pdf/xpdf/pdf-document.cc

   1 /* -*- Mode: C++; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8; c-indent-level: 8 -*- */
   2 /* pdfdocument.h: Implementation of EvDocument for PDF
   3  * Copyright (C) 2004, Red Hat, Inc.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2, or (at your option)
   8  * any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program; if not, write to the Free Software
  17  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  18  */
  19
  20 #include "gpdf-g-switch.h"
  21 #include "pdf-document.h"
  22 #include "ev-ps-exporter.h"
  23 #include "ev-document-find.h"
  24 #include "gpdf-g-switch.h"
  25
  26 #include "GlobalParams.h"
  27 #include "GDKSplashOutputDev.h"
  28 #include "PDFDoc.h"
  29 #include "PSOutputDev.h"
  30
  31 typedef struct
  32 {
  33         PdfDocument *document;
  34         gunichar *ucs4;
  35         glong ucs4_len;
  36         guint idle;
  37         /* full results are only possible for the rendered current page */
  38         int current_page;
  39         GArray *current_page_results;
  40         guchar *other_page_flags; /* length n_pages + 1, first element ignored */
  41         int start_page;   /* skip this one as we iterate, since we did it first */
  42         int search_page;  /* the page we're searching now */
  43         TextOutputDev *output_dev;
  44 } PdfDocumentSearch;
  45
  46 typedef struct _PdfDocumentClass PdfDocumentClass;
  47
  48 #define PDF_DOCUMENT_CLASS(klass)     (G_TYPE_CHECK_CLASS_CAST ((klass), PDF_TYPE_DOCUMENT, PdfDocumentClass))
  49 #define PDF_IS_DOCUMENT_CLASS(klass)  (G_TYPE_CHECK_CLASS_TYPE ((klass), PDF_TYPE_DOCUMENT))
  50 #define PDF_DOCUMENT_GET_CLASS(obj)   (G_TYPE_INSTANCE_GET_CLASS ((obj), PDF_TYPE_DOCUMENT, PdfDocumentClass))
  51
  52 struct _PdfDocumentClass
  53 {
  54         GObjectClass parent_class;
  55 };
  56
  57 struct _PdfDocument
  58 {
  59         GObject parent_instance;
  60
  61         int page;
  62         int page_x_offset;
  63         int page_y_offset;
  64         double scale;
  65         GdkDrawable *target;
  66
  67         GDKSplashOutputDev *out;
  68         PSOutputDev *ps_out;
  69         PDFDoc *doc;
  70
  71         gboolean page_valid;
  72
  73         PdfDocumentSearch *search;
  74 };
  75
  76 static void pdf_document_document_iface_init    (EvDocumentIface     *iface);
  77 static void pdf_document_ps_exporter_iface_init (EvPSExporterIface   *iface);
  78 static void pdf_document_find_iface_init        (EvDocumentFindIface *iface);
  79 static void pdf_document_search_free            (PdfDocumentSearch   *search);
  80 static void pdf_document_search_page_changed    (PdfDocumentSearch   *search);
  81
  82 G_DEFINE_TYPE_WITH_CODE (PdfDocument, pdf_document, G_TYPE_OBJECT,
  83                          {
  84                                  G_IMPLEMENT_INTERFACE (EV_TYPE_DOCUMENT,
  85                                                         pdf_document_document_iface_init);
  86                                  G_IMPLEMENT_INTERFACE (EV_TYPE_PS_EXPORTER,
  87                                                         pdf_document_ps_exporter_iface_init);
  88                                  G_IMPLEMENT_INTERFACE (EV_TYPE_DOCUMENT_FIND,
  89                                                         pdf_document_find_iface_init);
  90                          });
  91
  92 static gboolean
  93 document_validate_page (PdfDocument *pdf_document)
  94 {
  95         if (!pdf_document->page_valid) {
  96                 pdf_document->doc->displayPage (pdf_document->out, pdf_document->page,
  97                                                 72 * pdf_document->scale,
  98                                                 72 * pdf_document->scale,
  99                                                 0, gTrue, gTrue);
 100
 101                 pdf_document->page_valid = TRUE;
 102
 103                 /* Update the search results available to the app since
 104                  * we only provide full results on the current page
 105                  */
 106                 if (pdf_document->search)
 107                         pdf_document_search_page_changed (pdf_document->search);
