html2doc 1.1.0 → 1.1.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.github/workflows/rake.yml +1 -11
- data/.hound.yml +3 -1
- data/.rubocop.yml +4 -8
- data/Gemfile +2 -2
- data/Rakefile +1 -1
- data/bin/html2doc +1 -2
- data/bin/rspec +1 -1
- data/html2doc.gemspec +8 -9
- data/lib/html2doc/base.rb +48 -46
- data/lib/html2doc/lists.rb +47 -42
- data/lib/html2doc/math.rb +100 -79
- data/lib/html2doc/mime.rb +41 -34
- data/lib/html2doc/notes.rb +42 -36
- data/lib/html2doc/version.rb +1 -1
- data/lib/html2doc.rb +0 -3
- data/spec/html2doc_spec.rb +566 -521
- metadata +42 -42
data/spec/html2doc_spec.rb
CHANGED
|
@@ -1,275 +1,276 @@
|
|
|
1
1
|
require "base64"
|
|
2
2
|
|
|
3
|
-
def html_input(
|
|
3
|
+
def html_input(xml)
|
|
4
4
|
<<~HTML
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
|
|
5
|
+
<html><head><title>blank</title>
|
|
6
|
+
<meta name="Originator" content="Me"/>
|
|
7
|
+
</head>
|
|
8
|
+
<body>
|
|
9
|
+
#{xml}
|
|
10
|
+
</body></html>
|
|
11
11
|
HTML
|
|
12
12
|
end
|
|
13
13
|
|
|
14
|
-
def html_input_no_title(
|
|
14
|
+
def html_input_no_title(xml)
|
|
15
15
|
<<~HTML
|
|
16
|
-
|
|
17
|
-
|
|
18
|
-
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
16
|
+
<html><head>
|
|
17
|
+
<meta name="Originator" content="Me"/>
|
|
18
|
+
</head>
|
|
19
|
+
<body>
|
|
20
|
+
#{xml}
|
|
21
|
+
</body></html>
|
|
22
22
|
HTML
|
|
23
23
|
end
|
|
24
24
|
|
|
25
|
-
def html_input_empty_head(
|
|
25
|
+
def html_input_empty_head(xml)
|
|
26
26
|
<<~HTML
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
27
|
+
<html><head></head>
|
|
28
|
+
<body>
|
|
29
|
+
#{xml}
|
|
30
|
+
</body></html>
|
|
31
31
|
HTML
|
|
32
32
|
end
|
|
33
33
|
|
|
34
|
-
WORD_HDR = <<~HDR
|
|
35
|
-
MIME-Version: 1.0
|
|
36
|
-
Content-Type: multipart/related; boundary="----=_NextPart_"
|
|
37
|
-
|
|
38
|
-
------=_NextPart_
|
|
39
|
-
Content-ID: <test.htm>
|
|
40
|
-
Content-Disposition: inline; filename="test.htm"
|
|
41
|
-
Content-Type: text/html; charset="utf-8"
|
|
42
|
-
|
|
43
|
-
<?xml version="1.0"?>
|
|
44
|
-
<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:v="urn:schemas-microsoft-com:vml" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><!--[if gte mso 9]>
|
|
45
|
-
<xml>
|
|
46
|
-
<w:WordDocument>
|
|
47
|
-
<w:View>Print</w:View>
|
|
48
|
-
<w:Zoom>100</w:Zoom>
|
|
49
|
-
<w:DoNotOptimizeForBrowser/>
|
|
50
|
-
</w:WordDocument>
|
|
51
|
-
</xml>
|
|
52
|
-
<![endif]-->
|
|
53
|
-
<meta http-equiv=Content-Type content="text/html; charset=utf-8"/>
|
|
54
|
-
|
|
55
|
-
|
|
56
|
-
<title>blank</title><style><![CDATA[
|
|
57
|
-
|
|
34
|
+
WORD_HDR = <<~HDR.freeze
|
|
35
|
+
MIME-Version: 1.0
|
|
36
|
+
Content-Type: multipart/related; boundary="----=_NextPart_"
|
|
37
|
+
|
|
38
|
+
------=_NextPart_
|
|
39
|
+
Content-ID: <test.htm>
|
|
40
|
+
Content-Disposition: inline; filename="test.htm"
|
|
41
|
+
Content-Type: text/html; charset="utf-8"
|
|
42
|
+
|
|
43
|
+
<?xml version="1.0"?>
|
|
44
|
+
<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:v="urn:schemas-microsoft-com:vml" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><!--[if gte mso 9]>
|
|
45
|
+
<xml>
|
|
46
|
+
<w:WordDocument>
|
|
47
|
+
<w:View>Print</w:View>
|
|
48
|
+
<w:Zoom>100</w:Zoom>
|
|
49
|
+
<w:DoNotOptimizeForBrowser/>
|
|
50
|
+
</w:WordDocument>
|
|
51
|
+
</xml>
|
|
52
|
+
<![endif]-->
|
|
53
|
+
<meta http-equiv=Content-Type content="text/html; charset=utf-8"/>
|
|
54
|
+
|
|
55
|
+
<link rel=File-List href="cid:filelist.xml"/>
|
|
56
|
+
<title>blank</title><style><![CDATA[
|
|
57
|
+
<!--
|
|
58
58
|
HDR
|
|
59
59
|
|
|
60
|
-
WORD_HDR_END = <<~HDR
|
|
61
|
-
-->
|
|
62
|
-
]]></style>
|
|
63
|
-
<meta name="Originator" content="Me"/>
|
|
64
|
-
</head>
|
|
60
|
+
WORD_HDR_END = <<~HDR.freeze
|
|
61
|
+
-->
|
|
62
|
+
]]></style>
|
|
63
|
+
<meta name="Originator" content="Me"/>
|
|
64
|
+
</head>
|
|
65
65
|
HDR
|
|
66
66
|
|
|
67
|
-
def word_body(
|
|
67
|
+
def word_body(xml, footnote)
|
|
68
68
|
<<~BODY
|
|
69
|
-
<body>
|
|
70
|
-
|
|
71
|
-
|
|
69
|
+
<body>
|
|
70
|
+
#{xml}
|
|
71
|
+
#{footnote}</body></html>
|
|
72
72
|
BODY
|
|
73
73
|
end
|
|
74
74
|
|
|
75
|
-
WORD_FTR1 = <<~FTR
|
|
75
|
+
WORD_FTR1 = <<~FTR.freeze
|
|
76
76
|
------=_NextPart_
|
|
77
|
-
Content-ID: <filelist.xml>
|
|
78
|
-
Content-Disposition: inline; filename="filelist.xml"
|
|
79
|
-
Content-Transfer-Encoding: base64
|
|
80
|
-
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
77
|
+
Content-ID: <filelist.xml>
|
|
78
|
+
Content-Disposition: inline; filename="filelist.xml"
|
|
79
|
+
Content-Transfer-Encoding: base64
|
|
80
|
+
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
81
81
|
|
|
82
|
-
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
83
|
-
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
|
|
84
|
-
bGVsaXN0LnhtbCIvPgo8L3htbD4K
|
|
82
|
+
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
83
|
+
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
|
|
84
|
+
bGVsaXN0LnhtbCIvPgo8L3htbD4K
|
|
85
85
|
|
|
86
|
-
------=_NextPart_--
|
|
86
|
+
------=_NextPart_--
|
|
87
87
|
FTR
|
|
88
88
|
|
|
89
|
-
WORD_FTR2 = <<~FTR
|
|
89
|
+
WORD_FTR2 = <<~FTR.freeze
|
|
90
|
+
------=_NextPart_
|
|
91
|
+
Content-ID: <filelist.xml>
|
|
92
|
+
Content-Disposition: inline; filename="filelist.xml"
|
|
93
|
+
Content-Transfer-Encoding: base64
|
|
94
|
+
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
95
|
+
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
96
|
+
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
|
|
97
|
+
bGVsaXN0LnhtbCIvPgogIDxvOkZpbGUgSFJlZj0iaGVhZGVyLmh0bWwiLz4KPC94bWw+Cg==
|
|
90
98
|
------=_NextPart_
|
|
91
|
-
Content-ID: <filelist.xml>
|
|
92
|
-
Content-Disposition: inline; filename="filelist.xml"
|
|
93
|
-
Content-Transfer-Encoding: base64
|
|
94
|
-
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
95
|
-
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
96
|
-
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9ImZp
|
|
97
|
-
bGVsaXN0LnhtbCIvPgogIDxvOkZpbGUgSFJlZj0iaGVhZGVyLmh0bWwiLz4KPC94bWw+Cg==
|
|
98
|
-
------=_NextPart_
|
|
99
99
|
FTR
|
|
100
100
|
|
|
101
|
-
WORD_FTR3 = <<~FTR
|
|
102
|
-
------=_NextPart_
|
|
103
|
-
Content-ID: <filelist.xml>
|
|
104
|
-
Content-Disposition: inline; filename="filelist.xml"
|
|
105
|
-
Content-Transfer-Encoding: base64
|
|
106
|
-
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
107
|
-
|
|
108
|
-
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
109
|
-
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9IjFh
|
|
110
|
-
YzIwNjVmLTAzZjAtNGM3YS1iOWE2LTkyZTgyMDU5MWJmMC5wbmciLz4KICA8bzpGaWxlIEhSZWY9
|
|
111
|
-
ImZpbGVsaXN0LnhtbCIvPgo8L3htbD4K
|
|
112
|
-
------=_NextPart_
|
|
113
|
-
Content-ID: <cb7b0d19-891e-4634-815a-570d019d454c.png>
|
|
114
|
-
Content-Disposition: inline; filename="cb7b0d19-891e-4634-815a-570d019d454c.png"
|
|
115
|
-
Content-Transfer-Encoding: base64
|
|
116
|
-
Content-Type: image/png
|
|
117
|
-
------=_NextPart_--
|
|
101
|
+
WORD_FTR3 = <<~FTR.freeze
|
|
102
|
+
------=_NextPart_
|
|
103
|
+
Content-ID: <filelist.xml>
|
|
104
|
+
Content-Disposition: inline; filename="filelist.xml"
|
|
105
|
+
Content-Transfer-Encoding: base64
|
|
106
|
+
Content-Type: #{Html2Doc::mime_type('filelist.xml')}
|
|
107
|
+
|
|
108
|
+
PHhtbCB4bWxuczpvPSJ1cm46c2NoZW1hcy1taWNyb3NvZnQtY29tOm9mZmljZTpvZmZpY2UiPgog
|
|
109
|
+
ICAgICAgIDxvOk1haW5GaWxlIEhSZWY9Ii4uL3Rlc3QuaHRtIi8+ICA8bzpGaWxlIEhSZWY9IjFh
|
|
110
|
+
YzIwNjVmLTAzZjAtNGM3YS1iOWE2LTkyZTgyMDU5MWJmMC5wbmciLz4KICA8bzpGaWxlIEhSZWY9
|
|
111
|
+
ImZpbGVsaXN0LnhtbCIvPgo8L3htbD4K
|
|
112
|
+
------=_NextPart_
|
|
113
|
+
Content-ID: <cb7b0d19-891e-4634-815a-570d019d454c.png>
|
|
114
|
+
Content-Disposition: inline; filename="cb7b0d19-891e-4634-815a-570d019d454c.png"
|
|
115
|
+
Content-Transfer-Encoding: base64
|
|
116
|
+
Content-Type: image/png
|
|
117
|
+
------=_NextPart_--
|
|
118
118
|
FTR
|
|
119
119
|
|
|
120
|
-
HEADERHTML
|
|
121
|
-
<html xmlns:v="urn:schemas-microsoft-com:vml"
|
|
122
|
-
xmlns:o="urn:schemas-microsoft-com:office:office"
|
|
123
|
-
xmlns:w="urn:schemas-microsoft-com:office:word"
|
|
124
|
-
xmlns:m="http://schemas.microsoft.com/office/2004/12/omml"
|
|
125
|
-
xmlns:mv="http://macVmlSchemaUri" xmlns="http://www.w3.org/TR/REC-html40">
|
|
126
|
-
<head>
|
|
127
|
-
<meta name=Title content="">
|
|
128
|
-
<meta name=Keywords content="">
|
|
129
|
-
<meta http-equiv=Content-Type content="text/html; charset=utf-8">
|
|
130
|
-
<meta name=ProgId content=Word.Document>
|
|
131
|
-
<meta name=Generator content="Microsoft Word 15">
|
|
132
|
-
<meta name=Originator content="Microsoft Word 15">
|
|
133
|
-
<link id=Main-File rel=Main-File href="FILENAME.html">
|
|
134
|
-
<!--[if gte mso 9]><xml>
|
|
135
|
-
<o:shapedefaults v:ext="edit" spidmax="2049"/>
|
|
136
|
-
</xml><![endif]-->
|
|
137
|
-
</head>
|
|
138
|
-
<body lang=EN link=blue vlink="#954F72">
|
|
139
|
-
<div style='mso-element:footnote-separator' id=fs>
|
|
140
|
-
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
