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