From 0d35f3759707c689c6474a8888c288752556c92b Mon Sep 17 00:00:00 2001
From: danigm <danigm@wadobo.com>
Date: Tue, 25 Jan 2011 10:28:12 +0100
Subject: [PATCH] [glib] added get_text_attributes

---
 glib/poppler-page.cc                |  112 +++++++++++++++++++++++++++++++++++
 glib/poppler-page.h                 |    3 +
 glib/reference/poppler-sections.txt |    1 +
 3 files changed, 116 insertions(+), 0 deletions(-)

diff --git a/glib/poppler-page.cc b/glib/poppler-page.cc
index 064e4af..d0fc218 100644
--- a/glib/poppler-page.cc
+++ b/glib/poppler-page.cc
@@ -2203,3 +2203,115 @@ poppler_page_get_text_layout (PopplerPage       *page,
 
   return TRUE;
 }
+
+/**
+ * poppler_page_get_text_attributes:
+ * @page: A #PopplerPage
+ * @attributes: (out) (array length=n_attributes) (transfer container): return location for an array of #PangoAttrList
+ * @n_attributes: length of returned array
+ *
+ * Obtains the attributes of the text as a list of #PangoAttrList
+ * Each array element must be freed with pango_attr_list_unref () when done.
+ * And the array must be freed with g_free ()
+ *
+ * The position in the array represents an offset in the text returned by
+ * poppler_page_get_text ()
+ *
+ * Return value: %TRUE if the page contains text, %FALSE otherwise
+ *
+ * Since: 0.16
+ **/
+gboolean
+poppler_page_get_text_attributes  (PopplerPage     *page,
+                                   PangoAttrList ***attributes,
+                                   guint           *n_attributes)
+{
+  TextPage *text;
+  TextWordList *wordlist;
+  TextWord *word;
+  PangoAttrList *attr_list;
+  gchar *fontname;
+  gboolean subset = FALSE;
+  gint i, j, offset = 0;
+
+  g_return_val_if_fail (POPPLER_IS_PAGE (page), FALSE);
+
+  *n_attributes = 0;
+
+  text = poppler_page_get_text_page (page);
+  wordlist = text->makeWordList (gFalse);
+
+  if (wordlist->getLength () <= 0)
+    {
+      delete wordlist;
+      return FALSE;
+    }
+
+  // Getting the array size
+  for (i = 0; i < wordlist->getLength (); i++)
+    {
+      word = wordlist->get (i);
+      *n_attributes += word->getLength () + 1;
+    }
+
+  *attributes = (PangoAttrList**) g_malloc (sizeof (PangoAttrList*) * (*n_attributes));
+
+  // Calculating each word attributes
+  for (i = 0; i < wordlist->getLength (); i++)
+    {
+      word = wordlist->get (i);
+      fontname = word->getFontName()->getCString();
+      gdouble r, g, b;
+      PangoAttribute *font_attr, *size_attr, *underline_attr, *color_attr;
+
+      // check for a font subset name: capital letters followed by a '+' sign
+      if (fontname) {
+        gchar *tmpfontname = fontname;
+        gint i;
+        for (i=0; *(tmpfontname + i); i++) {
+          if (*(tmpfontname + i) < 'A' || *(tmpfontname + i) > 'Z') {
+            break;
+          }
+        }
+        subset = i > 0 && *(tmpfontname + i) == '+';
+      }
+
+      if (subset) {
+        while (*fontname && *fontname != '+') fontname++;
+        fontname++;
+      }
+
+      // word attributes
+      font_attr = pango_attr_family_new (fontname);
+      size_attr = pango_attr_size_new ((int)word->getFontSize());
+      underline_attr = pango_attr_underline_new (word->isUnderlined() ? PANGO_UNDERLINE_SINGLE : PANGO_UNDERLINE_NONE);
+      word->getColor (&r, &g, &b);
+      color_attr = pango_attr_foreground_new ((int)(r * 65535), (int)(g * 65535), (int)(b * 65535));
+
+      // each char of the word has the same attributes
+      for (j = 0; j < word->getLength (); j++)
+        {
+          (*attributes)[offset] = pango_attr_list_new ();
+          attr_list = (*attributes)[offset];
+          offset++;
+
+          pango_attr_list_insert (attr_list, pango_attribute_copy (font_attr));
+          pango_attr_list_insert (attr_list, pango_attribute_copy (size_attr));
+          pango_attr_list_insert (attr_list, pango_attribute_copy (underline_attr));
+          pango_attr_list_insert (attr_list, pango_attribute_copy (color_attr));
+        }
+
+      pango_attribute_destroy (font_attr);
+      pango_attribute_destroy (size_attr);
+      pango_attribute_destroy (underline_attr);
+      pango_attribute_destroy (color_attr);
+
+      // space or breakline
+      (*attributes)[offset] = pango_attr_list_new ();
+      offset++;
+    }
+
+  delete wordlist;
+
+  return TRUE;
+}
diff --git a/glib/poppler-page.h b/glib/poppler-page.h
index d40c0ee..f662b15 100644
--- a/glib/poppler-page.h
+++ b/glib/poppler-page.h
@@ -128,6 +128,9 @@ void 		      poppler_page_get_crop_box 	 (PopplerPage        *page,
 gboolean               poppler_page_get_text_layout      (PopplerPage        *page,
                                                           PopplerRectangle  **rectangles,
                                                           guint              *n_rectangles);
+gboolean               poppler_page_get_text_attributes  (PopplerPage        *page,
+                                                          PangoAttrList    ***attributes,
+                                                          guint              *n_attributes);
 
 /* A rectangle on a page, with coordinates in PDF points. */
 #define POPPLER_TYPE_RECTANGLE             (poppler_rectangle_get_type ())
diff --git a/glib/reference/poppler-sections.txt b/glib/reference/poppler-sections.txt
index 6ad9ee2..14521e5 100644
--- a/glib/reference/poppler-sections.txt
+++ b/glib/reference/poppler-sections.txt
@@ -38,6 +38,7 @@ poppler_page_get_selected_text
 poppler_page_find_text
 poppler_page_get_text
 poppler_page_get_text_layout
+poppler_page_get_text_attributes
 poppler_page_get_link_mapping
 poppler_page_free_link_mapping
 poppler_page_get_image_mapping
-- 
1.7.4.rc2.258.g844d7