link.c

   1 /* lot's of code c&p-ed directly from mupdf */
   2 #ifdef _WIN32
   3 #define WIN32_LEAN_AND_MEAN
   4 #include <windows.h>
   5 #include <winsock2.h>
   6 #define ssize_t int
   7 #pragma warning (disable:4244)
   8 #pragma warning (disable:4996)
   9 #pragma warning (disable:4995)
  10 #endif
  11
  12 #ifdef _MSC_VER
  13 #include <errno.h>
  14 #include <stdio.h>
  15 #include <string.h>
  16 #include <stdlib.h>
  17 static void __declspec (noreturn) err (int exitcode, const char *fmt, ...)
  18 {
  19     va_list ap;
  20     int errcode;
  21
  22     errcode = errno;
  23     va_start (ap, fmt);
  24     vfprintf (stderr, fmt, ap);
  25     va_end (ap);
  26     fprintf (stderr, ": %s\n", strerror (errno));
  27     exit (exitcode);
  28 }
  29 static void __declspec (noreturn) errx (int exitcode, const char *fmt, ...)
  30 {
  31     va_list ap;
  32     int errcode;
  33
  34     errcode = errno;
  35     va_start (ap, fmt);
  36     vfprintf (stderr, fmt, ap);
  37     va_end (ap);
  38     fputc ('\n', stderr);
  39     exit (exitcode);
  40 }
  41 static void __declspec (noreturn) sockerr (int exitcode, const char *fmt, ...)
  42 {
  43     va_list ap;
  44     int errcode;
  45
  46     errcode = errno;
  47     va_start (ap, fmt);
  48     vfprintf (stderr, fmt, ap);
  49     va_end (ap);
  50     fprintf (stderr, ": wsaerror %lx\n", WSAGetLastError ());
  51     exit (exitcode);
  52 }
  53 #else
  54 #define _GNU_SOURCE
  55 #include <err.h>
  56 #define sockerr err
  57 #endif
  58 #include <regex.h>
  59 #include <errno.h>
  60 #include <ctype.h>
  61 #include <stdio.h>
  62 #include <stdarg.h>
  63 #include <stdlib.h>
  64 #include <string.h>
  65 #ifndef _WIN32
  66 #include <pthread.h>
  67 #include <sys/time.h>
  68 #include <sys/types.h>
  69 #include <sys/socket.h>
  70 #include <sys/select.h>
  71 #endif
  72
  73 #ifdef __APPLE__
  74 #include <OpenGL/gl.h>
  75 #else
  76 #include <GL/gl.h>
  77 #endif
  78
  79 #ifndef GL_TEXTURE_RECTANGLE_ARB
  80 #define GL_TEXTURE_RECTANGLE_ARB          0x84F5
  81 #endif
  82
  83 #include <caml/fail.h>
  84 #include <caml/alloc.h>
  85 #include <caml/memory.h>
  86 #include <caml/unixsupport.h>
  87
  88 #include <fitz.h>
  89 #include <mupdf.h>
  90
  91 #if 0
  92 #define lprintf printf
  93 #else
  94 #define lprintf(...)
  95 #endif
  96
  97 #define ARSERT(cond) for (;;) {                         \
  98     if (!(cond)) {                                      \
  99         errx (1, "%s:%d " #cond, __FILE__, __LINE__);   \
 100     }                                                   \
 101     break;                                              \
 102 }
 103
 104 struct slice {
 105     int texindex;
 106     int w, h;
 107 };
 108
 109 struct page {
 110     int pageno;
 111     int slicecount;
 112     fz_textspan *text;
 113     fz_pixmap *pixmap;
 114     pdf_page *drawpage;
 115     struct pagedim *pagedim;
 116     struct page *prev;
 117     struct slice slices[];
 118 };
 119
 120 struct pagedim {
 121     int pageno;
 122     int rotate;
 123     fz_rect box;
 124     fz_bbox bbox;
 125     fz_matrix ctm;
 126 };
 127
 128 struct {
 129     int sock;
 130     int sliceheight;
 131     struct page *pages;
 132     struct pagedim *pagedims;
 133     int pagecount;
 134     int pagedimcount;
 135     pdf_xref *xref;
 136     fz_glyphcache *cache;
 137     int w, h;
 138
 139     int texindex;
 140     int texcount;
 141     GLuint *texids;
 142
 143     GLenum texform;
 144     GLenum texty;
 145
 146     int lotsamemory;
 147
 148     int *pagetbl;
 149     struct {
 150         int w, h;
 151         struct slice *slice;
 152     } *texowners;
 153
 154     int rotate;
 155
 156 #ifdef _WIN32
 157     HANDLE thread;
 158 #else
 159     pthread_t thread;
 160 #endif
 161 } state;
 162
 163 #ifdef _WIN32
 164 static CRITICAL_SECTION critsec;
 165
 166 static void lock (void *unused)
 167 {
 168     (void) unused;
 169     EnterCriticalSection (&critsec);
 170 }
 171
 172 static void unlock (void *unused)
 173 {
 174     (void) unused;
 175     LeaveCriticalSection (&critsec);
 176 }
 177
 178 static int trylock (void *unused)
 179 {
 180     return TryEnterCriticalSection (&critsec) == 0;
 181 }
 182 #else
 183 static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
 184
 185 static void lock (const char *cap)
 186 {
 187     int ret = pthread_mutex_lock (&mutex);
 188     if (ret) {
 189         errx (1, "%s: pthread_mutex_lock: %s", cap, strerror (ret));
 190     }
 191 }
 192
 193 static void unlock (const char *cap)
 194 {
 195     int ret = pthread_mutex_unlock (&mutex);
 196     if (ret) {
 197         errx (1, "%s: pthread_mutex_unlock: %s", cap, strerror (ret));
 198     }
 199 }
 200
 201 static int trylock (const char *cap)
 202 {
 203     int ret = pthread_mutex_trylock (&mutex);
 204
 205     if (ret && ret != EBUSY) {
 206         errx (1, "%s: pthread_mutex_trylock: %s", cap, strerror (ret));
 207     }
 208     return ret == EBUSY;
 209 }
 210 #endif
 211
 212 static void *parse_pointer (const char *cap, const char *s)
 213 {
 214     int ret;
 215     void *ptr;
 216
 217     ret = sscanf (s, "%p", &ptr);
 218     if (ret != 1) {
 219         errx (1, "%s: cannot parse pointer in `%s'", cap, s);
 220     }
 221     return ptr;
 222 }
 223
 224 static int hasdata (int sock)
 225 {
 226     int n;
 227     fd_set s;
 228     struct timeval tv;
 229     FD_ZERO (&s);
 230     FD_SET (sock, &s);
 231     tv.tv_sec = 0;
 232     tv.tv_usec = 0;
