43341922 |
1 | /* |
2 | * PDF backend for Halibut |
3 | */ |
4 | |
5 | #include <assert.h> |
6 | #include "halibut.h" |
7 | #include "paper.h" |
8 | |
9 | #define TREE_BRANCH 2 /* max branching factor in page tree */ |
10 | |
11 | paragraph *pdf_config_filename(char *filename) |
12 | { |
e4ea58f8 |
13 | return cmdline_cfg_simple("pdf-filename", filename, NULL); |
43341922 |
14 | } |
15 | |
16 | typedef struct object_Tag object; |
17 | typedef struct objlist_Tag objlist; |
18 | |
19 | struct object_Tag { |
20 | objlist *list; |
21 | object *next; |
22 | int number; |
23 | rdstringc main, stream; |
24 | int size, fileoff; |
25 | char *final; |
26 | }; |
27 | |
28 | struct objlist_Tag { |
29 | int number; |
30 | object *head, *tail; |
31 | }; |
32 | |
33 | static object *new_object(objlist *list); |
34 | static void objtext(object *o, char const *text); |
35 | static void objstream(object *o, char const *text); |
7c8c4239 |
36 | static void pdf_string(void (*add)(object *, char const *), |
37 | object *, char const *); |
ce1b04aa |
38 | static void pdf_string_len(void (*add)(object *, char const *), |
39 | object *, char const *, int); |
43341922 |
40 | static void objref(object *o, object *dest); |
993ade7a |
41 | static char *pdf_outline_convert(wchar_t *s, int *len); |
43341922 |
42 | |
3600720b |
43 | static int is_std_font(char const *name); |
44 | |
43341922 |
45 | static void make_pages_node(object *node, object *parent, page_data *first, |
2ff8419e |
46 | page_data *last, object *resources, |
47 | object *mediabox); |
f0e51ce1 |
48 | static int make_outline(object *parent, outline_element *start, int n, |
49 | int open); |
09358aa7 |
50 | static int pdf_versionid(FILE *fp, word *words); |
43341922 |
51 | |
52 | void pdf_backend(paragraph *sourceform, keywordlist *keywords, |
53 | indexdata *idx, void *vdoc) { |
54 | document *doc = (document *)vdoc; |
55 | int font_index; |
56 | font_encoding *fe; |
57 | page_data *page; |
58 | int pageno; |
59 | FILE *fp; |
60 | char *filename; |
61 | paragraph *p; |
62 | objlist olist; |
993ade7a |
63 | object *o, *info, *cat, *outlines, *pages, *resources, *mediabox; |
43341922 |
64 | int fileoff; |
65 | |
66 | IGNORE(keywords); |
67 | IGNORE(idx); |
68 | |
69 | filename = dupstr("output.pdf"); |
70 | for (p = sourceform; p; p = p->next) { |
7738a2fb |
71 | if (p->type == para_Config) { |
43341922 |
72 | if (!ustricmp(p->keyword, L"pdf-filename")) { |
73 | sfree(filename); |
e4ea58f8 |
74 | filename = dupstr(adv(p->origkeyword)); |
43341922 |
75 | } |
76 | } |
77 | } |
78 | |
79 | olist.head = olist.tail = NULL; |
80 | olist.number = 1; |
81 | |
993ade7a |
82 | { |
83 | char buf[256]; |
84 | |
85 | info = new_object(&olist); |
86 | objtext(info, "<<\n"); |
87 | if (doc->n_outline_elements > 0) { |
88 | char *title; |
89 | int titlelen; |
90 | |
91 | title = |
92 | pdf_outline_convert(doc->outline_elements->pdata->outline_title, |
93 | &titlelen); |
94 | objtext(info, "/Title "); |
95 | pdf_string_len(objtext, info, title, titlelen); |
96 | sfree(title); |
97 | objtext(info, "\n"); |
98 | } |
99 | objtext(info, "/Producer "); |
100 | sprintf(buf, "Halibut, %s", version); |
