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 struct stuff
 381 {
 382     fz_obj *resources;
 383     fz_obj *mediabox;
 384     fz_obj *cropbox;
 385     fz_obj *rotate;
 386 };
 387
 388 static void
 389 recurse_page (fz_obj *node, int bias, int *pagesp, struct stuff inherit)
 390 {
 391     fz_obj *type;
 392     fz_obj *kids;
 393     fz_obj *count;
 394     char *typestr;
 395     int pages = 0;
 396     int i;
 397
 398     if (!fz_isdict(node))
 399     {
 400         fz_warn("pagetree node is missing, igoring missing pages...");
 401         return;
 402     }
 403
 404     type = fz_dictgets(node, "Type");
 405     kids = fz_dictgets(node, "Kids");
 406     count = fz_dictgets(node, "Count");
 407
 408     if (fz_isname(type))
 409         typestr = fz_toname(type);
 410     else
 411     {
 412         fz_warn("pagetree node (%d %d R) lacks required type", fz_tonum(node), fz_togen(node));
 413
 414         kids = fz_dictgets(node, "Kids");
 415         if (kids)
 416         {
 417             fz_warn("guessing it may be a pagetree node, continuing...");
 418             typestr = "Pages";
 419         }
 420         else
 421         {
 422             fz_warn("guessing it may be a page, continuing...");
 423             typestr = "Page";
 424         }
 425     }
 426
 427     if (!strcmp(typestr, "Page")) {
 428         int rotate;
 429         fz_obj *obj;
 430         fz_rect box;
 431         struct pagedim *p;
 432         int pageno = *pagesp;
 433
 434         state.pagetbl[pageno + bias] = fz_tonum (node);
 435
 436         obj = fz_dictgets (node, "CropBox");
 437         if (!obj) obj = inherit.cropbox;
 438         if (!fz_isarray (obj)) {
 439             obj = fz_dictgets (node, "MediaBox");
 440             if (!obj) obj = inherit.mediabox;
 441
 442             if (!fz_isarray (obj)) {
 443                 die (fz_throw ("cannot find page bounds %d (%d Rd)",
 444                                fz_tonum (node), fz_togen (node)));
 445             }
 446         }
 447         box = pdf_torect (obj);
 448
 449         obj = fz_dictgets (node, "Rotate");
 450         if (!obj) obj = inherit.rotate;
 451         if (fz_isint (obj)) {
 452             rotate = fz_toint (obj);
 453         }
 454         else  {
 455             rotate = 0;
 456         }
 457
 458         p = &state.pagedims[state.pagedimcount - 1];
 459         if ((state.pagedimcount == 0)
 460             || (p->rotate != rotate || memcmp (&p->box, &box, sizeof (box)))) {
 461             size_t size;
 462
 463             size = (state.pagedimcount + 1) * sizeof (*state.pagedims);
 464             state.pagedims = realloc (state.pagedims, size);
 465             if (!state.pagedims) {
 466                 err (1, "realloc pagedims to %zu (%d elems)",
 467                      size, state.pagedimcount + 1);
 468             }
 469             p = &state.pagedims[state.pagedimcount++];
 470             p->rotate = rotate;
 471             p->box = box;
 472             p->pageno = pageno + bias;
 473         }
 474         (*pagesp)++;
 475     }
 476     else if (!strcmp(typestr, "Pages"))
 477     {
 478         fz_obj *inh;
 479
 480         if (!fz_isarray(kids))
 481             fz_warn("page tree node contains no pages");
 482
 483         pdf_logpage("subtree (%d %d R) {\n", fz_tonum(node), fz_togen(node));
 484
 485         inh = fz_dictgets(node, "Resources");
 486         if (inh) inherit.resources = inh;
 487
 488         inh = fz_dictgets(node, "MediaBox");
 489         if (inh) inherit.mediabox = inh;
 490
 491         inh = fz_dictgets(node, "CropBox");
 492         if (inh) inherit.cropbox = inh;
 493
 494         inh = fz_dictgets(node, "Rotate");
 495         if (inh) inherit.rotate = inh;
 496
 497         for (i = 0; i < fz_arraylen(kids); i++)
 498         {
 499             fz_obj *obj = fz_arrayget(kids, i);
 500
 501             /* prevent infinite recursion possible in maliciously crafted PDFs */
 502             if (obj == node)
 503             {
 504                 fz_warn("cyclic page tree");
 505                 return;
 506             }
 507
 508             recurse_page (obj, *pagesp + bias, &pages, inherit);
 509         }
 510
 511         if (pages != fz_toint(count))
 512         {
 513             fz_warn("page tree node contains incorrect number of pages, continuing...");
 514             count = fz_newint(pages);
