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 /* fugly as hell and GCC specific but... */
  74 #ifdef _BIG_ENDIAN
  75 #define GL_GLEXT_PROTOTYPES
  76 #endif
  77
  78 #include <GL/gl.h>
  79 #ifndef _WIN32
  80 #include <GL/glext.h>
  81 #else
  82 #define GL_TEXTURE_RECTANGLE_ARB          0x84F5
  83 #define GL_FRAGMENT_SHADER_ATI            0x8920
  84 #define GL_UNSIGNED_INT_8_8_8_8           0x8035
  85 #endif
  86
  87 #include <caml/fail.h>
  88 #include <caml/alloc.h>
  89 #include <caml/memory.h>
  90 #include <caml/unixsupport.h>
  91
  92 #include <fitz.h>
  93 #include <mupdf.h>
  94
  95 #if 0
  96 #define lprintf printf
  97 #else
  98 #define lprintf(...)
  99 #endif
 100
 101 #define ARSERT(cond) for (;;) {                         \
 102     if (!(cond)) {                                      \
 103         errx (1, "%s:%d " #cond, __FILE__, __LINE__);   \
 104     }                                                   \
 105     break;                                              \
 106 }
 107
 108 struct slice {
 109     int texindex;
 110     int w, h;
 111 };
 112
 113 struct page {
 114     int pageno;
 115     int slicecount;
 116     fz_textspan *text;
 117     fz_pixmap *pixmap;
 118     pdf_page *drawpage;
 119     struct pagedim *pagedim;
 120     struct page *prev;
 121     struct slice slices[];
 122 };
 123
 124 struct pagedim {
 125     int pageno;
 126     int rotate;
 127     fz_rect box;
 128     fz_bbox bbox;
 129     fz_matrix ctm;
 130 };
 131
 132 struct {
 133     int sock;
 134     int sliceheight;
 135     struct page *pages;
 136     struct pagedim *pagedims;
 137     int pagecount;
 138     int pagedimcount;
 139     pdf_xref *xref;
 140     fz_glyphcache *cache;
 141     int w, h;
 142
 143     int useatifs;
 144
 145     int texindex;
 146     int texcount;
 147     GLuint *texids;
 148
 149     GLenum texform;
 150     GLenum texty;
 151
 152     int lotsamemory;
 153
 154     int *pagetbl;
 155     struct {
 156         int w, h;
 157         struct slice *slice;
 158     } *texowners;
 159
 160 #ifdef _WIN32
 161     HANDLE thread;
 162 #else
 163     pthread_t thread;
 164 #endif
 165 } state;
 166
 167 #ifdef _WIN32
 168 static CRITICAL_SECTION critsec;
 169
 170 static void lock (void *unused)
 171 {
 172     (void) unused;
 173     EnterCriticalSection (&critsec);
 174 }
 175
 176 static void unlock (void *unused)
 177 {
 178     (void) unused;
 179     LeaveCriticalSection (&critsec);
 180 }
 181
 182 static int trylock (void *unused)
 183 {
 184     return TryEnterCriticalSection (&critsec) == 0;
 185 }
 186 #else
 187 static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
 188
 189 static void lock (const char *cap)
 190 {
 191     int ret = pthread_mutex_lock (&mutex);
 192     if (ret) {
 193         errx (1, "%s: pthread_mutex_lock: %s", cap, strerror (ret));
 194     }
 195 }
 196
 197 static void unlock (const char *cap)
 198 {
 199     int ret = pthread_mutex_unlock (&mutex);
 200     if (ret) {
 201         errx (1, "%s: pthread_mutex_unlock: %s", cap, strerror (ret));
 202     }
 203 }
 204
 205 static int trylock (const char *cap)
 206 {
 207     int ret = pthread_mutex_trylock (&mutex);
 208
 209     if (ret && ret != EBUSY) {
 210         errx (1, "%s: pthread_mutex_trylock: %s", cap, strerror (ret));
 211     }
 212     return ret == EBUSY;
 213 }
 214 #endif
 215
 216 static void *parse_pointer (const char *cap, const char *s)
 217 {
 218     int ret;
 219     void *ptr;
 220
 221     ret = sscanf (s, "%p", &ptr);
 222     if (ret != 1) {
 223         errx (1, "%s: cannot parse pointer in `%s'", cap, s);
 224     }
 225     return ptr;
 226 }
 227
 228 static int hasdata (int sock)
 229 {
 230     int n;
 231     fd_set s;
 232     struct timeval tv;
 233     FD_ZERO (&s);
 234     FD_SET (sock, &s);
 235     tv.tv_sec = 0;
 236     tv.tv_usec = 0;
 237     n = select (sock + 1, &s, NULL, NULL, &tv);
 238     if (n == 0) return 0;
 239     if (n == 1) return 1;
 240     sockerr (1, "hasdata: select error ret=%d", n);
 241 }
 242
 243 static double now (void)
 244 {
 245     struct timeval tv;
 246
 247     if (gettimeofday (&tv, NULL)) {
 248         err (1, "gettimeofday");
 249     }
 250     return tv.tv_sec + tv.tv_usec*1e-6;
 251 }
 252
 253 static void readdata (int fd, char *p, int size)
 254 {
 255     ssize_t n;
 256
 257     n = recv (fd, p, size, 0);
 258     if (n - size) {
 259         if (!n) errx (1, "EOF while reading");
 260         sockerr (1, "recv (req %d, ret %zd)", size, n);
 261     }
 262 }
 263
 264 static void writedata (int fd, char *p, int size)
 265 {
 266     char buf[4];
 267     ssize_t n;
 268
 269     buf[0] = (size >> 24) & 0xff;
 270     buf[1] = (size >> 16) & 0xff;
 271     buf[2] = (size >>  8) & 0xff;
 272     buf[3] = (size >>  0) & 0xff;
 273
 274     n = send (fd, buf, 4, 0);
 275     if (n != 4) {
 276         if (!n) errx (1, "EOF while writing length");
 277         sockerr (1, "send %zd", n);
 278     }
 279
 280     n = send (fd, p, size, 0);
 281     if (n - size) {
 282         if (!n) errx (1, "EOF while writing data");
 283         sockerr (1, "send (req %d, ret %zd)", size, n);
 284     }
 285 }
 286
 287 static void
 288 #ifdef __GNUC__
 289 __attribute__ ((format (printf, 2, 3)))
 290 #endif
 291     printd (int fd, const char *fmt, ...)
