Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
98.06% |
101 / 103 |
|
93.33% |
14 / 15 |
CRAP | |
0.00% |
0 / 1 |
| Document | |
98.06% |
101 / 103 |
|
93.33% |
14 / 15 |
35 | |
0.00% |
0 / 1 |
| read | |
100.00% |
29 / 29 |
|
100.00% |
1 / 1 |
7 | |||
| markOpening | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| markClosing | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| markBackslash | |
66.67% |
4 / 6 |
|
0.00% |
0 / 1 |
2.15 | |||
| markNewline | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
2 | |||
| flush | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| flushControl | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| flushText | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
5 | |||
| setControl | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| pushText | |
100.00% |
5 / 5 |
|
100.00% |
1 / 1 |
3 | |||
| parseControl | |
100.00% |
27 / 27 |
|
100.00% |
1 / 1 |
3 | |||
| readParagraph | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| readStyle | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| readSkip | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| readText | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * This file is part of PHPWord - A pure PHP library for reading and writing |
| 4 | * word processing documents. |
| 5 | * |
| 6 | * PHPWord is free software distributed under the terms of the GNU Lesser |
| 7 | * General Public License version 3 as published by the Free Software Foundation. |
| 8 | * |
| 9 | * For the full copyright and license information, please read the LICENSE |
| 10 | * file that was distributed with this source code. For the full list of |
| 11 | * contributors, visit https://github.com/PHPOffice/PHPWord/contributors. |
| 12 | * |
| 13 | * @see https://github.com/PHPOffice/PHPWord |
| 14 | * |
| 15 | * @license http://www.gnu.org/licenses/lgpl.txt LGPL version 3 |
| 16 | */ |
| 17 | |
| 18 | namespace PhpOffice\PhpWord\Reader\RTF; |
| 19 | |
| 20 | use PhpOffice\PhpWord\PhpWord; |
| 21 | use PhpOffice\PhpWord\SimpleType\Jc; |
| 22 | |
| 23 | /** |
| 24 | * RTF document reader. |
| 25 | * |
| 26 | * References: |
| 27 | * - How to Write an RTF Reader http://latex2rtf.sourceforge.net/rtfspec_45.html |
| 28 | * - PHP rtfclass by Markus Fischer https://github.com/mfn/rtfclass |
| 29 | * - JavaScript RTF-parser by LazyGyu https://github.com/lazygyu/RTF-parser |
| 30 | * |
| 31 | * @since 0.11.0 |
| 32 | * |
| 33 | * @SuppressWarnings(PHPMD.UnusedPrivateMethod) |
| 34 | */ |
| 35 | class Document |
| 36 | { |
| 37 | /** @const int */ |
| 38 | const PARA = 'readParagraph'; |
| 39 | const STYL = 'readStyle'; |
| 40 | const SKIP = 'readSkip'; |
| 41 | |
| 42 | /** |
| 43 | * PhpWord object. |
| 44 | * |
| 45 | * @var \PhpOffice\PhpWord\PhpWord |
| 46 | */ |
| 47 | private $phpWord; |
| 48 | |
| 49 | /** |
| 50 | * Section object. |
| 51 | * |
| 52 | * @var \PhpOffice\PhpWord\Element\Section |
| 53 | */ |
| 54 | private $section; |
| 55 | |
| 56 | /** |
| 57 | * Textrun object. |
| 58 | * |
| 59 | * @var \PhpOffice\PhpWord\Element\TextRun |
| 60 | */ |
| 61 | private $textrun; |
| 62 | |
| 63 | /** |
| 64 | * RTF content. |
| 65 | * |
| 66 | * @var string |
| 67 | */ |
| 68 | public $rtf; |
| 69 | |
| 70 | /** |
| 71 | * Content length. |
| 72 | * |
| 73 | * @var int |
| 74 | */ |
| 75 | private $length = 0; |
| 76 | |
| 77 | /** |
| 78 | * Character index. |
| 79 | * |
| 80 | * @var int |
| 81 | */ |
| 82 | private $offset = 0; |
| 83 | |
| 84 | /** |
| 85 | * Current control word. |
| 86 | * |
| 87 | * @var string |
| 88 | */ |
| 89 | private $control = ''; |
| 90 | |
| 91 | /** |
| 92 | * Text content. |
| 93 | * |
| 94 | * @var string |
| 95 | */ |
| 96 | private $text = ''; |
| 97 | |
| 98 | /** |
| 99 | * Parsing a control word flag. |
| 100 | * |
| 101 | * @var bool |
| 102 | */ |
| 103 | private $isControl = false; |
| 104 | |
| 105 | /** |
| 106 | * First character flag: watch out for control symbols. |
| 107 | * |
| 108 | * @var bool |
| 109 | */ |
| 110 | private $isFirst = false; |
| 111 | |
| 112 | /** |
| 113 | * Group groups. |
| 114 | * |
| 115 | * @var array |
| 116 | */ |
| 117 | private $groups = []; |
| 118 | |
| 119 | /** |
