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 | * - index |
16 | * |
a8070b46 |
17 | * - header/footer? Page numbers at least would be handy. Fully |
18 | * configurable footer can wait, though. |
19 | * |
43341922 |
20 | * That should bring us to the same level of functionality that |
21 | * original-Halibut had, and the same in PDF plus the obvious |
22 | * interactive navigation features. After that, in future work: |
23 | * |
24 | * - linearised PDF, perhaps? |
25 | * |
26 | * - I'm uncertain of whether I need to include a ToUnicode CMap |
27 | * in each of my font definitions in PDF. Currently things (by |
28 | * which I mean cut and paste out of acroread) seem to be |
29 | * working fairly happily without it, but I don't know. |
30 | * |
31 | * - configurability |
32 | * |
33 | * - title pages |
34 | */ |
35 | |
36 | #include <assert.h> |
37 | #include <stdio.h> |
38 | |
39 | #include "halibut.h" |
40 | #include "paper.h" |
41 | |
be76d597 |
42 | typedef struct paper_conf_Tag paper_conf; |
c6536773 |
43 | typedef struct paper_idx_Tag paper_idx; |
be76d597 |
44 | |
45 | struct paper_conf_Tag { |
46 | int paper_width; |
47 | int paper_height; |
48 | int left_margin; |
49 | int top_margin; |
50 | int right_margin; |
51 | int bottom_margin; |
52 | int indent_list_bullet; |
53 | int indent_list; |
54 | int indent_quote; |
55 | int base_leading; |
56 | int base_para_spacing; |
57 | int chapter_top_space; |
58 | int sect_num_left_space; |
59 | int chapter_underline_depth; |
60 | int chapter_underline_thickness; |
61 | int rule_thickness; |
62 | int base_font_size; |
2bfd1b76 |
63 | int contents_indent_step; |
64 | int contents_margin; |
65 | int leader_separation; |
c6536773 |
66 | int index_gutter; |
67 | int index_cols; |
68 | int index_minsep; |
be76d597 |
69 | /* These are derived from the above */ |
70 | int base_width; |
71 | int page_height; |
c6536773 |
72 | int index_colwidth; |
be76d597 |
73 | /* Fonts used in the configuration */ |
74 | font_data *tr, *ti, *hr, *hi, *cr, *co, *cb; |
75 | }; |
76 | |
c6536773 |
77 | struct paper_idx_Tag { |
78 | /* |
79 | * Word list giving the page numbers on which this index entry |
80 | * appears. Also the last word in the list, for ease of |
81 | * construction. |
82 | */ |
83 | word *words; |
84 | word *lastword; |
85 | /* |
86 | * The last page added to the list (so we can ensure we don't |
87 | * add one twice). |
88 | */ |
89 | page_data *lastpage; |
90 | }; |
91 | |
43341922 |
92 | static font_data *make_std_font(font_list *fontlist, char const *name); |
93 | static void wrap_paragraph(para_data *pdata, word *words, |
94 | int w, int i1, int i2); |
95 | static page_data *page_breaks(line_data *first, line_data *last, |
c6536773 |
96 | int page_height, int ncols, int headspace); |
2bfd1b76 |
97 | static int render_string(page_data *page, font_data *font, int fontsize, |
98 | int x, int y, wchar_t *str); |
99 | static int render_line(line_data *ldata, int left_x, int top_y, |
c6536773 |
100 | xref_dest *dest, keywordlist *keywords, indexdata *idx); |
101 | static void render_para(para_data *pdata, paper_conf *conf, |
102 | keywordlist *keywords, indexdata *idx, |
103 | paragraph *index_placeholder, page_data *index_page); |
515d216b |
104 | static int paper_width_simple(para_data *pdata, word *text); |
be76d597 |
105 | static para_data *code_paragraph(int indent, word *words, paper_conf *conf); |
106 | static para_data *rule_paragraph(int indent, paper_conf *conf); |
23765aeb |
107 | static void add_rect_to_page(page_data *page, int x, int y, int w, int h); |
2bfd1b76 |
108 | static para_data *make_para_data(int ptype, int paux, int indent, int rmargin, |
be76d597 |
109 | word *pkwtext, word *pkwtext2, word *pwords, |
110 | paper_conf *conf); |
111 | static void standard_line_spacing(para_data *pdata, paper_conf *conf); |
112 | static wchar_t *prepare_outline_title(word *first, wchar_t *separator, |
113 | word *second); |
2bfd1b76 |
114 | static word *fake_word(wchar_t *text); |
c6536773 |
115 | static word *fake_space_word(void); |
2bfd1b76 |
116 | static word *prepare_contents_title(word *first, wchar_t *separator, |
117 | word *second); |
c6536773 |
118 | static void fold_into_page(page_data *dest, page_data *src, int right_shift); |
43341922 |
119 | |
120 | void *paper_pre_backend(paragraph *sourceform, keywordlist *keywords, |
121 | indexdata *idx) { |
122 | paragraph *p; |
123 | document *doc; |
2bfd1b76 |
124 | int indent, used_contents; |
be76d597 |
125 | para_data *pdata, *firstpara = NULL, *lastpara = NULL; |
2bfd1b76 |
126 | para_data *firstcont, *lastcont; |
c6536773 |
127 | line_data *firstline, *lastline, *firstcontline, *lastcontline; |
43341922 |
128 | page_data *pages; |
129 | font_list *fontlist; |
be76d597 |
130 | paper_conf *conf; |
c6536773 |
131 | int has_index; |
132 | int pagenum; |
133 | paragraph index_placeholder_para; |
134 | page_data *first_index_page; |
43341922 |
135 | |
136 | /* |
137 | * FIXME: All these things ought to become configurable. |
138 | */ |
be76d597 |
139 | conf = mknew(paper_conf); |
140 | conf->paper_width = 595 * 4096; |
141 | conf->paper_height = 841 * 4096; |
142 | conf->left_margin = 72 * 4096; |
143 | conf->top_margin = 72 * 4096; |
144 | conf->right_margin = 72 * 4096; |
145 | conf->bottom_margin = 108 * 4096; |
146 | conf->indent_list_bullet = 6 * 4096; |
147 | conf->indent_list = 24 * 4096; |
148 | conf->indent_quote = 18 * 4096; |
149 | conf->base_leading = 4096; |
150 | conf->base_para_spacing = 10 * 4096; |
151 | conf->chapter_top_space = 72 * 4096; |
152 | conf->sect_num_left_space = 12 * 4096; |
153 | conf->chapter_underline_depth = 14 * 4096; |
154 | conf->chapter_underline_thickness = 3 * 4096; |
155 | conf->rule_thickness = 1 * 4096; |
156 | conf->base_font_size = 12; |
2bfd1b76 |
157 | conf->contents_indent_step = 24 * 4096; |
158 | conf->contents_margin = 84 * 4096; |
159 | conf->leader_separation = 12 * 4096; |
c6536773 |
160 | conf->index_gutter = 36 * 4096; |
161 | conf->index_cols = 2; |
162 | conf->index_minsep = 18 * 4096; |
be76d597 |
163 | |
164 | conf->base_width = |
165 | conf->paper_width - conf->left_margin - conf->right_margin; |
166 | conf->page_height = |
167 | conf->paper_height - conf->top_margin - conf->bottom_margin; |
c6536773 |
168 | conf->index_colwidth = |
169 | (conf->base_width - (conf->index_cols-1) * conf->index_gutter) |
170 | / conf->index_cols; |
43341922 |
171 | |
43341922 |
172 | IGNORE(idx); /* FIXME */ |
43341922 |
173 | |
174 | /* |
175 | * First, set up some font structures. |
176 | */ |
177 | fontlist = mknew(font_list); |
178 | fontlist->head = fontlist->tail = NULL; |
be76d597 |
179 | conf->tr = make_std_font(fontlist, "Times-Roman"); |
180 | conf->ti = make_std_font(fontlist, "Times-Italic"); |
181 | conf->hr = make_std_font(fontlist, "Helvetica-Bold"); |
182 | conf->hi = make_std_font(fontlist, "Helvetica-BoldOblique"); |
183 | conf->cr = make_std_font(fontlist, "Courier"); |
184 | conf->co = make_std_font(fontlist, "Courier-Oblique"); |
185 | conf->cb = make_std_font(fontlist, "Courier-Bold"); |
43341922 |
186 | |
187 | /* |
c6536773 |
188 | * Set up a data structure to collect page numbers for each |
189 | * index entry. |
190 | */ |
191 | { |
192 | int i; |
193 | indexentry *entry; |
194 | |
195 | has_index = FALSE; |
196 | |
197 | for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++) { |
198 | paper_idx *pi = mknew(paper_idx); |
199 | |
200 | has_index = TRUE; |
201 | |
202 | pi->words = pi->lastword = NULL; |
203 | pi->lastpage = NULL; |
204 | |
205 | entry->backend_data = pi; |
206 | } |
207 | } |
208 | |
209 | /* |
2bfd1b76 |
210 | * Format the contents entry for each heading. |
211 | */ |
212 | { |
213 | word *contents_title; |
214 | contents_title = fake_word(L"Contents"); |
215 | |
216 | firstcont = make_para_data(para_UnnumberedChapter, 0, 0, 0, |
217 | NULL, NULL, contents_title, conf); |
218 | lastcont = firstcont; |
219 | lastcont->next = NULL; |
220 | firstcontline = firstcont->first; |
221 | lastcontline = lastcont->last; |
222 | for (p = sourceform; p; p = p->next) { |
223 | word *words; |
224 | int indent; |
225 | |
226 | switch (p->type) { |
227 | case para_Chapter: |
228 | case para_Appendix: |
229 | case para_UnnumberedChapter: |
230 | case para_Heading: |
231 | case para_Subsect: |
232 | switch (p->type) { |
233 | case para_Chapter: |
234 | case para_Appendix: |
235 | words = prepare_contents_title(p->kwtext, L": ", p->words); |
236 | indent = 0; |
237 | break; |
238 | case para_UnnumberedChapter: |
239 | words = prepare_contents_title(NULL, NULL, p->words); |
240 | indent = 0; |
241 | break; |
242 | case para_Heading: |
243 | case para_Subsect: |
244 | words = prepare_contents_title(p->kwtext2, L" ", p->words); |
245 | indent = (p->aux + 1) * conf->contents_indent_step; |
246 | break; |
247 | } |
248 | pdata = make_para_data(para_Normal, p->aux, indent, |
249 | conf->contents_margin, |
250 | NULL, NULL, words, conf); |
251 | pdata->next = NULL; |
252 | pdata->contents_entry = p; |
