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_displaylist *list;
 316     fz_device *idev, *mdev;
 317     struct pagedim *pagedim;
 318
 319     start = now ();
 320     /* printd (state.sock, "T rendering %d", pageno); */
 321     pdf_flushxref (state.xref, 0);
 322
 323     pagedim = &state.pagedims[pindex];
 324     slicecount = (pagedim->bbox.y1 - pagedim->bbox.y0
 325                   + state.sliceheight - 1) / state.sliceheight;
 326     slicecount += slicecount == 0;
 327
 328     page = calloc (sizeof (*page)
 329                    + (slicecount * sizeof (struct slice)), 1);
 330     if (!page) {
 331         err (1, "calloc page %d\n", pageno);
 332     }
 333     page->slicecount = slicecount;
 334     page->prev = state.pages;
 335     state.pages = page;
 336
 337     pageobj = pdf_getpageobject (state.xref, pageno);
 338     if (!pageobj)
 339         die (fz_throw ("cannot retrieve info from page %d", pageno));
 340
 341     error = pdf_loadpage (&drawpage, state.xref, pageobj);
 342     if (error)
 343         die (error);
 344
 345     page->pixmap = fz_newpixmapwithrect (pdf_devicergb, pagedim->bbox);
 346     if (error)
 347         die (error);
 348     fz_clearpixmap (page->pixmap, 0xFF);
 349
 350     list = fz_newdisplaylist ();
 351     if (!list)
 352         die (fz_throw ("fz_newdisplaylist failed"));
 353
 354     mdev = fz_newlistdevice (list);
 355     error = pdf_runcontentstream (mdev, fz_identity (), state.xref,
 356                                   drawpage->resources,
 357                                   drawpage->contents);
 358     if (error)
 359         die (error);
 360     fz_freedevice (mdev);
 361
 362     idev = fz_newdrawdevice (state.cache, page->pixmap);
 363     if (!idev)
 364         die (fz_throw ("fz_newdrawdevice failed"));
 365     fz_executedisplaylist (list, idev, pagedim->ctm);
 366     fz_freedevice (idev);
 367
 368     fz_freedisplaylist (list);
 369
 370     page->drawpage = drawpage;
 371     page->pagedim = pagedim;
 372     page->pageno = pageno;
 373     subdivide (page);
 374     end = now ();
 375
 376     if (!state.lotsamemory) {
 377         pdf_agestoreditems (state.xref->store);
 378         pdf_evictageditems (state.xref->store);
 379     }
 380
 381     /* printd (state.sock, "T rendering %d took %f sec", pageno, end - start); */
 382     return page;
 383 }
 384
 385 static void initpdims (void)
 386 {
 387     int pageno;
 388     double start, end;
 389
 390     start = now ();
 391     for (pageno = 0; pageno < state.pagecount; ++pageno) {
 392         int rotate;
 393         fz_rect box;
 394         struct pagedim *p;
 395         fz_obj *obj, *pageobj;
 396
 397         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 398
 399         obj = fz_dictgets (pageobj, "CropBox");
 400         if (!fz_isarray (obj)) {
 401             obj = fz_dictgets (pageobj, "MediaBox");
 402             if (!fz_isarray (obj)) {
 403                 die (fz_throw ("cannot find page bounds %d (%d Rd)",
 404                                fz_tonum (pageobj), fz_togen (pageobj)));
 405             }
 406         }
 407         box = pdf_torect (obj);
 408
 409         obj = fz_dictgets (pageobj, "Rotate");
 410         if (fz_isint (obj)) {
 411             rotate = fz_toint (obj);
 412         }
 413         else  {
 414             rotate = 0;
 415         }
 416
 417         p = &state.pagedims[state.pagedimcount - 1];
 418         if ((state.pagedimcount == 0)
 419             || (p->rotate != rotate || memcmp (&p->box, &box, sizeof (box)))) {
 420             size_t size;
 421
 422             size = (state.pagedimcount + 1) * sizeof (*state.pagedims);
 423             state.pagedims = realloc (state.pagedims, size);
 424             if (!state.pagedims) {
 425                 err (1, "realloc pagedims to %zu (%d elems)",
 426                      size, state.pagedimcount + 1);
 427             }
 428             p = &state.pagedims[state.pagedimcount++];
 429             p->rotate = rotate;
 430             p->box = box;
 431             p->pageno = pageno;
 432         }
 433     }
 434     end = now ();
 435     printd (state.sock, "T Processed %d pages in %f seconds",
 436             state.pagecount, end - start);
 437 }
 438
 439 static void layout (void)
 440 {
 441     int pindex;
 442     fz_matrix ctm;
