mdw@git.distorted.org.uk Git - sgt/halibut/blob - bk_pdf.c

   1 /*
   2  * PDF backend for Halibut
   3  */
   4
   5 #include <assert.h>
   6 #include "halibut.h"
   7 #include "paper.h"
   8 #include "deflate.h"
   9
  10 #define TREE_BRANCH 8                  /* max branching factor in page tree */
  11
  12 paragraph *pdf_config_filename(char *filename)
  13 {
  14     return cmdline_cfg_simple("pdf-filename", filename, NULL);
  15 }
  16
  17 typedef struct object_Tag object;
  18 typedef struct objlist_Tag objlist;
  19
  20 struct object_Tag {
  21     objlist *list;
  22     object *next;
  23     int number;
  24     rdstringc main, stream;
  25     int size, fileoff;
  26     char *final;
  27 };
  28
  29 struct objlist_Tag {
  30     int number;
  31     object *head, *tail;
  32 };
  33
  34 static object *new_object(objlist *list);
  35 static void objtext(object *o, char const *text);
  36 static void objstream(object *o, char const *text);
  37 static void pdf_string(void (*add)(object *, char const *),
  38                        object *, char const *);
  39 static void pdf_string_len(void (*add)(object *, char const *),
  40                            object *, char const *, int);
  41 static void objref(object *o, object *dest);
  42 static void objdest(object *o, page_data *p);
  43 static char *pdf_outline_convert(wchar_t *s, int *len);
  44
  45 static int is_std_font(char const *name);
  46
  47 static void make_pages_node(object *node, object *parent, page_data *first,
  48                             page_data *last, object *resources,
  49                             object *mediabox);
  50 static int make_outline(object *parent, outline_element *start, int n,
  51                         int open);
  52 static int pdf_versionid(FILE *fp, word *words);
  53
  54 void pdf_backend(paragraph *sourceform, keywordlist *keywords,
  55                  indexdata *idx, void *vdoc) {
  56     document *doc = (document *)vdoc;
  57     int font_index;
  58     font_encoding *fe;
  59     page_data *page;
  60     int pageno;
  61     FILE *fp;
  62     char *filename;
  63     paragraph *p;
  64     objlist olist;
  65     object *o, *info, *cat, *outlines, *pages, *resources, *mediabox;
  66     int fileoff;
  67
  68     IGNORE(keywords);
  69     IGNORE(idx);
  70
  71     filename = dupstr("output.pdf");
  72     for (p = sourceform; p; p = p->next) {
  73         if (p->type == para_Config) {
  74             if (!ustricmp(p->keyword, L"pdf-filename")) {
  75                 sfree(filename);
  76                 filename = dupstr(adv(p->origkeyword));
  77             }
  78         }
  79     }
  80
  81     olist.head = olist.tail = NULL;
  82     olist.number = 1;
  83
  84     {
  85         char buf[256];
  86
  87         info = new_object(&olist);
  88         objtext(info, "<<\n");
  89         if (doc->n_outline_elements > 0) {
  90             char *title;
  91             int titlelen;
  92
  93             title =
  94                pdf_outline_convert(doc->outline_elements->pdata->outline_title,
  95                                         &titlelen);
  96             objtext(info, "/Title ");
  97             pdf_string_len(objtext, info, title, titlelen);
  98             sfree(title);
  99             objtext(info, "\n");
 100         }
 101         objtext(info, "/Producer ");
 102         sprintf(buf, "Halibut, %s", version);
 103         pdf_string(objtext, info, buf);
 104         objtext(info, "\n>>\n");
 105     }
 106
 107     cat = new_object(&olist);
 108     if (doc->n_outline_elements > 0)
 109         outlines = new_object(&olist);
 110     else
 111         outlines = NULL;
 112     pages = new_object(&olist);
 113     resources = new_object(&olist);
 114
 115     /*
 116      * The catalogue just contains references to the outlines and
 117      * pages objects, and the pagelabels dictionary.