253 | lastcont->next = pdata; |
254 | lastcont = pdata; |
255 | |
256 | /* |
257 | * Link all contents line structures together into |
258 | * a big list. |
259 | */ |
260 | if (pdata->first) { |
261 | if (lastcontline) { |
262 | lastcontline->next = pdata->first; |
263 | pdata->first->prev = lastcontline; |
264 | } else { |
265 | firstcontline = pdata->first; |
266 | pdata->first->prev = NULL; |
267 | } |
268 | lastcontline = pdata->last; |
269 | lastcontline->next = NULL; |
270 | } |
271 | |
272 | break; |
273 | } |
274 | } |
c6536773 |
275 | |
276 | /* |
277 | * And one extra one, for the index. |
278 | */ |
279 | if (has_index) { |
280 | pdata = make_para_data(para_Normal, 0, 0, |
281 | conf->contents_margin, |
282 | NULL, NULL, fake_word(L"Index"), conf); |
283 | pdata->next = NULL; |
284 | pdata->contents_entry = &index_placeholder_para; |
285 | lastcont->next = pdata; |
286 | lastcont = pdata; |
287 | |
288 | if (pdata->first) { |
289 | if (lastcontline) { |
290 | lastcontline->next = pdata->first; |
291 | pdata->first->prev = lastcontline; |
292 | } else { |
293 | firstcontline = pdata->first; |
294 | pdata->first->prev = NULL; |
295 | } |
296 | lastcontline = pdata->last; |
297 | lastcontline->next = NULL; |
298 | } |
299 | } |
2bfd1b76 |
300 | } |
301 | |
302 | /* |
303 | * Do the main paragraph formatting. |
43341922 |
304 | */ |
305 | indent = 0; |
2bfd1b76 |
306 | used_contents = FALSE; |
43341922 |
307 | firstline = lastline = NULL; |
308 | for (p = sourceform; p; p = p->next) { |
309 | p->private_data = NULL; |
310 | |
311 | switch (p->type) { |
312 | /* |
313 | * These paragraph types are either invisible or don't |
314 | * define text in the normal sense. Either way, they |
315 | * don't require wrapping. |
316 | */ |
317 | case para_IM: |
318 | case para_BR: |
43341922 |
319 | case para_Biblio: |
320 | case para_NotParaType: |
321 | case para_Config: |
322 | case para_VersionID: |
323 | case para_NoCite: |
324 | break; |
325 | |
326 | /* |
327 | * These paragraph types don't require wrapping, but |
328 | * they do affect the line width to which we wrap the |
329 | * rest of the paragraphs, so we need to pay attention. |
330 | */ |
331 | case para_LcontPush: |
be76d597 |
332 | indent += conf->indent_list; break; |
43341922 |
333 | case para_LcontPop: |
be76d597 |
334 | indent -= conf->indent_list; assert(indent >= 0); break; |
43341922 |
335 | case para_QuotePush: |
be76d597 |
336 | indent += conf->indent_quote; break; |
43341922 |
337 | case para_QuotePop: |
be76d597 |
338 | indent -= conf->indent_quote; assert(indent >= 0); break; |
43341922 |
339 | |
340 | /* |
341 | * This paragraph type is special. Process it |
342 | * specially. |
343 | */ |
344 | case para_Code: |
be76d597 |
345 | pdata = code_paragraph(indent, p->words, conf); |
515d216b |
346 | p->private_data = pdata; |
39a0cfb9 |
347 | if (pdata->first != pdata->last) { |
348 | pdata->first->penalty_after += 100000; |
349 | pdata->last->penalty_before += 100000; |
350 | } |
43341922 |
351 | break; |
352 | |
353 | /* |
87bd6353 |
354 | * This paragraph is also special. |
355 | */ |
356 | case para_Rule: |
be76d597 |
357 | pdata = rule_paragraph(indent, conf); |
87bd6353 |
358 | p->private_data = pdata; |
359 | break; |
360 | |
361 | /* |
43341922 |
362 | * All of these paragraph types require wrapping in the |
363 | * ordinary way. So we must supply a set of fonts, a |
364 | * line width and auxiliary information (e.g. bullet |
365 | * text) for each one. |
366 | */ |
367 | case para_Chapter: |
368 | case para_Appendix: |
369 | case para_UnnumberedChapter: |
370 | case para_Heading: |
371 | case para_Subsect: |
372 | case para_Normal: |
373 | case para_BiblioCited: |
374 | case para_Bullet: |
375 | case para_NumberedList: |
376 | case para_DescribedThing: |
377 | case para_Description: |
378 | case para_Copyright: |
379 | case para_Title: |
2bfd1b76 |
380 | pdata = make_para_data(p->type, p->aux, indent, 0, |
be76d597 |
381 | p->kwtext, p->kwtext2, p->words, conf); |
43341922 |
382 | |
43341922 |
383 | p->private_data = pdata; |
384 | |
515d216b |
385 | break; |
386 | } |
387 | |
388 | if (p->private_data) { |
389 | pdata = (para_data *)p->private_data; |
390 | |
43341922 |
391 | /* |
2bfd1b76 |
392 | * If this is the first non-title heading, we link the |
393 | * contents section in before it. |
394 | */ |
395 | if (!used_contents && pdata->outline_level > 0) { |
396 | used_contents = TRUE; |
397 | if (lastpara) |
398 | lastpara->next = firstcont; |
399 | else |
400 | firstpara = firstcont; |
401 | lastpara = lastcont; |
402 | assert(lastpara->next == NULL); |
403 | |
404 | if (lastline) { |
405 | lastline->next = firstcontline; |
406 | firstcontline->prev = lastline; |
407 | } else { |
408 | firstline = firstcontline; |
409 | firstcontline->prev = NULL; |
410 | } |
411 | assert(lastcontline != NULL); |
412 | lastline = lastcontline; |
413 | lastline->next = NULL; |
414 | } |
415 | |
416 | /* |
515d216b |
417 | * Link all line structures together into a big list. |
418 | */ |
43341922 |
419 | if (pdata->first) { |
420 | if (lastline) { |
421 | lastline->next = pdata->first; |
422 | pdata->first->prev = lastline; |
423 | } else { |
424 | firstline = pdata->first; |
425 | pdata->first->prev = NULL; |
426 | } |
427 | lastline = pdata->last; |
2bfd1b76 |
428 | lastline->next = NULL; |
43341922 |
429 | } |
be76d597 |
430 | |
431 | /* |
432 | * Link all paragraph structures together similarly. |
433 | */ |
434 | pdata->next = NULL; |
435 | if (lastpara) |
436 | lastpara->next = pdata; |
437 | else |
438 | firstpara = pdata; |
439 | lastpara = pdata; |
43341922 |
440 | } |
441 | } |
442 | |
443 | /* |
444 | * Now we have an enormous linked list of every line of text in |
445 | * the document. Break it up into pages. |
446 | */ |
c6536773 |
447 | pages = page_breaks(firstline, lastline, conf->page_height, 0, 0); |
43341922 |
448 | |
449 | /* |
2bfd1b76 |
450 | * Number the pages. |
451 | */ |
452 | { |
c6536773 |
453 | char buf[40]; |
2bfd1b76 |
454 | page_data *page; |
c6536773 |
455 | |
456 | pagenum = 0; |
457 | |
2bfd1b76 |
458 | for (page = pages; page; page = page->next) { |
c6536773 |
459 | sprintf(buf, "%d", ++pagenum); |
2bfd1b76 |
460 | page->number = ufroma_dup(buf); |
461 | } |
c6536773 |
462 | |
463 | if (has_index) { |
464 | first_index_page = mknew(page_data); |
465 | first_index_page->next = first_index_page->prev = NULL; |
466 | first_index_page->first_line = NULL; |
467 | first_index_page->last_line = NULL; |
468 | first_index_page->first_text = first_index_page->last_text = NULL; |
469 | first_index_page->first_xref = first_index_page->last_xref = NULL; |
470 | first_index_page->first_rect = first_index_page->last_rect = NULL; |
471 | |
472 | /* And don't forget the as-yet-uncreated index. */ |
473 | sprintf(buf, "%d", ++pagenum); |
474 | first_index_page->number = ufroma_dup(buf); |
475 | } |
2bfd1b76 |
476 | } |
477 | |
478 | /* |
43341922 |
479 | * Now we're ready to actually lay out the pages. We do this by |
480 | * looping over _paragraphs_, since we may need to track cross- |
481 | * references between lines and even across pages. |
482 | */ |
c6536773 |
483 | for (pdata = firstpara; pdata; pdata = pdata->next) |
484 | render_para(pdata, conf, keywords, idx, |
485 | &index_placeholder_para, first_index_page); |
486 | |
487 | /* |
488 | * Now we've laid out the main body pages, we should have |
489 | * acquired a full set of page numbers for the index. |
490 | */ |
491 | if (has_index) { |
492 | int i; |
493 | indexentry *entry; |
494 | word *index_title; |
495 | para_data *firstidx, *lastidx; |
496 | line_data *firstidxline, *lastidxline, *ldata; |
497 | page_data *ipages, *ipages2, *page; |
a0768d17 |
498 | |
c6536773 |
499 | /* |
500 | * Create a set of paragraphs for the index. |
501 | */ |
502 | index_title = fake_word(L"Index"); |
503 | |
504 | firstidx = make_para_data(para_UnnumberedChapter, 0, 0, 0, |
505 | NULL, NULL, index_title, conf); |
506 | lastidx = firstidx; |
507 | lastidx->next = NULL; |
508 | firstidxline = firstidx->first; |
509 | lastidxline = lastidx->last; |
510 | for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++) { |
511 | paper_idx *pi = (paper_idx *)entry->backend_data; |
512 | para_data *text, *pages; |
513 | |
514 | text = make_para_data(para_Normal, 0, 0, |
515 | conf->base_width - conf->index_colwidth, |
516 | NULL, NULL, entry->text, conf); |
517 | |
518 | pages = make_para_data(para_Normal, 0, 0, |
519 | conf->base_width - conf->index_colwidth, |
520 | NULL, NULL, pi->words, conf); |
521 | |
522 | text->justification = LEFT; |
523 | pages->justification = RIGHT; |
524 | text->last->space_after = pages->first->space_before = |
525 | conf->base_leading / 2; |
526 | |
527 | pages->last->space_after = text->first->space_before = |
528 | conf->base_leading; |
529 | |
530 | assert(text->first); |
531 | assert(pages->first); |
532 | assert(lastidxline); |
533 | assert(lastidx); |
a0768d17 |
534 | |
a0768d17 |
535 | /* |
c6536773 |
