1 | <?php |
---|
2 | // |
---|
3 | // FPDI - Version 1.4 |
---|
4 | // |
---|
5 | // Copyright 2004-2010 Setasign - Jan Slabon |
---|
6 | // |
---|
7 | // Licensed under the Apache License, Version 2.0 (the "License"); |
---|
8 | // you may not use this file except in compliance with the License. |
---|
9 | // You may obtain a copy of the License at |
---|
10 | // |
---|
11 | // http://www.apache.org/licenses/LICENSE-2.0 |
---|
12 | // |
---|
13 | // Unless required by applicable law or agreed to in writing, software |
---|
14 | // distributed under the License is distributed on an "AS IS" BASIS, |
---|
15 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
---|
16 | // See the License for the specific language governing permissions and |
---|
17 | // limitations under the License. |
---|
18 | // |
---|
19 | |
---|
20 | require_once('pdf_parser.php'); |
---|
21 | |
---|
22 | class fpdi_pdf_parser extends pdf_parser { |
---|
23 | |
---|
24 | /** |
---|
25 | * Pages |
---|
26 | * Index beginns at 0 |
---|
27 | * |
---|
28 | * @var array |
---|
29 | */ |
---|
30 | var $pages; |
---|
31 | |
---|
32 | /** |
---|
33 | * Page count |
---|
34 | * @var integer |
---|
35 | */ |
---|
36 | var $page_count; |
---|
37 | |
---|
38 | /** |
---|
39 | * actual page number |
---|
40 | * @var integer |
---|
41 | */ |
---|
42 | var $pageno; |
---|
43 | |
---|
44 | /** |
---|
45 | * PDF Version of imported Document |
---|
46 | * @var string |
---|
47 | */ |
---|
48 | var $pdfVersion; |
---|
49 | |
---|
50 | /** |
---|
51 | * FPDI Reference |
---|
52 | * @var object |
---|
53 | */ |
---|
54 | var $fpdi; |
---|
55 | |
---|
56 | /** |
---|
57 | * Available BoxTypes |
---|
58 | * |
---|
59 | * @var array |
---|
60 | */ |
---|
61 | var $availableBoxes = array('/MediaBox', '/CropBox', '/BleedBox', '/TrimBox', '/ArtBox'); |
---|
62 | |
---|
63 | /** |
---|
64 | * Constructor |
---|
65 | * |
---|
66 | * @param string $filename Source-Filename |
---|
67 | * @param object $fpdi Object of type fpdi |
---|
68 | */ |
---|
69 | function fpdi_pdf_parser($filename, &$fpdi) { |
---|
70 | $this->fpdi =& $fpdi; |
---|
71 | |
---|
72 | parent::pdf_parser($filename); |
---|
73 | |
---|
74 | // resolve Pages-Dictonary |
---|
75 | $pages = $this->pdf_resolve_object($this->c, $this->root[1][1]['/Pages']); |
---|
76 | |
---|
77 | // Read pages |
---|
78 | $this->read_pages($this->c, $pages, $this->pages); |
---|
79 | |
---|
80 | // count pages; |
---|
81 | $this->page_count = count($this->pages); |
---|
82 | } |
---|
83 | |
---|
84 | /** |
---|
85 | * Overwrite parent::error() |
---|
86 | * |
---|
87 | * @param string $msg Error-Message |
---|
88 | */ |
---|
89 | function error($msg) { |
---|
90 | $this->fpdi->error($msg); |
---|
91 | } |
---|
92 | |
---|
93 | /** |
---|
94 | * Get pagecount from sourcefile |
---|
95 | * |
---|
96 | * @return int |
---|
97 | */ |
---|
98 | function getPageCount() { |
---|
99 | return $this->page_count; |
---|
100 | } |
---|
101 | |
---|
102 | |
---|
103 | /** |
---|
104 | * Set pageno |
---|
105 | * |
---|
106 | * @param int $pageno Pagenumber to use |
---|
107 | */ |
---|
108 | function setPageno($pageno) { |
---|
109 | $pageno = ((int) $pageno) - 1; |
---|
110 | |
---|
111 | if ($pageno < 0 || $pageno >= $this->getPageCount()) { |
---|
112 | $this->fpdi->error('Pagenumber is wrong!'); |
---|
113 | } |
---|
114 | |
---|
