path: root/office/zathura-pdf-mupdf/patches/0005-Locate-and-extract-images-from-pages.patch

From c3820b67cc2742e531481c7255bfa3bbdbe36a2a Mon Sep 17 00:00:00 2001
From: Moritz Lipp <mlq@pwmt.org>
Date: Sun, 19 Apr 2015 23:24:39 +0200
Subject: [PATCH 5/6] Locate and extract images from pages

---
 image.c  | 163 ++++++++++++++++++++++++++++-----------------------------------
 plugin.c |   3 +-
 plugin.h |  14 ++++++
 select.c |   3 --
 utils.c  |   4 ++
 5 files changed, 92 insertions(+), 95 deletions(-)

diff --git a/image.c b/image.c
index ef2f67b..97d4143 100644
--- a/image.c
+++ b/image.c
@@ -7,10 +7,9 @@
 #include <mupdf/pdf.h>
 
 #include "plugin.h"
+#include "utils.h"
 
 static void pdf_zathura_image_free(zathura_image_t* image);
-static void get_images(zathura_page_t* page, pdf_obj* dict, girara_list_t* list);
-static void get_resources(zathura_page_t* page, pdf_obj* resource, girara_list_t* list);
 
 girara_list_t*
 pdf_page_images_get(zathura_page_t* page, mupdf_page_t* mupdf_page, zathura_error_t* error)
@@ -30,16 +29,7 @@ pdf_page_images_get(zathura_page_t* page, mupdf_page_t* mupdf_page, zathura_erro
 
   mupdf_document_t* mupdf_document = zathura_document_get_data(document);
 
-  pdf_obj* page_object = pdf_load_object(mupdf_document->ctx, (pdf_document*) mupdf_document->document, zathura_page_get_index(page), 0);
-  if (page_object == NULL) {
-    goto error_free;
-  }
-
-  pdf_obj* resource = pdf_dict_gets(mupdf_document->ctx, page_object, "Resources");
-  if (resource == NULL) {
-    goto error_free;
-  }
-
+  /* Setup image list */
   list = girara_list_new();
   if (list == NULL) {
     if (error != NULL) {
@@ -50,7 +40,25 @@ pdf_page_images_get(zathura_page_t* page, mupdf_page_t* mupdf_page, zathura_erro
 
   girara_list_set_free_function(list, (girara_free_function_t) pdf_zathura_image_free);
 
-  get_resources(page, resource, list);
+  /* Extract images */
+  mupdf_page_extract_text(mupdf_document, mupdf_page);
+
+  fz_page_block* block;
+  for (block = mupdf_page->text->blocks; block < mupdf_page->text->blocks + mupdf_page->text->len; block++) {
+    if (block->type == FZ_PAGE_BLOCK_IMAGE) {
+      fz_image_block *image_block = block->u.image;
+
+      zathura_image_t* zathura_image = g_malloc(sizeof(zathura_image_t));
+
+      zathura_image->position.x1 = image_block->bbox.x0;
+      zathura_image->position.y1 = image_block->bbox.y0;
+      zathura_image->position.x2 = image_block->bbox.x1;
+      zathura_image->position.y2 = image_block->bbox.y1;
+      zathura_image->data        = image_block->image;
+
+      girara_list_append(list, zathura_image);
+    }
+  }
 
   return list;
 
@@ -69,109 +77,82 @@ error_ret:
   return NULL;
 }
 
-
-static void
-pdf_zathura_image_free(zathura_image_t* image)
+cairo_surface_t*
+pdf_page_image_get_cairo(zathura_page_t* page, mupdf_page_t* mupdf_page,
+    zathura_image_t* image, zathura_error_t* error)
 {
-  if (image == NULL) {
-    return;
+  if (page == NULL || mupdf_page == NULL || image == NULL || image->data == NULL) {
+    if (error != NULL) {
+      *error = ZATHURA_ERROR_INVALID_ARGUMENTS;
+    }
+    goto error_ret;
   }
 
-  g_free(image);
-}
+  fz_image* mupdf_image = (fz_image*) image->data;
 
-static void
-get_images(zathura_page_t* page, pdf_obj* dict, girara_list_t* list)
-{
-  if (dict == NULL || list == NULL) {
-    return;
-  }
+  fz_pixmap* pixmap = NULL;
+  cairo_surface_t* surface = NULL;
 
