From 1d9244c1cbebc32c21495921a3c70bd7f5c95e4f Mon Sep 17 00:00:00 2001
From: Daniel Garcia <danigm@yaco.es>
Date: Tue, 15 Jun 2010 16:08:00 +0200
Subject: [PATCH] poppler_page_get_text_layout in poppler/glib

 Returns an array of PopplerRectangle items and each Rectangle is a
 text character position.

 The position in this array represent the offset in text returned by
 poppler_page_get_text
---
 glib/poppler-page.cc |   91 ++++++++++++++++++++++++++++++++++++++++++++++++-
 glib/poppler-page.h  |    4 ++-
 2 files changed, 92 insertions(+), 3 deletions(-)

diff --git a/glib/poppler-page.cc b/glib/poppler-page.cc
index 14601a6..6778419 100644
--- a/glib/poppler-page.cc
+++ b/glib/poppler-page.cc
@@ -1722,14 +1722,101 @@ poppler_annot_mapping_free (PopplerAnnotMapping *mapping)
   g_slice_free (PopplerAnnotMapping, 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;
 }
 
+/**
+ * poppler_page_get_text_layout:
+ * @page: A #PopplerPage
+ *
+ * Returns an array of #PopplerRectangle items
+ * This array must be freed with g_free () when done.
+ *
+ * The position in the array represent the offset in text returned by
+ * poppler_page_get_text
+ *
+ * Return value: An array of #PopplerRectangle, n_areas is the size of
+ * the array.
+ **/
+gboolean
+poppler_page_get_text_layout (PopplerPage       *page,
+                              PopplerRectangle **areas,
+                              guint             *n_areas)
+{
+  TextPage *text;
+  TextWordList *wordlist;
+  TextWord *word, *nextword;
+  PopplerRectangle *rect;
+  int i, j, offset = 0;
+  gdouble x1, y1, x2, y2;
+  gdouble x3, y3, x4, y4;
+
+  *n_areas = 0;
+
+  text = poppler_page_get_text_page (page);
+  wordlist = text->makeWordList (gFalse);
+
+  // Getting the array size
+  for (i=0; i < wordlist->getLength (); i++)
+  {
+    word = wordlist->get (i);
+    *n_areas += word->getLength () + 1;
+  }
+
+  // Creating areas
+  *areas = g_new (PopplerRectangle, *n_areas);
+
+  // Calculating each char position
+  for (i=0; i < wordlist->getLength (); i++)
+  {
+    word = wordlist->get (i);
+    for (j=0; j < word->getLength (); j++)
+    {
+      rect = *areas + offset;
+      word->getCharBBox (j,
+                         &(rect->x1),
+                         &(rect->y1),
+                         &(rect->x2),
+                         &(rect->y2));
+      offset++;
+    }
+
+    // adding spaces and break lines
+    nextword = word->getNext ();
+    if (nextword)
+    {
+      rect = *areas + offset;
+      word->getBBox (&x1, &y1, &x2, &y2);
+      nextword->getBBox (&x3, &y3, &x4, &y4);
+      // space is from one wort to other and with the same height as
+      // first word.
+      rect->x1 = x2;
+      rect->y1 = y1;
+      rect->x2 = x3;
+      rect->y2 = y2;
+    }
+    else
+    {
+      // end of line
+      rect = *areas + offset;
+      word->getBBox (&x1, &y1, &x2, &y2);
+      rect->x1 = x2;
+      rect->y1 = y2;
+      rect->x2 = x2;
+      rect->y2 = y2;
+    }
+    offset++;
+  }
+
+  delete wordlist;
+
+  return gTrue;
+}
diff --git a/glib/poppler-page.h b/glib/poppler-page.h
index 20dc20f..eaa9547 100644
--- a/glib/poppler-page.h
+++ b/glib/poppler-page.h
@@ -114,7 +114,9 @@ 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);
-
+gboolean               poppler_page_get_text_layout      (PopplerPage        *page,
+                                                          PopplerRectangle  **areas,
+                                                          guint              *n_areas);
 
 /* A rectangle on a page, with coordinates in PDF points. */
 #define POPPLER_TYPE_RECTANGLE             (poppler_rectangle_get_type ())
-- 
1.7.1.333.g6339f67