|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
use PhpSpellcheck\Text; |
|
6
|
|
|
use PhpSpellcheck\TextProcessor\MarkdownRemover; |
|
7
|
|
|
use PHPUnit\Framework\TestCase; |
|
8
|
|
|
|
|
9
|
|
|
class MarkdownRemoverTest extends TestCase |
|
10
|
|
|
{ |
|
11
|
|
|
public function testShouldStripOutRemainingMarkdown(): void |
|
12
|
|
|
{ |
|
13
|
|
|
$string = '*Javascript* developers are the _best_.'; |
|
14
|
|
|
$expected = 'Javascript developers are the best.'; |
|
15
|
|
|
|
|
16
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
17
|
|
|
} |
|
18
|
|
|
|
|
19
|
|
|
public function testShouldLeaveNonMatchingMarkdownMarkdown(): void |
|
20
|
|
|
{ |
|
21
|
|
|
$string = '*Javascript* developers* are the _best_.'; |
|
22
|
|
|
$expected = 'Javascript developers* are the best.'; |
|
23
|
|
|
|
|
24
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
public function testShouldLeaveNonMatchingMarkdownButStripEmptyAnchors(): void |
|
28
|
|
|
{ |
|
29
|
|
|
$string = '*Javascript* [developers]()* are the _best_.'; |
|
30
|
|
|
$expected = 'Javascript developers* are the best.'; |
|
31
|
|
|
|
|
32
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
public function testShouldStripHtml(): void |
|
36
|
|
|
{ |
|
37
|
|
|
$string = '<p>Hello World</p>'; |
|
38
|
|
|
$expected = 'Hello World'; |
|
39
|
|
|
|
|
40
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
public function testShouldStripAnchors(): void |
|
44
|
|
|
{ |
|
45
|
|
|
$string = '*Javascript* [developers](https://engineering.condenast.io/)* are the _best_.'; |
|
46
|
|
|
$expected = 'Javascript developers* are the best.'; |
|
47
|
|
|
|
|
48
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
public function testShouldStripImgTags(): void |
|
52
|
|
|
{ |
|
53
|
|
|
$string = '*Javascript* developers are the _best_.'; |
|
54
|
|
|
$expected = 'Javascript developers are the best.'; |
|
55
|
|
|
|
|
56
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
public function testShouldUseTheAltTextOfAnImageIfItIsProvided(): void |
|
60
|
|
|
{ |
|
61
|
|
|
$string = ''; |
|
62
|
|
|
$expected = 'This is the alt-text'; |
|
63
|
|
|
|
|
64
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
public function testShouldStripCodeTags(): void |
|
68
|
|
|
{ |
|
69
|
|
|
$string = 'In `Getting Started` we set up `something` foo.'; |
|
70
|
|
|
$expected = 'In Getting Started we set up something foo.'; |
|
71
|
|
|
|
|
72
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
public function testShouldLeaveHashtagsInHeadings(): void |
|
76
|
|
|
{ |
|
77
|
|
|
$string = '## This #heading contains #hashtags'; |
|
78
|
|
|
$expected = 'This #heading contains #hashtags'; |
|
79
|
|
|
|
|
80
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
public function testShouldRemoveHeadingsTrailingHashtags(): void |
|
84
|
|
|
{ |
|
85
|
|
|
$string = '## This #heading contains #hashtags ##'; |
|
86
|
|
|
$expected = 'This #heading contains #hashtags'; |
|
87
|
|
|
|
|
88
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
public function testShouldRemoveHeadingsHashtags(): void |
|
92
|
|
|
{ |
|
93
|
|
|
$string = '## This #heading contains #hashtags'; |
|
94
|
|
|
$expected = 'This #heading contains #hashtags'; |
|
95
|
|
|
|
|
96
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
public function testShouldRemoveEmphasis(): void |
|
100
|
|
|
{ |
|
101
|
|
|
$string = 'I italicized an *I* and it _made_ me *sad*.'; |
|
102
|
|
|
$expected = 'I italicized an I and it made me sad.'; |
|
103
|
|
|
|
|
104
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
public function testShouldRemoveDoubleEmphasis(): void |
|
108
|
|
|
{ |
|
109
|
|
|
$string = '**this sentence has __double styling__**'; |
|
110
|
|
|
$expected = 'this sentence has double styling'; |
|
111
|
|
|
|
|
112
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
public function testShouldRemoveHorizontalRules(): void |
|
116
|
|
|
{ |
|
117
|
|
|
$string = "Some text on a line\n\n---\n\nA line below"; |
|
118
|
|
|
$expected = "Some text on a line\n\n\n\n\nA line below"; |
|
119
|
|
|
|
|
120
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
public function testShouldRemoveHorizontalRulesAndReplaceByAtLeastABreakLine(): void |
|
124
|
|
|
{ |
|
125
|
|
|