 233     n = select (sock + 1, &s, NULL, NULL, &tv);
 234     if (n == 0) return 0;
 235     if (n == 1) return 1;
 236     sockerr (1, "hasdata: select error ret=%d", n);
 237 }
 238
 239 static double now (void)
 240 {
 241     struct timeval tv;
 242
 243     if (gettimeofday (&tv, NULL)) {
 244         err (1, "gettimeofday");
 245     }
 246     return tv.tv_sec + tv.tv_usec*1e-6;
 247 }
 248
 249 static void readdata (int fd, char *p, int size)
 250 {
 251     ssize_t n;
 252
 253     n = recv (fd, p, size, 0);
 254     if (n - size) {
 255         if (!n) errx (1, "EOF while reading");
 256         sockerr (1, "recv (req %d, ret %zd)", size, n);
 257     }
 258 }
 259
 260 static void writedata (int fd, char *p, int size)
 261 {
 262     char buf[4];
 263     ssize_t n;
 264
 265     buf[0] = (size >> 24) & 0xff;
 266     buf[1] = (size >> 16) & 0xff;
 267     buf[2] = (size >>  8) & 0xff;
 268     buf[3] = (size >>  0) & 0xff;
 269
 270     n = send (fd, buf, 4, 0);
 271     if (n != 4) {
 272         if (!n) errx (1, "EOF while writing length");
 273         sockerr (1, "send %zd", n);
 274     }
 275
 276     n = send (fd, p, size, 0);
 277     if (n - size) {
 278         if (!n) errx (1, "EOF while writing data");
 279         sockerr (1, "send (req %d, ret %zd)", size, n);
 280     }
 281 }
 282
 283 static void
 284 #ifdef __GNUC__
 285 __attribute__ ((format (printf, 2, 3)))
 286 #endif
 287     printd (int fd, const char *fmt, ...)
 288 {
 289     int len;
 290     va_list ap;
 291     char buf[200];
 292
 293     va_start (ap, fmt);
 294     len = vsnprintf (buf, sizeof (buf), fmt, ap);
 295     va_end (ap);
 296     writedata (fd, buf, len);
 297 }
 298
 299 static void die (fz_error error)
 300 {
 301     fz_catch (error, "aborting");
 302     if (state.xref)
 303        pdf_closexref (state.xref);
 304     exit (1);
 305 }
 306
 307 static void openxref (char *filename)
 308 {
 309     int fd;
 310     fz_error error;
 311     fz_stream *file;
 312
 313     fd = open (filename, O_BINARY | O_RDONLY, 0666);
 314     if (fd < 0)
 315         die (fz_throw ("cannot open file '%s'", filename));
 316
 317     file = fz_openfile (fd);
 318     state.xref = pdf_openxref (file);
 319     if (!state.xref)
 320         die (fz_throw ("cannot open PDF file '%s'", filename));
 321     fz_dropstream (file);
 322
 323     if (pdf_needspassword (state.xref)) {
 324         die (fz_throw ("password protected"));
 325     }
 326
 327     error = pdf_loadpagetree (state.xref);
 328     if (error) {
 329         die (fz_throw ("cannot load page tree"));
 330     }
 331
 332     state.pagecount = pdf_getpagecount (state.xref);
 333     state.pagetbl = stat_alloc (state.pagecount * sizeof (*state.pagetbl));
 334 }
 335
 336 static int readlen (int fd)
 337 {
 338     ssize_t n;
 339     char p[4];
 340
 341     n = recv (fd, p, 4, 0);
 342     if (n != 4) {
 343         if (!n) errx (1, "EOF while reading length");
 344         sockerr (1, "read %zd", n);
 345     }
 346
 347     return (p[0] << 24) | (p[1] << 16) | (p[2] << 8) | p[3];
 348 }
 349
 350 static void freepage (struct page *page)
 351 {
 352     int i;
 353     struct page *p;
 354
 355     fz_droppixmap (page->pixmap);
 356     for (p = state.pages; p; p = p->prev) {
 357         if (p->prev == page) {
 358             p->prev = page->prev;
 359             break;
 360         }
 361     }
 362     for (i = 0; i < page->slicecount; ++i) {
 363         struct slice *s = &page->slices[i];
 364         if (s->texindex != -1) {
 365             if (state.texowners[s->texindex].slice == s) {
 366                 state.texowners[s->texindex].slice = NULL;
 367                 ARSERT (state.texowners[s->texindex].w == s->w);
 368                 ARSERT (state.texowners[s->texindex].h >= s->h);
 369             }
 370         }
 371     }
 372     if (page->text) {
 373         fz_freetextspan (page->text);
 374     }
 375     if (page->drawpage) {
 376         pdf_freepage (page->drawpage);
 377     }
 378
 379     free (page);
 380 }
 381
 382 static void subdivide (struct page *p)
 383 {
 384     int i;
 385     int h = p->pixmap->h;
 386     int th = MIN (h, state.sliceheight);
 387
 388     for (i = 0; i < p->slicecount; ++i) {
 389         struct slice *s = &p->slices[i];
 390         s->texindex = -1;
 391         s->h = MIN (th, h);
 392         s->w = p->pixmap->w;
 393         h -= th;
 394     }
 395 }
 396
 397 static void *render (int pageno, int pindex)
 398 {
 399     fz_error error;
 400     int slicecount;
 401     fz_obj *pageobj;
 402     struct page *page;
 403     double start, end;
 404     pdf_page *drawpage;
 405     fz_device *idev;
 406     struct pagedim *pagedim;
 407
 408     start = now ();
 409     printd (state.sock, "V rendering %d", pageno);
 410     pdf_flushxref (state.xref, 0);
 411
 412     pagedim = &state.pagedims[pindex];
 413     slicecount = (pagedim->bbox.y1 - pagedim->bbox.y0
 414                   + state.sliceheight - 1) / state.sliceheight;
 415     slicecount += slicecount == 0;
 416
 417     page = calloc (sizeof (*page)
 418                    + (slicecount * sizeof (struct slice)), 1);
 419     if (!page) {
 420         err (1, "calloc page %d\n", pageno);
 421     }
 422     page->slicecount = slicecount;
 423     page->prev = state.pages;
 424     state.pages = page;
 425
 426     pageobj = pdf_getpageobject (state.xref, pageno);
 427     if (!pageobj)
 428         die (fz_throw ("cannot retrieve info from page %d", pageno));
 429