-  if (page == NULL) {
-    return;
+  pixmap = fz_new_pixmap_from_image(mupdf_page->ctx, mupdf_image, 0, 0);
+  if (pixmap == NULL) {
+    goto error_free;
   }
 
-  zathura_document_t* document = zathura_page_get_document(page);
-
-  if (document == NULL) {
-    return;
+  surface = cairo_image_surface_create(CAIRO_FORMAT_RGB24, mupdf_image->w, mupdf_image->h);
+  if (surface == NULL) {
+    goto error_free;
   }
 
-  mupdf_document_t* mupdf_document = zathura_document_get_data(document);
+  unsigned char* surface_data = cairo_image_surface_get_data(surface);
+  int rowstride = cairo_image_surface_get_stride(surface);
 
-  for (int i = 0; i < pdf_dict_len(mupdf_document->ctx, dict); i++) {
-    pdf_obj* image_dict = pdf_dict_get_val(mupdf_document->ctx, dict, i);
-    if (pdf_is_dict(mupdf_document->ctx, image_dict) == 0) {
-      continue;
-    }
+  unsigned char* s = fz_pixmap_samples(mupdf_page->ctx, pixmap);
+  unsigned int n   = fz_pixmap_components(mupdf_page->ctx, pixmap);
 
-    pdf_obj* type = pdf_dict_gets(mupdf_document->ctx, image_dict, "Subtype");
-    if (strcmp(pdf_to_name(mupdf_document->ctx, type), "Image") != 0) {
-      continue;
-    }
+  for (unsigned int y = 0; y < fz_pixmap_height(mupdf_page->ctx, pixmap); y++) {
+    for (unsigned int x = 0; x < fz_pixmap_width(mupdf_page->ctx, pixmap); x++) {
+      guchar* p = surface_data + y * rowstride + x * 4;
 
-    bool duplicate = false;
-    GIRARA_LIST_FOREACH(list, zathura_image_t*, iter, image)
-      if (image->data == image_dict) {
-        duplicate = true;
-        break;
+      // RGB
+      if (n == 4) {
+        p[0] = s[2];
+        p[1] = s[1];
+        p[2] = s[0];
+      // Gray-scale or mask
+      } else {
+        p[0] = s[0];
+        p[1] = s[0];
+        p[2] = s[0];
       }
-    GIRARA_LIST_FOREACH_END(list, zathura_image_t*, iter, image);
-
-    if (duplicate == true) {
-      continue;
+      s += n;
     }
+  }
 
-    pdf_obj* width  = pdf_dict_gets(mupdf_document->ctx, image_dict, "Width");
-    pdf_obj* height = pdf_dict_gets(mupdf_document->ctx, image_dict, "Height");
-
-    zathura_image_t* zathura_image = g_malloc(sizeof(zathura_image_t));
-
-    fprintf(stderr, "image\n");
+  fz_drop_pixmap(mupdf_page->ctx, pixmap);
 
-    // FIXME: Get correct image coordinates
-    zathura_image->data        = image_dict;
-    zathura_image->position.x1 = 0;
-    zathura_image->position.x2 = pdf_to_int(mupdf_document->ctx, width);
-    zathura_image->position.y1 = 0;
-    zathura_image->position.y2 = pdf_to_int(mupdf_document->ctx, height);
+  return surface;
 
-    girara_list_append(list, zathura_image);
-  }
-}
+error_free:
 
-static void
-get_resources(zathura_page_t* page, pdf_obj* resource, girara_list_t* list)
-{
-  if (resource == NULL || list == NULL) {
-    return;
+  if (pixmap != NULL) {
+    fz_drop_pixmap(mupdf_page->ctx, pixmap);
   }
 
-  if (page == NULL) {
-    return;
+  if (surface != NULL) {
+    cairo_surface_destroy(surface);
   }
 
-  zathura_document_t* document = zathura_page_get_document(page);
-
-  if (document == NULL) {
-    return;
-  }
+error_ret:
 
-  mupdf_document_t* mupdf_document = zathura_document_get_data(document);
+  return NULL;
+}
 
