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 objstream_len(object *o, char const *text, size_t len);
  38 static void pdf_string(void (*add)(object *, char const *),
  39                        object *, char const *);
  40 static void pdf_string_len(void (*add)(object *, char const *),
  41                            object *, char const *, int);
  42 static void objref(object *o, object *dest);
  43 static void objdest(object *o, page_data *p);
  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                 size_t len;
 242                 char *ffbuf;
 243
 244                 pf_part1((font_info *)fi, &ffbuf, &len);
 245                 objstream_len(fontfile, ffbuf, len);
 246                 sfree(ffbuf);
 247                 sprintf(buf, "<<\n/Length1 %lu\n", (unsigned long)len);
 248                 objtext(fontfile, buf);
 249                 pf_part2((font_info *)fi, &ffbuf, &len);
 250                 objstream_len(fontfile, ffbuf, len);
 251                 sfree(ffbuf);
 252                 sprintf(buf, "/Length2 %lu\n", (unsigned long)len);
 253                 objtext(fontfile, buf);
 254                 objtext(fontfile, "/Length3 0\n");
 255                 objtext(fontdesc, "/FontFile ");
 256                 objref(fontdesc, fontfile);
 257             }
 258             objtext(fontdesc, "\n>>\n");
 259         }
 260
 261         objtext(font, "\n>>\n");
 262     }
 263     objtext(resources, ">>\n>>\n");
 264
 265     {
 266         char buf[255];
 267         mediabox = new_object(&olist);
 268         sprintf(buf, "[0 0 %g %g]\n",
 269                 doc->paper_width / FUNITS_PER_PT,
 270                 doc->paper_height / FUNITS_PER_PT);
 271         objtext(mediabox, buf);
 272     }
 273
 274     /*
 275      * Define the page objects for each page, and get each one
 276      * ready to have a `Parent' specification added to it.
 277      */
 278     for (page = doc->pages; page; page = page->next) {
 279         object *opage;
 280
 281         opage = new_object(&olist);
 282         page->spare = opage;
 283         objtext(opage, "<<\n/Type /Page\n");
 284     }
 285
 286     /*
 287      * Recursively build the page tree.
 288      */
 289     make_pages_node(pages, NULL, doc->pages, NULL, resources, mediabox);
 290
 291     /*
 292      * Create and render the individual pages.
 293      */
 294     pageno = 0;
 295     for (page = doc->pages; page; page = page->next) {
 296         object *opage, *cstr;
 297         rect *r;
 298         text_fragment *frag, *frag_end;
 299         char buf[256];
 300         int x, y, lx, ly;
 301
 302         opage = (object *)page->spare;
 303         /*
 304          * At this point the page dictionary is already
 305          * half-written, with /Type and /Parent already present. We
 306          * continue from there.
 307          */
 308
 309         /*
 310          * The PDF spec says /Resources is required, but also says
 311          * that it's inheritable and may be omitted if it's present
 312          * in a Pages node. In our case it is: it's present in the
 313          * topmost /Pages node because we carefully put it there.
 314          * So we don't need a /Resources entry here.  The same applies
 315          * to /MediaBox.
 316          */
 317
 318         /*
 319          * Now we're ready to define a content stream containing
 320          * the actual text on the page.
 321          */
 322         cstr = new_object(&olist);
 323         objtext(opage, "/Contents ");
 324         objref(opage, cstr);
 325         objtext(opage, "\n");
 326
 327         /*
 328          * Render any rectangles on the page.
 329          */
 330         for (r = page->first_rect; r; r = r->next) {
 331             char buf[512];
 332             sprintf(buf, "%g %g %g %g re f\n",
 333                     r->x / FUNITS_PER_PT, r->y / FUNITS_PER_PT,
 334                     r->w / FUNITS_PER_PT, r->h / FUNITS_PER_PT);
 335             objstream(cstr, buf);
 336         }
 337
 338         objstream(cstr, "BT\n");
 339
 340         /*
 341          * PDF tracks two separate current positions: the position
 342          * given in the `line matrix' and the position given in the
 343          * `text matrix'. We must therefore track both as well.
 344          * They start off at -1 (unset).
 345          */
 346         lx = ly = -1;
 347         x = y = -1;
 348
 349         frag = page->first_text;
 350         while (frag) {
 351             /*
 352              * For compactness, I'm going to group text fragments
 353              * into subsequences that use the same font+size. So
 354              * first find the end of this subsequence.
 355              */
 356             for (frag_end = frag;
 357                  (frag_end &&
 358                   frag_end->fe == frag->fe &&
 359                   frag_end->fontsize == frag->fontsize);
 360                  frag_end = frag_end->next);
 361
 362             /*
 363              * Now select the text fragment, and prepare to display
 364              * the text.
