43341922 |
1 | /* |
2 | * Paper printing pre-backend for Halibut. |
3 | * |
4 | * This module does all the processing common to both PostScript |
5 | * and PDF output: selecting fonts, line wrapping and page breaking |
6 | * in accordance with font metrics, laying out the contents and |
7 | * index pages, generally doing all the page layout. After this, |
8 | * bk_ps.c and bk_pdf.c should only need to do linear translations |
9 | * into their literal output format. |
10 | */ |
11 | |
12 | /* |
13 | * To be done: |
14 | * |
43341922 |
15 | * - set up contents section now we know what sections begin on |
16 | * which pages |
17 | * |
43341922 |
18 | * - index |
19 | * |
a8070b46 |
20 | * - header/footer? Page numbers at least would be handy. Fully |
21 | * configurable footer can wait, though. |
22 | * |
43341922 |
23 | * That should bring us to the same level of functionality that |
24 | * original-Halibut had, and the same in PDF plus the obvious |
25 | * interactive navigation features. After that, in future work: |
26 | * |
27 | * - linearised PDF, perhaps? |
28 | * |
29 | * - I'm uncertain of whether I need to include a ToUnicode CMap |
30 | * in each of my font definitions in PDF. Currently things (by |
31 | * which I mean cut and paste out of acroread) seem to be |
32 | * working fairly happily without it, but I don't know. |
33 | * |
34 | * - configurability |
35 | * |
36 | * - title pages |
37 | */ |
38 | |
39 | #include <assert.h> |
40 | #include <stdio.h> |
41 | |
42 | #include "halibut.h" |
43 | #include "paper.h" |
44 | |
be76d597 |
45 | typedef struct paper_conf_Tag paper_conf; |
46 | |
47 | struct paper_conf_Tag { |
48 | int paper_width; |
49 | int paper_height; |
50 | int left_margin; |
51 | int top_margin; |
52 | int right_margin; |
53 | int bottom_margin; |
54 | int indent_list_bullet; |
55 | int indent_list; |
56 | int indent_quote; |
57 | int base_leading; |
58 | int base_para_spacing; |
59 | int chapter_top_space; |
60 | int sect_num_left_space; |
61 | int chapter_underline_depth; |
62 | int chapter_underline_thickness; |
63 | int rule_thickness; |
64 | int base_font_size; |
2bfd1b76 |
65 | int contents_indent_step; |
66 | int contents_margin; |
67 | int leader_separation; |
be76d597 |
68 | /* These are derived from the above */ |
69 | int base_width; |
70 | int page_height; |
71 | /* Fonts used in the configuration */ |
72 | font_data *tr, *ti, *hr, *hi, *cr, *co, *cb; |
73 | }; |
74 | |
43341922 |
75 | static font_data *make_std_font(font_list *fontlist, char const *name); |
76 | static void wrap_paragraph(para_data *pdata, word *words, |
77 | int w, int i1, int i2); |
78 | static page_data *page_breaks(line_data *first, line_data *last, |
79 | int page_height); |
2bfd1b76 |
80 | static int render_string(page_data *page, font_data *font, int fontsize, |
81 | int x, int y, wchar_t *str); |
82 | static int render_line(line_data *ldata, int left_x, int top_y, |
83 | xref_dest *dest, keywordlist *keywords); |
515d216b |
84 | static int paper_width_simple(para_data *pdata, word *text); |
be76d597 |
85 | static para_data *code_paragraph(int indent, word *words, paper_conf *conf); |
86 | static para_data *rule_paragraph(int indent, paper_conf *conf); |
23765aeb |
87 | static void add_rect_to_page(page_data *page, int x, int y, int w, int h); |
2bfd1b76 |
88 | static para_data *make_para_data(int ptype, int paux, int indent, int rmargin, |
be76d597 |
89 | word *pkwtext, word *pkwtext2, word *pwords, |
90 | paper_conf *conf); |
91 | static void standard_line_spacing(para_data *pdata, paper_conf *conf); |
92 | static wchar_t *prepare_outline_title(word *first, wchar_t *separator, |
93 | word *second); |
2bfd1b76 |
94 | static word *fake_word(wchar_t *text); |
95 | static word *prepare_contents_title(word *first, wchar_t *separator, |
96 | word *second); |
43341922 |
97 | |
98 | void *paper_pre_backend(paragraph *sourceform, keywordlist *keywords, |
99 | indexdata *idx) { |
100 | paragraph *p; |
101 | document *doc; |
2bfd1b76 |
102 | int indent, used_contents; |
be76d597 |
103 | para_data *pdata, *firstpara = NULL, *lastpara = NULL; |
2bfd1b76 |
104 | para_data *firstcont, *lastcont; |
105 | line_data *ldata, *firstline, *lastline, *firstcontline, *lastcontline; |
43341922 |
106 | page_data *pages; |
107 | font_list *fontlist; |
be76d597 |
108 | paper_conf *conf; |
43341922 |
109 | |
110 | /* |
111 | * FIXME: All these things ought to become configurable. |
112 | */ |
be76d597 |
113 | conf = mknew(paper_conf); |
114 | conf->paper_width = 595 * 4096; |
115 | conf->paper_height = 841 * 4096; |
116 | conf->left_margin = 72 * 4096; |
117 | conf->top_margin = 72 * 4096; |
118 | conf->right_margin = 72 * 4096; |
119 | conf->bottom_margin = 108 * 4096; |
120 | conf->indent_list_bullet = 6 * 4096; |
121 | conf->indent_list = 24 * 4096; |
122 | conf->indent_quote = 18 * 4096; |
123 | conf->base_leading = 4096; |
124 | conf->base_para_spacing = 10 * 4096; |
125 | conf->chapter_top_space = 72 * 4096; |
126 | conf->sect_num_left_space = 12 * 4096; |
127 | conf->chapter_underline_depth = 14 * 4096; |
128 | conf->chapter_underline_thickness = 3 * 4096; |
129 | conf->rule_thickness = 1 * 4096; |
130 | conf->base_font_size = 12; |
2bfd1b76 |
131 | conf->contents_indent_step = 24 * 4096; |
132 | conf->contents_margin = 84 * 4096; |
133 | conf->leader_separation = 12 * 4096; |
be76d597 |
134 | |
135 | conf->base_width = |
136 | conf->paper_width - conf->left_margin - conf->right_margin; |
137 | conf->page_height = |
138 | conf->paper_height - conf->top_margin - conf->bottom_margin; |
43341922 |
139 | |
43341922 |
140 | IGNORE(idx); /* FIXME */ |
43341922 |
141 | |
142 | /* |
143 | * First, set up some font structures. |
144 | */ |
145 | fontlist = mknew(font_list); |
146 | fontlist->head = fontlist->tail = NULL; |
be76d597 |
147 | conf->tr = make_std_font(fontlist, "Times-Roman"); |
148 | conf->ti = make_std_font(fontlist, "Times-Italic"); |
149 | conf->hr = make_std_font(fontlist, "Helvetica-Bold"); |
150 | conf->hi = make_std_font(fontlist, "Helvetica-BoldOblique"); |
151 | conf->cr = make_std_font(fontlist, "Courier"); |
152 | conf->co = make_std_font(fontlist, "Courier-Oblique"); |
153 | conf->cb = make_std_font(fontlist, "Courier-Bold"); |
43341922 |
154 | |
155 | /* |
2bfd1b76 |
156 | * Format the contents entry for each heading. |
157 | */ |
158 | { |
159 | word *contents_title; |
160 | contents_title = fake_word(L"Contents"); |
161 | |
162 | firstcont = make_para_data(para_UnnumberedChapter, 0, 0, 0, |
163 | NULL, NULL, contents_title, conf); |
164 | lastcont = firstcont; |
165 | lastcont->next = NULL; |
166 | firstcontline = firstcont->first; |
167 | lastcontline = lastcont->last; |
168 | for (p = sourceform; p; p = p->next) { |
169 | word *words; |
170 | int indent; |
171 | |
172 | switch (p->type) { |
173 | case para_Chapter: |
174 | case para_Appendix: |
175 | case para_UnnumberedChapter: |
176 | case para_Heading: |
177 | case para_Subsect: |
178 | switch (p->type) { |
179 | case para_Chapter: |
180 | case para_Appendix: |