 443     fz_rect box, box2;
 444     double zoom, w;
 445     struct pagedim *p = state.pagedims;
 446
 447     pindex = 0;
 448     for (pindex = 0; pindex < state.pagedimcount; ++pindex, ++p) {
 449         box.x0 = MIN (p->box.x0, p->box.x1);
 450         box.y0 = MIN (p->box.y0, p->box.y1);
 451         box.x1 = MAX (p->box.x0, p->box.x1);
 452         box.y1 = MAX (p->box.y0, p->box.y1);
 453
 454         ctm = fz_identity ();
 455         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 456         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 457         box2 = fz_transformrect (ctm, box);
 458         w = box2.x1 - box2.x0;
 459
 460         zoom = (state.w / w);
 461         ctm = fz_identity ();
 462         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 463         ctm = fz_concat (ctm, fz_scale (zoom, -zoom));
 464         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 465         p->bbox = fz_roundrect (fz_transformrect (ctm, box));
 466         memcpy (&p->ctm, &ctm, sizeof (ctm));
 467     }
 468
 469     while (p-- != state.pagedims)  {
 470         printd (state.sock, "l %d %d %d",
 471                 p->pageno, p->bbox.x1 - p->bbox.x0, p->bbox.y1 - p->bbox.y0);
 472     }
 473 }
 474
 475 static void recurse_outline (pdf_outline *outline, int level)
 476 {
 477     while (outline) {
 478         int i, num;
 479         fz_obj *obj;
 480         int top = 0;
 481         int pageno = -1;
 482
 483         if (!outline->link) goto next;
 484
 485         obj = outline->link->dest;
 486         if (fz_isindirect (obj)) {
 487             num = fz_tonum (obj);
 488
 489             for (i = 0; i < state.pagecount; ++i)  {
 490                 if (state.pagetbl[i] == num) {
 491                     pageno = i;
 492                     break;
 493                 }
 494             }
 495         }
 496         else if (fz_isarray (obj)) {
 497             fz_obj *obj2;
 498
 499             obj2 = fz_arrayget (obj, 0);
 500             if (fz_isint (obj2)) {
 501                 pageno = fz_toint (obj2);
 502             }
 503             else {
 504                 num = fz_tonum (obj2);
 505                 for (i = 0; i < state.pagecount; ++i)  {
 506                     if (state.pagetbl[i] == num) {
 507                         pageno = i;
 508                         break;
 509                     }
 510                 }
 511             }
 512
 513             if (fz_arraylen (obj) > 3) {
 514                 fz_point p;
 515                 struct pagedim *pagedim = state.pagedims;
 516
 517                 for (i = 0; i < state.pagedimcount; ++i) {
 518                     if (state.pagedims[i].pageno > pageno)
 519                         break;
 520                     pagedim = &state.pagedims[i];
 521                 }
 522
 523                 p.x = fz_toint (fz_arrayget (obj, 2));
 524                 p.y = fz_toint (fz_arrayget (obj, 3));
 525                 p = fz_transformpoint (pagedim->ctm, p);
 526                 top = p.y;
 527             }
 528         }
 529
 530         lprintf ("%*c%s %d\n", level, ' ', outline->title, pageno);
 531         printd (state.sock, "o %d %d %d %s",
 532                 level, pageno, top, outline->title);
 533     next:
 534         if (outline->child) {
 535             recurse_outline (outline->child, level + 1);
 536         }
 537         outline = outline->next;
 538     }
 539 }
 540
 541 static void process_outline (void)
 542 {
 543     pdf_outline *outline;
 544
 545     outline = pdf_loadoutline (state.xref);
 546     if (outline) {
 547         recurse_outline (outline, 0);
 548         pdf_freeoutline (outline);
 549     }
 550 }
 551
 552 static int comparespans (const void *l, const void *r)
 553 {
 554     fz_textspan *const*ls = l;
 555     fz_textspan *const*rs = r;
 556
 557     return (*ls)->text->bbox.y0 - (*rs)->text->bbox.y0;
 558 }
 559
 560 /* wishful thinking function */
 561 static void search (regex_t *re, int pageno, int y, int forward)
 562 {
 563     int i, j;
 564     int ret;
 565     char *p;
 566     char buf[256];
 567     fz_error error;
 568     fz_obj *pageobj;
 569     fz_device *tdev;
 570     pdf_page *drawpage;
 571     fz_textspan *text, *span, **pspan;
 572     struct pagedim *pdim, *pdimprev;
 573     int stop = 0;
