1
|
|
|
from __future__ import unicode_literals |
2
|
|
|
|
3
|
|
|
|
4
|
|
|
import re |
5
|
|
|
from builtins import str |
6
|
|
|
from CommonMark.common import escape_xml |
7
|
|
|
from CommonMark.render.renderer import Renderer |
8
|
|
|
|
9
|
|
|
|
10
|
|
|
reUnsafeProtocol = re.compile( |
11
|
|
|
r'^javascript:|vbscript:|file:|data:', re.IGNORECASE) |
12
|
|
|
reSafeDataProtocol = re.compile( |
13
|
|
|
r'^data:image\/(?:png|gif|jpeg|webp)', re.IGNORECASE) |
14
|
|
|
|
15
|
|
|
|
16
|
|
|
def potentially_unsafe(url): |
17
|
|
|
return re.match(reUnsafeProtocol, url) and \ |
18
|
|
|
(not re.match(reSafeDataProtocol, url)) |
19
|
|
|
|
20
|
|
|
|
21
|
|
|
class HtmlRenderer(Renderer): |
22
|
|
|
def __init__(self, options={}): |
23
|
|
|
# by default, soft breaks are rendered as newlines in HTML |
24
|
|
|
options['softbreak'] = options.get('softbreak') or '\n' |
25
|
|
|
# set to "<br />" to make them hard breaks |
26
|
|
|
# set to " " if you want to ignore line wrapping in source |
27
|
|
|
|
28
|
|
|
self.disable_tags = 0 |
29
|
|
|
self.last_out = '\n' |
30
|
|
|
self.options = options |
31
|
|
|
|
32
|
|
|
def escape(self, text, preserve_entities): |
33
|
|
|
return escape_xml(text, preserve_entities) |
34
|
|
|
|
35
|
|
|
def tag(self, name, attrs=None, selfclosing=None): |
36
|
|
|
"""Helper function to produce an HTML tag.""" |
37
|
|
|
if self.disable_tags > 0: |
38
|
|
|
return |
39
|
|
|
|
40
|
|
|
self.buf += '<' + name |
41
|
|
|
if attrs and len(attrs) > 0: |
42
|
|
|
for attrib in attrs: |
43
|
|
|
self.buf += ' ' + attrib[0] + '="' + attrib[1] + '"' |
44
|
|
|
|
45
|
|
|
if selfclosing: |
46
|
|
|
self.buf += ' /' |
47
|
|
|
|
48
|
|
|
self.buf += '>' |
49
|
|
|
self.last_out = '>' |
50
|
|
|
|
51
|
|
|
# Node methods # |
52
|
|
|
|
53
|
|
|
def text(self, node, entering=None): |
54
|
|
|
self.out(node.literal) |
55
|
|
|
|
56
|
|
|
def softbreak(self, node=None, entering=None): |
57
|
|
|
self.lit(self.options['softbreak']) |
58
|
|
|
|
59
|
|
|
def linebreak(self, node=None, entering=None): |
60
|
|
|
self.tag('br', [], True) |
61
|
|
|
self.cr() |
62
|
|
|
|
63
|
|
|
def link(self, node, entering): |
64
|
|
|
attrs = self.attrs(node) |
65
|
|
|
if entering: |
66
|
|
|
if not (self.options.get('safe') and |
67
|
|
|
potentially_unsafe(node.destination)): |
68
|
|
|
attrs.append(['href', self.escape(node.destination, True)]) |
69
|
|
|
|
70
|
|
|
if node.title: |
71
|
|
|
attrs.append(['title', self.escape(node.title, True)]) |
72
|
|
|
|
73
|
|
|
self.tag('a', attrs) |
74
|
|
|
else: |
75
|
|
|
self.tag('/a') |
76
|
|
|
|
77
|
|
|
def image(self, node, entering): |
78
|
|
|
if entering: |
79
|
|
|
if self.disable_tags == 0: |
80
|
|
|
if self.options.get('safe') and \ |
81
|
|
|
potentially_unsafe(node.destination): |
82
|
|
|
self.lit('<img src="" alt="') |
83
|
|
|
else: |
84
|
|
|
self.lit('<img src="' + |
85
|
|
|
self.escape(node.destination, True) + |
86
|
|
|
'" alt="') |
87
|
|
|
self.disable_tags += 1 |
88
|
|
|
else: |
89
|
|
|
self.disable_tags -= 1 |
90
|
|
|
if self.disable_tags == 0: |
91
|
|
|
if node.title: |
92
|
|
|
self.lit('" title="' + self.escape(node.title, True)) |
93
|
|
|
self.lit('" />') |
94
|
|
|
|
95
|
|
|
def emph(self, node, entering): |
96
|
|
|
self.tag('em' if entering else '/em') |
97
|
|
|
|
98
|
|
|
def strong(self, node, entering): |
99
|
|
|
self.tag('strong' if entering else '/strong') |
100
|
|
|
|
101
|
|
|
def paragraph(self, node, entering): |
102
|
|
|
grandparent = node.parent.parent |
103
|
|
|
attrs = self.attrs(node) |
104
|
|
|
if grandparent is not None and grandparent.t == 'list': |
105
|
|
|
if grandparent.list_data['tight']: |
106
|
|
|
return |
107
|
|
|
|
108
|
|
|
if entering: |
109
|
|
|
self.cr() |
110
|
|
|
