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 | * |
a8070b46 |
15 | * - implement some simple graphics |
16 | * * I had an underline below chapter headings in the original |
17 | * Perl version, and I thought it looked rather nice |
18 | * * also we need para_Rule. |
19 | * |
43341922 |
20 | * - set up contents section now we know what sections begin on |
21 | * which pages |
22 | * |
43341922 |
23 | * - do PDF outline |
24 | * |
43341922 |
25 | * - index |
26 | * |
a8070b46 |
27 | * - header/footer? Page numbers at least would be handy. Fully |
28 | * configurable footer can wait, though. |
29 | * |
43341922 |
30 | * That should bring us to the same level of functionality that |
31 | * original-Halibut had, and the same in PDF plus the obvious |
32 | * interactive navigation features. After that, in future work: |
33 | * |
34 | * - linearised PDF, perhaps? |
35 | * |
36 | * - I'm uncertain of whether I need to include a ToUnicode CMap |
37 | * in each of my font definitions in PDF. Currently things (by |
38 | * which I mean cut and paste out of acroread) seem to be |
39 | * working fairly happily without it, but I don't know. |
40 | * |
41 | * - configurability |
42 | * |
43 | * - title pages |
44 | */ |
45 | |
46 | #include <assert.h> |
47 | #include <stdio.h> |
48 | |
49 | #include "halibut.h" |
50 | #include "paper.h" |
51 | |
52 | static font_data *make_std_font(font_list *fontlist, char const *name); |
53 | static void wrap_paragraph(para_data *pdata, word *words, |
54 | int w, int i1, int i2); |
55 | static page_data *page_breaks(line_data *first, line_data *last, |
56 | int page_height); |
138d7ffb |
57 | static void render_line(line_data *ldata, int left_x, int top_y, |
58 | xref_dest *dest, keywordlist *keywords); |
515d216b |
59 | static int paper_width_simple(para_data *pdata, word *text); |
60 | static void code_paragraph(para_data *pdata, |
61 | font_data *fn, font_data *fi, font_data *fb, |
62 | int font_size, int indent, word *words); |
43341922 |
63 | |
64 | void *paper_pre_backend(paragraph *sourceform, keywordlist *keywords, |
65 | indexdata *idx) { |
66 | paragraph *p; |
67 | document *doc; |
515d216b |
68 | int indent, extra_indent, firstline_indent, aux_indent; |
43341922 |
69 | para_data *pdata; |
70 | line_data *ldata, *firstline, *lastline; |
515d216b |
71 | font_data *tr, *ti, *hr, *hi, *cr, *co, *cb; |
43341922 |
72 | page_data *pages; |
73 | font_list *fontlist; |
515d216b |
74 | word *aux, *aux2; |
43341922 |
75 | |
76 | /* |
77 | * FIXME: All these things ought to become configurable. |
78 | */ |
79 | int paper_width = 595 * 4096; |
80 | int paper_height = 841 * 4096; |
81 | int left_margin = 72 * 4096; |
82 | int top_margin = 72 * 4096; |
83 | int right_margin = 72 * 4096; |
84 | int bottom_margin = 108 * 4096; |
85 | int indent_list_bullet = 6 * 4096; |
86 | int indent_list = 24 * 4096; |
87 | int indent_quote = 18 * 4096; |
88 | int base_leading = 4096; |
89 | int base_para_spacing = 10 * 4096; |
515d216b |
90 | int chapter_top_space = 72 * 4096; |
91 | int sect_num_left_space = 12 * 4096; |
43341922 |
92 | |
93 | int base_width = paper_width - left_margin - right_margin; |
94 | int page_height = paper_height - top_margin - bottom_margin; |
95 | |
43341922 |
96 | IGNORE(idx); /* FIXME */ |
43341922 |
97 | |
98 | /* |
99 | * First, set up some font structures. |
100 | */ |
101 | fontlist = mknew(font_list); |
102 | fontlist->head = fontlist->tail = NULL; |
103 | tr = make_std_font(fontlist, "Times-Roman"); |
104 | ti = make_std_font(fontlist, "Times-Italic"); |
515d216b |
105 | hr = make_std_font(fontlist, "Helvetica-Bold"); |
106 | hi = make_std_font(fontlist, "Helvetica-BoldOblique"); |
43341922 |
107 | cr = make_std_font(fontlist, "Courier"); |
515d216b |
108 | co = make_std_font(fontlist, "Courier-Oblique"); |
109 | cb = make_std_font(fontlist, "Courier-Bold"); |
43341922 |
110 | |
111 | /* |
112 | * Go through and break up each paragraph into lines. |
113 | */ |
114 | indent = 0; |
115 | firstline = lastline = NULL; |
116 | for (p = sourceform; p; p = p->next) { |
117 | p->private_data = NULL; |
118 | |
119 | switch (p->type) { |
120 | /* |
121 | * These paragraph types are either invisible or don't |
122 | * define text in the normal sense. Either way, they |
123 | * don't require wrapping. |
124 | */ |
125 | case para_IM: |
126 | case para_BR: |
127 | case para_Rule: |
128 | case para_Biblio: |
129 | case para_NotParaType: |
130 | case para_Config: |
131 | case para_VersionID: |
132 | case para_NoCite: |
133 | break; |
134 | |
135 | /* |
136 | * These paragraph types don't require wrapping, but |
137 | * they do affect the line width to which we wrap the |
138 | * rest of the paragraphs, so we need to pay attention. |
139 | */ |
140 | case para_LcontPush: |
141 | indent += indent_list; break; |
142 | case para_LcontPop: |
143 | indent -= indent_list; assert(indent >= 0); break; |
144 | case para_QuotePush: |
145 | indent += indent_quote; break; |