 574     int niters = 0;
 575     int nspans;
 576     double start, end;
 577
 578     start = now ();
 579     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 580         if (niters++ == 5) {
 581             if (!state.lotsamemory) {
 582                 pdf_agestoreditems (state.xref->store);
 583                 pdf_evictageditems (state.xref->store);
 584             }
 585             niters = 0;
 586             if (hasdata (state.sock)) {
 587                 printd (state.sock, "T attention requested aborting search at %d",
 588                         pageno);
 589                 stop = 1;
 590             }
 591             else {
 592                 printd (state.sock, "T searching in page %d", pageno);
 593             }
 594         }
 595         pdimprev = NULL;
 596         for (i = 0; i < state.pagedimcount; ++i) {
 597             pdim = &state.pagedims[i];
 598             if (pdim->pageno == pageno)  {
 599                 goto found;
 600             }
 601             if (pdim->pageno > pageno) {
 602                 pdim = pdimprev;
 603                 goto found;
 604             }
 605             pdimprev = pdim;
 606         }
 607         pdim = pdimprev;
 608     found:
 609
 610         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 611         if (!pageobj)
 612             die (fz_throw ("cannot retrieve info from page %d", pageno));
 613
 614         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 615         if (error)
 616             die (error);
 617
 618         text = fz_newtextspan ();
 619         tdev = fz_newtextdevice (text);
 620         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 621                                       drawpage->resources,
 622                                       drawpage->contents);
 623         if (error) die (error);
 624         fz_freedevice (tdev);
 625
 626         nspans = 0;
 627         for (span = text; span; span = span->next) {
 628             nspans++;
 629         }
 630         pspan = malloc (sizeof (void *) * nspans);
 631         if (!pspan) {
 632             err (1, "malloc span pointers %zu", sizeof (void *) * nspans);
 633         }
 634         for (i = 0, span = text; span; span = span->next, ++i) {
 635             pspan[i] = span;
 636         }
 637         qsort (pspan, nspans, sizeof (fz_textspan *), comparespans);
 638
 639         j = forward ? 0 : nspans - 1;
 640         while (nspans--) {
 641             regmatch_t rm;
 642
 643             span = pspan[j];
 644             j += forward ? 1 : -1;
 645             p = buf;
 646             /* XXX: spans are not sorted "visually" */
 647             for (i = 0; i < MIN (span->len, sizeof (buf) - 1); ++i) {
 648                 if (forward) {
 649                     if (span->text[i].bbox.y0 < y + 1) {
 650                         continue;
 651                     }
 652                 }
 653                 else {
 654                     if (span->text[i].bbox.y0 > y - 1) {
 655                         continue;
 656                     }
 657                 }
 658                 if (span->text[i].c < 256) {
 659                     *p++ = span->text[i].c;
 660                 }
 661                 else  {
 662                     *p++ = '?';
 663                 }
 664             }
 665             if (p == buf) {
 666                 continue;
 667             }
 668             *p++ = 0;
 669
 670             ret = regexec (re, buf, 1, &rm, 0);
 671             if (ret)  {
 672                 if (ret != REG_NOMATCH) {
 673                     size_t size;
 674                     char errbuf[80];
 675                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 676                     printd (state.sock,
 677                             "T regexec error `%.*s'",
 678                             (int) size, errbuf);
 679                     fz_freetextspan (text);
 680                     pdf_freepage (drawpage);
 681                     free (pspan);
 682                     return;
 683                 }
 684             }
 685             else  {
 686                 fz_rect r;
 687
 688                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 689                 r.y0 = span->text[rm.rm_so].bbox.y0;
 690                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 691                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 692