181 | words = prepare_contents_title(p->kwtext, L": ", p->words); |
182 | indent = 0; |
183 | break; |
184 | case para_UnnumberedChapter: |
185 | words = prepare_contents_title(NULL, NULL, p->words); |
186 | indent = 0; |
187 | break; |
188 | case para_Heading: |
189 | case para_Subsect: |
190 | words = prepare_contents_title(p->kwtext2, L" ", p->words); |
191 | indent = (p->aux + 1) * conf->contents_indent_step; |
192 | break; |
193 | } |
194 | pdata = make_para_data(para_Normal, p->aux, indent, |
195 | conf->contents_margin, |
196 | NULL, NULL, words, conf); |
197 | pdata->next = NULL; |
198 | pdata->contents_entry = p; |
199 | lastcont->next = pdata; |
200 | lastcont = pdata; |
201 | |
202 | /* |
203 | * Link all contents line structures together into |
204 | * a big list. |
205 | */ |
206 | if (pdata->first) { |
207 | if (lastcontline) { |
208 | lastcontline->next = pdata->first; |
209 | pdata->first->prev = lastcontline; |
210 | } else { |
211 | firstcontline = pdata->first; |
212 | pdata->first->prev = NULL; |
213 | } |
214 | lastcontline = pdata->last; |
215 | lastcontline->next = NULL; |
216 | } |
217 | |
218 | break; |
219 | } |
220 | } |
221 | } |
222 | |
223 | /* |
224 | * Do the main paragraph formatting. |
43341922 |
225 | */ |
226 | indent = 0; |
2bfd1b76 |
227 | used_contents = FALSE; |
43341922 |
228 | firstline = lastline = NULL; |
229 | for (p = sourceform; p; p = p->next) { |
230 | p->private_data = NULL; |
231 | |
232 | switch (p->type) { |
233 | /* |
234 | * These paragraph types are either invisible or don't |
235 | * define text in the normal sense. Either way, they |
236 | * don't require wrapping. |
237 | */ |
238 | case para_IM: |
239 | case para_BR: |
43341922 |
240 | case para_Biblio: |
241 | case para_NotParaType: |
242 | case para_Config: |
243 | case para_VersionID: |
244 | case para_NoCite: |
245 | break; |
246 | |
247 | /* |
248 | * These paragraph types don't require wrapping, but |
249 | * they do affect the line width to which we wrap the |
250 | * rest of the paragraphs, so we need to pay attention. |
251 | */ |
252 | case para_LcontPush: |
be76d597 |
253 | indent += conf->indent_list; break; |
43341922 |
254 | case para_LcontPop: |
be76d597 |
255 | indent -= conf->indent_list; assert(indent >= 0); break; |
43341922 |
256 | case para_QuotePush: |
be76d597 |
257 | indent += conf->indent_quote; break; |
43341922 |
258 | case para_QuotePop: |
be76d597 |
259 | indent -= conf->indent_quote; assert(indent >= 0); break; |
43341922 |
260 | |
261 | /* |
262 | * This paragraph type is special. Process it |
263 | * specially. |
264 | */ |
265 | case para_Code: |
be76d597 |
266 | pdata = code_paragraph(indent, p->words, conf); |
515d216b |
267 | p->private_data = pdata; |
39a0cfb9 |
268 | if (pdata->first != pdata->last) { |
269 | pdata->first->penalty_after += 100000; |
270 | pdata->last->penalty_before += 100000; |
271 | } |
43341922 |
272 | break; |
273 | |
274 | /* |
87bd6353 |
275 | * This paragraph is also special. |
276 | */ |
277 | case para_Rule: |
be76d597 |
278 | pdata = rule_paragraph(indent, conf); |
87bd6353 |
279 | p->private_data = pdata; |
280 | break; |
281 | |
282 | /* |
43341922 |
283 | * All of these paragraph types require wrapping in the |
284 | * ordinary way. So we must supply a set of fonts, a |
285 | * line width and auxiliary information (e.g. bullet |
286 | * text) for each one. |
287 | */ |
288 | case para_Chapter: |
289 | case para_Appendix: |
290 | case para_UnnumberedChapter: |
291 | case para_Heading: |
292 | case para_Subsect: |
293 | case para_Normal: |
294 | case para_BiblioCited: |
295 | case para_Bullet: |
296 | case para_NumberedList: |
297 | case para_DescribedThing: |
298 | case para_Description: |
299 | case para_Copyright: |
300 | case para_Title: |
2bfd1b76 |
301 | pdata = make_para_data(p->type, p->aux, indent, 0, |
be76d597 |
302 | p->kwtext, p->kwtext2, p->words, conf); |
43341922 |
303 | |
43341922 |
304 | p->private_data = pdata; |
305 | |
515d216b |
306 | break; |
307 | } |
308 | |
309 | if (p->private_data) { |
310 | pdata = (para_data *)p->private_data; |
311 | |
43341922 |
312 | /* |
2bfd1b76 |
313 | * If this is the first non-title heading, we link the |
314 | * contents section in before it. |
315 | */ |
316 | if (!used_contents && pdata->outline_level > 0) { |
317 | used_contents = TRUE; |
318 | if (lastpara) |
319 | lastpara->next = firstcont; |
320 | else |
321 | firstpara = firstcont; |
322 | lastpara = lastcont; |
323 | assert(lastpara->next == NULL); |
324 | |
325 | if (lastline) { |
326 | lastline->next = firstcontline; |
327 | firstcontline->prev = lastline; |
328 | } else { |
329 | firstline = firstcontline; |
330 | firstcontline->prev = NULL; |
331 | } |
332 | assert(lastcontline != NULL); |
333 | lastline = lastcontline; |
334 | lastline->next = NULL; |
335 | } |
336 | |
337 | /* |
515d216b |
338 | * Link all line structures together into a big list. |
339 | */ |
43341922 |
340 | if (pdata->first) { |
341 | if (lastline) { |
342 | lastline->next = pdata->first; |
343 | pdata->first->prev = lastline; |
344 | } else { |
345 | firstline = pdata->first; |
346 | pdata->first->prev = NULL; |
347 | } |
348 | lastline = pdata->last; |
2bfd1b76 |
349 | lastline->next = NULL; |
43341922 |
350 | } |
be76d597 |
351 | |
352 | /* |
353 | * Link all paragraph structures together similarly. |
354 | */ |
355 | pdata->next = NULL; |
356 | if (lastpara) |
357 | lastpara->next = pdata; |
358 | else |
359 | firstpara = pdata; |
360 | lastpara = pdata; |
43341922 |
361 | } |
362 | } |
363 | |
364 | /* |
365 | * Now we have an enormous linked list of every line of text in |
366 | * the document. Break it up into pages. |
367 | */ |
be76d597 |
368 | pages = page_breaks(firstline, lastline, conf->page_height); |
43341922 |
369 | |
370 | /* |
2bfd1b76 |
371 | * Number the pages. |
372 | */ |
373 | { |
374 | page_data *page; |
375 | int num = 0; |
376 | for (page = pages; page; page = page->next) { |
377 | char buf[40]; |
378 | sprintf(buf, "%d", ++num); |
379 | page->number = ufroma_dup(buf); |
380 | } |
381 | } |
382 | |
383 | /* |
43341922 |
384 | * Now we're ready to actually lay out the pages. We do this by |
385 | * looping over _paragraphs_, since we may need to track cross- |
386 | * references between lines and even across pages. |
387 | */ |
be76d597 |
388 | for (pdata = firstpara; pdata; pdata = pdata->next) { |
2bfd1b76 |
389 | int last_x; |
be76d597 |
390 | xref_dest dest; |
391 | dest.type = NONE; |
392 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
2bfd1b76 |
393 | last_x = render_line(ldata, conf->left_margin, |
394 | conf->paper_height - conf->top_margin, |
395 | &dest, keywords); |
be76d597 |
396 | if (ldata == pdata->last) |
397 | break; |
398 | } |
87bd6353 |
399 | |
be76d597 |
400 | /* |
2bfd1b76 |
401 | * If this is a contents entry, add leaders and a page |
402 | * number. |
403 | */ |
404 | if (pdata->contents_entry) { |
405 | word *w; |
406 | para_data *target; |
407 | wchar_t *num; |
408 | int wid; |
409 | int x; |
410 | |
