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