 365              */
 366             objstream(cstr, "/");
 367             objstream(cstr, frag->fe->name);
 368             sprintf(buf, " %d Tf ", frag->fontsize);
 369             objstream(cstr, buf);
 370
 371             while (frag && frag != frag_end) {
 372                 /*
 373                  * Place the text position for the first piece of
 374                  * text.
 375                  */
 376                 if (lx < 0) {
 377                     sprintf(buf, "1 0 0 1 %g %g Tm ",
 378                             frag->x/FUNITS_PER_PT, frag->y/FUNITS_PER_PT);
 379                 } else {
 380                     sprintf(buf, "%g %g Td ",
 381                             (frag->x - lx)/FUNITS_PER_PT,
 382                             (frag->y - ly)/FUNITS_PER_PT);
 383                 }
 384                 objstream(cstr, buf);
 385                 lx = x = frag->x;
 386                 ly = y = frag->y;
 387
 388                 /*
 389                  * See if we're going to use Tj (show a single
 390                  * string) or TJ (show an array of strings with
 391                  * x-spacings between them). We determine this by
 392                  * seeing if there's more than one text fragment in
 393                  * sequence with the same y-coordinate.
 394                  */
 395                 if (frag->next && frag->next != frag_end &&
 396                     frag->next->y == y) {
 397                     /*
 398                      * The TJ strategy.
 399                      */
 400                     objstream(cstr, "[");
 401                     while (frag && frag != frag_end && frag->y == y) {
 402                         if (frag->x != x) {
 403                             sprintf(buf, "%g",
 404                                     (x - frag->x) * 1000.0 /
 405                                     (FUNITS_PER_PT * frag->fontsize));
 406                             objstream(cstr, buf);
 407                         }
 408                         pdf_string(objstream, cstr, frag->text);
 409                         x = frag->x + frag->width;
 410                         frag = frag->next;
 411                     }
 412                     objstream(cstr, "]TJ\n");
 413                 } else
 414                 {
 415                     /*
 416                      * The Tj strategy.
 417                      */
 418                     pdf_string(objstream, cstr, frag->text);
 419                     objstream(cstr, "Tj\n");
 420                     frag = frag->next;
 421                 }
 422             }
 423         }
 424         objstream(cstr, "ET");
 425
 426         /*
 427          * Also, we want an annotation dictionary containing the
 428          * cross-references from this page.
 429          */
 430         if (page->first_xref) {
 431             xref *xr;
 432             objtext(opage, "/Annots [\n");
 433
 434             for (xr = page->first_xref; xr; xr = xr->next) {
 435                 char buf[256];
 436
 437                 objtext(opage, "<</Subtype/Link\n/Rect[");
 438                 sprintf(buf, "%g %g %g %g",
 439                         xr->lx / FUNITS_PER_PT, xr->by / FUNITS_PER_PT,
 440                         xr->rx / FUNITS_PER_PT, xr->ty / FUNITS_PER_PT);
 441                 objtext(opage, buf);
 442                 objtext(opage, "]/Border[0 0 0]\n");
 443
 444                 if (xr->dest.type == PAGE) {
 445                     objtext(opage, "/Dest");
 446                     objdest(opage, xr->dest.page);
 447                 } else {
 448                     objtext(opage, "/A<</S/URI/URI");
 449                     pdf_string(objtext, opage, xr->dest.url);
 450                     objtext(opage, ">>");
 451                 }
 452
 453                 objtext(opage, ">>\n");
 454             }
 455
 456             objtext(opage, "]\n");
 457         }
 458
 459         objtext(opage, ">>\n");
 460     }
 461
 462     /*
 463      * Set up the outlines dictionary.
 464      */
 465     if (outlines) {
 466         int topcount;
 467         char buf[80];
 468
 469         objtext(outlines, "<<\n/Type /Outlines\n");
 470         topcount = make_outline(outlines, doc->outline_elements,
 471                                 doc->n_outline_elements, TRUE);
 472         sprintf(buf, "/Count %d\n>>\n", topcount);
 473         objtext(outlines, buf);
 474     }
 475
 476     /*
 477      * Assemble the final linear form of every object.