 292 {
 293     int len;
 294     va_list ap;
 295     char buf[200];
 296
 297     va_start (ap, fmt);
 298     len = vsnprintf (buf, sizeof (buf), fmt, ap);
 299     va_end (ap);
 300     writedata (fd, buf, len);
 301 }
 302
 303 static void die (fz_error error)
 304 {
 305     fz_catch (error, "aborting");
 306     if (state.xref)
 307        pdf_closexref (state.xref);
 308     exit (1);
 309 }
 310
 311 static void openxref (char *filename)
 312 {
 313     int fd;
 314     fz_error error;
 315     fz_stream *file;
 316
 317     fd = open (filename, O_BINARY | O_RDONLY, 0666);
 318     if (fd < 0)
 319         die (fz_throw ("cannot open file '%s'", filename));
 320
 321     file = fz_openfile (fd);
 322     state.xref = pdf_openxref (file);
 323     if (!state.xref)
 324         die (fz_throw ("cannot open PDF file '%s'", filename));
 325     fz_dropstream (file);
 326
 327     if (pdf_needspassword (state.xref)) {
 328         die (fz_throw ("password protected"));
 329     }
 330
 331     error = pdf_loadpagetree (state.xref);
 332     if (error) {
 333         die (fz_throw ("cannot load page tree"));
 334     }
 335
 336     state.pagecount = pdf_getpagecount (state.xref);
 337     state.pagetbl = stat_alloc (state.pagecount * sizeof (*state.pagetbl));
 338 }
 339
 340 static int readlen (int fd)
 341 {
 342     ssize_t n;
 343     char p[4];
 344
 345     n = recv (fd, p, 4, 0);
 346     if (n != 4) {
 347         if (!n) errx (1, "EOF while reading length");
 348         sockerr (1, "read %zd", n);
 349     }
 350
 351     return (p[0] << 24) | (p[1] << 16) | (p[2] << 8) | p[3];
 352 }
 353
 354 static void freepage (struct page *page)
 355 {
 356     int i;
 357     struct page *p;
 358
 359     fz_droppixmap (page->pixmap);
 360     for (p = state.pages; p; p = p->prev) {
 361         if (p->prev == page) {
 362             p->prev = page->prev;
 363             break;
 364         }
 365     }
 366     for (i = 0; i < page->slicecount; ++i) {
 367         struct slice *s = &page->slices[i];
 368         if (s->texindex != -1) {
 369             if (state.texowners[s->texindex].slice == s) {
 370                 state.texowners[s->texindex].slice = NULL;
 371                 ARSERT (state.texowners[s->texindex].w == s->w);
 372                 ARSERT (state.texowners[s->texindex].h >= s->h);
 373             }
 374         }
 375     }
 376     if (page->text) {
 377         fz_freetextspan (page->text);
 378     }
 379     if (page->drawpage) {
 380         pdf_freepage (page->drawpage);
 381     }
 382
 383     free (page);
 384 }
 385
 386 static void subdivide (struct page *p)
 387 {
 388     int i;
 389     int h = p->pixmap->h;
 390     int th = MIN (h, state.sliceheight);
 391
 392     for (i = 0; i < p->slicecount; ++i) {
 393         struct slice *s = &p->slices[i];
 394         s->texindex = -1;
 395         s->h = MIN (th, h);
 396         s->w = p->pixmap->w;
 397         h -= th;
 398     }
 399 }
 400
 401 static void *render (int pageno, int pindex)
 402 {
 403     fz_error error;
 404     int slicecount;
 405     fz_obj *pageobj;
 406     struct page *page;
 407     double start, end;
 408     pdf_page *drawpage;
 409     fz_device *idev;
 410     struct pagedim *pagedim;
 411
 412     start = now ();
 413     printd (state.sock, "V rendering %d", pageno);
 414     pdf_flushxref (state.xref, 0);
 415
 416     pagedim = &state.pagedims[pindex];
 417     slicecount = (pagedim->bbox.y1 - pagedim->bbox.y0
 418                   + state.sliceheight - 1) / state.sliceheight;
 419     slicecount += slicecount == 0;
 420
 421     page = calloc (sizeof (*page)
 422                    + (slicecount * sizeof (struct slice)), 1);
 423     if (!page) {
 424         err (1, "calloc page %d\n", pageno);
 425     }
 426     page->slicecount = slicecount;
 427     page->prev = state.pages;
 428     state.pages = page;
 429
 430     pageobj = pdf_getpageobject (state.xref, pageno);
 431     if (!pageobj)
 432         die (fz_throw ("cannot retrieve info from page %d", pageno));
 433
 434     error = pdf_loadpage (&drawpage, state.xref, pageobj);
 435     if (error)
 436         die (error);
 437
 438     page->pixmap = fz_newpixmapwithrect (pdf_devicergb, pagedim->bbox);
 439     if (error)
 440         die (error);
 441     fz_clearpixmap (page->pixmap, 0xFF);
 442
 443     idev = fz_newdrawdevice (state.cache, page->pixmap);
 444     if (!idev)
 445         die (fz_throw ("fz_newdrawdevice failed"));
 446     error = pdf_runcontentstream (idev, pagedim->ctm, state.xref,
 447                                   drawpage->resources,
 448                                   drawpage->contents);
 449     fz_freedevice (idev);
 450
 451     page->drawpage = drawpage;
 452     page->pagedim = pagedim;
 453     page->pageno = pageno;
 454     subdivide (page);
 455     end = now ();
 456
 457     if (!state.lotsamemory) {
 458         pdf_agestoreditems (state.xref->store);
 459         pdf_evictageditems (state.xref->store);
 460     }
 461
 462     printd (state.sock, "V rendering %d took %f sec", pageno, end - start);
 463     return page;
 464 }
 465
 466 static void initpdims (void)
 467 {
 468     int pageno;
 469     double start, end;
 470
 471     start = now ();
 472     for (pageno = 0; pageno < state.pagecount; ++pageno) {
 473         int rotate;
 474         fz_rect box;
 475         struct pagedim *p;
 476         fz_obj *obj, *pageobj;
 477