 118      */
 119     objtext(cat, "<<\n/Type /Catalog");
 120     if (outlines) {
 121         objtext(cat, "\n/Outlines ");
 122         objref(cat, outlines);
 123     }
 124     objtext(cat, "\n/Pages ");
 125     objref(cat, pages);
 126     /* Halibut just numbers pages 1, 2, 3, ... */
 127     objtext(cat, "\n/PageLabels<</Nums[0<</S/D>>]>>");
 128     if (outlines)
 129         objtext(cat, "\n/PageMode /UseOutlines");
 130     objtext(cat, "\n>>\n");
 131
 132     /*
 133      * Set up the resources dictionary, which mostly means
 134      * providing all the font objects and names to call them by.
 135      */
 136     font_index = 0;
 137     objtext(resources, "<<\n/ProcSet [/PDF/Text]\n/Font <<\n");
 138     for (fe = doc->fonts->head; fe; fe = fe->next) {
 139         char fname[40];
 140         int i, prev;
 141         object *font;
 142
 143         sprintf(fname, "f%d", font_index++);
 144         fe->name = dupstr(fname);
 145
 146         font = new_object(&olist);
 147
 148         objtext(resources, "/");
 149         objtext(resources, fe->name);
 150         objtext(resources, " ");
 151         objref(resources, font);
 152         objtext(resources, "\n");
 153
 154         objtext(font, "<<\n/Type /Font\n/Subtype /Type1\n/Name /");
 155         objtext(font, fe->name);
 156         objtext(font, "\n/BaseFont /");
 157         objtext(font, fe->font->info->name);
 158         objtext(font, "\n/Encoding <<\n/Type /Encoding\n/Differences [");
 159
 160         for (i = 0; i < 256; i++) {
 161             char buf[20];
 162             if (!fe->vector[i])
 163                 continue;
 164             if (i != prev + 1) {
 165                 sprintf(buf, "\n%d", i);
 166                 objtext(font, buf);
 167             }
 168             objtext(font, i % 8 ? "/" : "\n/");
 169             objtext(font, fe->vector[i]);
 170             prev = i;
 171         }
 172
 173         objtext(font, "\n]\n>>\n");
 174
 175 #define FF_FIXEDPITCH   0x00000001
 176 #define FF_SERIF        0x00000002
 177 #define FF_SYMBOLIC     0x00000004
 178 #define FF_SCRIPT       0x00000008
 179 #define FF_NONSYMBOLIC  0x00000020
 180 #define FF_ITALIC       0x00000040
 181 #define FF_ALLCAP       0x00010000
 182 #define FF_SMALLCAP     0x00020000
 183 #define FF_FORCEBOLD    0x00040000
 184
 185         if (!is_std_font(fe->font->info->name)){
 186             object *widths = new_object(&olist);
 187             object *fontdesc = new_object(&olist);
 188             int firstchar = -1, lastchar = -1;
 189             char buf[80];
 190             font_info const *fi = fe->font->info;
 191             int flags;
 192             for (i = 0; i < 256; i++)
 193                 if (fe->indices[i] >= 0) {
 194                     if (firstchar < 0) firstchar = i;
 195                     lastchar = i;
 196                 }
 197             sprintf(buf, "/FirstChar %d\n/LastChar %d\n/Widths ",
 198                     firstchar, lastchar);
 199             objtext(font, buf);
 200             objref(font, widths);
 201             objtext(font, "\n");
 202             objtext(widths, "[\n");
 203             for (i = firstchar; i <= lastchar; i++) {
 204                 double width;
 205                 if (fe->indices[i] < 0)
 206                     width = 0.0;
 207                 else
 208                     width = fi->widths[fe->indices[i]];
 209                 sprintf(buf, "%g\n", 1000.0 * width / FUNITS_PER_PT);
 210                 objtext(widths, buf);
 211             }
 212             objtext(widths, "]\n");
 213             objtext(font, "/FontDescriptor ");
 214             objref(font, fontdesc);