146 | case para_QuotePop: |
147 | indent -= indent_quote; assert(indent >= 0); break; |
148 | |
149 | /* |
150 | * This paragraph type is special. Process it |
151 | * specially. |
152 | */ |
153 | case para_Code: |
515d216b |
154 | pdata = mknew(para_data); |
155 | code_paragraph(pdata, cr, co, cb, 12, indent, p->words); |
156 | p->private_data = pdata; |
39a0cfb9 |
157 | if (pdata->first != pdata->last) { |
158 | pdata->first->penalty_after += 100000; |
159 | pdata->last->penalty_before += 100000; |
160 | } |
43341922 |
161 | break; |
162 | |
163 | /* |
164 | * All of these paragraph types require wrapping in the |
165 | * ordinary way. So we must supply a set of fonts, a |
166 | * line width and auxiliary information (e.g. bullet |
167 | * text) for each one. |
168 | */ |
169 | case para_Chapter: |
170 | case para_Appendix: |
171 | case para_UnnumberedChapter: |
172 | case para_Heading: |
173 | case para_Subsect: |
174 | case para_Normal: |
175 | case para_BiblioCited: |
176 | case para_Bullet: |
177 | case para_NumberedList: |
178 | case para_DescribedThing: |
179 | case para_Description: |
180 | case para_Copyright: |
181 | case para_Title: |
182 | pdata = mknew(para_data); |
183 | |
184 | /* |
515d216b |
185 | * Choose fonts for this paragraph. |
186 | * |
187 | * FIXME: All of this ought to be completely |
188 | * user-configurable. |
43341922 |
189 | */ |
515d216b |
190 | switch (p->type) { |
191 | case para_Title: |
192 | pdata->fonts[FONT_NORMAL] = hr; |
193 | pdata->sizes[FONT_NORMAL] = 24; |
194 | pdata->fonts[FONT_EMPH] = hi; |
195 | pdata->sizes[FONT_EMPH] = 24; |
196 | pdata->fonts[FONT_CODE] = cb; |
197 | pdata->sizes[FONT_CODE] = 24; |
198 | break; |
199 | |
200 | case para_Chapter: |
201 | case para_Appendix: |
202 | case para_UnnumberedChapter: |
203 | pdata->fonts[FONT_NORMAL] = hr; |
204 | pdata->sizes[FONT_NORMAL] = 20; |
205 | pdata->fonts[FONT_EMPH] = hi; |
206 | pdata->sizes[FONT_EMPH] = 20; |
207 | pdata->fonts[FONT_CODE] = cb; |
208 | pdata->sizes[FONT_CODE] = 20; |
209 | break; |
210 | |
211 | case para_Heading: |
212 | case para_Subsect: |
213 | pdata->fonts[FONT_NORMAL] = hr; |
214 | pdata->fonts[FONT_EMPH] = hi; |
215 | pdata->fonts[FONT_CODE] = cb; |
216 | pdata->sizes[FONT_NORMAL] = |
217 | pdata->sizes[FONT_EMPH] = |
218 | pdata->sizes[FONT_CODE] = |
219 | (p->aux == 0 ? 16 : p->aux == 1 ? 14 : 13); |
220 | break; |
221 | |
222 | case para_Normal: |
223 | case para_BiblioCited: |
224 | case para_Bullet: |
225 | case para_NumberedList: |
226 | case para_DescribedThing: |
227 | case para_Description: |
228 | case para_Copyright: |
229 | pdata->fonts[FONT_NORMAL] = tr; |
230 | pdata->sizes[FONT_NORMAL] = 12; |
231 | pdata->fonts[FONT_EMPH] = ti; |
232 | pdata->sizes[FONT_EMPH] = 12; |
233 | pdata->fonts[FONT_CODE] = cr; |
234 | pdata->sizes[FONT_CODE] = 12; |
235 | break; |
236 | } |
43341922 |
237 | |
238 | /* |
515d216b |
239 | * Also select an indentation level depending on the |
240 | * paragraph type (list paragraphs other than |
43341922 |
241 | * para_DescribedThing need extra indent). |
242 | * |
515d216b |
243 | * (FIXME: Perhaps at some point we might even arrange |
244 | * for the user to be able to request indented first |
245 | * lines in paragraphs.) |
43341922 |
246 | */ |
515d216b |
247 | if (p->type == para_Bullet || |
248 | p->type == para_NumberedList || |
249 | p->type == para_Description) { |
250 | extra_indent = firstline_indent = indent_list; |
251 | } else { |
252 | extra_indent = firstline_indent = 0; |
253 | } |
254 | |
255 | /* |
256 | * Find the auxiliary text for this paragraph. |
257 | */ |
258 | aux = aux2 = NULL; |
259 | aux_indent = 0; |
260 | |
261 | switch (p->type) { |
262 | case para_Chapter: |
263 | case para_Appendix: |
264 | case para_Heading: |
265 | case para_Subsect: |
266 | /* |
267 | * For some heading styles (FIXME: be able to |
268 | * configure which), the auxiliary text contains |
269 | * the chapter number and is arranged to be |
270 | * right-aligned a few points left of the primary |
271 | * margin. For other styles, the auxiliary text is |
272 | * the full chapter _name_ and takes up space |
273 | * within the (wrapped) chapter title, meaning that |
274 | * we must move the first line indent over to make |
275 | * space for it. |
276 | */ |
277 | if (p->type == para_Heading || p->type == para_Subsect) { |
278 | int len; |
279 | |
280 | aux = p->kwtext2; |
281 | len = paper_width_simple(pdata, p->kwtext2); |
282 | aux_indent = -len - sect_num_left_space; |
283 | } else { |
284 | aux = p->kwtext; |
285 | aux2 = mknew(word); |
286 | aux2->next = NULL; |
287 | aux2->alt = NULL; |
288 | aux2->type = word_Normal; |
289 | aux2->text = ustrdup(L": "); |
290 | aux2->breaks = FALSE; |
291 | aux2->aux = 0; |
292 | aux_indent = 0; |
293 | |
294 | firstline_indent += paper_width_simple(pdata, aux); |
295 | firstline_indent += paper_width_simple(pdata, aux2); |
296 | } |
297 | break; |
298 | |