101 | pdf_string(objtext, info, buf); |
102 | objtext(info, "\n>>\n"); |
103 | } |
104 | |
43341922 |
105 | cat = new_object(&olist); |
a615a778 |
106 | if (doc->n_outline_elements > 0) |
107 | outlines = new_object(&olist); |
108 | else |
109 | outlines = NULL; |
43341922 |
110 | pages = new_object(&olist); |
111 | resources = new_object(&olist); |
112 | |
113 | /* |
43341922 |
114 | * The catalogue just contains references to the outlines and |
115 | * pages objects. |
116 | */ |
a615a778 |
117 | objtext(cat, "<<\n/Type /Catalog"); |
118 | if (outlines) { |
119 | objtext(cat, "\n/Outlines "); |
120 | objref(cat, outlines); |
121 | } |
43341922 |
122 | objtext(cat, "\n/Pages "); |
123 | objref(cat, pages); |
a615a778 |
124 | if (outlines) |
125 | objtext(cat, "\n/PageMode /UseOutlines"); |
126 | objtext(cat, "\n>>\n"); |
43341922 |
127 | |
128 | /* |
129 | * Set up the resources dictionary, which mostly means |
130 | * providing all the font objects and names to call them by. |
131 | */ |
132 | font_index = 0; |
133 | objtext(resources, "<<\n/Font <<\n"); |
134 | for (fe = doc->fonts->head; fe; fe = fe->next) { |
135 | char fname[40]; |
136 | int i; |
137 | object *font; |
138 | |
139 | sprintf(fname, "f%d", font_index++); |
140 | fe->name = dupstr(fname); |
141 | |
142 | font = new_object(&olist); |
143 | |
144 | objtext(resources, "/"); |
145 | objtext(resources, fe->name); |
146 | objtext(resources, " "); |
147 | objref(resources, font); |
148 | objtext(resources, "\n"); |
149 | |
150 | objtext(font, "<<\n/Type /Font\n/Subtype /Type1\n/Name /"); |
151 | objtext(font, fe->name); |
152 | objtext(font, "\n/BaseFont /"); |
ba0fe3ec |
153 | objtext(font, fe->font->info->name); |
43341922 |
154 | objtext(font, "\n/Encoding <<\n/Type /Encoding\n/Differences ["); |
155 | |
156 | for (i = 0; i < 256; i++) { |
157 | char buf[20]; |
158 | if (!fe->vector[i]) |
159 | continue; |
160 | sprintf(buf, "\n%d /", i); |
161 | objtext(font, buf); |
162 | objtext(font, fe->vector[i] ? fe->vector[i] : ".notdef"); |
163 | } |
164 | |
165 | objtext(font, "\n]\n>>\n"); |
166 | |
3600720b |
167 | if (!is_std_font(fe->font->info->name)){ |
43341922 |
168 | object *widths = new_object(&olist); |
9fb062fc |
169 | int firstchar = -1, lastchar = -1; |
170 | char buf[80]; |
171 | for (i = 0; i < 256; i++) |
172 | if (fe->indices[i] >= 0) { |
173 | if (firstchar < 0) firstchar = i; |
174 | lastchar = i; |
175 | } |
176 | sprintf(buf, "/FirstChar %d\n/LastChar %d\n/Widths ", |
177 | firstchar, lastchar); |
178 | objtext(font, buf); |
43341922 |
179 | objref(font, widths); |
180 | objtext(font, "\n"); |
181 | objtext(widths, "[\n"); |
9fb062fc |
182 | for (i = firstchar; i <= lastchar; i++) { |
43341922 |
183 | char buf[80]; |
184 | double width; |
185 | if (fe->indices[i] < 0) |
186 | width = 0.0; |
187 | else |
ba0fe3ec |
188 | width = fe->font->info->widths[fe->indices[i]]; |
17c71b41 |
189 | sprintf(buf, "%g\n", 1000.0 * width / FUNITS_PER_PT); |
43341922 |
190 | objtext(widths, buf); |
191 | } |
192 | objtext(widths, "]\n"); |
193 | } |
194 | |
195 | objtext(font, ">>\n"); |
196 | } |
197 | objtext(resources, ">>\n>>\n"); |
198 | |
2ff8419e |