141
|
-
normal'><span lang=EN-GB><span style='mso-special-character:footnote-separator'><![if !supportFootnotes]>
|
|
142
|
-
<hr align=left size=1 width="33%">
|
|
143
|
-
<![endif]></span></span></p>
|
|
144
|
-
</div>
|
|
145
|
-
<div style='mso-element:footnote-continuation-separator' id=fcs>
|
|
146
|
-
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
147
|
-
normal'><span lang=EN-GB><span style='mso-special-character:footnote-continuation-separator'><![if !supportFootnotes]>
|
|
148
|
-
<hr align=left size=1>
|
|
149
|
-
<![endif]></span></span></p>
|
|
150
|
-
</div>
|
|
151
|
-
<div style='mso-element:endnote-separator' id=es>
|
|
152
|
-
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
153
|
-
normal'><span lang=EN-GB><span style='mso-special-character:footnote-separator'><![if !supportFootnotes]>
|
|
154
|
-
<hr align=left size=1 width="33%">
|
|
155
|
-
<![endif]></span></span></p>
|
|
156
|
-
</div>
|
|
157
|
-
<div style='mso-element:endnote-continuation-separator' id=ecs>
|
|
158
|
-
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
159
|
-
normal'><span lang=EN-GB><span style='mso-special-character:footnote-continuation-separator'><![if !supportFootnotes]>
|
|
160
|
-
<hr align=left size=1>
|
|
161
|
-
<![endif]></span></span></p>
|
|
162
|
-
</div>
|
|
163
|
-
<div style='mso-element:header' id=eh1>
|
|
164
|
-
<p class=MsoHeader align=left style='text-align:left;line-height:12.0pt;
|
|
165
|
-
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
166
|
-
</div>
|
|
167
|
-
<div style='mso-element:header' id=h1>
|
|
168
|
-
<p class=MsoHeader style='margin-bottom:18.0pt'><span lang=EN-GB
|
|
169
|
-
style='font-size:10.0pt;mso-bidi-font-size:11.0pt;font-weight:normal'>©
|
|
170
|
-
ISO/IEC&nbsp;2016&nbsp;– All rights reserved</span><span lang=EN-GB
|
|
171
|
-
style='font-weight:normal'><o:p></o:p></span></p>
|
|
172
|
-
</div>
|
|
173
|
-
<div style='mso-element:footer' id=ef1>
|
|
174
|
-
<p class=MsoFooter style='margin-top:12.0pt;line-height:12.0pt;mso-line-height-rule:
|
|
175
|
-
exactly'><!--[if supportFields]><b style='mso-bidi-font-weight:normal'><span
|
|
176
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
177
|
-
style='mso-element:field-begin'></span><span
|
|
178
|
-
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
179
|
-
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
180
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
181
|
-
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>2</span></span></b><!--[if supportFields]><b
|
|
182
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
183
|
-
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
184
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
185
|
-
style='mso-tab-count:1'>                                                                                                                                                                           </span>©
|
|
186
|
-
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
187
|
-
</div>
|
|
188
|
-
<div style='mso-element:header' id=eh2>
|
|
189
|
-
<p class=MsoHeader align=left style='text-align:left;line-height:12.0pt;
|
|
190
|
-
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
191
|
-
</div>
|
|
192
|
-
<div style='mso-element:header' id=h2>
|
|
193
|
-
<p class=MsoHeader align=right style='text-align:right;line-height:12.0pt;
|
|
194
|
-
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
195
|
-
</div>
|
|
196
|
-
<div style='mso-element:footer' id=ef2>
|
|
197
|
-
<p class=MsoFooter style='line-height:12.0pt;mso-line-height-rule:exactly'><!--[if supportFields]><span
|
|
198
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
199
|
-
style='mso-element:field-begin'></span><span
|
|
200
|
-
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
201
|
-
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span><![endif]--><span
|
|
202
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
203
|
-
style='mso-no-proof:yes'>ii</span></span><!--[if supportFields]><span
|
|
204
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
205
|
-
style='mso-element:field-end'></span></span><![endif]--><span lang=EN-GB
|
|
206
|
-
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span style='mso-tab-count:
|
|
207
|
-
1'>                                                                                                                                                                           </span>©
|
|
208
|
-
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
209
|
-
</div>
|
|
210
|
-
<div style='mso-element:footer' id=f2>
|
|
211
|
-
<p class=MsoFooter style='line-height:12.0pt'><span lang=EN-GB
|
|
212
|
-
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'>© ISO/IEC&nbsp;2016&nbsp;– All
|
|
213
|
-
rights reserved<span style='mso-tab-count:1'>                                                                                                                                                                          </span></span><!--[if supportFields]><span
|
|
214
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
215
|
-
style='mso-element:field-begin'></span> PAGE<span style='mso-spacerun:yes'>  
|
|
216
|
-
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span><![endif]--><span
|
|
217
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
218
|
-
style='mso-no-proof:yes'>iii</span></span><!--[if supportFields]><span
|
|
219
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
220
|
-
style='mso-element:field-end'></span></span><![endif]--><span lang=EN-GB
|
|
221
|
-
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><o:p></o:p></span></p>
|
|
222
|
-
</div>
|
|
223
|
-
<div style='mso-element:footer' id=ef3>
|
|
224
|
-
<p class=MsoFooter style='margin-top:12.0pt;line-height:12.0pt;mso-line-height-rule:
|
|
225
|
-
exactly'><!--[if supportFields]><b style='mso-bidi-font-weight:normal'><span
|
|
226
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
227
|
-
style='mso-element:field-begin'></span><span
|
|
228
|
-
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
229
|
-
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
230
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
231
|
-
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>2</span></span></b><!--[if supportFields]><b
|
|
232
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
233
|
-
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
234
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
235
|
-
style='mso-tab-count:1'>                                                                                                                                                                           </span>©
|
|
236
|
-
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
237
|
-
</div>
|
|
238
|
-
<div style='mso-element:footer' id=f3>
|
|
239
|
-
<p class=MsoFooter style='line-height:12.0pt'><span lang=EN-GB
|
|
240
|
-
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'>© ISO/IEC&nbsp;2016&nbsp;– All
|
|
241
|
-
rights reserved<span style='mso-tab-count:1'>                                                                                                                                                                           </span></span><!--[if supportFields]><b
|
|
242
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
243
|
-
mso-bidi-font-size:11.0pt'><span style='mso-element:field-begin'></span>
|
|
244
|
-
PAGE<span style='mso-spacerun:yes'>   </span>\\* MERGEFORMAT <span
|
|
245
|
-
style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
246
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
247
|
-
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>3</span></span></b><!--[if supportFields]><b
|
|
248
|
-
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
249
|
-
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
250
|
-
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><o:p></o:p></span></p>
|
|
251
|
-
</div>
|
|
252
|
-
</body>
|
|
253
|
-
</html>
|
|
120
|
+
HEADERHTML = <<~FTR.freeze
|
|
121
|
+
<html xmlns:v="urn:schemas-microsoft-com:vml"
|
|
122
|
+
xmlns:o="urn:schemas-microsoft-com:office:office"
|
|
123
|
+
xmlns:w="urn:schemas-microsoft-com:office:word"
|
|
124
|
+
xmlns:m="http://schemas.microsoft.com/office/2004/12/omml"
|
|
125
|
+
xmlns:mv="http://macVmlSchemaUri" xmlns="http://www.w3.org/TR/REC-html40">
|
|
126
|
+
<head>
|
|
127
|
+
<meta name=Title content="">
|
|
128
|
+
<meta name=Keywords content="">
|
|
129
|
+
<meta http-equiv=Content-Type content="text/html; charset=utf-8">
|
|
130
|
+
<meta name=ProgId content=Word.Document>
|
|
131
|
+
<meta name=Generator content="Microsoft Word 15">
|
|
132
|
+
<meta name=Originator content="Microsoft Word 15">
|
|
133
|
+
<link id=Main-File rel=Main-File href="FILENAME.html">
|
|
134
|
+
<!--[if gte mso 9]><xml>
|
|
135
|
+
<o:shapedefaults v:ext="edit" spidmax="2049"/>
|
|
136
|
+
</xml><![endif]-->
|
|
137
|
+
</head>
|
|
138
|
+
<body lang=EN link=blue vlink="#954F72">
|
|
139
|
+