 693                 if (!stop) {
 694                     printd (state.sock, "F %d %d %f %f %f %f",
 695                             pageno, 1,
 696                             r.x0, r.y0,
 697                             r.x1, r.y1);
 698                 }
 699                 else  {
 700                     printd (state.sock, "R %d %d %f %f %f %f",
 701                             pageno, 2,
 702                             r.x0, r.y0,
 703                             r.x1, r.y1);
 704                 }
 705                 printd (state.sock, "T found at %d `%.*s' %f in %f sec",
 706                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 707                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 708                         now () - start);
 709                 stop = 1;
 710             }
 711         }
 712         if (forward) {
 713             pageno += 1;
 714             y = 0;
 715         }
 716         else {
 717             pageno -= 1;
 718             y = INT_MAX;
 719         }
 720         fz_freetextspan (text);
 721         pdf_freepage (drawpage);
 722         free (pspan);
 723     }
 724     end = now ();
 725     if (!stop)  {
 726         printd (state.sock, "T no matches %f sec", end - start);
 727     }
 728     printd (state.sock, "D");
 729 }
 730
 731 static void *mainloop (void *unused)
 732 {
 733     char *p = NULL;
 734     int len, ret, oldlen = 0;
 735
 736     for (;;) {
 737         len = readlen (state.sock);
 738         if (len == 0) {
 739             errx (1, "readlen returned 0");
 740         }
 741
 742         if (oldlen < len + 1) {
 743             p = realloc (p, len + 1);
 744             if (!p) {
 745                 err (1, "realloc %d failed", len + 1);
 746             }
 747             oldlen = len + 1;
 748         }
 749         readdata (state.sock, p, len);
 750         p[len] = 0;
 751
 752         if (!strncmp ("open", p, 4)) {
 753             fz_obj *obj;
 754             char *filename = p + 5;
 755
 756             openxref (filename);
 757             initpdims ();
 758
 759             obj = fz_dictgets (state.xref->trailer, "Info");
 760             if (obj) {
 761                 obj = fz_dictgets (obj, "Title");
 762                 if (obj) {
 763                     printd (state.sock, "t %s", pdf_toutf8 (obj));
 764                 }
 765             }
 766         }
 767         else if (!strncmp ("free", p, 4)) {
 768             void *ptr;
 769
 770             ret = sscanf (p + 4, " %p", &ptr);
 771             if (ret != 1) {
 772                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 773             }
 774             lock ("free");
 775             freepage (ptr);
 776             unlock ("free");
 777             printd (state.sock, "d");
 778         }
 779         else if (!strncmp ("search", p, 6)) {
 780             int icase, pageno, y, ret, len2, forward;
 781             char *pattern;
 782             regex_t re;
 783
 784             ret = sscanf (p + 6, " %d %d %d %d,%n",
 785                           &icase, &pageno, &y, &forward, &len2);
 786             if (ret != 4) {
 787                 errx (1, "malformed search `%s' ret=%d", p, ret);
 788             }
 789
 790             pattern = p + 6 + len2;
 791             ret = regcomp (&re, pattern,
 792                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 793             if (ret) {
 794                 char errbuf[80];
 795                 size_t size;
 796
 797                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 798                 printd (state.sock, "T regcomp failed `%.*s'",
 799                         (int) size, errbuf);
 800             }
 801             else  {
 802                 search (&re, pageno, y, forward);
 803                 regfree (&re);
 804             }
 805         }
 806         else if (!strncmp ("geometry", p, 8)) {
 807             int w, h;
 808
 809             printd (state.sock, "c");
 810             ret = sscanf (p + 8, " %d %d", &w, &h);
 811             if (ret != 2) {
 812                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 813             }
 814             state.h = h;
 815             if (w != state.w) {
 816                 int i;