 478         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 479
 480         obj = fz_dictgets (pageobj, "CropBox");
 481         if (!fz_isarray (obj)) {
 482             obj = fz_dictgets (pageobj, "MediaBox");
 483             if (!fz_isarray (obj)) {
 484                 die (fz_throw ("cannot find page bounds %d (%d Rd)",
 485                                fz_tonum (pageobj), fz_togen (pageobj)));
 486             }
 487         }
 488         box = pdf_torect (obj);
 489
 490         obj = fz_dictgets (pageobj, "Rotate");
 491         if (fz_isint (obj)) {
 492             rotate = fz_toint (obj);
 493         }
 494         else  {
 495             rotate = 0;
 496         }
 497
 498         state.pagetbl[pageno] = fz_tonum (state.xref->pagerefs[pageno]);
 499         p = &state.pagedims[state.pagedimcount - 1];
 500         if ((state.pagedimcount == 0)
 501             || (p->rotate != rotate || memcmp (&p->box, &box, sizeof (box)))) {
 502             size_t size;
 503
 504             size = (state.pagedimcount + 1) * sizeof (*state.pagedims);
 505             state.pagedims = realloc (state.pagedims, size);
 506             if (!state.pagedims) {
 507                 err (1, "realloc pagedims to %zu (%d elems)",
 508                      size, state.pagedimcount + 1);
 509             }
 510             p = &state.pagedims[state.pagedimcount++];
 511             p->rotate = rotate;
 512             p->box = box;
 513             p->pageno = pageno;
 514         }
 515     }
 516     end = now ();
 517     printd (state.sock, "T Processed %d pages in %f seconds",
 518             state.pagecount, end - start);
 519 }
 520
 521 static void layout (void)
 522 {
 523     int pindex;
 524     fz_matrix ctm;
 525     fz_rect box, box2;
 526     double zoom, w;
 527     struct pagedim *p = state.pagedims;
 528
 529     pindex = 0;
 530     for (pindex = 0; pindex < state.pagedimcount; ++pindex, ++p) {
 531         box.x0 = MIN (p->box.x0, p->box.x1);
 532         box.y0 = MIN (p->box.y0, p->box.y1);
 533         box.x1 = MAX (p->box.x0, p->box.x1);
 534         box.y1 = MAX (p->box.y0, p->box.y1);
 535
 536         ctm = fz_identity ();
 537         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 538         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 539         box2 = fz_transformrect (ctm, box);
 540         w = box2.x1 - box2.x0;
 541
 542         zoom = (state.w / w);
 543         ctm = fz_identity ();
 544         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 545         ctm = fz_concat (ctm, fz_scale (zoom, -zoom));
 546         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 547         p->bbox = fz_roundrect (fz_transformrect (ctm, box));
 548         memcpy (&p->ctm, &ctm, sizeof (ctm));
 549     }
 550
 551     while (p-- != state.pagedims)  {
 552         printd (state.sock, "l %d %d %d",
 553                 p->pageno, p->bbox.x1 - p->bbox.x0, p->bbox.y1 - p->bbox.y0);
 554     }
 555 }
 556
 557 static void recurse_outline (pdf_outline *outline, int level)
 558 {
 559     while (outline) {
 560         int i, num;
 561         fz_obj *obj;
 562         int top = 0;
 563         int pageno = -1;
 564
 565         if (!outline->link) goto next;
 566
 567         obj = outline->link->dest;
 568         if (fz_isindirect (obj)) {
 569             num = fz_tonum (obj);
 570
 571             for (i = 0; i < state.pagecount; ++i)  {
 572                 if (state.pagetbl[i] == num) {
 573                     pageno = i;
 574                     break;
 575                 }
 576             }
 577         }
 578         else if (fz_isarray (obj)) {
 579             fz_obj *obj2;
 580
 581             obj2 = fz_arrayget (obj, 0);
 582             if (fz_isint (obj2)) {
 583                 pageno = fz_toint (obj2);
 584             }
 585             else {
 586                 num = fz_tonum (obj2);
 587                 for (i = 0; i < state.pagecount; ++i)  {
 588                     if (state.pagetbl[i] == num) {
 589                         pageno = i;
 590                         break;
 591                     }
 592                 }
 593             }
 594
 595             if (fz_arraylen (obj) > 3) {
 596                 fz_point p;
 597                 struct pagedim *pagedim = state.pagedims;
 598
 599                 for (i = 0; i < state.pagedimcount; ++i) {
 600                     if (state.pagedims[i].pageno > pageno)
 601                         break;
 602                     pagedim = &state.pagedims[i];
 603                 }
 604
 605                 p.x = fz_toint (fz_arrayget (obj, 2));
 606                 p.y = fz_toint (fz_arrayget (obj, 3));
 607                 p = fz_transformpoint (pagedim->ctm, p);
 608                 top = p.y;
 609             }
 610         }
 611
 612         lprintf ("%*c%s %d\n", level, ' ', outline->title, pageno);
 613         printd (state.sock, "o %d %d %d %s",
 614                 level, pageno, top, outline->title);
 615     next:
 616         if (outline->child) {
 617             recurse_outline (outline->child, level + 1);
 618         }
 619         outline = outline->next;
 620     }
 621 }
 622
 623 static void process_outline (void)
 624 {
 625     pdf_outline *outline;
 626
 627     outline = pdf_loadoutline (state.xref);
 628     if (outline) {
 629         recurse_outline (outline, 0);
 630         pdf_freeoutline (outline);
 631     }
 632 }
 633
