link.c

   1 /* lot's of code c&p-ed directly from mupdf */
   2
   3 #define _GNU_SOURCE
   4 #include <err.h>
   5 #include <regex.h>
   6 #include <errno.h>
   7 #include <ctype.h>
   8 #include <stdio.h>
   9 #include <stdarg.h>
  10 #include <stdlib.h>
  11 #include <string.h>
  12 #include <pthread.h>
  13 #include <sys/poll.h>
  14 #include <sys/time.h>
  15
  16 /* fugly as hell and GCC specific but... */
  17 #ifdef _BIG_ENDIAN
  18 #define GL_GLEXT_PROTOTYPES
  19 #endif
  20
  21 #include <GL/gl.h>
  22 #include <GL/glext.h>
  23
  24 #include <caml/fail.h>
  25 #include <caml/alloc.h>
  26 #include <caml/memory.h>
  27 #include <caml/unixsupport.h>
  28
  29 #include <fitz.h>
  30 #include <mupdf.h>
  31
  32 #if 0
  33 #define lprintf printf
  34 #else
  35 #define lprintf(...)
  36 #endif
  37
  38 #define ARSERT(cond) for (;;) {                         \
  39     if (!(cond)) {                                      \
  40         errx (1, "%s:%d " #cond, __FILE__, __LINE__);   \
  41     }                                                   \
  42     break;                                              \
  43 }
  44
  45 struct slice {
  46     int texindex;
  47     int w, h;
  48 };
  49
  50 struct page {
  51     int pageno;
  52     int slicecount;
  53     fz_textspan *text;
  54     fz_pixmap *pixmap;
  55     pdf_page *drawpage;
  56     struct pagedim *pagedim;
  57     struct page *prev;
  58     struct slice slices[];
  59 };
  60
  61 struct pagedim {
  62     int pageno;
  63     int rotate;
  64     fz_rect box;
  65     fz_bbox bbox;
  66     fz_matrix ctm;
  67 };
  68
  69 struct {
  70     int sock;
  71     int sliceheight;
  72     pthread_t thread;
  73     struct page *pages;
  74     struct pagedim *pagedims;
  75     int pagecount;
  76     int pagedimcount;
  77     pdf_xref *xref;
  78     fz_glyphcache *cache;
  79     int w, h;
  80
  81     int useatifs;
  82
  83     int texindex;
  84     int texcount;
  85     GLuint *texids;
  86
  87     GLenum texform;
  88     GLenum texty;
  89
  90     int lotsamemory;
  91
  92     int *pagetbl;
  93     struct {
  94         int w, h;
  95         struct slice *slice;
  96     } *texowners;
  97 } state;
  98
  99 static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
 100
 101 static void lock (const char *cap)
 102 {
 103     int ret = pthread_mutex_lock (&mutex);
 104     if (ret) {
 105         errx (1, "%s: pthread_mutex_lock: %s", cap, strerror (ret));
 106     }
 107 }
 108
 109 static void unlock (const char *cap)
 110 {
 111     int ret = pthread_mutex_unlock (&mutex);
 112     if (ret) {
 113         errx (1, "%s: pthread_mutex_unlock: %s", cap, strerror (ret));
 114     }
 115 }
 116
 117 static int trylock (const char *cap)
 118 {
 119     int ret = pthread_mutex_trylock (&mutex);
 120
 121     if (ret && ret != EBUSY) {
 122         errx (1, "%s: pthread_mutex_trylock: %s", cap, strerror (ret));
 123     }
 124     return ret == EBUSY;
 125 }
 126
 127 static void *parse_pointer (const char *cap, const char *s)
 128 {
 129     int ret;
 130     void *ptr;
 131
 132     ret = sscanf (s, "%p", &ptr);
 133     if (ret != 1) {
 134         errx (1, "%s: cannot parse pointer in `%s'", cap, s);
 135     }
 136     return ptr;
 137 }
 138
 139 static int hasdata (int sock)
 140 {
 141     int ret;
 142     struct pollfd pfd;
 143
 144     pfd.fd = sock;
 145     pfd.events = POLLIN;
 146     ret = poll (&pfd, 1, 0);
 147     if (ret == 0) {
 148         return 0;
 149     }
 150     if (ret != 1) {
 151         err (1, "poll");
 152     }
 153     return pfd.revents & POLLIN;
 154 }
 155
 156 static double now (void)
 157 {
 158     struct timeval tv;
 159
 160     if (gettimeofday (&tv, NULL)) {
 161         err (1, "gettimeofday");
 162     }
 163     return tv.tv_sec + tv.tv_usec*1e-6;
 164 }
 165
 166 static void readdata (int fd, char *p, int size)
 167 {
 168     ssize_t n;
 169
 170     n = read (fd, p, size);
 171     if (n - size) {
 172         err (1, "read (req %d, ret %zd)", size, n);
 173     }
 174 }
 175
 176 static void writedata (int fd, char *p, int size)
 177 {
 178     char buf[4];
 179     ssize_t n;
 180
 181     buf[0] = (size >> 24) & 0xff;
 182     buf[1] = (size >> 16) & 0xff;
 183     buf[2] = (size >>  8) & 0xff;
 184     buf[3] = (size >>  0) & 0xff;
 185
 186     n = write (fd, buf, 4);
 187     if (n != 4) {
 188         err (1, "write %zd", n);
 189     }
 190
 191     n = write (fd, p, size);
 192     if (n - size) {
 193         err (1, "write (req %d, ret %zd)", size, n);
 194     }
 195 }
 196
 197 static void __attribute__ ((format (printf, 2, 3)))
 198     printd (int fd, const char *fmt, ...)