 515             fz_dictputs(node, "Count", count);
 516             fz_dropobj(count);
 517         }
 518
 519         pdf_logpage("%d pages\n", pages);
 520
 521         (*pagesp) += pages;
 522
 523         pdf_logpage("}\n");
 524     }
 525 }
 526
 527 static void initpdims (void)
 528 {
 529     fz_obj *catalog;
 530     fz_obj *pages;
 531     int count;
 532     double start, end;
 533     struct stuff inherit;
 534
 535     start = now ();
 536     catalog = fz_dictgets (state.xref->trailer, "Root");
 537     pages = fz_dictgets (catalog, "Pages");
 538
 539     inherit.resources = nil;
 540     inherit.mediabox = nil;
 541     inherit.cropbox = nil;
 542     inherit.rotate = nil;
 543
 544     count = 0;
 545     recurse_page (pages, 0, &count, inherit);
 546     end = now ();
 547     printd (state.sock, "T Processed %d pages in %f seconds",
 548             count, end - start);
 549 }
 550
 551 static void layout (void)
 552 {
 553     int pindex;
 554     fz_matrix ctm;
 555     fz_rect box, box2;
 556     double zoom, w;
 557     struct pagedim *p = state.pagedims;
 558
 559     pindex = 0;
 560     printd (state.sock, "c");
 561     for (pindex = 0; pindex < state.pagedimcount; ++pindex, ++p) {
 562         box.x0 = MIN (p->box.x0, p->box.x1);
 563         box.y0 = MIN (p->box.y0, p->box.y1);
 564         box.x1 = MAX (p->box.x0, p->box.x1);
 565         box.y1 = MAX (p->box.y0, p->box.y1);
 566
 567         ctm = fz_identity ();
 568         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 569         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 570         box2 = fz_transformrect (ctm, box);
 571         w = box2.x1 - box2.x0;
 572
 573         zoom = (state.w / w);
 574         ctm = fz_identity ();
 575         ctm = fz_concat (ctm, fz_translate (0, -box.y1));
 576         ctm = fz_concat (ctm, fz_scale (zoom, -zoom));
 577         ctm = fz_concat (ctm, fz_rotate (p->rotate));
 578         p->bbox = fz_roundrect (fz_transformrect (ctm, box));
 579         memcpy (&p->ctm, &ctm, sizeof (ctm));
 580     }
 581
 582     while (p-- != state.pagedims)  {
 583         printd (state.sock, "l %d %d %d",
 584                 p->pageno, p->bbox.x1 - p->bbox.x0, p->bbox.y1 - p->bbox.y0);
 585     }
 586
 587     printd (state.sock, "C %d", state.pagecount);
 588 }
 589
 590 static void recurse_outline (pdf_outline *outline, int level)
 591 {
 592     while (outline) {
 593         fz_obj *obj;
 594         int top = 0;
 595         int pageno = -1;
 596
 597         if (!outline->link) goto next;
 598
 599         obj = outline->link->dest;
 600         if (fz_isarray (obj)) {
 601             int i;
 602             int num;
 603             fz_obj *obj2;
 604             struct pagedim *pagedim = state.pagedims;
 605
 606             obj2 = fz_arrayget (obj, 0);
 607             if (fz_isint (obj2)) {
 608                 pageno = fz_toint (obj2);
 609             }
 610             else {
 611                 num = fz_tonum (obj2);
 612                 for (i = 0; i < state.pagecount; ++i)  {
 613                     if (state.pagetbl[i] == num) {
 614                         pageno = i;
 615                         break;
 616                     }
 617                 }
 618             }
 619
 620             for (i = 0; i < state.pagedimcount; ++i) {
 621                 if (state.pagedims[i].pageno > pageno)
 622                     break;
 623                 pagedim = &state.pagedims[i];
 624             }
 625
 626             if (fz_arraylen (obj) > 3) {
 627                 fz_point p;
 628
 629                 p.x = fz_toint (fz_arrayget (obj, 2));
 630                 p.y = fz_toint (fz_arrayget (obj, 3));
 631                 p = fz_transformpoint (pagedim->ctm, p);
 632                 top = p.y;
 633             }
 634         }
 635
 636         lprintf ("%*c%s %d\n", level, ' ', outline->title, pageno);
 637         printd (state.sock, "o %d %d %d %s",
 638                 level, pageno, top, outline->title);
 639     next:
 640         if (outline->child) {
 641             recurse_outline (outline->child, level + 1);
 642         }
 643         outline = outline->next;
 644     }
 645 }
 646
 647 static void process_outline (void)
 648 {
 649     pdf_outline *outline;
 650
 651     outline = pdf_loadoutline (state.xref);
 652     if (outline) {
 653         recurse_outline (outline, 0);
 654         pdf_dropoutline (outline);
