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 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 (pdf_xref *xref, fz_obj *node, 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] = 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 = *pagesp;
 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 (xref, obj, &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 (state.xref, pages, &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 /* wishful thinking function */
 626 static void search (regex_t *re, int pageno, int y, int forward)
 627 {
 628     int i;
 629     int ret;
 630     char *p;
 631     char buf[256];
 632     fz_error error;
 633     fz_obj *pageobj;
 634     fz_device *tdev;
 635     pdf_page *drawpage;
 636     fz_textspan *text, *span;
 637     struct pagedim *pdim, *pdimprev;
 638     int stop = 0;
 639     int niters = 0;
 640     double start, end;
 641
 642     start = now ();
 643     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 644         if (niters++ == 5) {
 645             niters = 0;
 646             if (hasdata (state.sock)) {
 647                 printd (state.sock, "T \"attention requested aborting search at %d\"",
 648                         pageno);
 649                 stop = 1;
 650             }
 651             else {
 652                 printd (state.sock, "T \"searching in page %d\"", pageno);
 653             }
 654         }
 655         pdimprev = NULL;
 656         for (i = 0; i < state.pagedimcount; ++i) {
 657             pdim = &state.pagedims[i];
 658             if (pdim->pageno == pageno)  {
 659                 goto found;
 660             }
 661             if (pdim->pageno > pageno) {
 662                 pdim = pdimprev;
 663                 goto found;
 664             }
 665             pdimprev = pdim;
 666         }
 667         pdim = pdimprev;
 668     found:
 669
 670         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 671         if (!pageobj)
 672             die (fz_throw ("cannot retrieve info from page %d", pageno));
 673
 674         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 675         if (error)
 676             die (error);
 677
 678         text = fz_newtextspan ();
 679         tdev = fz_newtextdevice (text);
 680         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 681                                       drawpage->resources,
 682                                       drawpage->contents);
 683         if (error) die (error);
 684         fz_freedevice (tdev);
 685
 686         for (span = text; span; span = span->next) {
 687             regmatch_t rm;
 688
 689             p = buf;
 690             /* XXX: spans are not sorted "visually" */
 691             for (i = 0; i < MIN (span->len, sizeof (buf) - 1); ++i) {
 692                 if (forward) {
 693                     if (span->text[i].bbox.y0 < y + 1) {
 694                         continue;
 695                     }
 696                 }
 697                 else {
 698                     if (span->text[i].bbox.y0 > y - 1) {
 699                         continue;
 700                     }
 701                 }
 702                 if (span->text[i].c < 256) {
 703                     *p++ = span->text[i].c;
 704                 }
 705                 else  {
 706                     *p++ = '?';
 707                 }
 708             }
 709             if (p == buf) {
 710                 continue;
 711             }
 712             *p++ = 0;
 713
 714             ret = regexec (re, buf, 1, &rm, 0);
 715             if (ret)  {
 716                 if (ret != REG_NOMATCH) {
 717                     size_t size;
 718                     char errbuf[80];
 719                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 720                     printd (state.sock,
 721                             "T \"regexec error `%.*s'\"",
 722                             (int) size, errbuf);
 723                 }
 724             }
 725             else  {
 726                 fz_rect r;
 727
 728                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 729                 r.y0 = span->text[rm.rm_so].bbox.y0;
 730                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 731                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 732