 215             objtext(fontdesc, "<<\n/Type /FontDescriptor\n/Name /");
 216             objtext(fontdesc, fi->name);
 217             flags = 0;
 218             if (fi->italicangle) flags |= FF_ITALIC;
 219             flags |= FF_NONSYMBOLIC;
 220             sprintf(buf, "\n/Flags %d\n", flags);
 221             objtext(fontdesc, buf);
 222             sprintf(buf, "/FontBBox [%g %g %g %g]\n", fi->fontbbox[0],
 223                     fi->fontbbox[1], fi->fontbbox[2], fi->fontbbox[3]);
 224             objtext(fontdesc, buf);
 225             sprintf(buf, "/ItalicAngle %g\n", fi->italicangle);
 226             objtext(fontdesc, buf);
 227             sprintf(buf, "/Ascent %g\n", fi->ascent);
 228             objtext(fontdesc, buf);
 229             sprintf(buf, "/Descent %g\n", fi->descent);
 230             objtext(fontdesc, buf);
 231             sprintf(buf, "/CapHeight %g\n", fi->capheight);
 232             objtext(fontdesc, buf);
 233             sprintf(buf, "/XHeight %g\n", fi->xheight);
 234             objtext(fontdesc, buf);
 235             sprintf(buf, "/StemH %g\n", fi->stemh);
 236             objtext(fontdesc, buf);
 237             sprintf(buf, "/StemV %g\n", fi->stemv);
 238             objtext(fontdesc, buf);
 239             if (fi->fp) {
 240                 object *fontfile = new_object(&olist);
 241                 char buf[513];
 242                 size_t len;
 243                 rewind(fi->fp);
 244                 do {
 245                     len = fread(buf, 1, sizeof(buf)-1, fi->fp);
 246                     buf[len] = 0;
 247                     objstream(fontfile, buf);
 248                 } while (len == sizeof(buf)-1);
 249                 objtext(fontdesc, "/FontFile ");
 250                 objref(fontdesc, fontfile);
 251             }
 252             objtext(fontdesc, "\n>>\n");
 253         }
 254
 255         objtext(font, "\n>>\n");
 256     }
 257     objtext(resources, ">>\n>>\n");
 258
 259     {
 260         char buf[255];
 261         mediabox = new_object(&olist);
 262         sprintf(buf, "[0 0 %g %g]\n",
 263                 doc->paper_width / FUNITS_PER_PT,
 264                 doc->paper_height / FUNITS_PER_PT);
 265         objtext(mediabox, buf);
 266     }
 267
 268     /*
 269      * Define the page objects for each page, and get each one
 270      * ready to have a `Parent' specification added to it.
 271      */
 272     for (page = doc->pages; page; page = page->next) {
 273         object *opage;
 274
 275         opage = new_object(&olist);
 276         page->spare = opage;
 277         objtext(opage, "<<\n/Type /Page\n");
 278     }
 279
 280     /*
 281      * Recursively build the page tree.
 282      */
 283     make_pages_node(pages, NULL, doc->pages, NULL, resources, mediabox);
 284
 285     /*
 286      * Create and render the individual pages.
 287      */
 288     pageno = 0;
 289     for (page = doc->pages; page; page = page->next) {
 290         object *opage, *cstr;
 291         rect *r;
 292         text_fragment *frag, *frag_end;
 293         char buf[256];
 294         int x, y, lx, ly;
 295
 296         opage = (object *)page->spare;
 297         /*
 298          * At this point the page dictionary is already
 299          * half-written, with /Type and /Parent already present. We
 300          * continue from there.
 301          */
 302
 303         /*
 304          * The PDF spec says /Resources is required, but also says
 305          * that it's inheritable and may be omitted if it's present
 306          * in a Pages node. In our case it is: it's present in the
 307          * topmost /Pages node because we carefully put it there.