411 | assert(pdata->contents_entry->private_data); |
412 | target = (para_data *)pdata->contents_entry->private_data; |
413 | num = target->first->page->number; |
414 | |
415 | w = fake_word(num); |
416 | wid = paper_width_simple(pdata, w); |
417 | sfree(w); |
418 | |
419 | render_string(pdata->last->page, |
420 | pdata->fonts[FONT_NORMAL], |
421 | pdata->sizes[FONT_NORMAL], |
422 | conf->paper_width - conf->right_margin - wid, |
423 | (conf->paper_height - conf->top_margin - |
424 | pdata->last->ypos), num); |
425 | |
426 | for (x = 0; x < conf->base_width; x += conf->leader_separation) |
427 | if (x - conf->leader_separation > last_x - conf->left_margin && |
428 | x + conf->leader_separation < conf->base_width - wid) |
429 | render_string(pdata->last->page, |
430 | pdata->fonts[FONT_NORMAL], |
431 | pdata->sizes[FONT_NORMAL], |
432 | conf->left_margin + x, |
433 | (conf->paper_height - conf->top_margin - |
434 | pdata->last->ypos), L"."); |
435 | } |
436 | |
437 | /* |
be76d597 |
438 | * Render any rectangle (chapter title underline or rule) |
439 | * that goes with this paragraph. |
440 | */ |
441 | switch (pdata->rect_type) { |
442 | case RECT_CHAPTER_UNDERLINE: |
443 | add_rect_to_page(pdata->last->page, |
444 | conf->left_margin, |
445 | (conf->paper_height - conf->top_margin - |
446 | pdata->last->ypos - |
447 | conf->chapter_underline_depth), |
448 | conf->base_width, |
449 | conf->chapter_underline_thickness); |
450 | break; |
451 | case RECT_RULE: |
452 | add_rect_to_page(pdata->first->page, |
453 | conf->left_margin + pdata->first->xpos, |
454 | (conf->paper_height - conf->top_margin - |
455 | pdata->last->ypos - |
456 | pdata->last->line_height), |
457 | conf->base_width - pdata->first->xpos, |
458 | pdata->last->line_height); |
459 | break; |
460 | default: /* placate gcc */ |
461 | break; |
43341922 |
462 | } |
463 | } |
464 | |
f0e51ce1 |
465 | /* |
466 | * Start putting together the overall document structure we're |
467 | * going to return. |
468 | */ |
43341922 |
469 | doc = mknew(document); |
470 | doc->fonts = fontlist; |
471 | doc->pages = pages; |
be76d597 |
472 | doc->paper_width = conf->paper_width; |
473 | doc->paper_height = conf->paper_height; |
f0e51ce1 |
474 | |
475 | /* |
476 | * Collect the section heading paragraphs into a document |
477 | * outline. This is slightly fiddly because the Title paragraph |
478 | * isn't required to be at the start, although all the others |
479 | * must be in order. |
480 | */ |
481 | { |
482 | int osize = 20; |
483 | |
484 | doc->outline_elements = mknewa(outline_element, osize); |
485 | doc->n_outline_elements = 0; |
486 | |
487 | /* First find the title. */ |
be76d597 |
488 | for (pdata = firstpara; pdata; pdata = pdata->next) { |
489 | if (pdata->outline_level == 0) { |
f0e51ce1 |
490 | doc->outline_elements[0].level = 0; |
be76d597 |
491 | doc->outline_elements[0].pdata = pdata; |
f0e51ce1 |
492 | doc->n_outline_elements++; |
493 | break; |
494 | } |
495 | } |
496 | |
497 | /* Then collect the rest. */ |
be76d597 |
498 | for (pdata = firstpara; pdata; pdata = pdata->next) { |
499 | if (pdata->outline_level > 0) { |
f0e51ce1 |
500 | if (doc->n_outline_elements >= osize) { |
501 | osize += 20; |
502 | doc->outline_elements = |
503 | resize(doc->outline_elements, osize); |
504 | } |
505 | |
be76d597 |
506 | doc->outline_elements[doc->n_outline_elements].level = |
507 | pdata->outline_level; |
508 | doc->outline_elements[doc->n_outline_elements].pdata = pdata; |
f0e51ce1 |
509 | doc->n_outline_elements++; |
f0e51ce1 |
510 | } |
511 | } |
512 | } |
513 | |
be76d597 |
514 | sfree(conf); |
515 | |
43341922 |
516 | return doc; |
517 | } |
518 | |
2bfd1b76 |
519 | static para_data *make_para_data(int ptype, int paux, int indent, int rmargin, |
be76d597 |
520 | word *pkwtext, word *pkwtext2, word *pwords, |
521 | paper_conf *conf) |
522 | { |
523 | para_data *pdata; |
524 | line_data *ldata; |
525 | int extra_indent, firstline_indent, aux_indent; |
526 | word *aux, *aux2; |
527 | |
528 | pdata = mknew(para_data); |
529 | pdata->outline_level = -1; |
530 | pdata->outline_title = NULL; |
531 | pdata->rect_type = RECT_NONE; |
2bfd1b76 |
532 | pdata->contents_entry = NULL; |
be76d597 |
533 | |
534 | /* |
535 | * Choose fonts for this paragraph. |
536 | * |
537 | * FIXME: All of this ought to be completely |
538 | * user-configurable. |
539 | */ |
540 | switch (ptype) { |
541 | case para_Title: |
542 | pdata->fonts[FONT_NORMAL] = conf->hr; |
543 | pdata->sizes[FONT_NORMAL] = 24; |
544 | pdata->fonts[FONT_EMPH] = conf->hi; |
545 | pdata->sizes[FONT_EMPH] = 24; |
546 | pdata->fonts[FONT_CODE] = conf->cb; |
547 | pdata->sizes[FONT_CODE] = 24; |
548 | pdata->outline_level = 0; |
549 | break; |
550 | |
551 | case para_Chapter: |
552 | case para_Appendix: |
553 | case para_UnnumberedChapter: |
554 | pdata->fonts[FONT_NORMAL] = conf->hr; |
555 | pdata->sizes[FONT_NORMAL] = 20; |
556 | pdata->fonts[FONT_EMPH] = conf->hi; |
557 | pdata->sizes[FONT_EMPH] = 20; |
558 | pdata->fonts[FONT_CODE] = conf->cb; |
559 | pdata->sizes[FONT_CODE] = 20; |
560 | pdata->outline_level = 1; |
561 | break; |
562 | |
563 | case para_Heading: |
564 | case para_Subsect: |
565 | pdata->fonts[FONT_NORMAL] = conf->hr; |
566 | pdata->fonts[FONT_EMPH] = conf->hi; |
567 | pdata->fonts[FONT_CODE] = conf->cb; |
568 | pdata->sizes[FONT_NORMAL] = |
569 | pdata->sizes[FONT_EMPH] = |
570 | pdata->sizes[FONT_CODE] = |
571 | (paux == 0 ? 16 : paux == 1 ? 14 : 13); |
572 | pdata->outline_level = 2 + paux; |
573 | break; |
574 | |
575 | case para_Normal: |
576 | case para_BiblioCited: |
577 | case para_Bullet: |
578 | case para_NumberedList: |
579 | case para_DescribedThing: |
580 | case para_Description: |
581 | case para_Copyright: |
582 | pdata->fonts[FONT_NORMAL] = conf->tr; |
583 | pdata->sizes[FONT_NORMAL] = 12; |
584 | pdata->fonts[FONT_EMPH] = conf->ti; |
585 | pdata->sizes[FONT_EMPH] = 12; |
586 | pdata->fonts[FONT_CODE] = conf->cr; |
587 | pdata->sizes[FONT_CODE] = 12; |
588 | break; |
589 | } |
590 | |
591 | /* |
592 | * Also select an indentation level depending on the |
593 | * paragraph type (list paragraphs other than |
594 | * para_DescribedThing need extra indent). |
595 | * |
596 | * (FIXME: Perhaps at some point we might even arrange |
597 | * for the user to be able to request indented first |
598 | * lines in paragraphs.) |
599 | */ |
600 | if (ptype == para_Bullet || |
601 | ptype == para_NumberedList || |
602 | ptype == para_Description) { |
603 | extra_indent = firstline_indent = conf->indent_list; |
604 | } else { |
605 | extra_indent = firstline_indent = 0; |
606 | } |
607 | |
608 | /* |
609 | * Find the auxiliary text for this paragraph. |
610 | */ |
611 | aux = aux2 = NULL; |
612 | aux_indent = 0; |
613 | |
614 | switch (ptype) { |
615 | case para_Chapter: |
616 | case para_Appendix: |
617 | case para_Heading: |
618 | case para_Subsect: |
619 | /* |
620 | * For some heading styles (FIXME: be able to |
621 | * configure which), the auxiliary text contains |
622 | * the chapter number and is arranged to be |
