Mercurial > repos > bcclaywell > argo_navis

annotate venv/lib/python2.7/site-packages/docutils/utils/punctuation_chars.py @ 0:d67268158946 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
author bcclaywell
date Mon, 12 Oct 2015 17:43:33 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
1 #!/usr/bin/env python
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
2 # -*- coding: utf-8 -*-
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
3 # :Copyright: © 2011 Günter Milde.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
4 # :License: Released under the terms of the `2-Clause BSD license`_, in short:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
5 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
6 # Copying and distribution of this file, with or without modification,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
7 # are permitted in any medium without royalty provided the copyright
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
8 # notice and this notice are preserved.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
9 # This file is offered as-is, without any warranty.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
10 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
11 # .. _2-Clause BSD license: http://www.spdx.org/licenses/BSD-2-Clause
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
12
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
13 # :Id: $Id: punctuation_chars.py 7668 2013-06-04 12:46:30Z milde $
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
14
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
15 import sys, re
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
16 import unicodedata
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
17
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
18 # punctuation characters around inline markup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
19 # ===========================================
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
20 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
21 # This module provides the lists of characters for the implementation of
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
22 # the `inline markup recognition rules`_ in the reStructuredText parser
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
23 # (states.py)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
24 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
25 # .. _inline markup recognition rules:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
26 # ../../docs/ref/rst/restructuredtext.html#inline-markup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
27
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
28 # Docutils punctuation category sample strings
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
29 # --------------------------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
30 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
31 # The sample strings are generated by punctuation_samples() and put here
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
32 # literal to avoid the time-consuming generation with every Docutils run.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
33 # As the samples are used inside ``[ ]`` in regular expressions, hyphen and
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
34 # square brackets are escaped. ::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
35
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
36 openers = (u'"\'(<\\[{\u0f3a\u0f3c\u169b\u2045\u207d\u208d\u2329\u2768'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
37 u'\u276a\u276c\u276e\u2770\u2772\u2774\u27c5\u27e6\u27e8\u27ea'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
38 u'\u27ec\u27ee\u2983\u2985\u2987\u2989\u298b\u298d\u298f\u2991'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
39 u'\u2993\u2995\u2997\u29d8\u29da\u29fc\u2e22\u2e24\u2e26\u2e28'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
40 u'\u3008\u300a\u300c\u300e\u3010\u3014\u3016\u3018\u301a\u301d'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
41 u'\u301d\ufd3e\ufe17\ufe35\ufe37\ufe39\ufe3b\ufe3d\ufe3f\ufe41'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
42 u'\ufe43\ufe47\ufe59\ufe5b\ufe5d\uff08\uff3b\uff5b\uff5f\uff62'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
43 u'\xab\u2018\u201c\u2039\u2e02\u2e04\u2e09\u2e0c\u2e1c\u2e20'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
44 u'\u201a\u201e\xbb\u2019\u201d\u203a\u2e03\u2e05\u2e0a\u2e0d'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
45 u'\u2e1d\u2e21\u201b\u201f')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
46 closers = (u'"\')>\\]}\u0f3b\u0f3d\u169c\u2046\u207e\u208e\u232a\u2769'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
47 u'\u276b\u276d\u276f\u2771\u2773\u2775\u27c6\u27e7\u27e9\u27eb'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
48 u'\u27ed\u27ef\u2984\u2986\u2988\u298a\u298c\u298e\u2990\u2992'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
49 u'\u2994\u2996\u2998\u29d9\u29db\u29fd\u2e23\u2e25\u2e27\u2e29'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
50 u'\u3009\u300b\u300d\u300f\u3011\u3015\u3017\u3019\u301b\u301e'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
51 u'\u301f\ufd3f\ufe18\ufe36\ufe38\ufe3a\ufe3c\ufe3e\ufe40\ufe42'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
52 u'\ufe44\ufe48\ufe5a\ufe5c\ufe5e\uff09\uff3d\uff5d\uff60\uff63'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
53 u'\xbb\u2019\u201d\u203a\u2e03\u2e05\u2e0a\u2e0d\u2e1d\u2e21'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
54 u'\u201b\u201f\xab\u2018\u201c\u2039\u2e02\u2e04\u2e09\u2e0c'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
55 u'\u2e1c\u2e20\u201a\u201e')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
56 delimiters = (u'\\-/:\u058a\xa1\xb7\xbf\u037e\u0387\u055a-\u055f\u0589'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
57 u'\u05be\u05c0\u05c3\u05c6\u05f3\u05f4\u0609\u060a\u060c'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
58 u'\u060d\u061b\u061e\u061f\u066a-\u066d\u06d4\u0700-\u070d'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
59 u'\u07f7-\u07f9\u0830-\u083e\u0964\u0965\u0970\u0df4\u0e4f'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
60 u'\u0e5a\u0e5b\u0f04-\u0f12\u0f85\u0fd0-\u0fd4\u104a-\u104f'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
61 u'\u10fb\u1361-\u1368\u1400\u166d\u166e\u16eb-\u16ed\u1735'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
62 u'\u1736\u17d4-\u17d6\u17d8-\u17da\u1800-\u180a\u1944\u1945'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
63 u'\u19de\u19df\u1a1e\u1a1f\u1aa0-\u1aa6\u1aa8-\u1aad\u1b5a-'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
64 u'\u1b60\u1c3b-\u1c3f\u1c7e\u1c7f\u1cd3\u2010-\u2017\u2020-'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
65 u'\u2027\u2030-\u2038\u203b-\u203e\u2041-\u2043\u2047-'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
