doc/docchecker
author Matt Mackall <mpm@selenic.com>
Mon, 16 May 2016 17:21:25 -0500
branchstable
changeset 29157 854556c5f3bf
parent 28811 1a623585a658
child 29168 8f2805ce93d9
permissions -rwxr-xr-x
Added signature for changeset aaabed77791a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
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.
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
     9
import sys
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    10
import re
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    11
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    12
leadingline = re.compile(r'(^\s*)(\S.*)$')
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    13
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    14
checks = [
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    15
  (r""":hg:`[^`]*'[^`]*`""",
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    16
    """warning: please avoid nesting ' in :hg:`...`"""),
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    17
  (r'\w:hg:`',
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    18
    'warning: please have a space before :hg:'),
28811
1a623585a658 docchecker: try to reject single quotes
timeless <timeless@mozdev.org>
parents: 28810
diff changeset
    19
  (r"""(?:[^a-z][^'.])hg ([^,;"`]*'(?!hg)){2}""",
1a623585a658 docchecker: try to reject single quotes
timeless <timeless@mozdev.org>
parents: 28810
diff changeset
    20
    '''warning: please use " instead of ' for hg ... "..."'''),
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    21
]
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    22
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    23
def check(line):
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    24
    messages = []
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    25
    for match, msg in checks:
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    26
        if re.search(match, line):
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    27
            messages.append(msg)
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    28
    if messages:
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    29
        print(line)
28810
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    30
        for msg in messages:
9934362978e1 docchecker: report context line at most once
timeless <timeless@mozdev.org>
parents: 28049
diff changeset
    31
            print(msg)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    32
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    33
def work(file):
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    34
    (llead, lline) = ('', '')
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    35
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    36
    for line in file:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    37
        # this section unwraps lines
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    38
        match = leadingline.match(line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    39
        if not match:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    40
            check(lline)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    41
            (llead, lline) = ('', '')
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    42
            continue
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    43
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    44
        lead, line = match.group(1), match.group(2)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    45
        if (lead == llead):
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    46
            if (lline != ''):
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    47
                lline += ' ' + line
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    48
            else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    49
                lline = line
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    50
        else:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    51
            check(lline)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    52
            (llead, lline) = (lead, line)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    53
    check(lline)
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    54
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    55
def main():
28049
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    56
    for f in sys.argv[1:]:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    57
        try:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    58
            with open(f) as file:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    59
                work(file)
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    60
        except BaseException as e:
c00f67c15c5a docchecker: use indentation of 4 spaces
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 28048
diff changeset
    61
            print("failed to process %s: %s" % (f, e))
27730
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    62
ed86fb2a4187 docchecker: introduce a way to check for poor markup
timeless <timeless@mozdev.org>
parents:
diff changeset
    63
main()