623 | * right-aligned a few points left of the primary |
624 | * margin. For other styles, the auxiliary text is |
625 | * the full chapter _name_ and takes up space |
626 | * within the (wrapped) chapter title, meaning that |
627 | * we must move the first line indent over to make |
628 | * space for it. |
629 | */ |
630 | if (ptype == para_Heading || ptype == para_Subsect) { |
631 | int len; |
632 | |
633 | aux = pkwtext2; |
634 | len = paper_width_simple(pdata, pkwtext2); |
635 | aux_indent = -len - conf->sect_num_left_space; |
636 | |
637 | pdata->outline_title = |
638 | prepare_outline_title(pkwtext2, L" ", pwords); |
639 | } else { |
640 | aux = pkwtext; |
2bfd1b76 |
641 | aux2 = fake_word(L": "); |
be76d597 |
642 | aux_indent = 0; |
643 | |
644 | firstline_indent += paper_width_simple(pdata, aux); |
645 | firstline_indent += paper_width_simple(pdata, aux2); |
646 | |
647 | pdata->outline_title = |
648 | prepare_outline_title(pkwtext, L": ", pwords); |
649 | } |
650 | break; |
651 | |
652 | case para_Bullet: |
653 | /* |
654 | * Auxiliary text consisting of a bullet. (FIXME: |
655 | * configurable bullet.) |
656 | */ |
2bfd1b76 |
657 | aux = fake_word(L"\x2022"); |
be76d597 |
658 | aux_indent = indent + conf->indent_list_bullet; |
659 | break; |
660 | |
661 | case para_NumberedList: |
662 | /* |
663 | * Auxiliary text consisting of the number followed |
664 | * by a (FIXME: configurable) full stop. |
665 | */ |
666 | aux = pkwtext; |
2bfd1b76 |
667 | aux2 = fake_word(L"."); |
be76d597 |
668 | aux_indent = indent + conf->indent_list_bullet; |
669 | break; |
670 | |
671 | case para_BiblioCited: |
672 | /* |
673 | * Auxiliary text consisting of the bibliography |
674 | * reference text, and a trailing space. |
675 | */ |
676 | aux = pkwtext; |
2bfd1b76 |
677 | aux2 = fake_word(L" "); |
be76d597 |
678 | aux_indent = indent; |
679 | firstline_indent += paper_width_simple(pdata, aux); |
680 | firstline_indent += paper_width_simple(pdata, aux2); |
681 | break; |
682 | } |
683 | |
684 | if (pdata->outline_level >= 0 && !pdata->outline_title) { |
685 | pdata->outline_title = |
686 | prepare_outline_title(NULL, NULL, pwords); |
687 | } |
688 | |
2bfd1b76 |
689 | wrap_paragraph(pdata, pwords, conf->base_width - rmargin, |
be76d597 |
690 | indent + firstline_indent, |
691 | indent + extra_indent); |
692 | |
693 | pdata->first->aux_text = aux; |
694 | pdata->first->aux_text_2 = aux2; |
695 | pdata->first->aux_left_indent = aux_indent; |
696 | |
697 | /* |
698 | * Line breaking penalties. |
699 | */ |
700 | switch (ptype) { |
701 | case para_Chapter: |
702 | case para_Appendix: |
703 | case para_Heading: |
704 | case para_Subsect: |
705 | case para_UnnumberedChapter: |
706 | /* |
707 | * Fixed and large penalty for breaking straight |
708 | * after a heading; corresponding bonus for |
709 | * breaking straight before. |
710 | */ |
711 | pdata->first->penalty_before = -500000; |
712 | pdata->last->penalty_after = 500000; |
713 | for (ldata = pdata->first; ldata; ldata = ldata->next) |
714 | ldata->penalty_after = 500000; |
715 | break; |
716 | |
717 | case para_DescribedThing: |
718 | /* |
719 | * This is treated a bit like a small heading: |
720 | * there's a penalty for breaking after it (i.e. |
721 | * between it and its description), and a bonus for |
722 | * breaking before it (actually _between_ list |
723 | * items). |
724 | */ |
725 | pdata->first->penalty_before = -200000; |
726 | pdata->last->penalty_after = 200000; |
727 | break; |
728 | |
729 | default: |
730 | /* |
731 | * Most paragraph types: widow/orphan control by |
732 | * discouraging breaking one line from the end of |
733 | * any paragraph. |
734 | */ |
735 | if (pdata->first != pdata->last) { |
736 | pdata->first->penalty_after = 100000; |
737 | pdata->last->penalty_before = 100000; |
738 | } |
739 | break; |
740 | } |
741 | |
742 | standard_line_spacing(pdata, conf); |
743 | |
744 | /* |
745 | * Some kinds of section heading require a page break before |
746 | * them and an underline after. |
747 | */ |
748 | if (ptype == para_Title || |
749 | ptype == para_Chapter || |
750 | ptype == para_Appendix || |
751 | ptype == para_UnnumberedChapter) { |
752 | pdata->first->page_break = TRUE; |
753 | pdata->first->space_before = conf->chapter_top_space; |
754 | pdata->last->space_after += |
755 | (conf->chapter_underline_depth + |
756 | conf->chapter_underline_thickness); |
757 | pdata->rect_type = RECT_CHAPTER_UNDERLINE; |
758 | } |
759 | |
760 | return pdata; |
761 | } |
762 | |
763 | static void standard_line_spacing(para_data *pdata, paper_conf *conf) |
764 | { |
765 | line_data *ldata; |
766 | |
767 | /* |
768 | * Set the line spacing for each line in this paragraph. |
769 | */ |
770 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
771 | if (ldata == pdata->first) |
772 | ldata->space_before = conf->base_para_spacing / 2; |
773 | else |
774 | ldata->space_before = conf->base_leading / 2; |
775 | if (ldata == pdata->last) |
776 | ldata->space_after = conf->base_para_spacing / 2; |
777 | else |
778 | ldata->space_after = conf->base_leading / 2; |
779 | ldata->page_break = FALSE; |
780 | } |
781 | } |
782 | |
43341922 |
783 | static font_encoding *new_font_encoding(font_data *font) |
784 | { |
785 | font_encoding *fe; |
786 | int i; |
787 | |
788 | fe = mknew(font_encoding); |
789 | fe->next = NULL; |
790 | |
791 | if (font->list->tail) |
792 | font->list->tail->next = fe; |
793 | else |
794 | font->list->head = fe; |
795 | font->list->tail = fe; |
796 | |
797 | fe->font = font; |
798 | fe->free_pos = 0x21; |
799 | |
800 | for (i = 0; i < 256; i++) { |
801 | fe->vector[i] = NULL; |
802 | fe->indices[i] = -1; |
803 | fe->to_unicode[i] = 0xFFFF; |
804 | } |
805 | |
806 | return fe; |
807 | } |
808 | |
809 | static font_data *make_std_font(font_list *fontlist, char const *name) |
810 | { |
811 | const int *widths; |
812 | int nglyphs; |
813 | font_data *f; |
814 | font_encoding *fe; |
815 | int i; |
816 | |
817 | widths = ps_std_font_widths(name); |
818 | if (!widths) |
819 | return NULL; |
820 | |
821 | for (nglyphs = 0; ps_std_glyphs[nglyphs] != NULL; nglyphs++); |
822 | |
823 | f = mknew(font_data); |
824 | |
825 | f->list = fontlist; |
826 | f->name = name; |
827 | f->nglyphs = nglyphs; |
828 | f->glyphs = ps_std_glyphs; |
829 | f->widths = widths; |
830 | f->subfont_map = mknewa(subfont_map_entry, nglyphs); |
831 | |
832 | /* |
833 | * Our first subfont will contain all of US-ASCII. This isn't |
834 | * really necessary - we could just create custom subfonts |
835 | * precisely as the whim of render_string dictated - but |
836 | * instinct suggests that it might be nice to have the text in |
837 | * the output files look _marginally_ recognisable. |
838 | */ |
839 | fe = new_font_encoding(f); |
840 | fe->free_pos = 0xA1; /* only the top half is free */ |
841 | f->latest_subfont = fe; |
842 | |
843 | for (i = 0; i < (int)lenof(f->bmp); i++) |
844 | f->bmp[i] = 0xFFFF; |
845 | |
846 | for (i = 0; i < nglyphs; i++) { |
847 | wchar_t ucs; |
848 | ucs = ps_glyph_to_unicode(f->glyphs[i]); |