 478      */
 479     for (o = olist.head; o; o = o->next) {
 480         rdstringc rs = {0, 0, NULL};
 481         char text[80];
 482         deflate_compress_ctx *zcontext;
 483         void *zbuf;
 484         int zlen;
 485
 486         sprintf(text, "%d 0 obj\n", o->number);
 487         rdaddsc(&rs, text);
 488
 489         if (o->stream.text) {
 490             char buf[100];
 491             sprintf(buf, "stream%06d", o->number);
 492             fp = fopen(buf, "wb");
 493             fwrite(o->stream.text, 1, o->stream.pos, fp);
 494             fclose(fp);
 495             zcontext = deflate_compress_new(DEFLATE_TYPE_ZLIB);
 496             deflate_compress_data(zcontext, o->stream.text, o->stream.pos,
 497                                   DEFLATE_END_OF_DATA, &zbuf, &zlen);
 498             deflate_compress_free(zcontext);
 499             if (!o->main.text)
 500                 rdaddsc(&o->main, "<<\n");
 501             sprintf(text, "/Filter/FlateDecode\n/Length %d\n>>\n", zlen);
 502             rdaddsc(&o->main, text);
 503         }
 504
 505         assert(o->main.text);
 506         rdaddsc(&rs, o->main.text);
 507         sfree(o->main.text);
 508
 509         if (rs.text[rs.pos-1] != '\n')
 510             rdaddc(&rs, '\n');
 511
 512         if (o->stream.text) {
 513             rdaddsc(&rs, "stream\n");
 514             rdaddsn(&rs, zbuf, zlen);
 515             rdaddsc(&rs, "\nendstream\n");
 516             sfree(o->stream.text);
 517             sfree(zbuf);
 518         }
 519
 520         rdaddsc(&rs, "endobj\n");
 521
 522         o->final = rs.text;
 523         o->size = rs.pos;
 524     }
 525
 526     /*
 527      * Write out the PDF file.
 528      */
 529
 530     fp = fopen(filename, "wb");
 531     if (!fp) {
 532         error(err_cantopenw, filename);
 533         return;
 534     }
 535
 536     /*
 537      * Header. I'm going to put the version IDs in the header as
 538      * well, simply in PDF comments.  The PDF Reference also suggests
 539      * that binary PDF files contain four top-bit-set characters in
 540      * the second line.
 541      */
 542     fileoff = fprintf(fp, "%%PDF-1.3\n%% L\xc3\xba\xc3\xb0""a\n");
 543     for (p = sourceform; p; p = p->next)
 544         if (p->type == para_VersionID)
 545             fileoff += pdf_versionid(fp, p->words);
 546
 547     /*
 548      * Body
 549      */
 550     for (o = olist.head; o; o = o->next) {
 551         o->fileoff = fileoff;
 552         fwrite(o->final, 1, o->size, fp);
 553         fileoff += o->size;
 554     }
 555
 556     /*
 557      * Cross-reference table
 558      */
 559     fprintf(fp, "xref\n");
 560     assert(olist.head->number == 1);
 561     fprintf(fp, "0 %d\n", olist.tail->number + 1);
 562     fprintf(fp, "0000000000 65535 f \n");
 563     for (o = olist.head; o; o = o->next) {
 564         char entry[40];
 565         sprintf(entry, "%010d 00000 n \n", o->fileoff);
 566         assert(strlen(entry) == 20);
 567         fputs(entry, fp);
 568     }
 569
 570     /*
 571      * Trailer
 572      */
 573     fprintf(fp, "trailer\n<<\n/Size %d\n/Root %d 0 R\n/Info %d 0 R\n>>\n",
 574             olist.tail->number + 1, cat->number, info->number);
 575     fprintf(fp, "startxref\n%d\n%%%%EOF\n", fileoff);
 576
 577     fclose(fp);
 578
 579     sfree(filename);
 580 }
 581
 582 static object *new_object(objlist *list)
 583 {
 584     object *obj = snew(object);
 585
 586     obj->list = list;
 587
 588     obj->main.text = NULL;
 589     obj->main.pos = obj->main.size = 0;
 590     obj->stream.text = NULL;
 591     obj->stream.pos = obj->stream.size = 0;
 592
 593     obj->number = list->number++;
 594
 595     obj->next = NULL;
 596     if (list->tail)
 597         list->tail->next = obj;
 598     else
 599         list->head = obj;
 600     list->tail = obj;
 601
 602     obj->size = 0;
 603     obj->final = NULL;
 604
 605     return obj;
 606 }
 607
 608 static void objtext(object *o, char const *text)
 609 {
 610     rdaddsc(&o->main, text);
 611 }
 612
 613 static void objstream_len(object *o, char const *text, size_t len)
 614 {
 615     rdaddsn(&o->stream, text, len);
 616 }
 617
 618 static void objstream(object *o, char const *text)
 619 {
 620     rdaddsc(&o->stream, text);
 621 }
 622
 623 static void objref(object *o, object *dest)
 624 {
 625     char buf[40];
 626     sprintf(buf, "%d 0 R", dest->number);
