|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2016 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Plugins\FancyPants; |
|
9
|
|
|
|
|
10
|
|
|
use s9e\TextFormatter\Plugins\ParserBase; |
|
11
|
|
|
|
|
12
|
|
|
class Parser extends ParserBase |
|
13
|
|
|
{ |
|
14
|
|
|
/** |
|
15
|
|
|
* @var bool Whether currrent test contains a double quote character |
|
16
|
|
|
*/ |
|
17
|
|
|
protected $hasDoubleQuote; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var bool Whether currrent test contains a single quote character |
|
21
|
|
|
*/ |
|
22
|
|
|
protected $hasSingleQuote; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var string Text being parsed |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $text; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* {@inheritdoc} |
|
31
|
|
|
*/ |
|
32
|
65 |
|
public function parse($text, array $matches) |
|
33
|
|
|
{ |
|
34
|
65 |
|
$this->text = $text; |
|
35
|
65 |
|
$this->hasSingleQuote = (strpos($text, "'") !== false); |
|
36
|
65 |
|
$this->hasDoubleQuote = (strpos($text, '"') !== false); |
|
37
|
|
|
|
|
38
|
65 |
|
$this->parseSingleQuotes(); |
|
39
|
65 |
|
$this->parseSymbolsAfterDigits(); |
|
40
|
65 |
|
$this->parseSingleQuotePairs(); |
|
41
|
65 |
|
$this->parseDoubleQuotePairs(); |
|
42
|
65 |
|
$this->parseDashesAndEllipses(); |
|
43
|
65 |
|
$this->parseSymbolsInParentheses(); |
|
44
|
|
|
|
|
45
|
65 |
|
unset($this->text); |
|
46
|
65 |
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Add a fancy replacement tag |
|
50
|
|
|
* |
|
51
|
|
|
* @param integer $tagPos Position of the tag in the text |
|
52
|
|
|
* @param integer $tagLen Length of text consumed by the tag |
|
53
|
|
|
* @param string $chr Replacement character |
|
54
|
|
|
* @param integer $prio Tag's priority |
|
55
|
|
|
* @return \s9e\TextFormatter\Parser\Tag |
|
56
|
|
|
*/ |
|
57
|
65 |
|
protected function addTag($tagPos, $tagLen, $chr, $prio = 0) |
|
58
|
|
|
{ |
|
59
|
65 |
|
$tag = $this->parser->addSelfClosingTag($this->config['tagName'], $tagPos, $tagLen, $prio); |
|
60
|
65 |
|
$tag->setAttribute($this->config['attrName'], $chr); |
|
61
|
|
|
|
|
62
|
65 |
|
return $tag; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Parse dashes and ellipses |
|
67
|
|
|
* |
|
68
|
|
|
* Does en dash –, em dash — and ellipsis … |
|
69
|
|
|
* |
|
70
|
|
|
* @return void |
|
71
|
|
|
*/ |
|
72
|
65 |
|
protected function parseDashesAndEllipses() |
|
73
|
|
|
{ |
|
74
|
65 |
|
if (strpos($this->text, '...') === false && strpos($this->text, '--') === false) |
|
75
|
65 |
|
{ |
|
76
|
53 |
|
return; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
$chrs = [ |
|
80
|
12 |
|
'--' => "\xE2\x80\x93", |
|
81
|
12 |
|
'---' => "\xE2\x80\x94", |
|
82
|
|
|
'...' => "\xE2\x80\xA6" |
|
83
|
12 |
|
]; |
|
84
|
12 |
|
$regexp = '/---?|\\.\\.\\./S'; |
|
85
|
12 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE); |
|
86
|
12 |
|