 655     }
 656 }
 657
 658 static int comparespans (const void *l, const void *r)
 659 {
 660     fz_textspan *const*ls = l;
 661     fz_textspan *const*rs = r;
 662
 663     return (*ls)->text->bbox.y0 - (*rs)->text->bbox.y0;
 664 }
 665
 666 /* wishful thinking function */
 667 static void search (regex_t *re, int pageno, int y, int forward)
 668 {
 669     int i, j;
 670     int ret;
 671     char *p;
 672     char buf[256];
 673     fz_error error;
 674     fz_obj *pageobj;
 675     fz_device *tdev;
 676     pdf_page *drawpage;
 677     fz_textspan *text, *span, **pspan;
 678     struct pagedim *pdim, *pdimprev;
 679     int stop = 0;
 680     int niters = 0;
 681     int nspans;
 682     double start, end;
 683
 684     start = now ();
 685     while (pageno >= 0 && pageno < state.pagecount && !stop) {
 686         if (niters++ == 5) {
 687             if (!state.lotsamemory) {
 688                 pdf_agestoreditems (state.xref->store);
 689                 pdf_evictageditems (state.xref->store);
 690             }
 691             niters = 0;
 692             if (hasdata (state.sock)) {
 693                 printd (state.sock, "T attention requested aborting search at %d",
 694                         pageno);
 695                 stop = 1;
 696             }
 697             else {
 698                 printd (state.sock, "T searching in page %d", pageno);
 699             }
 700         }
 701         pdimprev = NULL;
 702         for (i = 0; i < state.pagedimcount; ++i) {
 703             pdim = &state.pagedims[i];
 704             if (pdim->pageno == pageno)  {
 705                 goto found;
 706             }
 707             if (pdim->pageno > pageno) {
 708                 pdim = pdimprev;
 709                 goto found;
 710             }
 711             pdimprev = pdim;
 712         }
 713         pdim = pdimprev;
 714     found:
 715
 716         pageobj = pdf_getpageobject (state.xref, pageno + 1);
 717         if (!pageobj)
 718             die (fz_throw ("cannot retrieve info from page %d", pageno));
 719
 720         error = pdf_loadpage (&drawpage, state.xref, pageobj);
 721         if (error)
 722             die (error);
 723
 724         text = fz_newtextspan ();
 725         tdev = fz_newtextdevice (text);
 726         error = pdf_runcontentstream (tdev, pdim->ctm, state.xref,
 727                                       drawpage->resources,
 728                                       drawpage->contents);
 729         if (error) die (error);
 730         fz_freedevice (tdev);
 731
 732         nspans = 0;
 733         for (span = text; span; span = span->next) {
 734             nspans++;
 735         }
 736         pspan = malloc (sizeof (void *) * nspans);
 737         if (!pspan) {
 738             err (1, "malloc span pointers %zu", sizeof (void *) * nspans);
 739         }
 740         for (i = 0, span = text; span; span = span->next, ++i) {
 741             pspan[i] = span;
 742         }
 743         qsort (pspan, nspans, sizeof (fz_textspan *), comparespans);
 744
 745         j = forward ? 0 : nspans - 1;
 746         while (nspans--) {
 747             regmatch_t rm;
 748
 749             span = pspan[j];
 750             j += forward ? 1 : -1;
 751             p = buf;
 752             /* XXX: spans are not sorted "visually" */
 753             for (i = 0; i < MIN (span->len, sizeof (buf) - 1); ++i) {
 754                 if (forward) {
 755                     if (span->text[i].bbox.y0 < y + 1) {
 756                         continue;
 757                     }
 758                 }
 759                 else {
 760                     if (span->text[i].bbox.y0 > y - 1) {
 761                         continue;
 762                     }
 763                 }
 764                 if (span->text[i].c < 256) {
 765                     *p++ = span->text[i].c;
 766                 }
 767                 else  {
 768                     *p++ = '?';
 769                 }
 770             }
 771             if (p == buf) {
 772                 continue;
 773             }
 774             *p++ = 0;
 775
 776             ret = regexec (re, buf, 1, &rm, 0);
 777             if (ret)  {
 778                 if (ret != REG_NOMATCH) {
 779                     size_t size;
 780                     char errbuf[80];
 781                     size = regerror (ret, re, errbuf, sizeof (errbuf));
 782                     printd (state.sock,