 733                 if (!stop) {
 734                     printd (state.sock, "F %d %d %f %f %f %f",
 735                             pageno, 1,
 736                             r.x0, r.y0,
 737                             r.x1, r.y1);
 738                 }
 739                 else  {
 740                     printd (state.sock, "R %d %d %f %f %f %f",
 741                             pageno, 2,
 742                             r.x0, r.y0,
 743                             r.x1, r.y1);
 744                 }
 745                 printd (state.sock, "T \"found at %d `%.*s' %f in %f sec\"",
 746                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 747                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 748                         now () - start);
 749                 stop = 1;
 750             }
 751         }
 752         if (forward) {
 753             pageno += 1;
 754             y = 0;
 755         }
 756         else {
 757             pageno -= 1;
 758             y = INT_MAX;
 759         }
 760         fz_freetextspan (text);
 761         pdf_droppage (drawpage);
 762     }
 763     end = now ();
 764     if (!stop)  {
 765         printd (state.sock, "T \"no matches %f sec\"", end - start);
 766     }
 767     printd (state.sock, "d");
 768 }
 769
 770 static void *mainloop (void *unused)
 771 {
 772     char *p = NULL;
 773     int len, ret, oldlen = 0;
 774
 775     for (;;) {
 776         len = readlen (state.sock);
 777         if (len == 0) {
 778             errx (1, "readlen returned 0");
 779         }
 780
 781         if (oldlen < len + 1) {
 782             p = realloc (p, len + 1);
 783             if (!p) {
 784                 err (1, "realloc %d failed", len + 1);
 785             }
 786             oldlen = len + 1;
 787         }
 788         readdata (state.sock, p, len);
 789         p[len] = 0;
 790
 791         if (!strncmp ("open", p, 4)) {
 792             char *filename = p + 5;
 793
 794             openxref (filename);
 795             initpdims ();
 796         }
 797         else if (!strncmp ("free", p, 4)) {
 798             void *ptr;
 799
 800             ret = sscanf (p + 4, " %p", &ptr);
 801             if (ret != 1) {
 802                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 803             }
 804             lock ("free");
 805             freepage (ptr);
 806             unlock ("free");
 807             printd (state.sock, "d");
 808         }
 809         else if (!strncmp ("search", p, 6)) {
 810             int icase, pageno, y, ret, len2, forward;
 811             char *pattern;
 812             regex_t re;
 813
 814             ret = sscanf (p + 6, " %d %d %d %d %n",
 815                           &icase, &pageno, &y, &forward, &len2);
 816             if (ret != 4) {
 817                 errx (1, "malformed search `%s' ret=%d", p, ret);
 818             }
 819
 820             pattern = p + 6 + len2;
 821             ret = regcomp (&re, pattern,
 822                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 823             if (ret) {
 824                 char errbuf[80];
 825                 size_t size;
 826
 827                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 828                 printd (state.sock, "T \"regcomp failed `%.*s'\"", (int) size, errbuf);
 829             }
 830             else  {
 831                 search (&re, pageno, y, forward);
 832                 regfree (&re);
 833             }
 834         }
 835         else if (!strncmp ("geometry", p, 8)) {
 836             int w, h;
 837
 838             ret = sscanf (p + 8, " %d %d", &w, &h);
 839             if (ret != 2) {
 840                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 841             }
 842             state.h = h;
 843             if (w != state.w) {
 844                 int i;
 845                 state.w = w;
 846                 for (i = 0; i < state.texcount; ++i)  {
 847                     state.texowners[i].slice = NULL;
 848                 }
 849             }
 850             lock ("geometry");
 851             layout ();
 852             process_outline ();
 853             unlock ("geometry");