 108         }
 109 }
 110
 111 static gboolean
 112 pdf_document_load (EvDocument  *document,
 113                    const char  *uri,
 114                    GError     **error)
 115 {
 116         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 117         PDFDoc *newDoc;
 118         int err;
 119         char *filename;
 120         GString *filename_g;
 121
 122         if (!globalParams) {
 123                 globalParams = new GlobalParams("/etc/xpdfrc");
 124                 globalParams->setupBaseFontsFc(NULL);
 125         }
 126
 127         filename = g_filename_from_uri (uri, NULL, error);
 128         if (!filename)
 129                 return FALSE;
 130
 131         filename_g = new GString (filename);
 132         g_free (filename);
 133
 134         // open the PDF file, assumes ownership of filename_g
 135         newDoc = new PDFDoc(filename_g, 0, 0);
 136
 137         if (!newDoc->isOk()) {
 138                 err = newDoc->getErrorCode();
 139                 delete newDoc;
 140
 141                 /* FIXME: Add a real error enum to EvDocument */
 142                 g_set_error (error, G_FILE_ERROR,
 143                              G_FILE_ERROR_FAILED,
 144                              "Failed to load document (error %d) '%s'\n",
 145                              err,
 146                              uri);
 147
 148                 return FALSE;
 149         }
 150
 151         if (pdf_document->doc)
 152                 delete pdf_document->doc;
 153         pdf_document->doc = newDoc;
 154
 155         pdf_document->page = 1;
 156
 157         if (pdf_document->out)
 158                 pdf_document->out->startDoc(pdf_document->doc->getXRef());
 159
 160         pdf_document->page_valid = FALSE;
 161
 162         return TRUE;
 163 }
 164
 165 static int
 166 pdf_document_get_n_pages (EvDocument  *document)
 167 {
 168         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 169
 170         if (pdf_document->doc)
 171                 return pdf_document->doc->getNumPages();
 172         else
 173                 return 1;
 174 }
 175
 176 static void
 177 pdf_document_set_page (EvDocument  *document,
 178                        int          page)
 179 {
 180         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 181
 182         page = CLAMP (page, 1, ev_document_get_n_pages (document));
 183
 184         if (page != pdf_document->page) {
 185                 pdf_document->page = page;
 186                 pdf_document->page_valid = FALSE;
 187         }
 188
 189 }
 190
 191 static int
 192 pdf_document_get_page (EvDocument  *document)
 193 {
 194         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 195
 196         return pdf_document->page;
 197 }
 198
 199 static void
 200 redraw_callback (void *data)
 201 {
 202         /* Need to hook up through a EvDocument callback? */
 203 }
 204
 205 static void
 206 pdf_document_set_target (EvDocument  *document,
 207                          GdkDrawable *target)
 208 {
 209         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 210
 211         if (pdf_document->target != target) {
 212                 if (pdf_document->target)
 213                         g_object_unref (pdf_document->target);
 214
 215                 pdf_document->target = target;
 216
 217                 if (pdf_document->target)
 218                         g_object_ref (pdf_document->target);
 219
 220                 if (pdf_document->out) {
 221                         delete pdf_document->out;
 222                         pdf_document->out = NULL;
 223                 }
 224
 225                 if (pdf_document->target) {
 226                         pdf_document->out = new GDKSplashOutputDev (gdk_drawable_get_screen (pdf_document->target),
 227                                                          redraw_callback, (void*) document);
 228
 229                         if (pdf_document->doc)
 230                                 pdf_document->out->startDoc(pdf_document->doc->getXRef());
 231
 232                 }
 233
 234                 pdf_document->page_valid = FALSE;
 235         }
 236 }
 237
 238 static void