 308          * So we don't need a /Resources entry here.  The same applies
 309          * to /MediaBox.
 310          */
 311
 312         /*
 313          * Now we're ready to define a content stream containing
 314          * the actual text on the page.
 315          */
 316         cstr = new_object(&olist);
 317         objtext(opage, "/Contents ");
 318         objref(opage, cstr);
 319         objtext(opage, "\n");
 320
 321         /*
 322          * Render any rectangles on the page.
 323          */
 324         for (r = page->first_rect; r; r = r->next) {
 325             char buf[512];
 326             sprintf(buf, "%g %g %g %g re f\n",
 327                     r->x / FUNITS_PER_PT, r->y / FUNITS_PER_PT,
 328                     r->w / FUNITS_PER_PT, r->h / FUNITS_PER_PT);
 329             objstream(cstr, buf);
 330         }
 331
 332         objstream(cstr, "BT\n");
 333
 334         /*
 335          * PDF tracks two separate current positions: the position
 336          * given in the `line matrix' and the position given in the
 337          * `text matrix'. We must therefore track both as well.
 338          * They start off at -1 (unset).
 339          */
 340         lx = ly = -1;
 341         x = y = -1;
 342
 343         frag = page->first_text;
 344         while (frag) {
 345             /*
 346              * For compactness, I'm going to group text fragments
 347              * into subsequences that use the same font+size. So
 348              * first find the end of this subsequence.
 349              */
 350             for (frag_end = frag;
 351                  (frag_end &&
 352                   frag_end->fe == frag->fe &&
 353                   frag_end->fontsize == frag->fontsize);
 354                  frag_end = frag_end->next);
 355
 356             /*
 357              * Now select the text fragment, and prepare to display
 358              * the text.
 359              */
 360             objstream(cstr, "/");
 361             objstream(cstr, frag->fe->name);
 362             sprintf(buf, " %d Tf ", frag->fontsize);
 363             objstream(cstr, buf);
 364
 365             while (frag && frag != frag_end) {
 366                 /*
 367                  * Place the text position for the first piece of
 368                  * text.
 369                  */
 370                 if (lx < 0) {
 371                     sprintf(buf, "1 0 0 1 %g %g Tm ",
 372                             frag->x/FUNITS_PER_PT, frag->y/FUNITS_PER_PT);
 373                 } else {
 374                     sprintf(buf, "%g %g Td ",
 375                             (frag->x - lx)/FUNITS_PER_PT,
 376                             (frag->y - ly)/FUNITS_PER_PT);
 377                 }
 378                 objstream(cstr, buf);
 379                 lx = x = frag->x;
 380                 ly = y = frag->y;
 381
 382                 /*
 383                  * See if we're going to use Tj (show a single
 384                  * string) or TJ (show an array of strings with
 385                  * x-spacings between them). We determine this by
 386                  * seeing if there's more than one text fragment in
 387                  * sequence with the same y-coordinate.
 388                  */
 389                 if (frag->next && frag->next != frag_end &&
 390                     frag->next->y == y) {
 391                     /*
 392                      * The TJ strategy.
 393                      */
 394                     objstream(cstr, "[");
 395                     while (frag && frag != frag_end && frag->y == y) {
 396                         if (frag->x != x) {
 397                             sprintf(buf, "%g",
 398                                     (x - frag->x) * 1000.0 /
 399                                     (FUNITS_PER_PT * frag->fontsize));
 400                             objstream(cstr, buf);
 401                         }
 402                         pdf_string(objstream, cstr, frag->text);
 403                         x = frag->x + frag->width;
 404                         frag = frag->next;
 405                     }
 406                     objstream(cstr, "]TJ\n");
 407                 } else
 408                 {
 409                     /*
 410                      * The Tj strategy.
 411                      */
 412                     pdf_string(objstream, cstr, frag->text);
 413                     objstream(cstr, "Tj\n");
 414                     frag = frag->next;
 415                 }
 416             }
 417         }
 418         objstream(cstr, "ET");
 419
 420         /*
 421          * Also, we want an annotation dictionary containing the
 422          * cross-references from this page.