 634 static int comparespans (const void *l, const void *r)
 635 {
 636     fz_textspan const *const*ls = l;
 637     fz_textspan const *const*rs = r;
 638     return (*ls)->text->bbox.y0 - (*rs)->text->bbox.y0;
 639 }
 640
 641 /* wishful thinking function */
 642 static void search (regex_t *re, int pageno, int y, int forward)
 643 {
 644     int i, j;
 645     int ret;
 646     char *p;
 647     char buf[256];
 648     fz_error error;
 649     fz_obj *pageobj;
 650     fz_device *tdev;
 651     pdf_page *drawpage;
 652     fz_textspan *text, *span, **pspan;
 653     struct pagedim *pdim, *pdimprev;
 654     int stop = 0;
 655     int niters = 0;
 656     int nspans;
 657     double start, end;
 658
 659     start = now ();
 660     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 661         if (niters++ == 5) {
 662             if (!state.lotsamemory) {
 663                 pdf_agestoreditems (state.xref->store);
 664                 pdf_evictageditems (state.xref->store);
 665             }
 666             niters = 0;
 667             if (hasdata (state.sock)) {
 668                 printd (state.sock, "T attention requested aborting search at %d",
 669                         pageno);
 670                 stop = 1;
 671             }
 672             else {
 673                 printd (state.sock, "T searching in page %d", pageno);
 674             }
 675         }
 676         pdimprev = NULL;
 677         for (i = 0; i < state.pagedimcount; ++i) {
 678             pdim = &state.pagedims[i];
 679             if (pdim->pageno == pageno)  {
 680                 goto found;
 681             }
 682             if (pdim->pageno > pageno) {
 683                 pdim = pdimprev;
 684                 goto found;
 685             }
 686             pdimprev = pdim;
 687         }
 688         pdim = pdimprev;
 689     found:
 690
 691         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 692         if (!pageobj)
 693             die (fz_throw ("cannot retrieve info from page %d", pageno));
 694
 695         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 696         if (error)
 697             die (error);
 698
 699         text = fz_newtextspan ();
 700         tdev = fz_newtextdevice (text);
 701         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 702                                       drawpage->resources,
 703                                       drawpage->contents);
 704         if (error) die (error);
 705         fz_freedevice (tdev);
 706
 707         nspans = 0;
 708         for (span = text; span; span = span->next) {
 709             nspans++;
 710         }
 711         pspan = malloc (sizeof (void *) * nspans);
 712         if (!pspan) {
 713             err (1, "malloc span pointers %zu", sizeof (void *) * nspans);
 714         }
 715         for (i = 0, span = text; span; span = span->next, ++i) {
 716             pspan[i] = span;
 717         }
 718         qsort (pspan, nspans, sizeof (fz_textspan *), comparespans);
 719
 720         j = forward ? 0 : nspans - 1;
 721         while (nspans--) {
 722             regmatch_t rm;
 723
 724             span = pspan[j];
 725             j += forward ? 1 : -1;
 726             p = buf;
 727             for (i = 0; i < MIN (span->len, (int) sizeof (buf) - 1); ++i) {
 728                 if (forward) {
 729                     if (span->text[i].bbox.y0 < y + 1) {
 730                         continue;
 731                     }
 732                 }
 733                 else {
 734                     if (span->text[i].bbox.y0 > y - 1) {
 735                         continue;
 736                     }
 737                 }
 738                 if (span->text[i].c < 256) {
 739                     *p++ = span->text[i].c;
 740                 }
 741                 else  {
 742                     *p++ = '?';
 743                 }
 744             }
 745             if (p == buf) {
 746                 continue;
 747             }
 748             *p++ = 0;
 749
 750             ret = regexec (re, buf, 1, &rm, 0);
 751             if (ret)  {
 752                 if (ret != REG_NOMATCH) {
 753                     size_t size;
 754                     char errbuf[80];
 755                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 756                     printd (state.sock,
 757                             "T regexec error `%.*s'",
 758                             (int) size, errbuf);
 759                     fz_freetextspan (text);
 760                     pdf_freepage (drawpage);
 761                     free (pspan);
 762                     return;
 763                 }
 764             }
 765             else  {
 766                 fz_rect r;
 767
 768                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 769                 r.y0 = span->text[rm.rm_so].bbox.y0;
 770                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 771                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 772
 773                 if (!stop) {
 774                     printd (state.sock, "F %d %d %f %f %f %f",
 775                             pageno, 1,
 776                             r.x0, r.y0,
 777                             r.x1, r.y1);
 778                 }
 779                 else  {
 780                     printd (state.sock, "R %d %d %f %f %f %f",
 781                             pageno, 2,
 782                             r.x0, r.y0,
