hgext/highlight/highlight.py
author Benoit Boissinot <benoit.boissinot@ens-lyon.org>
Sat, 06 Dec 2008 15:40:04 +0100
changeset 7472 9d457bb38de5
parent 7120 db7557359636
child 7948 de377b1a9a84
permissions -rw-r--r--
convert/cvs: do not compare None with int
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
6938
ce94b3236ea4 highlight: split code to improve startup times
Patrick Mezard <pmezard@gmail.com>
parents: 6666
diff changeset
     1
# highlight extension implementation file
ce94b3236ea4 highlight: split code to improve startup times
Patrick Mezard <pmezard@gmail.com>
parents: 6666
diff changeset
     2
#
ce94b3236ea4 highlight: split code to improve startup times
Patrick Mezard <pmezard@gmail.com>
parents: 6666
diff changeset
     3
# The original module was split in an interface and an implementation
ce94b3236ea4 highlight: split code to improve startup times
Patrick Mezard <pmezard@gmail.com>
parents: 6666
diff changeset
     4
# file to defer pygments loading and speedup extension setup.
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
     5
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
     6
from mercurial import demandimport
6394
55bc0a035e1f highlight: some small cleanups
Dirkjan Ochtman <dirkjan@ochtman.nl>
parents: 6393
diff changeset
     7
demandimport.ignore.extend(['pkgutil', 'pkg_resources', '__main__',])
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
     8
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
     9
from mercurial import util
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    10
from mercurial.templatefilters import filters
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    11
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    12
from pygments import highlight
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    13
from pygments.util import ClassNotFound
6199
0068809347d7 highlight: forgot import
Brendan Cully <brendan@kublai.com>
parents: 6198
diff changeset
    14
from pygments.lexers import guess_lexer, guess_lexer_for_filename, TextLexer
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    15
from pygments.formatters import HtmlFormatter
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    16
6485
938319418d8c highlight: Generate pygments style sheet dynamically
Isaac Jurado <diptongo@gmail.com>
parents: 6394
diff changeset
    17
SYNTAX_CSS = ('\n<link rel="stylesheet" href="{url}highlightcss" '
5533
6cf7d7fe7d3d highlight: clean up coding style a little
Bryan O'Sullivan <bos@serpentine.com>
parents: 5532
diff changeset
    18
              'type="text/css" />')
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    19
6394
55bc0a035e1f highlight: some small cleanups
Dirkjan Ochtman <dirkjan@ochtman.nl>
parents: 6393
diff changeset
    20
def pygmentize(field, fctx, style, tmpl):
55bc0a035e1f highlight: some small cleanups
Dirkjan Ochtman <dirkjan@ochtman.nl>
parents: 6393
diff changeset
    21
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    22
    # append a <link ...> to the syntax highlighting css
5616
88ca3e0fb6e5 highlight: adapt to hgweb_mode refactoring
Christian Ebert <blacktrash@gmx.net>
parents: 5533
diff changeset
    23
    old_header = ''.join(tmpl('header'))
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    24
    if SYNTAX_CSS not in old_header:
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    25
        new_header =  old_header + SYNTAX_CSS
5616
88ca3e0fb6e5 highlight: adapt to hgweb_mode refactoring
Christian Ebert <blacktrash@gmx.net>
parents: 5533
diff changeset
    26
        tmpl.cache['header'] = new_header
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    27
6194
fe54e7501de1 highlight: bail out if file is binary
Brendan Cully <brendan@kublai.com>
parents: 6193
diff changeset
    28
    text = fctx.data()
fe54e7501de1 highlight: bail out if file is binary
Brendan Cully <brendan@kublai.com>
parents: 6193
diff changeset
    29
    if util.binary(text):
fe54e7501de1 highlight: bail out if file is binary
Brendan Cully <brendan@kublai.com>
parents: 6193
diff changeset
    30
        return
fe54e7501de1 highlight: bail out if file is binary
Brendan Cully <brendan@kublai.com>
parents: 6193
diff changeset
    31
7120
db7557359636 highlight: convert text to local before passing to pygmentize (issue1341)
Christian Ebert <blacktrash@gmx.net>
parents: 6938
diff changeset
    32
    # avoid UnicodeDecodeError in pygments
db7557359636 highlight: convert text to local before passing to pygmentize (issue1341)
Christian Ebert <blacktrash@gmx.net>
parents: 6938
diff changeset
    33
    text = util.tolocal(text)
db7557359636 highlight: convert text to local before passing to pygmentize (issue1341)
Christian Ebert <blacktrash@gmx.net>
parents: 6938
diff changeset
    34
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    35
    # To get multi-line strings right, we can't format line-by-line
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    36
    try:
6662
22c303a514f8 Truncate input to 1K when using pygments guess_lexer.
Brendan Cully <brendan@kublai.com>
parents: 6494
diff changeset
    37
        lexer = guess_lexer_for_filename(fctx.path(), text[:1024],
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    38
                                         encoding=util._encoding)
6494
c30849d4c8ba highlight: backward compatibility with pygments 0.5.1
Benoit Allard <benoit@aeteurope.nl>
parents: 6212
diff changeset
    39
    except (ClassNotFound, ValueError):
6198
358cc9cf54db highlight: guess by text when path name is ambiguous
Brendan Cully <brendan@kublai.com>
parents: 6197
diff changeset
    40
        try:
6662
22c303a514f8 Truncate input to 1K when using pygments guess_lexer.
Brendan Cully <brendan@kublai.com>
parents: 6494
diff changeset
    41
            lexer = guess_lexer(text[:1024], encoding=util._encoding)
6494
c30849d4c8ba highlight: backward compatibility with pygments 0.5.1
Benoit Allard <benoit@aeteurope.nl>
parents: 6212
diff changeset
    42
        except (ClassNotFound, ValueError):
6198
358cc9cf54db highlight: guess by text when path name is ambiguous
Brendan Cully <brendan@kublai.com>
parents: 6197
diff changeset
    43
            lexer = TextLexer(encoding=util._encoding)
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    44
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    45
    formatter = HtmlFormatter(style=style, encoding=util._encoding)
5532
40a06e39f010 extension for synax highlighting in the hgweb file revision view
Adam Hupp <adam@hupp.org>
parents:
diff changeset
    46
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    47
    colorized = highlight(text, lexer, formatter)
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    48
    # strip wrapping div
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    49
    colorized = colorized[:colorized.find('\n</pre>')]
6197
f6565f7d9489 highlight: make div trimmer work when lexer fails to identify text.
Brendan Cully <brendan@kublai.com>
parents: 6196
diff changeset
    50
    colorized = colorized[colorized.find('<pre>')+5:]
6196
fc9535ae6fe2 highlight: use iter() instead of generator comprehension
Brendan Cully <brendan@kublai.com>
parents: 6194
diff changeset
    51
    coloriter = iter(colorized.splitlines())
6193
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    52
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    53
    filters['colorize'] = lambda x: coloriter.next()
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    54
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    55
    oldl = tmpl.cache[field]
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    56
    newl = oldl.replace('line|escape', 'line|colorize')
2344da8eb9b4 highlight: support annotate, and reduce layering violations.
Brendan Cully <brendan@kublai.com>
parents: 5991
diff changeset
    57
    tmpl.cache[field] = newl