 854             printd (state.sock, "d");
 855         }
 856         else if (!strncmp ("render", p, 6)) {
 857             int pageno, pindex, w, h, ret;
 858             struct page *page;
 859
 860             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 861             if (ret != 4) {
 862                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 863             }
 864
 865             page = render (pageno, pindex);
 866             printd (state.sock, "r %d %d %d %p\n",
 867                     pageno,
 868                     state.w,
 869                     state.h,
 870                     page);
 871         }
 872         else {
 873             errx (1, "unknown command %.*s", len, p);
 874         }
 875     }
 876     return NULL;
 877 }
 878
 879 static void upload2 (struct page *page, int slicenum, const char *cap)
 880 {
 881     int i;
 882     int w, h;
 883     double start, end;
 884     struct slice *slice = &page->slices[slicenum];
 885
 886     w = page->pixmap->w;
 887     h = page->pixmap->h;
 888
 889     ARSERT (w == slice->w);
 890     if (slice->texindex != -1
 891         && state.texowners[slice->texindex].slice == slice) {
 892         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 893     }
 894     else  {
 895         int subimage = 0;
 896         int index = (state.texindex++ % state.texcount);
 897         size_t offset = 0;
 898
 899         for (i = 0; i < slicenum; ++i) {
 900             offset += w * page->slices[i].h * 4;
 901         }
 902
 903         if (state.texowners[index].w == slice->w) {
 904             if (state.texowners[index].h >= slice->h ) {
 905                 subimage = 1;
 906             }
 907             else {
 908                 state.texowners[index].h = slice->h;
 909             }
 910         }
 911         else  {
 912             state.texowners[index].h = slice->h;
 913         }
 914
 915         state.texowners[index].slice = slice;
 916         state.texowners[index].w = slice->w;
 917         slice->texindex = index;
 918
 919         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 920         start = now ();
 921         if (subimage) {
 922             {
 923                 GLenum err = glGetError ();
 924                 if (err != GL_NO_ERROR) {
 925                     printf ("\e[0;31mERROR1 %d %d %#x\e[0m\n", w, slice->h, err);
 926                     abort ();
 927                 }
 928             }
 929             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
 930                              0,
 931                              0,
 932                              0,
 933                              w,
 934                              slice->h,
 935                              state.texform,
 936                              state.texty,
 937                              page->pixmap->samples + offset
 938                 );
 939             {
 940                 GLenum err = glGetError ();
 941                 if (err != GL_NO_ERROR) {
 942                     printf ("\e[0;31mERROR %d %d %#x\e[0m\n", w, slice->h, err);
 943                     abort ();
 944                 }
 945             }
 946         }
 947         else {
 948             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
 949                           0,
 950                           GL_RGBA8,
 951                           w,
 952                           slice->h,
 953                           0,
 954                           state.texform,
 955                           state.texty,
 956                           page->pixmap->samples + offset
 957                 );
 958         }
 959
 960         end = now ();
 961         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
 962                  subimage ? "sub" : "img",
 963                  page->pageno, slicenum,
 964                  slice->w, slice->h,
 965                  state.texids[slice->texindex],
 966                  end - start);
 967     }
 968 }
 969
 970 CAMLprim value ml_preload (value ptr_v)
 971 {
 972     int i;
 973     int ret;
 974     void *ptr;
 975     CAMLparam1 (ptr_v);
 976     char *s = String_val (ptr_v);
 977     struct page *page;
 978
 979     if (trylock ("ml_preload")) {
 980         goto done;