 783                             "T regexec error `%.*s'",
 784                             (int) size, errbuf);
 785                     fz_freetextspan (text);
 786                     pdf_droppage (drawpage);
 787                     free (pspan);
 788                     return;
 789                 }
 790             }
 791             else  {
 792                 fz_rect r;
 793
 794                 r.x0 = span->text[rm.rm_so].bbox.x0 - pdim->bbox.x0;
 795                 r.y0 = span->text[rm.rm_so].bbox.y0;
 796                 r.x1 = span->text[rm.rm_eo - 1].bbox.x1 - pdim->bbox.x0;
 797                 r.y1 = span->text[rm.rm_eo - 1].bbox.y1;
 798
 799                 if (!stop) {
 800                     printd (state.sock, "F %d %d %f %f %f %f",
 801                             pageno, 1,
 802                             r.x0, r.y0,
 803                             r.x1, r.y1);
 804                 }
 805                 else  {
 806                     printd (state.sock, "R %d %d %f %f %f %f",
 807                             pageno, 2,
 808                             r.x0, r.y0,
 809                             r.x1, r.y1);
 810                 }
 811                 printd (state.sock, "T found at %d `%.*s' %f in %f sec",
 812                         pageno, rm.rm_eo - rm.rm_so, &buf[rm.rm_so],
 813                         span->text[0].bbox.y0 - drawpage->mediabox.y0,
 814                         now () - start);
 815                 stop = 1;
 816             }
 817         }
 818         if (forward) {
 819             pageno += 1;
 820             y = 0;
 821         }
 822         else {
 823             pageno -= 1;
 824             y = INT_MAX;
 825         }
 826         fz_freetextspan (text);
 827         pdf_droppage (drawpage);
 828         free (pspan);
 829     }
 830     end = now ();
 831     if (!stop)  {
 832         printd (state.sock, "T no matches %f sec", end - start);
 833     }
 834     printd (state.sock, "D");
 835 }
 836
 837 static void *mainloop (void *unused)
 838 {
 839     char *p = NULL;
 840     int len, ret, oldlen = 0;
 841
 842     for (;;) {
 843         len = readlen (state.sock);
 844         if (len == 0) {
 845             errx (1, "readlen returned 0");
 846         }
 847
 848         if (oldlen < len + 1) {
 849             p = realloc (p, len + 1);
 850             if (!p) {
 851                 err (1, "realloc %d failed", len + 1);
 852             }
 853             oldlen = len + 1;
 854         }
 855         readdata (state.sock, p, len);
 856         p[len] = 0;
 857
 858         if (!strncmp ("open", p, 4)) {
 859             char *filename = p + 5;
 860
 861             openxref (filename);
 862             initpdims ();
 863         }
 864         else if (!strncmp ("free", p, 4)) {
 865             void *ptr;
 866
 867             ret = sscanf (p + 4, " %p", &ptr);
 868             if (ret != 1) {
 869                 errx (1, "malformed free `%.*s' ret=%d", len, p, ret);
 870             }
 871             lock ("free");
 872             freepage (ptr);
 873             unlock ("free");
 874             printd (state.sock, "d");
 875         }
 876         else if (!strncmp ("search", p, 6)) {
 877             int icase, pageno, y, ret, len2, forward;
 878             char *pattern;
 879             regex_t re;
 880
 881             ret = sscanf (p + 6, " %d %d %d %d,%n",
 882                           &icase, &pageno, &y, &forward, &len2);
 883             if (ret != 4) {
 884                 errx (1, "malformed search `%s' ret=%d", p, ret);
 885             }
 886
 887             pattern = p + 6 + len2;
 888             ret = regcomp (&re, pattern,
 889                            REG_EXTENDED | (icase ? REG_ICASE : 0));
 890             if (ret) {
 891                 char errbuf[80];
 892                 size_t size;
 893
 894                 size = regerror (ret, &re, errbuf, sizeof (errbuf));
 895                 printd (state.sock, "T regcomp failed `%.*s'",
 896                         (int) size, errbuf);
 897             }
 898             else  {
 899                 search (&re, pageno, y, forward);
 900                 regfree (&re);
 901             }
 902         }
 903         else if (!strncmp ("geometry", p, 8)) {
 904             int w, h;
 905
 906             ret = sscanf (p + 8, " %d %d", &w, &h);
 907             if (ret != 2) {