299 | case para_Bullet: |
300 | /* |
301 | * Auxiliary text consisting of a bullet. (FIXME: |
302 | * configurable bullet.) |
303 | */ |
304 | aux = mknew(word); |
305 | aux->next = NULL; |
306 | aux->alt = NULL; |
307 | aux->type = word_Normal; |
308 | aux->text = ustrdup(L"\x2022"); |
309 | aux->breaks = FALSE; |
310 | aux->aux = 0; |
311 | aux_indent = indent + indent_list_bullet; |
312 | break; |
313 | |
314 | case para_NumberedList: |
315 | /* |
316 | * Auxiliary text consisting of the number followed |
317 | * by a (FIXME: configurable) full stop. |
318 | */ |
319 | aux = p->kwtext; |
320 | aux2 = mknew(word); |
321 | aux2->next = NULL; |
322 | aux2->alt = NULL; |
323 | aux2->type = word_Normal; |
324 | aux2->text = ustrdup(L"."); |
325 | aux2->breaks = FALSE; |
326 | aux2->aux = 0; |
327 | aux_indent = indent + indent_list_bullet; |
328 | break; |
329 | |
330 | case para_BiblioCited: |
331 | /* |
332 | * Auxiliary text consisting of the bibliography |
333 | * reference text, and a trailing space. |
334 | */ |
335 | aux = p->kwtext; |
336 | aux2 = mknew(word); |
337 | aux2->next = NULL; |
338 | aux2->alt = NULL; |
339 | aux2->type = word_Normal; |
340 | aux2->text = ustrdup(L" "); |
341 | aux2->breaks = FALSE; |
342 | aux2->aux = 0; |
343 | aux_indent = indent; |
344 | firstline_indent += paper_width_simple(pdata, aux); |
345 | firstline_indent += paper_width_simple(pdata, aux2); |
346 | break; |
347 | } |
43341922 |
348 | |
349 | wrap_paragraph(pdata, p->words, base_width, |
350 | indent + firstline_indent, |
351 | indent + extra_indent); |
352 | |
43341922 |
353 | p->private_data = pdata; |
354 | |
515d216b |
355 | pdata->first->aux_text = aux; |
356 | pdata->first->aux_text_2 = aux2; |
357 | pdata->first->aux_left_indent = aux_indent; |
358 | |
39a0cfb9 |
359 | /* |
360 | * Line breaking penalties. |
361 | */ |
362 | switch (p->type) { |
363 | case para_Chapter: |
364 | case para_Appendix: |
365 | case para_Heading: |
366 | case para_Subsect: |
367 | case para_UnnumberedChapter: |
368 | /* |
369 | * Fixed and large penalty for breaking straight |
370 | * after a heading; corresponding bonus for |
371 | * breaking straight before. |
372 | */ |
373 | pdata->first->penalty_before = -500000; |
374 | pdata->last->penalty_after = 500000; |
375 | for (ldata = pdata->first; ldata; ldata = ldata->next) |
376 | ldata->penalty_after = 500000; |
377 | break; |
378 | |
379 | case para_DescribedThing: |
380 | /* |
381 | * This is treated a bit like a small heading: |
382 | * there's a penalty for breaking after it (i.e. |
383 | * between it and its description), and a bonus for |
384 | * breaking before it (actually _between_ list |
385 | * items). |
386 | */ |
387 | pdata->first->penalty_before = -200000; |
388 | pdata->last->penalty_after = 200000; |
389 | break; |
390 | |
391 | default: |
392 | /* |
393 | * Most paragraph types: widow/orphan control by |
394 | * discouraging breaking one line from the end of |
395 | * any paragraph. |
396 | */ |
397 | if (pdata->first != pdata->last) { |
398 | pdata->first->penalty_after = 100000; |
399 | pdata->last->penalty_before = 100000; |
400 | } |
401 | break; |
402 | } |
403 | |
515d216b |
404 | break; |
405 | } |
406 | |
407 | if (p->private_data) { |
408 | pdata = (para_data *)p->private_data; |
409 | |
43341922 |
410 | /* |
411 | * Set the line spacing for each line in this paragraph. |
412 | */ |
413 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
414 | if (ldata == pdata->first) |
415 | ldata->space_before = base_para_spacing / 2; |
416 | else |
417 | ldata->space_before = base_leading / 2; |
418 | if (ldata == pdata->last) |
419 | ldata->space_after = base_para_spacing / 2; |
420 | else |
421 | ldata->space_after = base_leading / 2; |
422 | ldata->page_break = FALSE; |
423 | } |
424 | |
425 | /* |
515d216b |
426 | * Some kinds of section heading do require a page |
427 | * break before them. |
43341922 |
428 | */ |
515d216b |
429 | if (p->type == para_Title || |
430 | p->type == para_Chapter || |
431 | p->type == para_Appendix || |
432 | p->type == para_UnnumberedChapter) { |
433 | pdata->first->page_break = TRUE; |
434 | pdata->first->space_before = chapter_top_space; |
435 | } |
43341922 |
436 | |
515d216b |
437 | /* |
438 | * Link all line structures together into a big list. |
439 | */ |
43341922 |
440 | if (pdata->first) { |
441 | if (lastline) { |
442 | lastline->next = pdata->first; |
443 | pdata->first->prev = lastline; |
444 | } else { |
445 | firstline = pdata->first; |
446 | pdata->first->prev = NULL; |
447 | } |
448 | lastline = pdata->last; |
449 | } |
450 | } |
451 | } |
452 | |
453 | /* |
454 | * Now we have an enormous linked list of every line of text in |
455 | * the document. Break it up into pages. |
456 | */ |
457 | pages = page_breaks(firstline, lastline, page_height); |
458 | |
459 | /* |
460 | * Now we're ready to actually lay out the pages. We do this by |
461 | * looping over _paragraphs_, since we may need to track cross- |