115 | $this->pageno = $pageno; |
---|
116 | } |
---|
117 | |
---|
118 | /** |
---|
119 | * Get page-resources from current page |
---|
120 | * |
---|
121 | * @return array |
---|
122 | */ |
---|
123 | function getPageResources() { |
---|
124 | return $this->_getPageResources($this->pages[$this->pageno]); |
---|
125 | } |
---|
126 | |
---|
127 | /** |
---|
128 | * Get page-resources from /Page |
---|
129 | * |
---|
130 | * @param array $obj Array of pdf-data |
---|
131 | */ |
---|
132 | function _getPageResources ($obj) { // $obj = /Page |
---|
133 | $obj = $this->pdf_resolve_object($this->c, $obj); |
---|
134 | |
---|
135 | // If the current object has a resources |
---|
136 | // dictionary associated with it, we use |
---|
137 | // it. Otherwise, we move back to its |
---|
138 | // parent object. |
---|
139 | if (isset ($obj[1][1]['/Resources'])) { |
---|
140 | $res = $this->pdf_resolve_object($this->c, $obj[1][1]['/Resources']); |
---|
141 | if ($res[0] == PDF_TYPE_OBJECT) |
---|
142 | return $res[1]; |
---|
143 | return $res; |
---|
144 | } else { |
---|
145 | if (!isset ($obj[1][1]['/Parent'])) { |
---|
146 | return false; |
---|
147 | } else { |
---|
148 | $res = $this->_getPageResources($obj[1][1]['/Parent']); |
---|
149 | if ($res[0] == PDF_TYPE_OBJECT) |
---|
150 | return $res[1]; |
---|
151 | return $res; |
---|
152 | } |
---|
153 | } |
---|
154 | } |
---|
155 | |
---|
156 | |
---|
157 | /** |
---|
158 | * Get content of current page |
---|
159 | * |
---|
160 | * If more /Contents is an array, the streams are concated |
---|
161 | * |
---|
162 | * @return string |
---|
163 | */ |
---|
164 | function getContent() { |
---|
165 | $buffer = ''; |
---|
166 | |
---|
167 | if (isset($this->pages[$this->pageno][1][1]['/Contents'])) { |
---|
168 | $contents = $this->_getPageContent($this->pages[$this->pageno][1][1]['/Contents']); |
---|
169 | foreach($contents AS $tmp_content) { |
---|
170 | $buffer .= $this->_rebuildContentStream($tmp_content) . ' '; |
---|
171 | } |
---|
172 | } |
---|
173 | |
---|
174 | return $buffer; |
---|
175 | } |
---|
176 | |
---|
177 | |
---|
178 | /** |
---|
179 | * Resolve all content-objects |
---|
180 | * |
---|
181 | * @param array $content_ref |
---|
182 | * @return array |
---|
183 | */ |
---|
184 | function _getPageContent($content_ref) { |
---|
185 | $contents = array(); |
---|
186 | |
---|
187 | if ($content_ref[0] == PDF_TYPE_OBJREF) { |
---|
188 | $content = $this->pdf_resolve_object($this->c, $content_ref); |
---|
189 | if ($content[1][0] == PDF_TYPE_ARRAY) { |
---|
190 | $contents = $this->_getPageContent($content[1]); |
---|
191 | } else { |
---|
192 | $contents[] = $content; |
---|
193 | } |
---|
194 | } else if ($content_ref[0] == PDF_TYPE_ARRAY) { |
---|
195 | foreach ($content_ref[1] AS $tmp_content_ref) { |
---|
196 | $contents = array_merge($contents,$this->_getPageContent($tmp_content_ref)); |
---|
197 | } |
---|
198 | } |
---|
199 | |
---|
200 | return $contents; |
---|
201 | } |
---|
202 | |
---|
203 | |
---|
204 | /** |
---|
205 | * Rebuild content-streams |
---|
206 | * |
---|
207 | * @param array $obj |
---|
208 | * @return string |
---|
209 | */ |
---|
210 | function _rebuildContentStream($obj) { |
---|
211 | $filters = array(); |
---|
212 | |
---|
213 | if (isset($obj[1][1]['/Filter'])) { |
---|
214 | $_filter = $obj[1][1]['/Filter']; |
---|
215 | |
---|