$string = "Some text on a line\n---\nA line below"; |
|
126
|
|
|
$expected = "Some text on a line\n\n\nA line below"; |
|
127
|
|
|
|
|
128
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
public function testShouldRemoveHorizontalRulesWithSpaceSeparatedAsterisks(): void |
|
132
|
|
|
{ |
|
133
|
|
|
$string = "Some text on a line\n\n* * *\n\nA line below"; |
|
134
|
|
|
$expected = "Some text on a line\n\n\n\n\nA line below"; |
|
135
|
|
|
|
|
136
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
public function testShouldRemoveBlockquotes(): void |
|
140
|
|
|
{ |
|
141
|
|
|
$string = '>I am a blockquote'; |
|
142
|
|
|
$expected = 'I am a blockquote'; |
|
143
|
|
|
|
|
144
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
public function testShouldRemoveBlockquotesWithSpaces(): void |
|
148
|
|
|
{ |
|
149
|
|
|
$string = '> I am a blockquote'; |
|
150
|
|
|
$expected = 'I am a blockquote'; |
|
151
|
|
|
|
|
152
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
|
|
public function testShouldRemoveIndentedBlockquotes(): void |
|
156
|
|
|
{ |
|
157
|
|
|
$string = ' > I am a blockquote'; |
|
158
|
|
|
$expected = 'I am a blockquote'; |
|
159
|
|
|
|
|
160
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
161
|
|
|
|
|
162
|
|
|
$string = ' > I am a blockquote'; |
|
163
|
|
|
|
|
164
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
public function testShouldNotRemoveGreaterThanSigns(): void |
|
168
|
|
|
{ |
|
169
|
|
|
$tests = [ |
|
170
|
|
|
['string' => '100 > 0', 'expected' => '100 > 0'], |
|
171
|
|
|
['string' => '100 >= 0', 'expected' => '100 >= 0'], |
|
172
|
|
|
['string' => '100>0', 'expected' => '100>0'], |
|
173
|
|
|
['string' => '> 100 > 0', 'expected' => '100 > 0'], |
|
174
|
|
|
['string' => '1 < 100', 'expected' => '1 < 100'], |
|
175
|
|
|
['string' => '1 <= 100', 'expected' => '1 <= 100'], |
|
176
|
|
|
]; |
|
177
|
|
|
|
|
178
|
|
|
foreach ($tests as $test) { |
|
179
|
|
|
$this->assertSame( |
|
180
|
|
|
$test['expected'], |
|
181
|
|
|
(new MarkdownRemover())->process(Text::utf8($test['string']))->getContent() |
|
182
|
|
|
); |
|
183
|
|
|
} |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
|
|
public function testShouldStripUnorderedListLeaders(): void |
|
187
|
|
|
{ |
|
188
|
|
|
$string = "Some text on a line\n\n* A list Item\n* Another list item"; |
|
189
|
|
|
$expected = "Some text on a line\n\nA list Item\nAnother list item"; |
|
190
|
|
|
|
|
191
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
public function testShouldStripOrderedListLeaders(): void |
|
195
|
|
|
{ |
|
196
|
|
|
$string = "Some text on a line\n\n* A list Item\n* Another list item"; |
|
197
|
|
|
$expected = "Some text on a line\n\nA list Item\nAnother list item"; |
|
198
|
|
|
|
|
199
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
200
|
|
|
} |
|
201
|
|
|
|
|
202
|
|
|
public function testShouldStripOrderedListLeadersKeepingIndentation(): void |
|
203
|
|
|
{ |
|
204
|
|
|
$string = "Some text on a line\n\n* A list Item\n * Another list item"; |
|
205
|
|
|
$expected = "Some text on a line\n\n A list Item\n Another list item"; |
|
206
|
|
|
|
|
207
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
|
|
public function testShouldStripCodeBlocks(): void |
|
211
|
|
|
{ |
|
212
|
|
|
$string = <<<CODE |
|
213
|
|
|
```js |
|
214
|
|
|
test |
|
215
|
|
|
``` |
|
216
|
|
|
CODE; |
|
217
|
|
|
|
|
218
|
|
|
$expected = <<<CODE |
|
219
|
|
|
|
|
220
|
|
|
test |
|
221
|
|
|
|
|
222
|
|
|
CODE; |
|
223
|
|
|
|
|
224
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
|
|
public function testShouldHandleParagraphsWithMarkdown(): void |
|
228
|
|
|
{ |
|
229
|
|
|
$string = "\n## This is a heading ##\n\nThis is a paragraph with [a link](http://www.disney.com/).\n\n### This is another heading\n\nIn `Getting Started` we set up `something` foo.\n\n * Some list\n * With items\n * Even indented"; |
|
230
|
|
|
$expected = "\nThis is a heading\n\nThis is a paragraph with a link.\n\nThis is another heading\n\nIn Getting Started we set up something foo.\n\n Some list\n With items\n Even indented"; |
|
231
|
|
|
|
|
232
|
|
|
$this->assertSame($expected, (new MarkdownRemover())->process(Text::utf8($string))->getContent()); |
|
233
|
|
|
} |
|
234
|
|
|
} |
|
235
|
|
|
|