 430     error = pdf_loadpage (&drawpage, state.xref, pageobj);
 431     if (error)
 432         die (error);
 433
 434     page->pixmap = fz_newpixmapwithrect (pdf_devicergb, pagedim->bbox);
 435     if (error)
 436         die (error);
 437     fz_clearpixmap (page->pixmap, 0xFF);
 438
 439     idev = fz_newdrawdevice (state.cache, page->pixmap);
 440     if (!idev)
 441         die (fz_throw ("fz_newdrawdevice failed"));
 442     error = pdf_runcontentstream (idev, pagedim->ctm, state.xref,
 443                                   drawpage->resources,
 444                                   drawpage->contents);
 445     fz_freedevice (idev);
 446
 447     page->drawpage = drawpage;
 448     page->pagedim = pagedim;
 449     page->pageno = pageno;
 450     subdivide (page);
 451     end = now ();
 452
 453     if (!state.lotsamemory) {
 454         pdf_agestoreditems (state.xref->store);
 455         pdf_evictageditems (state.xref->store);
 456     }
 457
 458     printd (state.sock, "V rendering %d took %f sec", pageno, end - start);
 459     return page;
 460 }
 461
 462 static void initpdims (void)
 463 {
 464     int pageno;
 465     double start, end;
 466
 467     start = now ();
 468     for (pageno = 0; pageno < state.pagecount; ++pageno) {
 469         int rotate;
 470         fz_rect box;
 471         struct pagedim *p;
 472         fz_obj *obj, *pageobj;
 473
 474         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 475
 476         obj = fz_dictgets (pageobj, "CropBox");
 477         if (!fz_isarray (obj)) {
 478             obj = fz_dictgets (pageobj, "MediaBox");
 479             if (!fz_isarray (obj)) {
 480                 die (fz_throw ("cannot find page bounds %d (%d Rd)",
 481                                fz_tonum (pageobj), fz_togen (pageobj)));
 482             }
 483         }
 484         box = pdf_torect (obj);
 485
 486         obj = fz_dictgets (pageobj, "Rotate");
 487         if (fz_isint (obj)) {
 488             rotate = fz_toint (obj);
 489         }
 490         else  {
 491             rotate = 0;
 492         }
 493         rotate += state.rotate;
 494
 495         state.pagetbl[pageno] = fz_tonum (state.xref->pagerefs[pageno]);
 496         p = &state.pagedims[state.pagedimcount - 1];
 497         if ((state.pagedimcount == 0)
 498             || (p->rotate != rotate || memcmp (&p->box, &box, sizeof (box)))) {
 499             size_t size;
 500
 501             size = (state.pagedimcount + 1) * sizeof (*state.pagedims);
 502             state.pagedims = realloc (state.pagedims, size);
 503             if (!state.pagedims) {
 504                 err (1, "realloc pagedims to %zu (%d elems)",
 505                      size, state.pagedimcount + 1);
 506             }
 507             p = &state.pagedims[state.pagedimcount++];
 508             p->rotate = rotate;
 509             p->box = box;
 510             p->pageno = pageno;
 511         }
 512     }
 513     end = now ();
 514     printd (state.sock, "T Processed %d pages in %f seconds",
 515             state.pagecount, end - start);
 516 }
 517
 518 static void layout (void)
 519 {
 520     int pindex;
 521     fz_matrix ctm;
 522     fz_rect box, box2;
 523     double zoom, w;
 524     struct pagedim *p = state.pagedims;
 525
 526     pindex = 0;
 527     for (pindex = 0; pindex < state.pagedimcount; ++pindex, ++p) {
 528         box.x0 = MIN (p->box.x0, p->box.x1);
 529         box.y0 = MIN (p->box.y0, p->box.y1);
 530         box.x1 = MAX (p->box.x0, p->box.x1);
 531         box.y1 = MAX (p->box.y0, p->box.y1);
 532
 533         ctm = fz_identity ();
 534         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 535         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 536         box2 = fz_transformrect (ctm, box);
 537         w = box2.x1 - box2.x0;
 538
 539         zoom = (state.w / w);
 540         ctm = fz_identity ();
 541         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 542         ctm = fz_concat (ctm, fz_scale (zoom, -zoom));
 543         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 544         p->bbox = fz_roundrect (fz_transformrect (ctm, box));
 545         memcpy (&p->ctm, &ctm, sizeof (ctm));
 546     }
 547
 548     while (p-- != state.pagedims)  {
 549         printd (state.sock, "l %d %d %d",
 550                 p->pageno, p->bbox.x1 - p->bbox.x0, p->bbox.y1 - p->bbox.y0);
 551     }
 552 }
 553
 554 static void recurse_outline (pdf_outline *outline, int level)
 555 {
 556     while (outline) {
 557         int i, num;
 558         fz_obj *obj;
 559         int top = 0;
 560         int pageno = -1;
 561
 562         if (!outline->link) goto next;
 563
 564         obj = outline->link->dest;
 565         if (fz_isindirect (obj)) {
 566             num = fz_tonum (obj);
 567
 568             for (i = 0; i < state.pagecount; ++i)  {
 569                 if (state.pagetbl[i] == num) {
 570                     pageno = i;
 571                     break;
 572                 }
 573             }
 574         }
 575         else if (fz_isarray (obj)) {
 576             fz_obj *obj2;
 577
 578             obj2 = fz_arrayget (obj, 0);
 579             if (fz_isint (obj2)) {
 580                 pageno = fz_toint (obj2);
 581             }
 582             else {
 583                 num = fz_tonum (obj2);
 584                 for (i = 0; i < state.pagecount; ++i)  {
 585                     if (state.pagetbl[i] == num) {
 586                         pageno = i;
 587                         break;
 588                     }
