]>
Commit | Line | Data |
---|---|---|
1 | <?php | |
2 | // | |
3 | // FPDI - Version 1.2 | |
4 | // | |
5 | // Copyright 2004-2007 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 | ||
21 | class fpdi_pdf_parser extends pdf_parser { | |
22 | ||
23 | /** | |
24 | * Pages | |
25 | * Index beginns at 0 | |
26 | * | |
27 | * @var array | |
28 | */ | |
29 | var $pages; | |
30 | ||
31 | /** | |
32 | * Page count | |
33 | * @var integer | |
34 | */ | |
35 | var $page_count; | |
36 | ||
37 | /** | |
38 | * actual page number | |
39 | * @var integer | |
40 | */ | |
41 | var $pageno; | |
42 | ||
43 | ||
44 | /** | |
45 | * FPDI Reference | |
46 | * @var object | |
47 | */ | |
48 | var $fpdi; | |
49 | ||
50 | /** | |
51 | * Available BoxTypes | |
52 | * | |
53 | * @var array | |
54 | */ | |
55 | var $availableBoxes = array("/MediaBox","/CropBox","/BleedBox","/TrimBox","/ArtBox"); | |
56 | ||
57 | /** | |
58 | * Constructor | |
59 | * | |
60 | * @param string $filename Source-Filename | |
61 | * @param object $fpdi Object of type fpdi | |
62 | */ | |
63 | function fpdi_pdf_parser($filename,&$fpdi) { | |
64 | $this->fpdi =& $fpdi; | |
65 | $this->filename = $filename; | |
66 | ||
67 | parent::pdf_parser($filename); | |
68 | if ($this->success == false) { return false; } | |
69 | ||
70 | // resolve Pages-Dictonary | |
71 | $pages = $this->pdf_resolve_object($this->c, $this->root[1][1]['/Pages']); | |
72 | if ($this->success == false) { return false; } | |
73 | ||
74 | // Read pages | |
75 | $this->read_pages($this->c, $pages, $this->pages); | |
76 | if ($this->success == false) { return false; } | |
77 | ||
78 | // count pages; | |
79 | $this->page_count = count($this->pages); | |
80 | } | |
81 | ||
82 | ||
83 | /** | |
84 | * Get pagecount from sourcefile | |
85 | * | |
86 | * @return int | |
87 | */ | |
88 | function getPageCount() { | |
89 | return $this->page_count; | |
90 | } | |
91 | ||
92 | ||
93 | /** | |
94 | * Set pageno | |
95 | * | |
96 | * @param int $pageno Pagenumber to use | |
97 | */ | |
98 | function setPageno($pageno) { | |
99 | $pageno = ((int) $pageno) - 1; | |
100 | ||
101 | if ($pageno < 0 || $pageno >= $this->getPageCount()) { | |
102 | $this->fpdi->error("Pagenumber is wrong!"); | |
103 | } | |
104 | ||
105 | $this->pageno = $pageno; | |
106 | } | |
107 | ||
108 | /** | |
109 | * Get page-resources from current page | |
110 | * | |
111 | * @return array | |
112 | */ | |
113 | function getPageResources() { | |
114 | return $this->_getPageResources($this->pages[$this->pageno]); | |
115 | } | |
116 | ||
117 | /** | |
118 | * Get page-resources from /Page | |
119 | * | |
120 | * @param array $obj Array of pdf-data | |
121 | */ | |
122 | function _getPageResources ($obj) { // $obj = /Page | |
123 | $obj = $this->pdf_resolve_object($this->c, $obj); | |
124 | ||
125 | // If the current object has a resources | |
126 | // dictionary associated with it, we use | |
127 | // it. Otherwise, we move back to its | |
128 | // parent object. | |
129 | if (isset ($obj[1][1]['/Resources'])) { | |
130 | $res = $this->pdf_resolve_object($this->c, $obj[1][1]['/Resources']); | |
131 | if ($res[0] == PDF_TYPE_OBJECT) | |
132 | return $res[1]; | |
133 | return $res; | |
134 | } else { | |
135 | if (!isset ($obj[1][1]['/Parent'])) { | |
136 | return false; | |
137 | } else { | |
138 | $res = $this->_getPageResources($obj[1][1]['/Parent']); | |
139 | if ($res[0] == PDF_TYPE_OBJECT) | |
140 | return $res[1]; | |