 423          */
 424         if (page->first_xref) {
 425             xref *xr;
 426             objtext(opage, "/Annots [\n");
 427
 428             for (xr = page->first_xref; xr; xr = xr->next) {
 429                 char buf[256];
 430
 431                 objtext(opage, "<</Subtype/Link\n/Rect[");
 432                 sprintf(buf, "%g %g %g %g",
 433                         xr->lx / FUNITS_PER_PT, xr->by / FUNITS_PER_PT,
 434                         xr->rx / FUNITS_PER_PT, xr->ty / FUNITS_PER_PT);
 435                 objtext(opage, buf);
 436                 objtext(opage, "]/Border[0 0 0]\n");
 437
 438                 if (xr->dest.type == PAGE) {
 439                     objtext(opage, "/Dest");
 440                     objdest(opage, xr->dest.page);
 441                 } else {
 442                     objtext(opage, "/A<</S/URI/URI");
 443                     pdf_string(objtext, opage, xr->dest.url);
 444                     objtext(opage, ">>");
 445                 }
 446
 447                 objtext(opage, ">>\n");
 448             }
 449
 450             objtext(opage, "]\n");
 451         }
 452
 453         objtext(opage, ">>\n");
 454     }
 455
 456     /*
 457      * Set up the outlines dictionary.
 458      */
 459     if (outlines) {
 460         int topcount;
 461         char buf[80];
 462
 463         objtext(outlines, "<<\n/Type /Outlines\n");
 464         topcount = make_outline(outlines, doc->outline_elements,
 465                                 doc->n_outline_elements, TRUE);
 466         sprintf(buf, "/Count %d\n>>\n", topcount);
 467         objtext(outlines, buf);
 468     }
 469
 470     /*
 471      * Assemble the final linear form of every object.
 472      */
 473     for (o = olist.head; o; o = o->next) {
 474         rdstringc rs = {0, 0, NULL};
 475         char text[80];
 476         deflate_compress_ctx *zcontext;
 477         void *zbuf;
 478         int zlen;
 479
 480         sprintf(text, "%d 0 obj\n", o->number);
 481         rdaddsc(&rs, text);
 482
 483         if (!o->main.text && o->stream.text) {
 484             zcontext = deflate_compress_new(DEFLATE_TYPE_ZLIB);
 485             deflate_compress_data(zcontext, o->stream.text, o->stream.pos,
 486                                   DEFLATE_END_OF_DATA, &zbuf, &zlen);
 487             deflate_compress_free(zcontext);
 488             sprintf(text, "<<\n/Filter/FlateDecode\n/Length %d\n>>\n", zlen);
 489             rdaddsc(&o->main, text);
 490         }
 491
 492         assert(o->main.text);
 493         rdaddsc(&rs, o->main.text);
 494         sfree(o->main.text);
 495
 496         if (rs.text[rs.pos-1] != '\n')
 497             rdaddc(&rs, '\n');
 498
 499         if (o->stream.text) {
 500             rdaddsc(&rs, "stream\n");
 501             rdaddsn(&rs, zbuf, zlen);
 502             rdaddsc(&rs, "\nendstream\n");
 503             sfree(o->stream.text);
 504             sfree(zbuf);
 505         }
 506
 507         rdaddsc(&rs, "endobj\n");
 508
 509         o->final = rs.text;
 510         o->size = rs.pos;
 511     }
 512
 513     /*
 514      * Write out the PDF file.
 515      */
 516
 517     fp = fopen(filename, "wb");
 518     if (!fp) {
 519         error(err_cantopenw, filename);
 520         return;
 521     }
 522
 523     /*
 524      * Header. I'm going to put the version IDs in the header as
 525      * well, simply in PDF comments.  The PDF Reference also suggests
 526      * that binary PDF files contain four top-bit-set characters in
 527      * the second line.