<div style='mso-element:footnote-separator' id=fs>
|
|
140
|
+
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
141
|
+
normal'><span lang=EN-GB><span style='mso-special-character:footnote-separator'><![if !supportFootnotes]>
|
|
142
|
+
<hr align=left size=1 width="33%">
|
|
143
|
+
<![endif]></span></span></p>
|
|
144
|
+
</div>
|
|
145
|
+
<div style='mso-element:footnote-continuation-separator' id=fcs>
|
|
146
|
+
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
147
|
+
normal'><span lang=EN-GB><span style='mso-special-character:footnote-continuation-separator'><![if !supportFootnotes]>
|
|
148
|
+
<hr align=left size=1>
|
|
149
|
+
<![endif]></span></span></p>
|
|
150
|
+
</div>
|
|
151
|
+
<div style='mso-element:endnote-separator' id=es>
|
|
152
|
+
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
153
|
+
normal'><span lang=EN-GB><span style='mso-special-character:footnote-separator'><![if !supportFootnotes]>
|
|
154
|
+
<hr align=left size=1 width="33%">
|
|
155
|
+
<![endif]></span></span></p>
|
|
156
|
+
</div>
|
|
157
|
+
<div style='mso-element:endnote-continuation-separator' id=ecs>
|
|
158
|
+
<p class=MsoNormal style='margin-bottom:0cm;margin-bottom:.0001pt;line-height:
|
|
159
|
+
normal'><span lang=EN-GB><span style='mso-special-character:footnote-continuation-separator'><![if !supportFootnotes]>
|
|
160
|
+
<hr align=left size=1>
|
|
161
|
+
<![endif]></span></span></p>
|
|
162
|
+
</div>
|
|
163
|
+
<div style='mso-element:header' id=eh1>
|
|
164
|
+
<p class=MsoHeader align=left style='text-align:left;line-height:12.0pt;
|
|
165
|
+
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
166
|
+
</div>
|
|
167
|
+
<div style='mso-element:header' id=h1>
|
|
168
|
+
<p class=MsoHeader style='margin-bottom:18.0pt'><span lang=EN-GB
|
|
169
|
+
style='font-size:10.0pt;mso-bidi-font-size:11.0pt;font-weight:normal'>©
|
|
170
|
+
ISO/IEC&nbsp;2016&nbsp;– All rights reserved</span><span lang=EN-GB
|
|
171
|
+
style='font-weight:normal'><o:p></o:p></span></p>
|
|
172
|
+
</div>
|
|
173
|
+
<div style='mso-element:footer' id=ef1>
|
|
174
|
+
<p class=MsoFooter style='margin-top:12.0pt;line-height:12.0pt;mso-line-height-rule:
|
|
175
|
+
exactly'><!--[if supportFields]><b style='mso-bidi-font-weight:normal'><span
|
|
176
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
177
|
+
style='mso-element:field-begin'></span><span
|
|
178
|
+
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
179
|
+
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
180
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
181
|
+
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>2</span></span></b><!--[if supportFields]><b
|
|
182
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
183
|
+
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
184
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
185
|
+
style='mso-tab-count:1'>                                                                                                                                                                           </span>©
|
|
186
|
+
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
187
|
+
</div>
|
|
188
|
+
<div style='mso-element:header' id=eh2>
|
|
189
|
+
<p class=MsoHeader align=left style='text-align:left;line-height:12.0pt;
|
|
190
|
+
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
191
|
+
</div>
|
|
192
|
+
<div style='mso-element:header' id=h2>
|
|
193
|
+
<p class=MsoHeader align=right style='text-align:right;line-height:12.0pt;
|
|
194
|
+
mso-line-height-rule:exactly'><span lang=EN-GB>ISO/IEC&nbsp;CD 17301-1:2016(E)</span></p>
|
|
195
|
+
</div>
|
|
196
|
+
<div style='mso-element:footer' id=ef2>
|
|
197
|
+
<p class=MsoFooter style='line-height:12.0pt;mso-line-height-rule:exactly'><!--[if supportFields]><span
|
|
198
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
199
|
+
style='mso-element:field-begin'></span><span
|
|
200
|
+
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
201
|
+
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span><![endif]--><span
|
|
202
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
203
|
+
style='mso-no-proof:yes'>ii</span></span><!--[if supportFields]><span
|
|
204
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
205
|
+
style='mso-element:field-end'></span></span><![endif]--><span lang=EN-GB
|
|
206
|
+
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span style='mso-tab-count:
|
|
207
|
+
1'>                                                                                                                                                                           </span>©
|
|
208
|
+
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
209
|
+
</div>
|
|
210
|
+
<div style='mso-element:footer' id=f2>
|
|
211
|
+
<p class=MsoFooter style='line-height:12.0pt'><span lang=EN-GB
|
|
212
|
+
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'>© ISO/IEC&nbsp;2016&nbsp;– All
|
|
213
|
+
rights reserved<span style='mso-tab-count:1'>                                                                                                                                                                          </span></span><!--[if supportFields]><span
|
|
214
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
215
|
+
style='mso-element:field-begin'></span> PAGE<span style='mso-spacerun:yes'>  
|
|
216
|
+
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span><![endif]--><span
|
|
217
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
218
|
+
style='mso-no-proof:yes'>iii</span></span><!--[if supportFields]><span
|
|
219
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
220
|
+
style='mso-element:field-end'></span></span><![endif]--><span lang=EN-GB
|
|
221
|
+
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><o:p></o:p></span></p>
|
|
222
|
+
</div>
|
|
223
|
+
<div style='mso-element:footer' id=ef3>
|
|
224
|
+
<p class=MsoFooter style='margin-top:12.0pt;line-height:12.0pt;mso-line-height-rule:
|
|
225
|
+
exactly'><!--[if supportFields]><b style='mso-bidi-font-weight:normal'><span
|
|
226
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
227
|
+
style='mso-element:field-begin'></span><span
|
|
228
|
+
style='mso-spacerun:yes'> </span>PAGE<span style='mso-spacerun:yes'>  
|
|
229
|
+
</span>\\* MERGEFORMAT <span style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
230
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
231
|
+
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>2</span></span></b><!--[if supportFields]><b
|
|
232
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
233
|
+
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
234
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><span
|
|
235
|
+
style='mso-tab-count:1'>                                                                                                                                                                           </span>©
|
|
236
|
+
ISO/IEC&nbsp;2016&nbsp;– All rights reserved<o:p></o:p></span></p>
|
|
237
|
+
</div>
|
|
238
|
+
<div style='mso-element:footer' id=f3>
|
|
239
|
+
<p class=MsoFooter style='line-height:12.0pt'><span lang=EN-GB
|
|
240
|
+
style='font-size:10.0pt;mso-bidi-font-size:11.0pt'>© ISO/IEC&nbsp;2016&nbsp;– All
|
|
241
|
+
rights reserved<span style='mso-tab-count:1'>                                                                                                                                                                           </span></span><!--[if supportFields]><b
|
|
242
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
243
|
+
mso-bidi-font-size:11.0pt'><span style='mso-element:field-begin'></span>
|
|
244
|
+
PAGE<span style='mso-spacerun:yes'>   </span>\\* MERGEFORMAT <span
|
|
245
|
+
style='mso-element:field-separator'></span></span></b><![endif]--><b
|
|
246
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
247
|
+
mso-bidi-font-size:11.0pt'><span style='mso-no-proof:yes'>3</span></span></b><!--[if supportFields]><b
|
|
248
|
+
style='mso-bidi-font-weight:normal'><span lang=EN-GB style='font-size:10.0pt;
|
|
249
|
+
mso-bidi-font-size:11.0pt'><span style='mso-element:field-end'></span></span></b><![endif]--><span
|
|
250
|
+
lang=EN-GB style='font-size:10.0pt;mso-bidi-font-size:11.0pt'><o:p></o:p></span></p>
|
|
251
|
+
</div>
|
|
252
|
+
</body>
|
|
253
|
+
</html>
|
|
254
254
|
FTR
|
|
255
255
|
|
|
256
|
-
ASCII_MATH='<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=1</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>i</m:t></m:r></m:e><m:sup><m:r><m:t>3</m:t></m:r></m:sup></m:sSup></m:e></m:nary><m:r><m:t>=</m:t></m:r><m:sSup><m:e><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:f><m:fPr><m:type m:val="bar"></m:type></m:fPr><m:num><m:r><m:t>n</m:t></m:r><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:r><m:t>n+1</m:t></m:r></m:e></m:d></m:num><m:den><m:r><m:t>2</m:t></m:r></m:den></m:f></m:e></m:d></m:e><m:sup><m:r><m:t>2</m:t></m:r></m:sup></m:sSup>'
|
|
256
|
+
