link.c

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