 817                 state.w = w;
 818                 for (i = 0; i < state.texcount; ++i)  {
 819                     state.texowners[i].slice = NULL;
 820                 }
 821             }
 822             lock ("geometry");
 823             layout ();
 824             process_outline ();
 825             unlock ("geometry");
 826             printd (state.sock, "C %d", state.pagecount);
 827         }
 828         else if (!strncmp ("render", p, 6)) {
 829             int pageno, pindex, w, h, ret;
 830             struct page *page;
 831
 832             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 833             if (ret != 4) {
 834                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 835             }
 836
 837             page = render (pageno, pindex);
 838             printd (state.sock, "r %d %d %d %p\n",
 839                     pageno,
 840                     state.w,
 841                     state.h,
 842                     page);
 843         }
 844         else {
 845             errx (1, "unknown command %.*s", len, p);
 846         }
 847     }
 848     return NULL;
 849 }
 850
 851 static void upload2 (struct page *page, int slicenum, const char *cap)
 852 {
 853     int i;
 854     int w, h;
 855     double start, end;
 856     struct slice *slice = &page->slices[slicenum];
 857
 858     w = page->pixmap->w;
 859     h = page->pixmap->h;
 860
 861     ARSERT (w == slice->w);
 862     if (slice->texindex != -1
 863         && state.texowners[slice->texindex].slice == slice) {
 864         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 865     }
 866     else  {
 867         int subimage = 0;
 868         int index = (state.texindex++ % state.texcount);
 869         size_t offset = 0;
 870
 871         for (i = 0; i < slicenum; ++i) {
 872             offset += w * page->slices[i].h * 4;
 873         }
 874
 875         if (state.texowners[index].w == slice->w) {
 876             if (state.texowners[index].h >= slice->h ) {
 877                 subimage = 1;
 878             }
 879             else {
 880                 state.texowners[index].h = slice->h;
 881             }
 882         }
 883         else  {
 884             state.texowners[index].h = slice->h;
 885         }
 886
 887         state.texowners[index].slice = slice;
 888         state.texowners[index].w = slice->w;
 889         slice->texindex = index;
 890
 891         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 892         start = now ();
 893         if (subimage) {
 894             {
 895                 GLenum err = glGetError ();
 896                 if (err != GL_NO_ERROR) {
 897                     printf ("\e[0;31mERROR1 %d %d %#x\e[0m\n", w, slice->h, err);
 898                     abort ();
 899                 }
 900             }
 901             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
 902                              0,
 903                              0,
 904                              0,
 905                              w,
 906                              slice->h,
 907                              state.texform,
 908                              state.texty,
 909                              page->pixmap->samples + offset
 910                 );
 911             {
 912                 GLenum err = glGetError ();
 913                 if (err != GL_NO_ERROR) {
 914                     printf ("\e[0;31mERROR %d %d %#x\e[0m\n", w, slice->h, err);
 915                     abort ();
 916                 }
 917             }
 918         }
 919         else {
 920             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
 921                           0,
 922                           GL_RGBA8,
 923                           w,
 924                           slice->h,
 925                           0,
 926                           state.texform,
 927                           state.texty,
 928                           page->pixmap->samples + offset
 929                 );
 930         }
 931
 932         end = now ();
 933         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
 934                  subimage ? "sub" : "img",
 935                  page->pageno, slicenum,
 936                  slice->w, slice->h,
 937                  state.texids[slice->texindex],
 938                  end - start);