ASCII_MATH = '<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=1</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>i</m:t></m:r></m:e><m:sup><m:r><m:t>3</m:t></m:r></m:sup></m:sSup></m:e></m:nary><m:r><m:t>=</m:t></m:r><m:sSup><m:e><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:f><m:fPr><m:type m:val="bar"></m:type></m:fPr><m:num><m:r><m:t>n</m:t></m:r><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:r><m:t>n+1</m:t></m:r></m:e></m:d></m:num><m:den><m:r><m:t>2</m:t></m:r></m:den></m:f></m:e></m:d></m:e><m:sup><m:r><m:t>2</m:t></m:r></m:sup></m:sSup>'.freeze
|
|
257
257
|
|
|
258
|
-
DEFAULT_STYLESHEET = File.read("lib/html2doc/wordstyle.css",
|
|
258
|
+
DEFAULT_STYLESHEET = File.read("lib/html2doc/wordstyle.css",
|
|
259
|
+
encoding: "utf-8").freeze
|
|
259
260
|
|
|
260
|
-
def guid_clean(
|
|
261
|
-
|
|
261
|
+
def guid_clean(xml)
|
|
262
|
+
xml.gsub(/NextPart_[0-9a-f.]+/, "NextPart_")
|
|
262
263
|
end
|
|
263
264
|
|
|
264
|
-
def image_clean(
|
|
265
|
-
|
|
266
|
-
gsub(%r{[0-9a-f-]+\.gif}, "image.gif")
|
|
267
|
-
gsub(%r{[0-9a-f-]+\.(jpeg|jpg)}, "image.jpg")
|
|
268
|
-
gsub(%r{------=_NextPart_\s+Content-Location: file:///C:/Doc/test_files/image\.(png|gif).*?\s-----=_NextPart_}m, "------=_NextPart_")
|
|
269
|
-
gsub(%r{Content-Type: image/(png|gif|jpeg)[^-]*------=_NextPart_-?-?}m, "")
|
|
270
|
-
gsub(%r{ICAgICAg[^-]*-----}m, "-----")
|
|
271
|
-
gsub(%r{\s*</img>\s*}m, "</img>")
|
|
272
|
-
gsub(%r{</body>\s*</html>}m, "</body></html>")
|
|
265
|
+
def image_clean(xml)
|
|
266
|
+
xml.gsub(%r{[0-9a-f-]+\.png}, "image.png")
|
|
267
|
+
.gsub(%r{[0-9a-f-]+\.gif}, "image.gif")
|
|
268
|
+
.gsub(%r{[0-9a-f-]+\.(jpeg|jpg)}, "image.jpg")
|
|
269
|
+
.gsub(%r{------=_NextPart_\s+Content-Location: file:///C:/Doc/test_files/image\.(png|gif).*?\s-----=_NextPart_}m, "------=_NextPart_")
|
|
270
|
+
.gsub(%r{Content-Type: image/(png|gif|jpeg)[^-]*------=_NextPart_-?-?}m, "")
|
|
271
|
+
.gsub(%r{ICAgICAg[^-]*-----}m, "-----")
|
|
272
|
+
.gsub(%r{\s*</img>\s*}m, "</img>")
|
|
273
|
+
.gsub(%r{</body>\s*</html>}m, "</body></html>")
|
|
273
274
|
end
|
|
274
275
|
|
|
275
276
|
RSpec.describe Html2Doc do
|
|
@@ -279,11 +280,11 @@ RSpec.describe Html2Doc do
|
|
|
279
280
|
|
|
280
281
|
it "processes a blank document" do
|
|
281
282
|
Html2Doc.process(html_input(""), filename: "test")
|
|
282
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
283
|
-
to match_fuzzy(<<~OUTPUT)
|
|
284
|
-
|
|
285
|
-
|
|
286
|
-
|
|
283
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
284
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
285
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
286
|
+
#{word_body('', '<div style="mso-element:footnote-list"/>')} #{WORD_FTR1}
|
|
287
|
+
OUTPUT
|
|
287
288
|
end
|
|
288
289
|
|
|
289
290
|
it "removes any temp files" do
|
|
@@ -295,54 +296,67 @@ RSpec.describe Html2Doc do
|
|
|
295
296
|
end
|
|
296
297
|
|
|
297
298
|
it "processes a stylesheet in an HTML document with a title" do
|
|
298
|
-
Html2Doc.process(html_input(""),
|
|
299
|
-
|
|
300
|
-
|
|
301
|
-
|
|
302
|
-
|
|
303
|
-
|
|
299
|
+
Html2Doc.process(html_input(""),
|
|
300
|
+
filename: "test", stylesheet: "lib/html2doc/wordstyle.css")
|
|
301
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
302
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
303
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
304
|
+
#{word_body('', '<div style="mso-element:footnote-list"/>')} #{WORD_FTR1}
|
|
305
|
+
OUTPUT
|
|
304
306
|
end
|
|
305
307
|
|
|
306
308
|
it "processes a stylesheet in an HTML document without a title" do
|
|
307
|
-
Html2Doc.process(html_input_no_title(""),
|
|
308
|
-
|
|
309
|
-
|
|
310
|
-
|
|
311
|
-
|
|
312
|
-
|
|
313
|
-
|
|
309
|
+
Html2Doc.process(html_input_no_title(""),
|
|
310
|
+
filename: "test", stylesheet: "lib/html2doc/wordstyle.css")
|
|
311
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
312
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
313
|
+
#{WORD_HDR.sub('<title>blank</title>', '')}
|
|
314
|
+
#{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
315
|
+
#{word_body('', '<div style="mso-element:footnote-list"/>')} #{WORD_FTR1}
|
|
316
|
+
OUTPUT
|
|
314
317
|
end
|
|
315
318
|
|
|
316
319
|
it "processes a stylesheet in an HTML document with an empty head" do
|
|
317
|
-
Html2Doc.process(html_input_empty_head(""),
|
|
318
|
-
|
|
319
|
-
|
|
320
|
-
|
|
321
|
-
|
|
322
|
-
|
|
323
|
-
|
|
324
|
-
|
|
320
|
+
Html2Doc.process(html_input_empty_head(""),
|
|
321
|
+
filename: "test", stylesheet: "lib/html2doc/wordstyle.css")
|
|
322
|
+
word_hdr_end = WORD_HDR_END
|
|
323
|
+
.sub(%(<meta name="Originator" content="Me"/>\n), "")
|
|
324
|
+
.sub("</style>\n</head>", "</style></head>")
|
|
325
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
326
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
327
|
+
#{WORD_HDR.sub('<title>blank</title>', '')}
|
|
328
|
+
#{DEFAULT_STYLESHEET}
|
|
329
|
+
#{word_hdr_end}
|
|
330
|
+
#{word_body('', '<div style="mso-element:footnote-list"/>')} #{WORD_FTR1}
|
|
331
|
+
OUTPUT
|
|
325
332
|
end
|
|
326
333
|
|
|
327
334
|
it "processes a header" do
|
|
328
|
-
Html2Doc.process(html_input(""),
|
|
335
|
+
Html2Doc.process(html_input(""),
|
|
336
|
+
filename: "test", header_file: "spec/header.html")
|
|
329
337
|
html = guid_clean(File.read("test.doc", encoding: "utf-8"))
|
|
330
|
-
hdr = Base64.decode64(
|
|
331
|
-
|
|
332
|
-
|
|
333
|
-
|
|
334
|
-
|
|
335
|
-
|
|
336
|
-
|
|
337
|
-
expect(
|
|
338
|
-
|
|
339
|
-
|
|
340
|
-
|
|
341
|
-
|
|
338
|
+
hdr = Base64.decode64(
|
|
339
|
+
html
|
|
340
|
+
.sub(%r{^.*Content-Location: file:///C:/Doc/test_files/header.html}, "")
|
|
341
|
+
.sub(%r{^.*Content-Type: text/html charset="utf-8"}m, "")
|
|
342
|
+
.sub(%r{------=_NextPart_--.*$}m, ""),
|
|
343
|
+
).force_encoding("UTF-8")
|
|
344
|
+
# expect(hdr.gsub(/\xa0/, " ")).to match_fuzzy(HEADERHTML)
|
|
345
|
+
expect(HTMLEntities.new.encode(hdr, :hexadecimal)
|
|
346
|
+
.gsub(/</, "<").gsub(/>/, ">")
|
|
347
|
+
.gsub(/'/, "'").gsub(/"/, '"')
|
|
348
|
+
.gsub(/
/, "
").gsub(/
/, "\n"))
|
|
349
|
+
.to match_fuzzy(HEADERHTML)
|
|
350
|
+
expect(html.sub(%r{Content-ID: <header.html>.*$}m, ""))
|
|
351
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
352
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET.gsub(/url\("[^"]+"\)/, 'url(cid:header.html)')}
|
|
353
|
+
#{WORD_HDR_END} #{word_body('', '<div style="mso-element:footnote-list"/>')} #{WORD_FTR2}
|
|
354
|
+
OUTPUT
|
|
342
355
|
end
|
|
343
356
|
|
|
344
357
|
it "processes a header with an image" do
|
|
345
|
-
Html2Doc.process(html_input(""),
|
|
358
|
+
Html2Doc.process(html_input(""),
|
|
359
|
+
filename: "test", header_file: "spec/header_img.html")
|
|
346
360
|
doc = guid_clean(File.read("test.doc", encoding: "utf-8"))
|
|
347
361
|
expect(doc).to match(%r{Content-Type: image/png})
|
|
348
362
|
expect(doc).to match(%r{iVBORw0KGgoAAAANSUhEUgAAA5cAAAN7CAYAAADRE24cAAAgAElEQVR4XuydB5gUxdaGC65gTogB})
|
|
@@ -351,158 +365,174 @@ RSpec.describe Html2Doc do
|
|
|
351
365
|
it "processes a header with an image with absolute path" do
|
|
352
366
|
doc = File.read("spec/header_img.html", encoding: "utf-8")
|
|
353
367
|
File.open("spec/header_img1.html", "w:UTF-8") do |f|
|
|
354
|
-
f.write
|
|
368
|
+
f.write(
|
|
369
|
+
doc.sub(%r{spec/19160-6.png},
|
|
370
|
+
File.expand_path(File.join(File.dirname(__FILE__), "19160-6.png"))),
|
|
371
|
+
)
|
|
355
372
|
end
|
|
356
|
-
Html2Doc.process(html_input(""),
|
|
373
|
+
Html2Doc.process(html_input(""),
|
|
374
|
+
filename: "test", header_file: "spec/header_img1.html")
|
|
357
375
|
doc = guid_clean(File.read("test.doc", encoding: "utf-8"))
|
|
358
376
|
expect(doc).to match(%r{Content-Type: image/png})
|
|
359
377
|
expect(doc).to match(%r{iVBORw0KGgoAAAANSUhEUgAAA5cAAAN7CAYAAADRE24cAAAgAElEQVR4XuydB5gUxdaGC65gTogB})
|
|
360
378
|
end
|
|
361
379
|
|
|
362
|
-
|
|
363
380
|
it "processes a populated document" do
|
|
364
381
|
simple_body = "<h1>Hello word!</h1>
|
|
365
382
|
<div>This is a very simple document</div>"
|
|
366
383
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
367
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
368
|
-
to match_fuzzy(<<~OUTPUT)
|
|
369
|
-
|
|
370
|
-
|
|
371
|
-
|
|
372
|
-
|
|
384
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
385
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
386
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
387
|
+
#{word_body(simple_body, '<div style="mso-element:footnote-list"/>')}
|
|
388
|
+
#{WORD_FTR1}
|
|
389
|
+
OUTPUT
|
|
373
390
|
end
|
|
374
391
|
|
|
375
392
|
it "processes AsciiMath" do
|
|
376
|
-
Html2Doc.process(html_input(%[<div>{{sum_(i=1)^n i^3=((n(n+1))/2)^2 text("integer"))}}</div>]),
|
|
377
|
-
|
|
378
|
-
|
|
379
|
-
|
|
380
|
-
|
|
381
|
-
|
|
382
|
-
|
|
383
|
-
|
|
384
|
-
|
|
385
|
-
|
|
386
|
-
|
|
387
|
-
|
|
388
|
-
|
|
389
|
-
|
|
390
|
-
|
|
391
|
-
|
|
392
|
-
|
|
393
|
-
|
|
394
|
-
|
|
395
|
-
|
|
396
|
-
|
|
397
|
-
|
|
398
|
-
|
|
399
|
-
|
|
400
|
-
|
|
393
|
+