199 | { |
200 | char buf[255]; |
201 | mediabox = new_object(&olist); |
202 | sprintf(buf, "[0 0 %g %g]\n", |
203 | doc->paper_width / FUNITS_PER_PT, |
204 | doc->paper_height / FUNITS_PER_PT); |
205 | objtext(mediabox, buf); |
206 | } |
207 | |
43341922 |
208 | /* |
209 | * Define the page objects for each page, and get each one |
210 | * ready to have a `Parent' specification added to it. |
211 | */ |
212 | for (page = doc->pages; page; page = page->next) { |
213 | object *opage; |
214 | |
215 | opage = new_object(&olist); |
216 | page->spare = opage; |
217 | objtext(opage, "<<\n/Type /Page\n"); |
218 | } |
219 | |
220 | /* |
221 | * Recursively build the page tree. |
222 | */ |
2ff8419e |
223 | make_pages_node(pages, NULL, doc->pages, NULL, resources, mediabox); |
43341922 |
224 | |
225 | /* |
226 | * Create and render the individual pages. |
227 | */ |
228 | pageno = 0; |
229 | for (page = doc->pages; page; page = page->next) { |
230 | object *opage, *cstr; |
23765aeb |
231 | rect *r; |
7c8c4239 |
232 | text_fragment *frag, *frag_end; |
43341922 |
233 | char buf[256]; |
7c8c4239 |
234 | int x, y, lx, ly; |
43341922 |
235 | |
236 | opage = (object *)page->spare; |
237 | /* |
238 | * At this point the page dictionary is already |
239 | * half-written, with /Type and /Parent already present. We |
240 | * continue from there. |
241 | */ |
242 | |
243 | /* |
244 | * The PDF spec says /Resources is required, but also says |
245 | * that it's inheritable and may be omitted if it's present |
246 | * in a Pages node. In our case it is: it's present in the |
247 | * topmost /Pages node because we carefully put it there. |
2ff8419e |
248 | * So we don't need a /Resources entry here. The same applies |
249 | * to /MediaBox. |
43341922 |
250 | */ |
43341922 |
251 | |
252 | /* |
253 | * Now we're ready to define a content stream containing |
254 | * the actual text on the page. |
255 | */ |
256 | cstr = new_object(&olist); |
257 | objtext(opage, "/Contents "); |
258 | objref(opage, cstr); |
259 | objtext(opage, "\n"); |
260 | |
23765aeb |
261 | /* |
262 | * Render any rectangles on the page. |
263 | */ |
264 | for (r = page->first_rect; r; r = r->next) { |
265 | char buf[512]; |
17c71b41 |
266 | sprintf(buf, "%g %g %g %g re f\n", |
267 | r->x / FUNITS_PER_PT, r->y / FUNITS_PER_PT, |
268 | r->w / FUNITS_PER_PT, r->h / FUNITS_PER_PT); |
23765aeb |
269 | objstream(cstr, buf); |
270 | } |
271 | |
43341922 |
272 | objstream(cstr, "BT\n"); |
43341922 |
273 | |
7c8c4239 |
274 | /* |
275 | * PDF tracks two separate current positions: the position |
276 | * given in the `line matrix' and the position given in the |
277 | * `text matrix'. We must therefore track both as well. |
278 | * They start off at -1 (unset). |
279 | */ |
280 | lx = ly = -1; |
281 | x = y = -1; |
282 | |
283 | frag = page->first_text; |
284 | while (frag) { |
285 | /* |
286 | * For compactness, I'm going to group text fragments |
287 | * into subsequences that use the same font+size. So |
288 | * first find the end of this subsequence. |
289 | */ |
290 | for (frag_end = frag; |
291 | (frag_end && |
292 | frag_end->fe == frag->fe && |