 981     }
 982     ret = sscanf (s, "%p", &ptr);
 983     if (ret != 1) {
 984         errx (1, "cannot parse pointer `%s'", s);
 985     }
 986
 987     page = ptr;
 988     for (i = 0; i < page->slicecount; ++i) {
 989         upload2 (ptr, i, "preload");
 990     }
 991
 992     unlock ("ml_preload");
 993  done:
 994     CAMLreturn (Val_unit);
 995 }
 996
 997 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
 998                         value py_v, value ptr_v)
 999 {
1000     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
1001     int dispy = Int_val (dispy_v);
1002     int w = Int_val (w_v);
1003     int h = Int_val (h_v);
1004     int py = Int_val (py_v);
1005     char *s = String_val (ptr_v);
1006     int ret;
1007     void *ptr;
1008     struct page *page;
1009     int slicenum = 0;
1010
1011     if (trylock ("ml_draw")) {
1012         goto done;
1013     }
1014
1015     ret = sscanf (s, "%p", &ptr);
1016     if (ret != 1) {
1017         errx (1, "cannot parse pointer `%s'", s);
1018     }
1019     page = ptr;
1020
1021     w = page->pixmap->w;
1022
1023     ARSERT (h >= 0 && "ml_draw wrong h");
1024
1025     glEnable (GL_TEXTURE_RECTANGLE_ARB);
1026     if (state.useatifs) {
1027         glEnable (GL_FRAGMENT_SHADER_ATI);
1028     }
1029
1030     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
1031         struct slice *slice = &page->slices[slicenum];
1032         if (slice->h > py) {
1033             break;
1034         }
1035         py -= slice->h;
1036     }
1037
1038     h = MIN (state.h, h);
1039     while (h) {
1040         int th;
1041         struct slice *slice = &page->slices[slicenum];
1042
1043         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
1044
1045         th = MIN (h, slice->h - py);
1046         upload2 (page, slicenum, "upload");
1047
1048         glBegin (GL_QUADS);
1049         {
1050             glTexCoord2i (0, py);
1051             glVertex2i (0, dispy);
1052
1053             glTexCoord2i (w, py);
1054             glVertex2i (w, dispy);
1055
1056             glTexCoord2i (w, py+th);
1057             glVertex2i (w, dispy + th);
1058
1059             glTexCoord2i (0, py+th);
1060             glVertex2i (0, dispy + th);
1061         }
1062         glEnd ();
1063
1064         h -= th;
1065         py = 0;
1066         dispy += th;
1067         slicenum += 1;
1068     }
1069
1070     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1071     if (state.useatifs) {
1072         glDisable (GL_FRAGMENT_SHADER_ATI);
1073     }
1074
1075     unlock ("ml_draw");
1076  done:
1077     CAMLreturn (Val_unit);
1078 }
1079
1080 static pdf_link *getlink (struct page *page, int x, int y)
1081 {
1082     fz_point p;
1083     fz_matrix ctm;
1084     pdf_link *link;
1085
1086     p.x = x;
1087     p.y = y;
1088
1089     ctm = fz_invertmatrix (page->pagedim->ctm);
1090     p = fz_transformpoint (ctm, p);
1091
1092     for (link = page->drawpage->links; link; link = link->next) {
1093         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1094             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1095                 if (link->kind == PDF_LGOTO) {
1096                     return link;
1097                 }
1098             }
1099         }
1100     }
1101     return NULL;
1102 }
1103
1104 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1105 {
1106     CAMLparam3 (ptr_v, x_v, y_v);
1107     char *s = String_val (ptr_v);
1108     int ret;
1109
1110     if (trylock ("ml_checklink")) {
1111         ret = 0;
1112     }
1113     else {
1114         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1115                                Int_val (x_v), Int_val (y_v));
1116         unlock ("ml_checklink");
1117     }
1118     CAMLreturn (Val_bool (ret));
1119 }
1120
1121 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1122 {
1123     CAMLparam3 (ptr_v, x_v, y_v);
1124     CAMLlocal2 (ret_v, tup_v);
1125     pdf_link *link;
1126     struct page *page;
1127     char *s = String_val (ptr_v);
1128
1129     if (trylock ("ml_gettext")) {
1130         ret_v = Val_int (0);
1131         goto done;
1132     }
1133
1134     page = parse_pointer ("ml_getlink", s);
1135
1136     link = getlink (page, Int_val (x_v), Int_val (y_v));
1137     if (link) {