849 | assert(ucs != 0xFFFF); |
850 | f->bmp[ucs] = i; |
851 | if (ucs >= 0x20 && ucs <= 0x7E) { |
852 | fe->vector[ucs] = f->glyphs[i]; |
853 | fe->indices[ucs] = i; |
854 | fe->to_unicode[ucs] = ucs; |
855 | f->subfont_map[i].subfont = fe; |
856 | f->subfont_map[i].position = ucs; |
857 | } else { |
858 | /* |
859 | * This character is not yet assigned to a subfont. |
860 | */ |
861 | f->subfont_map[i].subfont = NULL; |
862 | f->subfont_map[i].position = 0; |
863 | } |
864 | } |
865 | |
866 | return f; |
867 | } |
868 | |
869 | static int string_width(font_data *font, wchar_t const *string, int *errs) |
870 | { |
871 | int width = 0; |
872 | |
873 | if (errs) |
874 | *errs = 0; |
875 | |
876 | for (; *string; string++) { |
877 | int index; |
878 | |
879 | index = font->bmp[(unsigned short)*string]; |
880 | if (index == 0xFFFF) { |
881 | if (errs) |
882 | *errs = 1; |
883 | } else { |
884 | width += font->widths[index]; |
885 | } |
886 | } |
887 | |
888 | return width; |
889 | } |
890 | |
faad4952 |
891 | static int paper_width_internal(void *vctx, word *word, int *nspaces); |
43341922 |
892 | |
893 | struct paper_width_ctx { |
894 | int minspacewidth; |
895 | para_data *pdata; |
896 | }; |
897 | |
faad4952 |
898 | static int paper_width_list(void *vctx, word *text, word *end, int *nspaces) { |
43341922 |
899 | int w = 0; |
faad4952 |
900 | while (text && text != end) { |
901 | w += paper_width_internal(vctx, text, nspaces); |
43341922 |
902 | text = text->next; |
903 | } |
904 | return w; |
905 | } |
906 | |
faad4952 |
907 | static int paper_width_internal(void *vctx, word *word, int *nspaces) |
43341922 |
908 | { |
909 | struct paper_width_ctx *ctx = (struct paper_width_ctx *)vctx; |
910 | int style, type, findex, width, errs; |
911 | wchar_t *str; |
912 | |
913 | switch (word->type) { |
914 | case word_HyperLink: |
915 | case word_HyperEnd: |
916 | case word_UpperXref: |
917 | case word_LowerXref: |
918 | case word_XrefEnd: |
919 | case word_IndexRef: |
920 | return 0; |
921 | } |
922 | |
923 | style = towordstyle(word->type); |
924 | type = removeattr(word->type); |
925 | |
926 | findex = (style == word_Normal ? FONT_NORMAL : |
927 | style == word_Emph ? FONT_EMPH : |
928 | FONT_CODE); |
929 | |
930 | if (type == word_Normal) { |
931 | str = word->text; |
932 | } else if (type == word_WhiteSpace) { |
faad4952 |
933 | if (findex != FONT_CODE) { |
934 | if (nspaces) |
935 | (*nspaces)++; |
43341922 |
936 | return ctx->minspacewidth; |
faad4952 |
937 | } else |
43341922 |
938 | str = L" "; |
939 | } else /* if (type == word_Quote) */ { |
940 | if (word->aux == quote_Open) |
941 | str = L"\x2018"; /* FIXME: configurability! */ |
942 | else |
943 | str = L"\x2019"; /* FIXME: configurability! */ |
944 | } |
945 | |
946 | width = string_width(ctx->pdata->fonts[findex], str, &errs); |
947 | |
948 | if (errs && word->alt) |
faad4952 |
949 | return paper_width_list(vctx, word->alt, NULL, nspaces); |
43341922 |
950 | else |
951 | return ctx->pdata->sizes[findex] * width; |
952 | } |
953 | |
faad4952 |
954 | static int paper_width(void *vctx, word *word) |
955 | { |
956 | return paper_width_internal(vctx, word, NULL); |
957 | } |
958 | |
515d216b |
959 | static int paper_width_simple(para_data *pdata, word *text) |
960 | { |
961 | struct paper_width_ctx ctx; |
962 | |
963 | ctx.pdata = pdata; |
964 | ctx.minspacewidth = |
965 | (pdata->sizes[FONT_NORMAL] * |
966 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
967 | |
968 | return paper_width_list(&ctx, text, NULL, NULL); |
969 | } |
970 | |
43341922 |
971 | static void wrap_paragraph(para_data *pdata, word *words, |
972 | int w, int i1, int i2) |
973 | { |
974 | wrappedline *wrapping, *p; |
975 | int spacewidth; |
976 | struct paper_width_ctx ctx; |
977 | int line_height; |
978 | |
979 | /* |
980 | * We're going to need to store the line height in every line |
981 | * structure we generate. |
982 | */ |
983 | { |
984 | int i; |
985 | line_height = 0; |
986 | for (i = 0; i < NFONTS; i++) |
987 | if (line_height < pdata->sizes[i]) |
988 | line_height = pdata->sizes[i]; |
989 | line_height *= 4096; |
990 | } |
991 | |
992 | spacewidth = (pdata->sizes[FONT_NORMAL] * |
993 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
994 | if (spacewidth == 0) { |
995 | /* |
996 | * A font without a space?! Disturbing. I hope this never |
997 | * comes up, but I'll make a random guess anyway and set my |
998 | * space width to half the point size. |
999 | */ |
1000 | spacewidth = pdata->sizes[FONT_NORMAL] * 4096 / 2; |
1001 | } |
1002 | |
1003 | /* |
1004 | * I'm going to set the _minimum_ space width to 3/5 of the |
1005 | * standard one, and use the standard one as the optimum. |
1006 | */ |
1007 | ctx.minspacewidth = spacewidth * 3 / 5; |
1008 | ctx.pdata = pdata; |
1009 | |
1010 | wrapping = wrap_para(words, w - i1, w - i2, paper_width, &ctx, spacewidth); |
1011 | |
1012 | /* |
1013 | * Having done the wrapping, we now concoct a set of line_data |
1014 | * structures. |
1015 | */ |
1016 | pdata->first = pdata->last = NULL; |
1017 | |
1018 | for (p = wrapping; p; p = p->next) { |
1019 | line_data *ldata; |
1020 | word *wd; |
1021 | int len, wid, spaces; |
1022 | |
1023 | ldata = mknew(line_data); |
1024 | |
1025 | ldata->pdata = pdata; |
1026 | ldata->first = p->begin; |
faad4952 |
1027 | ldata->end = p->end; |
43341922 |
1028 | ldata->line_height = line_height; |
1029 | |
1030 | ldata->xpos = (p == wrapping ? i1 : i2); |
1031 | |
1032 | if (pdata->last) { |
1033 | pdata->last->next = ldata; |
1034 | ldata->prev = pdata->last; |
1035 | } else { |
1036 | pdata->first = ldata; |
1037 | ldata->prev = NULL; |
1038 | } |
1039 | ldata->next = NULL; |
1040 | pdata->last = ldata; |
1041 | |
43341922 |
1042 | spaces = 0; |
faad4952 |
1043 | len = paper_width_list(&ctx, ldata->first, ldata->end, &spaces); |
1044 | wid = (p == wrapping ? w - i1 : w - i2); |
43341922 |
1045 | wd = ldata->first; |
43341922 |
1046 | |
faad4952 |
1047 | ldata->hshortfall = wid - len; |
1048 | ldata->nspaces = spaces; |
1049 | /* |
1050 | * This tells us how much the space width needs to |
1051 | * change from _min_spacewidth. But we want to store |
1052 | * its difference from the _natural_ space width, to |
1053 | * make the text rendering easier. |
1054 | */ |
1055 | ldata->hshortfall += ctx.minspacewidth * spaces; |
1056 | ldata->hshortfall -= spacewidth * spaces; |
1057 | /* |
1058 | * Special case: on the last line of a paragraph, we |
1059 | * never stretch spaces. |
1060 | */ |
1061 | if (ldata->hshortfall > 0 && !p->next) |
1062 | ldata->hshortfall = 0; |
43341922 |
1063 | |
1064 | ldata->aux_text = NULL; |
515d216b |
1065 | ldata->aux_text_2 = NULL; |
43341922 |
1066 | ldata->aux_left_indent = 0; |
39a0cfb9 |
1067 | ldata->penalty_before = ldata->penalty_after = 0; |
43341922 |
1068 | } |
1069 | |
1070 | } |
1071 | |
1072 | static page_data *page_breaks(line_data *first, line_data *last, |
1073 | int page_height) |
1074 | { |
1075 | line_data *l, *m; |
1076 | page_data *ph, *pt; |
1077 | |
1078 | /* |