 199 {
 200     int len;
 201     va_list ap;
 202     char buf[200];
 203
 204     va_start (ap, fmt);
 205     len = vsnprintf (buf, sizeof (buf), fmt, ap);
 206     va_end (ap);
 207     writedata (fd, buf, len);
 208 }
 209
 210 static void die (fz_error error)
 211 {
 212     fz_catch (error, "aborting");
 213     if (state.xref)
 214        pdf_closexref (state.xref);
 215     exit (1);
 216 }
 217
 218 static void openxref (char *filename)
 219 {
 220     int fd;
 221     fz_stream *file;
 222
 223     fd = open (filename, O_BINARY | O_RDONLY, 0666);
 224     if (fd < 0)
 225         die (fz_throw ("cannot open file '%s'", filename));
 226
 227     file = fz_openfile (fd);
 228     state.xref = pdf_openxref (file);
 229     if (!state.xref)
 230         die (fz_throw ("cannot open PDF file '%s'", filename));
 231     fz_dropstream (file);
 232
 233     if (pdf_needspassword (state.xref)) {
 234         die (fz_throw ("password protected"));
 235     }
 236
 237     state.pagecount = pdf_getpagecount (state.xref);
 238     state.pagetbl = stat_alloc (state.pagecount * sizeof (*state.pagetbl));
 239 }
 240
 241 static int readlen (int fd)
 242 {
 243     ssize_t n;
 244     char p[4];
 245
 246     n = read (fd, p, 4);
 247     if (n != 4) {
 248         err (1, "read %zd", n);
 249     }
 250
 251     return (p[0] << 24) | (p[1] << 16) | (p[2] << 8) | p[3];
 252 }
 253
 254 static void freepage (struct page *page)
 255 {
 256     int i;
 257     struct page *p;
 258
 259     fz_droppixmap (page->pixmap);
 260     for (p = state.pages; p; p = p->prev) {
 261         if (p->prev == page) {
 262             p->prev = page->prev;
 263             break;
 264         }
 265     }
 266     for (i = 0; i < page->slicecount; ++i) {
 267         struct slice *s = &page->slices[i];
 268         if (s->texindex != -1) {
 269             if (state.texowners[s->texindex].slice == s) {
 270                 state.texowners[s->texindex].slice = NULL;
 271                 ARSERT (state.texowners[s->texindex].w == s->w);
 272                 ARSERT (state.texowners[s->texindex].h >= s->h);
 273             }
 274         }
 275     }
 276     if (page->text) {
 277         fz_freetextspan (page->text);
 278     }
 279     if (page->drawpage) {
 280         pdf_droppage (page->drawpage);
 281     }
 282
 283     free (page);
 284 }
 285
 286 static void subdivide (struct page *p)
 287 {
 288     int i;
 289     int h = p->pixmap->h;
 290     int th = MIN (h, state.sliceheight);
 291
 292     for (i = 0; i < p->slicecount; ++i) {
 293         struct slice *s = &p->slices[i];
 294         s->texindex = -1;
 295         s->h = MIN (th, h);
 296         s->w = p->pixmap->w;
 297         h -= th;
 298     }
 299 }
 300
 301 static void *render (int pageno, int pindex)
 302 {
 303     fz_error error;
 304     int slicecount;
 305     fz_obj *pageobj;
 306     struct page *page;
 307     double start, end;
 308     pdf_page *drawpage;
 309     fz_displaylist *list;
 310     fz_device *idev, *mdev;
 311     struct pagedim *pagedim;
 312
 313     start = now ();
 314     /* printd (state.sock, "T rendering %d", pageno); */
 315     pdf_flushxref (state.xref, 0);
 316
 317     pagedim = &state.pagedims[pindex];
 318     slicecount = (pagedim->bbox.y1 - pagedim->bbox.y0
 319                   + state.sliceheight - 1) / state.sliceheight;
 320     slicecount += slicecount == 0;
 321
 322     page = calloc (sizeof (*page)
 323                    + (slicecount * sizeof (struct slice)), 1);
 324     if (!page) {
 325         err (1, "calloc page %d\n", pageno);
 326     }
 327     page->slicecount = slicecount;
 328     page->prev = state.pages;
 329     state.pages = page;
 330
 331     pageobj = pdf_getpageobject (state.xref, pageno);
 332     if (!pageobj)
 333         die (fz_throw ("cannot retrieve info from page %d", pageno));
 334
 335     error = pdf_loadpage (&drawpage, state.xref, pageobj);
 336     if (error)
 337         die (error);
 338
 339     page->pixmap = fz_newpixmapwithrect (pdf_devicergb, pagedim->bbox);
 340     if (error)
 341         die (error);
 342     fz_clearpixmap (page->pixmap, 0xFF);
 343
 344     list = fz_newdisplaylist ();
 345     if (!list)
 346         die (fz_throw ("fz_newdisplaylist failed"));
 347
 348     mdev = fz_newlistdevice (list);
 349     error = pdf_runcontentstream (mdev, fz_identity (), state.xref,
 350                                   drawpage->resources,
 351                                   drawpage->contents);
 352     if (error)
 353         die (error);
 354     fz_freedevice (mdev);
 355
 356     idev = fz_newdrawdevice (state.cache, page->pixmap);
 357     if (!idev)
 358         die (fz_throw ("fz_newdrawdevice failed"));
 359     fz_executedisplaylist (list, idev, pagedim->ctm);
 360     fz_freedevice (idev);
 361
 362     fz_freedisplaylist (list);
 363
 364     page->drawpage = drawpage;
 365     page->pagedim = pagedim;
 366     page->pageno = pageno;
 367     subdivide (page);
 368     end = now ();
 369
 370     if (!state.lotsamemory) {
 371         pdf_agestoreditems (state.xref->store);
 372         pdf_evictageditems (state.xref->store);
 373     }
 374
 375     /* printd (state.sock, "T rendering %d took %f sec", pageno, end - start); */
 376     return page;
 377 }
 378
 379 /* almost verbatim copy of pdf_getpagecountimp */
 380 struct stuff
 381 {
 382     fz_obj *resources;
 383     fz_obj *mediabox;
 384     fz_obj *cropbox;
 385     fz_obj *rotate;
 386 };
 387
 388 static void
 389 recurse_page (fz_obj *node, int bias, int *pagesp, struct stuff inherit)
 390 {
 391     fz_obj *type;
 392     fz_obj *kids;
 393     fz_obj *count;
 394     char *typestr;
 395     int pages = 0;
 396     int i;
 397
 398     if (!fz_isdict(node))
 399     {
 400         fz_warn("pagetree node is missing, igoring missing pages...");
 401         return;
 402     }
 403
 404     type = fz_dictgets(node, "Type");