foreach ($matches[0] as $m) |
|
87
|
|
|
{ |
|
88
|
12 |
|
$this->addTag($m[1], strlen($m[0]), $chrs[$m[0]]); |
|
89
|
12 |
|
} |
|
90
|
12 |
|
} |
|
91
|
|
|
|
|
92
|
|
|
/** |
|
93
|
|
|
* Parse pairs of double quotes |
|
94
|
|
|
* |
|
95
|
|
|
* Does quote pairs “” -- must be done separately to handle nesting |
|
96
|
|
|
* |
|
97
|
|
|
* @return void |
|
98
|
|
|
*/ |
|
99
|
65 |
|
protected function parseDoubleQuotePairs() |
|
100
|
|
|
{ |
|
101
|
65 |
|
if ($this->hasDoubleQuote) |
|
102
|
65 |
|
{ |
|
103
|
11 |
|
$this->parseQuotePairs( |
|
104
|
11 |
|
'/(?<![0-9\\pL])"[^"\\n]+"(?![0-9\\pL])/uS', |
|
105
|
11 |
|
"\xE2\x80\x9C", |
|
106
|
|
|
"\xE2\x80\x9D" |
|
107
|
11 |
|
); |
|
108
|
11 |
|
} |
|
109
|
65 |
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Parse pairs of quotes |
|
113
|
|
|
* |
|
114
|
|
|
* @param string $regexp Regexp used to identify quote pairs |
|
115
|
|
|
* @param string $leftQuote Fancy replacement for left quote |
|
116
|
|
|
* @param string $rightQuote Fancy replacement for right quote |
|
117
|
|
|
* @return void |
|
118
|
|
|
*/ |
|
119
|
37 |
|
protected function parseQuotePairs($regexp, $leftQuote, $rightQuote) |
|
120
|
|
|
{ |
|
121
|
37 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE); |
|
122
|
37 |
|
foreach ($matches[0] as $m) |
|
123
|
|
|
{ |
|
124
|
10 |
|
$left = $this->addTag($m[1], 1, $leftQuote); |
|
125
|
10 |
|
$right = $this->addTag($m[1] + strlen($m[0]) - 1, 1, $rightQuote); |
|
126
|
|
|
|
|
127
|
|
|
// Cascade left tag's invalidation to the right so that if we skip the left quote, |
|
128
|
|
|
// the right quote remains untouched |
|
129
|
10 |
|
$left->cascadeInvalidationTo($right); |
|
130
|
37 |
|
} |
|
131
|
37 |
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Parse pairs of single quotes |
|
135
|
|
|
* |
|
136
|
|
|
* Does quote pairs ‘’ must be done separately to handle nesting |
|
137
|
|
|
* |
|
138
|
|
|
* @return void |
|
139
|
|
|
*/ |
|
140
|
65 |
|
protected function parseSingleQuotePairs() |
|
141
|
|
|
{ |
|
142
|
65 |
|
if ($this->hasSingleQuote) |
|
143
|
65 |
|
{ |
|
144
|
30 |
|
$this->parseQuotePairs( |
|
145
|
30 |
|
"/(?<![0-9\\pL])'[^'\\n]+'(?![0-9\\pL])/uS", |
|
146
|
30 |
|
"\xE2\x80\x98", |
|
147
|
|
|
"\xE2\x80\x99" |
|
148
|
30 |
|
); |
|
149
|
30 |
|
} |
|
150
|
65 |
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* Parse single quotes in general |
|
154
|
|
|
* |
|
155
|
|
|
* Does apostrophes ’ after a letter or at the beginning of a word or a couple of digits |
|
156
|
|
|
* |
|
157
|
|
|
* @return void |
|
158
|
|
|
*/ |
|
159
|
65 |
|
protected function parseSingleQuotes() |
|
160
|
|
|
{ |
|
161
|
65 |
|
if (!$this->hasSingleQuote) |
|
162
|
65 |
|
{ |
|
163
|
35 |
|
return; |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
30 |
|
$regexp = "/(?<=\\pL)'|(?<!\\S)'(?=\\pL|[0-9]{2})/uS"; |
|
167
|
30 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE); |
|
168
