1 | r""" |
---|
2 | Convert a restructured text document to html. |
---|
3 | |
---|
4 | Inline math markup can uses the *math* directive, or it can use latex |
---|
5 | style *\$expression\$*. Math can be rendered using simple html and |
---|
6 | unicode, or with mathjax. |
---|
7 | """ |
---|
8 | |
---|
9 | import re |
---|
10 | from contextlib import contextmanager |
---|
11 | |
---|
12 | # CRUFT: locale.getlocale() fails on some versions of OS X |
---|
13 | # See https://bugs.python.org/issue18378 |
---|
14 | import locale |
---|
15 | if hasattr(locale, '_parse_localename'): |
---|
16 | try: |
---|
17 | locale._parse_localename('UTF-8') |
---|
18 | except ValueError: |
---|
19 | _old_parse_localename = locale._parse_localename |
---|
20 | def _parse_localename(localename): |
---|
21 | code = locale.normalize(localename) |
---|
22 | if code == 'UTF-8': |
---|
23 | return None, code |
---|
24 | else: |
---|
25 | return _old_parse_localename(localename) |
---|
26 | locale._parse_localename = _parse_localename |
---|
27 | |
---|
28 | from docutils.core import publish_parts |
---|
29 | from docutils.writers.html4css1 import HTMLTranslator |
---|
30 | from docutils.nodes import SkipNode |
---|
31 | |
---|
32 | def wxview(html, url="", size=(850, 540)): |
---|
33 | import wx |
---|
34 | from wx.html2 import WebView |
---|
35 | frame = wx.Frame(None, -1, size=size) |
---|
36 | view = WebView.New(frame) |
---|
37 | view.SetPage(html, url) |
---|
38 | frame.Show() |
---|
39 | return frame |
---|
40 | |
---|
41 | def view_rst(filename): |
---|
42 | from os.path import expanduser |
---|
43 | with open(expanduser(filename)) as fid: |
---|
44 | rst = fid.read() |
---|
45 | html = rst2html(rst) |
---|
46 | wxview(html) |
---|
47 | |
---|
48 | def rst2html(rst, part="whole", math_output="mathjax"): |
---|
49 | r""" |
---|
50 | Convert restructured text into simple html. |
---|
51 | |
---|
52 | Valid *math_output* formats for formulas include: |
---|
53 | - html |
---|
54 | - mathml |
---|
55 | - mathjax |
---|
56 | See `http://docutils.sourceforge.net/docs/user/config.html#math-output`_ |
---|
57 | for details. |
---|
58 | |
---|
59 | The following *part* choices are available: |
---|
60 | - whole: the entire html document |
---|
61 | - html_body: document division with title and contents and footer |
---|
62 | - body: contents only |
---|
63 | |
---|
64 | There are other parts, but they don't make sense alone: |
---|
65 | |
---|
66 | subtitle, version, encoding, html_prolog, header, meta, |
---|
67 | html_title, title, stylesheet, html_subtitle, html_body, |
---|
68 | body, head, body_suffix, fragment, docinfo, html_head, |
---|
69 | head_prefix, body_prefix, footer, body_pre_docinfo, whole |
---|
70 | """ |
---|
71 | # Ick! mathjax doesn't work properly with math-output, and the |
---|
72 | # others don't work properly with math_output! |
---|
73 | if math_output == "mathjax": |
---|
74 | settings = {"math_output": math_output} |
---|
75 | else: |
---|
76 | settings = {"math-output": math_output} |
---|
77 | |
---|
78 | # TODO: support stylesheets |
---|
79 | #html_root = "/full/path/to/_static/" |
---|
80 | #sheets = [html_root+s for s in ["basic.css","classic.css"]] |
---|
81 | #settings["embed_styesheet"] = True |
---|
82 | #settings["stylesheet_path"] = sheets |
---|
83 | |
---|
84 | # math2html and mathml do not support \frac12 |
---|
85 | rst = replace_compact_fraction(rst) |
---|
86 | |
---|
87 | # mathml, html do not support \tfrac |
---|
88 | if math_output in ("mathml", "html"): |
---|
89 | rst = rst.replace(r'\tfrac', r'\frac') |
---|
90 | |
---|