293 | frag_end->fontsize == frag->fontsize); |
294 | frag_end = frag_end->next); |
295 | |
296 | /* |
297 | * Now select the text fragment, and prepare to display |
298 | * the text. |
299 | */ |
43341922 |
300 | objstream(cstr, "/"); |
301 | objstream(cstr, frag->fe->name); |
7c8c4239 |
302 | sprintf(buf, " %d Tf ", frag->fontsize); |
43341922 |
303 | objstream(cstr, buf); |
304 | |
7c8c4239 |
305 | while (frag && frag != frag_end) { |
306 | /* |
307 | * Place the text position for the first piece of |
308 | * text. |
309 | */ |
310 | if (lx < 0) { |
311 | sprintf(buf, "1 0 0 1 %g %g Tm ", |
17c71b41 |
312 | frag->x/FUNITS_PER_PT, frag->y/FUNITS_PER_PT); |
7c8c4239 |
313 | } else { |
314 | sprintf(buf, "%g %g Td ", |
17c71b41 |
315 | (frag->x - lx)/FUNITS_PER_PT, |
316 | (frag->y - ly)/FUNITS_PER_PT); |
7c8c4239 |
317 | } |
43341922 |
318 | objstream(cstr, buf); |
7c8c4239 |
319 | lx = x = frag->x; |
320 | ly = y = frag->y; |
321 | |
322 | /* |
323 | * See if we're going to use Tj (show a single |
324 | * string) or TJ (show an array of strings with |
325 | * x-spacings between them). We determine this by |
326 | * seeing if there's more than one text fragment in |
327 | * sequence with the same y-coordinate. |
328 | */ |
329 | if (frag->next && frag->next != frag_end && |
330 | frag->next->y == y) { |
331 | /* |
332 | * The TJ strategy. |
333 | */ |
334 | objstream(cstr, "["); |
335 | while (frag && frag != frag_end && frag->y == y) { |
336 | if (frag->x != x) { |
337 | sprintf(buf, "%g", |
338 | (x - frag->x) * 1000.0 / |
17c71b41 |
339 | (FUNITS_PER_PT * frag->fontsize)); |
7c8c4239 |
340 | objstream(cstr, buf); |
341 | } |
342 | pdf_string(objstream, cstr, frag->text); |
343 | x = frag->x + frag->width; |
344 | frag = frag->next; |
345 | } |
346 | objstream(cstr, "]TJ\n"); |
347 | } else |
348 | { |
349 | /* |
350 | * The Tj strategy. |
351 | */ |
352 | pdf_string(objstream, cstr, frag->text); |
353 | objstream(cstr, "Tj\n"); |
354 | frag = frag->next; |
355 | } |
43341922 |
356 | } |
43341922 |
357 | } |
358 | objstream(cstr, "ET"); |
359 | |
138d7ffb |
360 | /* |
361 | * Also, we want an annotation dictionary containing the |
362 | * cross-references from this page. |
363 | */ |
364 | if (page->first_xref) { |
365 | xref *xr; |
366 | objtext(opage, "/Annots [\n"); |
367 | |
368 | for (xr = page->first_xref; xr; xr = xr->next) { |
369 | object *annot; |
370 | char buf[256]; |
371 | |
372 | annot = new_object(&olist); |
373 | objref(opage, annot); |
374 | objtext(opage, "\n"); |
375 | |
376 | objtext(annot, "<<\n/Type /Annot\n/Subtype /Link\n/Rect ["); |
377 | sprintf(buf, "%g %g %g %g", |
17c71b41 |
378 | xr->lx / FUNITS_PER_PT, xr->by / FUNITS_PER_PT, |
379 | xr->rx / FUNITS_PER_PT, xr->ty / FUNITS_PER_PT); |
138d7ffb |
380 | objtext(annot, buf); |
381 | objtext(annot, "]\n/Border [0 0 0]\n"); |
382 | |
383 | if (xr->dest.type == PAGE) { |
384 | objtext(annot, "/Dest ["); |
385 | objref(annot, (object *)xr->dest.page->spare); |
386 | objtext(annot, " /XYZ null null null]\n"); |
387 | } else { |
7c8c4239 |
388 | objtext(annot, "/A <<\n/Type /Action\n/S /URI\n/URI "); |
389 | pdf_string(objtext, annot, xr->dest.url); |