536 | * If feasible, fold the two halves of the index entry |
537 | * together. |
a0768d17 |
538 | */ |
c6536773 |
539 | if (text->last->real_shortfall + pages->first->real_shortfall > |
540 | conf->index_colwidth + conf->index_minsep) { |
541 | text->last->space_after = -1; |
542 | pages->first->space_before = -pages->first->line_height+1; |
a0768d17 |
543 | } |
544 | |
c6536773 |
545 | lastidx->next = text; |
546 | text->next = pages; |
547 | pages->next = NULL; |
548 | lastidx = pages; |
549 | |
550 | /* |
551 | * Link all index line structures together into |
552 | * a big list. |
553 | */ |
554 | text->last->next = pages->first; |
555 | pages->first->prev = text->last; |
556 | |
557 | lastidxline->next = text->first; |
558 | text->first->prev = lastidxline; |
559 | |
560 | lastidxline = pages->last; |
561 | |
562 | /* |
563 | * Breaking an index entry anywhere is so bad that I |
564 | * think I'm going to forbid it totally. |
565 | */ |
566 | for (ldata = text->first; ldata && ldata->next; |
567 | ldata = ldata->next) { |
568 | ldata->next->space_before += ldata->space_after + 1; |
569 | ldata->space_after = -1; |
570 | } |
be76d597 |
571 | } |
87bd6353 |
572 | |
be76d597 |
573 | /* |
c6536773 |
574 | * Now break the index into pages. |
2bfd1b76 |
575 | */ |
c6536773 |
576 | ipages = page_breaks(firstidxline, firstidxline, conf->page_height, |
577 | 0, 0); |
578 | ipages2 = page_breaks(firstidxline->next, lastidxline, |
579 | conf->page_height, |
580 | conf->index_cols, |
581 | firstidxline->space_before + |
582 | firstidxline->line_height + |
583 | firstidxline->space_after); |
2bfd1b76 |
584 | |
c6536773 |
585 | /* |
586 | * This will have put each _column_ of the index on a |
587 | * separate page, which isn't what we want. Fold the pages |
588 | * back together. |
589 | */ |
590 | page = ipages2; |
591 | while (page) { |
592 | int i; |
593 | |
594 | for (i = 1; i < conf->index_cols; i++) |
595 | if (page->next) { |
596 | page_data *tpage; |
597 | |
598 | fold_into_page(page, page->next, |
599 | i * (conf->index_colwidth + |
600 | conf->index_gutter)); |
601 | tpage = page->next; |
602 | page->next = page->next->next; |
603 | if (page->next) |
604 | page->next->prev = page; |
605 | sfree(tpage); |
606 | } |
2bfd1b76 |
607 | |
c6536773 |
608 | page = page->next; |
2bfd1b76 |
609 | } |
c6536773 |
610 | /* Also fold the heading on to the same page as the index items. */ |
611 | fold_into_page(ipages, ipages2, 0); |
612 | ipages->next = ipages2->next; |
613 | if (ipages->next) |
614 | ipages->next->prev = ipages; |
615 | sfree(ipages2); |
616 | fold_into_page(first_index_page, ipages, 0); |
617 | first_index_page->next = ipages->next; |
618 | if (first_index_page->next) |
619 | first_index_page->next->prev = first_index_page; |
620 | sfree(ipages); |
621 | ipages = first_index_page; |
2bfd1b76 |
622 | |
623 | /* |
c6536773 |
624 | * Number the index pages, except the already-numbered |
625 | * first one. |
be76d597 |
626 | */ |
c6536773 |
627 | for (page = ipages->next; page; page = page->next) { |
628 | char buf[40]; |
629 | sprintf(buf, "%d", ++pagenum); |
630 | page->number = ufroma_dup(buf); |
43341922 |
631 | } |
c6536773 |
632 | |
633 | /* |
634 | * Render the index pages. |
635 | */ |
636 | for (pdata = firstidx; pdata; pdata = pdata->next) |
637 | render_para(pdata, conf, keywords, idx, |
638 | &index_placeholder_para, first_index_page); |
639 | |
640 | /* |
641 | * Link the index page list on to the end of the main page |
642 | * list. |
643 | */ |
644 | if (!pages) |
645 | pages = ipages; |
646 | else { |
647 | for (page = pages; page->next; page = page->next); |
648 | page->next = ipages; |
649 | } |
650 | |
651 | /* |
652 | * Same with the paragraph list, which will cause the index |
653 | * to be mentioned in the document outline. |
654 | */ |
655 | if (!firstpara) |
656 | firstpara = firstidx; |
657 | else |
658 | lastpara->next = firstidx; |
659 | lastpara = lastidx; |
43341922 |
660 | } |
661 | |
f0e51ce1 |
662 | /* |
663 | * Start putting together the overall document structure we're |
664 | * going to return. |
665 | */ |
43341922 |
666 | doc = mknew(document); |
667 | doc->fonts = fontlist; |
668 | doc->pages = pages; |
be76d597 |
669 | doc->paper_width = conf->paper_width; |
670 | doc->paper_height = conf->paper_height; |
f0e51ce1 |
671 | |
672 | /* |
673 | * Collect the section heading paragraphs into a document |
674 | * outline. This is slightly fiddly because the Title paragraph |
675 | * isn't required to be at the start, although all the others |
676 | * must be in order. |
677 | */ |
678 | { |
679 | int osize = 20; |
680 | |
681 | doc->outline_elements = mknewa(outline_element, osize); |
682 | doc->n_outline_elements = 0; |
683 | |
684 | /* First find the title. */ |
be76d597 |
685 | for (pdata = firstpara; pdata; pdata = pdata->next) { |
686 | if (pdata->outline_level == 0) { |
f0e51ce1 |
687 | doc->outline_elements[0].level = 0; |
be76d597 |
688 | doc->outline_elements[0].pdata = pdata; |
f0e51ce1 |
689 | doc->n_outline_elements++; |
690 | break; |
691 | } |
692 | } |
693 | |
694 | /* Then collect the rest. */ |
be76d597 |
695 | for (pdata = firstpara; pdata; pdata = pdata->next) { |
696 | if (pdata->outline_level > 0) { |
f0e51ce1 |
697 | if (doc->n_outline_elements >= osize) { |
698 | osize += 20; |
699 | doc->outline_elements = |
700 | resize(doc->outline_elements, osize); |
701 | } |
702 | |
be76d597 |
703 | doc->outline_elements[doc->n_outline_elements].level = |
704 | pdata->outline_level; |
705 | doc->outline_elements[doc->n_outline_elements].pdata = pdata; |
f0e51ce1 |
706 | doc->n_outline_elements++; |
f0e51ce1 |
707 | } |
708 | } |
709 | } |
710 | |
be76d597 |
711 | sfree(conf); |
712 | |
43341922 |
713 | return doc; |
714 | } |
715 | |
2bfd1b76 |
716 | static para_data *make_para_data(int ptype, int paux, int indent, int rmargin, |
be76d597 |
717 | word *pkwtext, word *pkwtext2, word *pwords, |
718 | paper_conf *conf) |
719 | { |
720 | para_data *pdata; |
721 | line_data *ldata; |
722 | int extra_indent, firstline_indent, aux_indent; |
723 | word *aux, *aux2; |
724 | |
725 | pdata = mknew(para_data); |
726 | pdata->outline_level = -1; |
727 | pdata->outline_title = NULL; |
728 | pdata->rect_type = RECT_NONE; |
2bfd1b76 |
729 | pdata->contents_entry = NULL; |
c6536773 |
730 | pdata->justification = JUST; |
be76d597 |
731 | |
732 | /* |
733 | * Choose fonts for this paragraph. |
734 | * |
735 | * FIXME: All of this ought to be completely |
736 | * user-configurable. |
737 | */ |
738 | switch (ptype) { |
739 | case para_Title: |
740 | pdata->fonts[FONT_NORMAL] = conf->hr; |
741 | pdata->sizes[FONT_NORMAL] = 24; |
742 | pdata->fonts[FONT_EMPH] = conf->hi; |
743 | pdata->sizes[FONT_EMPH] = 24; |
744 | pdata->fonts[FONT_CODE] = conf->cb; |
745 | pdata->sizes[FONT_CODE] = 24; |
746 | pdata->outline_level = 0; |
747 | break; |
748 | |
749 | case para_Chapter: |
750 | case para_Appendix: |
751 | case para_UnnumberedChapter: |
752 | pdata->fonts[FONT_NORMAL] = conf->hr; |
753 | pdata->sizes[FONT_NORMAL] = 20; |
754 | pdata->fonts[FONT_EMPH] = conf->hi; |
755 | pdata->sizes[FONT_EMPH] = 20; |
756 | pdata->fonts[FONT_CODE] = conf->cb; |
757 | pdata->sizes[FONT_CODE] = 20; |
758 | pdata->outline_level = 1; |
759 | break; |
760 | |
761 | case para_Heading: |
762 | case para_Subsect: |
763 | pdata->fonts[FONT_NORMAL] = conf->hr; |
764 | pdata->fonts[FONT_EMPH] = conf->hi; |
765 | pdata->fonts[FONT_CODE] = conf->cb; |
766 | pdata->sizes[FONT_NORMAL] = |
767 | pdata->sizes[FONT_EMPH] = |
768 | pdata->sizes[FONT_CODE] = |
769 | (paux == 0 ? 16 : paux == 1 ? 14 : 13); |
770 | pdata->outline_level = 2 + paux; |
771 | break; |
772 | |
773 | case para_Normal: |
774 | case para_BiblioCited: |
775 | case para_Bullet: |
776 | case para_NumberedList: |
777 | case para_DescribedThing: |
778 | case para_Description: |
779 | case para_Copyright: |
780 | pdata->fonts[FONT_NORMAL] = conf->tr; |
781 | pdata->sizes[FONT_NORMAL] = 12; |
782 | pdata->fonts[FONT_EMPH] = conf->ti; |
783 | pdata->sizes[FONT_EMPH] = 12; |
784 | pdata->fonts[FONT_CODE] = conf->cr; |
785 | pdata->sizes[FONT_CODE] = 12; |
786 | break; |
787 | } |
788 | |
789 | /* |
790 | * Also select an indentation level depending on the |
791 | * paragraph type (list paragraphs other than |
792 | * para_DescribedThing need extra indent). |
793 | * |
794 | * (FIXME: Perhaps at some point we might even arrange |
795 | * for the user to be able to request indented first |
796 | * lines in paragraphs.) |
797 | */ |
798 | if (ptype == para_Bullet || |
799 | ptype == para_NumberedList || |
800 | ptype == para_Description) { |
801 | extra_indent = firstline_indent = conf->indent_list; |
802 | } else { |
803 | extra_indent = firstline_indent = 0; |
804 | } |
805 | |
806 | /* |
807 | * Find the auxiliary text for this paragraph. |
808 | */ |
809 | aux = aux2 = NULL; |
810 | aux_indent = 0; |
811 | |
812 | switch (ptype) { |
813 | case para_Chapter: |
814 | case para_Appendix: |
815 | case para_Heading: |