 939     }
 940 }
 941
 942 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
 943                         value py_v, value ptr_v)
 944 {
 945     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
 946     int dispy = Int_val (dispy_v);
 947     int w = Int_val (w_v);
 948     int h = Int_val (h_v);
 949     int py = Int_val (py_v);
 950     char *s = String_val (ptr_v);
 951     int ret;
 952     void *ptr;
 953     struct page *page;
 954     int slicenum = 0;
 955
 956     if (trylock ("ml_draw")) {
 957         goto done;
 958     }
 959
 960     ret = sscanf (s, "%p", &ptr);
 961     if (ret != 1) {
 962         errx (1, "cannot parse pointer `%s'", s);
 963     }
 964     page = ptr;
 965
 966     w = page->pixmap->w;
 967
 968     ARSERT (h >= 0 && "ml_draw wrong h");
 969
 970     glEnable (GL_TEXTURE_RECTANGLE_ARB);
 971     if (state.useatifs) {
 972         glEnable (GL_FRAGMENT_SHADER_ATI);
 973     }
 974
 975     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
 976         struct slice *slice = &page->slices[slicenum];
 977         if (slice->h > py) {
 978             break;
 979         }
 980         py -= slice->h;
 981     }
 982
 983     h = MIN (state.h, h);
 984     while (h) {
 985         int th;
 986         struct slice *slice = &page->slices[slicenum];
 987
 988         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
 989
 990         th = MIN (h, slice->h - py);
 991         upload2 (page, slicenum, "upload");
 992
 993         glBegin (GL_QUADS);
 994         {
 995             glTexCoord2i (0, py);
 996             glVertex2i (0, dispy);
 997
 998             glTexCoord2i (w, py);
 999             glVertex2i (w, dispy);
1000
1001             glTexCoord2i (w, py+th);
1002             glVertex2i (w, dispy + th);
1003
1004             glTexCoord2i (0, py+th);
1005             glVertex2i (0, dispy + th);
1006         }
1007         glEnd ();
1008
1009         h -= th;
1010         py = 0;
1011         dispy += th;
1012         slicenum += 1;
1013     }
1014
1015     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1016     if (state.useatifs) {
1017         glDisable (GL_FRAGMENT_SHADER_ATI);
1018     }
1019
1020     unlock ("ml_draw");
1021  done:
1022     CAMLreturn (Val_unit);
1023 }
1024
1025 static pdf_link *getlink (struct page *page, int x, int y)
1026 {
1027     fz_point p;
1028     fz_matrix ctm;
1029     pdf_link *link;
1030
1031     p.x = x;
1032     p.y = y;
1033
1034     ctm = fz_invertmatrix (page->pagedim->ctm);
1035     p = fz_transformpoint (ctm, p);
1036
1037     for (link = page->drawpage->links; link; link = link->next) {
1038         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1039             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1040                 if (link->kind == PDF_LGOTO) {
1041                     return link;
1042                 }
1043             }
1044         }
1045     }
1046     return NULL;
1047 }
1048
1049 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1050 {
1051     CAMLparam3 (ptr_v, x_v, y_v);
1052     char *s = String_val (ptr_v);
1053     int ret;
1054
1055     if (trylock ("ml_checklink")) {
1056         ret = 0;
1057     }
1058     else {
1059         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1060                                Int_val (x_v), Int_val (y_v));
1061         unlock ("ml_checklink");
1062     }
1063     CAMLreturn (Val_bool (ret));
1064 }
1065
1066 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1067 {
1068     CAMLparam3 (ptr_v, x_v, y_v);
1069     CAMLlocal2 (ret_v, tup_v);
1070     pdf_link *link;
1071     struct page *page;
1072     char *s = String_val (ptr_v);
1073
1074     if (trylock ("ml_getlink")) {
1075         ret_v = Val_int (0);
1076         goto done;
1077     }
1078
1079     page = parse_pointer ("ml_getlink", s);
1080
1081     link = getlink (page, Int_val (x_v), Int_val (y_v));
1082     if (link) {
1083         int pageno;
1084         fz_point p;
1085         fz_obj *obj;
1086
1087         pageno = -1;
1088         obj = fz_arrayget (link->dest, 0);
1089         if (fz_isindirect (obj)) {
1090             pageno = pdf_findpageobject (state.xref, obj) - 1;
1091         }
1092         else if (fz_isint (obj)) {
1093             pageno = fz_toint (obj);
1094         }
1095
1096         if (fz_arraylen (link->dest) > 3) {