 627     rdaddsc(&o->main, buf);
 628 }
 629
 630 static void objdest(object *o, page_data *p) {
 631     objtext(o, "[");
 632     objref(o, (object *)p->spare);
 633     objtext(o, "/XYZ null null null]");
 634 }
 635
 636 static char const * const stdfonts[] = {
 637     "Times-Roman", "Times-Bold", "Times-Italic", "Times-BoldItalic",
 638     "Helvetica", "Helvetica-Bold", "Helvetica-Oblique","Helvetica-BoldOblique",
 639     "Courier", "Courier-Bold", "Courier-Oblique", "Courier-BoldOblique",
 640     "Symbol", "ZapfDingbats"
 641 };
 642
 643 static int is_std_font(char const *name) {
 644     unsigned i;
 645     for (i = 0; i < lenof(stdfonts); i++)
 646         if (strcmp(name, stdfonts[i]) == 0)
 647             return TRUE;
 648     return FALSE;
 649 }
 650
 651 static void make_pages_node(object *node, object *parent, page_data *first,
 652                             page_data *last, object *resources,
 653                             object *mediabox)
 654 {
 655     int count;
 656     page_data *page;
 657     char buf[80];
 658
 659     objtext(node, "<<\n/Type /Pages\n");
 660     if (parent) {
 661         objtext(node, "/Parent ");
 662         objref(node, parent);
 663         objtext(node, "\n");
 664     }
 665
 666     /*
 667      * Count the pages in this stretch, to see if there are few
 668      * enough to reference directly.
 669      */
 670     count = 0;
 671     for (page = first; page; page = page->next) {
 672         count++;
 673         if (page == last)
 674             break;
 675     }
 676
 677     sprintf(buf, "/Count %d\n/Kids [\n", count);
 678     objtext(node, buf);
 679
 680     if (count > TREE_BRANCH) {
 681         int i;
 682         page_data *thisfirst, *thislast;
 683
 684         page = first;
 685
 686         for (i = 0; i < TREE_BRANCH; i++) {
 687             int number = (i+1) * count / TREE_BRANCH - i * count / TREE_BRANCH;
 688             thisfirst = page;
 689             while (number--) {
 690                 thislast = page;
 691                 page = page->next;
 692             }
 693
 694             if (thisfirst == thislast) {
 695                 objref(node, (object *)thisfirst->spare);
 696                 objtext((object *)thisfirst->spare, "/Parent ");
 697                 objref((object *)thisfirst->spare, node);
 698                 objtext((object *)thisfirst->spare, "\n");
 699             } else {
 700                 object *newnode = new_object(node->list);
 701                 make_pages_node(newnode, node, thisfirst, thislast,
 702                                 NULL, NULL);
 703                 objref(node, newnode);
 704             }
 705             objtext(node, "\n");
 706         }
 707
 708         assert(thislast == last || page == NULL);
 709
 710     } else {
 711         for (page = first; page; page = page->next) {
 712             objref(node, (object *)page->spare);
 713             objtext(node, "\n");
 714             objtext((object *)page->spare, "/Parent ");
 715             objref((object *)page->spare, node);
 716             objtext((object *)page->spare, "\n");
 717             if (page == last)
 718                 break;
 719         }
 720     }
 721
 722     objtext(node, "]\n");
 723
 724     if (resources) {
 725         objtext(node, "/Resources ");
 726         objref(node, resources);
 727         objtext(node, "\n");
 728     }
 729     if (mediabox) {
 730         objtext(node, "/MediaBox ");
 731         objref(node, mediabox);
 732         objtext(node, "\n");
 733     }
 734
 735     objtext(node, ">>\n");
 736 }
 737
 738 /*
 739  * In text on the page, PDF uses the PostScript font model, which
 740  * means that glyphs are identified by PS strings and hence font
 741  * encoding can be managed independently of the supplied encoding
 742  * of the font. However, in the document outline, the PDF spec
 743  * encodes in either PDFDocEncoding (a custom superset of
 744  * ISO-8859-1) or UTF-16BE.
 745  */
 746 char *pdf_outline_convert(wchar_t *s, int *len) {
 747     char *ret;
 748
 749     ret = utoa_careful_dup(s, CS_PDF);
 750
 751     /*
 752      * Very silly special case: if the returned string begins with
 753      * FE FF, then the PDF reader will mistake it for a UTF-16BE
 754      * string. So in this case we give up on PDFDocEncoding and
 755      * encode it in UTF-16 straight away.