 908                 errx (1, "malformed geometry `%.*s' ret=%d", len, p, ret);
 909             }
 910             state.h = h;
 911             if (w != state.w) {
 912                 int i;
 913                 state.w = w;
 914                 for (i = 0; i < state.texcount; ++i)  {
 915                     state.texowners[i].slice = NULL;
 916                 }
 917             }
 918             lock ("geometry");
 919             layout ();
 920             process_outline ();
 921             unlock ("geometry");
 922             printd (state.sock, "d");
 923         }
 924         else if (!strncmp ("render", p, 6)) {
 925             int pageno, pindex, w, h, ret;
 926             struct page *page;
 927
 928             ret = sscanf (p + 6, " %d %d %d %d", &pageno, &pindex, &w, &h);
 929             if (ret != 4) {
 930                 errx (1, "bad render line `%.*s' ret=%d", len, p, ret);
 931             }
 932
 933             page = render (pageno, pindex);
 934             printd (state.sock, "r %d %d %d %p\n",
 935                     pageno,
 936                     state.w,
 937                     state.h,
 938                     page);
 939         }
 940         else {
 941             errx (1, "unknown command %.*s", len, p);
 942         }
 943     }
 944     return NULL;
 945 }
 946
 947 static void upload2 (struct page *page, int slicenum, const char *cap)
 948 {
 949     int i;
 950     int w, h;
 951     double start, end;
 952     struct slice *slice = &page->slices[slicenum];
 953
 954     w = page->pixmap->w;
 955     h = page->pixmap->h;
 956
 957     ARSERT (w == slice->w);
 958     if (slice->texindex != -1
 959         && state.texowners[slice->texindex].slice == slice) {
 960         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 961     }
 962     else  {
 963         int subimage = 0;
 964         int index = (state.texindex++ % state.texcount);
 965         size_t offset = 0;
 966
 967         for (i = 0; i < slicenum; ++i) {
 968             offset += w * page->slices[i].h * 4;
 969         }
 970
 971         if (state.texowners[index].w == slice->w) {
 972             if (state.texowners[index].h >= slice->h ) {
 973                 subimage = 1;
 974             }
 975             else {
 976                 state.texowners[index].h = slice->h;
 977             }
 978         }
 979         else  {
 980             state.texowners[index].h = slice->h;
 981         }
 982
 983         state.texowners[index].slice = slice;
 984         state.texowners[index].w = slice->w;
 985         slice->texindex = index;
 986
 987         glBindTexture (GL_TEXTURE_RECTANGLE_ARB, state.texids[slice->texindex]);
 988         start = now ();
 989         if (subimage) {
 990             {
 991                 GLenum err = glGetError ();
 992                 if (err != GL_NO_ERROR) {
 993                     printf ("\e[0;31mERROR1 %d %d %#x\e[0m\n", w, slice->h, err);
 994                     abort ();
 995                 }
 996             }
 997             glTexSubImage2D (GL_TEXTURE_RECTANGLE_ARB,
 998                              0,
 999                              0,
1000                              0,
1001                              w,
1002                              slice->h,
1003                              state.texform,
1004                              state.texty,
1005                              page->pixmap->samples + offset
1006                 );
1007             {
1008                 GLenum err = glGetError ();
1009                 if (err != GL_NO_ERROR) {
1010                     printf ("\e[0;31mERROR %d %d %#x\e[0m\n", w, slice->h, err);
1011                     abort ();
1012                 }
1013             }
1014         }
1015         else {
1016             glTexImage2D (GL_TEXTURE_RECTANGLE_ARB,
1017                           0,
1018                           GL_RGBA8,
1019                           w,
1020                           slice->h,
1021                           0,
1022                           state.texform,
1023                           state.texty,
1024                           page->pixmap->samples + offset
1025                 );
1026         }
1027
1028         end = now ();
1029         lprintf ("%s[%d] slice=%d(%d,%d) texid=%d %f sec\n",
1030                  subimage ? "sub" : "img",
1031                  page->pageno, slicenum,
1032                  slice->w, slice->h,