816 | case para_Subsect: |
817 | /* |
818 | * For some heading styles (FIXME: be able to |
819 | * configure which), the auxiliary text contains |
820 | * the chapter number and is arranged to be |
821 | * right-aligned a few points left of the primary |
822 | * margin. For other styles, the auxiliary text is |
823 | * the full chapter _name_ and takes up space |
824 | * within the (wrapped) chapter title, meaning that |
825 | * we must move the first line indent over to make |
826 | * space for it. |
827 | */ |
828 | if (ptype == para_Heading || ptype == para_Subsect) { |
829 | int len; |
830 | |
831 | aux = pkwtext2; |
832 | len = paper_width_simple(pdata, pkwtext2); |
833 | aux_indent = -len - conf->sect_num_left_space; |
834 | |
835 | pdata->outline_title = |
836 | prepare_outline_title(pkwtext2, L" ", pwords); |
837 | } else { |
838 | aux = pkwtext; |
2bfd1b76 |
839 | aux2 = fake_word(L": "); |
be76d597 |
840 | aux_indent = 0; |
841 | |
842 | firstline_indent += paper_width_simple(pdata, aux); |
843 | firstline_indent += paper_width_simple(pdata, aux2); |
844 | |
845 | pdata->outline_title = |
846 | prepare_outline_title(pkwtext, L": ", pwords); |
847 | } |
848 | break; |
849 | |
850 | case para_Bullet: |
851 | /* |
852 | * Auxiliary text consisting of a bullet. (FIXME: |
853 | * configurable bullet.) |
854 | */ |
2bfd1b76 |
855 | aux = fake_word(L"\x2022"); |
be76d597 |
856 | aux_indent = indent + conf->indent_list_bullet; |
857 | break; |
858 | |
859 | case para_NumberedList: |
860 | /* |
861 | * Auxiliary text consisting of the number followed |
862 | * by a (FIXME: configurable) full stop. |
863 | */ |
864 | aux = pkwtext; |
2bfd1b76 |
865 | aux2 = fake_word(L"."); |
be76d597 |
866 | aux_indent = indent + conf->indent_list_bullet; |
867 | break; |
868 | |
869 | case para_BiblioCited: |
870 | /* |
871 | * Auxiliary text consisting of the bibliography |
872 | * reference text, and a trailing space. |
873 | */ |
874 | aux = pkwtext; |
2bfd1b76 |
875 | aux2 = fake_word(L" "); |
be76d597 |
876 | aux_indent = indent; |
877 | firstline_indent += paper_width_simple(pdata, aux); |
878 | firstline_indent += paper_width_simple(pdata, aux2); |
879 | break; |
880 | } |
881 | |
882 | if (pdata->outline_level >= 0 && !pdata->outline_title) { |
883 | pdata->outline_title = |
884 | prepare_outline_title(NULL, NULL, pwords); |
885 | } |
886 | |
2bfd1b76 |
887 | wrap_paragraph(pdata, pwords, conf->base_width - rmargin, |
be76d597 |
888 | indent + firstline_indent, |
889 | indent + extra_indent); |
890 | |
891 | pdata->first->aux_text = aux; |
892 | pdata->first->aux_text_2 = aux2; |
893 | pdata->first->aux_left_indent = aux_indent; |
894 | |
895 | /* |
896 | * Line breaking penalties. |
897 | */ |
898 | switch (ptype) { |
899 | case para_Chapter: |
900 | case para_Appendix: |
901 | case para_Heading: |
902 | case para_Subsect: |
903 | case para_UnnumberedChapter: |
904 | /* |
905 | * Fixed and large penalty for breaking straight |
906 | * after a heading; corresponding bonus for |
907 | * breaking straight before. |
908 | */ |
909 | pdata->first->penalty_before = -500000; |
910 | pdata->last->penalty_after = 500000; |
911 | for (ldata = pdata->first; ldata; ldata = ldata->next) |
912 | ldata->penalty_after = 500000; |
913 | break; |
914 | |
915 | case para_DescribedThing: |
916 | /* |
917 | * This is treated a bit like a small heading: |
918 | * there's a penalty for breaking after it (i.e. |
919 | * between it and its description), and a bonus for |
920 | * breaking before it (actually _between_ list |
921 | * items). |
922 | */ |
923 | pdata->first->penalty_before = -200000; |
924 | pdata->last->penalty_after = 200000; |
925 | break; |
926 | |
927 | default: |
928 | /* |
929 | * Most paragraph types: widow/orphan control by |
930 | * discouraging breaking one line from the end of |
931 | * any paragraph. |
932 | */ |
933 | if (pdata->first != pdata->last) { |
934 | pdata->first->penalty_after = 100000; |
935 | pdata->last->penalty_before = 100000; |
936 | } |
937 | break; |
938 | } |
939 | |
940 | standard_line_spacing(pdata, conf); |
941 | |
942 | /* |
943 | * Some kinds of section heading require a page break before |
944 | * them and an underline after. |
945 | */ |
946 | if (ptype == para_Title || |
947 | ptype == para_Chapter || |
948 | ptype == para_Appendix || |
949 | ptype == para_UnnumberedChapter) { |
950 | pdata->first->page_break = TRUE; |
951 | pdata->first->space_before = conf->chapter_top_space; |
952 | pdata->last->space_after += |
953 | (conf->chapter_underline_depth + |
954 | conf->chapter_underline_thickness); |
955 | pdata->rect_type = RECT_CHAPTER_UNDERLINE; |
956 | } |
957 | |
958 | return pdata; |
959 | } |
960 | |
961 | static void standard_line_spacing(para_data *pdata, paper_conf *conf) |
962 | { |
963 | line_data *ldata; |
964 | |
965 | /* |
966 | * Set the line spacing for each line in this paragraph. |
967 | */ |
968 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
969 | if (ldata == pdata->first) |
970 | ldata->space_before = conf->base_para_spacing / 2; |
971 | else |
972 | ldata->space_before = conf->base_leading / 2; |
973 | if (ldata == pdata->last) |
974 | ldata->space_after = conf->base_para_spacing / 2; |
975 | else |
976 | ldata->space_after = conf->base_leading / 2; |
977 | ldata->page_break = FALSE; |
978 | } |
979 | } |
980 | |
43341922 |
981 | static font_encoding *new_font_encoding(font_data *font) |
982 | { |
983 | font_encoding *fe; |
984 | int i; |
985 | |
986 | fe = mknew(font_encoding); |
987 | fe->next = NULL; |
988 | |
989 | if (font->list->tail) |
990 | font->list->tail->next = fe; |
991 | else |
992 | font->list->head = fe; |
993 | font->list->tail = fe; |
994 | |
995 | fe->font = font; |
996 | fe->free_pos = 0x21; |
997 | |
998 | for (i = 0; i < 256; i++) { |
999 | fe->vector[i] = NULL; |
1000 | fe->indices[i] = -1; |
1001 | fe->to_unicode[i] = 0xFFFF; |
1002 | } |
1003 | |
1004 | return fe; |
1005 | } |
1006 | |
1007 | static font_data *make_std_font(font_list *fontlist, char const *name) |
1008 | { |
1009 | const int *widths; |
1010 | int nglyphs; |
1011 | font_data *f; |
1012 | font_encoding *fe; |
1013 | int i; |
1014 | |
1015 | widths = ps_std_font_widths(name); |
1016 | if (!widths) |
1017 | return NULL; |
1018 | |
1019 | for (nglyphs = 0; ps_std_glyphs[nglyphs] != NULL; nglyphs++); |
1020 | |
1021 | f = mknew(font_data); |
1022 | |
1023 | f->list = fontlist; |
1024 | f->name = name; |
1025 | f->nglyphs = nglyphs; |
1026 | f->glyphs = ps_std_glyphs; |
1027 | f->widths = widths; |
1028 | f->subfont_map = mknewa(subfont_map_entry, nglyphs); |
1029 | |
1030 | /* |
1031 | * Our first subfont will contain all of US-ASCII. This isn't |
1032 | * really necessary - we could just create custom subfonts |
1033 | * precisely as the whim of render_string dictated - but |
1034 | * instinct suggests that it might be nice to have the text in |
1035 | * the output files look _marginally_ recognisable. |
1036 | */ |
1037 | fe = new_font_encoding(f); |
1038 | fe->free_pos = 0xA1; /* only the top half is free */ |
1039 | f->latest_subfont = fe; |
1040 | |
1041 | for (i = 0; i < (int)lenof(f->bmp); i++) |
1042 | f->bmp[i] = 0xFFFF; |
1043 | |
1044 | for (i = 0; i < nglyphs; i++) { |
1045 | wchar_t ucs; |
1046 | ucs = ps_glyph_to_unicode(f->glyphs[i]); |
1047 | assert(ucs != 0xFFFF); |
1048 | f->bmp[ucs] = i; |
1049 | if (ucs >= 0x20 && ucs <= 0x7E) { |
1050 | fe->vector[ucs] = f->glyphs[i]; |
1051 | fe->indices[ucs] = i; |
1052 | fe->to_unicode[ucs] = ucs; |
1053 | f->subfont_map[i].subfont = fe; |
1054 | f->subfont_map[i].position = ucs; |
1055 | } else { |
1056 | /* |
1057 | * This character is not yet assigned to a subfont. |
1058 | */ |
1059 | f->subfont_map[i].subfont = NULL; |
1060 | f->subfont_map[i].position = 0; |
1061 | } |
1062 | } |
1063 | |
1064 | return f; |
1065 | } |
1066 | |
1067 | static int string_width(font_data *font, wchar_t const *string, int *errs) |
1068 | { |
1069 | int width = 0; |
1070 | |
1071 | if (errs) |
1072 | *errs = 0; |
1073 | |
1074 | for (; *string; string++) { |
1075 | int index; |
1076 | |
1077 | index = font->bmp[(unsigned short)*string]; |
1078 | if (index == 0xFFFF) { |
1079 | if (errs) |
1080 | *errs = 1; |
1081 | } else { |
1082 | width += font->widths[index]; |
1083 | } |
1084 | } |
1085 | |
1086 | return width; |
1087 | } |
1088 | |
faad4952 |
1089 | static int paper_width_internal(void *vctx, word *word, int *nspaces); |
43341922 |
1090 | |
1091 | struct paper_width_ctx { |
1092 | int minspacewidth; |
1093 | para_data *pdata; |
1094 | }; |
1095 | |
faad4952 |
1096 | static int paper_width_list(void *vctx, word *text, word *end, int *nspaces) { |
43341922 |
1097 | int w = 0; |
faad4952 |
1098 | while (text && text != end) { |
1099 | w += paper_width_internal(vctx, text, nspaces); |
43341922 |
1100 | text = text->next; |
1101 | } |
1102 | return w; |
1103 | } |
1104 | |
faad4952 |
1105 | static int paper_width_internal(void *vctx, word *word, int *nspaces) |
43341922 |
1106 | { |