66 u'\u2051\u2053\u2055-\u205e\u2cf9-\u2cfc\u2cfe\u2cff\u2e00'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
67 u'\u2e01\u2e06-\u2e08\u2e0b\u2e0e-\u2e1b\u2e1e\u2e1f\u2e2a-'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
68 u'\u2e2e\u2e30\u2e31\u3001-\u3003\u301c\u3030\u303d\u30a0'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
69 u'\u30fb\ua4fe\ua4ff\ua60d-\ua60f\ua673\ua67e\ua6f2-\ua6f7'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
70 u'\ua874-\ua877\ua8ce\ua8cf\ua8f8-\ua8fa\ua92e\ua92f\ua95f'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
71 u'\ua9c1-\ua9cd\ua9de\ua9df\uaa5c-\uaa5f\uaade\uaadf\uabeb'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
72 u'\ufe10-\ufe16\ufe19\ufe30-\ufe32\ufe45\ufe46\ufe49-\ufe4c'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
73 u'\ufe50-\ufe52\ufe54-\ufe58\ufe5f-\ufe61\ufe63\ufe68\ufe6a'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
74 u'\ufe6b\uff01-\uff03\uff05-\uff07\uff0a\uff0c-\uff0f\uff1a'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
75 u'\uff1b\uff1f\uff20\uff3c\uff61\uff64\uff65')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
76 if sys.maxunicode >= 0x10FFFF: # "wide" build
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
77 delimiters += (u'\U00010100\U00010101\U0001039f\U000103d0\U00010857'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
78 u'\U0001091f\U0001093f\U00010a50-\U00010a58\U00010a7f'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
79 u'\U00010b39-\U00010b3f\U000110bb\U000110bc\U000110be-'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
80 u'\U000110c1\U00012470-\U00012473')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
81 closing_delimiters = u'\\\\.,;!?'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
82
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
83
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
84 # Matching open/close quotes
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
85 # --------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
86
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
87 # Rule (5) requires determination of matching open/close pairs. However,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
88 # the pairing of open/close quotes is ambigue due to different typographic
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
89 # conventions in different languages.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
90
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
91 quote_pairs = {u'\xbb': u'\xbb', # Swedish
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
92 u'\u2018': u'\u201a', # Greek
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
93 u'\u2019': u'\u2019', # Swedish
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
94 u'\u201a': u'\u2018\u2019', # German, Polish
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
95 u'\u201c': u'\u201e', # German
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
96 u'\u201e': u'\u201c\u201d',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
97 u'\u201d': u'\u201d', # Swedish
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
98 u'\u203a': u'\u203a', # Swedish
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
99 }
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
100
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
101 def match_chars(c1, c2):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
102 try:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
103 i = openers.index(c1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
104 except ValueError: # c1 not in openers
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
105 return False
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
106 return c2 == closers[i] or c2 in quote_pairs.get(c1, '')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
107
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
108
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
109 # Running this file as a standalone module checks the definitions against a
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
110 # re-calculation::
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
111
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
112 if __name__ == '__main__':
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
113
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
114
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
115 # Unicode punctuation character categories
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
116 # ----------------------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
117
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
118 unicode_punctuation_categories = {
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
119 # 'Pc': 'Connector', # not used in Docutils inline markup recognition
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
120 'Pd': 'Dash',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
121 'Ps': 'Open',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
122 'Pe': 'Close',
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
123 'Pi': 'Initial quote', # may behave like Ps or Pe depending on usage
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
124 'Pf': 'Final quote', # may behave like Ps or Pe depending on usage
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
125 'Po': 'Other'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
126 }
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
127 """Unicode character categories for punctuation"""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
128
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
129
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
130 # generate character pattern strings
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
131 # ==================================
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
132
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
133 def unicode_charlists(categories, cp_min=0, cp_max=None):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
134 """Return dictionary of Unicode character lists.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
135
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
136 For each of the `catagories`, an item contains a list with all Unicode
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
137 characters with `cp_min` <= code-point <= `cp_max` that belong to
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
138 the category.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
139
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
140 The default values check every code-point supported by Python
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
141 (`sys.maxint` is 0x10FFFF in a "wide" build and 0xFFFF in a "narrow"
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
142 build, i.e. ucs4 and ucs2 respectively).