1079 | * Page breaking is done by a close analogue of the optimal |
1080 | * paragraph wrapping algorithm used by wrap_para(). We work |
1081 | * backwards from the end of the document line by line; for |
1082 | * each line, we contemplate every possible number of lines we |
1083 | * could put on a page starting with that line, determine a |
1084 | * cost function for each one, add it to the pre-computed cost |
1085 | * function for optimally page-breaking everything after that |
1086 | * page, and pick the best option. |
1087 | * |
1088 | * Since my line_data structures are only used for this |
1089 | * purpose, I might as well just store the algorithm data |
1090 | * directly in them. |
1091 | */ |
1092 | |
1093 | for (l = last; l; l = l->prev) { |
1094 | int minheight, text = 0, space = 0; |
1095 | int cost; |
1096 | |
1097 | l->bestcost = -1; |
1098 | for (m = l; m; m = m->next) { |
1099 | if (m != l && m->page_break) |
1100 | break; /* we've gone as far as we can */ |
1101 | |
1102 | if (m != l) |
1103 | space += m->prev->space_after; |
1104 | if (m != l || m->page_break) |
1105 | space += m->space_before; |
1106 | text += m->line_height; |
1107 | minheight = text + space; |
1108 | |
1109 | if (m != l && minheight > page_height) |
1110 | break; |
1111 | |
1112 | /* |
1113 | * Compute the cost of this arrangement, as the square |
1114 | * of the amount of wasted space on the page. |
1115 | * Exception: if this is the last page before a |
1116 | * mandatory break or the document end, we don't |
1117 | * penalise a large blank area. |
1118 | */ |
1119 | if (m->next && !m->next->page_break) |
1120 | { |
1121 | int x = page_height - minheight; |
1122 | int xf; |
1123 | |
1124 | xf = x & 0xFF; |
1125 | x >>= 8; |
1126 | |
1127 | cost = x*x; |
1128 | cost += (x * xf) >> 8; |
1129 | } else |
1130 | cost = 0; |
1131 | |
39a0cfb9 |
1132 | if (m->next && !m->next->page_break) { |
1133 | cost += m->penalty_after; |
1134 | cost += m->next->penalty_before; |
1135 | } |
43341922 |
1136 | |
1137 | if (m->next && !m->next->page_break) |
1138 | cost += m->next->bestcost; |
43341922 |
1139 | if (l->bestcost == -1 || l->bestcost > cost) { |
1140 | /* |
1141 | * This is the best option yet for this starting |
1142 | * point. |
1143 | */ |
1144 | l->bestcost = cost; |
1145 | if (m->next && !m->next->page_break) |
faad4952 |
1146 | l->vshortfall = page_height - minheight; |
43341922 |
1147 | else |
faad4952 |
1148 | l->vshortfall = 0; |
43341922 |
1149 | l->text = text; |
1150 | l->space = space; |
1151 | l->page_last = m; |
1152 | } |
1153 | } |
1154 | } |
1155 | |
1156 | /* |
1157 | * Now go through the line list forwards and assemble the |
1158 | * actual pages. |
1159 | */ |
1160 | ph = pt = NULL; |
1161 | |
1162 | l = first; |
1163 | while (l) { |
1164 | page_data *page; |
1165 | int text, space; |
1166 | |
1167 | page = mknew(page_data); |
1168 | page->next = NULL; |
1169 | page->prev = pt; |
1170 | if (pt) |
1171 | pt->next = page; |
1172 | else |
1173 | ph = page; |
1174 | pt = page; |
1175 | |
1176 | page->first_line = l; |
1177 | page->last_line = l->page_last; |
1178 | |
1179 | page->first_text = page->last_text = NULL; |
138d7ffb |
1180 | page->first_xref = page->last_xref = NULL; |
23765aeb |
1181 | page->first_rect = page->last_rect = NULL; |
138d7ffb |
1182 | |
43341922 |
1183 | /* |
1184 | * Now assign a y-coordinate to each line on the page. |
1185 | */ |
1186 | text = space = 0; |
1187 | for (l = page->first_line; l; l = l->next) { |
1188 | if (l != page->first_line) |
1189 | space += l->prev->space_after; |
1190 | if (l != page->first_line || l->page_break) |
1191 | space += l->space_before; |
1192 | text += l->line_height; |
1193 | |
1194 | l->page = page; |
1195 | l->ypos = text + space + |
faad4952 |
1196 | space * (float)page->first_line->vshortfall / |
43341922 |
1197 | page->first_line->space; |
1198 | |
1199 | if (l == page->last_line) |
1200 | break; |
1201 | } |
1202 | |
1203 | l = page->last_line->next; |
1204 | } |
1205 | |
1206 | return ph; |
1207 | } |
1208 | |
23765aeb |
1209 | static void add_rect_to_page(page_data *page, int x, int y, int w, int h) |
1210 | { |
1211 | rect *r = mknew(rect); |
1212 | |
1213 | r->next = NULL; |
1214 | if (page->last_rect) |
1215 | page->last_rect->next = r; |
1216 | else |
1217 | page->first_rect = r; |
1218 | page->last_rect = r; |
1219 | |
1220 | r->x = x; |
1221 | r->y = y; |
1222 | r->w = w; |
1223 | r->h = h; |
1224 | } |
1225 | |
43341922 |
1226 | static void add_string_to_page(page_data *page, int x, int y, |
1227 | font_encoding *fe, int size, char *text) |
1228 | { |
1229 | text_fragment *frag; |
1230 | |
1231 | frag = mknew(text_fragment); |
1232 | frag->next = NULL; |
1233 | |
1234 | if (page->last_text) |
1235 | page->last_text->next = frag; |
1236 | else |
1237 | page->first_text = frag; |
1238 | page->last_text = frag; |
1239 | |
1240 | frag->x = x; |
1241 | frag->y = y; |
1242 | frag->fe = fe; |
1243 | frag->fontsize = size; |
1244 | frag->text = dupstr(text); |
1245 | } |
1246 | |
1247 | /* |
1248 | * Returns the updated x coordinate. |
1249 | */ |
1250 | static int render_string(page_data *page, font_data *font, int fontsize, |
1251 | int x, int y, wchar_t *str) |
1252 | { |
1253 | char *text; |
1254 | int textpos, textwid, glyph; |
1255 | font_encoding *subfont = NULL, *sf; |
1256 | |
1257 | text = mknewa(char, 1 + ustrlen(str)); |
1258 | textpos = textwid = 0; |
1259 | |
1260 | while (*str) { |
1261 | glyph = font->bmp[*str]; |
1262 | |
1263 | if (glyph == 0xFFFF) |
1264 | continue; /* nothing more we can do here */ |
1265 | |
1266 | /* |
1267 | * Find which subfont this character is going in. |
1268 | */ |
1269 | sf = font->subfont_map[glyph].subfont; |
1270 | |
1271 | if (!sf) { |
1272 | int c; |
1273 | |
1274 | /* |
1275 | * This character is not yet in a subfont. Assign one. |
1276 | */ |
1277 | if (font->latest_subfont->free_pos >= 0x100) |
1278 | font->latest_subfont = new_font_encoding(font); |
1279 | |
1280 | c = font->latest_subfont->free_pos++; |
1281 | if (font->latest_subfont->free_pos == 0x7F) |
1282 | font->latest_subfont->free_pos = 0xA1; |
1283 | |
1284 | font->subfont_map[glyph].subfont = font->latest_subfont; |
1285 | font->subfont_map[glyph].position = c; |
1286 | font->latest_subfont->vector[c] = font->glyphs[glyph]; |
1287 | font->latest_subfont->indices[c] = glyph; |
1288 | font->latest_subfont->to_unicode[c] = *str; |
1289 | |
1290 | sf = font->latest_subfont; |
1291 | } |
1292 | |
1293 | if (!subfont || sf != subfont) { |
1294 | if (subfont) { |
1295 | text[textpos] = '\0'; |
1296 | add_string_to_page(page, x, y, subfont, fontsize, text); |
1297 | x += textwid; |
1298 | } else { |
1299 | assert(textpos == 0); |
1300 | } |
1301 | textpos = 0; |
1302 | subfont = sf; |
1303 | } |
1304 | |
1305 | text[textpos++] = font->subfont_map[glyph].position; |
1306 | textwid += font->widths[glyph] * fontsize; |
1307 | |
1308 | str++; |
1309 | } |
1310 | |
1311 | if (textpos > 0) { |
1312 | text[textpos] = '\0'; |
1313 | add_string_to_page(page, x, y, subfont, fontsize, text); |