 783                             r.x1, r.y1);
 784                 }
 785                 printd (state.sock, "T found at %d `%.*s' %f in %f sec",
 786                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 787                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 788                         now () - start);
 789                 stop = 1;
 790             }
 791         }
 792         if (forward) {
 793             pageno += 1;
 794             y = 0;
 795         }
 796         else {
 797             pageno -= 1;
 798             y = INT_MAX;
 799         }
 800         fz_freetextspan (text);
 801         pdf_freepage (drawpage);
 802         free (pspan);
 803     }
 804     end = now ();
 805     if (!stop)  {
 806         printd (state.sock, "T no matches %f sec", end - start);
 807     }
 808     printd (state.sock, "D");
 809 }
 810
 811 static
 812 #ifdef _WIN32
 813 DWORD _stdcall
 814 #else
 815 void *
 816 #endif
 817 mainloop (void *unused)
 818 {
 819     char *p = NULL;
 820     int len, ret, oldlen = 0;
 821
 822     for (;;) {
 823         len = readlen (state.sock);
 824         if (len == 0) {
 825             errx (1, "readlen returned 0");
 826         }
 827
 828         if (oldlen < len + 1) {
 829             p = realloc (p, len + 1);
 830             if (!p) {
 831                 err (1, "realloc %d failed", len + 1);
 832             }
 833             oldlen = len + 1;
 834         }
 835         readdata (state.sock, p, len);
 836         p[len] = 0;
 837
 838         if (!strncmp ("open", p, 4)) {
 839             fz_obj *obj;
 840             char *filename = p + 5;
 841
 842             openxref (filename);
 843             initpdims ();
 844
 845             obj = fz_dictgets (state.xref->trailer, "Info");
 846             if (obj) {
 847                 obj = fz_dictgets (obj, "Title");
 848                 if (obj) {
 849                     printd (state.sock, "t %s", pdf_toutf8 (obj));
 850                 }
 851             }
 852         }
 853         else if (!strncmp ("free", p, 4)) {
 854             void *ptr;
 855
 856             ret = sscanf (p + 4, " %p", &ptr);
 857             if (ret != 1) {
 858                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 859             }
 860             lock ("free");
 861             freepage (ptr);
 862             unlock ("free");
 863             printd (state.sock, "d");
 864         }
 865         else if (!strncmp ("search", p, 6)) {
 866             int icase, pageno, y, ret, len2, forward;
 867             char *pattern;
 868             regex_t re;
 869
 870             ret = sscanf (p + 6, " %d %d %d %d,%n",
 871                           &icase, &pageno, &y, &forward, &len2);
 872             if (ret != 4) {
 873                 errx (1, "malformed search `%s' ret=%d", p, ret);
 874             }
 875
 876             pattern = p + 6 + len2;
 877             ret = regcomp (&re, pattern,
 878                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 879             if (ret) {
 880                 char errbuf[80];
 881                 size_t size;
 882
 883                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 884                 printd (state.sock, "T regcomp failed `%.*s'",
 885                         (int) size, errbuf);
 886             }
 887             else  {
 888                 search (&re, pageno, y, forward);
 889                 regfree (&re);
 890             }
 891         }
 892         else if (!strncmp ("geometry", p, 8)) {
 893             int w, h;
 894
 895             printd (state.sock, "c");
 896             ret = sscanf (p + 8, " %d %d", &w, &h);
 897             if (ret != 2) {
 898                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 899             }
 900             state.h = h;
 901             if (w != state.w) {
 902                 int i;
 903                 state.w = w;
 904                 for (i = 0; i < state.texcount; ++i)  {
 905                     state.texowners[i].slice = NULL;
 906                 }
 907             }
 908             lock ("geometry");
 909             layout ();
 910             process_outline ();
 911             unlock ("geometry");
 912             printd (state.sock, "C %d", state.pagecount);
 913         }
 914         else if (!strncmp ("render", p, 6)) {
 915             int pageno, pindex, w, h, ret;
 916             struct page *page;
 917
 918             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 919             if (ret != 4) {
 920                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 921             }
 922
 923             page = render (pageno, pindex);
 924             printd (state.sock, "r %d %d %d %p\n",
 925                     pageno,
 926                     state.w,
 927                     state.h,
 928                     page);
 929         }
 930         else {
 931             errx (1, "unknown command %.*s", len, p);
 932         }
 933     }
 934     return 0;
 935 }
 936
 937 static void upload2 (struct page *page, int slicenum, const char *cap)
 938 {
 939     int i;
 940     int w, h;
 941     double start, end;
 942     struct slice *slice = &page->slices[slicenum];
 943
 944     w = page->pixmap->w;
 945     h = page->pixmap->h;
 946
 947     ARSERT (w == slice->w);
 948     if (slice->texindex != -1
 949         && state.texowners[slice->texindex].slice == slice) {