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
143 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
144 # Determine highest code point with one of the given categories
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
145 # (may shorten the search time considerably if there are many
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
146 # categories with not too high characters):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
147 if cp_max is None:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
148 cp_max = max(x for x in xrange(sys.maxunicode+1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
149 if unicodedata.category(unichr(x)) in categories)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
150 # print cp_max # => 74867 for unicode_punctuation_categories
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
151 charlists = {}
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
152 for cat in categories:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
153 charlists[cat] = [unichr(x) for x in xrange(cp_min, cp_max+1)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
154 if unicodedata.category(unichr(x)) == cat]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
155 return charlists
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
156
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
157
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
158 # Character categories in Docutils
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
159 # --------------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
160
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
161 def punctuation_samples():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
162
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
163 """Docutils punctuation category sample strings.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
164
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
165 Return list of sample strings for the categories "Open", "Close",
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
166 "Delimiters" and "Closing-Delimiters" used in the `inline markup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
167 recognition rules`_.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
168 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
169
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
170 # Lists with characters in Unicode punctuation character categories
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
171 cp_min = 160 # ASCII chars have special rules for backwards compatibility
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
172 ucharlists = unicode_charlists(unicode_punctuation_categories, cp_min)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
173
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
174 # match opening/closing characters
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
175 # --------------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
176 # Rearange the lists to ensure matching characters at the same
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
177 # index position.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
178
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
179 # low quotation marks are also used as closers (e.g. in Greek)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
180 # move them to category Pi:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
181 ucharlists['Ps'].remove(u'‚') # 201A SINGLE LOW-9 QUOTATION MARK
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
182 ucharlists['Ps'].remove(u'„') # 201E DOUBLE LOW-9 QUOTATION MARK
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
183 ucharlists['Pi'] += [u'‚', u'„']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
184
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
185 ucharlists['Pi'].remove(u'‛') # 201B SINGLE HIGH-REVERSED-9 QUOTATION MARK
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
186 ucharlists['Pi'].remove(u'‟') # 201F DOUBLE HIGH-REVERSED-9 QUOTATION MARK
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
187 ucharlists['Pf'] += [u'‛', u'‟']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
188
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
189 # 301F LOW DOUBLE PRIME QUOTATION MARK misses the opening pendant:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
190 ucharlists['Ps'].insert(ucharlists['Pe'].index(u'\u301f'), u'\u301d')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
191
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
192 # print u''.join(ucharlists['Ps']).encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
193 # print u''.join(ucharlists['Pe']).encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
194 # print u''.join(ucharlists['Pi']).encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
195 # print u''.join(ucharlists['Pf']).encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
196
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
197 # The Docutils character categories
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
198 # ---------------------------------
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
199 #
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
200 # The categorization of ASCII chars is non-standard to reduce
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
201 # both false positives and need for escaping. (see `inline markup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
202 # recognition rules`_)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
203
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
204 # allowed before markup if there is a matching closer
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
205 openers = [u'"\'(<\\[{']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
206 for cat in ('Ps', 'Pi', 'Pf'):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
207 openers.extend(ucharlists[cat])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
208
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
209 # allowed after markup if there is a matching opener
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
210 closers = [u'"\')>\\]}']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
211 for cat in ('Pe', 'Pf', 'Pi'):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
212 closers.extend(ucharlists[cat])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
213
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
214 # non-matching, allowed on both sides
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
215 delimiters = [u'\\-/:']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
216 for cat in ('Pd', 'Po'):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
217 delimiters.extend(ucharlists[cat])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
218
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
219 # non-matching, after markup
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
220 closing_delimiters = [r'\\.,;!?']
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
221
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
222 # # Test open/close matching:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
223 # for i in range(min(len(openers),len(closers))):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
224 # print '%4d %s %s' % (i, openers[i].encode('utf8'),
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
225 # closers[i].encode('utf8'))
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
226
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
227 return [u''.join(chars) for chars in (openers, closers, delimiters,
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
228 closing_delimiters)]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
229
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
230 def separate_wide_chars(s):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
231 """Return (s1,s2) with characters above 0xFFFF in s2"""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
232 maxunicode_narrow = 0xFFFF
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
233 l1 = [ch for ch in s if ord(ch) <= maxunicode_narrow]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
234 l2 = [ch for ch in s if ord(ch) > maxunicode_narrow]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
235 return ''.join(l1), ''.join(l2)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
236
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
237 def mark_intervals(s):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
238 """Return s with shortcut notation for runs of consecutive characters
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
239
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
240 Sort string and replace 'cdef' by 'c-f' and similar.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
241 """
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
242 l =[]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
243 s = [ord(ch) for ch in s]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
244 s.sort()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
245 for n in s:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
246 try:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
247 if l[-1][-1]+1 == n:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
248 l[-1].append(n)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
249 else:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
250 l.append([n])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
251 except IndexError:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
252 l.append([n])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
253
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
254 l2 = []
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
255 for i in l:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
256 i = [unichr(n) for n in i]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
257 if len(i) > 2:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
258 i = i[0], u'-', i[-1]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
259 l2.extend(i)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
260
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
261 return ''.join(l2)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
262
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
263 def wrap_string(s, startstring= "(",
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
264 endstring = ")", wrap=65):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
265 """Line-wrap a unicode string literal definition."""