1138         int pageno;
1139         fz_point p;
1140         fz_obj *obj;
1141
1142         pageno = -1;
1143         obj = fz_arrayget (link->dest, 0);
1144         if (fz_isindirect (obj)) {
1145             pageno = pdf_findpageobject (state.xref, obj) - 1;
1146         }
1147         else if (fz_isint (obj)) {
1148             pageno = fz_toint (obj);
1149         }
1150
1151         if (fz_arraylen (link->dest) > 3) {
1152             p.x = fz_toint (fz_arrayget (link->dest, 2));
1153             p.y = fz_toint (fz_arrayget (link->dest, 3));
1154             p = fz_transformpoint (page->pagedim->ctm, p);
1155         }
1156         else  {
1157             p.x = 0.0;
1158             p.y = 0.0;
1159         }
1160
1161         tup_v = caml_alloc_tuple (2);
1162         ret_v = caml_alloc_small (1, 1);
1163         Field (tup_v, 0) = Val_int (pageno);
1164         Field (tup_v, 1) = Val_int (p.y);
1165         Field (ret_v, 0) = tup_v;
1166     }
1167     else {
1168         ret_v = Val_int (0);
1169     }
1170     unlock ("ml_getlink");
1171
1172  done:
1173     CAMLreturn (ret_v);
1174 }
1175
1176 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1177 {
1178     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1179     fz_matrix ctm;
1180     fz_point p1, p2;
1181     struct page *page;
1182     fz_textspan *span;
1183     char *s = String_val (ptr_v);
1184     int rectsel = Bool_val (rectsel_v);
1185     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1186
1187     /* stop GCC from complaining about uninitialized variables */
1188     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1189
1190     if (trylock ("ml_gettext")) {
1191         goto done;
1192     }
1193
1194     page = parse_pointer ("ml_gettext", s);
1195
1196     oy = Int_val (oy_v);
1197     p1.x = Int_val (Field (rect_v, 0));
1198     p1.y = Int_val (Field (rect_v, 1));
1199     p2.x = Int_val (Field (rect_v, 2));
1200     p2.y = Int_val (Field (rect_v, 3));
1201
1202     if (0) {
1203         glEnable (GL_BLEND);
1204         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1205         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1206         glColor4f (0, 0, 0, 0.2);
1207         glRecti (p1.x, p1.y, p2.x, p2.y);
1208         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1209         glDisable (GL_BLEND);
1210     }
1211
1212     ctm = page->pagedim->ctm;
1213     if (!page->text) {
1214         fz_error error;
1215         fz_device *tdev;
1216
1217         page->text = fz_newtextspan ();
1218         tdev = fz_newtextdevice (page->text);
1219         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1220                                       page->drawpage->resources,
1221                                       page->drawpage->contents);
1222         if (error) die (error);
1223         fz_freedevice (tdev);
1224     }
1225
1226     printf ("\ec");
1227
1228     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1229     p1.x += page->pixmap->x;
1230     p1.y += page->pixmap->y;
1231     p2.x += page->pixmap->x;
1232     p2.y += page->pixmap->y;
1233     x0 = p1.x;
1234     y0 = p1.y;
1235     x1 = p2.x;
1236     y1 = p2.y;
1237     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1238
1239     for (span = page->text; span; span = span->next) {
1240         int seen = 0;
1241
1242         /* fz_debugtextspanxml (span); */
1243         for (i = 0; i < span->len; ++i) {
1244             long c;
1245
1246             bx0 = span->text[i].bbox.x0;
1247             bx1 = span->text[i].bbox.x1;
1248             by0 = span->text[i].bbox.y0 + oy;
1249             by1 = span->text[i].bbox.y1 + oy;
1250
1251             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1252                 if (!seen) {
1253                     rx0 = bx0 - page->pixmap->x;
1254                     rx1 = bx1 - page->pixmap->x;
1255                     ry0 = by0;
1256                     ry1 = by1;
1257                 }
1258
1259                 seen = 1;
1260                 c = span->text[i].c;
1261                 if (c < 256) {
1262                     if ((isprint (c) && !isspace (c))) {
1263                         if (!rectsel) {
1264                             bx0 -= page->pixmap->x;