 405     kids = fz_dictgets(node, "Kids");
 406     count = fz_dictgets(node, "Count");
 407
 408     if (fz_isname(type))
 409         typestr = fz_toname(type);
 410     else
 411     {
 412         fz_warn("pagetree node (%d %d R) lacks required type", fz_tonum(node), fz_togen(node));
 413
 414         kids = fz_dictgets(node, "Kids");
 415         if (kids)
 416         {
 417             fz_warn("guessing it may be a pagetree node, continuing...");
 418             typestr = "Pages";
 419         }
 420         else
 421         {
 422             fz_warn("guessing it may be a page, continuing...");
 423             typestr = "Page";
 424         }
 425     }
 426
 427     if (!strcmp(typestr, "Page")) {
 428         int rotate;
 429         fz_obj *obj;
 430         fz_rect box;
 431         struct pagedim *p;
 432         int pageno = *pagesp;
 433
 434         state.pagetbl[pageno + bias] = fz_tonum (node);
 435
 436         obj = fz_dictgets (node, "CropBox");
 437         if (!obj) obj = inherit.cropbox;
 438         if (!fz_isarray (obj)) {
 439             obj = fz_dictgets (node, "MediaBox");
 440             if (!obj) obj = inherit.mediabox;
 441
 442             if (!fz_isarray (obj)) {
 443                 die (fz_throw ("cannot find page bounds %d (%d Rd)",
 444                                fz_tonum (node), fz_togen (node)));
 445             }
 446         }
 447         box = pdf_torect (obj);
 448
 449         obj = fz_dictgets (node, "Rotate");
 450         if (!obj) obj = inherit.rotate;
 451         if (fz_isint (obj)) {
 452             rotate = fz_toint (obj);
 453         }
 454         else  {
 455             rotate = 0;
 456         }
 457
 458         p = &state.pagedims[state.pagedimcount - 1];
 459         if ((state.pagedimcount == 0)
 460             || (p->rotate != rotate || memcmp (&p->box, &box, sizeof (box)))) {
 461             size_t size;
 462
 463             size = (state.pagedimcount + 1) * sizeof (*state.pagedims);
 464             state.pagedims = realloc (state.pagedims, size);
 465             if (!state.pagedims) {
 466                 err (1, "realloc pagedims to %zu (%d elems)",
 467                      size, state.pagedimcount + 1);
 468             }
 469             p = &state.pagedims[state.pagedimcount++];
 470             p->rotate = rotate;
 471             p->box = box;
 472             p->pageno = pageno + bias;
 473         }
 474         (*pagesp)++;
 475     }
 476     else if (!strcmp(typestr, "Pages"))
 477     {
 478         fz_obj *inh;
 479
 480         if (!fz_isarray(kids))
 481             fz_warn("page tree node contains no pages");
 482
 483         pdf_logpage("subtree (%d %d R) {\n", fz_tonum(node), fz_togen(node));
 484
 485         inh = fz_dictgets(node, "Resources");
 486         if (inh) inherit.resources = inh;
 487
 488         inh = fz_dictgets(node, "MediaBox");
 489         if (inh) inherit.mediabox = inh;
 490
 491         inh = fz_dictgets(node, "CropBox");
 492         if (inh) inherit.cropbox = inh;
 493
 494         inh = fz_dictgets(node, "Rotate");
 495         if (inh) inherit.rotate = inh;
 496
 497         for (i = 0; i < fz_arraylen(kids); i++)
 498         {
 499             fz_obj *obj = fz_arrayget(kids, i);
 500
 501             /* prevent infinite recursion possible in maliciously crafted PDFs */
 502             if (obj == node)
 503             {
 504                 fz_warn("cyclic page tree");
 505                 return;
 506             }
 507
 508             recurse_page (obj, *pagesp + bias, &pages, inherit);
 509         }
 510
 511         if (pages != fz_toint(count))
 512         {
 513             fz_warn("page tree node contains incorrect number of pages, continuing...");
 514             count = fz_newint(pages);
 515             fz_dictputs(node, "Count", count);
 516             fz_dropobj(count);
 517         }
 518
 519         pdf_logpage("%d pages\n", pages);
 520
 521         (*pagesp) += pages;
 522
 523         pdf_logpage("}\n");
 524     }
 525 }
 526
 527 static void initpdims (void)
 528 {
 529     fz_obj *catalog;
 530     fz_obj *pages;
 531     int count;
 532     double start, end;
 533     struct stuff inherit;
 534
 535     start = now ();
 536     catalog = fz_dictgets (state.xref->trailer, "Root");
 537     pages = fz_dictgets (catalog, "Pages");
 538
 539     inherit.resources = nil;
 540     inherit.mediabox = nil;
 541     inherit.cropbox = nil;
 542     inherit.rotate = nil;
 543
 544     count = 0;
 545     recurse_page (pages, 0, &count, inherit);
 546     end = now ();
 547     printd (state.sock, "T Processed %d pages in %f seconds",
 548             count, end - start);
 549 }
 550
 551 static void layout (void)
 552 {
 553     int pindex;
 554     fz_matrix ctm;
 555     fz_rect box, box2;
 556     double zoom, w;
 557     struct pagedim *p = state.pagedims;
 558
 559     pindex = 0;
 560     printd (state.sock, "c");
 561     for (pindex = 0; pindex < state.pagedimcount; ++pindex, ++p) {
 562         box.x0 = MIN (p->box.x0, p->box.x1);
 563         box.y0 = MIN (p->box.y0, p->box.y1);
 564         box.x1 = MAX (p->box.x0, p->box.x1);
 565         box.y1 = MAX (p->box.y0, p->box.y1);
 566
 567         ctm = fz_identity ();
 568         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 569         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 570         box2 = fz_transformrect (ctm, box);
 571         w = box2.x1 - box2.x0;
 572
 573         zoom = (state.w / w);
 574         ctm = fz_identity ();
 575         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 576         ctm = fz_concat (ctm, fz_scale (zoom, -zoom));
 577         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 578         p->bbox = fz_roundrect (fz_transformrect (ctm, box));
 579         memcpy (&p->ctm, &ctm, sizeof (ctm));
 580     }
 581
 582     while (p-- != state.pagedims)  {