 239 pdf_document_set_scale (EvDocument  *document,
 240                         double       scale)
 241 {
 242         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 243
 244         if (pdf_document->scale != scale) {
 245                 pdf_document->scale = scale;
 246                 pdf_document->page_valid = FALSE;
 247         }
 248 }
 249
 250 static void
 251 pdf_document_set_page_offset (EvDocument  *document,
 252                               int          x,
 253                               int          y)
 254 {
 255         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 256
 257         pdf_document->page_x_offset = x;
 258         pdf_document->page_y_offset = y;
 259 }
 260
 261 static void
 262 pdf_document_get_page_size (EvDocument   *document,
 263                             int          *width,
 264                             int          *height)
 265 {
 266         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 267
 268         if (document_validate_page (pdf_document)) {
 269                 if (width)
 270                         *width = pdf_document->out->getBitmapWidth();
 271                 if (height)
 272                         *height = pdf_document->out->getBitmapHeight();
 273         } else {
 274                 if (width)
 275                         *width = 1;
 276                 if (height)
 277                         *height = 1;
 278         }
 279 }
 280
 281 static void
 282 pdf_document_render (EvDocument  *document,
 283                      int          clip_x,
 284                      int          clip_y,
 285                      int          clip_width,
 286                      int          clip_height)
 287 {
 288         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 289         GdkRectangle page;
 290         GdkRectangle draw;
 291
 292         if (!document_validate_page (pdf_document) || !pdf_document->target)
 293                 return;
 294
 295         page.x = pdf_document->page_x_offset;
 296         page.y = pdf_document->page_y_offset;
 297         page.width = pdf_document->out->getBitmapWidth();
 298         page.height = pdf_document->out->getBitmapHeight();
 299
 300         draw.x = clip_x;
 301         draw.y = clip_y;
 302         draw.width = clip_width;
 303         draw.height = clip_height;
 304
 305         if (gdk_rectangle_intersect (&page, &draw, &draw))
 306                 pdf_document->out->redraw (draw.x - page.x, draw.y - page.y,
 307                                            pdf_document->target,
 308                                            draw.x, draw.y,
 309                                            draw.width, draw.height);
 310 }
 311
 312 static void
 313 pdf_document_search_emit_found (PdfDocumentSearch *search)
 314 {
 315         PdfDocument *pdf_document = search->document;
 316         int n_pages;
 317         double pages_done;
 318         GArray *tmp_results;
 319         int i;
 320
 321         n_pages = ev_document_get_n_pages (EV_DOCUMENT (search->document));
 322         if (search->search_page > search->start_page) {
 323                 pages_done = search->search_page - search->start_page;
 324         } else {
 325                 pages_done = n_pages - search->start_page + search->search_page;
 326         }
 327
 328         tmp_results = g_array_new (FALSE, FALSE, sizeof (EvFindResult));
 329         g_array_append_vals (tmp_results,
 330                              search->current_page_results->data,
 331                              search->current_page_results->len);
 332
 333         /* Now append a bogus element for each page that has a result in it,
 334          * that is not the current page
 335          */
 336         i = 1;
 337         while (i <= n_pages) {
 338                 if (i != pdf_document->page &&
 339                     search->other_page_flags[i]) {
 340                         EvFindResult result;
 341
 342                         result.page_num = i;
 343
 344                         /* Use bogus coordinates, again we can't get coordinates
 345                          * until this is the current page because TextOutputDev