Html2Doc.process(html_input(%[<div>{{sum_(i=1)^n i^3=((n(n+1))/2)^2 text("integer"))}}</div>]),
|
|
394
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
395
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
396
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
397
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
398
|
+
#{word_body(%{
|
|
399
|
+
<div><m:oMath>
|
|
400
|
+
<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=1</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>i</m:t></m:r></m:e><m:sup><m:r><m:t>3</m:t></m:r></m:sup></m:sSup></m:e></m:nary><span style="font-style:normal;"><m:r><m:rPr><m:sty m:val="p"></m:sty></m:rPr><m:t>=</m:t></m:r></span><m:sSup><m:e><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:f><m:fPr><m:type m:val="bar"></m:type></m:fPr><m:num><m:r><m:t>n</m:t></m:r><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:r><m:t>n+1</m:t></m:r></m:e></m:d></m:num><m:den><m:r><m:t>2</m:t></m:r></m:den></m:f></m:e></m:d></m:e><m:sup><m:r><m:t>2</m:t></m:r></m:sup></m:sSup><m:r><m:rPr><m:nor></m:nor></m:rPr><m:t>"integer"</m:t></m:r><span style="font-style:normal;"><m:r><m:rPr><m:sty m:val="p"></m:sty></m:rPr><m:t>)</m:t></m:r></span>
|
|
401
|
+
</m:oMath>
|
|
402
|
+
</div>}, '<div style="mso-element:footnote-list"/>')}
|
|
403
|
+
#{WORD_FTR1}
|
|
404
|
+
OUTPUT
|
|
405
|
+
end
|
|
406
|
+
|
|
407
|
+
it "processes mstyle" do
|
|
408
|
+
Html2Doc.process(html_input(%[<div>{{bb (-log_2 (p_u)) bb "BB" bbb "BBB" cc "CC" bcc "BCC" tt "TT" fr "FR" bfr "BFR" sf "SF" bsf "BSFα" sfi "SFI" sfbi "SFBIα" bii "BII" ii "II"}}</div>]),
|
|
409
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
410
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
411
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
412
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
413
|
+
#{word_body(%{
|
|
414
|
+
<div><m:oMath>
|
|
415
|
+
<span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:sty m:val="b"></m:sty></m:rPr><m:t>−</m:t></m:r></span><m:sSub><m:e><span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:sty m:val="b"></m:sty></m:rPr><m:t>log</m:t></m:r></span></m:e><m:sub><span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:sty m:val="b"></m:sty></m:rPr><m:t>2</m:t></m:r></span></m:sub></m:sSub><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:sSub><m:e><span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:sty m:val="b"></m:sty></m:rPr><m:t>p</m:t></m:r></span></m:e><m:sub><span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:sty m:val="b"></m:sty></m:rPr><m:t>u</m:t></m:r></span></m:sub></m:sSub></m:e></m:d><span style="font-style:normal;font-weight:bold;"><m:r><m:rPr><m:nor></m:nor><m:sty m:val="b"></m:sty></m:rPr><m:t>BB</m:t></m:r></span><m:r><m:rPr><m:nor></m:nor><m:scr m:val="double-struck"></m:scr><m:sty m:val="p"></m:sty></m:rPr><m:t>𝔹𝔹𝔹</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="script"></m:scr></m:rPr><m:t>𝒞𝒞</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="script"></m:scr><m:sty m:val="b"></m:sty></m:rPr><m:t>𝓑𝓒𝓒</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="monospace"></m:scr><m:sty m:val="p"></m:sty></m:rPr><m:t>𝚃𝚃</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="fraktur"></m:scr><m:sty m:val="p"></m:sty></m:rPr><m:t>𝔉ℜ</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="fraktur"></m:scr><m:sty m:val="b"></m:sty></m:rPr><m:t>𝕭𝕱𝕽</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="sans-serif"></m:scr><m:sty m:val="p"></m:sty></m:rPr><m:t>𝖲𝖥</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="sans-serif"></m:scr><m:sty m:val="b"></m:sty></m:rPr><m:t>𝗕𝗦𝗙𝝰</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="sans-serif"></m:scr></m:rPr><m:t>𝖲𝖥𝖨</m:t></m:r><m:r><m:rPr><m:nor></m:nor><m:scr m:val="sans-serif"></m:scr><m:sty m:val="bi"></m:sty></m:rPr><m:t>𝙎𝙁𝘽𝙄𝞪</m:t></m:r><span class="nostem" style="font-weight:bold;"><em></em><m:r><m:rPr><m:nor></m:nor><m:sty m:val="bi"></m:sty></m:rPr><m:t>BII</m:t></m:r></span><span class="nostem"><em></em><m:r><m:rPr><m:nor></m:nor><m:sty m:val="i"></m:sty></m:rPr><m:t>II</m:t></m:r></span>
|
|
416
|
+
</m:oMath>
|
|
417
|
+
</div>}, '<div style="mso-element:footnote-list"/>')}
|
|
418
|
+
#{WORD_FTR1}
|
|
419
|
+
OUTPUT
|
|
401
420
|
end
|
|
402
421
|
|
|
403
422
|
it "processes spaces in AsciiMath" do
|
|
404
|
-
Html2Doc.process(html_input(%[<div>{{text " integer ")}}</div>]),
|
|
405
|
-
|
|
406
|
-
|
|
407
|
-
|
|
408
|
-
|
|
409
|
-
|
|
410
|
-
|
|
411
|
-
|
|
412
|
-
|
|
413
|
-
|
|
414
|
-
|
|
423
|
+
Html2Doc.process(html_input(%[<div>{{text " integer ")}}</div>]),
|
|
424
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
425
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
426
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
427
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
428
|
+
#{word_body('
|
|
429
|
+
<div><m:oMath>
|
|
430
|
+
<m:r><m:t>text</m:t></m:r><m:r><m:rPr><m:nor></m:nor></m:rPr><m:t> integer </m:t></m:r><span style="font-style:normal;"><m:r><m:rPr><m:sty m:val="p"></m:sty></m:rPr><m:t>)</m:t></m:r></span>
|
|
431
|
+
</m:oMath>
|
|
432
|
+
</div>', '<div style="mso-element:footnote-list"/>')}
|
|
433
|
+
#{WORD_FTR1}
|
|
434
|
+
OUTPUT
|
|
415
435
|
end
|
|
416
436
|
|
|
417
437
|
it "processes spaces in MathML mtext" do
|
|
418
438
|
Html2Doc.process(html_input("<div><math xmlns='http://www.w3.org/1998/Math/MathML'>
|
|
419
439
|
<mrow><mi>H</mi><mtext> original </mtext><mi>J</mi></mrow>
|
|
420
|
-
</math></div>"),
|
|
421
|
-
|
|
422
|
-
|
|
423
|
-
|
|
424
|
-
|
|
425
|
-
|
|
426
|
-
|
|
427
|
-
|
|
428
|
-
|
|
429
|
-
|
|
440
|
+
</math></div>"),
|
|
441
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
442
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
443
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
444
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
445
|
+
#{word_body('<div><m:oMath>
|
|
446
|
+
<m:r><m:t>H</m:t></m:r><m:r><m:rPr><m:nor></m:nor></m:rPr><m:t> original </m:t></m:r><m:r><m:t>J</m:t></m:r>
|
|
447
|
+
</m:oMath>
|
|
448
|
+
</div>', '<div style="mso-element:footnote-list"/>')}
|
|
449
|
+
#{WORD_FTR1}
|
|
450
|
+
OUTPUT
|
|
430
451
|
end
|
|
431
452
|
|
|
432
453
|
it "unwraps accent in MathML" do
|
|
433
454
|
Html2Doc.process(html_input("<div><math xmlns='http://www.w3.org/1998/Math/MathML'>
|
|
434
455
|
<mover accent='true'><mrow><mi>p</mi></mrow><mrow><mo>^</mo></mrow></mover>
|
|
435
456
|
</math></div>"), filename: "test", asciimathdelims: ["{{", "}}"])
|
|
436
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
437
|
-
to match_fuzzy(<<~OUTPUT)
|
|
438
|
-
|
|
439
|
-
|
|
440
|
-
|
|
441
|
-
|
|
442
|
-
|
|
443
|
-
|
|
444
|
-
|
|
457
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
458
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
459
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
460
|
+
#{word_body('<div><m:oMath>
|
|
461
|
+
<m:acc><m:accPr><m:chr m:val="^"></m:chr></m:accPr><m:e><m:r><m:t>p</m:t></m:r></m:e></m:acc>
|
|
462
|
+
</m:oMath>
|
|
463
|
+
</div>', '<div style="mso-element:footnote-list"/>')}
|
|
464
|
+
#{WORD_FTR1}
|
|
465
|
+
OUTPUT
|
|
445
466
|
end
|
|
446
467
|
|
|
447
468
|
it "left-aligns AsciiMath" do
|
|
448
|
-
Html2Doc.process(html_input("<div style='text-align:left;'>{{sum_(i=1)^n i^3=((n(n+1))/2)^2}}</div>"),
|
|
449
|
-
|
|
450
|
-
|
|
451
|
-
|
|
452
|
-
|
|
453
|
-
|
|
454
|
-
|
|
455
|
-
|
|
456
|
-
|
|
457
|
-
|
|
458
|
-
|
|
469
|
+
Html2Doc.process(html_input("<div style='text-align:left;'>{{sum_(i=1)^n i^3=((n(n+1))/2)^2}}</div>"),
|
|
470
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
471
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
472
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
473
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
474
|
+
#{word_body(%{
|
|
475
|
+
<div style="text-align:left;"><m:oMathPara><m:oMathParaPr><m:jc m:val="left"/></m:oMathParaPr><m:oMath>
|
|
476
|
+
<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=1</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>i</m:t></m:r></m:e><m:sup><m:r><m:t>3</m:t></m:r></m:sup></m:sSup></m:e></m:nary><span style="font-style:normal;"><m:r><m:rPr><m:sty m:val="p"></m:sty></m:rPr><m:t>=</m:t></m:r></span><m:sSup><m:e><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:f><m:fPr><m:type m:val="bar"></m:type></m:fPr><m:num><m:r><m:t>n</m:t></m:r><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:r><m:t>n+1</m:t></m:r></m:e></m:d></m:num><m:den><m:r><m:t>2</m:t></m:r></m:den></m:f></m:e></m:d></m:e><m:sup><m:r><m:t>2</m:t></m:r></m:sup></m:sSup>
|
|
477
|
+
</m:oMath>
|
|
478
|
+
</m:oMathPara></div>}, '<div style="mso-element:footnote-list"/>')}
|
|
479
|
+
#{WORD_FTR1}
|
|
480
|
+
OUTPUT
|
|
459
481
|
end
|
|
460
482
|
|
|
461
483
|
it "right-aligns AsciiMath" do
|
|
462
|
-
Html2Doc.process(html_input("<div style='text-align:right;'>{{sum_(i=1)^n i^3=((n(n+1))/2)^2}}</div>"),
|
|
463
|
-
|
|
464
|
-
|
|
465
|
-
|
|
466
|
-
|
|
467
|
-
|
|
468
|