 950         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 951     }
 952     else  {
 953         int subimage = 0;
 954         int index = (state.texindex++ % state.texcount);
 955         size_t offset = 0;
 956
 957         for (i = 0; i < slicenum; ++i) {
 958             offset += w * page->slices[i].h * 4;
 959         }
 960
 961         if (state.texowners[index].w == slice->w) {
 962             if (state.texowners[index].h >= slice->h ) {
 963                 subimage = 1;
 964             }
 965             else {
 966                 state.texowners[index].h = slice->h;
 967             }
 968         }
 969         else  {
 970             state.texowners[index].h = slice->h;
 971         }
 972
 973         state.texowners[index].slice = slice;
 974         state.texowners[index].w = slice->w;
 975         slice->texindex = index;
 976
 977         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 978         start = now ();
 979         if (subimage) {
 980             {
 981                 GLenum err = glGetError ();
 982                 if (err != GL_NO_ERROR) {
 983                     printf ("\033[0;31mERROR1 %d %d %#x\033[0m\n", w, slice->h, err);
 984                     abort ();
 985                 }
 986             }
 987             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
 988                              0,
 989                              0,
 990                              0,
 991                              w,
 992                              slice->h,
 993                              state.texform,
 994                              state.texty,
 995                              page->pixmap->samples + offset
 996                 );
 997             {
 998                 GLenum err = glGetError ();
 999                 if (err != GL_NO_ERROR) {
1000                     printf ("\033[0;31mERROR %d %d %#x\033[0m\n", w, slice->h, err);
1001                     abort ();
1002                 }
1003             }
1004         }
1005         else {
1006             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
1007                           0,
1008                           GL_RGBA8,
1009                           w,
1010                           slice->h,
1011                           0,
1012                           state.texform,
1013                           state.texty,
1014                           page->pixmap->samples + offset
1015                 );
1016         }
1017
1018         end = now ();
1019         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
1020                  subimage ? "sub" : "img",
1021                  page->pageno, slicenum,
1022                  slice->w, slice->h,
1023                  state.texids[slice->texindex],
1024                  end - start);
1025     }
1026 }
1027
1028 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
1029                         value py_v, value ptr_v)
1030 {
1031     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
1032     int dispy = Int_val (dispy_v);
1033     int w = Int_val (w_v);
1034     int h = Int_val (h_v);
1035     int py = Int_val (py_v);
1036     char *s = String_val (ptr_v);
1037     int ret;
1038     void *ptr;
1039     struct page *page;
1040     int slicenum = 0;
1041
1042     if (trylock ("ml_draw")) {
1043         goto done;
1044     }
1045
1046     ret = sscanf (s, "%p", &ptr);
1047     if (ret != 1) {
1048         errx (1, "cannot parse pointer `%s'", s);
1049     }
1050     page = ptr;
1051
1052     w = page->pixmap->w;
1053
1054     ARSERT (h >= 0 && "ml_draw wrong h");
1055
1056     glEnable (GL_TEXTURE_RECTANGLE_ARB);
1057     if (state.useatifs) {
1058         glEnable (GL_FRAGMENT_SHADER_ATI);
1059     }
1060
1061     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
1062         struct slice *slice = &page->slices[slicenum];
1063         if (slice->h > py) {
1064             break;
1065         }
1066         py -= slice->h;
1067     }
1068
1069     h = MIN (state.h, h);
1070     while (h) {
1071         int th;
1072         struct slice *slice = &page->slices[slicenum];
1073
1074         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
1075
1076         th = MIN (h, slice->h - py);
1077         upload2 (page, slicenum, "upload");
1078
1079         glBegin (GL_QUADS);
1080         {
1081             glTexCoord2i (0, py);
1082             glVertex2i (0, dispy);
1083
1084             glTexCoord2i (w, py);
1085             glVertex2i (w, dispy);
1086
1087             glTexCoord2i (w, py+th);
1088             glVertex2i (w, dispy + th);
1089
1090             glTexCoord2i (0, py+th);
1091             glVertex2i (0, dispy + th);
1092         }
1093         glEnd ();
1094
1095         h -= th;
1096         py = 0;
1097         dispy += th;
1098         slicenum += 1;
1099     }
1100
1101     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1102     if (state.useatifs) {
1103         glDisable (GL_FRAGMENT_SHADER_ATI);
1104     }
1105
1106     unlock ("ml_draw");
1107  done:
1108     CAMLreturn (Val_unit);
1109 }
1110
1111 static pdf_link *getlink (struct page *page, int x, int y)
1112 {
1113     fz_point p;
1114     fz_matrix ctm;
1115     pdf_link *link;
1116
1117     p.x = x;
1118     p.y = y;
1119
1120     ctm = fz_invertmatrix (page->pagedim->ctm);
1121     p = fz_transformpoint (ctm, p);
1122
1123     for (link = page->drawpage->links; link; link = link->next) {