141 | return $res; | |
142 | } | |
143 | } | |
144 | } | |
145 | ||
146 | ||
147 | /** | |
148 | * Get content of current page | |
149 | * | |
150 | * If more /Contents is an array, the streams are concated | |
151 | * | |
152 | * @return string | |
153 | */ | |
154 | function getContent() { | |
155 | $buffer = ""; | |
156 | ||
157 | if (isset($this->pages[$this->pageno][1][1]['/Contents'])) { | |
158 | $contents = $this->_getPageContent($this->pages[$this->pageno][1][1]['/Contents']); | |
159 | foreach($contents AS $tmp_content) { | |
160 | $buffer .= $this->_rebuildContentStream($tmp_content).' '; | |
161 | } | |
162 | } | |
163 | ||
164 | return $buffer; | |
165 | } | |
166 | ||
167 | ||
168 | /** | |
169 | * Resolve all content-objects | |
170 | * | |
171 | * @param array $content_ref | |
172 | * @return array | |
173 | */ | |
174 | function _getPageContent($content_ref) { | |
175 | $contents = array(); | |
176 | ||
177 | if ($content_ref[0] == PDF_TYPE_OBJREF) { | |
178 | $content = $this->pdf_resolve_object($this->c, $content_ref); | |
179 | if ($content[1][0] == PDF_TYPE_ARRAY) { | |
180 | $contents = $this->_getPageContent($content[1]); | |
181 | } else { | |
182 | $contents[] = $content; | |
183 | } | |
184 | } else if ($content_ref[0] == PDF_TYPE_ARRAY) { | |
185 | foreach ($content_ref[1] AS $tmp_content_ref) { | |
186 | $contents = array_merge($contents,$this->_getPageContent($tmp_content_ref)); | |
187 | } | |
188 | } | |
189 | ||
190 | return $contents; | |
191 | } | |
192 | ||
193 | ||
194 | /** | |
195 | * Rebuild content-streams | |
196 | * | |
197 | * @param array $obj | |
198 | * @return string | |
199 | */ | |
200 | function _rebuildContentStream($obj) { | |
201 | $filters = array(); | |
202 | ||
203 | if (isset($obj[1][1]['/Filter'])) { | |
204 | $_filter = $obj[1][1]['/Filter']; | |
205 | ||
206 | if ($_filter[0] == PDF_TYPE_TOKEN) { | |
207 | $filters[] = $_filter; | |
208 | } else if ($_filter[0] == PDF_TYPE_ARRAY) { | |
209 | $filters = $_filter[1]; | |
210 | } | |
211 | } | |
212 | ||
213 | $stream = $obj[2][1]; | |
214 | ||
215 | foreach ($filters AS $_filter) { | |
216 | switch ($_filter[1]) { | |
217 | case "/FlateDecode": | |
218 | if (function_exists('gzuncompress')) { | |
219 | $stream = (strlen($stream) > 0) ? @gzuncompress($stream) : ''; | |
220 | } else { | |
221 | $this->fpdi->error(sprintf("To handle %s filter, please compile php with zlib support.",$_filter[1])); | |
222 | } | |
223 | if ($stream === false) { | |
224 | $this->fpdi->error("Error while decompressing stream."); | |
225 | } | |
226 | break; | |
227 | // mPDF 4.2.003 | |
228 | case '/LZWDecode': | |
229 | include_once(_MPDF_PATH.'mpdfi/filters/FilterLZW.php'); | |
230 | // mPDF 5.0 Removed pass by reference =& | |
231 | $decoder = new FilterLZW(); | |
232 | $stream = $decoder->decode($stream); | |
233 | break; | |
234 | case '/ASCII85Decode': | |
235 | include_once(_MPDF_PATH.'mpdfi/filters/FilterASCII85.php'); | |
236 | // mPDF 5.0 Removed pass by reference =& | |
237 | $decoder = new FilterASCII85(); | |
238 | $stream = $decoder->decode($stream); | |
239 | break; | |
240 | case null: | |
241 | $stream = $stream; | |
242 | break; | |
243 | default: | |
244 | $this->fpdi->error(sprintf("Unsupported Filter: %s",$_filter[1])); | |
245 | } | |
246 | } | |
247 | ||
248 | return $stream; | |
249 | } | |
250 | ||
251 | ||
252 | /** | |
253 | * Get a Box from a page | |