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
266 c = len(startstring)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
267 contstring = "'\n" + ' ' * len(startstring) + "u'"
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
268 l = [startstring]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
269 for ch in s:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
270 c += 1
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
271 if ch == '\\' and c > wrap:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
272 c = len(startstring)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
273 ch = contstring + ch
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
274 l.append(ch)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
275 l.append(endstring)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
276 return ''.join(l)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
277
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
278
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
279 # print results
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
280 # =============
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
281
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
282 # (re) create and compare the samples:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
283
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
284 (o, c, d, cd) = punctuation_samples()
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
285 o, o_wide = separate_wide_chars(o)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
286 c, c_wide = separate_wide_chars(c)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
287 d, d_wide = separate_wide_chars(d)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
288 d = d[:5] + mark_intervals(d[5:])
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
289 d_wide = mark_intervals(d_wide)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
290 if sys.maxunicode >= 0x10FFFF: # "wide" build
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
291 d += d_wide
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
292 if o != openers:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
293 print '- openers = ur"""%s"""' % openers.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
294 print '+ openers = ur"""%s"""' % o.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
295 if o_wide:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
296 print '+ openers-wide = ur"""%s"""' % o_wide.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
297 if c != closers:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
298 print '- closers = ur"""%s"""' % closers.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
299 print '+ closers = ur"""%s"""' % c.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
300 if c_wide:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
301 print '+ closers-wide = ur"""%s"""' % c_wide.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
302 if d != delimiters:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
303 print '- delimiters = ur"%s"' % delimiters.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
304 print '+ delimiters = ur"%s"' % d.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
305 if cd != closing_delimiters:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
306 print '- closing_delimiters = ur"%s"' % closing_delimiters.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
307 print '+ closing_delimiters = ur"%s"' % cd.encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
308 # closing_delimiters are all ASCII characters
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
309
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
310 # Print literal code to define the character sets:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
311
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
312 # `openers` and `closers` must be verbose and keep order because they are
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
313 # also used in `match_chars()`.
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
314 print wrap_string(repr(o), startstring='openers = (')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
315 print wrap_string(repr(c), startstring='closers = (')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
316 # delimiters: sort and use shortcut for intervals (saves ~150 characters):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
317 print wrap_string(repr(d), startstring='delimiters = (')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
318 # add characters in the upper plane only in a "wide" build:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
319 print 'if sys.maxunicode >= 0x10FFFF: # "wide" build'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
320 print wrap_string(repr(d_wide), startstring=' delimiters += (')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
321 print 'closing_delimiters =', repr(cd)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
322
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
323 # test prints
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
324
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
325 # print "wide" Unicode characters:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
326 # ucharlists = unicode_charlists(unicode_punctuation_categories)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
327 # for key in ucharlists:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
328 # if key.endswith('wide'):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
329 # print key, ucharlists[key]
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
330
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
331 # print 'openers = ', repr(openers)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
332 # print 'closers = ', repr(closers)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
333 # print 'delimiters = ', repr(delimiters)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
334 # print 'closing_delimiters = ', repr(closing_delimiters)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
335
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
336 # ucharlists = unicode_charlists(unicode_punctuation_categories)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
337 # for cat, chars in ucharlists.items():
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
338 # # print cat, chars
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
339 # # compact output (visible with a comprehensive font):
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
340 # print (u":%s: %s" % (cat, u''.join(chars))).encode('utf8')
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
341
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
342 # verbose print
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
343
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
344 # print 'openers:'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
345 # for ch in openers:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
346 # print ch.encode('utf8'), unicodedata.name(ch)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
347 # print 'closers:'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
348 # for ch in closers:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
349 # print ch.encode('utf8'), unicodedata.name(ch)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
350 # print 'delimiters:'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
351 # for ch in delimiters:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
352 # print ch.encode('utf8'), unicodedata.name(ch)
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
353 # print 'closing_delimiters:'
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
354 # for ch in closing_delimiters:
d67268158946 planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff changeset
355 # print ch.encode('utf8'), unicodedata.name(ch)