462 | * references between lines and even across pages. |
463 | */ |
464 | for (p = sourceform; p; p = p->next) { |
465 | pdata = (para_data *)p->private_data; |
466 | |
467 | if (pdata) { |
138d7ffb |
468 | xref_dest dest; |
469 | dest.type = NONE; |
43341922 |
470 | for (ldata = pdata->first; ldata; ldata = ldata->next) { |
138d7ffb |
471 | render_line(ldata, left_margin, paper_height - top_margin, |
472 | &dest, keywords); |
43341922 |
473 | if (ldata == pdata->last) |
474 | break; |
475 | } |
476 | } |
477 | } |
478 | |
479 | doc = mknew(document); |
480 | doc->fonts = fontlist; |
481 | doc->pages = pages; |
482 | doc->paper_width = paper_width; |
483 | doc->paper_height = paper_height; |
484 | return doc; |
485 | } |
486 | |
487 | static font_encoding *new_font_encoding(font_data *font) |
488 | { |
489 | font_encoding *fe; |
490 | int i; |
491 | |
492 | fe = mknew(font_encoding); |
493 | fe->next = NULL; |
494 | |
495 | if (font->list->tail) |
496 | font->list->tail->next = fe; |
497 | else |
498 | font->list->head = fe; |
499 | font->list->tail = fe; |
500 | |
501 | fe->font = font; |
502 | fe->free_pos = 0x21; |
503 | |
504 | for (i = 0; i < 256; i++) { |
505 | fe->vector[i] = NULL; |
506 | fe->indices[i] = -1; |
507 | fe->to_unicode[i] = 0xFFFF; |
508 | } |
509 | |
510 | return fe; |
511 | } |
512 | |
513 | static font_data *make_std_font(font_list *fontlist, char const *name) |
514 | { |
515 | const int *widths; |
516 | int nglyphs; |
517 | font_data *f; |
518 | font_encoding *fe; |
519 | int i; |
520 | |
521 | widths = ps_std_font_widths(name); |
522 | if (!widths) |
523 | return NULL; |
524 | |
525 | for (nglyphs = 0; ps_std_glyphs[nglyphs] != NULL; nglyphs++); |
526 | |
527 | f = mknew(font_data); |
528 | |
529 | f->list = fontlist; |
530 | f->name = name; |
531 | f->nglyphs = nglyphs; |
532 | f->glyphs = ps_std_glyphs; |
533 | f->widths = widths; |
534 | f->subfont_map = mknewa(subfont_map_entry, nglyphs); |
535 | |
536 | /* |
537 | * Our first subfont will contain all of US-ASCII. This isn't |
538 | * really necessary - we could just create custom subfonts |
539 | * precisely as the whim of render_string dictated - but |
540 | * instinct suggests that it might be nice to have the text in |
541 | * the output files look _marginally_ recognisable. |
542 | */ |
543 | fe = new_font_encoding(f); |
544 | fe->free_pos = 0xA1; /* only the top half is free */ |
545 | f->latest_subfont = fe; |
546 | |
547 | for (i = 0; i < (int)lenof(f->bmp); i++) |
548 | f->bmp[i] = 0xFFFF; |
549 | |
550 | for (i = 0; i < nglyphs; i++) { |
551 | wchar_t ucs; |
552 | ucs = ps_glyph_to_unicode(f->glyphs[i]); |
553 | assert(ucs != 0xFFFF); |
554 | f->bmp[ucs] = i; |
555 | if (ucs >= 0x20 && ucs <= 0x7E) { |
556 | fe->vector[ucs] = f->glyphs[i]; |
557 | fe->indices[ucs] = i; |
558 | fe->to_unicode[ucs] = ucs; |
559 | f->subfont_map[i].subfont = fe; |
560 | f->subfont_map[i].position = ucs; |
561 | } else { |
562 | /* |
563 | * This character is not yet assigned to a subfont. |
564 | */ |
565 | f->subfont_map[i].subfont = NULL; |
566 | f->subfont_map[i].position = 0; |
567 | } |
568 | } |
569 | |
570 | return f; |
571 | } |
572 | |
573 | static int string_width(font_data *font, wchar_t const *string, int *errs) |
574 | { |
575 | int width = 0; |
576 | |
577 | if (errs) |
578 | *errs = 0; |
579 | |
580 | for (; *string; string++) { |
581 | int index; |
582 | |
583 | index = font->bmp[(unsigned short)*string]; |
584 | if (index == 0xFFFF) { |
585 | if (errs) |
586 | *errs = 1; |
587 | } else { |
588 | width += font->widths[index]; |
589 | } |
590 | } |
591 | |
592 | return width; |
593 | } |
594 | |
faad4952 |
595 | static int paper_width_internal(void *vctx, word *word, int *nspaces); |
43341922 |
596 | |
597 | struct paper_width_ctx { |
598 | int minspacewidth; |
599 | para_data *pdata; |
600 | }; |
601 | |
faad4952 |
602 | static int paper_width_list(void *vctx, word *text, word *end, int *nspaces) { |
43341922 |
603 | int w = 0; |
faad4952 |
604 | while (text && text != end) { |
605 | w += paper_width_internal(vctx, text, nspaces); |
43341922 |
606 | text = text->next; |
607 | } |
608 | return w; |
609 | } |
610 | |
faad4952 |
611 | static int paper_width_internal(void *vctx, word *word, int *nspaces) |
43341922 |
612 | { |
613 | struct paper_width_ctx *ctx = (struct paper_width_ctx *)vctx; |
614 | int style, type, findex, width, errs; |
615 | wchar_t *str; |
616 | |
617 | switch (word->type) { |
618 | case word_HyperLink: |
619 | case word_HyperEnd: |
620 | case word_UpperXref: |
621 | case word_LowerXref: |
622 | case word_XrefEnd: |
623 | case word_IndexRef: |
624 | return 0; |
625 | } |
626 | |
627 | style = towordstyle(word->type); |
628 | type = removeattr(word->type); |
629 | |
630 | findex = (style == word_Normal ? FONT_NORMAL : |
631 | style == word_Emph ? FONT_EMPH : |
632 | FONT_CODE); |
633 | |
634 | if (type == word_Normal) { |
635 | str = word->text; |