 346                          * isn't good enough
 347                          */
 348                         result.highlight_area.x = -1;
 349                         result.highlight_area.y = -1;
 350                         result.highlight_area.width = 1;
 351                         result.highlight_area.height = 1;
 352
 353                         g_array_append_val (tmp_results, result);
 354                 }
 355
 356                 ++i;
 357         }
 358
 359         ev_document_find_found (EV_DOCUMENT_FIND (pdf_document),
 360                                 (EvFindResult*) tmp_results->data,
 361                                 tmp_results->len,
 362                                 pages_done / (double) n_pages);
 363
 364         g_array_free (tmp_results, TRUE);
 365 }
 366
 367 static void
 368 pdf_document_search_page_changed (PdfDocumentSearch   *search)
 369 {
 370         PdfDocument *pdf_document = search->document;
 371         int current_page;
 372         EvFindResult result;
 373         int xMin, yMin, xMax, yMax;
 374
 375         current_page = pdf_document->page;
 376
 377         if (!pdf_document->page_valid) {
 378                 /* we can't do anything until displayPage() */
 379                 search->current_page = -1;
 380                 return;
 381         }
 382
 383         if (search->current_page == current_page)
 384                 return;
 385
 386         /* We need to create current_page_results for the new current page */
 387         g_array_set_size (search->current_page_results, 0);
 388
 389         if (pdf_document->out->findText (search->ucs4, search->ucs4_len,
 390                                          gTrue, gTrue, // startAtTop, stopAtBottom
 391                                          gFalse, gFalse, // startAtLast, stopAtLast
 392                                          &xMin, &yMin, &xMax, &yMax)) {
 393                 result.page_num = pdf_document->page;
 394
 395                 result.highlight_area.x = xMin;
 396                 result.highlight_area.y = yMin;
 397                 result.highlight_area.width = xMax - xMin;
 398                 result.highlight_area.height = yMax - yMin;
 399
 400                 g_array_append_val (search->current_page_results, result);
 401
 402                 /* Now find further results */
 403
 404                 while (pdf_document->out->findText (search->ucs4, search->ucs4_len,
 405                                                     gFalse, gTrue,
 406                                                     gTrue, gFalse,
 407                                                     &xMin, &yMin, &xMax, &yMax)) {
 408
 409                         result.page_num = pdf_document->page;
 410
 411                         result.highlight_area.x = xMin;
 412                         result.highlight_area.y = yMin;
 413                         result.highlight_area.width = xMax - xMin;
 414                         result.highlight_area.height = yMax - yMin;
 415
 416                         g_array_append_val (search->current_page_results, result);
 417                 }
 418         }
 419
 420         /* needed for the initial current page since we don't search
 421          * it in the idle
 422          */
 423         search->other_page_flags[current_page] =
 424                 search->current_page_results->len > 0;
 425
 426         pdf_document_search_emit_found (search);
 427 }
 428
 429 static gboolean
 430 pdf_document_search_idle_callback (void *data)
 431 {
 432         PdfDocumentSearch *search = (PdfDocumentSearch*) data;
 433         PdfDocument *pdf_document = search->document;
 434         int n_pages;
 435         double xMin, yMin, xMax, yMax;
 436         gboolean found;
 437
 438         /* Note that PDF page count is 1 through n_pages INCLUSIVE
 439          * like a real book. We are looking to add one result for each
 440          * page with a match, because the coordinates are meaningless
 441          * with TextOutputDev, so we just want to flag matching pages
 442          * and then when the user switches to the current page, we
 443          * will emit "found" again with the real results.
