt Harbison <matt_harbison@yahoo.com>
parents: 29169
diff changeset
    80
            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
    81
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41006
diff changeset
    82
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    83
main()