1107 | struct paper_width_ctx *ctx = (struct paper_width_ctx *)vctx; |
1108 | int style, type, findex, width, errs; |
1109 | wchar_t *str; |
1110 | |
1111 | switch (word->type) { |
1112 | case word_HyperLink: |
1113 | case word_HyperEnd: |
1114 | case word_UpperXref: |
1115 | case word_LowerXref: |
1116 | case word_XrefEnd: |
1117 | case word_IndexRef: |
1118 | return 0; |
1119 | } |
1120 | |
1121 | style = towordstyle(word->type); |
1122 | type = removeattr(word->type); |
1123 | |
1124 | findex = (style == word_Normal ? FONT_NORMAL : |
1125 | style == word_Emph ? FONT_EMPH : |
1126 | FONT_CODE); |
1127 | |
1128 | if (type == word_Normal) { |
1129 | str = word->text; |
1130 | } else if (type == word_WhiteSpace) { |
faad4952 |
1131 | if (findex != FONT_CODE) { |
1132 | if (nspaces) |
1133 | (*nspaces)++; |
43341922 |
1134 | return ctx->minspacewidth; |
faad4952 |
1135 | } else |
43341922 |
1136 | str = L" "; |
1137 | } else /* if (type == word_Quote) */ { |
1138 | if (word->aux == quote_Open) |
1139 | str = L"\x2018"; /* FIXME: configurability! */ |
1140 | else |
1141 | str = L"\x2019"; /* FIXME: configurability! */ |
1142 | } |
1143 | |
1144 | width = string_width(ctx->pdata->fonts[findex], str, &errs); |
1145 | |
1146 | if (errs && word->alt) |
faad4952 |
1147 | return paper_width_list(vctx, word->alt, NULL, nspaces); |
43341922 |
1148 | else |
1149 | return ctx->pdata->sizes[findex] * width; |
1150 | } |
1151 | |
faad4952 |
1152 | static int paper_width(void *vctx, word *word) |
1153 | { |
1154 | return paper_width_internal(vctx, word, NULL); |
1155 | } |
1156 | |
515d216b |
1157 | static int paper_width_simple(para_data *pdata, word *text) |
1158 | { |
1159 | struct paper_width_ctx ctx; |
1160 | |
1161 | ctx.pdata = pdata; |
1162 | ctx.minspacewidth = |
1163 | (pdata->sizes[FONT_NORMAL] * |
1164 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
1165 | |
1166 | return paper_width_list(&ctx, text, NULL, NULL); |
1167 | } |
1168 | |
43341922 |
1169 | static void wrap_paragraph(para_data *pdata, word *words, |
1170 | int w, int i1, int i2) |
1171 | { |
1172 | wrappedline *wrapping, *p; |
1173 | int spacewidth; |
1174 | struct paper_width_ctx ctx; |
1175 | int line_height; |
1176 | |
1177 | /* |
1178 | * We're going to need to store the line height in every line |
1179 | * structure we generate. |
1180 | */ |
1181 | { |
1182 | int i; |
1183 | line_height = 0; |
1184 | for (i = 0; i < NFONTS; i++) |
1185 | if (line_height < pdata->sizes[i]) |
1186 | line_height = pdata->sizes[i]; |
1187 | line_height *= 4096; |
1188 | } |
1189 | |
1190 | spacewidth = (pdata->sizes[FONT_NORMAL] * |
1191 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
1192 | if (spacewidth == 0) { |
1193 | /* |
1194 | * A font without a space?! Disturbing. I hope this never |
1195 | * comes up, but I'll make a random guess anyway and set my |
1196 | * space width to half the point size. |
1197 | */ |
1198 | spacewidth = pdata->sizes[FONT_NORMAL] * 4096 / 2; |
1199 | } |
1200 | |
1201 | /* |
1202 | * I'm going to set the _minimum_ space width to 3/5 of the |
1203 | * standard one, and use the standard one as the optimum. |
1204 | */ |
1205 | ctx.minspacewidth = spacewidth * 3 / 5; |
1206 | ctx.pdata = pdata; |
1207 | |
1208 | wrapping = wrap_para(words, w - i1, w - i2, paper_width, &ctx, spacewidth); |
1209 | |
1210 | /* |
1211 | * Having done the wrapping, we now concoct a set of line_data |
1212 | * structures. |
1213 | */ |
1214 | pdata->first = pdata->last = NULL; |
1215 | |
1216 | for (p = wrapping; p; p = p->next) { |
1217 | line_data *ldata; |
1218 | word *wd; |
1219 | int len, wid, spaces; |
1220 | |
1221 | ldata = mknew(line_data); |
1222 | |
1223 | ldata->pdata = pdata; |
1224 | ldata->first = p->begin; |
faad4952 |
1225 | ldata->end = p->end; |
43341922 |
1226 | ldata->line_height = line_height; |
1227 | |
1228 | ldata->xpos = (p == wrapping ? i1 : i2); |
1229 | |
1230 | if (pdata->last) { |
1231 | pdata->last->next = ldata; |
1232 | ldata->prev = pdata->last; |
1233 | } else { |
1234 | pdata->first = ldata; |
1235 | ldata->prev = NULL; |
1236 | } |
1237 | ldata->next = NULL; |
1238 | pdata->last = ldata; |
1239 | |
43341922 |
1240 | spaces = 0; |
faad4952 |
1241 | len = paper_width_list(&ctx, ldata->first, ldata->end, &spaces); |
1242 | wid = (p == wrapping ? w - i1 : w - i2); |
43341922 |
1243 | wd = ldata->first; |
43341922 |
1244 | |
faad4952 |
1245 | ldata->hshortfall = wid - len; |
1246 | ldata->nspaces = spaces; |
1247 | /* |
1248 | * This tells us how much the space width needs to |
1249 | * change from _min_spacewidth. But we want to store |
1250 | * its difference from the _natural_ space width, to |
1251 | * make the text rendering easier. |
1252 | */ |
1253 | ldata->hshortfall += ctx.minspacewidth * spaces; |
1254 | ldata->hshortfall -= spacewidth * spaces; |
c6536773 |
1255 | ldata->real_shortfall = ldata->hshortfall; |
faad4952 |
1256 | /* |
1257 | * Special case: on the last line of a paragraph, we |
1258 | * never stretch spaces. |
1259 | */ |
1260 | if (ldata->hshortfall > 0 && !p->next) |
1261 | ldata->hshortfall = 0; |
43341922 |
1262 | |
1263 | ldata->aux_text = NULL; |
515d216b |
1264 | ldata->aux_text_2 = NULL; |
43341922 |
1265 | ldata->aux_left_indent = 0; |
39a0cfb9 |
1266 | ldata->penalty_before = ldata->penalty_after = 0; |
43341922 |
1267 | } |
1268 | |
1269 | } |
1270 | |
1271 | static page_data *page_breaks(line_data *first, line_data *last, |
c6536773 |
1272 | int page_height, int ncols, int headspace) |
43341922 |
1273 | { |
1274 | line_data *l, *m; |
1275 | page_data *ph, *pt; |
c6536773 |
1276 | int n, n1, this_height; |
43341922 |
1277 | |
1278 | /* |
1279 | * Page breaking is done by a close analogue of the optimal |
1280 | * paragraph wrapping algorithm used by wrap_para(). We work |
1281 | * backwards from the end of the document line by line; for |
1282 | * each line, we contemplate every possible number of lines we |
1283 | * could put on a page starting with that line, determine a |
1284 | * cost function for each one, add it to the pre-computed cost |
1285 | * function for optimally page-breaking everything after that |
1286 | * page, and pick the best option. |
1287 | * |
c6536773 |
1288 | * This is made slightly more complex by the fact that we have |
1289 | * a multi-column index with a heading at the top of the |
1290 | * _first_ page, meaning that the first _ncols_ pages must have |
1291 | * a different length. Hence, we must do the wrapping ncols+1 |
1292 | * times over, hypothetically trying to put every subsequence |
1293 | * on every possible page. |
1294 | * |
43341922 |
1295 | * Since my line_data structures are only used for this |
1296 | * purpose, I might as well just store the algorithm data |
1297 | * directly in them. |
1298 | */ |
1299 | |
1300 | for (l = last; l; l = l->prev) { |
c6536773 |
1301 | l->bestcost = mknewa(int, ncols+1); |
1302 | l->vshortfall = mknewa(int, ncols+1); |
1303 | l->text = mknewa(int, ncols+1); |
1304 | l->space = mknewa(int, ncols+1); |
1305 | l->page_last = mknewa(line_data *, ncols+1); |
1306 | |
1307 | for (n = 0; n <= ncols; n++) { |
1308 | int minheight, text = 0, space = 0; |
1309 | int cost; |
1310 | |
1311 | n1 = (n < ncols ? n+1 : ncols); |
1312 | if (n < ncols) |
1313 | this_height = page_height - headspace; |
1314 | else |
1315 | this_height = page_height; |
1316 | |
1317 | l->bestcost[n] = -1; |
1318 | for (m = l; m; m = m->next) { |
1319 | if (m != l && m->page_break) |
1320 | break; /* we've gone as far as we can */ |
1321 | |
1322 | if (m != l) { |
1323 | if (m->prev->space_after > 0) |
1324 | space += m->prev->space_after; |
1325 | else |
1326 | text += m->prev->space_after; |
1327 | } |
1328 | if (m != l || m->page_break) { |
1329 | if (m->space_before > 0) |
1330 | space += m->space_before; |
1331 | else |
1332 | text += m->space_before; |
1333 | } |
1334 | text += m->line_height; |
1335 | minheight = text + space; |
43341922 |
1336 | |
c6536773 |
1337 | if (m != l && minheight > this_height) |
1338 | break; |
43341922 |
1339 | |
c6536773 |
1340 | /* |
1341 | * If the space after this paragraph is _negative_ |
1342 | * (which means the next line is folded on to this |
1343 | * one, which happens in the index), we absolutely |
1344 | * cannot break here. |
1345 | */ |
1346 | if (m->space_after >= 0) { |
43341922 |
1347 | |
c6536773 |
1348 | /* |
1349 | * Compute the cost of this arrangement, as the |
1350 | * square of the amount of wasted space on the |
1351 | * page. Exception: if this is the last page |
1352 | * before a mandatory break or the document |
1353 | * end, we don't penalise a large blank area. |
1354 | */ |
1355 | if (m != last && m->next && !m->next->page_break) |
1356 | { |
1357 | int x = this_height - minheight; |
1358 | int xf; |
1359 | |
1360 | xf = x & 0xFF; |
1361 | x >>= 8; |
1362 | |
1363 | cost = x*x; |
1364 | cost += (x * xf) >> 8; |
1365 | } else |
1366 | cost = 0; |
1367 | |
1368 | if (m != last && m->next && !m->next->page_break) { |