 583         printd (state.sock, "l %d %d %d",
 584                 p->pageno, p->bbox.x1 - p->bbox.x0, p->bbox.y1 - p->bbox.y0);
 585     }
 586
 587     printd (state.sock, "C %d", state.pagecount);
 588 }
 589
 590 static void recurse_outline (pdf_outline *outline, int level)
 591 {
 592     while (outline) {
 593         int i, num;
 594         fz_obj *obj;
 595         int top = 0;
 596         int pageno = -1;
 597
 598         if (!outline->link) goto next;
 599
 600         obj = outline->link->dest;
 601         if (fz_isindirect (obj)) {
 602             num = fz_tonum (obj);
 603
 604             for (i = 0; i < state.pagecount; ++i)  {
 605                 if (state.pagetbl[i] == num) {
 606                     pageno = i;
 607                     break;
 608                 }
 609             }
 610         }
 611         else if (fz_isarray (obj)) {
 612             fz_obj *obj2;
 613
 614             obj2 = fz_arrayget (obj, 0);
 615             if (fz_isint (obj2)) {
 616                 pageno = fz_toint (obj2);
 617             }
 618             else {
 619                 num = fz_tonum (obj2);
 620                 for (i = 0; i < state.pagecount; ++i)  {
 621                     if (state.pagetbl[i] == num) {
 622                         pageno = i;
 623                         break;
 624                     }
 625                 }
 626             }
 627
 628             if (fz_arraylen (obj) > 3) {
 629                 fz_point p;
 630                 struct pagedim *pagedim = state.pagedims;
 631
 632                 for (i = 0; i < state.pagedimcount; ++i) {
 633                     if (state.pagedims[i].pageno > pageno)
 634                         break;
 635                     pagedim = &state.pagedims[i];
 636                 }
 637
 638                 p.x = fz_toint (fz_arrayget (obj, 2));
 639                 p.y = fz_toint (fz_arrayget (obj, 3));
 640                 p = fz_transformpoint (pagedim->ctm, p);
 641                 top = p.y;
 642             }
 643         }
 644
 645         lprintf ("%*c%s %d\n", level, ' ', outline->title, pageno);
 646         printd (state.sock, "o %d %d %d %s",
 647                 level, pageno, top, outline->title);
 648     next:
 649         if (outline->child) {
 650             recurse_outline (outline->child, level + 1);
 651         }
 652         outline = outline->next;
 653     }
 654 }
 655
 656 static void process_outline (void)
 657 {
 658     pdf_outline *outline;
 659
 660     outline = pdf_loadoutline (state.xref);
 661     if (outline) {
 662         recurse_outline (outline, 0);
 663         pdf_dropoutline (outline);
 664     }
 665 }
 666
 667 static int comparespans (const void *l, const void *r)
 668 {
 669     fz_textspan *const*ls = l;
 670     fz_textspan *const*rs = r;
 671
 672     return (*ls)->text->bbox.y0 - (*rs)->text->bbox.y0;
 673 }
 674
 675 /* wishful thinking function */
 676 static void search (regex_t *re, int pageno, int y, int forward)
 677 {
 678     int i, j;
 679     int ret;
 680     char *p;
 681     char buf[256];
 682     fz_error error;
 683     fz_obj *pageobj;
 684     fz_device *tdev;
 685     pdf_page *drawpage;
 686     fz_textspan *text, *span, **pspan;
 687     struct pagedim *pdim, *pdimprev;
 688     int stop = 0;
 689     int niters = 0;
 690     int nspans;
 691     double start, end;
 692
 693     start = now ();
 694     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 695         if (niters++ == 5) {
 696             if (!state.lotsamemory) {
 697                 pdf_agestoreditems (state.xref->store);
 698                 pdf_evictageditems (state.xref->store);
 699             }
 700             niters = 0;
 701             if (hasdata (state.sock)) {
 702                 printd (state.sock, "T attention requested aborting search at %d",
 703                         pageno);
 704                 stop = 1;
 705             }
 706             else {
 707                 printd (state.sock, "T searching in page %d", pageno);
 708             }
 709         }
 710         pdimprev = NULL;
 711         for (i = 0; i < state.pagedimcount; ++i) {
 712             pdim = &state.pagedims[i];
 713             if (pdim->pageno == pageno)  {
 714                 goto found;
 715             }
 716             if (pdim->pageno > pageno) {
 717                 pdim = pdimprev;
 718                 goto found;
 719             }
 720             pdimprev = pdim;
 721         }
 722         pdim = pdimprev;
 723     found:
 724
 725         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 726         if (!pageobj)
 727             die (fz_throw ("cannot retrieve info from page %d", pageno));
 728
 729         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 730         if (error)
 731             die (error);
 732
 733         text = fz_newtextspan ();
 734         tdev = fz_newtextdevice (text);
 735         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 736                                       drawpage->resources,
 737                                       drawpage->contents);
 738         if (error) die (error);
 739         fz_freedevice (tdev);
 740
 741         nspans = 0;
 742         for (span = text; span; span = span->next) {
 743             nspans++;
 744         }
 745         pspan = malloc (sizeof (void *) * nspans);
 746         if (!pspan) {
 747             err (1, "malloc span pointers %zu", sizeof (void *) * nspans);
 748         }
 749         for (i = 0, span = text; span; span = span->next, ++i) {
 750             pspan[i] = span;
 751         }
 752         qsort (pspan, nspans, sizeof (fz_textspan *), comparespans);
 753
 754         j = forward ? 0 : nspans - 1;
 755         while (nspans--) {
 756             regmatch_t rm;
 757
 758             span = pspan[j];
 759             j += forward ? 1 : -1;
 760             p = buf;
 761             /* XXX: spans are not sorted "visually" */