1265                             bx1 -= page->pixmap->x;
1266                             glEnable (GL_BLEND);
1267                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1268                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1269                             glColor4f (0.5, 0.5, 0.0, 0.6);
1270                             glRecti (bx0, by0, bx1, by1);
1271                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1272                             glDisable (GL_BLEND);
1273                         }
1274                         if (isprint (c) || c ==' ') {
1275                             rx1 = bx1;
1276                             ry1 = by1;
1277                         }
1278                     }
1279                     putc (c, stdout);
1280                 }
1281                 else  {
1282                     putc ('?', stdout);
1283                 }
1284             }
1285         }
1286
1287         if (rectsel) {
1288             if (seen) {
1289                 glEnable (GL_BLEND);
1290                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1291                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1292                 glColor4f (0.5, 0.5, 0.0, 0.6);
1293                 glRecti (rx0, ry0, rx1, ry1);
1294                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1295                 glDisable (GL_BLEND);
1296             }
1297         }
1298
1299         if (seen && span->eol) {
1300             x0 = page->pixmap->x;
1301             putc ('\n', stdout);
1302         }
1303     }
1304     unlock ("ml_gettext");
1305
1306  done:
1307     CAMLreturn (Val_unit);
1308 }
1309
1310 static void initgl (void)
1311 {
1312 #ifdef _BIG_ENDIAN
1313     if (strstr ((char *) glGetString (GL_EXTENSIONS),
1314                 "GL_ATI_fragment_shader")) {
1315         /* Here, with MESA, rv280, powerpc32: BGRA(rev) is slow while
1316            ABGR is fast, so fix things in the shader */
1317         state.texform = GL_ABGR_EXT;
1318         state.texty = GL_UNSIGNED_INT_8_8_8_8;
1319
1320         glBindFragmentShaderATI (1);
1321         glBeginFragmentShaderATI ();
1322         {
1323             glSampleMapATI (GL_REG_0_ATI, GL_TEXTURE0_ARB, GL_SWIZZLE_STR_ATI);
1324
1325             glColorFragmentOp1ATI (GL_MOV_ATI,
1326                                    GL_REG_1_ATI, GL_RED_BIT_ATI, GL_NONE,
1327                                    GL_REG_0_ATI, GL_BLUE, GL_NONE);
1328             glColorFragmentOp1ATI (GL_MOV_ATI,
1329                                    GL_REG_1_ATI, GL_BLUE_BIT_ATI, GL_NONE,
1330                                    GL_REG_0_ATI, GL_RED, GL_NONE);
1331             glColorFragmentOp1ATI (
1332                 GL_MOV_ATI,
1333                 GL_REG_0_ATI, GL_RED_BIT_ATI | GL_BLUE_BIT_ATI, GL_NONE,
1334                 GL_REG_1_ATI, GL_NONE, GL_NONE
1335                 );
1336         }
1337         glEndFragmentShaderATI ();
1338         state.useatifs = 1;
1339     }
1340     else {
1341         state.texform = GL_BGRA_EXT;
1342         state.texty = GL_UNSIGNED_INT_8_8_8_8_REV;
1343     }
1344 #else
1345     state.texform = GL_BGRA_EXT;
1346     state.texty = GL_UNSIGNED_INT_8_8_8_8;
1347 #endif
1348 }
1349
1350 CAMLprim value ml_init (value sock_v)
1351 {
1352     int ret;
1353     CAMLparam1 (sock_v);
1354
1355     state.texcount = 128;
1356     state.sliceheight = 64;
1357
1358     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1359     if (!state.texids) {
1360         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1361     }
1362
1363     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1364     if (!state.texowners) {
1365         err (1, "calloc texowners %zu",
1366              state.texcount * sizeof (*state.texowners));
1367     }
1368
1369     glGenTextures (state.texcount, state.texids);
1370
1371     state.sock = Int_val (sock_v);
1372     initgl ();
1373
1374     state.cache = fz_newglyphcache ();
1375     if (!state.cache) {
1376         errx (1, "fz_newglyphcache failed");
1377     }
1378
1379     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1380     if (ret) {
1381         errx (1, "pthread_create: %s", strerror (errno));
1382     }
1383
1384     CAMLreturn (Val_unit);
1385 }