390 | objtext(annot, "\n>>\n"); |
138d7ffb |
391 | } |
392 | |
393 | objtext(annot, ">>\n"); |
394 | } |
395 | |
396 | objtext(opage, "]\n"); |
397 | } |
398 | |
43341922 |
399 | objtext(opage, ">>\n"); |
400 | } |
401 | |
402 | /* |
f0e51ce1 |
403 | * Set up the outlines dictionary. |
404 | */ |
a615a778 |
405 | if (outlines) { |
f0e51ce1 |
406 | int topcount; |
407 | char buf[80]; |
408 | |
409 | objtext(outlines, "<<\n/Type /Outlines\n"); |
410 | topcount = make_outline(outlines, doc->outline_elements, |
411 | doc->n_outline_elements, TRUE); |
412 | sprintf(buf, "/Count %d\n>>\n", topcount); |
413 | objtext(outlines, buf); |
414 | } |
415 | |
416 | /* |
43341922 |
417 | * Assemble the final linear form of every object. |
418 | */ |
419 | for (o = olist.head; o; o = o->next) { |
420 | rdstringc rs = {0, 0, NULL}; |
421 | char text[80]; |
422 | |
423 | sprintf(text, "%d 0 obj\n", o->number); |
424 | rdaddsc(&rs, text); |
425 | |
426 | if (!o->main.text && o->stream.text) { |
427 | sprintf(text, "<<\n/Length %d\n>>\n", o->stream.pos); |
428 | rdaddsc(&o->main, text); |
429 | } |
430 | |
431 | assert(o->main.text); |
432 | rdaddsc(&rs, o->main.text); |
433 | sfree(o->main.text); |
434 | |
435 | if (rs.text[rs.pos-1] != '\n') |
436 | rdaddc(&rs, '\n'); |
437 | |
438 | if (o->stream.text) { |
439 | /* |
440 | * FIXME: If we ever start compressing stream data then |
441 | * it will have zero bytes in it, so we'll have to be |
442 | * more careful than this. |
443 | */ |
444 | rdaddsc(&rs, "stream\n"); |
445 | rdaddsc(&rs, o->stream.text); |
446 | rdaddsc(&rs, "\nendstream\n"); |
447 | sfree(o->stream.text); |
448 | } |
449 | |
450 | rdaddsc(&rs, "endobj\n"); |
451 | |
452 | o->final = rs.text; |
453 | o->size = rs.pos; |
454 | } |
455 | |
456 | /* |
457 | * Write out the PDF file. |
458 | */ |
459 | |
460 | fp = fopen(filename, "wb"); |
461 | if (!fp) { |
462 | error(err_cantopenw, filename); |
463 | return; |
464 | } |
465 | |
466 | /* |
09358aa7 |
467 | * Header. I'm going to put the version IDs in the header as |
468 | * well, simply in PDF comments. |
43341922 |
469 | */ |
470 | fileoff = fprintf(fp, "%%PDF-1.3\n"); |
09358aa7 |
471 | for (p = sourceform; p; p = p->next) |
472 | if (p->type == para_VersionID) |
473 | fileoff += pdf_versionid(fp, p->words); |
43341922 |
474 | |
475 | /* |
476 | * Body |
477 | */ |
478 | for (o = olist.head; o; o = o->next) { |
479 | o->fileoff = fileoff; |
480 | fwrite(o->final, 1, o->size, fp); |
481 | fileoff += o->size; |
482 | } |
483 | |
484 | /* |
485 | * Cross-reference table |
486 | */ |
487 | fprintf(fp, "xref\n"); |
488 | assert(olist.head->number == 1); |
489 | fprintf(fp, "0 %d\n", olist.tail->number + 1); |
490 | fprintf(fp, "0000000000 65535 f \n"); |
491 | for (o = olist.head; o; o = o->next) { |
492 | char entry[40]; |
493 | sprintf(entry, "%010d 00000 n \n", o->fileoff); |
494 | assert(strlen(entry) == 20); |
495 | fputs(entry, fp); |
496 | } |
497 | |
498 | /* |
499 | * Trailer |
500 | */ |
993ade7a |
501 | fprintf(fp, "trailer\n<<\n/Size %d\n/Root %d 0 R\n/Info %d 0 R\n>>\n", |
502 | olist.tail->number + 1, cat->number, info->number); |
43341922 |