1314 | x += textwid; |
1315 | } |
1316 | |
1317 | return x; |
1318 | } |
1319 | |
1320 | /* |
1321 | * Returns the updated x coordinate. |
1322 | */ |
138d7ffb |
1323 | static int render_text(page_data *page, para_data *pdata, line_data *ldata, |
1324 | int x, int y, word *text, word *text_end, xref **xr, |
1325 | int shortfall, int nspaces, int *nspace, |
1326 | keywordlist *keywords) |
43341922 |
1327 | { |
faad4952 |
1328 | while (text && text != text_end) { |
43341922 |
1329 | int style, type, findex, errs; |
1330 | wchar_t *str; |
138d7ffb |
1331 | xref_dest dest; |
43341922 |
1332 | |
1333 | switch (text->type) { |
138d7ffb |
1334 | /* |
1335 | * Start a cross-reference. |
1336 | */ |
43341922 |
1337 | case word_HyperLink: |
43341922 |
1338 | case word_UpperXref: |
1339 | case word_LowerXref: |
138d7ffb |
1340 | |
1341 | if (text->type == word_HyperLink) { |
1342 | dest.type = URL; |
1343 | dest.url = utoa_dup(text->text); |
1344 | dest.page = NULL; |
1345 | } else { |
1346 | keyword *kwl = kw_lookup(keywords, text->text); |
1347 | para_data *pdata; |
1348 | |
1349 | if (kwl) { |
1350 | assert(kwl->para->private_data); |
1351 | pdata = (para_data *) kwl->para->private_data; |
1352 | dest.type = PAGE; |
1353 | dest.page = pdata->first->page; |
1354 | dest.url = NULL; |
1355 | } else { |
1356 | /* |
1357 | * Shouldn't happen, but *shrug* |
1358 | */ |
1359 | dest.type = NONE; |
1360 | dest.page = NULL; |
1361 | dest.url = NULL; |
1362 | } |
1363 | } |
1364 | if (dest.type != NONE) { |
1365 | *xr = mknew(xref); |
1366 | (*xr)->dest = dest; /* structure copy */ |
1367 | if (page->last_xref) |
1368 | page->last_xref->next = *xr; |
1369 | else |
1370 | page->first_xref = *xr; |
1371 | page->last_xref = *xr; |
23765aeb |
1372 | (*xr)->next = NULL; |
138d7ffb |
1373 | |
1374 | /* |
1375 | * FIXME: Ideally we should have, and use, some |
1376 | * vertical font metric information here so that |
1377 | * our cross-ref rectangle can take account of |
1378 | * descenders and the font's cap height. This will |
1379 | * do for the moment, but it isn't ideal. |
1380 | */ |
1381 | (*xr)->lx = (*xr)->rx = x; |
1382 | (*xr)->by = y; |
1383 | (*xr)->ty = y + ldata->line_height; |
1384 | } |
1385 | goto nextword; |
1386 | |
1387 | /* |
1388 | * Finish extending a cross-reference box. |
1389 | */ |
1390 | case word_HyperEnd: |
43341922 |
1391 | case word_XrefEnd: |
138d7ffb |
1392 | *xr = NULL; |
1393 | goto nextword; |
1394 | |
43341922 |
1395 | case word_IndexRef: |
1396 | goto nextword; |
1397 | /* |
a8070b46 |
1398 | * FIXME: we should do something with this. |
43341922 |
1399 | */ |
1400 | } |
1401 | |
1402 | style = towordstyle(text->type); |
1403 | type = removeattr(text->type); |
1404 | |
1405 | findex = (style == word_Normal ? FONT_NORMAL : |
1406 | style == word_Emph ? FONT_EMPH : |
1407 | FONT_CODE); |
1408 | |
1409 | if (type == word_Normal) { |
1410 | str = text->text; |
1411 | } else if (type == word_WhiteSpace) { |
1412 | x += pdata->sizes[findex] * |
1413 | string_width(pdata->fonts[findex], L" ", NULL); |
faad4952 |
1414 | if (nspaces && findex != FONT_CODE) { |
1415 | x += (*nspace+1) * shortfall / nspaces; |
1416 | x -= *nspace * shortfall / nspaces; |
1417 | (*nspace)++; |
1418 | } |
43341922 |
1419 | goto nextword; |
1420 | } else /* if (type == word_Quote) */ { |
1421 | if (text->aux == quote_Open) |
1422 | str = L"\x2018"; /* FIXME: configurability! */ |
1423 | else |
1424 | str = L"\x2019"; /* FIXME: configurability! */ |
1425 | } |
1426 | |
1427 | (void) string_width(pdata->fonts[findex], str, &errs); |
1428 | |
1429 | if (errs && text->alt) |
138d7ffb |
1430 | x = render_text(page, pdata, ldata, x, y, text->alt, NULL, |
1431 | xr, shortfall, nspaces, nspace, keywords); |
43341922 |
1432 | else |
1433 | x = render_string(page, pdata->fonts[findex], |
1434 | pdata->sizes[findex], x, y, str); |
1435 | |
138d7ffb |
1436 | if (*xr) |
1437 | (*xr)->rx = x; |
1438 | |
43341922 |
1439 | nextword: |
43341922 |
1440 | text = text->next; |
1441 | } |
1442 | |
1443 | return x; |
1444 | } |
1445 | |
2bfd1b76 |
1446 | /* |
1447 | * Returns the last x position used on the line. |
1448 | */ |
1449 | static int render_line(line_data *ldata, int left_x, int top_y, |
1450 | xref_dest *dest, keywordlist *keywords) |
43341922 |
1451 | { |
faad4952 |
1452 | int nspace; |
138d7ffb |
1453 | xref *xr; |
2bfd1b76 |
1454 | int ret = 0; |
138d7ffb |
1455 | |
faad4952 |
1456 | if (ldata->aux_text) { |
515d216b |
1457 | int x; |
138d7ffb |
1458 | xr = NULL; |
faad4952 |
1459 | nspace = 0; |
515d216b |
1460 | x = render_text(ldata->page, ldata->pdata, ldata, |
1461 | left_x + ldata->aux_left_indent, |
1462 | top_y - ldata->ypos, |
1463 | ldata->aux_text, NULL, &xr, 0, 0, &nspace, keywords); |
1464 | if (ldata->aux_text_2) |
1465 | render_text(ldata->page, ldata->pdata, ldata, |
1466 | x, top_y - ldata->ypos, |
1467 | ldata->aux_text_2, NULL, &xr, 0, 0, &nspace, keywords); |
faad4952 |
1468 | } |
1469 | nspace = 0; |
138d7ffb |
1470 | |
87bd6353 |
1471 | if (ldata->first) { |
138d7ffb |
1472 | /* |
87bd6353 |
1473 | * There might be a cross-reference carried over from a |
1474 | * previous line. |
138d7ffb |
1475 | */ |
87bd6353 |
1476 | if (dest->type != NONE) { |
1477 | xr = mknew(xref); |
1478 | xr->next = NULL; |
1479 | xr->dest = *dest; /* structure copy */ |
1480 | if (ldata->page->last_xref) |
1481 | ldata->page->last_xref->next = xr; |
1482 | else |
1483 | ldata->page->first_xref = xr; |
1484 | ldata->page->last_xref = xr; |
1485 | xr->lx = xr->rx = left_x + ldata->xpos; |
1486 | xr->by = top_y - ldata->ypos; |
1487 | xr->ty = top_y - ldata->ypos + ldata->line_height; |
1488 | } else |
1489 | xr = NULL; |
1490 | |
2bfd1b76 |
1491 | ret = render_text(ldata->page, ldata->pdata, ldata, |
1492 | left_x + ldata->xpos, |
1493 | top_y - ldata->ypos, ldata->first, ldata->end, &xr, |
1494 | ldata->hshortfall, ldata->nspaces, &nspace, |
1495 | keywords); |
87bd6353 |
1496 | |
1497 | if (xr) { |
1498 | /* |
1499 | * There's a cross-reference continued on to the next line. |
1500 | */ |
1501 | *dest = xr->dest; |
1502 | } else |
1503 | dest->type = NONE; |
1504 | } |
2bfd1b76 |
1505 | |
1506 | return ret; |
43341922 |
1507 | } |
515d216b |
1508 | |
be76d597 |
1509 | static para_data *code_paragraph(int indent, word *words, paper_conf *conf) |
515d216b |
1510 | { |
be76d597 |
1511 | para_data *pdata = mknew(para_data); |
1512 | |
515d216b |
1513 | /* |
1514 | * For code paragraphs, I'm going to hack grievously and |
1515 | * pretend the three normal fonts are the three code paragraph |
1516 | * fonts. |
1517 | */ |
be76d597 |
1518 | pdata->fonts[FONT_NORMAL] = conf->cb; |
1519 | pdata->fonts[FONT_EMPH] = conf->co; |
1520 | pdata->fonts[FONT_CODE] = conf->cb; |
515d216b |
1521 | pdata->sizes[FONT_NORMAL] = |
1522 | pdata->sizes[FONT_EMPH] = |
be76d597 |
1523 | pdata->sizes[FONT_CODE] = 12; |
515d216b |
1524 | |
1525 | pdata->first = pdata->last = NULL; |
be76d597 |
1526 | pdata->outline_level = -1; |
1527 | pdata->rect_type = RECT_NONE; |