 589                 }
 590             }
 591
 592             if (fz_arraylen (obj) > 3) {
 593                 fz_point p;
 594                 struct pagedim *pagedim = state.pagedims;
 595
 596                 for (i = 0; i < state.pagedimcount; ++i) {
 597                     if (state.pagedims[i].pageno > pageno)
 598                         break;
 599                     pagedim = &state.pagedims[i];
 600                 }
 601
 602                 p.x = fz_toint (fz_arrayget (obj, 2));
 603                 p.y = fz_toint (fz_arrayget (obj, 3));
 604                 p = fz_transformpoint (pagedim->ctm, p);
 605                 top = p.y;
 606             }
 607         }
 608
 609         lprintf ("%*c%s %d\n", level, ' ', outline->title, pageno);
 610         printd (state.sock, "o %d %d %d %s",
 611                 level, pageno, top, outline->title);
 612     next:
 613         if (outline->child) {
 614             recurse_outline (outline->child, level + 1);
 615         }
 616         outline = outline->next;
 617     }
 618 }
 619
 620 static void process_outline (void)
 621 {
 622     pdf_outline *outline;
 623
 624     outline = pdf_loadoutline (state.xref);
 625     if (outline) {
 626         recurse_outline (outline, 0);
 627         pdf_freeoutline (outline);
 628     }
 629 }
 630
 631 static int comparespans (const void *l, const void *r)
 632 {
 633     fz_textspan const *const*ls = l;
 634     fz_textspan const *const*rs = r;
 635     return (*ls)->text->bbox.y0 - (*rs)->text->bbox.y0;
 636 }
 637
 638 /* wishful thinking function */
 639 static void search (regex_t *re, int pageno, int y, int forward)
 640 {
 641     int i, j;
 642     int ret;
 643     char *p;
 644     char buf[256];
 645     fz_error error;
 646     fz_obj *pageobj;
 647     fz_device *tdev;
 648     pdf_page *drawpage;
 649     fz_textspan *text, *span, **pspan;
 650     struct pagedim *pdim, *pdimprev;
 651     int stop = 0;
 652     int niters = 0;
 653     int nspans;
 654     double start, end;
 655
 656     start = now ();
 657     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 658         if (niters++ == 5) {
 659             if (!state.lotsamemory) {
 660                 pdf_agestoreditems (state.xref->store);
 661                 pdf_evictageditems (state.xref->store);
 662             }
 663             niters = 0;
 664             if (hasdata (state.sock)) {
 665                 printd (state.sock, "T attention requested aborting search at %d",
 666                         pageno);
 667                 stop = 1;
 668             }
 669             else {
 670                 printd (state.sock, "T searching in page %d", pageno);
 671             }
 672         }
 673         pdimprev = NULL;
 674         for (i = 0; i < state.pagedimcount; ++i) {
 675             pdim = &state.pagedims[i];
 676             if (pdim->pageno == pageno)  {
 677                 goto found;
 678             }
 679             if (pdim->pageno > pageno) {
 680                 pdim = pdimprev;
 681                 goto found;
 682             }
 683             pdimprev = pdim;
 684         }
 685         pdim = pdimprev;
 686     found:
 687
 688         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 689         if (!pageobj)
 690             die (fz_throw ("cannot retrieve info from page %d", pageno));
 691
 692         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 693         if (error)
 694             die (error);
 695
 696         text = fz_newtextspan ();
 697         tdev = fz_newtextdevice (text);
 698         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 699                                       drawpage->resources,
 700                                       drawpage->contents);
 701         if (error) die (error);
 702         fz_freedevice (tdev);
 703
 704         nspans = 0;
 705         for (span = text; span; span = span->next) {
 706             nspans++;
 707         }
 708         pspan = malloc (sizeof (void *) * nspans);
 709         if (!pspan) {
 710             err (1, "malloc span pointers %zu", sizeof (void *) * nspans);
 711         }
 712         for (i = 0, span = text; span; span = span->next, ++i) {
 713             pspan[i] = span;
 714         }
 715         qsort (pspan, nspans, sizeof (fz_textspan *), comparespans);
 716
 717         j = forward ? 0 : nspans - 1;
 718         while (nspans--) {
 719             regmatch_t rm;
 720
 721             span = pspan[j];
 722             j += forward ? 1 : -1;
 723             p = buf;
 724             for (i = 0; i < MIN (span->len, (int) sizeof (buf) - 1); ++i) {
 725                 if (forward) {
 726                     if (span->text[i].bbox.y0 < y + 1) {
 727                         continue;
 728                     }
 729                 }
 730                 else {
 731                     if (span->text[i].bbox.y0 > y - 1) {
 732                         continue;
 733                     }
 734                 }
 735                 if (span->text[i].c < 256) {
 736                     *p++ = span->text[i].c;
 737                 }
 738                 else  {
 739                     *p++ = '?';
 740                 }
 741             }
 742             if (p == buf) {
 743                 continue;
 744             }
 745             *p++ = 0;
 746
 747             ret = regexec (re, buf, 1, &rm, 0);
 748             if (ret)  {
 749                 if (ret != REG_NOMATCH) {
 750                     size_t size;
 751                     char errbuf[80];