 528      */
 529     fileoff = fprintf(fp, "%%PDF-1.3\n%% L\xc3\xba\xc3\xb0""a\n");
 530     for (p = sourceform; p; p = p->next)
 531         if (p->type == para_VersionID)
 532             fileoff += pdf_versionid(fp, p->words);
 533
 534     /*
 535      * Body
 536      */
 537     for (o = olist.head; o; o = o->next) {
 538         o->fileoff = fileoff;
 539         fwrite(o->final, 1, o->size, fp);
 540         fileoff += o->size;
 541     }
 542
 543     /*
 544      * Cross-reference table
 545      */
 546     fprintf(fp, "xref\n");
 547     assert(olist.head->number == 1);
 548     fprintf(fp, "0 %d\n", olist.tail->number + 1);
 549     fprintf(fp, "0000000000 65535 f \n");
 550     for (o = olist.head; o; o = o->next) {
 551         char entry[40];
 552         sprintf(entry, "%010d 00000 n \n", o->fileoff);
 553         assert(strlen(entry) == 20);
 554         fputs(entry, fp);
 555     }
 556
 557     /*
 558      * Trailer
 559      */
 560     fprintf(fp, "trailer\n<<\n/Size %d\n/Root %d 0 R\n/Info %d 0 R\n>>\n",
 561             olist.tail->number + 1, cat->number, info->number);
 562     fprintf(fp, "startxref\n%d\n%%%%EOF\n", fileoff);
 563
 564     fclose(fp);
 565
 566     sfree(filename);
 567 }
 568
 569 static object *new_object(objlist *list)
 570 {
 571     object *obj = snew(object);
 572
 573     obj->list = list;
 574
 575     obj->main.text = NULL;
 576     obj->main.pos = obj->main.size = 0;
 577     obj->stream.text = NULL;
 578     obj->stream.pos = obj->stream.size = 0;
 579
 580     obj->number = list->number++;
 581
 582     obj->next = NULL;
 583     if (list->tail)
 584         list->tail->next = obj;
 585     else
 586         list->head = obj;
 587     list->tail = obj;
 588
 589     obj->size = 0;
 590     obj->final = NULL;
 591
 592     return obj;
 593 }
 594
 595 static void objtext(object *o, char const *text)
 596 {
 597     rdaddsc(&o->main, text);
 598 }
 599
 600 static void objstream(object *o, char const *text)
 601 {
 602     rdaddsc(&o->stream, text);
 603 }
 604
 605 static void objref(object *o, object *dest)
 606 {
 607     char buf[40];
 608     sprintf(buf, "%d 0 R", dest->number);
 609     rdaddsc(&o->main, buf);
 610 }
 611
 612 static void objdest(object *o, page_data *p) {
 613     objtext(o, "[");
 614     objref(o, (object *)p->spare);
 615     objtext(o, "/XYZ null null null]");
 616 }
 617
 618 static char const * const stdfonts[] = {
 619     "Times-Roman", "Times-Bold", "Times-Italic", "Times-BoldItalic",
 620     "Helvetica", "Helvetica-Bold", "Helvetica-Oblique","Helvetica-BoldOblique",
 621     "Courier", "Courier-Bold", "Courier-Oblique", "Courier-BoldOblique",
 622     "Symbol", "ZapfDingbats"
 623 };
 624
 625 static int is_std_font(char const *name) {
 626     unsigned i;
 627     for (i = 0; i < lenof(stdfonts); i++)
 628         if (strcmp(name, stdfonts[i]) == 0)
 629             return TRUE;
 630     return FALSE;
 631 }
 632
 633 static void make_pages_node(object *node, object *parent, page_data *first,
 634                             page_data *last, object *resources,
 635                             object *mediabox)
 636 {
 637     int count;
 638     page_data *page;
 639     char buf[80];
 640
 641     objtext(node, "<<\n/Type /Pages\n");
 642     if (parent) {
 643         objtext(node, "/Parent ");
 644         objref(node, parent);
 645         objtext(node, "\n");
 646     }
 647
 648     /*
 649      * Count the pages in this stretch, to see if there are few
 650      * enough to reference directly.
