[gtksourceview] SearchContext: fix bug multi-byte characters

[Sébastien Wilmet <swilmet src gnome org>]

commit b7adba6815cda2664fe2724885bdcc263cf95ea3
Author: Sébastien Wilmet <swilmet gnome org>
Date:   Sat Jan 3 15:34:58 2015 +0100

    SearchContext: fix bug multi-byte characters
    
    The @start_position parameter of g_regex_match_full(), g_regex_replace()
    etc is in _bytes_, not characters.
    
    gtk_text_iter_get_visible_text() is used because it is the function used
    elsewhere in SearchContext to retrieve the text to search.

 gtksourceview/gtksourcesearchcontext.c |   12 ++++++++++--
 tests/test-search-context.c            |   16 ++++++++++++++++
 2 files changed, 26 insertions(+), 2 deletions(-)
---
diff --git a/gtksourceview/gtksourcesearchcontext.c b/gtksourceview/gtksourcesearchcontext.c
index 4c6c245..8195af8 100644
--- a/gtksourceview/gtksourcesearchcontext.c
+++ b/gtksourceview/gtksourcesearchcontext.c
@@ -592,6 +592,7 @@ get_text_search_flags (GtkSourceSearchContext *search)
        return flags;
 }
 
+/* @start_pos is in bytes. */
 static void
 regex_search_get_real_start (GtkSourceSearchContext *search,
                             const GtkTextIter      *start,
@@ -599,16 +600,23 @@ regex_search_get_real_start (GtkSourceSearchContext *search,
                             gint                   *start_pos)
 {
        gint max_lookbehind = g_regex_get_max_lookbehind (search->priv->regex);
+       gint i;
+       gchar *text;
 
        *real_start = *start;
 
-       for (*start_pos = 0; *start_pos < max_lookbehind; (*start_pos)++)
+       for (i = 0; i < max_lookbehind; i++)
        {
                if (!gtk_text_iter_backward_char (real_start))
                {
                        break;
                }
        }
+
+       text = gtk_text_iter_get_visible_text (real_start, start);
+       *start_pos = strlen (text);
+
+       g_free (text);
 }
 
 static GRegexMatchFlags
@@ -1858,7 +1866,7 @@ regex_search_scan_segment (GtkSourceSearchContext *search,
 
        DEBUG ({
               g_print ("\n*** regex search - scan segment ***\n");
-              g_print ("start position in the subject: %d\n", start_pos);
+              g_print ("start position in the subject (in bytes): %d\n", start_pos);
        });
 
        match_options = regex_search_get_match_options (&real_start, segment_end);
diff --git a/tests/test-search-context.c b/tests/test-search-context.c
index daa1678..8e3782b 100644
--- a/tests/test-search-context.c
+++ b/tests/test-search-context.c
@@ -1038,6 +1038,22 @@ test_regex_at_word_boundaries (void)
        g_assert_cmpstr (content, ==, "&bb");
        g_free (content);
 
+       /* Test replace multi-byte character */
+
+       gtk_text_buffer_set_text (text_buffer, "–aa", -1);
+       gtk_source_search_settings_set_search_text (settings, "aa");
+       flush_queue ();
+
+       gtk_text_buffer_get_iter_at_offset (text_buffer, &match_start, 1);
+       gtk_text_buffer_get_end_iter (text_buffer, &match_end);
+       gtk_source_search_context_replace (context, &match_start, &match_end, "bb", -1, NULL);
+
+       gtk_text_buffer_get_start_iter (text_buffer, &start);
+       gtk_text_buffer_get_end_iter (text_buffer, &end);
+       content = gtk_text_iter_get_visible_text (&start, &end);
+       g_assert_cmpstr (content, ==, "–bb");
+       g_free (content);
+
        g_object_unref (source_buffer);
        g_object_unref (settings);
        g_object_unref (context);