self.tag('p', attrs) |
111
|
|
|
else: |
112
|
|
|
self.tag('/p') |
113
|
|
|
self.cr() |
114
|
|
|
|
115
|
|
|
def heading(self, node, entering): |
116
|
|
|
tagname = 'h' + str(node.level) |
117
|
|
|
attrs = self.attrs(node) |
118
|
|
|
if entering: |
119
|
|
|
self.cr() |
120
|
|
|
self.tag(tagname, attrs) |
121
|
|
|
else: |
122
|
|
|
self.tag('/' + tagname) |
123
|
|
|
self.cr() |
124
|
|
|
|
125
|
|
|
def code(self, node, entering): |
126
|
|
|
self.tag('code') |
127
|
|
|
self.out(node.literal) |
128
|
|
|
self.tag('/code') |
129
|
|
|
|
130
|
|
|
def code_block(self, node, entering): |
131
|
|
|
info_words = node.info.split() if node.info else [] |
132
|
|
|
attrs = self.attrs(node) |
133
|
|
|
if len(info_words) > 0 and len(info_words[0]) > 0: |
134
|
|
|
attrs.append(['class', 'language-' + |
135
|
|
|
self.escape(info_words[0], True)]) |
136
|
|
|
|
137
|
|
|
self.cr() |
138
|
|
|
self.tag('pre') |
139
|
|
|
self.tag('code', attrs) |
140
|
|
|
self.out(node.literal) |
141
|
|
|
self.tag('/code') |
142
|
|
|
self.tag('/pre') |
143
|
|
|
self.cr() |
144
|
|
|
|
145
|
|
|
def thematic_break(self, node, entering): |
146
|
|
|
attrs = self.attrs(node) |
147
|
|
|
self.cr() |
148
|
|
|
self.tag('hr', attrs, True) |
149
|
|
|
self.cr() |
150
|
|
|
|
151
|
|
|
def block_quote(self, node, entering): |
152
|
|
|
attrs = self.attrs(node) |
153
|
|
|
if entering: |
154
|
|
|
self.cr() |
155
|
|
|
self.tag('blockquote', attrs) |
156
|
|
|
self.cr() |
157
|
|
|
else: |
158
|
|
|
self.cr() |
159
|
|
|
self.tag('/blockquote') |
160
|
|
|
self.cr() |
161
|
|
|
|
162
|
|
|
def list(self, node, entering): |
163
|
|
|
tagname = 'ul' if node.list_data['type'] == 'bullet' else 'ol' |
164
|
|
|
attrs = self.attrs(node) |
165
|
|
|
if entering: |
166
|
|
|
start = node.list_data['start'] |
167
|
|
|
if start is not None and start != 1: |
168
|
|
|
attrs.append(['start', str(start)]) |
169
|
|
|
|
170
|
|
|
self.cr() |
171
|
|
|
self.tag(tagname, attrs) |
172
|
|
|
self.cr() |
173
|
|
|
else: |
174
|
|
|
self.cr() |
175
|
|
|
self.tag('/' + tagname) |
176
|
|
|
self.cr() |
177
|
|
|
|
178
|
|
|
def item(self, node, entering): |
179
|
|
|
attrs = self.attrs(node) |
180
|
|
|
if entering: |
181
|
|
|
self.tag('li', attrs) |
182
|
|
|
else: |
183
|
|
|
self.tag('/li') |
184
|
|
|
self.cr() |
185
|
|
|
|
186
|
|
|
def html_inline(self, node, entering): |
187
|
|
|
if self.options.get('safe'): |
188
|
|
|
self.lit('<!-- raw HTML omitted -->') |
189
|
|
|
else: |
190
|
|
|
self.lit(node.literal) |
191
|
|
|
|
192
|
|
|
def html_block(self, node, entering): |
193
|
|
|
self.cr() |
194
|
|
|
if self.options.get('safe'): |
195
|
|
|
self.lit('<!-- raw HTML omitted -->') |
196
|
|
|
else: |
197
|
|
|
self.lit(node.literal) |
198
|
|
|
self.cr() |
199
|
|
|
|
200
|
|
|
def custom_inline(self, node, entering): |
201
|
|
|
if entering and node.on_enter: |
202
|
|
|
self.lit(node.on_enter) |
203
|
|
|
elif (not entering) and node.on_exit: |
204
|
|
|
self.lit(node.on_exit) |
205
|
|
|
|
206
|
|
|
def custom_block(self, node, entering): |
207
|
|
|
self.cr() |
208
|
|
|
if entering and node.on_enter: |
209
|
|
|
self.lit(node.on_enter) |
210
|
|
|
elif (not entering) and node.on_exit: |
211
|
|
|
self.lit(node.on_exit) |
212
|
|
|
self.cr() |
213
|
|
|
|
214
|
|
|
# Helper methods # |
215
|
|
|
|
216
|
|
|
def out(self, s): |
217
|
|
|
self.lit(self.escape(s, False)) |
218
|
|
|
|
219
|
|
|
def attrs(self, node): |
220
|
|
|
att = [] |
221
|
|
|
if self.options.get('sourcepos'): |
222
|
|
|
pos = node.sourcepos |
223
|
|
|
if pos: |
224
|
|
|
att.append(['data-sourcepos', str(pos[0][0]) + ':' + |
225
|
|
|
str(pos[0][1]) + '-' + str(pos[1][0]) + ':' + |
226
|
|
|
str(pos[1][1])]) |
227
|
|
|
|
228
|
|
|
return att |
229
|
|
|
|