 651      */
 652     count = 0;
 653     for (page = first; page; page = page->next) {
 654         count++;
 655         if (page == last)
 656             break;
 657     }
 658
 659     sprintf(buf, "/Count %d\n/Kids [\n", count);
 660     objtext(node, buf);
 661
 662     if (count > TREE_BRANCH) {
 663         int i;
 664         page_data *thisfirst, *thislast;
 665
 666         page = first;
 667
 668         for (i = 0; i < TREE_BRANCH; i++) {
 669             int number = (i+1) * count / TREE_BRANCH - i * count / TREE_BRANCH;
 670             thisfirst = page;
 671             while (number--) {
 672                 thislast = page;
 673                 page = page->next;
 674             }
 675
 676             if (thisfirst == thislast) {
 677                 objref(node, (object *)thisfirst->spare);
 678                 objtext((object *)thisfirst->spare, "/Parent ");
 679                 objref((object *)thisfirst->spare, node);
 680                 objtext((object *)thisfirst->spare, "\n");
 681             } else {
 682                 object *newnode = new_object(node->list);
 683                 make_pages_node(newnode, node, thisfirst, thislast,
 684                                 NULL, NULL);
 685                 objref(node, newnode);
 686             }
 687             objtext(node, "\n");
 688         }
 689
 690         assert(thislast == last || page == NULL);
 691
 692     } else {
 693         for (page = first; page; page = page->next) {
 694             objref(node, (object *)page->spare);
 695             objtext(node, "\n");
 696             objtext((object *)page->spare, "/Parent ");
 697             objref((object *)page->spare, node);
 698             objtext((object *)page->spare, "\n");
 699             if (page == last)
 700                 break;
 701         }
 702     }
 703
 704     objtext(node, "]\n");
 705
 706     if (resources) {
 707         objtext(node, "/Resources ");
 708         objref(node, resources);
 709         objtext(node, "\n");
 710     }
 711     if (mediabox) {
 712         objtext(node, "/MediaBox ");
 713         objref(node, mediabox);
 714         objtext(node, "\n");
 715     }
 716
 717     objtext(node, ">>\n");
 718 }
 719
 720 /*
 721  * In text on the page, PDF uses the PostScript font model, which
 722  * means that glyphs are identified by PS strings and hence font
 723  * encoding can be managed independently of the supplied encoding
 724  * of the font. However, in the document outline, the PDF spec
 725  * encodes in either PDFDocEncoding (a custom superset of
 726  * ISO-8859-1) or UTF-16BE.
 727  */
 728 static char *pdf_outline_convert(wchar_t *s, int *len) {
 729     char *ret;
 730
 731     ret = utoa_careful_dup(s, CS_PDF);
 732
 733     /*
 734      * Very silly special case: if the returned string begins with
 735      * FE FF, then the PDF reader will mistake it for a UTF-16BE
 736      * string. So in this case we give up on PDFDocEncoding and
 737      * encode it in UTF-16 straight away.
 738      */
 739     if (ret && ret[0] == '\xFE' && ret[1] == '\xFF') {
 740         sfree(ret);
 741         ret = NULL;
 742     }
 743
 744     if (!ret) {
 745         ret = utoa_dup_len(s, CS_UTF16BE, len);
 746     } else {
 747         *len = strlen(ret);
 748     }
 749
 750     return ret;
 751 }
 752
 753 static int make_outline(object *parent, outline_element *items, int n,
 754                         int open)
 755 {
 756     int level, totalcount = 0;
 757     outline_element *itemp;
 758     object *curr, *prev = NULL, *first = NULL, *last = NULL;
 759
 760     assert(n > 0);
 761
 762     level = items->level;
 763
 764     while (n > 0) {
 765         char *title;
 766         int titlelen;
 767
 768         /*
 769          * Here we expect to be sitting on an item at the given
 770          * level. So we start by constructing an outline entry for
 771          * that item.