| 120 | * Parser flags; not used. |
| 121 | * |
| 122 | * @var array |
| 123 | */ |
| 124 | private $flags = []; |
| 125 | |
| 126 | /** |
| 127 | * Parse RTF content. |
| 128 | * |
| 129 | * - Marks controlling characters `{`, `}`, and `\` |
| 130 | * - Removes line endings |
| 131 | * - Builds control words and control symbols |
| 132 | * - Pushes every other character into the text queue |
| 133 | * |
| 134 | * @todo Use `fread` stream for scalability |
| 135 | */ |
| 136 | public function read(PhpWord $phpWord): void |
| 137 | { |
| 138 | $markers = [ |
| 139 | 123 => 'markOpening', // { |
| 140 | 125 => 'markClosing', // } |
| 141 | 92 => 'markBackslash', // \ |
| 142 | 10 => 'markNewline', // LF |
| 143 | 13 => 'markNewline', // CR |
| 144 | ]; |
| 145 | |
| 146 | $this->phpWord = $phpWord; |
| 147 | $this->section = $phpWord->addSection(); |
| 148 | $this->textrun = $this->section->addTextRun(); |
| 149 | $this->length = strlen($this->rtf); |
| 150 | |
| 151 | $this->flags['paragraph'] = true; // Set paragraph flag from the beginning |
| 152 | |
| 153 | // Walk each characters |
| 154 | while ($this->offset < $this->length) { |
| 155 | $char = $this->rtf[$this->offset]; |
| 156 | $ascii = ord($char); |
| 157 | |
| 158 | if (isset($markers[$ascii])) { // Marker found: {, }, \, LF, or CR |
| 159 | $markerFunction = $markers[$ascii]; |
| 160 | $this->$markerFunction(); |
| 161 | } else { |
| 162 | if (false === $this->isControl) { // Non control word: Push character |
| 163 | $this->pushText($char); |
| 164 | } else { |
| 165 | if (preg_match('/^[a-zA-Z0-9-]?$/', $char)) { // No delimiter: Buffer control |
| 166 | $this->control .= $char; |
| 167 | $this->isFirst = false; |
| 168 | } else { // Delimiter found: Parse buffered control |
| 169 | if ($this->isFirst) { |
| 170 | $this->isFirst = false; |
| 171 | } else { |
| 172 | if (' ' == $char) { // Discard space as a control word delimiter |
| 173 | $this->flushControl(true); |
| 174 | } |
| 175 | } |
| 176 | } |
| 177 | } |
| 178 | } |
| 179 | ++$this->offset; |
| 180 | } |
| 181 | $this->flushText(); |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Mark opening braket `{` character. |
| 186 | */ |
| 187 | private function markOpening(): void |
| 188 | { |
| 189 | $this->flush(true); |
| 190 | array_push($this->groups, $this->flags); |
| 191 | } |
| 192 | |
| 193 | /** |
| 194 | * Mark closing braket `}` character. |
| 195 | */ |
| 196 | private function markClosing(): void |
| 197 | { |
| 198 | $this->flush(true); |
| 199 | $this->flags = array_pop($this->groups); |
| 200 | } |
| 201 | |
| 202 | /** |
| 203 | * Mark backslash `\` character. |
| 204 | */ |
| 205 | private function markBackslash(): void |
| 206 | { |
| 207 | if ($this->isFirst) { |
| 208 | $this->setControl(false); |
| 209 | $this->text .= '\\'; |
| 210 | } else { |
| 211 | $this->flush(); |
| 212 | $this->setControl(true); |
| 213 | $this->control = ''; |
| 214 | } |
| 215 | } |
| 216 | |
| 217 | /** |
| 218 | * Mark newline character: Flush control word because it's not possible to span multiline. |
| 219 | */ |
| 220 | private function markNewline(): void |
| 221 | { |
| 222 | if ($this->isControl) { |
| 223 | $this->flushControl(true); |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | /** |
| 228 | * Flush control word or text. |
| 229 | * |
| 230 | * @param bool $isControl |
| 231 | */ |
| 232 | private function flush($isControl = false): void |
| 233 | { |
| 234 | if ($this->isControl) { |
| 235 | $this->flushControl($isControl); |
| 236 | } else { |
| 237 | $this->flushText(); |
| 238 | } |
| 239 | } |
| 240 | |
| 241 | /** |
| 242 | * Flush control word. |
| 243 | * |
| 244 | * @param bool $isControl |
| 245 | */ |
| 246 | private function flushControl($isControl = false): void |
| 247 | { |
| 248 | if (1 === preg_match('/^([A-Za-z]+)(-?[0-9]*) ?$/', $this->control, $match)) { |
| 249 | [, $control, $parameter] = $match; |
| 250 | $this->parseControl($control, $parameter); |
| 251 | } |
| 252 | |
| 253 | if (true === $isControl) { |
| 254 | $this->setControl(false); |
| 255 | } |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * Flush text in queue. |
| 260 | */ |
| 261 | private function flushText(): void |
| 262 | { |