503 | fprintf(fp, "startxref\n%d\n%%%%EOF\n", fileoff); |
504 | |
505 | fclose(fp); |
506 | |
507 | sfree(filename); |
508 | } |
509 | |
510 | static object *new_object(objlist *list) |
511 | { |
f1530049 |
512 | object *obj = snew(object); |
43341922 |
513 | |
514 | obj->list = list; |
515 | |
516 | obj->main.text = NULL; |
517 | obj->main.pos = obj->main.size = 0; |
518 | obj->stream.text = NULL; |
519 | obj->stream.pos = obj->stream.size = 0; |
520 | |
521 | obj->number = list->number++; |
522 | |
523 | obj->next = NULL; |
524 | if (list->tail) |
525 | list->tail->next = obj; |
526 | else |
527 | list->head = obj; |
528 | list->tail = obj; |
529 | |
530 | obj->size = 0; |
531 | obj->final = NULL; |
532 | |
533 | return obj; |
534 | } |
535 | |
536 | static void objtext(object *o, char const *text) |
537 | { |
538 | rdaddsc(&o->main, text); |
539 | } |
540 | |
541 | static void objstream(object *o, char const *text) |
542 | { |
543 | rdaddsc(&o->stream, text); |
544 | } |
545 | |
546 | static void objref(object *o, object *dest) |
547 | { |
548 | char buf[40]; |
549 | sprintf(buf, "%d 0 R", dest->number); |
550 | rdaddsc(&o->main, buf); |
551 | } |
552 | |
3600720b |
553 | static char const * const stdfonts[] = { |
554 | "Times-Roman", "Times-Bold", "Times-Italic", "Times-BoldItalic", |
555 | "Helvetica", "Helvetica-Bold", "Helvetica-Oblique","Helvetica-BoldOblique", |
556 | "Courier", "Courier-Bold", "Courier-Oblique", "Courier-BoldOblique", |
557 | "Symbol", "ZapfDingbats" |
558 | }; |
559 | |
560 | static int is_std_font(char const *name) { |
561 | unsigned i; |
562 | for (i = 0; i < lenof(stdfonts); i++) |
563 | if (strcmp(name, stdfonts[i]) == 0) |
564 | return TRUE; |
565 | return FALSE; |
566 | } |
567 | |
43341922 |
568 | static void make_pages_node(object *node, object *parent, page_data *first, |
2ff8419e |
569 | page_data *last, object *resources, |
570 | object *mediabox) |
43341922 |
571 | { |
572 | int count; |
573 | page_data *page; |
574 | char buf[80]; |
575 | |
576 | objtext(node, "<<\n/Type /Pages\n"); |
577 | if (parent) { |
578 | objtext(node, "/Parent "); |
579 | objref(node, parent); |
580 | objtext(node, "\n"); |
581 | } |
582 | |
583 | /* |
584 | * Count the pages in this stretch, to see if there are few |
585 | * enough to reference directly. |
586 | */ |
587 | count = 0; |
588 | for (page = first; page; page = page->next) { |
589 | count++; |
590 | if (page == last) |
591 | break; |
592 | } |
593 | |
594 | sprintf(buf, "/Count %d\n/Kids [\n", count); |
595 | objtext(node, buf); |
596 | |
597 | if (count > TREE_BRANCH) { |
598 | int i; |
599 | page_data *thisfirst, *thislast; |
600 | |
601 | page = first; |
602 | |
603 | for (i = 0; i < TREE_BRANCH; i++) { |
604 | int number = (i+1) * count / TREE_BRANCH - i * count / TREE_BRANCH; |
605 | thisfirst = page; |
606 | while (number--) { |
607 | thislast = page; |
608 | page = page->next; |
609 | } |
610 | |
611 | if (thisfirst == thislast) { |
612 | objref(node, (object *)thisfirst->spare); |
613 | objtext((object *)thisfirst->spare, "/Parent "); |
614 | objref((object *)thisfirst->spare, node); |
615 | objtext((object *)thisfirst->spare, "\n"); |
616 | } else { |