-  pdf_obj* x_object = pdf_dict_gets(mupdf_document->ctx, resource, "XObject");
-  if (x_object == NULL) {
+static void
+pdf_zathura_image_free(zathura_image_t* image)
+{
+  if (image == NULL) {
     return;
   }
 
-  get_images(page, x_object, list);
-
-  for (int i = 0; i < pdf_dict_len(mupdf_document->ctx, x_object); i++) {
-    pdf_obj* obj = pdf_dict_get_val(mupdf_document->ctx, x_object, i);
-    pdf_obj* subsrc = pdf_dict_gets(mupdf_document->ctx, obj, "Resources");
-    if (subsrc != NULL && pdf_objcmp(mupdf_document->ctx, resource, subsrc)) {
-      get_resources(page, subsrc, list);
-    }
-  }
+  g_free(image);
 }
-
diff --git a/plugin.c b/plugin.c
index 86cb8de..fef2c6a 100644
--- a/plugin.c
+++ b/plugin.c
@@ -17,12 +17,13 @@ register_functions(zathura_plugin_functions_t* functions)
   functions->page_links_get           = (zathura_plugin_page_links_get_t) pdf_page_links_get;
 #if 0
   functions->document_get_information = (zathura_plugin_document_get_information_t) pdf_document_get_information;
-  functions->page_images_get          = (zathura_plugin_page_images_get_t) pdf_page_images_get;
 #endif
+  functions->page_images_get          = (zathura_plugin_page_images_get_t) pdf_page_images_get;
   functions->page_get_text            = (zathura_plugin_page_get_text_t) pdf_page_get_text;
   functions->page_render              = (zathura_plugin_page_render_t) pdf_page_render;
 #if HAVE_CAIRO
   functions->page_render_cairo        = (zathura_plugin_page_render_cairo_t) pdf_page_render_cairo;
+  functions->page_image_get_cairo     = (zathura_plugin_page_image_get_cairo_t) pdf_page_image_get_cairo;
 #endif
 }
 
diff --git a/plugin.h b/plugin.h
index a229a74..8b6c74d 100644
--- a/plugin.h
+++ b/plugin.h
@@ -113,6 +113,20 @@ girara_list_t* pdf_page_links_get(zathura_page_t* page, mupdf_page_t* mupdf_page
  */
 girara_list_t* pdf_page_images_get(zathura_page_t* page, mupdf_page_t* mupdf_page, zathura_error_t* error);
 
+#if HAVE_CAIRO
+/**
+ * Gets the content of the image in a cairo surface
+ *
+ * @param page Page
+ * @param image Image identifier
+ * @param error Set to an error value (see \ref zathura_error_t) if an
+ *   error occured
+ * @return The cairo image surface or NULL if an error occured
+ */
+cairo_surface_t* pdf_page_image_get_cairo(zathura_page_t* page, mupdf_page_t*
+    mupdf_page, zathura_image_t* image, zathura_error_t* error);
+#endif
+
 /**
  * Get text for selection
  * @param page Page
diff --git a/select.c b/select.c
index 15f9258..c1e1437 100644
--- a/select.c
+++ b/select.c
@@ -7,9 +7,6 @@
 #include "plugin.h"
 #include "utils.h"
 
-void mupdf_page_extract_text(mupdf_document_t* mupdf_document,
-    mupdf_page_t* mupdf_page);
-
 char*
 pdf_page_get_text(zathura_page_t* page, mupdf_page_t* mupdf_page, zathura_rectangle_t rectangle, zathura_error_t* error)
 {
diff --git a/utils.c b/utils.c
index d45a31d..4a003b9 100644
--- a/utils.c
+++ b/utils.c
@@ -15,6 +15,10 @@ mupdf_page_extract_text(mupdf_document_t* mupdf_document, mupdf_page_t* mupdf_pa
 
   fz_try (mupdf_page->ctx) {
     text_device = fz_new_text_device(mupdf_page->ctx, mupdf_page->sheet, mupdf_page->text);
+
+    /* Disable FZ_IGNORE_IMAGE to collect image blocks */
+    fz_disable_device_hints(mupdf_page->ctx, text_device, FZ_IGNORE_IMAGE);
+
     fz_matrix ctm;
     fz_scale(&ctm, 1.0, 1.0);
     fz_run_page(mupdf_page->ctx, mupdf_page->page, text_device, &ctm, NULL);
-- 
1.8.4