i18n/posplit
author Matt Harbison <matt_harbison@yahoo.com>
Thu, 11 Mar 2021 19:21:58 -0500
branchstable
changeset 46694 d6601547f22b
parent 45830 c102b704edb5
child 48875 6000f5b25c9b
permissions -rwxr-xr-x
subrepo: handle unexpected file types from git gracefully This was flagged by pytype because `tar.extractfile(...)` can return None if the entry is not a file or symlink. I don't think that git supports other types, but better safe than sorry. Differential Revision: https://phab.mercurial-scm.org/D10179
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
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     2
#
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     3
# posplit - split messages in paragraphs on .po/.pot files
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     4
#
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     5
# license: MIT/X11/Expat
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     6
#
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
     7
29153
90d84e1e427a py3: make i18n/posplit use print_function
Pulkit Goyal <7895pulkit@gmail.com>
parents: 29152
diff changeset
     8
from __future__ import absolute_import, print_function
29152
c5057b7780dc py3: make i18n/posplit use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28074
diff changeset
     9
c5057b7780dc py3: make i18n/posplit use absolute_import
Pulkit Goyal <7895pulkit@gmail.com>
parents: 28074
diff changeset
    10
import polib
20359
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    11
import re
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    12
import sys
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    13
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    14
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    15
def addentry(po, entry, cache):
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    16
    e = cache.get(entry.msgid)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    17
    if e:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    18
        e.occurrences.extend(entry.occurrences)
39269
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    19
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    20
        # merge comments from entry
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    21
        for comment in entry.comment.split('\n'):
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    22
            if comment and comment not in e.comment:
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    23
                if not e.comment:
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    24
                    e.comment = comment
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    25
                else:
d0e8933d6dad i18n: merge i18n comments of translatable texts correctly
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 29153
diff changeset
    26
                    e.comment += '\n' + comment
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    27
    else:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    28
        po.append(entry)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    29
        cache[entry.msgid] = entry
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    30
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    31
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    32
def mkentry(orig, delta, msgid, msgstr):
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    33
    entry = polib.POEntry()
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    34
    entry.merge(orig)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    35
    entry.msgid = msgid or orig.msgid
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    36
    entry.msgstr = msgstr or orig.msgstr
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    37
    entry.occurrences = [(p, int(l) + delta) for (p, l) in orig.occurrences]
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    38
    return entry
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    39
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    40
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    41
if __name__ == "__main__":
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    42
    po = polib.pofile(sys.argv[1])
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    43
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    44
    cache = {}
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    45
    entries = po[:]
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    46
    po[:] = []
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    47
    findd = re.compile(r' *\.\. (\w+)::')  # for finding directives
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    48
    for entry in entries:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    49
        msgids = entry.msgid.split(u'\n\n')
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    50
        if entry.msgstr:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    51
            msgstrs = entry.msgstr.split(u'\n\n')
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    52
        else:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    53
            msgstrs = [u''] * len(msgids)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    54
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    55
        if len(msgids) != len(msgstrs):
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    56
            # places the whole existing translation as a fuzzy
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    57
            # translation for each paragraph, to give the
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    58
            # translator a chance to recover part of the old
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    59
            # translation - erasing extra paragraphs is
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    60
            # probably better than retranslating all from start
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    61
            if 'fuzzy' not in entry.flags:
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    62
                entry.flags.append('fuzzy')
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    63
            msgstrs = [entry.msgstr] * len(msgids)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    64
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    65
        delta = 0
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    66
        for msgid, msgstr in zip(msgids, msgstrs):
20361
3fe079d3a2b4 i18n: posplit removes the entry "::" from the pot file
Simon Heimberg <simohe@besonet.ch>
parents: 20359
diff changeset
    67
            if msgid and msgid != '::':
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    68
                newentry = mkentry(entry, delta, msgid, msgstr)
20359
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    69
                mdirective = findd.match(msgid)
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    70
                if mdirective:
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    71
                    if not msgid[mdirective.end() :].rstrip():
20362
1bce1078501d i18n: leave out entries which contain only a rst directive
Simon Heimberg <simohe@besonet.ch>
parents: 20361
diff changeset
    72
                        # only directive, nothing to translate here
28074
a1924bc6e267 i18n: calculate correct line number in source of messages to be translated
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 20363
diff changeset
    73
                        delta += 2
20362
1bce1078501d i18n: leave out entries which contain only a rst directive
Simon Heimberg <simohe@besonet.ch>
parents: 20361
diff changeset
    74
                        continue
20359
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    75
                    directive = mdirective.group(1)
20363
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    76
                    if directive in ('container', 'include'):
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    77
                        if msgid.rstrip('\n').count('\n') == 0:
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    78
                            # only rst syntax, nothing to translate
28074
a1924bc6e267 i18n: calculate correct line number in source of messages to be translated
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 20363
diff changeset
    79
                            delta += 2
20363
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    80
                            continue
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    81
                        else:
e3ee7ec85a15 i18n: leave out entries which contain only rst syntax
Simon Heimberg <simohe@besonet.ch>
parents: 20362
diff changeset
    82
                            # lines following directly, unexpected
43691
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    83
                            print(
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    84
                                'Warning: text follows line with directive'
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    85
                                ' %s' % directive
47ef023d0165 black: blacken scripts
Gregory Szorc <gregory.szorc@gmail.com>
parents: 41759
diff changeset
    86
                            )
20359
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    87
                    comment = 'do not translate: .. %s::' % directive
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    88
                    if not newentry.comment:
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    89
                        newentry.comment = comment
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    90
                    elif comment not in newentry.comment:
ff6ab0b2ebf7 i18n: posplit writes a warning for translators before rst directives
Simon Heimberg <simohe@besonet.ch>
parents: 11389
diff changeset
    91
                        newentry.comment += '\n' + comment
11389
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    92
                addentry(po, newentry, cache)
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    93
            delta += 2 + msgid.count('\n')
4fd49329a1b5 i18n: script for splitting large messages on .po/.pot files
Wagner Bruna <wbruna@yahoo.com>
parents:
diff changeset
    94
    po.save()