 752                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 753                     printd (state.sock,
 754                             "T regexec error `%.*s'",
 755                             (int) size, errbuf);
 756                     fz_freetextspan (text);
 757                     pdf_freepage (drawpage);
 758                     free (pspan);
 759                     return;
 760                 }
 761             }
 762             else  {
 763                 fz_rect r;
 764
 765                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 766                 r.y0 = span->text[rm.rm_so].bbox.y0;
 767                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 768                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 769
 770                 if (!stop) {
 771                     printd (state.sock, "F %d %d %f %f %f %f",
 772                             pageno, 1,
 773                             r.x0, r.y0,
 774                             r.x1, r.y1);
 775                 }
 776                 else  {
 777                     printd (state.sock, "R %d %d %f %f %f %f",
 778                             pageno, 2,
 779                             r.x0, r.y0,
 780                             r.x1, r.y1);
 781                 }
 782                 printd (state.sock, "T found at %d `%.*s' %f in %f sec",
 783                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 784                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 785                         now () - start);
 786                 stop = 1;
 787             }
 788         }
 789         if (forward) {
 790             pageno += 1;
 791             y = 0;
 792         }
 793         else {
 794             pageno -= 1;
 795             y = INT_MAX;
 796         }
 797         fz_freetextspan (text);
 798         pdf_freepage (drawpage);
 799         free (pspan);
 800     }
 801     end = now ();
 802     if (!stop)  {
 803         printd (state.sock, "T no matches %f sec", end - start);
 804     }
 805     printd (state.sock, "D");
 806 }
 807
 808 static
 809 #ifdef _WIN32
 810 DWORD _stdcall
 811 #else
 812 void *
 813 #endif
 814 mainloop (void *unused)
 815 {
 816     char *p = NULL;
 817     int len, ret, oldlen = 0;
 818
 819     for (;;) {
 820         len = readlen (state.sock);
 821         if (len == 0) {
 822             errx (1, "readlen returned 0");
 823         }
 824
 825         if (oldlen < len + 1) {
 826             p = realloc (p, len + 1);
 827             if (!p) {
 828                 err (1, "realloc %d failed", len + 1);
 829             }
 830             oldlen = len + 1;
 831         }
 832         readdata (state.sock, p, len);
 833         p[len] = 0;
 834
 835         if (!strncmp ("open", p, 4)) {
 836             fz_obj *obj;
 837             char *filename = p + 5;
 838
 839             openxref (filename);
 840             initpdims ();
 841
 842             obj = fz_dictgets (state.xref->trailer, "Info");
 843             if (obj) {
 844                 obj = fz_dictgets (obj, "Title");
 845                 if (obj) {
 846                     printd (state.sock, "t %s", pdf_toutf8 (obj));
 847                 }
 848             }
 849         }
 850         else if (!strncmp ("free", p, 4)) {
 851             void *ptr;
 852
 853             ret = sscanf (p + 4, " %p", &ptr);
 854             if (ret != 1) {
 855                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 856             }
 857             lock ("free");
 858             freepage (ptr);
 859             unlock ("free");
 860             printd (state.sock, "d");
 861         }
 862         else if (!strncmp ("search", p, 6)) {
 863             int icase, pageno, y, ret, len2, forward;
 864             char *pattern;
 865             regex_t re;
 866
 867             ret = sscanf (p + 6, " %d %d %d %d,%n",
 868                           &icase, &pageno, &y, &forward, &len2);
 869             if (ret != 4) {
 870                 errx (1, "malformed search `%s' ret=%d", p, ret);
 871             }
 872
 873             pattern = p + 6 + len2;
 874             ret = regcomp (&re, pattern,
 875                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 876             if (ret) {
 877                 char errbuf[80];
 878                 size_t size;
 879
 880                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 881                 printd (state.sock, "T regcomp failed `%.*s'",
 882                         (int) size, errbuf);
 883             }
 884             else  {
 885                 search (&re, pageno, y, forward);
 886                 regfree (&re);
 887             }
 888         }
 889         else if (!strncmp ("geometry", p, 8)) {
 890             int w, h;
 891
 892             printd (state.sock, "c");
 893             ret = sscanf (p + 8, " %d %d", &w, &h);
 894             if (ret != 2) {
 895                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 896             }
 897             state.h = h;
 898             if (w != state.w) {
 899                 int i;
 900                 state.w = w;
 901                 for (i = 0; i < state.texcount; ++i)  {
 902                     state.texowners[i].slice = NULL;
 903                 }
 904             }
 905             lock ("geometry");
 906             layout ();
 907             process_outline ();
 908             unlock ("geometry");
 909             printd (state.sock, "C %d", state.pagecount);