636 | } else if (type == word_WhiteSpace) { |
faad4952 |
637 | if (findex != FONT_CODE) { |
638 | if (nspaces) |
639 | (*nspaces)++; |
43341922 |
640 | return ctx->minspacewidth; |
faad4952 |
641 | } else |
43341922 |
642 | str = L" "; |
643 | } else /* if (type == word_Quote) */ { |
644 | if (word->aux == quote_Open) |
645 | str = L"\x2018"; /* FIXME: configurability! */ |
646 | else |
647 | str = L"\x2019"; /* FIXME: configurability! */ |
648 | } |
649 | |
650 | width = string_width(ctx->pdata->fonts[findex], str, &errs); |
651 | |
652 | if (errs && word->alt) |
faad4952 |
653 | return paper_width_list(vctx, word->alt, NULL, nspaces); |
43341922 |
654 | else |
655 | return ctx->pdata->sizes[findex] * width; |
656 | } |
657 | |
faad4952 |
658 | static int paper_width(void *vctx, word *word) |
659 | { |
660 | return paper_width_internal(vctx, word, NULL); |
661 | } |
662 | |
515d216b |
663 | static int paper_width_simple(para_data *pdata, word *text) |
664 | { |
665 | struct paper_width_ctx ctx; |
666 | |
667 | ctx.pdata = pdata; |
668 | ctx.minspacewidth = |
669 | (pdata->sizes[FONT_NORMAL] * |
670 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
671 | |
672 | return paper_width_list(&ctx, text, NULL, NULL); |
673 | } |
674 | |
43341922 |
675 | static void wrap_paragraph(para_data *pdata, word *words, |
676 | int w, int i1, int i2) |
677 | { |
678 | wrappedline *wrapping, *p; |
679 | int spacewidth; |
680 | struct paper_width_ctx ctx; |
681 | int line_height; |
682 | |
683 | /* |
684 | * We're going to need to store the line height in every line |
685 | * structure we generate. |
686 | */ |
687 | { |
688 | int i; |
689 | line_height = 0; |
690 | for (i = 0; i < NFONTS; i++) |
691 | if (line_height < pdata->sizes[i]) |
692 | line_height = pdata->sizes[i]; |
693 | line_height *= 4096; |
694 | } |
695 | |
696 | spacewidth = (pdata->sizes[FONT_NORMAL] * |
697 | string_width(pdata->fonts[FONT_NORMAL], L" ", NULL)); |
698 | if (spacewidth == 0) { |
699 | /* |
700 | * A font without a space?! Disturbing. I hope this never |
701 | * comes up, but I'll make a random guess anyway and set my |
702 | * space width to half the point size. |
703 | */ |
704 | spacewidth = pdata->sizes[FONT_NORMAL] * 4096 / 2; |
705 | } |
706 | |
707 | /* |
708 | * I'm going to set the _minimum_ space width to 3/5 of the |
709 | * standard one, and use the standard one as the optimum. |
710 | */ |
711 | ctx.minspacewidth = spacewidth * 3 / 5; |
712 | ctx.pdata = pdata; |
713 | |
714 | wrapping = wrap_para(words, w - i1, w - i2, paper_width, &ctx, spacewidth); |
715 | |
716 | /* |
717 | * Having done the wrapping, we now concoct a set of line_data |
718 | * structures. |
719 | */ |
720 | pdata->first = pdata->last = NULL; |
721 | |
722 | for (p = wrapping; p; p = p->next) { |
723 | line_data *ldata; |
724 | word *wd; |
725 | int len, wid, spaces; |
726 | |
727 | ldata = mknew(line_data); |
728 | |
729 | ldata->pdata = pdata; |
730 | ldata->first = p->begin; |
faad4952 |
731 | ldata->end = p->end; |
43341922 |
732 | ldata->line_height = line_height; |
733 | |
734 | ldata->xpos = (p == wrapping ? i1 : i2); |
735 | |
736 | if (pdata->last) { |
737 | pdata->last->next = ldata; |
738 | ldata->prev = pdata->last; |
739 | } else { |
740 | pdata->first = ldata; |
741 | ldata->prev = NULL; |
742 | } |
743 | ldata->next = NULL; |
744 | pdata->last = ldata; |
745 | |
43341922 |
746 | spaces = 0; |
faad4952 |
747 | len = paper_width_list(&ctx, ldata->first, ldata->end, &spaces); |
748 | wid = (p == wrapping ? w - i1 : w - i2); |
43341922 |
749 | wd = ldata->first; |
43341922 |
750 | |
faad4952 |
751 | ldata->hshortfall = wid - len; |
752 | ldata->nspaces = spaces; |
753 | /* |
754 | * This tells us how much the space width needs to |
755 | * change from _min_spacewidth. But we want to store |
756 | * its difference from the _natural_ space width, to |
757 | * make the text rendering easier. |
758 | */ |
759 | ldata->hshortfall += ctx.minspacewidth * spaces; |
760 | ldata->hshortfall -= spacewidth * spaces; |
761 | /* |
762 | * Special case: on the last line of a paragraph, we |
763 | * never stretch spaces. |
764 | */ |
765 | if (ldata->hshortfall > 0 && !p->next) |
766 | ldata->hshortfall = 0; |
43341922 |
767 | |
768 | ldata->aux_text = NULL; |
515d216b |
769 | ldata->aux_text_2 = NULL; |
43341922 |
770 | ldata->aux_left_indent = 0; |
39a0cfb9 |
771 | ldata->penalty_before = ldata->penalty_after = 0; |
43341922 |
772 | } |
773 | |
774 | } |
775 | |
776 | static page_data *page_breaks(line_data *first, line_data *last, |
777 | int page_height) |
778 | { |
779 | line_data *l, *m; |
780 | page_data *ph, *pt; |
781 | |
782 | /* |
783 | * Page breaking is done by a close analogue of the optimal |
784 | * paragraph wrapping algorithm used by wrap_para(). We work |
785 | * backwards from the end of the document line by line; for |
786 | * each line, we contemplate every possible number of lines we |
