Add backspace_deletes_character to PangoLogAttr. (#114483)

2003-08-14  Noah Levitt  <nlevitt@columbia.edu>

	* docs/tmpl/main.sgml:
	* pango/break.c:
	* pango/pango-break.h: Add backspace_deletes_character to
        PangoLogAttr. (#114483)

8 files changed, 72 insertions, 0 deletions

diff --git a/ChangeLog b/ChangeLog
index 5e5baf48..7ebc97c9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2003-08-14  Noah Levitt  <nlevitt@columbia.edu>
+
+	* docs/tmpl/main.sgml:
+	* pango/break.c:
+	* pango/pango-break.h: Add backspace_deletes_character to
+        PangoLogAttr. (#114483)
+
 Thu Aug 14 10:41:21 2003  Owen Taylor  <otaylor@redhat.com>
 
 	* pango/modules.c (init_modules): Call g_type_init()
diff --git a/ChangeLog.pre-1-10 b/ChangeLog.pre-1-10
index 5e5baf48..7ebc97c9 100644
--- a/ChangeLog.pre-1-10
+++ b/ChangeLog.pre-1-10
@@ -1,3 +1,10 @@
+2003-08-14  Noah Levitt  <nlevitt@columbia.edu>
+
+	* docs/tmpl/main.sgml:
+	* pango/break.c:
+	* pango/pango-break.h: Add backspace_deletes_character to
+        PangoLogAttr. (#114483)
+
 Thu Aug 14 10:41:21 2003  Owen Taylor  <otaylor@redhat.com>
 
 	* pango/modules.c (init_modules): Call g_type_init()
diff --git a/ChangeLog.pre-1-4 b/ChangeLog.pre-1-4
index 5e5baf48..7ebc97c9 100644
--- a/ChangeLog.pre-1-4
+++ b/ChangeLog.pre-1-4
@@ -1,3 +1,10 @@
+2003-08-14  Noah Levitt  <nlevitt@columbia.edu>
+
+	* docs/tmpl/main.sgml:
+	* pango/break.c:
+	* pango/pango-break.h: Add backspace_deletes_character to
+        PangoLogAttr. (#114483)
+
 Thu Aug 14 10:41:21 2003  Owen Taylor  <otaylor@redhat.com>
 
 	* pango/modules.c (init_modules): Call g_type_init()
diff --git a/ChangeLog.pre-1-6 b/ChangeLog.pre-1-6
index 5e5baf48..7ebc97c9 100644
--- a/ChangeLog.pre-1-6
+++ b/ChangeLog.pre-1-6
@@ -1,3 +1,10 @@
+2003-08-14  Noah Levitt  <nlevitt@columbia.edu>
+
+	* docs/tmpl/main.sgml:
+	* pango/break.c:
+	* pango/pango-break.h: Add backspace_deletes_character to
+        PangoLogAttr. (#114483)
+
 Thu Aug 14 10:41:21 2003  Owen Taylor  <otaylor@redhat.com>
 
 	* pango/modules.c (init_modules): Call g_type_init()
diff --git a/ChangeLog.pre-1-8 b/ChangeLog.pre-1-8
index 5e5baf48..7ebc97c9 100644
--- a/ChangeLog.pre-1-8
+++ b/ChangeLog.pre-1-8
@@ -1,3 +1,10 @@
+2003-08-14  Noah Levitt  <nlevitt@columbia.edu>
+
+	* docs/tmpl/main.sgml:
+	* pango/break.c:
+	* pango/pango-break.h: Add backspace_deletes_character to
+        PangoLogAttr. (#114483)
+
 Thu Aug 14 10:41:21 2003  Owen Taylor  <otaylor@redhat.com>
 