216 | if ($_filter[0] == PDF_TYPE_OBJREF) { |
---|
217 | $tmpFilter = $this->pdf_resolve_object($this->c, $_filter); |
---|
218 | $_filter = $tmpFilter[1]; |
---|
219 | } |
---|
220 | |
---|
221 | if ($_filter[0] == PDF_TYPE_TOKEN) { |
---|
222 | $filters[] = $_filter; |
---|
223 | } else if ($_filter[0] == PDF_TYPE_ARRAY) { |
---|
224 | $filters = $_filter[1]; |
---|
225 | } |
---|
226 | } |
---|
227 | |
---|
228 | $stream = $obj[2][1]; |
---|
229 | |
---|
230 | foreach ($filters AS $_filter) { |
---|
231 | switch ($_filter[1]) { |
---|
232 | case '/FlateDecode': |
---|
233 | // $stream .= "\x0F\x0D"; // in an errorious stream this suffix could work |
---|
234 | if (function_exists('gzuncompress')) { |
---|
235 | $stream = (strlen($stream) > 0) ? @gzuncompress($stream) : ''; |
---|
236 | } else { |
---|
237 | $this->error(sprintf('To handle %s filter, please compile php with zlib support.',$_filter[1])); |
---|
238 | } |
---|
239 | |
---|
240 | if ($stream === false) { |
---|
241 | $this->error('Error while decompressing stream.'); |
---|
242 | } |
---|
243 | break; |
---|
244 | case '/LZWDecode': |
---|
245 | include_once('filters/FilterLZW_FPDI.php'); |
---|
246 | $decoder = new FilterLZW_FPDI($this->fpdi); |
---|
247 | $stream = $decoder->decode($stream); |
---|
248 | break; |
---|
249 | case '/ASCII85Decode': |
---|
250 | include_once('filters/FilterASCII85_FPDI.php'); |
---|
251 | $decoder = new FilterASCII85_FPDI($this->fpdi); |
---|
252 | $stream = $decoder->decode($stream); |
---|
253 | break; |
---|
254 | case null: |
---|
255 | $stream = $stream; |
---|
256 | break; |
---|
257 | default: |
---|
258 | $this->error(sprintf('Unsupported Filter: %s',$_filter[1])); |
---|
259 | } |
---|
260 | } |
---|
261 | |
---|
262 | return $stream; |
---|
263 | } |
---|
264 | |
---|
265 | |
---|
266 | /** |
---|
267 | * Get a Box from a page |
---|
268 | * Arrayformat is same as used by fpdf_tpl |
---|
269 | * |
---|
270 | * @param array $page a /Page |
---|
271 | * @param string $box_index Type of Box @see $availableBoxes |
---|
272 | * @param float Scale factor from user space units to points |
---|
273 | * @return array |
---|
274 | */ |
---|
275 | function getPageBox($page, $box_index, $k) { |
---|
276 | $page = $this->pdf_resolve_object($this->c, $page); |
---|
277 | $box = null; |
---|
278 | if (isset($page[1][1][$box_index])) |
---|
279 | $box =& $page[1][1][$box_index]; |
---|
280 | |
---|
281 | if (!is_null($box) && $box[0] == PDF_TYPE_OBJREF) { |
---|
282 | $tmp_box = $this->pdf_resolve_object($this->c, $box); |
---|
283 | $box = $tmp_box[1]; |
---|
284 | } |
---|
285 | |
---|
286 | if (!is_null($box) && $box[0] == PDF_TYPE_ARRAY) { |
---|
287 | $b =& $box[1]; |
---|
288 | return array('x' => $b[0][1]/$k, |
---|
289 | 'y' => $b[1][1]/$k, |
---|
290 | 'w' => abs($b[0][1]-$b[2][1])/$k, |
---|
291 | 'h' => abs($b[1][1]-$b[3][1])/$k, |
---|
292 | 'llx' => min($b[0][1], $b[2][1])/$k, |
---|
293 | 'lly' => min($b[1][1], $b[3][1])/$k, |
---|
294 | 'urx' => max($b[0][1], $b[2][1])/$k, |
---|
295 | 'ury' => max($b[1][1], $b[3][1])/$k, |
---|
296 | ); |
---|
297 | } else if (!isset ($page[1][1]['/Parent'])) { |
---|
298 | return false; |
---|
299 | } else { |
---|
300 | return $this->getPageBox($this->pdf_resolve_object($this->c, $page[1][1]['/Parent']), $box_index, $k); |
---|
301 | } |
---|
302 | } |
---|
303 | |
---|