787 | * could put on a page starting with that line, determine a |
788 | * cost function for each one, add it to the pre-computed cost |
789 | * function for optimally page-breaking everything after that |
790 | * page, and pick the best option. |
791 | * |
792 | * Since my line_data structures are only used for this |
793 | * purpose, I might as well just store the algorithm data |
794 | * directly in them. |
795 | */ |
796 | |
797 | for (l = last; l; l = l->prev) { |
798 | int minheight, text = 0, space = 0; |
799 | int cost; |
800 | |
801 | l->bestcost = -1; |
802 | for (m = l; m; m = m->next) { |
803 | if (m != l && m->page_break) |
804 | break; /* we've gone as far as we can */ |
805 | |
806 | if (m != l) |
807 | space += m->prev->space_after; |
808 | if (m != l || m->page_break) |
809 | space += m->space_before; |
810 | text += m->line_height; |
811 | minheight = text + space; |
812 | |
813 | if (m != l && minheight > page_height) |
814 | break; |
815 | |
816 | /* |
817 | * Compute the cost of this arrangement, as the square |
818 | * of the amount of wasted space on the page. |
819 | * Exception: if this is the last page before a |
820 | * mandatory break or the document end, we don't |
821 | * penalise a large blank area. |
822 | */ |
823 | if (m->next && !m->next->page_break) |
824 | { |
825 | int x = page_height - minheight; |
826 | int xf; |
827 | |
828 | xf = x & 0xFF; |
829 | x >>= 8; |
830 | |
831 | cost = x*x; |
832 | cost += (x * xf) >> 8; |
833 | } else |
834 | cost = 0; |
835 | |
39a0cfb9 |
836 | if (m->next && !m->next->page_break) { |
837 | cost += m->penalty_after; |
838 | cost += m->next->penalty_before; |
839 | } |
43341922 |
840 | |
841 | if (m->next && !m->next->page_break) |
842 | cost += m->next->bestcost; |
43341922 |
843 | if (l->bestcost == -1 || l->bestcost > cost) { |
844 | /* |
845 | * This is the best option yet for this starting |
846 | * point. |
847 | */ |
848 | l->bestcost = cost; |
849 | if (m->next && !m->next->page_break) |
faad4952 |
850 | l->vshortfall = page_height - minheight; |
43341922 |
851 | else |
faad4952 |
852 | l->vshortfall = 0; |
43341922 |
853 | l->text = text; |
854 | l->space = space; |
855 | l->page_last = m; |
856 | } |
857 | } |
858 | } |
859 | |
860 | /* |
861 | * Now go through the line list forwards and assemble the |
862 | * actual pages. |
863 | */ |
864 | ph = pt = NULL; |
865 | |
866 | l = first; |
867 | while (l) { |
868 | page_data *page; |
869 | int text, space; |
870 | |
871 | page = mknew(page_data); |
872 | page->next = NULL; |
873 | page->prev = pt; |
874 | if (pt) |
875 | pt->next = page; |
876 | else |
877 | ph = page; |
878 | pt = page; |
879 | |
880 | page->first_line = l; |
881 | page->last_line = l->page_last; |
882 | |
883 | page->first_text = page->last_text = NULL; |
884 | |
138d7ffb |
885 | page->first_xref = page->last_xref = NULL; |
886 | |
43341922 |
887 | /* |
888 | * Now assign a y-coordinate to each line on the page. |
889 | */ |
890 | text = space = 0; |
891 | for (l = page->first_line; l; l = l->next) { |
892 | if (l != page->first_line) |
893 | space += l->prev->space_after; |
894 | if (l != page->first_line || l->page_break) |
895 | space += l->space_before; |
896 | text += l->line_height; |
897 | |
898 | l->page = page; |
899 | l->ypos = text + space + |
faad4952 |
900 | space * (float)page->first_line->vshortfall / |
43341922 |
901 | page->first_line->space; |
902 | |
903 | if (l == page->last_line) |
904 | break; |
905 | } |
906 | |
907 | l = page->last_line->next; |
908 | } |
909 | |
910 | return ph; |
911 | } |
912 | |
913 | static void add_string_to_page(page_data *page, int x, int y, |
914 | font_encoding *fe, int size, char *text) |
915 | { |
916 | text_fragment *frag; |
917 | |
918 | frag = mknew(text_fragment); |
919 | frag->next = NULL; |
920 | |
921 | if (page->last_text) |
922 | page->last_text->next = frag; |
923 | else |
924 | page->first_text = frag; |
925 | page->last_text = frag; |
926 | |
927 | frag->x = x; |
928 | frag->y = y; |
929 | frag->fe = fe; |
930 | frag->fontsize = size; |
931 | frag->text = dupstr(text); |
932 | } |
933 | |
934 | /* |
935 | * Returns the updated x coordinate. |
936 | */ |
937 | static int render_string(page_data *page, font_data *font, int fontsize, |
938 | int x, int y, wchar_t *str) |
939 | { |
940 | char *text; |
941 | int textpos, textwid, glyph; |
942 | font_encoding *subfont = NULL, *sf; |
943 | |
944 | text = mknewa(char, 1 + ustrlen(str)); |
945 | textpos = textwid = 0; |
946 | |
947 | while (*str) { |
948 | glyph = font->bmp[*str]; |
949 | |
950 | if (glyph == 0xFFFF) |
951 | continue; /* nothing more we can do here */ |
952 | |
953 | /* |
954 | * Find which subfont this character is going in. |
955 | */ |
956 | sf = font->subfont_map[glyph].subfont; |
957 | |
958 | if (!sf) { |
959 | int c; |
960 | |
961 | /* |
962 | * This character is not yet in a subfont. Assign one. |
963 | */ |
964 | if (font->latest_subfont->free_pos >= 0x100) |