 756      */
 757     if (ret && ret[0] == '\xFE' && ret[1] == '\xFF') {
 758         sfree(ret);
 759         ret = NULL;
 760     }
 761
 762     if (!ret) {
 763         ret = utoa_dup_len(s, CS_UTF16BE, len);
 764     } else {
 765         *len = strlen(ret);
 766     }
 767
 768     return ret;
 769 }
 770
 771 static int make_outline(object *parent, outline_element *items, int n,
 772                         int open)
 773 {
 774     int level, totalcount = 0;
 775     outline_element *itemp;
 776     object *curr, *prev = NULL, *first = NULL, *last = NULL;
 777
 778     assert(n > 0);
 779
 780     level = items->level;
 781
 782     while (n > 0) {
 783         char *title;
 784         int titlelen;
 785
 786         /*
 787          * Here we expect to be sitting on an item at the given
 788          * level. So we start by constructing an outline entry for
 789          * that item.
 790          */
 791         assert(items->level == level);
 792
 793         title = pdf_outline_convert(items->pdata->outline_title, &titlelen);
 794
 795         totalcount++;
 796         curr = new_object(parent->list);
 797         if (!first) first = curr;
 798         last = curr;
 799         objtext(curr, "<<\n/Title ");
 800         pdf_string_len(objtext, curr, title, titlelen);
 801         sfree(title);
 802         objtext(curr, "\n/Parent ");
 803         objref(curr, parent);
 804         objtext(curr, "\n/Dest");
 805         objdest(curr, items->pdata->first->page);
 806         objtext(curr, "\n");
 807         if (prev) {
 808             objtext(curr, "/Prev ");
 809             objref(curr, prev);
 810             objtext(curr, "\n");
 811
 812             objtext(prev, "/Next ");
 813             objref(prev, curr);
 814             objtext(prev, "\n>>\n");
 815         }
 816         prev = curr;
 817
 818         items++, n--;
 819         for (itemp = items; itemp < items+n && itemp->level > level;
 820              itemp++);
 821
 822         if (itemp > items) {
 823             char buf[80];
 824             int count = make_outline(curr, items, itemp - items, FALSE);
 825             if (!open)
 826                 count = -count;
 827             else
 828                 totalcount += count;
 829             sprintf(buf, "/Count %d\n", count);
 830             objtext(curr, buf);
 831         }
 832
 833         n -= itemp - items;
 834         items = itemp;
 835     }
 836     objtext(prev, ">>\n");
 837
 838     assert(first && last);
 839     objtext(parent, "/First ");
 840     objref(parent, first);
 841     objtext(parent, "\n/Last ");
 842     objref(parent, last);
 843     objtext(parent, "\n");
 844
 845     return totalcount;
 846 }
 847
 848 static int pdf_versionid(FILE *fp, word *words)
 849 {
 850     int ret;
 851
 852     ret = fprintf(fp, "%% ");
 853
 854     for (; words; words = words->next) {
 855         char *text;
 856         int type;
 857
 858         switch (words->type) {
 859           case word_HyperLink:
 860           case word_HyperEnd:
 861           case word_UpperXref:
 862           case word_LowerXref:
 863           case word_XrefEnd:
 864           case word_IndexRef:
 865             continue;
 866         }
 867
 868         type = removeattr(words->type);
 869
 870         switch (type) {
 871           case word_Normal:
 872             text = utoa_dup(words->text, CS_ASCII);
 873             break;
 874           case word_WhiteSpace:
 875             text = dupstr(" ");
 876             break;
 877           case word_Quote:
 878             text = dupstr("'");
 879             break;
 880         }
 881
 882         fputs(text, fp);
 883         ret += strlen(text);
 884         sfree(text);
 885     }
 886
 887     ret += fprintf(fp, "\n");
 888
 889     return ret;
 890 }
 891
 892 static void pdf_string_len(void (*add)(object *, char const *),
 893                            object *o, char const *str, int len)
 894 {
 895     char const *p;
 896
 897     add(o, "(");
 898     for (p = str; len > 0; p++, len--) {
 899         char c[10];
 900         if (*p < ' ' || *p > '~') {
 901             sprintf(c, "\\%03o", 0xFF & (int)*p);
 902         } else {
 903             int n = 0;
 904             if (*p == '\\' || *p == '(' || *p == ')')
 905                 c[n++] = '\\';
 906             c[n++] = *p;
 907             c[n] = '\0';
 908         }
 909         add(o, c);
 910     }
 911     add(o, ")");
 912 }
 913
 914 static void pdf_string(void (*add)(object *, char const *),
 915                        object *o, char const *str)
 916 {
 917     pdf_string_len(add, o, str, strlen(str));
 918 }