From 850e7563b4a50c838749bc728b99ba587e1ebe72 Mon Sep 17 00:00:00 2001
From: Daniel Garcia <danigm@yaco.es>
Date: Mon, 7 Jun 2010 13:46:03 +0200
Subject: [PATCH] poppler page get text mapping in poppler-glib

---
 glib/poppler-page.cc |  105 +++++++++++++++++++++++++++++++++++++++++++++++++-
 glib/poppler-page.h  |   15 +++++++
 glib/poppler.h       |    1 +
 3 files changed, 119 insertions(+), 2 deletions(-)

diff --git a/glib/poppler-page.cc b/glib/poppler-page.cc
index 39645bd..256872f 100644
--- a/glib/poppler-page.cc
+++ b/glib/poppler-page.cc
@@ -1743,14 +1743,115 @@ poppler_annot_mapping_free (PopplerAnnotMapping *mapping)
   g_free (mapping);
 }
 
-void 
+void
 poppler_page_get_crop_box (PopplerPage *page, PopplerRectangle *rect)
 {
   PDFRectangle* cropBox = page->page->getCropBox ();
-  
+
   rect->x1 = cropBox->x1;
   rect->x2 = cropBox->x2;
   rect->y1 = cropBox->y1;
   rect->y2 = cropBox->y2;
 }
 
+/* PopplerTextMapping type */
+POPPLER_DEFINE_BOXED_TYPE (PopplerTextMapping, poppler_text_mapping,
+			   poppler_text_mapping_copy,
+			   poppler_text_mapping_free)
+
+PopplerTextMapping *
+poppler_text_mapping_new (void)
+{
+  return (PopplerTextMapping *) g_new0 (PopplerTextMapping, 1);
+}
+
+PopplerTextMapping *
+poppler_text_mapping_copy (PopplerTextMapping *mapping)
+{
+  PopplerTextMapping *new_mapping;
+
+  new_mapping = poppler_text_mapping_new ();
+  *new_mapping = *mapping;
+
+  return new_mapping;
+}
+
+void
+poppler_text_mapping_free (PopplerTextMapping *mapping)
+{
+  g_free (mapping);
+}
+
+/**
+ * poppler_page_get_text_mapping:
+ * @page: A #PopplerPage
+ *
+ * Returns a list of #PopplerTextMapping items
+ * This list must be freed with poppler_page_free_text_mapping() when done.
+ *
+ * Return value: A #GList of #PopplerTextMapping
+ **/
+GList *
+poppler_page_get_text_mapping (PopplerPage *page)
+{
+  TextPage *text;
+  TextWordList *wordlist;
+  TextWord *word;
+  GList *mapping_list = NULL;
+  PopplerTextMapping *mapping;
+  int i, j, offset = 0;
+
+  text = poppler_page_get_text_page (page);
+  wordlist = text->makeWordList (gFalse);
+
+  for (i=0; i < wordlist->getLength (); i++)
+  {
+    word = wordlist->get(i);
+    for (j=0; j < word->getLength (); j++)
+    {
+      mapping = poppler_text_mapping_new ();
+      mapping->offset = offset++;
+
+      word->getCharBBox (j,
+                         &(mapping->area.x1),
+                         &(mapping->area.y1),
+                         &(mapping->area.x2),
+                         &(mapping->area.y2));
+
+      mapping_list = g_list_prepend (mapping_list, mapping);
+    }
+
+    // adding spaces and break lines
+    mapping = poppler_text_mapping_new ();
+    mapping->offset = offset++;
+
+    word->getBBox (&(mapping->area.x1),
+                   &(mapping->area.y1),
+                   &(mapping->area.x2),
+                   &(mapping->area.y2));
+
+    mapping->area.x1 = mapping->area.x2;
+    mapping->area.y1 = mapping->area.y2;
+
+    mapping_list = g_list_prepend (mapping_list, mapping);
+  }
+
+  return g_list_reverse (mapping_list);
+}
+
+/**
+ * poppler_page_free_text_mapping:
+ * @list: A list of #PopplerTextMapping<!-- -->s
+ *
+ * Frees a list of #PopplerTextMapping<!-- -->s allocated by
+ * poppler_page_get_text_mapping().
+ **/
+void
+poppler_page_free_text_mapping (GList *list)
+{
+  if (!list)
+    return;
+
+  g_list_foreach (list, (GFunc)poppler_text_mapping_free, NULL);
+  g_list_free (list);
+}
diff --git a/glib/poppler-page.h b/glib/poppler-page.h
index 20dc20f..288276f 100644
--- a/glib/poppler-page.h
+++ b/glib/poppler-page.h
@@ -114,6 +114,8 @@ GList                 *poppler_page_get_annot_mapping    (PopplerPage        *pa
 void                   poppler_page_free_annot_mapping   (GList              *list);
 void 		      poppler_page_get_crop_box 	 (PopplerPage        *page,
 							  PopplerRectangle   *rect);
+GList                 *poppler_page_get_text_mapping     (PopplerPage        *page);
+void                   poppler_page_free_text_mapping    (GList              *list);
 
 
 /* A rectangle on a page, with coordinates in PDF points. */
@@ -215,6 +217,19 @@ PopplerAnnotMapping *poppler_annot_mapping_new      (void);
 PopplerAnnotMapping *poppler_annot_mapping_copy     (PopplerAnnotMapping *mapping);
 void                 poppler_annot_mapping_free     (PopplerAnnotMapping *mapping);
 
+/* Mapping between areas on the current page and text */
+#define POPPLER_TYPE_TEXT_MAPPING                  (poppler_text_mapping_get_type ())
+struct _PopplerTextMapping
+{
+  PopplerRectangle area;
+  gint offset;
+};
+
+GType                poppler_text_mapping_get_type (void) G_GNUC_CONST;
+PopplerTextMapping  *poppler_text_mapping_new      (void);
+PopplerTextMapping  *poppler_text_mapping_copy     (PopplerTextMapping *mapping);
+void                 poppler_text_mapping_free     (PopplerTextMapping *mapping);
+
 G_END_DECLS
 
 #endif /* __POPPLER_PAGE_H__ */
diff --git a/glib/poppler.h b/glib/poppler.h
index 01bcafa..b9cd7dd 100644
--- a/glib/poppler.h
+++ b/glib/poppler.h
@@ -90,6 +90,7 @@ typedef struct _PopplerPageTransition      PopplerPageTransition;
 typedef struct _PopplerImageMapping        PopplerImageMapping;
 typedef struct _PopplerFormFieldMapping    PopplerFormFieldMapping;
 typedef struct _PopplerAnnotMapping        PopplerAnnotMapping;
+typedef struct _PopplerTextMapping         PopplerTextMapping;
 typedef struct _PopplerPage                PopplerPage;
 typedef struct _PopplerFontInfo            PopplerFontInfo;
 typedef struct _PopplerLayer               PopplerLayer;
-- 
1.7.1.226.g770c5