 762             for (i = 0; i < MIN (span->len, sizeof (buf) - 1); ++i) {
 763                 if (forward) {
 764                     if (span->text[i].bbox.y0 < y + 1) {
 765                         continue;
 766                     }
 767                 }
 768                 else {
 769                     if (span->text[i].bbox.y0 > y - 1) {
 770                         continue;
 771                     }
 772                 }
 773                 if (span->text[i].c < 256) {
 774                     *p++ = span->text[i].c;
 775                 }
 776                 else  {
 777                     *p++ = '?';
 778                 }
 779             }
 780             if (p == buf) {
 781                 continue;
 782             }
 783             *p++ = 0;
 784
 785             ret = regexec (re, buf, 1, &rm, 0);
 786             if (ret)  {
 787                 if (ret != REG_NOMATCH) {
 788                     size_t size;
 789                     char errbuf[80];
 790                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 791                     printd (state.sock,
 792                             "T regexec error `%.*s'",
 793                             (int) size, errbuf);
 794                     fz_freetextspan (text);
 795                     pdf_droppage (drawpage);
 796                     free (pspan);
 797                     return;
 798                 }
 799             }
 800             else  {
 801                 fz_rect r;
 802
 803                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 804                 r.y0 = span->text[rm.rm_so].bbox.y0;
 805                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 806                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 807
 808                 if (!stop) {
 809                     printd (state.sock, "F %d %d %f %f %f %f",
 810                             pageno, 1,
 811                             r.x0, r.y0,
 812                             r.x1, r.y1);
 813                 }
 814                 else  {
 815                     printd (state.sock, "R %d %d %f %f %f %f",
 816                             pageno, 2,
 817                             r.x0, r.y0,
 818                             r.x1, r.y1);
 819                 }
 820                 printd (state.sock, "T found at %d `%.*s' %f in %f sec",
 821                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 822                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 823                         now () - start);
 824                 stop = 1;
 825             }
 826         }
 827         if (forward) {
 828             pageno += 1;
 829             y = 0;
 830         }
 831         else {
 832             pageno -= 1;
 833             y = INT_MAX;
 834         }
 835         fz_freetextspan (text);
 836         pdf_droppage (drawpage);
 837         free (pspan);
 838     }
 839     end = now ();
 840     if (!stop)  {
 841         printd (state.sock, "T no matches %f sec", end - start);
 842     }
 843     printd (state.sock, "D");
 844 }
 845
 846 static void *mainloop (void *unused)
 847 {
 848     char *p = NULL;
 849     int len, ret, oldlen = 0;
 850
 851     for (;;) {
 852         len = readlen (state.sock);
 853         if (len == 0) {
 854             errx (1, "readlen returned 0");
 855         }
 856
 857         if (oldlen < len + 1) {
 858             p = realloc (p, len + 1);
 859             if (!p) {
 860                 err (1, "realloc %d failed", len + 1);
 861             }
 862             oldlen = len + 1;
 863         }
 864         readdata (state.sock, p, len);
 865         p[len] = 0;
 866
 867         if (!strncmp ("open", p, 4)) {
 868             char *filename = p + 5;
 869
 870             openxref (filename);
 871             initpdims ();
 872         }
 873         else if (!strncmp ("free", p, 4)) {
 874             void *ptr;
 875
 876             ret = sscanf (p + 4, " %p", &ptr);
 877             if (ret != 1) {
 878                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 879             }
 880             lock ("free");
 881             freepage (ptr);
 882             unlock ("free");
 883             printd (state.sock, "d");
 884         }
 885         else if (!strncmp ("search", p, 6)) {
 886             int icase, pageno, y, ret, len2, forward;
 887             char *pattern;
 888             regex_t re;
 889
 890             ret = sscanf (p + 6, " %d %d %d %d,%n",
 891                           &icase, &pageno, &y, &forward, &len2);
 892             if (ret != 4) {
 893                 errx (1, "malformed search `%s' ret=%d", p, ret);
 894             }
 895
 896             pattern = p + 6 + len2;
 897             ret = regcomp (&re, pattern,
 898                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 899             if (ret) {
 900                 char errbuf[80];
 901                 size_t size;
 902
 903                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 904                 printd (state.sock, "T regcomp failed `%.*s'",
 905                         (int) size, errbuf);
 906             }
 907             else  {
 908                 search (&re, pageno, y, forward);
 909                 regfree (&re);
 910             }
 911         }
 912         else if (!strncmp ("geometry", p, 8)) {
 913             int w, h;
 914
 915             ret = sscanf (p + 8, " %d %d", &w, &h);
 916             if (ret != 2) {
 917                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 918             }
 919             state.h = h;
 920             if (w != state.w) {
 921                 int i;
 922                 state.w = w;
 923                 for (i = 0; i < state.texcount; ++i)  {
 924                     state.texowners[i].slice = NULL;
 925                 }
 926             }
 927             lock ("geometry");