1033                  state.texids[slice->texindex],
1034                  end - start);
1035     }
1036 }
1037
1038 CAMLprim value ml_draw (value dispy_v, value w_v, value h_v,
1039                         value py_v, value ptr_v)
1040 {
1041     CAMLparam5 (dispy_v, w_v, h_v, py_v, ptr_v);
1042     int dispy = Int_val (dispy_v);
1043     int w = Int_val (w_v);
1044     int h = Int_val (h_v);
1045     int py = Int_val (py_v);
1046     char *s = String_val (ptr_v);
1047     int ret;
1048     void *ptr;
1049     struct page *page;
1050     int slicenum = 0;
1051
1052     if (trylock ("ml_draw")) {
1053         goto done;
1054     }
1055
1056     ret = sscanf (s, "%p", &ptr);
1057     if (ret != 1) {
1058         errx (1, "cannot parse pointer `%s'", s);
1059     }
1060     page = ptr;
1061
1062     w = page->pixmap->w;
1063
1064     ARSERT (h >= 0 && "ml_draw wrong h");
1065
1066     glEnable (GL_TEXTURE_RECTANGLE_ARB);
1067     if (state.useatifs) {
1068         glEnable (GL_FRAGMENT_SHADER_ATI);
1069     }
1070
1071     for (slicenum = 0; slicenum < page->slicecount; ++slicenum) {
1072         struct slice *slice = &page->slices[slicenum];
1073         if (slice->h > py) {
1074             break;
1075         }
1076         py -= slice->h;
1077     }
1078
1079     h = MIN (state.h, h);
1080     while (h) {
1081         int th;
1082         struct slice *slice = &page->slices[slicenum];
1083
1084         ARSERT (slicenum < page->slicecount && "ml_draw wrong slicenum");
1085
1086         th = MIN (h, slice->h - py);
1087         upload2 (page, slicenum, "upload");
1088
1089         glBegin (GL_QUADS);
1090         {
1091             glTexCoord2i (0, py);
1092             glVertex2i (0, dispy);
1093
1094             glTexCoord2i (w, py);
1095             glVertex2i (w, dispy);
1096
1097             glTexCoord2i (w, py+th);
1098             glVertex2i (w, dispy + th);
1099
1100             glTexCoord2i (0, py+th);
1101             glVertex2i (0, dispy + th);
1102         }
1103         glEnd ();
1104
1105         h -= th;
1106         py = 0;
1107         dispy += th;
1108         slicenum += 1;
1109     }
1110
1111     glDisable (GL_TEXTURE_RECTANGLE_ARB);
1112     if (state.useatifs) {
1113         glDisable (GL_FRAGMENT_SHADER_ATI);
1114     }
1115
1116     unlock ("ml_draw");
1117  done:
1118     CAMLreturn (Val_unit);
1119 }
1120
1121 static pdf_link *getlink (struct page *page, int x, int y)
1122 {
1123     fz_point p;
1124     fz_matrix ctm;
1125     pdf_link *link;
1126
1127     p.x = x;
1128     p.y = y;
1129
1130     ctm = fz_invertmatrix (page->pagedim->ctm);
1131     p = fz_transformpoint (ctm, p);
1132
1133     for (link = page->drawpage->links; link; link = link->next) {
1134         if (p.x >= link->rect.x0 && p.x <= link->rect.x1) {
1135             if (p.y >= link->rect.y0 && p.y <= link->rect.y1) {
1136                 if (link->kind == PDF_LGOTO) {
1137                     return link;
1138                 }
1139             }
1140         }
1141     }
1142     return NULL;
1143 }
1144
1145 CAMLprim value ml_checklink (value ptr_v, value x_v, value y_v)
1146 {
1147     CAMLparam3 (ptr_v, x_v, y_v);
1148     char *s = String_val (ptr_v);
1149     int ret;
1150
1151     if (trylock ("ml_checklink")) {
1152         ret = 0;
1153     }
1154     else {
1155         ret = NULL != getlink (parse_pointer ("ml_checklink", s),
1156                                Int_val (x_v), Int_val (y_v));
1157         unlock ("ml_checklink");
1158     }
1159     CAMLreturn (Val_bool (ret));
1160 }
1161
1162 CAMLprim value ml_getlink (value ptr_v, value x_v, value y_v)
1163 {
1164     CAMLparam3 (ptr_v, x_v, y_v);
1165     CAMLlocal2 (ret_v, tup_v);
1166     pdf_link *link;
1167     struct page *page;
1168     char *s = String_val (ptr_v);
1169
1170     if (trylock ("ml_gettext")) {
1171         ret_v = Val_int (0);
1172         goto done;
1173     }
1174
1175     page = parse_pointer ("ml_getlink", s);
1176
1177     link = getlink (page, Int_val (x_v), Int_val (y_v));
1178     if (link) {
1179         int pageno;
1180         fz_point p;
1181         fz_obj *obj;
1182
1183         pageno = -1;
1184         obj = fz_arrayget (link->dest, 0);
1185         if (fz_isindirect (obj)) {
1186             pageno = pdf_findpageobject (state.xref, obj) - 1;
1187         }
1188         else if (fz_isint (obj)) {
1189             pageno = fz_toint (obj);
1190         }
1191