 	* pango/modules.c (init_modules): Call g_type_init()
diff --git a/docs/tmpl/main.sgml b/docs/tmpl/main.sgml
index 9a7ef804..bb894008 100644
--- a/docs/tmpl/main.sgml
+++ b/docs/tmpl/main.sgml
@@ -308,6 +308,18 @@ about the attributes of a single character.
 @is_sentence_boundary: 
 @is_sentence_start: 
 @is_sentence_end: 
+@backspace_deletes_character: If set, backspace deletes one character
+                              rather than the entire grapheme cluster. This
+                              field is only meaningful on grapheme
+                              boundaries (where @is_cursor_position is
+                              set).  In some languages, the full grapheme
+                              (e.g.  letter + diacritics) is considered a
+                              unit, while in others, each decomposed
+                              character in the grapheme is a unit. In the
+                              default implementation of #pango_break, this
+                              bit is set on all grapheme boundaries except
+                              those following Latin, Cyrillic or Greek base
+                              characters.
 
 <!-- ##### FUNCTION pango_shape ##### -->
 <para>
diff --git a/pango/break.c b/pango/break.c
index 491e936f..bbe6b730 100644
--- a/pango/break.c
+++ b/pango/break.c
@@ -328,6 +328,11 @@ static int line_break_indexes[] = {
 #define HIRAGANA(wc) ((wc) >= 0x3040 && (wc) <= 0x309F)
 #define KATAKANA(wc) ((wc) >= 0x30A0 && (wc) <= 0x30FF)
 
+#define LATIN(wc) (((wc) >= 0x0020 && (wc) <= 0x02AF) || ((wc) >= 0x1E00 && (wc) <= 0x1EFF))
+#define CYRILLIC(wc) (((wc) >= 0x0400 && (wc) <= 0x052F))
+#define GREEK(wc) (((wc) >= 0x0370 && (wc) <= 0x3FF) || ((wc) >= 0x1F00 && (wc) <= 0x1FFF))
+#define BACKSPACE_DELETES_CHARACTER(wc) (!LATIN (wc) && !CYRILLIC (wc) && !GREEK (wc))
+
 
 /* p. 132-133 of Unicode spec table 5-6 will help understand this */
 typedef enum
@@ -406,6 +411,7 @@ pango_default_break (const gchar   *text,
   gboolean prev_was_break_space;
   WordType current_word_type = WordNone;
   gunichar last_word_letter = 0;
+  gunichar base_character = 0;
   SentenceState sentence_state = STATE_SENTENCE_OUTSIDE;
   /* Tracks what will be the end of the sentence if a period is
    * determined to actually be a sentence-ending period.
@@ -598,6 +604,13 @@ pango_default_break (const gchar   *text,
             }
         }
       
+      /* If this is a grapheme boundary, we have to decide if backspace
+       * deletes a character or the whole grapheme cluster */
+      if (attrs[i].is_cursor_position)
+        attrs[i].backspace_deletes_character = BACKSPACE_DELETES_CHARACTER (base_character);
+      else
+        attrs[i].backspace_deletes_character = FALSE;
+
       /* ---- Line breaking ---- */
 
       break_type = g_unichar_break_type (wc);
@@ -1277,6 +1290,13 @@ pango_default_break (const gchar   *text,
 
       prev_type = type;
       prev_wc = wc;
+
+      /* wc might not be a valid unicode base character, but really all we
+       * need to know is the last non-combining character */
+      if (type != G_UNICODE_COMBINING_MARK && 
+          type != G_UNICODE_ENCLOSING_MARK && 
+          type != G_UNICODE_NON_SPACING_MARK)
+        base_character = wc;
     }
 }
 
diff --git a/pango/pango-break.h b/pango/pango-break.h
index 3662009b..0ef4b5f3 100644
--- a/pango/pango-break.h
+++ b/pango/pango-break.h
@@ -64,6 +64,11 @@ struct _PangoLogAttr
   guint is_sentence_boundary : 1;
   guint is_sentence_start : 1;  /* first character in a sentence */
   guint is_sentence_end : 1;    /* first non-sentence char after a sentence */
+
+  /* if set, backspace deletes one character rather than
+   * the entire grapheme cluster  
+   */
+  guint backspace_deletes_character : 1; 
 };
 
 /* Determine information about cluster/word/line breaks in a string