 928             layout ();
 929             process_outline ();
 930             unlock ("geometry");
 931             printd (state.sock, "d");
 932         }
 933         else if (!strncmp ("render", p, 6)) {
 934             int pageno, pindex, w, h, ret;
 935             struct page *page;
 936
 937             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 938             if (ret != 4) {
 939                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 940             }
 941
 942             page = render (pageno, pindex);
 943             printd (state.sock, "r %d %d %d %p\n",
 944                     pageno,
 945                     state.w,
 946                     state.h,
 947                     page);
 948         }
 949         else {
 950             errx (1, "unknown command %.*s", len, p);
 951         }
 952     }
 953     return NULL;
 954 }
 955
 956 static void upload2 (struct page *page, int slicenum, const char *cap)
 957 {
 958     int i;
 959     int w, h;
 960     double start, end;
 961     struct slice *slice = &page->slices[slicenum];
 962
 963     w = page->pixmap->w;
 964     h = page->pixmap->h;
 965
 966     ARSERT (w == slice->w);
 967     if (slice->texindex != -1
 968         && state.texowners[slice->texindex].slice == slice) {
 969         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 970     }
 971     else  {
 972         int subimage = 0;
 973         int index = (state.texindex++ % state.texcount);
 974         size_t offset = 0;
 975
 976         for (i = 0; i < slicenum; ++i) {
 977             offset += w * page->slices[i].h * 4;
 978         }
 979
 980         if (state.texowners[index].w == slice->w) {
 981             if (state.texowners[index].h >= slice->h ) {
 982                 subimage = 1;
 983             }
 984             else {
 985                 state.texowners[index].h = slice->h;
 986             }
 987         }
 988         else  {
 989             state.texowners[index].h = slice->h;
 990         }
 991
 992         state.texowners[index].slice = slice;
 993         state.texowners[index].w = slice->w;
 994         slice->texindex = index;
 995
 996         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 997         start = now ();
 998         if (subimage) {
 999             {
1000                 GLenum err = glGetError ();
1001                 if (err != GL_NO_ERROR) {
1002                     printf ("\e[0;31mERROR1 %d %d %#x\e[0m\n", w, slice->h, err);
1003                     abort ();
1004                 }
1005             }
1006             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
1007                              0,
1008                              0,
1009                              0,
1010                              w,
1011                              slice->h,
1012                              state.texform,
1013                              state.texty,
1014                              page->pixmap->samples + offset
1015                 );
1016             {
1017                 GLenum err = glGetError ();
1018                 if (err != GL_NO_ERROR) {
1019                     printf ("\e[0;31mERROR %d %d %#x\e[0m\n", w, slice->h, err);
1020                     abort ();
1021                 }
1022             }
1023         }
1024         else {
1025             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
1026                           0,
1027                           GL_RGBA8,
1028                           w,
1029                           slice->h,
1030                           0,
1031                           state.texform,
1032                           state.texty,
1033                           page->pixmap->samples + offset
1034                 );
1035         }
1036
1037         end = now ();
1038         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
1039                  subimage ? "sub" : "img",
1040                  page->pageno, slicenum,
1041                  slice->w, slice->h,
1042                  state.texids[slice->texindex],
1043                  end - start);
1044     }
1045 }
1046
1047 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
1048                         value py_v, value ptr_v)
1049 {
1050     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
1051     int dispy = Int_val (dispy_v);
1052     int w = Int_val (w_v);
1053     int h = Int_val (h_v);
1054     int py = Int_val (py_v);
1055     char *s = String_val (ptr_v);
1056     int ret;
1057     void *ptr;
1058     struct page *page;
1059     int slicenum = 0;
1060
1061     if (trylock ("ml_draw")) {
1062         goto done;
1063     }
1064
1065     ret = sscanf (s, "%p", &ptr);
1066     if (ret != 1) {
1067         errx (1, "cannot parse pointer `%s'", s);
1068     }
1069     page = ptr;
1070
1071     w = page->pixmap->w;
1072
1073     ARSERT (h >= 0 && "ml_draw wrong h");
1074
1075     glEnable (GL_TEXTURE_RECTANGLE_ARB);
1076     if (state.useatifs) {
1077         glEnable (GL_FRAGMENT_SHADER_ATI);
1078     }
1079
1080     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
1081         struct slice *slice = &page->slices[slicenum];
1082         if (slice->h > py) {
1083             break;
1084         }
1085         py -= slice->h;
1086     }
1087
1088     h = MIN (state.h, h);
1089     while (h) {
1090         int th;
1091         struct slice *slice = &page->slices[slicenum];
1092
1093         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
1094
1095         th = MIN (h, slice->h - py);
1096         upload2 (page, slicenum, "upload");
1097
1098         glBegin (GL_QUADS);
1099         {
1100             glTexCoord2i (0, py);
1101             glVertex2i (0, dispy);
1102
1103             glTexCoord2i (w, py);
1104             glVertex2i (w, dispy);
1105
1106             glTexCoord2i (w, py+th);
1107             glVertex2i (w, dispy + th);
1108
1109             glTexCoord2i (0, py+th);