1369 | cost += m->penalty_after; |
1370 | cost += m->next->penalty_before; |
1371 | } |
43341922 |
1372 | |
c6536773 |
1373 | if (m != last && m->next && !m->next->page_break) |
1374 | cost += m->next->bestcost[n1]; |
1375 | if (l->bestcost[n] == -1 || l->bestcost[n] > cost) { |
1376 | /* |
1377 | * This is the best option yet for this |
1378 | * starting point. |
1379 | */ |
1380 | l->bestcost[n] = cost; |
1381 | if (m != last && m->next && !m->next->page_break) |
1382 | l->vshortfall[n] = this_height - minheight; |
1383 | else |
1384 | l->vshortfall[n] = 0; |
1385 | l->text[n] = text; |
1386 | l->space[n] = space; |
1387 | l->page_last[n] = m; |
1388 | } |
1389 | } |
43341922 |
1390 | |
c6536773 |
1391 | if (m == last) |
1392 | break; |
43341922 |
1393 | } |
1394 | } |
1395 | } |
1396 | |
1397 | /* |
1398 | * Now go through the line list forwards and assemble the |
1399 | * actual pages. |
1400 | */ |
1401 | ph = pt = NULL; |
1402 | |
1403 | l = first; |
c6536773 |
1404 | n = 0; |
43341922 |
1405 | while (l) { |
1406 | page_data *page; |
c6536773 |
1407 | int text, space, head; |
43341922 |
1408 | |
1409 | page = mknew(page_data); |
1410 | page->next = NULL; |
1411 | page->prev = pt; |
1412 | if (pt) |
1413 | pt->next = page; |
1414 | else |
1415 | ph = page; |
1416 | pt = page; |
1417 | |
1418 | page->first_line = l; |
c6536773 |
1419 | page->last_line = l->page_last[n]; |
43341922 |
1420 | |
1421 | page->first_text = page->last_text = NULL; |
138d7ffb |
1422 | page->first_xref = page->last_xref = NULL; |
23765aeb |
1423 | page->first_rect = page->last_rect = NULL; |
138d7ffb |
1424 | |
43341922 |
1425 | /* |
1426 | * Now assign a y-coordinate to each line on the page. |
1427 | */ |
1428 | text = space = 0; |
c6536773 |
1429 | head = (n < ncols ? headspace : 0); |
43341922 |
1430 | for (l = page->first_line; l; l = l->next) { |
c6536773 |
1431 | if (l != page->first_line) { |
1432 | if (l->prev->space_after > 0) |
1433 | space += l->prev->space_after; |
1434 | else |
1435 | text += l->prev->space_after; |
1436 | } |
1437 | if (l != page->first_line || l->page_break) { |
1438 | if (l->space_before > 0) |
1439 | space += l->space_before; |
1440 | else |
1441 | text += l->space_before; |
1442 | } |
43341922 |
1443 | text += l->line_height; |
1444 | |
1445 | l->page = page; |
c6536773 |
1446 | l->ypos = text + space + head + |
1447 | space * (float)page->first_line->vshortfall[n] / |
1448 | page->first_line->space[n]; |
43341922 |
1449 | |
1450 | if (l == page->last_line) |
1451 | break; |
1452 | } |
1453 | |
c6536773 |
1454 | l = page->last_line; |
1455 | if (l == last) |
1456 | break; |
1457 | l = l->next; |
1458 | |
1459 | n = (n < ncols ? n+1 : ncols); |
43341922 |
1460 | } |
1461 | |
1462 | return ph; |
1463 | } |
1464 | |
23765aeb |
1465 | static void add_rect_to_page(page_data *page, int x, int y, int w, int h) |
1466 | { |
1467 | rect *r = mknew(rect); |
1468 | |
1469 | r->next = NULL; |
1470 | if (page->last_rect) |
1471 | page->last_rect->next = r; |
1472 | else |
1473 | page->first_rect = r; |
1474 | page->last_rect = r; |
1475 | |
1476 | r->x = x; |
1477 | r->y = y; |
1478 | r->w = w; |
1479 | r->h = h; |
1480 | } |
1481 | |
43341922 |
1482 | static void add_string_to_page(page_data *page, int x, int y, |
1483 | font_encoding *fe, int size, char *text) |
1484 | { |
1485 | text_fragment *frag; |
1486 | |
1487 | frag = mknew(text_fragment); |
1488 | frag->next = NULL; |
1489 | |
1490 | if (page->last_text) |
1491 | page->last_text->next = frag; |
1492 | else |
1493 | page->first_text = frag; |
1494 | page->last_text = frag; |
1495 | |
1496 | frag->x = x; |
1497 | frag->y = y; |
1498 | frag->fe = fe; |
1499 | frag->fontsize = size; |
1500 | frag->text = dupstr(text); |
1501 | } |
1502 | |
1503 | /* |
1504 | * Returns the updated x coordinate. |
1505 | */ |
1506 | static int render_string(page_data *page, font_data *font, int fontsize, |
1507 | int x, int y, wchar_t *str) |
1508 | { |
1509 | char *text; |
1510 | int textpos, textwid, glyph; |
1511 | font_encoding *subfont = NULL, *sf; |
1512 | |
1513 | text = mknewa(char, 1 + ustrlen(str)); |
1514 | textpos = textwid = 0; |
1515 | |
1516 | while (*str) { |
1517 | glyph = font->bmp[*str]; |
1518 | |
1519 | if (glyph == 0xFFFF) |
1520 | continue; /* nothing more we can do here */ |
1521 | |
1522 | /* |
1523 | * Find which subfont this character is going in. |
1524 | */ |
1525 | sf = font->subfont_map[glyph].subfont; |
1526 | |
1527 | if (!sf) { |
1528 | int c; |
1529 | |
1530 | /* |
1531 | * This character is not yet in a subfont. Assign one. |
1532 | */ |
1533 | if (font->latest_subfont->free_pos >= 0x100) |
1534 | font->latest_subfont = new_font_encoding(font); |
1535 | |
1536 | c = font->latest_subfont->free_pos++; |
1537 | if (font->latest_subfont->free_pos == 0x7F) |
1538 | font->latest_subfont->free_pos = 0xA1; |
1539 | |
1540 | font->subfont_map[glyph].subfont = font->latest_subfont; |
1541 | font->subfont_map[glyph].position = c; |
1542 | font->latest_subfont->vector[c] = font->glyphs[glyph]; |
1543 | font->latest_subfont->indices[c] = glyph; |
1544 | font->latest_subfont->to_unicode[c] = *str; |
1545 | |
1546 | sf = font->latest_subfont; |
1547 | } |
1548 | |
1549 | if (!subfont || sf != subfont) { |
1550 | if (subfont) { |
1551 | text[textpos] = '\0'; |
1552 | add_string_to_page(page, x, y, subfont, fontsize, text); |
1553 | x += textwid; |
1554 | } else { |
1555 | assert(textpos == 0); |
1556 | } |
1557 | textpos = 0; |
1558 | subfont = sf; |
1559 | } |
1560 | |
1561 | text[textpos++] = font->subfont_map[glyph].position; |
1562 | textwid += font->widths[glyph] * fontsize; |
1563 | |
1564 | str++; |
1565 | } |
1566 | |
1567 | if (textpos > 0) { |
1568 | text[textpos] = '\0'; |
1569 | add_string_to_page(page, x, y, subfont, fontsize, text); |
1570 | x += textwid; |
1571 | } |
1572 | |
1573 | return x; |
1574 | } |
1575 | |
1576 | /* |
1577 | * Returns the updated x coordinate. |
1578 | */ |
138d7ffb |
1579 | static int render_text(page_data *page, para_data *pdata, line_data *ldata, |
1580 | int x, int y, word *text, word *text_end, xref **xr, |
1581 | int shortfall, int nspaces, int *nspace, |
c6536773 |
1582 | keywordlist *keywords, indexdata *idx) |
43341922 |
1583 | { |
faad4952 |
1584 | while (text && text != text_end) { |
43341922 |
1585 | int style, type, findex, errs; |
1586 | wchar_t *str; |
138d7ffb |
1587 | xref_dest dest; |
43341922 |
1588 | |
1589 | switch (text->type) { |
138d7ffb |
1590 | /* |
1591 | * Start a cross-reference. |
1592 | */ |
43341922 |
1593 | case word_HyperLink: |
43341922 |
1594 | case word_UpperXref: |
1595 | case word_LowerXref: |
138d7ffb |
1596 | |
1597 | if (text->type == word_HyperLink) { |
1598 | dest.type = URL; |
1599 | dest.url = utoa_dup(text->text); |
1600 | dest.page = NULL; |
1601 | } else { |
1602 | keyword *kwl = kw_lookup(keywords, text->text); |
1603 | para_data *pdata; |
1604 | |
1605 | if (kwl) { |
1606 | assert(kwl->para->private_data); |
1607 | pdata = (para_data *) kwl->para->private_data; |
1608 | dest.type = PAGE; |
1609 | dest.page = pdata->first->page; |
1610 | dest.url = NULL; |
1611 | } else { |
1612 | /* |
1613 | * Shouldn't happen, but *shrug* |
1614 | */ |
1615 | dest.type = NONE; |
1616 | dest.page = NULL; |
1617 | dest.url = NULL; |
1618 | } |
1619 | } |
1620 | if (dest.type != NONE) { |
1621 | *xr = mknew(xref); |
1622 | (*xr)->dest = dest; /* structure copy */ |
1623 | if (page->last_xref) |
1624 | page->last_xref->next = *xr; |
1625 | else |
1626 | page->first_xref = *xr; |
1627 | page->last_xref = *xr; |
23765aeb |
1628 | (*xr)->next = NULL; |
138d7ffb |
1629 | |
1630 | /* |
1631 | * FIXME: Ideally we should have, and use, some |
1632 | * vertical font metric information here so that |
1633 | * our cross-ref rectangle can take account of |
1634 | * descenders and the font's cap height. This will |
1635 | * do for the moment, but it isn't ideal. |
1636 | */ |
1637 | (*xr)->lx = (*xr)->rx = x; |
1638 | (*xr)->by = y; |
1639 | (*xr)->ty = y + ldata->line_height; |
1640 | } |
1641 | goto nextword; |
1642 | |
1643 | /* |
1644 | * Finish extending a cross-reference box. |
1645 | */ |
1646 | case word_HyperEnd: |
43341922 |
1647 | case word_XrefEnd: |
138d7ffb |
1648 | *xr = NULL; |
1649 | goto nextword; |
1650 | |
43341922 |
1651 | /* |
c6536773 |
1652 | * Add the current page number to the list of pages |
1653 | * referenced by an index entry. |
43341922 |
1654 | */ |
c6536773 |
1655 | case word_IndexRef: |
1656 | { |
1657 | indextag *tag; |
1658 | int i; |
1659 | |
1660 | tag = index_findtag(idx, text->text); |
1661 | if (!tag) |
1662 | goto nextword; |
1663 | |
1664 | for (i = 0; i < tag->nrefs; i++) { |
1665 | indexentry *entry = tag->refs[i]; |
1666 | paper_idx *pi = (paper_idx *)entry->backend_data; |
1667 | |
1668 | /* |
1669 | * If the same index term is indexed twice |
1670 | * within the same section, we only want to |
1671 | * mention it once in the index. |
1672 | */ |
1673 | if (pi->lastpage != page) { |
1674 | if (pi->lastword) { |