965 | font->latest_subfont = new_font_encoding(font); |
966 | |
967 | c = font->latest_subfont->free_pos++; |
968 | if (font->latest_subfont->free_pos == 0x7F) |
969 | font->latest_subfont->free_pos = 0xA1; |
970 | |
971 | font->subfont_map[glyph].subfont = font->latest_subfont; |
972 | font->subfont_map[glyph].position = c; |
973 | font->latest_subfont->vector[c] = font->glyphs[glyph]; |
974 | font->latest_subfont->indices[c] = glyph; |
975 | font->latest_subfont->to_unicode[c] = *str; |
976 | |
977 | sf = font->latest_subfont; |
978 | } |
979 | |
980 | if (!subfont || sf != subfont) { |
981 | if (subfont) { |
982 | text[textpos] = '\0'; |
983 | add_string_to_page(page, x, y, subfont, fontsize, text); |
984 | x += textwid; |
985 | } else { |
986 | assert(textpos == 0); |
987 | } |
988 | textpos = 0; |
989 | subfont = sf; |
990 | } |
991 | |
992 | text[textpos++] = font->subfont_map[glyph].position; |
993 | textwid += font->widths[glyph] * fontsize; |
994 | |
995 | str++; |
996 | } |
997 | |
998 | if (textpos > 0) { |
999 | text[textpos] = '\0'; |
1000 | add_string_to_page(page, x, y, subfont, fontsize, text); |
1001 | x += textwid; |
1002 | } |
1003 | |
1004 | return x; |
1005 | } |
1006 | |
1007 | /* |
1008 | * Returns the updated x coordinate. |
1009 | */ |
138d7ffb |
1010 | static int render_text(page_data *page, para_data *pdata, line_data *ldata, |
1011 | int x, int y, word *text, word *text_end, xref **xr, |
1012 | int shortfall, int nspaces, int *nspace, |
1013 | keywordlist *keywords) |
43341922 |
1014 | { |
faad4952 |
1015 | while (text && text != text_end) { |
43341922 |
1016 | int style, type, findex, errs; |
1017 | wchar_t *str; |
138d7ffb |
1018 | xref_dest dest; |
43341922 |
1019 | |
1020 | switch (text->type) { |
138d7ffb |
1021 | /* |
1022 | * Start a cross-reference. |
1023 | */ |
43341922 |
1024 | case word_HyperLink: |
43341922 |
1025 | case word_UpperXref: |
1026 | case word_LowerXref: |
138d7ffb |
1027 | |
1028 | if (text->type == word_HyperLink) { |
1029 | dest.type = URL; |
1030 | dest.url = utoa_dup(text->text); |
1031 | dest.page = NULL; |
1032 | } else { |
1033 | keyword *kwl = kw_lookup(keywords, text->text); |
1034 | para_data *pdata; |
1035 | |
1036 | if (kwl) { |
1037 | assert(kwl->para->private_data); |
1038 | pdata = (para_data *) kwl->para->private_data; |
1039 | dest.type = PAGE; |
1040 | dest.page = pdata->first->page; |
1041 | dest.url = NULL; |
1042 | } else { |
1043 | /* |
1044 | * Shouldn't happen, but *shrug* |
1045 | */ |
1046 | dest.type = NONE; |
1047 | dest.page = NULL; |
1048 | dest.url = NULL; |
1049 | } |
1050 | } |
1051 | if (dest.type != NONE) { |
1052 | *xr = mknew(xref); |
1053 | (*xr)->dest = dest; /* structure copy */ |
1054 | if (page->last_xref) |
1055 | page->last_xref->next = *xr; |
1056 | else |
1057 | page->first_xref = *xr; |
1058 | page->last_xref = *xr; |
1059 | |
1060 | /* |
1061 | * FIXME: Ideally we should have, and use, some |
1062 | * vertical font metric information here so that |
1063 | * our cross-ref rectangle can take account of |
1064 | * descenders and the font's cap height. This will |
1065 | * do for the moment, but it isn't ideal. |
1066 | */ |
1067 | (*xr)->lx = (*xr)->rx = x; |
1068 | (*xr)->by = y; |
1069 | (*xr)->ty = y + ldata->line_height; |
1070 | } |
1071 | goto nextword; |
1072 | |
1073 | /* |
1074 | * Finish extending a cross-reference box. |
1075 | */ |
1076 | case word_HyperEnd: |
43341922 |
1077 | case word_XrefEnd: |
138d7ffb |
1078 | *xr = NULL; |
1079 | goto nextword; |
1080 | |
43341922 |
1081 | case word_IndexRef: |
1082 | goto nextword; |
1083 | /* |
a8070b46 |
1084 | * FIXME: we should do something with this. |
43341922 |
1085 | */ |
1086 | } |
1087 | |
1088 | style = towordstyle(text->type); |
1089 | type = removeattr(text->type); |
1090 | |
1091 | findex = (style == word_Normal ? FONT_NORMAL : |
1092 | style == word_Emph ? FONT_EMPH : |
1093 | FONT_CODE); |
1094 | |
1095 | if (type == word_Normal) { |
1096 | str = text->text; |
1097 | } else if (type == word_WhiteSpace) { |
1098 | x += pdata->sizes[findex] * |
1099 | string_width(pdata->fonts[findex], L" ", NULL); |
faad4952 |
1100 | if (nspaces && findex != FONT_CODE) { |
1101 | x += (*nspace+1) * shortfall / nspaces; |
1102 | x -= *nspace * shortfall / nspaces; |
1103 | (*nspace)++; |
1104 | } |
43341922 |
1105 | goto nextword; |
1106 | } else /* if (type == word_Quote) */ { |
1107 | if (text->aux == quote_Open) |
1108 | str = L"\x2018"; /* FIXME: configurability! */ |
1109 | else |
1110 | str = L"\x2019"; /* FIXME: configurability! */ |
1111 | } |
1112 | |
1113 | (void) string_width(pdata->fonts[findex], str, &errs); |
1114 | |
1115 | if (errs && text->alt) |
138d7ffb |
1116 | x = render_text(page, pdata, ldata, x, y, text->alt, NULL, |
1117 | xr, shortfall, nspaces, nspace, keywords); |
43341922 |
1118 | else |
1119 | x = render_string(page, pdata->fonts[findex], |
1120 | pdata->sizes[findex], x, y, str); |
1121 | |
138d7ffb |
1122 | if (*xr) |