254 | * Arrayformat is same as used by fpdf_tpl | |
255 | * | |
256 | * @param array $page a /Page | |
257 | * @param string $box_index Type of Box @see $availableBoxes | |
258 | * @return array | |
259 | */ | |
260 | function getPageBox($page, $box_index) { | |
261 | $page = $this->pdf_resolve_object($this->c,$page); | |
262 | $box = null; | |
263 | if (isset($page[1][1][$box_index])) | |
264 | $box =& $page[1][1][$box_index]; | |
265 | ||
266 | if (!is_null($box) && $box[0] == PDF_TYPE_OBJREF) { | |
267 | $tmp_box = $this->pdf_resolve_object($this->c,$box); | |
268 | $box = $tmp_box[1]; | |
269 | } | |
270 | ||
271 | if (!is_null($box) && $box[0] == PDF_TYPE_ARRAY) { | |
272 | $b =& $box[1]; | |
273 | return array("x" => $b[0][1]/_MPDFK, | |
274 | "y" => $b[1][1]/_MPDFK, | |
275 | "w" => abs($b[0][1]-$b[2][1])/_MPDFK, | |
276 | "h" => abs($b[1][1]-$b[3][1])/_MPDFK); // mPDF 5.3.90 | |
277 | } else if (!isset ($page[1][1]['/Parent'])) { | |
278 | return false; | |
279 | } else { | |
280 | return $this->getPageBox($this->pdf_resolve_object($this->c, $page[1][1]['/Parent']), $box_index); | |
281 | } | |
282 | } | |
283 | ||
284 | function getPageBoxes($pageno) { | |
285 | return $this->_getPageBoxes($this->pages[$pageno-1]); | |
286 | } | |
287 | ||
288 | /** | |
289 | * Get all Boxes from /Page | |
290 | * | |
291 | * @param array a /Page | |
292 | * @return array | |
293 | */ | |
294 | function _getPageBoxes($page) { | |
295 | $boxes = array(); | |
296 | ||
297 | foreach($this->availableBoxes AS $box) { | |
298 | if ($_box = $this->getPageBox($page,$box)) { | |
299 | $boxes[$box] = $_box; | |
300 | } | |
301 | } | |
302 | ||
303 | return $boxes; | |
304 | } | |
305 | ||
306 | function getPageRotation($pageno) { | |
307 | return $this->_getPageRotation($this->pages[$pageno-1]); | |
308 | } | |
309 | ||
310 | function _getPageRotation ($obj) { // $obj = /Page | |
311 | $obj = $this->pdf_resolve_object($this->c, $obj); | |
312 | if (isset ($obj[1][1]['/Rotate'])) { | |
313 | $res = $this->pdf_resolve_object($this->c, $obj[1][1]['/Rotate']); | |
314 | if ($res[0] == PDF_TYPE_OBJECT) | |
315 | return $res[1]; | |
316 | return $res; | |
317 | } else { | |
318 | if (!isset ($obj[1][1]['/Parent'])) { | |
319 | return false; | |
320 | } else { | |
321 | $res = $this->_getPageRotation($obj[1][1]['/Parent']); | |
322 | if ($res[0] == PDF_TYPE_OBJECT) | |
323 | return $res[1]; | |
324 | return $res; | |
325 | } | |
326 | } | |
327 | } | |
328 | ||
329 | /** | |
330 | * Read all /Page(es) | |
331 | * | |
332 | * @param object pdf_context | |
333 | * @param array /Pages | |
334 | * @param array the result-array | |
335 | */ | |
336 | function read_pages (&$c, &$pages, &$result) { | |
337 | // Get the kids dictionary | |
338 | $kids = $this->pdf_resolve_object ($c, $pages[1][1]['/Kids']); | |
339 | ||
340 | if (!is_array($kids)) { | |
341 | // mPDF 4.0 | |
342 | $this->success = false; | |
343 | $this->errormsg = sprintf("Cannot find /Kids in current /Page-Dictionary"); | |
344 | return false; | |
345 | } | |
346 | foreach ($kids[1] as $v) { | |
347 | $pg = $this->pdf_resolve_object ($c, $v); | |
348 | if ($pg[1][1]['/Type'][1] === '/Pages') { | |
349 | // If one of the kids is an embedded | |
350 | // /Pages array, resolve it as well. | |
351 | $this->read_pages ($c, $pg, $result); | |
352 | } else { | |
353 | $result[] = $pg; | |
354 | } | |
355 | } | |
356 | } | |
357 | ||
358 | ||
359 | ||
360 | ||
361 | } | |
362 | ||
363 | ?> |