304 | /** |
---|
305 | * Get all page boxes by page no |
---|
306 | * |
---|
307 | * @param int The page number |
---|
308 | * @param float Scale factor from user space units to points |
---|
309 | * @return array |
---|
310 | */ |
---|
311 | function getPageBoxes($pageno, $k) { |
---|
312 | return $this->_getPageBoxes($this->pages[$pageno-1], $k); |
---|
313 | } |
---|
314 | |
---|
315 | /** |
---|
316 | * Get all boxes from /Page |
---|
317 | * |
---|
318 | * @param array a /Page |
---|
319 | * @return array |
---|
320 | */ |
---|
321 | function _getPageBoxes($page, $k) { |
---|
322 | $boxes = array(); |
---|
323 | |
---|
324 | foreach($this->availableBoxes AS $box) { |
---|
325 | if ($_box = $this->getPageBox($page, $box, $k)) { |
---|
326 | $boxes[$box] = $_box; |
---|
327 | } |
---|
328 | } |
---|
329 | |
---|
330 | return $boxes; |
---|
331 | } |
---|
332 | |
---|
333 | /** |
---|
334 | * Get the page rotation by pageno |
---|
335 | * |
---|
336 | * @param integer $pageno |
---|
337 | * @return array |
---|
338 | */ |
---|
339 | function getPageRotation($pageno) { |
---|
340 | return $this->_getPageRotation($this->pages[$pageno-1]); |
---|
341 | } |
---|
342 | |
---|
343 | function _getPageRotation($obj) { // $obj = /Page |
---|
344 | $obj = $this->pdf_resolve_object($this->c, $obj); |
---|
345 | if (isset ($obj[1][1]['/Rotate'])) { |
---|
346 | $res = $this->pdf_resolve_object($this->c, $obj[1][1]['/Rotate']); |
---|
347 | if ($res[0] == PDF_TYPE_OBJECT) |
---|
348 | return $res[1]; |
---|
349 | return $res; |
---|
350 | } else { |
---|
351 | if (!isset ($obj[1][1]['/Parent'])) { |
---|
352 | return false; |
---|
353 | } else { |
---|
354 | $res = $this->_getPageRotation($obj[1][1]['/Parent']); |
---|
355 | if ($res[0] == PDF_TYPE_OBJECT) |
---|
356 | return $res[1]; |
---|
357 | return $res; |
---|
358 | } |
---|
359 | } |
---|
360 | } |
---|
361 | |
---|
362 | /** |
---|
363 | * Read all /Page(es) |
---|
364 | * |
---|
365 | * @param object pdf_context |
---|
366 | * @param array /Pages |
---|
367 | * @param array the result-array |
---|
368 | */ |
---|
369 | function read_pages(&$c, &$pages, &$result) { |
---|
370 | // Get the kids dictionary |
---|
371 | $_kids = $this->pdf_resolve_object ($c, $pages[1][1]['/Kids']); |
---|
372 | |
---|
373 | if (!is_array($_kids)) |
---|
374 | $this->error('Cannot find /Kids in current /Page-Dictionary'); |
---|
375 | |
---|
376 | if ($_kids[1][0] == PDF_TYPE_ARRAY) { |
---|
377 | $kids = $_kids[1][1]; |
---|
378 | } else { |
---|
379 | $kids = $_kids[1]; |
---|
380 | } |
---|
381 | |
---|
382 | foreach ($kids as $v) { |
---|
383 | $pg = $this->pdf_resolve_object ($c, $v); |
---|
384 | if ($pg[1][1]['/Type'][1] === '/Pages') { |
---|
385 | // If one of the kids is an embedded |
---|
386 | // /Pages array, resolve it as well. |
---|
387 | $this->read_pages($c, $pg, $result); |
---|
388 | } else { |
---|
389 | $result[] = $pg; |
---|
390 | } |
---|
391 | } |
---|
392 | } |
---|
393 | |
---|
394 | |
---|
395 | |
---|
396 | /** |
---|
397 | * Get PDF-Version |
---|
398 | * |
---|
399 | * And reset the PDF Version used in FPDI if needed |
---|
400 | */ |
---|
401 | function getPDFVersion() { |
---|
402 | parent::getPDFVersion(); |
---|
403 | $this->fpdi->setPDFVersion(max($this->fpdi->getPDFVersion(), $this->pdfVersion)); |
---|
404 | } |
---|
405 | |
---|
406 | } |
---|