1097             p.x = fz_toint (fz_arrayget (link->dest, 2));
1098             p.y = fz_toint (fz_arrayget (link->dest, 3));
1099             p = fz_transformpoint (page->pagedim->ctm, p);
1100         }
1101         else  {
1102             p.x = 0.0;
1103             p.y = 0.0;
1104         }
1105
1106         tup_v = caml_alloc_tuple (2);
1107         ret_v = caml_alloc_small (1, 1);
1108         Field (tup_v, 0) = Val_int (pageno);
1109         Field (tup_v, 1) = Val_int (p.y);
1110         Field (ret_v, 0) = tup_v;
1111     }
1112     else {
1113         ret_v = Val_int (0);
1114     }
1115     unlock ("ml_getlink");
1116
1117  done:
1118     CAMLreturn (ret_v);
1119 }
1120
1121 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1122 {
1123     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1124     fz_matrix ctm;
1125     fz_point p1, p2;
1126     struct page *page;
1127     fz_textspan *span;
1128     char *s = String_val (ptr_v);
1129     int rectsel = Bool_val (rectsel_v);
1130     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1131
1132     /* stop GCC from complaining about uninitialized variables */
1133     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1134
1135     if (trylock ("ml_gettext")) {
1136         goto done;
1137     }
1138
1139     page = parse_pointer ("ml_gettext", s);
1140
1141     oy = Int_val (oy_v);
1142     p1.x = Int_val (Field (rect_v, 0));
1143     p1.y = Int_val (Field (rect_v, 1));
1144     p2.x = Int_val (Field (rect_v, 2));
1145     p2.y = Int_val (Field (rect_v, 3));
1146
1147     if (0) {
1148         glEnable (GL_BLEND);
1149         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1150         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1151         glColor4f (0, 0, 0, 0.2);
1152         glRecti (p1.x, p1.y, p2.x, p2.y);
1153         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1154         glDisable (GL_BLEND);
1155     }
1156
1157     ctm = page->pagedim->ctm;
1158     if (!page->text) {
1159         fz_error error;
1160         fz_device *tdev;
1161
1162         page->text = fz_newtextspan ();
1163         tdev = fz_newtextdevice (page->text);
1164         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1165                                       page->drawpage->resources,
1166                                       page->drawpage->contents);
1167         if (error) die (error);
1168         fz_freedevice (tdev);
1169     }
1170
1171     printf ("\ec");
1172
1173     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1174     p1.x += page->pixmap->x;
1175     p1.y += page->pixmap->y;
1176     p2.x += page->pixmap->x;
1177     p2.y += page->pixmap->y;
1178     x0 = p1.x;
1179     y0 = p1.y;
1180     x1 = p2.x;
1181     y1 = p2.y;
1182     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1183
1184     for (span = page->text; span; span = span->next) {
1185         int seen = 0;
1186
1187         /* fz_debugtextspanxml (span); */
1188         for (i = 0; i < span->len; ++i) {
1189             long c;
1190
1191             bx0 = span->text[i].bbox.x0;
1192             bx1 = span->text[i].bbox.x1;
1193             by0 = span->text[i].bbox.y0 + oy;
1194             by1 = span->text[i].bbox.y1 + oy;
1195
1196             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1197                 if (!seen) {
1198                     rx0 = bx0 - page->pixmap->x;
1199                     rx1 = bx1 - page->pixmap->x;
1200                     ry0 = by0;
1201                     ry1 = by1;
1202                 }
1203
1204                 seen = 1;
1205                 c = span->text[i].c;
1206                 if (c < 256) {
1207                     if ((isprint (c) && !isspace (c))) {
1208                         if (!rectsel) {
1209                             bx0 -= page->pixmap->x;
1210                             bx1 -= page->pixmap->x;
1211                             glEnable (GL_BLEND);
1212                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1213                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1214                             glColor4f (0.5, 0.5, 0.0, 0.6);
1215                             glRecti (bx0, by0, bx1, by1);
1216                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1217                             glDisable (GL_BLEND);