1192         if (fz_arraylen (link->dest) > 3) {
1193             p.x = fz_toint (fz_arrayget (link->dest, 2));
1194             p.y = fz_toint (fz_arrayget (link->dest, 3));
1195             p = fz_transformpoint (page->pagedim->ctm, p);
1196         }
1197         else  {
1198             p.x = 0.0;
1199             p.y = 0.0;
1200         }
1201
1202         tup_v = caml_alloc_tuple (2);
1203         ret_v = caml_alloc_small (1, 1);
1204         Field (tup_v, 0) = Val_int (pageno);
1205         Field (tup_v, 1) = Val_int (p.y);
1206         Field (ret_v, 0) = tup_v;
1207     }
1208     else {
1209         ret_v = Val_int (0);
1210     }
1211     unlock ("ml_getlink");
1212
1213  done:
1214     CAMLreturn (ret_v);
1215 }
1216
1217 CAMLprim value ml_gettext (value ptr_v, value rect_v, value oy_v, value rectsel_v)
1218 {
1219     CAMLparam4 (ptr_v, rect_v, oy_v, rect_v);
1220     fz_matrix ctm;
1221     fz_point p1, p2;
1222     struct page *page;
1223     fz_textspan *span;
1224     char *s = String_val (ptr_v);
1225     int rectsel = Bool_val (rectsel_v);
1226     int i, bx0, bx1, by0, by1, x0, x1, y0, y1, oy;
1227
1228     /* stop GCC from complaining about uninitialized variables */
1229     int rx0 = rx0, rx1 = rx1, ry0 = ry0, ry1 = ry1;
1230
1231     if (trylock ("ml_gettext")) {
1232         goto done;
1233     }
1234
1235     page = parse_pointer ("ml_gettext", s);
1236
1237     oy = Int_val (oy_v);
1238     p1.x = Int_val (Field (rect_v, 0));
1239     p1.y = Int_val (Field (rect_v, 1));
1240     p2.x = Int_val (Field (rect_v, 2));
1241     p2.y = Int_val (Field (rect_v, 3));
1242
1243     if (0) {
1244         glEnable (GL_BLEND);
1245         glPolygonMode (GL_FRONT_AND_BACK, GL_LINE);
1246         glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1247         glColor4f (0, 0, 0, 0.2);
1248         glRecti (p1.x, p1.y, p2.x, p2.y);
1249         glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1250         glDisable (GL_BLEND);
1251     }
1252
1253     ctm = page->pagedim->ctm;
1254     if (!page->text) {
1255         fz_error error;
1256         fz_device *tdev;
1257
1258         page->text = fz_newtextspan ();
1259         tdev = fz_newtextdevice (page->text);
1260         error = pdf_runcontentstream (tdev, page->pagedim->ctm, state.xref,
1261                                       page->drawpage->resources,
1262                                       page->drawpage->contents);
1263         if (error) die (error);
1264         fz_freedevice (tdev);
1265     }
1266
1267     printf ("\ec");
1268
1269     printf ("BBox %f %f %f %f\n", p1.x, p1.y, p2.x, p2.y);
1270     p1.x += page->pixmap->x;
1271     p1.y += page->pixmap->y;
1272     p2.x += page->pixmap->x;
1273     p2.y += page->pixmap->y;
1274     x0 = p1.x;
1275     y0 = p1.y;
1276     x1 = p2.x;
1277     y1 = p2.y;
1278     printf ("BBox %d %d %d %d %d %d\n", x0, y0, x1, y1, oy, page->pageno);
1279
1280     for (span = page->text; span; span = span->next) {
1281         int seen = 0;
1282
1283         /* fz_debugtextspanxml (span); */
1284         for (i = 0; i < span->len; ++i) {
1285             long c;
1286
1287             bx0 = span->text[i].bbox.x0;
1288             bx1 = span->text[i].bbox.x1;
1289             by0 = span->text[i].bbox.y0 + oy;
1290             by1 = span->text[i].bbox.y1 + oy;
1291
1292             if ((bx1 >= x0 && bx0 <= x1 && by1 >= y0 && by0 <= y1)) {
1293                 if (!seen) {
1294                     rx0 = bx0 - page->pixmap->x;
1295                     rx1 = bx1 - page->pixmap->x;
1296                     ry0 = by0;
1297                     ry1 = by1;
1298                 }
1299
1300                 seen = 1;
1301                 c = span->text[i].c;
1302                 if (c < 256) {
1303                     if ((isprint (c) && !isspace (c))) {
1304                         if (!rectsel) {
1305                             bx0 -= page->pixmap->x;
1306                             bx1 -= page->pixmap->x;
1307                             glEnable (GL_BLEND);
1308                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1309                             glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1310                             glColor4f (0.5, 0.5, 0.0, 0.6);
1311                             glRecti (bx0, by0, bx1, by1);
1312                             glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1313                             glDisable (GL_BLEND);