 910         }
 911         else if (!strncmp ("rotate", p, 6)) {
 912             float rotate;
 913
 914             printd (state.sock, "c");
 915             ret = sscanf (p + 6, " %f", &rotate);
 916             if (ret != 1) {
 917                 errx (1, "bad rotate line `%.*s' ret=%d", len, p, ret);
 918             }
 919             state.rotate = rotate;
 920             lock ("rotate");
 921             state.pagedimcount = 0;
 922             free (state.pagedims);
 923             state.pagedims = NULL;
 924             initpdims ();
 925             layout ();
 926             process_outline ();
 927             unlock ("rotate");
 928             printd (state.sock, "C %d", state.pagecount);
 929         }
 930         else if (!strncmp ("render", p, 6)) {
 931             int pageno, pindex, w, h, ret;
 932             struct page *page;
 933
 934             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 935             if (ret != 4) {
 936                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 937             }
 938
 939             page = render (pageno, pindex);
 940             printd (state.sock, "r %d %d %d %d %p\n",
 941                     pageno,
 942                     state.w,
 943                     state.h,
 944                     state.rotate,
 945                     page);
 946         }
 947         else {
 948             errx (1, "unknown command %.*s", len, p);
 949         }
 950     }
 951     return 0;
 952 }
 953
 954 static void upload2 (struct page *page, int slicenum, const char *cap)
 955 {
 956     int i;
 957     int w, h;
 958     double start, end;
 959     struct slice *slice = &page->slices[slicenum];
 960
 961     w = page->pixmap->w;
 962     h = page->pixmap->h;
 963
 964     ARSERT (w == slice->w);
 965     if (slice->texindex != -1
 966         && state.texowners[slice->texindex].slice == slice) {
 967         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 968     }
 969     else  {
 970         int subimage = 0;
 971         int index = (state.texindex++ % state.texcount);
 972         size_t offset = 0;
 973
 974         for (i = 0; i < slicenum; ++i) {
 975             offset += w * page->slices[i].h * 4;
 976         }
 977
 978         if (state.texowners[index].w == slice->w) {
 979             if (state.texowners[index].h >= slice->h ) {
 980                 subimage = 1;
 981             }
 982             else {
 983                 state.texowners[index].h = slice->h;
 984             }
 985         }
 986         else  {
 987             state.texowners[index].h = slice->h;
 988         }
 989
 990         state.texowners[index].slice = slice;
 991         state.texowners[index].w = slice->w;
 992         slice->texindex = index;
 993
 994         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 995         start = now ();
 996         if (subimage) {
 997             {
 998                 GLenum err = glGetError ();
 999                 if (err != GL_NO_ERROR) {
1000                     printf ("\033[0;31mERROR1 %d %d %#x\033[0m\n", w, slice->h, err);
1001                     abort ();
1002                 }
1003             }
1004             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
1005                              0,
1006                              0,
1007                              0,
1008                              w,
1009                              slice->h,
1010                              state.texform,
1011                              state.texty,
1012                              page->pixmap->samples + offset
1013                 );
1014             {
1015                 GLenum err = glGetError ();
1016                 if (err != GL_NO_ERROR) {
1017                     printf ("\033[0;31mERROR %d %d %#x\033[0m\n", w, slice->h, err);
1018                     abort ();
1019                 }
1020             }
1021         }
1022         else {
1023             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
1024                           0,
1025                           GL_RGBA8,
1026                           w,
1027                           slice->h,
1028                           0,
1029                           state.texform,
1030                           state.texty,
1031                           page->pixmap->samples + offset
1032                 );
1033         }
1034
1035         end = now ();
1036         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
1037                  subimage ? "sub" : "img",
1038                  page->pageno, slicenum,
1039                  slice->w, slice->h,
1040                  state.texids[slice->texindex],
1041                  end - start);
1042     }
1043 }
1044
1045 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
1046                         value py_v, value ptr_v)
1047 {
1048     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
1049     int dispy = Int_val (dispy_v);
1050     int w = Int_val (w_v);
1051     int h = Int_val (h_v);
1052     int py = Int_val (py_v);
1053     char *s = String_val (ptr_v);
1054     int ret;
1055     void *ptr;
1056     struct page *page;
1057     int slicenum = 0;
1058
1059     if (trylock ("ml_draw")) {
1060         goto done;
1061     }
1062
1063     ret = sscanf (s, "%p", &ptr);
1064     if (ret != 1) {
1065         errx (1, "cannot parse pointer `%s'", s);
1066     }
1067     page = ptr;
1068
1069     w = page->pixmap->w;
1070
1071     ARSERT (h >= 0 && "ml_draw wrong h");
1072