1675 | pi->lastword = pi->lastword->next = |
1676 | fake_word(L","); |
1677 | pi->lastword = pi->lastword->next = |
1678 | fake_space_word(); |
1679 | pi->lastword = pi->lastword->next = |
1680 | fake_word(page->number); |
1681 | } else { |
1682 | pi->lastword = pi->words = |
1683 | fake_word(page->number); |
1684 | } |
1685 | } |
1686 | |
1687 | pi->lastpage = page; |
1688 | } |
1689 | } |
1690 | goto nextword; |
43341922 |
1691 | } |
1692 | |
1693 | style = towordstyle(text->type); |
1694 | type = removeattr(text->type); |
1695 | |
1696 | findex = (style == word_Normal ? FONT_NORMAL : |
1697 | style == word_Emph ? FONT_EMPH : |
1698 | FONT_CODE); |
1699 | |
1700 | if (type == word_Normal) { |
1701 | str = text->text; |
1702 | } else if (type == word_WhiteSpace) { |
1703 | x += pdata->sizes[findex] * |
1704 | string_width(pdata->fonts[findex], L" ", NULL); |
faad4952 |
1705 | if (nspaces && findex != FONT_CODE) { |
1706 | x += (*nspace+1) * shortfall / nspaces; |
1707 | x -= *nspace * shortfall / nspaces; |
1708 | (*nspace)++; |
1709 | } |
43341922 |
1710 | goto nextword; |
1711 | } else /* if (type == word_Quote) */ { |
1712 | if (text->aux == quote_Open) |
1713 | str = L"\x2018"; /* FIXME: configurability! */ |
1714 | else |
1715 | str = L"\x2019"; /* FIXME: configurability! */ |
1716 | } |
1717 | |
1718 | (void) string_width(pdata->fonts[findex], str, &errs); |
1719 | |
1720 | if (errs && text->alt) |
138d7ffb |
1721 | x = render_text(page, pdata, ldata, x, y, text->alt, NULL, |
c6536773 |
1722 | xr, shortfall, nspaces, nspace, keywords, idx); |
43341922 |
1723 | else |
1724 | x = render_string(page, pdata->fonts[findex], |
1725 | pdata->sizes[findex], x, y, str); |
1726 | |
138d7ffb |
1727 | if (*xr) |
1728 | (*xr)->rx = x; |
1729 | |
43341922 |
1730 | nextword: |
43341922 |
1731 | text = text->next; |
1732 | } |
1733 | |
1734 | return x; |
1735 | } |
1736 | |
2bfd1b76 |
1737 | /* |
1738 | * Returns the last x position used on the line. |
1739 | */ |
1740 | static int render_line(line_data *ldata, int left_x, int top_y, |
c6536773 |
1741 | xref_dest *dest, keywordlist *keywords, indexdata *idx) |
43341922 |
1742 | { |
faad4952 |
1743 | int nspace; |
138d7ffb |
1744 | xref *xr; |
2bfd1b76 |
1745 | int ret = 0; |
138d7ffb |
1746 | |
faad4952 |
1747 | if (ldata->aux_text) { |
515d216b |
1748 | int x; |
138d7ffb |
1749 | xr = NULL; |
faad4952 |
1750 | nspace = 0; |
515d216b |
1751 | x = render_text(ldata->page, ldata->pdata, ldata, |
1752 | left_x + ldata->aux_left_indent, |
1753 | top_y - ldata->ypos, |
c6536773 |
1754 | ldata->aux_text, NULL, &xr, 0, 0, &nspace, |
1755 | keywords, idx); |
515d216b |
1756 | if (ldata->aux_text_2) |
1757 | render_text(ldata->page, ldata->pdata, ldata, |
1758 | x, top_y - ldata->ypos, |
c6536773 |
1759 | ldata->aux_text_2, NULL, &xr, 0, 0, &nspace, |
1760 | keywords, idx); |
faad4952 |
1761 | } |
1762 | nspace = 0; |
138d7ffb |
1763 | |
87bd6353 |
1764 | if (ldata->first) { |
138d7ffb |
1765 | /* |
87bd6353 |
1766 | * There might be a cross-reference carried over from a |
1767 | * previous line. |
138d7ffb |
1768 | */ |
87bd6353 |
1769 | if (dest->type != NONE) { |
1770 | xr = mknew(xref); |
1771 | xr->next = NULL; |
1772 | xr->dest = *dest; /* structure copy */ |
1773 | if (ldata->page->last_xref) |
1774 | ldata->page->last_xref->next = xr; |
1775 | else |
1776 | ldata->page->first_xref = xr; |
1777 | ldata->page->last_xref = xr; |
1778 | xr->lx = xr->rx = left_x + ldata->xpos; |
1779 | xr->by = top_y - ldata->ypos; |
1780 | xr->ty = top_y - ldata->ypos + ldata->line_height; |
1781 | } else |
1782 | xr = NULL; |
1783 | |
c6536773 |
1784 | { |
1785 | int extra_indent, shortfall, spaces; |
1786 | int just = ldata->pdata->justification; |
1787 | |
1788 | /* |
1789 | * All forms of justification become JUST when we have |
1790 | * to squeeze the paragraph. |
1791 | */ |
1792 | if (ldata->hshortfall < 0) |
1793 | just = JUST; |
1794 | |
1795 | switch (just) { |
1796 | case JUST: |
1797 | shortfall = ldata->hshortfall; |
1798 | spaces = ldata->nspaces; |
1799 | extra_indent = 0; |
1800 | break; |
1801 | case LEFT: |
1802 | shortfall = spaces = extra_indent = 0; |
1803 | break; |
1804 | case RIGHT: |
1805 | shortfall = spaces = 0; |
1806 | extra_indent = ldata->real_shortfall; |
1807 | break; |
1808 | } |
1809 | |
1810 | ret = render_text(ldata->page, ldata->pdata, ldata, |
1811 | left_x + ldata->xpos + extra_indent, |
1812 | top_y - ldata->ypos, ldata->first, ldata->end, |
1813 | &xr, shortfall, spaces, &nspace, |
1814 | keywords, idx); |
1815 | } |
87bd6353 |
1816 | |
1817 | if (xr) { |
1818 | /* |
1819 | * There's a cross-reference continued on to the next line. |
1820 | */ |
1821 | *dest = xr->dest; |
1822 | } else |
1823 | dest->type = NONE; |
1824 | } |
2bfd1b76 |
1825 | |
1826 | return ret; |
43341922 |
1827 | } |
515d216b |
1828 | |
c6536773 |
1829 | static void render_para(para_data *pdata, paper_conf *conf, |
1830 | keywordlist *keywords, indexdata *idx, |
1831 | paragraph *index_placeholder, page_data *index_page) |
1832 | { |
1833 | int last_x; |
1834 | xref *cxref; |
1835 | page_data *cxref_page; |
1836 | xref_dest dest; |
1837 | para_data *target; |
1838 | line_data *ldata; |
1839 | |
1840 | dest.type = NONE; |
1841 | cxref = NULL; |
1842 | cxref_page = NULL; |
1843 | |
1844 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
1845 | /* |
1846 | * If this is a contents entry, we expect to have a single |
1847 | * enormous cross-reference rectangle covering the whole |
1848 | * thing. (Unless, of course, it spans multiple pages.) |
1849 | */ |
1850 | if (pdata->contents_entry && ldata->page != cxref_page) { |
1851 | cxref_page = ldata->page; |
1852 | cxref = mknew(xref); |
1853 | cxref->next = NULL; |
1854 | cxref->dest.type = PAGE; |
1855 | if (pdata->contents_entry == index_placeholder) { |
1856 | cxref->dest.page = index_page; |
1857 | } else { |
1858 | assert(pdata->contents_entry->private_data); |
1859 | target = (para_data *)pdata->contents_entry->private_data; |
1860 | cxref->dest.page = target->first->page; |
1861 | } |
1862 | cxref->dest.url = NULL; |
1863 | if (ldata->page->last_xref) |
1864 | ldata->page->last_xref->next = cxref; |
1865 | else |
1866 | ldata->page->first_xref = cxref; |
1867 | ldata->page->last_xref = cxref; |
1868 | cxref->lx = conf->left_margin; |
1869 | cxref->rx = conf->paper_width - conf->right_margin; |
1870 | cxref->ty = conf->paper_height - conf->top_margin |
1871 | - ldata->ypos + ldata->line_height; |
1872 | } |
1873 | if (pdata->contents_entry) { |
1874 | assert(cxref != NULL); |
1875 | cxref->by = conf->paper_height - conf->top_margin |
1876 | - ldata->ypos; |
1877 | } |
1878 | |
1879 | last_x = render_line(ldata, conf->left_margin, |
1880 | conf->paper_height - conf->top_margin, |
1881 | &dest, keywords, idx); |
1882 | if (ldata == pdata->last) |
1883 | break; |
1884 | } |
1885 | |
1886 | /* |
1887 | * If this is a contents entry, add leaders and a page |
1888 | * number. |
1889 | */ |
1890 | if (pdata->contents_entry) { |
1891 | word *w; |
1892 | wchar_t *num; |
1893 | int wid; |
1894 | int x; |
1895 | |
1896 | if (pdata->contents_entry == index_placeholder) { |
1897 | num = index_page->number; |
1898 | } else { |
1899 | assert(pdata->contents_entry->private_data); |
1900 | target = (para_data *)pdata->contents_entry->private_data; |
1901 | num = target->first->page->number; |
1902 | } |
1903 | |
1904 | w = fake_word(num); |
1905 | wid = paper_width_simple(pdata, w); |
1906 | sfree(w); |
1907 | |
1908 | render_string(pdata->last->page, |
1909 | pdata->fonts[FONT_NORMAL], |
1910 | pdata->sizes[FONT_NORMAL], |
1911 | conf->paper_width - conf->right_margin - wid, |
1912 | (conf->paper_height - conf->top_margin - |
1913 | pdata->last->ypos), num); |
1914 | |
1915 | for (x = 0; x < conf->base_width; x += conf->leader_separation) |
1916 | if (x - conf->leader_separation > last_x - conf->left_margin && |
1917 | x + conf->leader_separation < conf->base_width - wid) |
1918 | render_string(pdata->last->page, |
1919 | pdata->fonts[FONT_NORMAL], |
1920 | pdata->sizes[FONT_NORMAL], |
1921 | conf->left_margin + x, |
1922 | (conf->paper_height - conf->top_margin - |
1923 | pdata->last->ypos), L"."); |
1924 | } |
1925 | |
1926 | /* |
1927 | * Render any rectangle (chapter title underline or rule) |
1928 | * that goes with this paragraph. |
1929 | */ |
1930 | switch (pdata->rect_type) { |
1931 | case RECT_CHAPTER_UNDERLINE: |
1932 | add_rect_to_page(pdata->last->page, |
1933 | conf->left_margin, |
1934 | (conf->paper_height - conf->top_margin - |
1935 | pdata->last->ypos - |
1936 | conf->chapter_underline_depth), |
1937 | conf->base_width, |
1938 | conf->chapter_underline_thickness); |
1939 | break; |
1940 | case RECT_RULE: |
1941 | add_rect_to_page(pdata->first->page, |
1942 | conf->left_margin + pdata->first->xpos, |
1943 | (conf->paper_height - conf->top_margin - |
1944 | pdata->last->ypos - |
1945 | pdata->last->line_height), |
1946 | conf->base_width - pdata->first->xpos, |
1947 | pdata->last->line_height); |