1110             glVertex2i (0, dispy + th);
1111         }
1112         glEnd ();
1113
1114         h -= th;
1115         py = 0;
1116         dispy += th;
1117         slicenum += 1;
1118     }
1119
1120     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1121     if (state.useatifs) {
1122         glDisable (GL_FRAGMENT_SHADER_ATI);
1123     }
1124
1125     unlock ("ml_draw");
1126  done:
1127     CAMLreturn (Val_unit);
1128 }
1129
1130 static pdf_link *getlink (struct page *page, int x, int y)
1131 {
1132     fz_point p;
1133     fz_matrix ctm;
1134     pdf_link *link;
1135
1136     p.x = x;
1137     p.y = y;
1138
1139     ctm = fz_invertmatrix (page->pagedim->ctm);
1140     p = fz_transformpoint (ctm, p);
1141
1142     for (link = page->drawpage->links; link; link = link->next) {
1143         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1144             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1145                 if (link->kind == PDF_LGOTO) {
1146                     return link;
1147                 }
1148             }
1149         }
1150     }
1151     return NULL;
1152 }
1153
1154 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1155 {
1156     CAMLparam3 (ptr_v, x_v, y_v);
1157     char *s = String_val (ptr_v);
1158     int ret;
1159
1160     if (trylock ("ml_checklink")) {
1161         ret = 0;
1162     }
1163     else {
1164         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1165                                Int_val (x_v), Int_val (y_v));
1166         unlock ("ml_checklink");
1167     }
1168     CAMLreturn (Val_bool (ret));
1169 }
1170
1171 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1172 {
1173     CAMLparam3 (ptr_v, x_v, y_v);
1174     CAMLlocal2 (ret_v, tup_v);
1175     pdf_link *link;
1176     struct page *page;
1177     char *s = String_val (ptr_v);
1178
1179     if (trylock ("ml_gettext")) {
1180         ret_v = Val_int (0);
1181         goto done;
1182     }
1183
1184     page = parse_pointer ("ml_getlink", s);
1185
1186     link = getlink (page, Int_val (x_v), Int_val (y_v));
1187     if (link) {
1188         int pageno;
1189         fz_point p;
1190         fz_obj *obj;
1191
1192         pageno = -1;
1193         obj = fz_arrayget (link->dest, 0);
1194         if (fz_isindirect (obj)) {
1195             pageno = pdf_findpageobject (state.xref, obj) - 1;
1196         }
1197         else if (fz_isint (obj)) {
1198             pageno = fz_toint (obj);
1199         }
1200
1201         if (fz_arraylen (link->dest) > 3) {
1202             p.x = fz_toint (fz_arrayget (link->dest, 2));
1203             p.y = fz_toint (fz_arrayget (link->dest, 3));
1204             p = fz_transformpoint (page->pagedim->ctm, p);
1205         }
1206         else  {
1207             p.x = 0.0;
1208             p.y = 0.0;
1209         }
1210
1211         tup_v = caml_alloc_tuple (2);
1212         ret_v = caml_alloc_small (1, 1);
1213         Field (tup_v, 0) = Val_int (pageno);
1214         Field (tup_v, 1) = Val_int (p.y);
1215         Field (ret_v, 0) = tup_v;
1216     }
1217     else {
1218         ret_v = Val_int (0);
1219     }
1220     unlock ("ml_getlink");
1221
1222  done:
1223     CAMLreturn (ret_v);
1224 }
1225
1226 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1227 {
1228     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1229     fz_matrix ctm;
1230     fz_point p1, p2;
1231     struct page *page;
1232     fz_textspan *span;
1233     char *s = String_val (ptr_v);
1234     int rectsel = Bool_val (rectsel_v);
1235     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1236
1237     /* stop GCC from complaining about uninitialized variables */
1238     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1239
1240     if (trylock ("ml_gettext")) {
1241         goto done;
1242     }
1243
1244     page = parse_pointer ("ml_gettext", s);
1245
1246     oy = Int_val (oy_v);
1247     p1.x = Int_val (Field (rect_v, 0));
1248     p1.y = Int_val (Field (rect_v, 1));
1249     p2.x = Int_val (Field (rect_v, 2));
1250     p2.y = Int_val (Field (rect_v, 3));
1251
1252     if (0) {
1253         glEnable (GL_BLEND);
1254         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1255         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1256         glColor4f (0, 0, 0, 0.2);
1257         glRecti (p1.x, p1.y, p2.x, p2.y);
1258         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1259         glDisable (GL_BLEND);
1260     }
1261
1262     ctm = page->pagedim->ctm;
1263     if (!page->text) {
1264         fz_error error;
1265         fz_device *tdev;
1266
1267         page->text = fz_newtextspan ();
1268         tdev = fz_newtextdevice (page->text);
1269         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1270                                       page->drawpage->resources,
1271                                       page->drawpage->contents);
1272         if (error) die (error);
1273         fz_freedevice (tdev);
1274     }
1275
1276     printf ("\ec");
1277
1278     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1279     p1.x += page->pixmap->x;
1280     p1.y += page->pixmap->y;
1281     p2.x += page->pixmap->x;
1282     p2.y += page->pixmap->y;
1283     x0 = p1.x;
1284     y0 = p1.y;
1285     x1 = p2.x;
1286     y1 = p2.y;
1287     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1288
1289     for (span = page->text; span; span = span->next) {
1290         int seen = 0;
1291
1292         /* fz_debugtextspanxml (span); */
1293         for (i = 0; i < span->len; ++i) {
1294             long c;
1295
1296             bx0 = span->text[i].bbox.x0;
1297             bx1 = span->text[i].bbox.x1;
1298             by0 = span->text[i].bbox.y0 + oy;
1299             by1 = span->text[i].bbox.y1 + oy;
1300
1301             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1302                 if (!seen) {
1303                     rx0 = bx0 - page->pixmap->x;