 444          */
 445         n_pages = ev_document_get_n_pages (EV_DOCUMENT (search->document));
 446
 447         if (search->search_page == search->start_page) {
 448                 goto end_search;
 449         }
 450
 451         if (search->output_dev == 0) {
 452                 /* First time through here... */
 453                 search->output_dev = new TextOutputDev (NULL, gTrue, gFalse, gFalse);
 454                 if (!search->output_dev->isOk()) {
 455                         goto end_search;
 456                 }
 457         }
 458
 459         pdf_document->doc->displayPage (search->output_dev,
 460                                         search->search_page,
 461                                         72, 72, 0, gTrue, gFalse);
 462
 463         if (search->output_dev->findText (search->ucs4,
 464                                           search->ucs4_len,
 465                                           gTrue, gTrue, // startAtTop, stopAtBottom
 466                                           gFalse, gFalse, // startAtLast, stopAtLast
 467                                           &xMin, &yMin, &xMax, &yMax)) {
 468                 /* This page has results */
 469                 search->other_page_flags[search->search_page] = TRUE;
 470
 471                 pdf_document_search_emit_found (search);
 472         }
 473
 474         search->search_page += 1;
 475         if (search->search_page > n_pages) {
 476                 /* wrap around */
 477                 search->search_page = 1;
 478         }
 479
 480         return TRUE;
 481
 482  end_search:
 483         /* We're done. */
 484         search->idle = 0; /* will return FALSE to remove */
 485         return FALSE;
 486 }
 487
 488 static void
 489 pdf_document_find_begin (EvDocumentFind   *document,
 490                          const char       *search_string,
 491                          gboolean          case_sensitive)
 492 {
 493         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 494         PdfDocumentSearch *search;
 495         int n_pages;
 496         gunichar *ucs4;
 497         glong ucs4_len;
 498
 499         /* FIXME handle case_sensitive (right now XPDF
 500          * code is always case insensitive for ASCII
 501          * and case sensitive for all other languaages)
 502          */
 503
 504         g_assert (sizeof (gunichar) == sizeof (Unicode));
 505         ucs4 = g_utf8_to_ucs4_fast (search_string, -1,
 506                                     &ucs4_len);
 507
 508         if (pdf_document->search &&
 509             pdf_document->search->ucs4_len == ucs4_len &&
 510             memcmp (pdf_document->search->ucs4,
 511                     ucs4,
 512                     sizeof (gunichar) * ucs4_len) == 0) {
 513                 /* Search is unchanged */
 514                 g_free (ucs4);
 515                 return;
 516         }
 517
 518         if (pdf_document->search) {
 519                 pdf_document_search_free (pdf_document->search);
 520                 pdf_document->search = NULL;
 521         }
 522
 523         search = g_new0 (PdfDocumentSearch, 1);
 524
 525         search->ucs4 = ucs4;
 526         search->ucs4_len = ucs4_len;
 527
 528         search->current_page_results = g_array_new (FALSE,
 529                                                     FALSE,
 530                                                     sizeof (EvFindResult));
 531         n_pages = ev_document_get_n_pages (EV_DOCUMENT (document));
 532
 533         /* This is an array of bool; with the first value ignored
 534          * so we can index by the based-at-1 page numbers
 535          */
 536         search->other_page_flags = g_new0 (guchar, n_pages + 1);
 537
 538         search->document = pdf_document;
 539
 540         search->idle = g_idle_add (pdf_document_search_idle_callback,
 541                                    search);
 542
 543         search->output_dev = 0;
 544
 545         search->start_page = pdf_document->page;
 546         search->search_page = search->start_page + 1;
 547         if (search->search_page > n_pages)
 548                 search->search_page = 1;
 549
 550         search->current_page = -1;
 551
 552         pdf_document->search = search;
 553
 554         /* Update for the current page right away */