1073     glEnable (GL_TEXTURE_RECTANGLE_ARB);
1074
1075     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
1076         struct slice *slice = &page->slices[slicenum];
1077         if (slice->h > py) {
1078             break;
1079         }
1080         py -= slice->h;
1081     }
1082
1083     h = MIN (state.h, h);
1084     while (h) {
1085         int th;
1086         struct slice *slice = &page->slices[slicenum];
1087
1088         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
1089
1090         th = MIN (h, slice->h - py);
1091         upload2 (page, slicenum, "upload");
1092
1093         glBegin (GL_QUADS);
1094         {
1095             glTexCoord2i (0, py);
1096             glVertex2i (0, dispy);
1097
1098             glTexCoord2i (w, py);
1099             glVertex2i (w, dispy);
1100
1101             glTexCoord2i (w, py+th);
1102             glVertex2i (w, dispy + th);
1103
1104             glTexCoord2i (0, py+th);
1105             glVertex2i (0, dispy + th);
1106         }
1107         glEnd ();
1108
1109         h -= th;
1110         py = 0;
1111         dispy += th;
1112         slicenum += 1;
1113     }
1114
1115     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1116
1117     unlock ("ml_draw");
1118  done:
1119     CAMLreturn (Val_unit);
1120 }
1121
1122 static pdf_link *getlink (struct page *page, int x, int y)
1123 {
1124     fz_point p;
1125     fz_matrix ctm;
1126     pdf_link *link;
1127
1128     p.x = x + page->pixmap->x;
1129     p.y = y + page->pixmap->y;
1130
1131     ctm = fz_invertmatrix (page->pagedim->ctm);
1132     p = fz_transformpoint (ctm, p);
1133
1134     for (link = page->drawpage->links; link; link = link->next) {
1135         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1136             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1137                 if (link->kind == PDF_LGOTO) {
1138                     return link;
1139                 }
1140             }
1141         }
1142     }
1143     return NULL;
1144 }
1145
1146 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1147 {
1148     CAMLparam3 (ptr_v, x_v, y_v);
1149     char *s = String_val (ptr_v);
1150     int ret;
1151
1152     if (trylock ("ml_checklink")) {
1153         ret = 0;
1154     }
1155     else {
1156         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1157                                Int_val (x_v), Int_val (y_v));
1158         unlock ("ml_checklink");
1159     }
1160     CAMLreturn (Val_bool (ret));
1161 }
1162
1163 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1164 {
1165     CAMLparam3 (ptr_v, x_v, y_v);
1166     CAMLlocal2 (ret_v, tup_v);
1167     pdf_link *link;
1168     struct page *page;
1169     char *s = String_val (ptr_v);
1170
1171     if (trylock ("ml_getlink")) {
1172         ret_v = Val_int (0);
1173         goto done;
1174     }
1175
1176     page = parse_pointer ("ml_getlink", s);
1177
1178     link = getlink (page, Int_val (x_v), Int_val (y_v));
1179     if (link) {
1180         int pageno;
1181         fz_point p;
1182         fz_obj *obj;
1183
1184         pageno = -1;
1185         obj = fz_arrayget (link->dest, 0);
1186         if (fz_isindirect (obj)) {
1187             pageno = pdf_findpageobject (state.xref, obj) - 1;
1188         }
1189         else if (fz_isint (obj)) {
1190             pageno = fz_toint (obj);
1191         }
1192
1193         if (fz_arraylen (link->dest) > 3) {
1194             p.x = fz_toint (fz_arrayget (link->dest, 2));
1195             p.y = fz_toint (fz_arrayget (link->dest, 3));
1196             p = fz_transformpoint (page->pagedim->ctm, p);
1197         }
1198         else  {
1199             p.x = 0.0;
1200             p.y = 0.0;
1201         }
1202
1203         tup_v = caml_alloc_tuple (2);
1204         ret_v = caml_alloc_small (1, 1);
1205         Field (tup_v, 0) = Val_int (pageno);
1206         Field (tup_v, 1) = Val_int (p.y);
1207         Field (ret_v, 0) = tup_v;
1208     }
1209     else {
1210         ret_v = Val_int (0);
1211     }
1212     unlock ("ml_getlink");
1213
1214  done:
1215     CAMLreturn (ret_v);
1216 }
1217
1218 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1219 {
1220     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1221     fz_matrix ctm;
1222     fz_point p1, p2;
1223     struct page *page;
1224     fz_textspan *span;
1225     char *s = String_val (ptr_v);
1226     int rectsel = Bool_val (rectsel_v);
1227     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1228
1229     /* stop GCC from complaining about uninitialized variables */
1230 #ifdef __GNUC__
1231     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1232 #else
1233     int rx0, rx1, ry0, ry1;
1234 #endif
1235
1236     if (trylock ("ml_gettext")) {
1237         goto done;
1238     }
1239
1240     page = parse_pointer ("ml_gettext", s);
1241
1242     oy = Int_val (oy_v);
1243     p1.x = Int_val (Field (rect_v, 0));
1244     p1.y = Int_val (Field (rect_v, 1));
1245     p2.x = Int_val (Field (rect_v, 2));
1246     p2.y = Int_val (Field (rect_v, 3));
1247
1248     if (0) {
1249         glEnable (GL_BLEND);
1250         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1251         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1252         glColor4f (0.0f, 0.0f, 0.0f, 0.2f);
1253         glRecti (p1.x, p1.y, p2.x, p2.y);
1254         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1255         glDisable (GL_BLEND);
1256     }
1257
1258     ctm = page->pagedim->ctm;
1259     if (!page->text) {
1260         fz_error error;
1261         fz_device *tdev;