1304                     rx1 = bx1 - page->pixmap->x;
1305                     ry0 = by0;
1306                     ry1 = by1;
1307                 }
1308
1309                 seen = 1;
1310                 c = span->text[i].c;
1311                 if (c < 256) {
1312                     if ((isprint (c) && !isspace (c))) {
1313                         if (!rectsel) {
1314                             bx0 -= page->pixmap->x;
1315                             bx1 -= page->pixmap->x;
1316                             glEnable (GL_BLEND);
1317                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1318                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1319                             glColor4f (0.5, 0.5, 0.0, 0.6);
1320                             glRecti (bx0, by0, bx1, by1);
1321                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1322                             glDisable (GL_BLEND);
1323                         }
1324                         if (isprint (c) || c ==' ') {
1325                             rx1 = bx1;
1326                             ry1 = by1;
1327                         }
1328                     }
1329                     putc (c, stdout);
1330                 }
1331                 else  {
1332                     putc ('?', stdout);
1333                 }
1334             }
1335         }
1336
1337         if (rectsel) {
1338             if (seen) {
1339                 glEnable (GL_BLEND);
1340                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1341                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1342                 glColor4f (0.5, 0.5, 0.0, 0.6);
1343                 glRecti (rx0, ry0, rx1, ry1);
1344                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1345                 glDisable (GL_BLEND);
1346             }
1347         }
1348
1349         if (seen && span->eol) {
1350             x0 = page->pixmap->x;
1351             putc ('\n', stdout);
1352         }
1353     }
1354     unlock ("ml_gettext");
1355
1356  done:
1357     CAMLreturn (Val_unit);
1358 }
1359
1360 CAMLprim value ml_getpagewh (value pagedimno_v)
1361 {
1362     CAMLparam1 (pagedimno_v);
1363     CAMLlocal1 (ret_v);
1364     int pagedimno = Int_val (pagedimno_v);
1365
1366     ret_v = caml_alloc_small (4 * Double_wosize, Double_array_tag);
1367     Store_double_field (ret_v, 0, state.pagedims[pagedimno].box.x0);
1368     Store_double_field (ret_v, 1, state.pagedims[pagedimno].box.x1);
1369     Store_double_field (ret_v, 2, state.pagedims[pagedimno].box.y0);
1370     Store_double_field (ret_v, 3, state.pagedims[pagedimno].box.y1);
1371     CAMLreturn (ret_v);
1372 }
1373
1374 CAMLprim value ml_gettitle (value unit_v)
1375 {
1376     CAMLparam1 (unit_v);
1377     CAMLlocal1 (title_v);
1378     fz_obj *obj;
1379
1380     obj = fz_dictgets (state.xref->trailer, "Info");
1381     title_v = Val_int (0);
1382     if (obj) {
1383         obj = fz_dictgets (obj, "Title");
1384         if (obj) {
1385             char *utf8 = pdf_toutf8 (obj);
1386             if (*utf8) {
1387                 title_v = caml_alloc_small (1, 1);
1388                 Store_field (title_v, 0, caml_copy_string (utf8));
1389             }
1390         }
1391     }
1392     CAMLreturn (title_v);
1393 }
1394
1395 static void initgl (void)
1396 {
1397 #ifdef _BIG_ENDIAN
1398     if (strstr ((char *) glGetString (GL_EXTENSIONS),
1399                 "GL_ATI_fragment_shader")) {
1400         /* Here, with MESA, rv280, powerpc32: BGRA(rev) is slow while
1401            ABGR is fast, so fix things in the shader */
1402         state.texform = GL_ABGR_EXT;
1403         state.texty = GL_UNSIGNED_INT_8_8_8_8;
1404
1405         glBindFragmentShaderATI (1);
1406         glBeginFragmentShaderATI ();
1407         {
1408             glSampleMapATI (GL_REG_0_ATI, GL_TEXTURE0_ARB, GL_SWIZZLE_STR_ATI);
1409
1410             glColorFragmentOp1ATI (GL_MOV_ATI,
1411                                    GL_REG_1_ATI, GL_RED_BIT_ATI, GL_NONE,
1412                                    GL_REG_0_ATI, GL_BLUE, GL_NONE);
1413             glColorFragmentOp1ATI (GL_MOV_ATI,
1414                                    GL_REG_1_ATI, GL_BLUE_BIT_ATI, GL_NONE,
1415                                    GL_REG_0_ATI, GL_RED, GL_NONE);
1416             glColorFragmentOp1ATI (
1417                 GL_MOV_ATI,
1418                 GL_REG_0_ATI, GL_RED_BIT_ATI | GL_BLUE_BIT_ATI, GL_NONE,
1419                 GL_REG_1_ATI, GL_NONE, GL_NONE
1420                 );
1421         }
1422         glEndFragmentShaderATI ();
1423         state.useatifs = 1;
1424     }
1425     else {
1426         state.texform = GL_BGRA_EXT;
1427         state.texty = GL_UNSIGNED_INT_8_8_8_8_REV;
1428     }
1429 #else
1430     state.texform = GL_BGRA_EXT;
1431     state.texty = GL_UNSIGNED_INT_8_8_8_8;
1432 #endif
1433 }
1434
1435 CAMLprim value ml_init (value sock_v)
1436 {
1437     int ret;
1438     CAMLparam1 (sock_v);
1439
1440     state.texcount = 128;
1441     state.sliceheight = 64;
1442
1443     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1444     if (!state.texids) {
1445         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1446     }
1447
1448     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1449     if (!state.texowners) {
1450         err (1, "calloc texowners %zu",
1451              state.texcount * sizeof (*state.texowners));
1452     }
1453
1454     glGenTextures (state.texcount, state.texids);
1455
1456     state.sock = Int_val (sock_v);
1457     initgl ();
1458
1459     state.cache = fz_newglyphcache ();
1460     if (!state.cache) {
1461         errx (1, "fz_newglyphcache failed");
1462     }
1463
1464     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1465     if (ret) {
1466         errx (1, "pthread_create: %s", strerror (errno));
1467     }
1468
1469     CAMLreturn (Val_unit);
1470 }