91 | rst = replace_dollar(rst) |
---|
92 | with suppress_html_errors(): |
---|
93 | parts = publish_parts(source=rst, writer_name='html', |
---|
94 | settings_overrides=settings) |
---|
95 | return parts[part] |
---|
96 | |
---|
97 | @contextmanager |
---|
98 | def suppress_html_errors(): |
---|
99 | r""" |
---|
100 | Context manager for keeping error reports out of the generated HTML. |
---|
101 | |
---|
102 | Within the context, system message nodes in the docutils parse tree |
---|
103 | will be ignored. After the context, the usual behaviour will be restored. |
---|
104 | """ |
---|
105 | visit_system_message = HTMLTranslator.visit_system_message |
---|
106 | HTMLTranslator.visit_system_message = _skip_node |
---|
107 | yield None |
---|
108 | HTMLTranslator.visit_system_message = visit_system_message |
---|
109 | |
---|
110 | def _skip_node(self, node): |
---|
111 | raise SkipNode |
---|
112 | |
---|
113 | |
---|
114 | _compact_fraction = re.compile(r"(\\[cdt]?frac)([0-9])([0-9])") |
---|
115 | def replace_compact_fraction(content): |
---|
116 | r""" |
---|
117 | Convert \frac12 to \frac{1}{2} for broken latex parsers |
---|
118 | """ |
---|
119 | return _compact_fraction.sub(r"\1{\2}{\3}", content) |
---|
120 | |
---|
121 | |
---|
122 | _dollar = re.compile(r"(?:^|(?<=\s|[(]))[$]([^\n]*?)(?<![\\])[$](?:$|(?=\s|[.,;)\\]))") |
---|
123 | _notdollar = re.compile(r"\\[$]") |
---|
124 | def replace_dollar(content): |
---|
125 | r""" |
---|
126 | Convert dollar signs to inline math markup in rst. |
---|
127 | """ |
---|
128 | content = _dollar.sub(r":math:`\1`", content) |
---|
129 | content = _notdollar.sub("$", content) |
---|
130 | return content |
---|
131 | |
---|
132 | |
---|
133 | def test_dollar(): |
---|
134 | """ |
---|
135 | Test substitution of dollar signs with equivalent RST math markup |
---|
136 | """ |
---|
137 | assert replace_dollar(u"no dollar") == u"no dollar" |
---|
138 | assert replace_dollar(u"$only$") == u":math:`only`" |
---|
139 | assert replace_dollar(u"$first$ is good") == u":math:`first` is good" |
---|
140 | assert replace_dollar(u"so is $last$") == u"so is :math:`last`" |
---|
141 | assert replace_dollar(u"and $mid$ too") == u"and :math:`mid` too" |
---|
142 | assert replace_dollar(u"$first$, $mid$, $last$") == u":math:`first`, :math:`mid`, :math:`last`" |
---|
143 | assert replace_dollar(u"dollar\\$ escape") == u"dollar$ escape" |
---|
144 | assert replace_dollar(u"dollar \\$escape\\$ too") == u"dollar $escape$ too" |
---|
145 | assert replace_dollar(u"spaces $in the$ math") == u"spaces :math:`in the` math" |
---|
146 | assert replace_dollar(u"emb\\ $ed$\\ ed") == u"emb\\ :math:`ed`\\ ed" |
---|
147 | assert replace_dollar(u"$first$a") == u"$first$a" |
---|
148 | assert replace_dollar(u"a$last$") == u"a$last$" |
---|
149 | assert replace_dollar(u"$37") == u"$37" |
---|
150 | assert replace_dollar(u"($37)") == u"($37)" |
---|
151 | assert replace_dollar(u"$37 - $43") == u"$37 - $43" |
---|
152 | assert replace_dollar(u"($37, $38)") == u"($37, $38)" |
---|
153 | assert replace_dollar(u"a $mid$dle a") == u"a $mid$dle a" |
---|
154 | assert replace_dollar(u"a ($in parens$) a") == u"a (:math:`in parens`) a" |
---|
155 | assert replace_dollar(u"a (again $in parens$) a") == u"a (again :math:`in parens`) a" |
---|
156 | |
---|
157 | def view_rst_app(filename): |
---|
158 | import wx # type: ignore |
---|
159 | app = wx.App() |
---|
160 | view_rst(filename) |
---|
161 | app.MainLoop() |
---|
162 | |
---|
163 | |
---|
164 | if __name__ == "__main__": |
---|
165 | import sys |
---|
166 | view_rst_app(sys.argv[1]) |
---|
167 | |
---|