1124         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1125             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1126                 if (link->kind == PDF_LGOTO) {
1127                     return link;
1128                 }
1129             }
1130         }
1131     }
1132     return NULL;
1133 }
1134
1135 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1136 {
1137     CAMLparam3 (ptr_v, x_v, y_v);
1138     char *s = String_val (ptr_v);
1139     int ret;
1140
1141     if (trylock ("ml_checklink")) {
1142         ret = 0;
1143     }
1144     else {
1145         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1146                                Int_val (x_v), Int_val (y_v));
1147         unlock ("ml_checklink");
1148     }
1149     CAMLreturn (Val_bool (ret));
1150 }
1151
1152 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1153 {
1154     CAMLparam3 (ptr_v, x_v, y_v);
1155     CAMLlocal2 (ret_v, tup_v);
1156     pdf_link *link;
1157     struct page *page;
1158     char *s = String_val (ptr_v);
1159
1160     if (trylock ("ml_getlink")) {
1161         ret_v = Val_int (0);
1162         goto done;
1163     }
1164
1165     page = parse_pointer ("ml_getlink", s);
1166
1167     link = getlink (page, Int_val (x_v), Int_val (y_v));
1168     if (link) {
1169         int pageno;
1170         fz_point p;
1171         fz_obj *obj;
1172
1173         pageno = -1;
1174         obj = fz_arrayget (link->dest, 0);
1175         if (fz_isindirect (obj)) {
1176             pageno = pdf_findpageobject (state.xref, obj) - 1;
1177         }
1178         else if (fz_isint (obj)) {
1179             pageno = fz_toint (obj);
1180         }
1181
1182         if (fz_arraylen (link->dest) > 3) {
1183             p.x = fz_toint (fz_arrayget (link->dest, 2));
1184             p.y = fz_toint (fz_arrayget (link->dest, 3));
1185             p = fz_transformpoint (page->pagedim->ctm, p);
1186         }
1187         else  {
1188             p.x = 0.0;
1189             p.y = 0.0;
1190         }
1191
1192         tup_v = caml_alloc_tuple (2);
1193         ret_v = caml_alloc_small (1, 1);
1194         Field (tup_v, 0) = Val_int (pageno);
1195         Field (tup_v, 1) = Val_int (p.y);
1196         Field (ret_v, 0) = tup_v;
1197     }
1198     else {
1199         ret_v = Val_int (0);
1200     }
1201     unlock ("ml_getlink");
1202
1203  done:
1204     CAMLreturn (ret_v);
1205 }
1206
1207 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1208 {
1209     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1210     fz_matrix ctm;
1211     fz_point p1, p2;
1212     struct page *page;
1213     fz_textspan *span;
1214     char *s = String_val (ptr_v);
1215     int rectsel = Bool_val (rectsel_v);
1216     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1217
1218     /* stop GCC from complaining about uninitialized variables */
1219 #ifdef __GNUC__
1220     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1221 #else
1222     int rx0, rx1, ry0, ry1;
1223 #endif
1224
1225     if (trylock ("ml_gettext")) {
1226         goto done;
1227     }
1228
1229     page = parse_pointer ("ml_gettext", s);
1230
1231     oy = Int_val (oy_v);
1232     p1.x = Int_val (Field (rect_v, 0));
1233     p1.y = Int_val (Field (rect_v, 1));
1234     p2.x = Int_val (Field (rect_v, 2));
1235     p2.y = Int_val (Field (rect_v, 3));
1236
1237     if (0) {
1238         glEnable (GL_BLEND);
1239         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1240         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1241         glColor4f (0.0f, 0.0f, 0.0f, 0.2f);
1242         glRecti (p1.x, p1.y, p2.x, p2.y);
1243         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1244         glDisable (GL_BLEND);
1245     }
1246
1247     ctm = page->pagedim->ctm;
1248     if (!page->text) {
1249         fz_error error;
1250         fz_device *tdev;
1251
1252         page->text = fz_newtextspan ();
1253         tdev = fz_newtextdevice (page->text);
1254         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1255                                       page->drawpage->resources,
1256                                       page->drawpage->contents);
1257         if (error) die (error);
1258         fz_freedevice (tdev);
1259     }
1260
1261     printf ("\033c");
1262
1263     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1264     p1.x += page->pixmap->x;
1265     p1.y += page->pixmap->y;
1266     p2.x += page->pixmap->x;
1267     p2.y += page->pixmap->y;
1268     x0 = p1.x;
1269     y0 = p1.y;
1270     x1 = p2.x;
1271     y1 = p2.y;
1272     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1273
1274     for (span = page->text; span; span = span->next) {
1275         int seen = 0;
1276
1277         /* fz_debugtextspanxml (span); */
1278         for (i = 0; i < span->len; ++i) {
1279             long c;
1280
1281             bx0 = span->text[i].bbox.x0;
1282             bx1 = span->text[i].bbox.x1;
1283             by0 = span->text[i].bbox.y0 + oy;
1284             by1 = span->text[i].bbox.y1 + oy;
1285
1286             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1287                 if (!seen) {
1288                     rx0 = bx0 - page->pixmap->x;
1289                     rx1 = bx1 - page->pixmap->x;
1290                     ry0 = by0;
1291                     ry1 = by1;
1292                 }
1293
1294                 seen = 1;
1295                 c = span->text[i].c;
1296                 if (c < 256) {