1218                         }
1219                         if (isprint (c) || c ==' ') {
1220                             rx1 = bx1;
1221                             ry1 = by1;
1222                         }
1223                     }
1224                     putc (c, stdout);
1225                 }
1226                 else  {
1227                     putc ('?', stdout);
1228                 }
1229             }
1230         }
1231
1232         if (rectsel) {
1233             if (seen) {
1234                 glEnable (GL_BLEND);
1235                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1236                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1237                 glColor4f (0.5, 0.5, 0.0, 0.6);
1238                 glRecti (rx0, ry0, rx1, ry1);
1239                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1240                 glDisable (GL_BLEND);
1241             }
1242         }
1243
1244         if (seen && span->eol) {
1245             x0 = page->pixmap->x;
1246             putc ('\n', stdout);
1247         }
1248     }
1249     unlock ("ml_gettext");
1250
1251  done:
1252     CAMLreturn (Val_unit);
1253 }
1254
1255 CAMLprim value ml_getpagewh (value pagedimno_v)
1256 {
1257     CAMLparam1 (pagedimno_v);
1258     CAMLlocal1 (ret_v);
1259     int pagedimno = Int_val (pagedimno_v);
1260
1261     ret_v = caml_alloc_small (4 * Double_wosize, Double_array_tag);
1262     Store_double_field (ret_v, 0, state.pagedims[pagedimno].box.x0);
1263     Store_double_field (ret_v, 1, state.pagedims[pagedimno].box.x1);
1264     Store_double_field (ret_v, 2, state.pagedims[pagedimno].box.y0);
1265     Store_double_field (ret_v, 3, state.pagedims[pagedimno].box.y1);
1266     CAMLreturn (ret_v);
1267 }
1268
1269 static void initgl (void)
1270 {
1271 #ifdef _BIG_ENDIAN
1272     if (strstr ((char *) glGetString (GL_EXTENSIONS),
1273                 "GL_ATI_fragment_shader")) {
1274         /* Here, with MESA, rv280, powerpc32: BGRA(rev) is slow while
1275            ABGR is fast, so fix things in the shader */
1276         state.texform = GL_ABGR_EXT;
1277         state.texty = GL_UNSIGNED_INT_8_8_8_8;
1278
1279         glBindFragmentShaderATI (1);
1280         glBeginFragmentShaderATI ();
1281         {
1282             glSampleMapATI (GL_REG_0_ATI, GL_TEXTURE0_ARB, GL_SWIZZLE_STR_ATI);
1283
1284             glColorFragmentOp1ATI (GL_MOV_ATI,
1285                                    GL_REG_1_ATI, GL_RED_BIT_ATI, GL_NONE,
1286                                    GL_REG_0_ATI, GL_BLUE, GL_NONE);
1287             glColorFragmentOp1ATI (GL_MOV_ATI,
1288                                    GL_REG_1_ATI, GL_BLUE_BIT_ATI, GL_NONE,
1289                                    GL_REG_0_ATI, GL_RED, GL_NONE);
1290             glColorFragmentOp1ATI (
1291                 GL_MOV_ATI,
1292                 GL_REG_0_ATI, GL_RED_BIT_ATI | GL_BLUE_BIT_ATI, GL_NONE,
1293                 GL_REG_1_ATI, GL_NONE, GL_NONE
1294                 );
1295         }
1296         glEndFragmentShaderATI ();
1297         state.useatifs = 1;
1298     }
1299     else {
1300         state.texform = GL_BGRA_EXT;
1301         state.texty = GL_UNSIGNED_INT_8_8_8_8_REV;
1302     }
1303 #else
1304     state.texform = GL_BGRA_EXT;
1305     state.texty = GL_UNSIGNED_INT_8_8_8_8;
1306 #endif
1307 }
1308
1309 CAMLprim value ml_init (value sock_v)
1310 {
1311     int ret;
1312     CAMLparam1 (sock_v);
1313
1314     state.texcount = 256;
1315     state.sliceheight = 16;
1316
1317     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1318     if (!state.texids) {
1319         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1320     }
1321
1322     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1323     if (!state.texowners) {
1324         err (1, "calloc texowners %zu",
1325              state.texcount * sizeof (*state.texowners));
1326     }
1327
1328     glGenTextures (state.texcount, state.texids);
1329
1330     state.sock = Int_val (sock_v);
1331     initgl ();
1332
1333     state.cache = fz_newglyphcache ();
1334     if (!state.cache) {
1335         errx (1, "fz_newglyphcache failed");
1336     }
1337
1338     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1339     if (ret) {
1340         errx (1, "pthread_create: %s", strerror (errno));
1341     }
1342
1343     CAMLreturn (Val_unit);
1344 }