-
|
|
469
|
-
|
|
470
|
-
|
|
471
|
-
|
|
472
|
-
|
|
484
|
+
Html2Doc.process(html_input("<div style='text-align:right;'>{{sum_(i=1)^n i^3=((n(n+1))/2)^2}}</div>"),
|
|
485
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
486
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
487
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
488
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
489
|
+
#{word_body(%{
|
|
490
|
+
<div style="text-align:right;"><m:oMathPara><m:oMathParaPr><m:jc m:val="right"/></m:oMathParaPr><m:oMath>
|
|
491
|
+
<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=1</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>i</m:t></m:r></m:e><m:sup><m:r><m:t>3</m:t></m:r></m:sup></m:sSup></m:e></m:nary><span style="font-style:normal;"><m:r><m:rPr><m:sty m:val="p"></m:sty></m:rPr><m:t>=</m:t></m:r></span><m:sSup><m:e><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:f><m:fPr><m:type m:val="bar"></m:type></m:fPr><m:num><m:r><m:t>n</m:t></m:r><m:d><m:dPr><m:sepChr m:val=","></m:sepChr></m:dPr><m:e><m:r><m:t>n+1</m:t></m:r></m:e></m:d></m:num><m:den><m:r><m:t>2</m:t></m:r></m:den></m:f></m:e></m:d></m:e><m:sup><m:r><m:t>2</m:t></m:r></m:sup></m:sSup>
|
|
492
|
+
</m:oMath>
|
|
493
|
+
</m:oMathPara></div>}, '<div style="mso-element:footnote-list"/>')}
|
|
494
|
+
#{WORD_FTR1}
|
|
495
|
+
OUTPUT
|
|
473
496
|
end
|
|
474
497
|
|
|
475
498
|
it "raises error in processing of broken AsciiMath" do
|
|
476
499
|
begin
|
|
477
|
-
|
|
500
|
+
expect do
|
|
501
|
+
Html2Doc.process(html_input(%[<div style='text-align:right;'>{{u_c = 6.6"unitsml(kHz)}}</div>]),
|
|
502
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
503
|
+
end.to output('parsing: u_c = 6.6"unitsml(kHz)').to_stderr
|
|
478
504
|
rescue StandardError
|
|
479
505
|
end
|
|
480
|
-
expect
|
|
506
|
+
expect do
|
|
507
|
+
Html2Doc.process(html_input(%[<div style='text-align:right;'>{{u_c = 6.6"unitsml(kHz)}}</div>]),
|
|
508
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
509
|
+
end.to raise_error(StandardError)
|
|
481
510
|
end
|
|
482
511
|
|
|
483
512
|
it "wraps msup after munderover in MathML" do
|
|
484
513
|
Html2Doc.process(html_input("<div><math xmlns='http://www.w3.org/1998/Math/MathML'>
|
|
485
|
-
<munderover><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>0</mn></mrow><mrow><mi>n</mi></mrow></munderover><msup><mn>2</mn><mrow><mi>i</mi></mrow></msup></math></div>"),
|
|
486
|
-
|
|
487
|
-
|
|
488
|
-
|
|
489
|
-
|
|
490
|
-
|
|
491
|
-
|
|
492
|
-
|
|
493
|
-
|
|
514
|
+
<munderover><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>0</mn></mrow><mrow><mi>n</mi></mrow></munderover><msup><mn>2</mn><mrow><mi>i</mi></mrow></msup></math></div>"),
|
|
515
|
+
filename: "test", asciimathdelims: ["{{", "}}"])
|
|
516
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
517
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
518
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
519
|
+
#{word_body('<div><m:oMath>
|
|
520
|
+
<m:nary><m:naryPr><m:chr m:val="∑"></m:chr><m:limLoc m:val="undOvr"></m:limLoc><m:grow m:val="on"></m:grow><m:subHide m:val="off"></m:subHide><m:supHide m:val="off"></m:supHide></m:naryPr><m:sub><m:r><m:t>i=0</m:t></m:r></m:sub><m:sup><m:r><m:t>n</m:t></m:r></m:sup><m:e><m:sSup><m:e><m:r><m:t>2</m:t></m:r></m:e><m:sup><m:r><m:t>i</m:t></m:r></m:sup></m:sSup></m:e></m:nary></m:oMath>
|
|
521
|
+
</div>', '<div style="mso-element:footnote-list"/>')}
|
|
522
|
+
#{WORD_FTR1}
|
|
523
|
+
OUTPUT
|
|
494
524
|
end
|
|
495
525
|
|
|
496
526
|
it "processes tabs" do
|
|
497
527
|
simple_body = "<h1>Hello word!</h1>
|
|
498
528
|
<div>This is a very &tab; simple document</div>"
|
|
499
529
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
500
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
501
|
-
to match_fuzzy(<<~OUTPUT)
|
|
502
|
-
|
|
503
|
-
|
|
504
|
-
|
|
505
|
-
|
|
530
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
531
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
532
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
533
|
+
#{word_body(simple_body.gsub(/&tab;/, %[<span style="mso-tab-count:1">  </span>]), '<div style="mso-element:footnote-list"/>')}
|
|
534
|
+
#{WORD_FTR1}
|
|
535
|
+
OUTPUT
|
|
506
536
|
end
|
|
507
537
|
|
|
508
538
|
it "makes unstyled paragraphs be MsoNormal" do
|
|
@@ -510,12 +540,12 @@ RSpec.describe Html2Doc do
|
|
|
510
540
|
<p>This is a very simple document</p>
|
|
511
541
|
<p class="x">This style stays</p>'
|
|
512
542
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
513
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
514
|
-
to match_fuzzy(<<~OUTPUT)
|
|
515
|
-
|
|
516
|
-
|
|
517
|
-
|
|
518
|
-
|
|
543
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
544
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
545
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
546
|
+
#{word_body(simple_body.gsub(/<p>/, %[<p class="MsoNormal">]), '<div style="mso-element:footnote-list"/>')}
|
|
547
|
+
#{WORD_FTR1}
|
|
548
|
+
OUTPUT
|
|
519
549
|
end
|
|
520
550
|
|
|
521
551
|
it "makes unstyled list entries be MsoNormal" do
|
|
@@ -525,12 +555,12 @@ RSpec.describe Html2Doc do
|
|
|
525
555
|
<li class="x">This style stays</li>
|
|
526
556
|
</ul>'
|
|
527
557
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
528
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
529
|
-
to match_fuzzy(<<~OUTPUT)
|
|
530
|
-
|
|
531
|
-
|
|
532
|
-
|
|
533
|
-
|
|
558
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
559
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
560
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
561
|
+
#{word_body(simple_body.gsub(/<li>/, %[<li class="MsoNormal">]), '<div style="mso-element:footnote-list"/>')}
|
|
562
|
+
#{WORD_FTR1}
|
|
563
|
+
OUTPUT
|
|
534
564
|
end
|
|
535
565
|
|
|
536
566
|
it "resizes images for height, in a file in a subdirectory" do
|
|
@@ -539,9 +569,9 @@ RSpec.describe Html2Doc do
|
|
|
539
569
|
testdoc = File.read("spec/test.doc", encoding: "utf-8")
|
|
540
570
|
expect(testdoc).to match(%r{Content-Type: image/png})
|
|
541
571
|
expect(image_clean(guid_clean(testdoc))).to match_fuzzy(<<~OUTPUT)
|
|
542
|
-
|
|
543
|
-
|
|
544
|
-
|
|
572
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
573
|
+
#{image_clean(word_body('<img src="cid:cb7b0d19-891e-4634-815a-570d019d454c.png" width="400" height="388"></img>', '<div style="mso-element:footnote-list"/>'))}
|
|
574
|
+
#{image_clean(WORD_FTR3)}
|
|
545
575
|
OUTPUT
|
|
546
576
|
end
|
|
547
577
|
|
|
@@ -551,9 +581,9 @@ RSpec.describe Html2Doc do
|
|
|
551
581
|
testdoc = File.read("test.doc", encoding: "utf-8")
|
|
552
582
|
expect(testdoc).to match(%r{Content-Type: image/gif})
|
|
553
583
|
expect(image_clean(guid_clean(testdoc))).to match_fuzzy(<<~OUTPUT)
|
|
554
|
-
|
|
555
|
-
|
|
556
|
-
|
|
584
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
585
|
+
#{image_clean(word_body('<img src="cid:cb7b0d19-891e-4634-815a-570d019d454c.gif" width="400" height="118"></img>', '<div style="mso-element:footnote-list"/>'))}
|
|
586
|
+
#{image_clean(WORD_FTR3).gsub(/image\.png/, 'image.gif')}
|
|
557
587
|
OUTPUT
|
|
558
588
|
end
|
|
559
589
|
|
|
@@ -563,41 +593,51 @@ RSpec.describe Html2Doc do
|
|
|
563
593
|
testdoc = File.read("test.doc", encoding: "utf-8")
|
|
564
594
|
expect(testdoc).to match(%r{Content-Type: image/jpeg})
|
|
565
595
|
expect(image_clean(guid_clean(testdoc))).to match_fuzzy(<<~OUTPUT)
|
|
566
|
-
|
|
567
|
-
|
|
568
|
-
|
|
596
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
597
|
+
#{image_clean(word_body('<img src="cid:cb7b0d19-891e-4634-815a-570d019d454c.jpg" width="208" height="680"></img>', '<div style="mso-element:footnote-list"/>'))}
|
|
598
|
+
#{image_clean(WORD_FTR3).gsub(/image\.png/, 'image.jpg')}
|
|
569
599
|
OUTPUT
|
|
570
600
|
end
|
|
571
601
|
|
|
572
602
|
it "resizes images with missing or auto sizes" do
|
|
573
603
|
image = { "src" => "spec/19160-8.jpg" }
|
|
574
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
604
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
605
|
+
.to eq [30, 100]
|
|
575
606
|
image["width"] = "20"
|
|
576
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
607
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
608
|
+
.to eq [20, 65]
|
|
577
609
|
image.delete("width")
|
|
578
610
|
image["height"] = "50"
|
|
579
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
611
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
612
|
+
.to eq [15, 50]
|
|
580
613
|
image.delete("height")
|
|
581
614
|
image["width"] = "500"
|
|
582
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
615
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
616
|
+
.to eq [30, 100]
|
|
583
617
|
image.delete("width")
|
|
584
618
|
image["height"] = "500"
|
|
585
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
619
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
620
|
+
.to eq [30, 100]
|
|
586
621
|
image["width"] = "20"
|
|
587
622
|
image["height"] = "auto"
|
|
588