617 | object *newnode = new_object(node->list); |
2ff8419e |
618 | make_pages_node(newnode, node, thisfirst, thislast, |
619 | NULL, NULL); |
43341922 |
620 | objref(node, newnode); |
621 | } |
622 | objtext(node, "\n"); |
623 | } |
624 | |
625 | assert(thislast == last || page == NULL); |
626 | |
627 | } else { |
628 | for (page = first; page; page = page->next) { |
629 | objref(node, (object *)page->spare); |
630 | objtext(node, "\n"); |
631 | objtext((object *)page->spare, "/Parent "); |
632 | objref((object *)page->spare, node); |
633 | objtext((object *)page->spare, "\n"); |
634 | if (page == last) |
635 | break; |
636 | } |
637 | } |
638 | |
639 | objtext(node, "]\n"); |
640 | |
641 | if (resources) { |
642 | objtext(node, "/Resources "); |
643 | objref(node, resources); |
644 | objtext(node, "\n"); |
645 | } |
2ff8419e |
646 | if (mediabox) { |
647 | objtext(node, "/MediaBox "); |
648 | objref(node, mediabox); |
649 | objtext(node, "\n"); |
650 | } |
43341922 |
651 | |
652 | objtext(node, ">>\n"); |
653 | } |
f0e51ce1 |
654 | |
655 | /* |
656 | * In text on the page, PDF uses the PostScript font model, which |
657 | * means that glyphs are identified by PS strings and hence font |
658 | * encoding can be managed independently of the supplied encoding |
659 | * of the font. However, in the document outline, the PDF spec |
ce1b04aa |
660 | * encodes in either PDFDocEncoding (a custom superset of |
661 | * ISO-8859-1) or UTF-16BE. |
f0e51ce1 |
662 | */ |
ce1b04aa |
663 | static char *pdf_outline_convert(wchar_t *s, int *len) { |
664 | char *ret; |
665 | |
666 | ret = utoa_careful_dup(s, CS_PDF); |
667 | |
668 | /* |
669 | * Very silly special case: if the returned string begins with |
670 | * FE FF, then the PDF reader will mistake it for a UTF-16BE |
671 | * string. So in this case we give up on PDFDocEncoding and |
672 | * encode it in UTF-16 straight away. |
673 | */ |
674 | if (ret && ret[0] == '\xFE' && ret[1] == '\xFF') { |
675 | sfree(ret); |
676 | ret = NULL; |
f0e51ce1 |
677 | } |
ce1b04aa |
678 | |
679 | if (!ret) { |
680 | ret = utoa_dup_len(s, CS_UTF16BE, len); |
681 | } else { |
682 | *len = strlen(ret); |
f0e51ce1 |
683 | } |
ce1b04aa |
684 | |
685 | return ret; |
f0e51ce1 |
686 | } |
687 | |
f0e51ce1 |
688 | static int make_outline(object *parent, outline_element *items, int n, |
689 | int open) |
690 | { |
691 | int level, totalcount = 0; |
692 | outline_element *itemp; |
693 | object *curr, *prev = NULL, *first = NULL, *last = NULL; |
f0e51ce1 |
694 | |
695 | assert(n > 0); |
696 | |
697 | level = items->level; |
698 | |
699 | while (n > 0) { |
7c8c4239 |
700 | char *title; |
ce1b04aa |
701 | int titlelen; |
f0e51ce1 |
702 | |
703 | /* |
704 | * Here we expect to be sitting on an item at the given |
705 | * level. So we start by constructing an outline entry for |
706 | * that item. |
707 | */ |
708 | assert(items->level == level); |
709 | |
ce1b04aa |
710 | title = pdf_outline_convert(items->pdata->outline_title, &titlelen); |
f0e51ce1 |
711 | |
712 | totalcount++; |
713 | curr = new_object(parent->list); |
714 | if (!first) first = curr; |
715 | last = curr; |
7c8c4239 |
716 | objtext(curr, "<<\n/Title "); |