 555         pdf_document_search_page_changed (search);
 556 }
 557
 558 static void
 559 pdf_document_find_cancel (EvDocumentFind   *document)
 560 {
 561         PdfDocument *pdf_document = PDF_DOCUMENT (document);
 562
 563         if (pdf_document->search) {
 564                 pdf_document_search_free (pdf_document->search);
 565                 pdf_document->search = NULL;
 566         }
 567 }
 568
 569 static void
 570 pdf_document_search_free (PdfDocumentSearch   *search)
 571 {
 572         if (search->idle != 0)
 573                 g_source_remove (search->idle);
 574
 575         if (search->output_dev)
 576                 delete search->output_dev;
 577
 578         g_array_free (search->current_page_results, TRUE);
 579         g_free (search->other_page_flags);
 580
 581         g_free (search->ucs4);
 582         g_free (search);
 583 }
 584
 585 static void
 586 pdf_document_ps_export_begin (EvPSExporter *exporter, const char *filename)
 587 {
 588         PdfDocument *document = PDF_DOCUMENT (exporter);
 589
 590         if (document->ps_out)
 591                 delete document->ps_out;
 592
 593         document->ps_out = new PSOutputDev ((char *)filename, document->doc->getXRef(),
 594                                             document->doc->getCatalog(), 1,
 595                                             ev_document_get_n_pages (EV_DOCUMENT (document)),
 596                                             psModePS);
 597 }
 598
 599 static void
 600 pdf_document_ps_export_do_page (EvPSExporter *exporter, int page)
 601 {
 602         PdfDocument *document = PDF_DOCUMENT (exporter);
 603
 604         document->doc->displayPage (document->ps_out, page,
 605                                     72.0, 72.0, 0, gTrue, gFalse);
 606 }
 607
 608 static void
 609 pdf_document_ps_export_end (EvPSExporter *exporter)
 610 {
 611         PdfDocument *document = PDF_DOCUMENT (exporter);
 612
 613         delete document->ps_out;
 614         document->ps_out = NULL;
 615 }
 616
 617 static void
 618 pdf_document_finalize (GObject *object)
 619 {
 620         PdfDocument *pdf_document = PDF_DOCUMENT (object);
 621
 622         if (pdf_document->search)
 623                 pdf_document_search_free (pdf_document->search);
 624
 625         if (pdf_document->target)
 626                 g_object_unref (pdf_document->target);
 627
 628         if (pdf_document->out)
 629                 delete pdf_document->out;
 630         if (pdf_document->ps_out)
 631                 delete pdf_document->ps_out;
 632         if (pdf_document->doc)
 633                 delete pdf_document->doc;
 634
 635 }
 636
 637 static void
 638 pdf_document_class_init (PdfDocumentClass *klass)
 639 {
 640         GObjectClass *gobject_class = G_OBJECT_CLASS (klass);
 641
 642         gobject_class->finalize = pdf_document_finalize;
 643 }
 644
 645 static void
 646 pdf_document_document_iface_init (EvDocumentIface *iface)
 647 {
 648         iface->load = pdf_document_load;
 649         iface->get_n_pages = pdf_document_get_n_pages;
 650         iface->set_page = pdf_document_set_page;
 651         iface->get_page = pdf_document_get_page;
 652         iface->set_scale = pdf_document_set_scale;
 653         iface->set_target = pdf_document_set_target;
 654         iface->set_page_offset = pdf_document_set_page_offset;
 655         iface->get_page_size = pdf_document_get_page_size;
 656         iface->render = pdf_document_render;
 657 }
 658
 659 static void
 660 pdf_document_ps_exporter_iface_init (EvPSExporterIface *iface)
 661 {
 662         iface->begin = pdf_document_ps_export_begin;
 663         iface->do_page = pdf_document_ps_export_do_page;
 664         iface->end = pdf_document_ps_export_end;
 665 }
 666
 667
 668 static void
 669 pdf_document_find_iface_init (EvDocumentFindIface *iface)
 670 {
 671         iface->begin = pdf_document_find_begin;
 672         iface->cancel = pdf_document_find_cancel;
 673 }
 674
 675 static void
 676 pdf_document_init (PdfDocument *pdf_document)
 677 {
 678         pdf_document->page = 1;
 679         pdf_document->page_x_offset = 0;
 680         pdf_document->page_y_offset = 0;
 681         pdf_document->scale = 1.;
 682
 683         pdf_document->page_valid = FALSE;
 684 }
 685