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
623
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
624
|
+
.to eq [20, 65]
|
|
589
625
|
image["width"] = "auto"
|
|
590
626
|
image["height"] = "50"
|
|
591
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
627
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
628
|
+
.to eq [15, 50]
|
|
592
629
|
image["width"] = "500"
|
|
593
630
|
image["height"] = "auto"
|
|
594
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
631
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
632
|
+
.to eq [30, 100]
|
|
595
633
|
image["width"] = "auto"
|
|
596
634
|
image["height"] = "500"
|
|
597
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
635
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
636
|
+
.to eq [30, 100]
|
|
598
637
|
image["width"] = "auto"
|
|
599
638
|
image["height"] = "auto"
|
|
600
|
-
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
639
|
+
expect(Html2Doc.image_resize(image, "spec/19160-8.jpg", 100, 100))
|
|
640
|
+
.to eq [30, 100]
|
|
601
641
|
end
|
|
602
642
|
|
|
603
643
|
it "does not move images if they are external URLs" do
|
|
@@ -605,49 +645,48 @@ RSpec.describe Html2Doc do
|
|
|
605
645
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
606
646
|
testdoc = File.read("test.doc", encoding: "utf-8")
|
|
607
647
|
expect(image_clean(guid_clean(testdoc))).to match_fuzzy(<<~OUTPUT)
|
|
608
|
-
|
|
609
|
-
|
|
610
|
-
|
|
648
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
649
|
+
#{image_clean(word_body('<img src="https://example.com/19160-6.png"></img>', '<div style="mso-element:footnote-list"/>'))}
|
|
650
|
+
#{image_clean(WORD_FTR1)}
|
|
611
651
|
OUTPUT
|
|
612
652
|
end
|
|
613
653
|
|
|
614
654
|
it "deals with absolute image locations" do
|
|
615
|
-
simple_body = %{<img src="#{
|
|
655
|
+
simple_body = %{<img src="#{__dir__}/19160-6.png">}
|
|
616
656
|
Html2Doc.process(html_input(simple_body), filename: "spec/test")
|
|
617
657
|
testdoc = File.read("spec/test.doc", encoding: "utf-8")
|
|
618
658
|
expect(testdoc).to match(%r{Content-Type: image/png})
|
|
619
659
|
expect(image_clean(guid_clean(testdoc))).to match_fuzzy(<<~OUTPUT)
|
|
620
|
-
|
|
621
|
-
|
|
622
|
-
|
|
660
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
661
|
+
#{image_clean(word_body('<img src="cid:cb7b0d19-891e-4634-815a-570d019d454c.png" width="400" height="388"></img>', '<div style="mso-element:footnote-list"/>'))}
|
|
662
|
+
#{image_clean(WORD_FTR3)}
|
|
623
663
|
OUTPUT
|
|
624
664
|
end
|
|
625
665
|
|
|
626
|
-
|
|
627
|
-
|
|
628
|
-
|
|
629
|
-
|
|
630
|
-
end
|
|
631
|
-
=end
|
|
666
|
+
# it "warns about SVG" do
|
|
667
|
+
# simple_body = '<img src="https://example.com/19160-6.svg">'
|
|
668
|
+
# expect{ Html2Doc.process(html_input(simple_body), filename: "test") }
|
|
669
|
+
# .to output("https://example.com/19160-6.svg: SVG not supported\n").to_stderr
|
|
670
|
+
# end
|
|
632
671
|
|
|
633
672
|
it "processes epub:type footnotes" do
|
|
634
|
-
simple_body = '<div>This is a very simple
|
|
673
|
+
simple_body = '<div>This is a very simple
|
|
635
674
|
document<a epub:type="footnote" href="#a1">1</a> allegedly<a epub:type="footnote" href="#a2">2</a></div>
|
|
636
675
|
<aside id="a1">Footnote</aside>
|
|
637
676
|
<aside id="a2">Other Footnote</aside>'
|
|
638
677
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
639
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
640
|
-
to match_fuzzy(<<~OUTPUT)
|
|
641
|
-
|
|
642
|
-
|
|
643
|
-
|
|
644
|
-
|
|
645
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
646
|
-
<div style="mso-element:footnote" id="ftn2">
|
|
647
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
648
|
-
</div>')}
|
|
649
|
-
|
|
650
|
-
|
|
678
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
679
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
680
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
681
|
+
#{word_body('<div>This is a very simple
|
|
682
|
+
document<a epub:type="footnote" href="#_ftn1" style="mso-footnote-id:ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a> allegedly<a epub:type="footnote" href="#_ftn2" style="mso-footnote-id:ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a></div>',
|
|
683
|
+
'<div style="mso-element:footnote-list"><div style="mso-element:footnote" id="ftn1">
|
|
684
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
685
|
+
<div style="mso-element:footnote" id="ftn2">
|
|
686
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
687
|
+
</div>')}
|
|
688
|
+
#{WORD_FTR1}
|
|
689
|
+
OUTPUT
|
|
651
690
|
end
|
|
652
691
|
|
|
653
692
|
it "processes class footnotes" do
|
|
@@ -656,38 +695,38 @@ RSpec.describe Html2Doc do
|
|
|
656
695
|
<aside id="a1">Footnote</aside>
|
|
657
696
|
<aside id="a2">Other Footnote</aside>'
|
|
658
697
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
659
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
660
|
-
to match_fuzzy(<<~OUTPUT)
|
|
661
|
-
|
|
662
|
-
|
|
663
|
-
|
|
664
|
-
|
|
665
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
666
|
-
<div style="mso-element:footnote" id="ftn2">
|
|
667
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
668
|
-
</div>')}
|
|
669
|
-
|
|
670
|
-
|
|
698
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
699
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
700
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
701
|
+
#{word_body('<div>This is a very simple
|
|
702
|
+
document<a class="footnote" href="#_ftn1" style="mso-footnote-id:ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a> allegedly<a class="footnote" href="#_ftn2" style="mso-footnote-id:ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a></div>',
|
|
703
|
+
'<div style="mso-element:footnote-list"><div style="mso-element:footnote" id="ftn1">
|
|
704
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
705
|
+
<div style="mso-element:footnote" id="ftn2">
|
|
706
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
707
|
+
</div>')}
|
|
708
|
+
#{WORD_FTR1}
|
|
709
|
+
OUTPUT
|
|
671
710
|
end
|
|
672
711
|
|
|
673
712
|
it "processes footnotes with text wrapping the footnote reference" do
|
|
674
|
-
|
|
713
|
+
simple_body = '<div>This is a very simple
|
|
675
714
|
document<a class="footnote" href="#a1">(<span class="MsoFootnoteReference">1</span>)</a> allegedly<a class="footnote" href="#a2">2</a></div>
|
|
676
715
|
<aside id="a1">Footnote</aside>
|
|
677
716
|
<aside id="a2">Other Footnote</aside>'
|
|
678
717
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
679
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
680
|
-
to match_fuzzy(<<~OUTPUT)
|
|
681
|
-
|
|
682
|
-
|
|
683
|
-
|
|
684
|
-
|
|
685
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference">(</span><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span><span class="MsoFootnoteReference">)</span></a>Footnote</p></div>
|
|
686
|
-
<div style="mso-element:footnote" id="ftn2">
|
|
687
|
-
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
688
|
-
</div>')}
|
|
689
|
-
|
|
690
|
-
|
|
718
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
719
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
720
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
721
|
+
#{word_body('<div>This is a very simple
|
|
722
|
+
document<a class="footnote" href="#_ftn1" style="mso-footnote-id:ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference">(</span><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span><span class="MsoFootnoteReference">)</span></a> allegedly<a class="footnote" href="#_ftn2" style="mso-footnote-id:ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a></div>',
|
|
723
|
+
'<div style="mso-element:footnote-list"><div style="mso-element:footnote" id="ftn1">
|
|
724
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference">(</span><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span><span class="MsoFootnoteReference">)</span></a>Footnote</p></div>
|
|
725
|
+
<div style="mso-element:footnote" id="ftn2">
|
|
726
|
+
<p id="" class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
727
|
+
</div>')}
|
|
728
|
+
#{WORD_FTR1}
|
|
729
|
+
OUTPUT
|
|
691
730
|
end
|
|
692
731
|
|
|
693
732
|
it "extracts paragraphs from footnotes" do
|
|
@@ -696,18 +735,18 @@ RSpec.describe Html2Doc do
|
|
|
696
735
|
<aside id="a1"><p>Footnote</p></aside>
|
|
697
736
|
<div id="a2"><p>Other Footnote</p></div>'
|
|
698
737
|
Html2Doc.process(html_input(simple_body), filename: "test")
|
|
699
|
-
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
700
|
-
to match_fuzzy(<<~OUTPUT)
|
|
701
|
-
|
|
702
|
-
|
|
703
|
-
|
|
704
|
-
|
|
705
|
-
<p class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
706
|
-
<div style="mso-element:footnote" id="ftn2">
|
|
707
|
-
<p class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
708
|
-
</div>')}
|
|
709
|
-
|
|
710
|
-
|
|
738
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
739
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
740