ce1b04aa |
717 | pdf_string_len(objtext, curr, title, titlelen); |
993ade7a |
718 | sfree(title); |
7c8c4239 |
719 | objtext(curr, "\n/Parent "); |
f0e51ce1 |
720 | objref(curr, parent); |
721 | objtext(curr, "\n/Dest ["); |
be76d597 |
722 | objref(curr, (object *)items->pdata->first->page->spare); |
f0e51ce1 |
723 | objtext(curr, " /XYZ null null null]\n"); |
724 | if (prev) { |
725 | objtext(curr, "/Prev "); |
726 | objref(curr, prev); |
727 | objtext(curr, "\n"); |
728 | |
729 | objtext(prev, "/Next "); |
730 | objref(prev, curr); |
731 | objtext(prev, "\n>>\n"); |
732 | } |
733 | prev = curr; |
734 | |
735 | items++, n--; |
736 | for (itemp = items; itemp < items+n && itemp->level > level; |
737 | itemp++); |
738 | |
739 | if (itemp > items) { |
740 | char buf[80]; |
741 | int count = make_outline(curr, items, itemp - items, FALSE); |
742 | if (!open) |
743 | count = -count; |
744 | else |
745 | totalcount += count; |
746 | sprintf(buf, "/Count %d\n", count); |
747 | objtext(curr, buf); |
748 | } |
749 | |
750 | n -= itemp - items; |
751 | items = itemp; |
752 | } |
753 | objtext(prev, ">>\n"); |
754 | |
755 | assert(first && last); |
756 | objtext(parent, "/First "); |
757 | objref(parent, first); |
758 | objtext(parent, "\n/Last "); |
759 | objref(parent, last); |
760 | objtext(parent, "\n"); |
761 | |
762 | return totalcount; |
763 | } |
09358aa7 |
764 | |
765 | static int pdf_versionid(FILE *fp, word *words) |
766 | { |
767 | int ret; |
768 | |
769 | ret = fprintf(fp, "%% "); |
770 | |
771 | for (; words; words = words->next) { |
772 | char *text; |
773 | int type; |
774 | |
775 | switch (words->type) { |
776 | case word_HyperLink: |
777 | case word_HyperEnd: |
778 | case word_UpperXref: |
779 | case word_LowerXref: |
780 | case word_XrefEnd: |
781 | case word_IndexRef: |
782 | continue; |
783 | } |
784 | |
785 | type = removeattr(words->type); |
786 | |
787 | switch (type) { |
788 | case word_Normal: |
e4ea58f8 |
789 | text = utoa_dup(words->text, CS_ASCII); |
09358aa7 |
790 | break; |
791 | case word_WhiteSpace: |
792 | text = dupstr(" "); |
793 | break; |
794 | case word_Quote: |
795 | text = dupstr("'"); |
796 | break; |
797 | } |
798 | |
799 | fputs(text, fp); |
800 | ret += strlen(text); |
801 | sfree(text); |
802 | } |
803 | |
804 | ret += fprintf(fp, "\n"); |
805 | |
806 | return ret; |
807 | } |
7c8c4239 |
808 | |
ce1b04aa |
809 | static void pdf_string_len(void (*add)(object *, char const *), |
810 | object *o, char const *str, int len) |
7c8c4239 |
811 | { |
812 | char const *p; |
813 | |
814 | add(o, "("); |
ce1b04aa |
815 | for (p = str; len > 0; p++, len--) { |
816 | char c[10]; |
817 | if (*p < ' ' || *p > '~') { |
818 | sprintf(c, "\\%03o", 0xFF & (int)*p); |
819 | } else { |
820 | int n = 0; |
821 | if (*p == '\\' || *p == '(' || *p == ')') |
822 | c[n++] = '\\'; |
823 | c[n++] = *p; |
824 | c[n] = '\0'; |
825 | } |
7c8c4239 |
826 | add(o, c); |
827 | } |
828 | add(o, ")"); |
829 | } |
ce1b04aa |
830 | |
831 | static void pdf_string(void (*add)(object *, char const *), |
832 | object *o, char const *str) |
833 | { |
834 | pdf_string_len(add, o, str, strlen(str)); |
835 | } |