2bfd1b76 |
1528 | pdata->contents_entry = NULL; |
515d216b |
1529 | |
1530 | for (; words; words = words->next) { |
1531 | wchar_t *t, *e, *start; |
1532 | word *lhead = NULL, *ltail = NULL, *w; |
1533 | line_data *ldata; |
1534 | int prev = -1, curr; |
1535 | |
1536 | t = words->text; |
1537 | if (words->next && words->next->type == word_Emph) { |
1538 | e = words->next->text; |
1539 | words = words->next; |
1540 | } else |
1541 | e = NULL; |
1542 | |
1543 | start = t; |
1544 | |
1545 | while (*start) { |
1546 | while (*t) { |
1547 | if (!e || !*e) |
1548 | curr = 0; |
1549 | else if (*e == L'i') |
1550 | curr = 1; |
1551 | else if (*e == L'b') |
1552 | curr = 2; |
1553 | else |
1554 | curr = 0; |
1555 | |
1556 | if (prev < 0) |
1557 | prev = curr; |
1558 | |
1559 | if (curr != prev) |
1560 | break; |
1561 | |
1562 | t++; |
1563 | if (e && *e) |
1564 | e++; |
1565 | } |
1566 | |
1567 | /* |
1568 | * We've isolated a maximal subsequence of the line |
1569 | * which has the same emphasis. Form it into a word |
1570 | * structure. |
1571 | */ |
1572 | w = mknew(word); |
1573 | w->next = NULL; |
1574 | w->alt = NULL; |
1575 | w->type = (prev == 0 ? word_WeakCode : |
1576 | prev == 1 ? word_Emph : word_Normal); |
1577 | w->text = mknewa(wchar_t, t-start+1); |
1578 | memcpy(w->text, start, (t-start) * sizeof(wchar_t)); |
1579 | w->text[t-start] = '\0'; |
1580 | w->breaks = FALSE; |
1581 | |
1582 | if (ltail) |
1583 | ltail->next = w; |
1584 | else |
1585 | lhead = w; |
1586 | ltail = w; |
1587 | |
1588 | start = t; |
1589 | prev = -1; |
1590 | } |
1591 | |
1592 | ldata = mknew(line_data); |
1593 | |
1594 | ldata->pdata = pdata; |
1595 | ldata->first = lhead; |
1596 | ldata->end = NULL; |
be76d597 |
1597 | ldata->line_height = conf->base_font_size * 4096; |
515d216b |
1598 | |
1599 | ldata->xpos = indent; |
1600 | |
1601 | if (pdata->last) { |
1602 | pdata->last->next = ldata; |
1603 | ldata->prev = pdata->last; |
1604 | } else { |
1605 | pdata->first = ldata; |
1606 | ldata->prev = NULL; |
1607 | } |
1608 | ldata->next = NULL; |
1609 | pdata->last = ldata; |
1610 | |
1611 | ldata->hshortfall = 0; |
1612 | ldata->nspaces = 0; |
1613 | ldata->aux_text = NULL; |
1614 | ldata->aux_text_2 = NULL; |
1615 | ldata->aux_left_indent = 0; |
39a0cfb9 |
1616 | /* General opprobrium for breaking in a code paragraph. */ |
1617 | ldata->penalty_before = ldata->penalty_after = 50000; |
515d216b |
1618 | } |
be76d597 |
1619 | |
1620 | standard_line_spacing(pdata, conf); |
1621 | |
1622 | return pdata; |
515d216b |
1623 | } |
87bd6353 |
1624 | |
be76d597 |
1625 | static para_data *rule_paragraph(int indent, paper_conf *conf) |
87bd6353 |
1626 | { |
be76d597 |
1627 | para_data *pdata = mknew(para_data); |
87bd6353 |
1628 | line_data *ldata; |
1629 | |
1630 | ldata = mknew(line_data); |
1631 | |
1632 | ldata->pdata = pdata; |
1633 | ldata->first = NULL; |
1634 | ldata->end = NULL; |
be76d597 |
1635 | ldata->line_height = conf->rule_thickness; |
87bd6353 |
1636 | |
1637 | ldata->xpos = indent; |
1638 | |
1639 | ldata->prev = NULL; |
1640 | ldata->next = NULL; |
1641 | |
1642 | ldata->hshortfall = 0; |
1643 | ldata->nspaces = 0; |
1644 | ldata->aux_text = NULL; |
1645 | ldata->aux_text_2 = NULL; |
1646 | ldata->aux_left_indent = 0; |
1647 | |
1648 | /* |
1649 | * Better to break after a rule than before it |
1650 | */ |
1651 | ldata->penalty_after += 100000; |
1652 | ldata->penalty_before += -100000; |
1653 | |
1654 | pdata->first = pdata->last = ldata; |
be76d597 |
1655 | pdata->outline_level = -1; |
1656 | pdata->rect_type = RECT_RULE; |
2bfd1b76 |
1657 | pdata->contents_entry = NULL; |
be76d597 |
1658 | |
1659 | standard_line_spacing(pdata, conf); |
1660 | |
1661 | return pdata; |
1662 | } |
1663 | |
1664 | /* |
1665 | * Plain-text-like formatting for outline titles. |
1666 | */ |
1667 | static void paper_rdaddw(rdstring *rs, word *text) { |
1668 | for (; text; text = text->next) switch (text->type) { |
1669 | case word_HyperLink: |
1670 | case word_HyperEnd: |
1671 | case word_UpperXref: |
1672 | case word_LowerXref: |
1673 | case word_XrefEnd: |
1674 | case word_IndexRef: |
1675 | break; |
1676 | |
1677 | case word_Normal: |
1678 | case word_Emph: |
1679 | case word_Code: |
1680 | case word_WeakCode: |
1681 | case word_WhiteSpace: |
1682 | case word_EmphSpace: |
1683 | case word_CodeSpace: |
1684 | case word_WkCodeSpace: |
1685 | case word_Quote: |
1686 | case word_EmphQuote: |
1687 | case word_CodeQuote: |
1688 | case word_WkCodeQuote: |
1689 | assert(text->type != word_CodeQuote && |
1690 | text->type != word_WkCodeQuote); |
1691 | if (towordstyle(text->type) == word_Emph && |
1692 | (attraux(text->aux) == attr_First || |
1693 | attraux(text->aux) == attr_Only)) |
1694 | rdadd(rs, L'_'); /* FIXME: configurability */ |
1695 | else if (towordstyle(text->type) == word_Code && |
1696 | (attraux(text->aux) == attr_First || |
1697 | attraux(text->aux) == attr_Only)) |
1698 | rdadd(rs, L'\''); /* FIXME: configurability */ |
1699 | if (removeattr(text->type) == word_Normal) { |
1700 | rdadds(rs, text->text); |
1701 | } else if (removeattr(text->type) == word_WhiteSpace) { |
1702 | rdadd(rs, L' '); |
1703 | } else if (removeattr(text->type) == word_Quote) { |
1704 | rdadd(rs, L'\''); /* fixme: configurability */ |
1705 | } |
1706 | if (towordstyle(text->type) == word_Emph && |
1707 | (attraux(text->aux) == attr_Last || |
1708 | attraux(text->aux) == attr_Only)) |
1709 | rdadd(rs, L'_'); /* FIXME: configurability */ |
1710 | else if (towordstyle(text->type) == word_Code && |
1711 | (attraux(text->aux) == attr_Last || |
1712 | attraux(text->aux) == attr_Only)) |
1713 | rdadd(rs, L'\''); /* FIXME: configurability */ |
1714 | break; |
1715 | } |
1716 | } |
1717 | |
1718 | static wchar_t *prepare_outline_title(word *first, wchar_t *separator, |
1719 | word *second) |
1720 | { |
1721 | rdstring rs = {0, 0, NULL}; |
1722 | |
1723 | if (first) |
1724 | paper_rdaddw(&rs, first); |
1725 | if (separator) |
1726 | rdadds(&rs, separator); |
1727 | if (second) |
1728 | paper_rdaddw(&rs, second); |
1729 | |
1730 | return rs.text; |
87bd6353 |
1731 | } |
2bfd1b76 |
1732 | |
1733 | static word *fake_word(wchar_t *text) |
1734 | { |
1735 | word *ret = mknew(word); |
1736 | ret->next = NULL; |
1737 | ret->alt = NULL; |
1738 | ret->type = word_Normal; |
1739 | ret->text = ustrdup(text); |
1740 | ret->breaks = FALSE; |
1741 | ret->aux = 0; |
1742 | return ret; |
1743 | } |
1744 | |
1745 | static word *prepare_contents_title(word *first, wchar_t *separator, |
1746 | word *second) |
1747 | { |
1748 | word *ret; |
1749 | word **wptr, *w; |
1750 | |
1751 | wptr = &ret; |
1752 | |
1753 | if (first) { |
1754 | w = dup_word_list(first); |
1755 | *wptr = w; |
1756 | while (w->next) |
1757 | w = w->next; |
1758 | wptr = &w->next; |
1759 | } |
1760 | |
1761 | if (separator) { |
1762 | w = fake_word(separator); |
1763 | *wptr = w; |
1764 | wptr = &w->next; |
1765 | } |
1766 | |
1767 | if (second) { |
1768 | *wptr = dup_word_list(second); |
1769 | } |
1770 | |
1771 | return ret; |
1772 | } |