1314                         }
1315                         if (isprint (c) || c ==' ') {
1316                             rx1 = bx1;
1317                             ry1 = by1;
1318                         }
1319                     }
1320                     putc (c, stdout);
1321                 }
1322                 else  {
1323                     putc ('?', stdout);
1324                 }
1325             }
1326         }
1327
1328         if (rectsel) {
1329             if (seen) {
1330                 glEnable (GL_BLEND);
1331                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1332                 glBlendFunc (GL_DST_ALPHA, GL_SRC_ALPHA);
1333                 glColor4f (0.5, 0.5, 0.0, 0.6);
1334                 glRecti (rx0, ry0, rx1, ry1);
1335                 glPolygonMode (GL_FRONT_AND_BACK, GL_FILL);
1336                 glDisable (GL_BLEND);
1337             }
1338         }
1339
1340         if (seen && span->eol) {
1341             x0 = page->pixmap->x;
1342             putc ('\n', stdout);
1343         }
1344     }
1345     unlock ("ml_gettext");
1346
1347  done:
1348     CAMLreturn (Val_unit);
1349 }
1350
1351 CAMLprim value ml_getpagewh (value pagedimno_v)
1352 {
1353     CAMLparam1 (pagedimno_v);
1354     CAMLlocal1 (ret_v);
1355     int pagedimno = Int_val (pagedimno_v);
1356
1357     ret_v = caml_alloc_small (4 * Double_wosize, Double_array_tag);
1358     Store_double_field (ret_v, 0, state.pagedims[pagedimno].box.x0);
1359     Store_double_field (ret_v, 1, state.pagedims[pagedimno].box.x1);
1360     Store_double_field (ret_v, 2, state.pagedims[pagedimno].box.y0);
1361     Store_double_field (ret_v, 3, state.pagedims[pagedimno].box.y1);
1362     CAMLreturn (ret_v);
1363 }
1364
1365 static void initgl (void)
1366 {
1367 #ifdef _BIG_ENDIAN
1368     if (strstr ((char *) glGetString (GL_EXTENSIONS),
1369                 "GL_ATI_fragment_shader")) {
1370         /* Here, with MESA, rv280, powerpc32: BGRA(rev) is slow while
1371            ABGR is fast, so fix things in the shader */
1372         state.texform = GL_ABGR_EXT;
1373         state.texty = GL_UNSIGNED_INT_8_8_8_8;
1374
1375         glBindFragmentShaderATI (1);
1376         glBeginFragmentShaderATI ();
1377         {
1378             glSampleMapATI (GL_REG_0_ATI, GL_TEXTURE0_ARB, GL_SWIZZLE_STR_ATI);
1379
1380             glColorFragmentOp1ATI (GL_MOV_ATI,
1381                                    GL_REG_1_ATI, GL_RED_BIT_ATI, GL_NONE,
1382                                    GL_REG_0_ATI, GL_BLUE, GL_NONE);
1383             glColorFragmentOp1ATI (GL_MOV_ATI,
1384                                    GL_REG_1_ATI, GL_BLUE_BIT_ATI, GL_NONE,
1385                                    GL_REG_0_ATI, GL_RED, GL_NONE);
1386             glColorFragmentOp1ATI (
1387                 GL_MOV_ATI,
1388                 GL_REG_0_ATI, GL_RED_BIT_ATI | GL_BLUE_BIT_ATI, GL_NONE,
1389                 GL_REG_1_ATI, GL_NONE, GL_NONE
1390                 );
1391         }
1392         glEndFragmentShaderATI ();
1393         state.useatifs = 1;
1394     }
1395     else {
1396         state.texform = GL_BGRA_EXT;
1397         state.texty = GL_UNSIGNED_INT_8_8_8_8_REV;
1398     }
1399 #else
1400     state.texform = GL_BGRA_EXT;
1401     state.texty = GL_UNSIGNED_INT_8_8_8_8;
1402 #endif
1403 }
1404
1405 CAMLprim value ml_init (value sock_v)
1406 {
1407     int ret;
1408     CAMLparam1 (sock_v);
1409
1410     state.texcount = 128;
1411     state.sliceheight = 64;
1412
1413     state.texids = calloc (state.texcount * sizeof (*state.texids), 1);
1414     if (!state.texids) {
1415         err (1, "calloc texids %zu", state.texcount * sizeof (*state.texids));
1416     }
1417
1418     state.texowners = calloc (state.texcount * sizeof (*state.texowners), 1);
1419     if (!state.texowners) {
1420         err (1, "calloc texowners %zu",
1421              state.texcount * sizeof (*state.texowners));
1422     }
1423
1424     glGenTextures (state.texcount, state.texids);
1425
1426     state.sock = Int_val (sock_v);
1427     initgl ();
1428
1429     state.cache = fz_newglyphcache ();
1430     if (!state.cache) {
1431         errx (1, "fz_newglyphcache failed");
1432     }
1433
1434     ret = pthread_create (&state.thread, NULL, mainloop, NULL);
1435     if (ret) {
1436         errx (1, "pthread_create: %s", strerror (errno));
1437     }
1438
1439     CAMLreturn (Val_unit);
1440 }