 772          */
 773         assert(items->level == level);
 774
 775         title = pdf_outline_convert(items->pdata->outline_title, &titlelen);
 776
 777         totalcount++;
 778         curr = new_object(parent->list);
 779         if (!first) first = curr;
 780         last = curr;
 781         objtext(curr, "<<\n/Title ");
 782         pdf_string_len(objtext, curr, title, titlelen);
 783         sfree(title);
 784         objtext(curr, "\n/Parent ");
 785         objref(curr, parent);
 786         objtext(curr, "\n/Dest");
 787         objdest(curr, items->pdata->first->page);
 788         objtext(curr, "\n");
 789         if (prev) {
 790             objtext(curr, "/Prev ");
 791             objref(curr, prev);
 792             objtext(curr, "\n");
 793
 794             objtext(prev, "/Next ");
 795             objref(prev, curr);
 796             objtext(prev, "\n>>\n");
 797         }
 798         prev = curr;
 799
 800         items++, n--;
 801         for (itemp = items; itemp < items+n && itemp->level > level;
 802              itemp++);
 803
 804         if (itemp > items) {
 805             char buf[80];
 806             int count = make_outline(curr, items, itemp - items, FALSE);
 807             if (!open)
 808                 count = -count;
 809             else
 810                 totalcount += count;
 811             sprintf(buf, "/Count %d\n", count);
 812             objtext(curr, buf);
 813         }
 814
 815         n -= itemp - items;
 816         items = itemp;
 817     }
 818     objtext(prev, ">>\n");
 819
 820     assert(first && last);
 821     objtext(parent, "/First ");
 822     objref(parent, first);
 823     objtext(parent, "\n/Last ");
 824     objref(parent, last);
 825     objtext(parent, "\n");
 826
 827     return totalcount;
 828 }
 829
 830 static int pdf_versionid(FILE *fp, word *words)
 831 {
 832     int ret;
 833
 834     ret = fprintf(fp, "%% ");
 835
 836     for (; words; words = words->next) {
 837         char *text;
 838         int type;
 839
 840         switch (words->type) {
 841           case word_HyperLink:
 842           case word_HyperEnd:
 843           case word_UpperXref:
 844           case word_LowerXref:
 845           case word_XrefEnd:
 846           case word_IndexRef:
 847             continue;
 848         }
 849
 850         type = removeattr(words->type);
 851
 852         switch (type) {
 853           case word_Normal:
 854             text = utoa_dup(words->text, CS_ASCII);
 855             break;
 856           case word_WhiteSpace:
 857             text = dupstr(" ");
 858             break;
 859           case word_Quote:
 860             text = dupstr("'");
 861             break;
 862         }
 863
 864         fputs(text, fp);
 865         ret += strlen(text);
 866         sfree(text);
 867     }
 868
 869     ret += fprintf(fp, "\n");
 870
 871     return ret;
 872 }
 873
 874 static void pdf_string_len(void (*add)(object *, char const *),
 875                            object *o, char const *str, int len)
 876 {
 877     char const *p;
 878
 879     add(o, "(");
 880     for (p = str; len > 0; p++, len--) {
 881         char c[10];
 882         if (*p < ' ' || *p > '~') {
 883             sprintf(c, "\\%03o", 0xFF & (int)*p);
 884         } else {
 885             int n = 0;
 886             if (*p == '\\' || *p == '(' || *p == ')')
 887                 c[n++] = '\\';
 888             c[n++] = *p;
 889             c[n] = '\0';
 890         }
 891         add(o, c);
 892     }
 893     add(o, ")");
 894 }
 895
 896 static void pdf_string(void (*add)(object *, char const *),
 897                        object *o, char const *str)
 898 {
 899     pdf_string_len(add, o, str, strlen(str));
 900 }