| 263 | if ($this->text != '') { |
| 264 | if (isset($this->flags['property'])) { // Set property |
| 265 | $this->flags['value'] = $this->text; |
| 266 | } else { // Set text |
| 267 | if (true === $this->flags['paragraph']) { |
| 268 | $this->flags['paragraph'] = false; |
| 269 | $this->flags['text'] = $this->text; |
| 270 | } |
| 271 | } |
| 272 | |
| 273 | // Add text if it's not flagged as skipped |
| 274 | if (!isset($this->flags['skipped'])) { |
| 275 | $this->readText(); |
| 276 | } |
| 277 | |
| 278 | $this->text = ''; |
| 279 | } |
| 280 | } |
| 281 | |
| 282 | /** |
| 283 | * Reset control word and first char state. |
| 284 | * |
| 285 | * @param bool $value |
| 286 | */ |
| 287 | private function setControl($value): void |
| 288 | { |
| 289 | $this->isControl = $value; |
| 290 | $this->isFirst = $value; |
| 291 | } |
| 292 | |
| 293 | /** |
| 294 | * Push text into queue. |
| 295 | * |
| 296 | * @param string $char |
| 297 | */ |
| 298 | private function pushText($char): void |
| 299 | { |
| 300 | if ('<' == $char) { |
| 301 | $this->text .= '<'; |
| 302 | } elseif ('>' == $char) { |
| 303 | $this->text .= '>'; |
| 304 | } else { |
| 305 | $this->text .= $char; |
| 306 | } |
| 307 | } |
| 308 | |
| 309 | /** |
| 310 | * Parse control. |
| 311 | * |
| 312 | * @param string $control |
| 313 | * @param string $parameter |
| 314 | */ |
| 315 | private function parseControl($control, $parameter): void |
| 316 | { |
| 317 | $controls = [ |
| 318 | 'par' => [self::PARA, 'paragraph', true], |
| 319 | 'b' => [self::STYL, 'font', 'bold', true], |
| 320 | 'i' => [self::STYL, 'font', 'italic', true], |
| 321 | 'u' => [self::STYL, 'font', 'underline', true], |
| 322 | 'strike' => [self::STYL, 'font', 'strikethrough', true], |
| 323 | 'fs' => [self::STYL, 'font', 'size', $parameter], |
| 324 | 'qc' => [self::STYL, 'paragraph', 'alignment', Jc::CENTER], |
| 325 | 'sa' => [self::STYL, 'paragraph', 'spaceAfter', $parameter], |
| 326 | 'fonttbl' => [self::SKIP, 'fonttbl', null], |
| 327 | 'colortbl' => [self::SKIP, 'colortbl', null], |
| 328 | 'info' => [self::SKIP, 'info', null], |
| 329 | 'generator' => [self::SKIP, 'generator', null], |
| 330 | 'title' => [self::SKIP, 'title', null], |
| 331 | 'subject' => [self::SKIP, 'subject', null], |
| 332 | 'category' => [self::SKIP, 'category', null], |
| 333 | 'keywords' => [self::SKIP, 'keywords', null], |
| 334 | 'comment' => [self::SKIP, 'comment', null], |
| 335 | 'shppict' => [self::SKIP, 'pic', null], |
| 336 | 'fldinst' => [self::SKIP, 'link', null], |
| 337 | ]; |
| 338 | |
| 339 | if (isset($controls[$control])) { |
| 340 | [$function] = $controls[$control]; |
| 341 | if (method_exists($this, $function)) { |
| 342 | $directives = $controls[$control]; |
| 343 | array_shift($directives); // remove the function variable; we won't need it |
| 344 | $this->$function($directives); |
| 345 | } |
| 346 | } |
| 347 | } |
| 348 | |
| 349 | /** |
| 350 | * Read paragraph. |
| 351 | * |
| 352 | * @param array $directives |
| 353 | */ |
| 354 | private function readParagraph($directives): void |
| 355 | { |
| 356 | [$property, $value] = $directives; |
| 357 | $this->textrun = $this->section->addTextRun(); |
| 358 | $this->flags[$property] = $value; |
| 359 | } |
| 360 | |
| 361 | /** |
| 362 | * Read style. |
| 363 | * |
| 364 | * @param array $directives |
| 365 | */ |
| 366 | private function readStyle($directives): void |
| 367 | { |
| 368 | [$style, $property, $value] = $directives; |
| 369 | $this->flags['styles'][$style][$property] = $value; |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * Read skip. |
| 374 | * |
| 375 | * @param array $directives |
| 376 | */ |
| 377 | private function readSkip($directives): void |
| 378 | { |
| 379 | [$property] = $directives; |
| 380 | $this->flags['property'] = $property; |
| 381 | $this->flags['skipped'] = true; |
| 382 | } |
| 383 | |
| 384 | /** |
| 385 | * Read text. |
| 386 | */ |
| 387 | private function readText(): void |
| 388 | { |
| 389 | $text = $this->textrun->addText($this->text); |
| 390 | if (isset($this->flags['styles']['font'])) { |
| 391 | $text->getFontStyle()->setStyleByArray($this->flags['styles']['font']); |
| 392 | } |
| 393 | } |
| 394 | } |