1123 | (*xr)->rx = x; |
1124 | |
43341922 |
1125 | nextword: |
43341922 |
1126 | text = text->next; |
1127 | } |
1128 | |
1129 | return x; |
1130 | } |
1131 | |
138d7ffb |
1132 | static void render_line(line_data *ldata, int left_x, int top_y, |
1133 | xref_dest *dest, keywordlist *keywords) |
43341922 |
1134 | { |
faad4952 |
1135 | int nspace; |
138d7ffb |
1136 | xref *xr; |
1137 | |
faad4952 |
1138 | if (ldata->aux_text) { |
515d216b |
1139 | int x; |
138d7ffb |
1140 | xr = NULL; |
faad4952 |
1141 | nspace = 0; |
515d216b |
1142 | x = render_text(ldata->page, ldata->pdata, ldata, |
1143 | left_x + ldata->aux_left_indent, |
1144 | top_y - ldata->ypos, |
1145 | ldata->aux_text, NULL, &xr, 0, 0, &nspace, keywords); |
1146 | if (ldata->aux_text_2) |
1147 | render_text(ldata->page, ldata->pdata, ldata, |
1148 | x, top_y - ldata->ypos, |
1149 | ldata->aux_text_2, NULL, &xr, 0, 0, &nspace, keywords); |
faad4952 |
1150 | } |
1151 | nspace = 0; |
138d7ffb |
1152 | |
1153 | /* |
1154 | * There might be a cross-reference carried over from a |
1155 | * previous line. |
1156 | */ |
1157 | if (dest->type != NONE) { |
1158 | xr = mknew(xref); |
1159 | xr->dest = *dest; /* structure copy */ |
1160 | if (ldata->page->last_xref) |
1161 | ldata->page->last_xref->next = xr; |
1162 | else |
1163 | ldata->page->first_xref = xr; |
1164 | ldata->page->last_xref = xr; |
1165 | xr->lx = xr->rx = left_x + ldata->xpos; |
1166 | xr->by = top_y - ldata->ypos; |
1167 | xr->ty = top_y - ldata->ypos + ldata->line_height; |
1168 | } else |
1169 | xr = NULL; |
1170 | |
1171 | render_text(ldata->page, ldata->pdata, ldata, left_x + ldata->xpos, |
1172 | top_y - ldata->ypos, ldata->first, ldata->end, &xr, |
1173 | ldata->hshortfall, ldata->nspaces, &nspace, keywords); |
1174 | |
1175 | if (xr) { |
1176 | /* |
1177 | * There's a cross-reference continued on to the next line. |
1178 | */ |
1179 | *dest = xr->dest; |
1180 | } else |
1181 | dest->type = NONE; |
43341922 |
1182 | } |
515d216b |
1183 | |
1184 | static void code_paragraph(para_data *pdata, |
1185 | font_data *fn, font_data *fi, font_data *fb, |
1186 | int font_size, int indent, word *words) |
1187 | { |
1188 | /* |
1189 | * For code paragraphs, I'm going to hack grievously and |
1190 | * pretend the three normal fonts are the three code paragraph |
1191 | * fonts. |
1192 | */ |
1193 | pdata->fonts[FONT_NORMAL] = fb; |
1194 | pdata->fonts[FONT_EMPH] = fi; |
1195 | pdata->fonts[FONT_CODE] = fn; |
1196 | pdata->sizes[FONT_NORMAL] = |
1197 | pdata->sizes[FONT_EMPH] = |
1198 | pdata->sizes[FONT_CODE] = font_size; |
1199 | |
1200 | pdata->first = pdata->last = NULL; |
1201 | |
1202 | for (; words; words = words->next) { |
1203 | wchar_t *t, *e, *start; |
1204 | word *lhead = NULL, *ltail = NULL, *w; |
1205 | line_data *ldata; |
1206 | int prev = -1, curr; |
1207 | |
1208 | t = words->text; |
1209 | if (words->next && words->next->type == word_Emph) { |
1210 | e = words->next->text; |
1211 | words = words->next; |
1212 | } else |
1213 | e = NULL; |
1214 | |
1215 | start = t; |
1216 | |
1217 | while (*start) { |
1218 | while (*t) { |
1219 | if (!e || !*e) |
1220 | curr = 0; |
1221 | else if (*e == L'i') |
1222 | curr = 1; |
1223 | else if (*e == L'b') |
1224 | curr = 2; |
1225 | else |
1226 | curr = 0; |
1227 | |
1228 | if (prev < 0) |
1229 | prev = curr; |
1230 | |
1231 | if (curr != prev) |
1232 | break; |
1233 | |
1234 | t++; |
1235 | if (e && *e) |
1236 | e++; |
1237 | } |
1238 | |
1239 | /* |
1240 | * We've isolated a maximal subsequence of the line |
1241 | * which has the same emphasis. Form it into a word |
1242 | * structure. |
1243 | */ |
1244 | w = mknew(word); |
1245 | w->next = NULL; |
1246 | w->alt = NULL; |
1247 | w->type = (prev == 0 ? word_WeakCode : |
1248 | prev == 1 ? word_Emph : word_Normal); |
1249 | w->text = mknewa(wchar_t, t-start+1); |
1250 | memcpy(w->text, start, (t-start) * sizeof(wchar_t)); |
1251 | w->text[t-start] = '\0'; |
1252 | w->breaks = FALSE; |
1253 | |
1254 | if (ltail) |
1255 | ltail->next = w; |
1256 | else |
1257 | lhead = w; |
1258 | ltail = w; |
1259 | |
1260 | start = t; |
1261 | prev = -1; |
1262 | } |
1263 | |
1264 | ldata = mknew(line_data); |
1265 | |
1266 | ldata->pdata = pdata; |
1267 | ldata->first = lhead; |
1268 | ldata->end = NULL; |
1269 | ldata->line_height = font_size * 4096; |
1270 | |
1271 | ldata->xpos = indent; |
1272 | |
1273 | if (pdata->last) { |
1274 | pdata->last->next = ldata; |
1275 | ldata->prev = pdata->last; |
1276 | } else { |
1277 | pdata->first = ldata; |
1278 | ldata->prev = NULL; |
1279 | } |
1280 | ldata->next = NULL; |
1281 | pdata->last = ldata; |
1282 | |
1283 | ldata->hshortfall = 0; |
1284 | ldata->nspaces = 0; |
1285 | ldata->aux_text = NULL; |
1286 | ldata->aux_text_2 = NULL; |
1287 | ldata->aux_left_indent = 0; |
39a0cfb9 |
1288 | /* General opprobrium for breaking in a code paragraph. */ |
1289 | ldata->penalty_before = ldata->penalty_after = 50000; |
515d216b |
1290 | } |
1291 | } |