1948 | break; |
1949 | default: /* placate gcc */ |
1950 | break; |
1951 | } |
1952 | } |
1953 | |
be76d597 |
1954 | static para_data *code_paragraph(int indent, word *words, paper_conf *conf) |
515d216b |
1955 | { |
be76d597 |
1956 | para_data *pdata = mknew(para_data); |
1957 | |
515d216b |
1958 | /* |
1959 | * For code paragraphs, I'm going to hack grievously and |
1960 | * pretend the three normal fonts are the three code paragraph |
1961 | * fonts. |
1962 | */ |
be76d597 |
1963 | pdata->fonts[FONT_NORMAL] = conf->cb; |
1964 | pdata->fonts[FONT_EMPH] = conf->co; |
36e64ed4 |
1965 | pdata->fonts[FONT_CODE] = conf->cr; |
515d216b |
1966 | pdata->sizes[FONT_NORMAL] = |
1967 | pdata->sizes[FONT_EMPH] = |
be76d597 |
1968 | pdata->sizes[FONT_CODE] = 12; |
515d216b |
1969 | |
1970 | pdata->first = pdata->last = NULL; |
be76d597 |
1971 | pdata->outline_level = -1; |
1972 | pdata->rect_type = RECT_NONE; |
2bfd1b76 |
1973 | pdata->contents_entry = NULL; |
c6536773 |
1974 | pdata->justification = LEFT; |
515d216b |
1975 | |
1976 | for (; words; words = words->next) { |
1977 | wchar_t *t, *e, *start; |
1978 | word *lhead = NULL, *ltail = NULL, *w; |
1979 | line_data *ldata; |
1980 | int prev = -1, curr; |
1981 | |
1982 | t = words->text; |
1983 | if (words->next && words->next->type == word_Emph) { |
1984 | e = words->next->text; |
1985 | words = words->next; |
1986 | } else |
1987 | e = NULL; |
1988 | |
1989 | start = t; |
1990 | |
1991 | while (*start) { |
1992 | while (*t) { |
1993 | if (!e || !*e) |
1994 | curr = 0; |
1995 | else if (*e == L'i') |
1996 | curr = 1; |
1997 | else if (*e == L'b') |
1998 | curr = 2; |
1999 | else |
2000 | curr = 0; |
2001 | |
2002 | if (prev < 0) |
2003 | prev = curr; |
2004 | |
2005 | if (curr != prev) |
2006 | break; |
2007 | |
2008 | t++; |
2009 | if (e && *e) |
2010 | e++; |
2011 | } |
2012 | |
2013 | /* |
2014 | * We've isolated a maximal subsequence of the line |
2015 | * which has the same emphasis. Form it into a word |
2016 | * structure. |
2017 | */ |
2018 | w = mknew(word); |
2019 | w->next = NULL; |
2020 | w->alt = NULL; |
2021 | w->type = (prev == 0 ? word_WeakCode : |
2022 | prev == 1 ? word_Emph : word_Normal); |
2023 | w->text = mknewa(wchar_t, t-start+1); |
2024 | memcpy(w->text, start, (t-start) * sizeof(wchar_t)); |
2025 | w->text[t-start] = '\0'; |
2026 | w->breaks = FALSE; |
2027 | |
2028 | if (ltail) |
2029 | ltail->next = w; |
2030 | else |
2031 | lhead = w; |
2032 | ltail = w; |
2033 | |
2034 | start = t; |
2035 | prev = -1; |
2036 | } |
2037 | |
2038 | ldata = mknew(line_data); |
2039 | |
2040 | ldata->pdata = pdata; |
2041 | ldata->first = lhead; |
2042 | ldata->end = NULL; |
be76d597 |
2043 | ldata->line_height = conf->base_font_size * 4096; |
515d216b |
2044 | |
2045 | ldata->xpos = indent; |
2046 | |
2047 | if (pdata->last) { |
2048 | pdata->last->next = ldata; |
2049 | ldata->prev = pdata->last; |
2050 | } else { |
2051 | pdata->first = ldata; |
2052 | ldata->prev = NULL; |
2053 | } |
2054 | ldata->next = NULL; |
2055 | pdata->last = ldata; |
2056 | |
2057 | ldata->hshortfall = 0; |
2058 | ldata->nspaces = 0; |
2059 | ldata->aux_text = NULL; |
2060 | ldata->aux_text_2 = NULL; |
2061 | ldata->aux_left_indent = 0; |
39a0cfb9 |
2062 | /* General opprobrium for breaking in a code paragraph. */ |
2063 | ldata->penalty_before = ldata->penalty_after = 50000; |
515d216b |
2064 | } |
be76d597 |
2065 | |
2066 | standard_line_spacing(pdata, conf); |
2067 | |
2068 | return pdata; |
515d216b |
2069 | } |
87bd6353 |
2070 | |
be76d597 |
2071 | static para_data *rule_paragraph(int indent, paper_conf *conf) |
87bd6353 |
2072 | { |
be76d597 |
2073 | para_data *pdata = mknew(para_data); |
87bd6353 |
2074 | line_data *ldata; |
2075 | |
2076 | ldata = mknew(line_data); |
2077 | |
2078 | ldata->pdata = pdata; |
2079 | ldata->first = NULL; |
2080 | ldata->end = NULL; |
be76d597 |
2081 | ldata->line_height = conf->rule_thickness; |
87bd6353 |
2082 | |
2083 | ldata->xpos = indent; |
2084 | |
2085 | ldata->prev = NULL; |
2086 | ldata->next = NULL; |
2087 | |
2088 | ldata->hshortfall = 0; |
2089 | ldata->nspaces = 0; |
2090 | ldata->aux_text = NULL; |
2091 | ldata->aux_text_2 = NULL; |
2092 | ldata->aux_left_indent = 0; |
2093 | |
2094 | /* |
2095 | * Better to break after a rule than before it |
2096 | */ |
2097 | ldata->penalty_after += 100000; |
2098 | ldata->penalty_before += -100000; |
2099 | |
2100 | pdata->first = pdata->last = ldata; |
be76d597 |
2101 | pdata->outline_level = -1; |
2102 | pdata->rect_type = RECT_RULE; |
2bfd1b76 |
2103 | pdata->contents_entry = NULL; |
c6536773 |
2104 | pdata->justification = LEFT; |
be76d597 |
2105 | |
2106 | standard_line_spacing(pdata, conf); |
2107 | |
2108 | return pdata; |
2109 | } |
2110 | |
2111 | /* |
2112 | * Plain-text-like formatting for outline titles. |
2113 | */ |
2114 | static void paper_rdaddw(rdstring *rs, word *text) { |
2115 | for (; text; text = text->next) switch (text->type) { |
2116 | case word_HyperLink: |
2117 | case word_HyperEnd: |
2118 | case word_UpperXref: |
2119 | case word_LowerXref: |
2120 | case word_XrefEnd: |
2121 | case word_IndexRef: |
2122 | break; |
2123 | |
2124 | case word_Normal: |
2125 | case word_Emph: |
2126 | case word_Code: |
2127 | case word_WeakCode: |
2128 | case word_WhiteSpace: |
2129 | case word_EmphSpace: |
2130 | case word_CodeSpace: |
2131 | case word_WkCodeSpace: |
2132 | case word_Quote: |
2133 | case word_EmphQuote: |
2134 | case word_CodeQuote: |
2135 | case word_WkCodeQuote: |
2136 | assert(text->type != word_CodeQuote && |
2137 | text->type != word_WkCodeQuote); |
2138 | if (towordstyle(text->type) == word_Emph && |
2139 | (attraux(text->aux) == attr_First || |
2140 | attraux(text->aux) == attr_Only)) |
2141 | rdadd(rs, L'_'); /* FIXME: configurability */ |
2142 | else if (towordstyle(text->type) == word_Code && |
2143 | (attraux(text->aux) == attr_First || |
2144 | attraux(text->aux) == attr_Only)) |
2145 | rdadd(rs, L'\''); /* FIXME: configurability */ |
2146 | if (removeattr(text->type) == word_Normal) { |
2147 | rdadds(rs, text->text); |
2148 | } else if (removeattr(text->type) == word_WhiteSpace) { |
2149 | rdadd(rs, L' '); |
2150 | } else if (removeattr(text->type) == word_Quote) { |
2151 | rdadd(rs, L'\''); /* fixme: configurability */ |
2152 | } |
2153 | if (towordstyle(text->type) == word_Emph && |
2154 | (attraux(text->aux) == attr_Last || |
2155 | attraux(text->aux) == attr_Only)) |
2156 | rdadd(rs, L'_'); /* FIXME: configurability */ |
2157 | else if (towordstyle(text->type) == word_Code && |
2158 | (attraux(text->aux) == attr_Last || |
2159 | attraux(text->aux) == attr_Only)) |
2160 | rdadd(rs, L'\''); /* FIXME: configurability */ |
2161 | break; |
2162 | } |
2163 | } |
2164 | |
2165 | static wchar_t *prepare_outline_title(word *first, wchar_t *separator, |
2166 | word *second) |
2167 | { |
2168 | rdstring rs = {0, 0, NULL}; |
2169 | |
2170 | if (first) |
2171 | paper_rdaddw(&rs, first); |
2172 | if (separator) |
2173 | rdadds(&rs, separator); |
2174 | if (second) |
2175 | paper_rdaddw(&rs, second); |
2176 | |
2177 | return rs.text; |
87bd6353 |
2178 | } |
2bfd1b76 |
2179 | |
2180 | static word *fake_word(wchar_t *text) |
2181 | { |
2182 | word *ret = mknew(word); |
2183 | ret->next = NULL; |
2184 | ret->alt = NULL; |
2185 | ret->type = word_Normal; |
2186 | ret->text = ustrdup(text); |
2187 | ret->breaks = FALSE; |
2188 | ret->aux = 0; |
2189 | return ret; |
2190 | } |
2191 | |
c6536773 |
2192 | static word *fake_space_word(void) |
2193 | { |
2194 | word *ret = mknew(word); |
2195 | ret->next = NULL; |
2196 | ret->alt = NULL; |
2197 | ret->type = word_WhiteSpace; |
2198 | ret->text = NULL; |
2199 | ret->breaks = TRUE; |
2200 | ret->aux = 0; |
2201 | return ret; |
2202 | } |
2203 | |
2bfd1b76 |
2204 | static word *prepare_contents_title(word *first, wchar_t *separator, |
2205 | word *second) |
2206 | { |
2207 | word *ret; |
2208 | word **wptr, *w; |
2209 | |
2210 | wptr = &ret; |
2211 | |
2212 | if (first) { |
2213 | w = dup_word_list(first); |
2214 | *wptr = w; |
2215 | while (w->next) |
2216 | w = w->next; |
2217 | wptr = &w->next; |
2218 | } |
2219 | |
2220 | if (separator) { |
2221 | w = fake_word(separator); |
2222 | *wptr = w; |
2223 | wptr = &w->next; |
2224 | } |
2225 | |
2226 | if (second) { |
2227 | *wptr = dup_word_list(second); |
2228 | } |
2229 | |
2230 | return ret; |
2231 | } |
c6536773 |
2232 | |
2233 | static void fold_into_page(page_data *dest, page_data *src, int right_shift) |
2234 | { |
2235 | line_data *ldata; |
2236 | |
2237 | if (!src->first_line) |
2238 | return; |
2239 | |
2240 | if (dest->last_line) { |
2241 | dest->last_line->next = src->first_line; |
2242 | src->first_line->prev = dest->last_line; |
2243 | } |
2244 | dest->last_line = src->last_line; |
2245 | |
2246 | for (ldata = src->first_line; ldata; ldata = ldata->next) { |
2247 | ldata->page = dest; |
2248 | ldata->xpos += right_shift; |
2249 | |
2250 | if (ldata == src->last_line) |
2251 | break; |
2252 | } |
2253 | } |