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