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
741
|
+
#{word_body('<div>This is a very simple
|
|
742
|
+
document<a class="footnote" href="#_ftn1" style="mso-footnote-id:ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a> allegedly<a class="footnote" href="#_ftn2" style="mso-footnote-id:ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a></div>',
|
|
743
|
+
'<div style="mso-element:footnote-list"><div style="mso-element:footnote" id="ftn1">
|
|
744
|
+
<p class="MsoFootnoteText"><a style="mso-footnote-id:ftn1" href="#_ftn1" name="_ftnref1" title="" id="_ftnref1"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Footnote</p></div>
|
|
745
|
+
<div style="mso-element:footnote" id="ftn2">
|
|
746
|
+
<p class="MsoFootnoteText"><a style="mso-footnote-id:ftn2" href="#_ftn2" name="_ftnref2" title="" id="_ftnref2"><span class="MsoFootnoteReference"><span style="mso-special-character:footnote"></span></span></a>Other Footnote</p></div>
|
|
747
|
+
</div>')}
|
|
748
|
+
#{WORD_FTR1}
|
|
749
|
+
OUTPUT
|
|
711
750
|
end
|
|
712
751
|
|
|
713
752
|
it "labels lists with list styles" do
|
|
@@ -715,38 +754,39 @@ RSpec.describe Html2Doc do
|
|
|
715
754
|
<div><ul id="0">
|
|
716
755
|
<li><div><p><ol id="1"><li><ul id="2"><li><p><ol id="3"><li><ol id="4"><li>A</li><li><p>B</p><p>B2</p></li><li>C</li></ol></li></ol></p></li></ul></li></ol></p></div></li><div><ul id="5"><li>C</li></ul></div>
|
|
717
756
|
BODY
|
|
718
|
-
Html2Doc.process(html_input(simple_body),
|
|
719
|
-
|
|
720
|
-
|
|
721
|
-
|
|
722
|
-
|
|
723
|
-
|
|
724
|
-
|
|
725
|
-
|
|
726
|
-
|
|
727
|
-
|
|
757
|
+
Html2Doc.process(html_input(simple_body),
|
|
758
|
+
filename: "test", liststyles: { ul: "l1", ol: "l2" })
|
|
759
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
760
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
761
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
762
|
+
#{word_body('<div>
|
|
763
|
+
<p style="mso-list:l1 level1 lfo1;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l2 level2 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level4 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpFirst">A</p><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpMiddle">B<p class="MsoListParagraphCxSpMiddle">B2</p></p><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpLast">C</p></p></p></p></div></p><div><p style="mso-list:l1 level1 lfo2;" class="MsoListParagraphCxSpFirst">C</p></div>
|
|
764
|
+
</div>',
|
|
765
|
+
'<div style="mso-element:footnote-list"/>')}
|
|
766
|
+
#{WORD_FTR1}
|
|
767
|
+
OUTPUT
|
|
728
768
|
end
|
|
729
769
|
|
|
730
|
-
|
|
731
770
|
it "restarts numbering of lists with list styles" do
|
|
732
771
|
simple_body = <<~BODY
|
|
733
772
|
<div>
|
|
734
773
|
<ol id="1"><li><div><p><ol id="2"><li><ul id="3"><li><p><ol id="4"><li><ol id="5"><li>A</li></ol></li></ol></p></li></ul></li></ol></p></div></li></ol>
|
|
735
774
|
<ol id="6"><li><div><p><ol id="7"><li><ul id="8"><li><p><ol id="9"><li><ol id="10"><li>A</li></ol></li></ol></p></li></ul></li></ol></p></div></li></ol></div>
|
|
736
775
|
BODY
|
|
737
|
-
Html2Doc.process(html_input(simple_body),
|
|
738
|
-
|
|
739
|
-
|
|
740
|
-
|
|
741
|
-
|
|
742
|
-
|
|
743
|
-
|
|
744
|
-
|
|
745
|
-
|
|
746
|
-
|
|
747
|
-
|
|
748
|
-
|
|
749
|
-
|
|
776
|
+
Html2Doc.process(html_input(simple_body),
|
|
777
|
+
filename: "test", liststyles: { ul: "l1", ol: "l2" })
|
|
778
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
779
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
780
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
781
|
+
#{word_body('<div>
|
|
782
|
+
<p style="mso-list:l2 level1 lfo1;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l2 level2 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level4 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpFirst">A</p></p></p></p></div></p>
|
|
783
|
+
<p style="mso-list:l2 level1 lfo2;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l2 level2 lfo2;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level4 lfo2;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level5 lfo2;" class="MsoListParagraphCxSpFirst">A</p></p></p></p></div></p></div>',
|
|
784
|
+
'<div style="mso-element:footnote-list"/>')}
|
|
785
|
+
#{WORD_FTR1}
|
|
786
|
+
OUTPUT
|
|
787
|
+
end
|
|
788
|
+
|
|
789
|
+
it "labels lists with multiple list styles" do
|
|
750
790
|
simple_body = <<~BODY
|
|
751
791
|
<div><ul class="steps" id="0">
|
|
752
792
|
<li><div><p><ol id="1"><li><ul id="2"><li><p><ol id="3"><li><ol id="4"><li>A</li><li><p>B</p><p>B2</p></li><li>C</li></ol></li></ol></p></li></ul></li></ol></p></div></li></ul></div>
|
|
@@ -755,19 +795,21 @@ RSpec.describe Html2Doc do
|
|
|
755
795
|
<div><ul class="other" id="10">
|
|
756
796
|
<li><div><p><ol id="11"><li><ul id="12"><li><p><ol id="13"><li><ol id="14"><li>A</li><li><p>B</p><p>B2</p></li><li>C</li></ol></li></ol></p></li></ul></li></ol></p></div></li></ul></div>
|
|
757
797
|
BODY
|
|
758
|
-
Html2Doc.process(html_input(simple_body),
|
|
759
|
-
|
|
760
|
-
|
|
761
|
-
|
|
762
|
-
|
|
763
|
-
|
|
764
|
-
|
|
765
|
-
|
|
766
|
-
|
|
767
|
-
|
|
768
|
-
|
|
769
|
-
|
|
770
|
-
|
|
798
|
+
Html2Doc.process(html_input(simple_body),
|
|
799
|
+
filename: "test",
|
|
800
|
+
liststyles: { ul: "l1", ol: "l2", steps: "l3" })
|
|
801
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
802
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
803
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
804
|
+
#{word_body('<div>
|
|
805
|
+
<p style="mso-list:l3 level1 lfo2;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l3 level2 lfo2;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l3 level4 lfo2;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l3 level5 lfo2;" class="MsoListParagraphCxSpFirst">A</p><p style="mso-list:l3 level5 lfo2;" class="MsoListParagraphCxSpMiddle">B<p class="MsoListParagraphCxSpMiddle">B2</p></p><p style="mso-list:l3 level5 lfo2;" class="MsoListParagraphCxSpLast">C</p></p></p></p></div></p></div>
|
|
806
|
+
<div>
|
|
807
|
+
<p style="mso-list:l1 level1 lfo1;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l2 level2 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level4 lfo1;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpFirst">A</p><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpMiddle">B<p class="MsoListParagraphCxSpMiddle">B2</p></p><p style="mso-list:l2 level5 lfo1;" class="MsoListParagraphCxSpLast">C</p></p></p></p></div></p></div>
|
|
808
|
+
<div>
|
|
809
|
+
<p style="mso-list:l1 level1 lfo3;" class="MsoListParagraphCxSpFirst"><div><p class="MsoNormal"><p style="mso-list:l2 level2 lfo3;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level4 lfo3;" class="MsoListParagraphCxSpFirst"><p style="mso-list:l2 level5 lfo3;" class="MsoListParagraphCxSpFirst">A</p><p style="mso-list:l2 level5 lfo3;" class="MsoListParagraphCxSpMiddle">B<p class="MsoListParagraphCxSpMiddle">B2</p></p><p style="mso-list:l2 level5 lfo3;" class="MsoListParagraphCxSpLast">C</p></p></p></p></div></p></div>',
|
|
810
|
+
'<div style="mso-element:footnote-list"/>')}
|
|
811
|
+
#{WORD_FTR1}
|
|
812
|
+
OUTPUT
|
|
771
813
|
end
|
|
772
814
|
|
|
773
815
|
it "replaces id attributes with explicit a@name bookmarks" do
|
|
@@ -777,27 +819,30 @@ RSpec.describe Html2Doc do
|
|
|
777
819
|
<p id="b"/>
|
|
778
820
|
</div>
|
|
779
821
|
BODY
|
|
780
|
-
Html2Doc.process(html_input(simple_body),
|
|
781
|
-
|
|
782
|
-
|
|
783
|
-
|
|
784
|
-
|
|
785
|
-
<
|
|
786
|
-
|
|
787
|
-
|
|
788
|
-
|
|
789
|
-
|
|
790
|
-
|
|
822
|
+
Html2Doc.process(html_input(simple_body),
|
|
823
|
+
filename: "test", liststyles: { ul: "l1", ol: "l2" })
|
|
824
|
+
expect(guid_clean(File.read("test.doc", encoding: "utf-8")))
|
|
825
|
+
.to match_fuzzy(<<~OUTPUT)
|
|
826
|
+
#{WORD_HDR} #{DEFAULT_STYLESHEET} #{WORD_HDR_END}
|
|
827
|
+
#{word_body('<div>
|
|
828
|
+
<p class="MsoNormal"><a name="a" id="a"></a>Hello</p>
|
|
829
|
+
<p class="MsoNormal"><a name="b" id="b"></a></p>
|
|
830
|
+
</div>',
|
|
831
|
+
'<div style="mso-element:footnote-list"/>')}
|
|
832
|
+
#{WORD_FTR1}
|
|
833
|
+
OUTPUT
|
|
791
834
|
end
|
|
792
835
|
|
|
793
836
|
it "test image base64 image encoding" do
|
|
794
837
|
simple_body = '<img src="19160-6.png">'
|
|
795
|
-
Html2Doc.process(html_input(simple_body),
|
|
838
|
+
Html2Doc.process(html_input(simple_body),
|
|
839
|
+
filename: "spec/test", debug: true)
|
|
796
840
|
testdoc = File.read("spec/test.doc", encoding: "utf-8")
|
|
797
841
|
base64_image = testdoc[/image\/png\n\n(.*?)\n\n----/m, 1].gsub!("\n", "")
|
|
798
842
|
base64_image_basename = testdoc[%r{Content-ID: <([0-9a-z\-]+)\.png}m, 1]
|
|
799
843
|
doc_bin_image = Base64.strict_decode64(base64_image)
|
|
800
|
-
file_bin_image = IO
|
|
844
|
+
file_bin_image = IO
|
|
845
|
+
.read("spec/test_files/#{base64_image_basename}.png", mode: "rb")
|
|
801
846
|
expect(doc_bin_image).to eq file_bin_image
|
|
802
847
|
FileUtils.rm_rf %w[spec/test_files spec/test.doc spec/test.htm]
|
|
803
848
|
end
|