1297                     if ((isprint (c) && !isspace (c))) {
1298                         if (!rectsel) {
1299                             bx0 -= page->pixmap->x;
1300                             bx1 -= page->pixmap->x;
1301                             glEnable (GL_BLEND);
1302                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1303                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1304                             glColor4f (0.5f, 0.5f, 0.0f, 0.6f);
1305                             glRecti (bx0, by0, bx1, by1);
1306                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1307                             glDisable (GL_BLEND);
1308                         }
1309                         if (isprint (c) || c ==' ') {
1310                             rx1 = bx1;
1311                             ry1 = by1;
1312                         }
1313                     }
1314                     putc (c, stdout);
1315                 }
1316                 else  {
1317                     putc ('?', stdout);
1318                 }
1319             }
1320         }
1321
1322         if (rectsel) {
1323             if (seen) {
1324                 glEnable (GL_BLEND);
1325                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1326                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1327                 glColor4f (0.5f, 0.5f, 0.0f, 0.6f);
1328                 glRecti (rx0, ry0, rx1, ry1);
1329                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1330                 glDisable (GL_BLEND);
1331             }
1332         }
1333
1334         if (seen && span->eol) {
1335             x0 = page->pixmap->x;
1336             putc ('\n', stdout);
1337         }
1338     }
1339     unlock ("ml_gettext");
1340
1341  done:
1342     CAMLreturn (Val_unit);
1343 }
1344
1345 CAMLprim value ml_getpagewh (value pagedimno_v)
1346 {
1347     CAMLparam1 (pagedimno_v);
1348     CAMLlocal1 (ret_v);
1349     int pagedimno = Int_val (pagedimno_v);
1350
1351     ret_v = caml_alloc_small (4 * Double_wosize, Double_array_tag);
1352     Store_double_field (ret_v, 0, state.pagedims[pagedimno].box.x0);
1353     Store_double_field (ret_v, 1, state.pagedims[pagedimno].box.x1);
1354     Store_double_field (ret_v, 2, state.pagedims[pagedimno].box.y0);
1355     Store_double_field (ret_v, 3, state.pagedims[pagedimno].box.y1);
1356     CAMLreturn (ret_v);
1357 }
1358
1359 static void initgl (void)
1360 {
1361 #ifdef _BIG_ENDIAN
1362     if (strstr ((char *) glGetString (GL_EXTENSIONS),
1363                 "GL_ATI_fragment_shader")) {
1364         /* Here, with MESA, rv280, powerpc32: BGRA(rev) is slow while
1365            ABGR is fast, so fix things in the shader */
1366         state.texform = GL_ABGR_EXT;
1367         state.texty = GL_UNSIGNED_INT_8_8_8_8;
1368
1369         glBindFragmentShaderATI (1);
1370         glBeginFragmentShaderATI ();
1371         {
1372             glSampleMapATI (GL_REG_0_ATI, GL_TEXTURE0_ARB, GL_SWIZZLE_STR_ATI);
1373
1374             glColorFragmentOp1ATI (GL_MOV_ATI,
1375                                    GL_REG_1_ATI, GL_RED_BIT_ATI, GL_NONE,
1376                                    GL_REG_0_ATI, GL_BLUE, GL_NONE);
1377             glColorFragmentOp1ATI (GL_MOV_ATI,
1378                                    GL_REG_1_ATI, GL_BLUE_BIT_ATI, GL_NONE,
1379                                    GL_REG_0_ATI, GL_RED, GL_NONE);
1380             glColorFragmentOp1ATI (
1381                 GL_MOV_ATI,
1382                 GL_REG_0_ATI, GL_RED_BIT_ATI | GL_BLUE_BIT_ATI, GL_NONE,
1383                 GL_REG_1_ATI, GL_NONE, GL_NONE
1384                 );
1385         }
1386         glEndFragmentShaderATI ();
1387         state.useatifs = 1;
1388     }
1389     else {
1390         state.texform = GL_BGRA_EXT;
1391         state.texty = GL_UNSIGNED_INT_8_8_8_8_REV;
1392     }
1393 #else
1394     state.texform = GL_BGRA_EXT;
1395     state.texty = GL_UNSIGNED_INT_8_8_8_8;
1396 #endif
1397 }
1398
1399 CAMLprim value ml_init (value sock_v)
1400 {
1401 #ifndef _WIN32
1402     int ret;
1403 #endif
1404     CAMLparam1 (sock_v);
1405
1406     state.texcount = 256;
1407     state.sliceheight = 24;
1408
1409     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1410     if (!state.texids) {
1411         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1412     }
1413
1414     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1415     if (!state.texowners) {
1416         err (1, "calloc texowners %zu",
1417              state.texcount * sizeof (*state.texowners));
1418     }
1419
1420     glGenTextures (state.texcount, state.texids);
1421
1422 #ifdef _WIN32
1423     state.sock = Socket_val (sock_v);
1424 #else
1425     state.sock = Int_val (sock_v);
1426 #endif
1427     initgl ();
1428
1429     state.cache = fz_newglyphcache ();
1430     if (!state.cache) {
1431         errx (1, "fz_newglyphcache failed");
1432     }
1433
1434 #ifdef _WIN32
1435     InitializeCriticalSection (&critsec);
1436     state.thread = CreateThread (NULL, 0, mainloop, NULL, 0, NULL);
1437     if (state.thread == INVALID_HANDLE_VALUE) {
1438         errx (1, "CreateThread failed: %lx", GetLastError ());
1439     }
1440 #else
1441     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1442     if (ret) {
1443         errx (1, "pthread_create: %s", strerror (errno));
1444     }
1445 #endif
1446
1447     CAMLreturn (Val_unit);
1448 }