1262
1263         page->text = fz_newtextspan ();
1264         tdev = fz_newtextdevice (page->text);
1265         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1266                                       page->drawpage->resources,
1267                                       page->drawpage->contents);
1268         if (error) die (error);
1269         fz_freedevice (tdev);
1270     }
1271
1272     printf ("\033c");
1273
1274     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1275     p1.x += page->pixmap->x;
1276     p1.y += page->pixmap->y;
1277     p2.x += page->pixmap->x;
1278     p2.y += page->pixmap->y;
1279     x0 = p1.x;
1280     y0 = p1.y;
1281     x1 = p2.x;
1282     y1 = p2.y;
1283     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1284
1285     for (span = page->text; span; span = span->next) {
1286         int seen = 0;
1287
1288         /* fz_debugtextspanxml (span); */
1289         for (i = 0; i < span->len; ++i) {
1290             long c;
1291
1292             bx0 = span->text[i].bbox.x0;
1293             bx1 = span->text[i].bbox.x1;
1294             by0 = span->text[i].bbox.y0 + oy;
1295             by1 = span->text[i].bbox.y1 + oy;
1296
1297             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1298                 if (!seen) {
1299                     rx0 = bx0 - page->pixmap->x;
1300                     rx1 = bx1 - page->pixmap->x;
1301                     ry0 = by0;
1302                     ry1 = by1;
1303                 }
1304
1305                 seen = 1;
1306                 c = span->text[i].c;
1307                 if (c < 256) {
1308                     if ((isprint (c) && !isspace (c))) {
1309                         if (!rectsel) {
1310                             bx0 -= page->pixmap->x;
1311                             bx1 -= page->pixmap->x;
1312                             glEnable (GL_BLEND);
1313                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1314                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1315                             glColor4f (0.5f, 0.5f, 0.0f, 0.6f);
1316                             glRecti (bx0, by0, bx1, by1);
1317                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1318                             glDisable (GL_BLEND);
1319                         }
1320                         if (isprint (c) || c ==' ') {
1321                             rx1 = bx1;
1322                             ry1 = by1;
1323                         }
1324                     }
1325                     putc (c, stdout);
1326                 }
1327                 else  {
1328                     putc ('?', stdout);
1329                 }
1330             }
1331         }
1332
1333         if (rectsel) {
1334             if (seen) {
1335                 glEnable (GL_BLEND);
1336                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1337                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1338                 glColor4f (0.5f, 0.5f, 0.0f, 0.6f);
1339                 glRecti (rx0, ry0, rx1, ry1);
1340                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1341                 glDisable (GL_BLEND);
1342             }
1343         }
1344
1345         if (seen && span->eol) {
1346             x0 = page->pixmap->x;
1347             putc ('\n', stdout);
1348         }
1349     }
1350     unlock ("ml_gettext");
1351
1352  done:
1353     CAMLreturn (Val_unit);
1354 }
1355
1356 CAMLprim value ml_getpagewh (value pagedimno_v)
1357 {
1358     CAMLparam1 (pagedimno_v);
1359     CAMLlocal1 (ret_v);
1360     int pagedimno = Int_val (pagedimno_v);
1361
1362     ret_v = caml_alloc_small (4 * Double_wosize, Double_array_tag);
1363     Store_double_field (ret_v, 0, state.pagedims[pagedimno].box.x0);
1364     Store_double_field (ret_v, 1, state.pagedims[pagedimno].box.x1);
1365     Store_double_field (ret_v, 2, state.pagedims[pagedimno].box.y0);
1366     Store_double_field (ret_v, 3, state.pagedims[pagedimno].box.y1);
1367     CAMLreturn (ret_v);
1368 }
1369
1370 CAMLprim value ml_init (value sock_v)
1371 {
1372 #ifndef _WIN32
1373     int ret;
1374 #endif
1375     CAMLparam1 (sock_v);
1376
1377     state.texcount = 256;
1378     state.sliceheight = 24;
1379     state.texform = GL_RGBA;
1380     state.texty = GL_UNSIGNED_BYTE;
1381
1382     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1383     if (!state.texids) {
1384         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1385     }
1386
1387     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1388     if (!state.texowners) {
1389         err (1, "calloc texowners %zu",
1390              state.texcount * sizeof (*state.texowners));
1391     }
1392
1393     glGenTextures (state.texcount, state.texids);
1394
1395 #ifdef _WIN32
1396     state.sock = Socket_val (sock_v);
1397 #else
1398     state.sock = Int_val (sock_v);
1399 #endif
1400
1401     state.cache = fz_newglyphcache ();
1402     if (!state.cache) {
1403         errx (1, "fz_newglyphcache failed");
1404     }
1405
1406 #ifdef _WIN32
1407     InitializeCriticalSection (&critsec);
1408     state.thread = CreateThread (NULL, 0, mainloop, NULL, 0, NULL);
1409     if (state.thread == INVALID_HANDLE_VALUE) {
1410         errx (1, "CreateThread failed: %lx", GetLastError ());
1411     }
1412 #else
1413     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1414     if (ret) {
1415         errx (1, "pthread_create: %s", strerror (errno));
1416     }
1417 #endif
1418
1419     CAMLreturn (Val_unit);
1420 }