|
30 |
|
foreach ($matches[0] as $m) |
|
169
|
|
|
{ |
|
170
|
|
|
// Give this tag a worse priority than default so that quote pairs take precedence |
|
171
|
20 |
|
$this->addTag($m[1], 1, "\xE2\x80\x99", 10); |
|
172
|
30 |
|
} |
|
173
|
30 |
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* Parse symbols found after digits |
|
177
|
|
|
* |
|
178
|
|
|
* Does symbols found after a digit: |
|
179
|
|
|
* - apostrophe ’ if it's followed by an "s" as in 80's |
|
180
|
|
|
* - prime ′ and double prime ″ |
|
181
|
|
|
* - multiply sign × if it's followed by an optional space and another digit |
|
182
|
|
|
* |
|
183
|
|
|
* @return void |
|
184
|
|
|
*/ |
|
185
|
65 |
|
protected function parseSymbolsAfterDigits() |
|
186
|
|
|
{ |
|
187
|
65 |
|
if (!$this->hasSingleQuote && !$this->hasDoubleQuote && strpos($this->text, 'x') === false) |
|
188
|
65 |
|
{ |
|
189
|
24 |
|
return; |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
41 |
|
$regexp = '/[0-9](?>\'s|["\']? ?x(?= ?[0-9])|["\'])/S'; |
|
193
|
41 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE); |
|
194
|
41 |
|
foreach ($matches[0] as $m) |
|
195
|
|
|
{ |
|
196
|
|
|
// Test for a multiply sign at the end |
|
197
|
15 |
|
if (substr($m[0], -1) === 'x') |
|
198
|
15 |
|
{ |
|
199
|
9 |
|
$this->addTag($m[1] + strlen($m[0]) - 1, 1, "\xC3\x97"); |
|
200
|
9 |
|
} |
|
201
|
|
|
|
|
202
|
|
|
// Test for a apostrophe/prime right after the digit |
|
203
|
15 |
|
$c = $m[0][1]; |
|
204
|
15 |
|
if ($c === "'" || $c === '"') |
|
205
|
15 |
|
{ |
|
206
|
11 |
|
if (substr($m[0], 1, 2) === "'s") |
|
207
|
11 |
|
{ |
|
208
|
|
|
// 80's -- use an apostrophe |
|
209
|
2 |
|
$chr = "\xE2\x80\x99"; |
|
210
|
2 |
|
} |
|
211
|
|
|
else |
|
212
|
|
|
{ |
|
213
|
|
|
// 12' or 12" -- use a prime |
|
214
|
9 |
|
$chr = ($c === "'") ? "\xE2\x80\xB2" : "\xE2\x80\xB3"; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
11 |
|
$this->addTag($m[1] + 1, 1, $chr); |
|
218
|
11 |
|
} |
|
219
|
41 |
|
} |
|
220
|
41 |
|
} |
|
221
|
|
|
|
|
222
|
|
|
/** |
|
223
|
|
|
* Parse symbols found in parentheses such as (c) |
|
224
|
|
|
* |
|
225
|
|
|
* Does symbols ©, ® and ™ |
|
226
|
|
|
* |
|
227
|
|
|
* @return void |
|
228
|
|
|
*/ |
|
229
|
65 |
|
protected function parseSymbolsInParentheses() |
|
230
|
|
|
{ |
|
231
|
65 |
|
if (strpos($this->text, '(') === false) |
|
232
|
65 |
|
{ |
|
233
|
53 |
|
return; |
|
234
|
|
|
} |
|
235
|
|
|
|
|
236
|
|
|
$chrs = [ |
|
237
|
12 |
|
'(c)' => "\xC2\xA9", |
|
238
|
12 |
|
'(r)' => "\xC2\xAE", |
|
239
|
|
|
'(tm)' => "\xE2\x84\xA2" |
|
240
|
12 |
|
]; |
|
241
|
12 |
|
$regexp = '/\\((?>c|r|tm)\\)/i'; |
|
242
|
12 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE); |
|
243
|
12 |
|
foreach ($matches[0] as $m) |
|
244
|
|
|
{ |
|
245
|
12 |
|
$this->addTag($m[1], strlen($m[0]), $chrs[strtr($m[0], 'CMRT', 'cmrt')]); |
|
246
|
12 |
|
} |
|
247
|
|
|
} |
|
248
|
|
|
} |
s9e /
TextFormatter
