Some UTF-8 fixes and experimental 16-bit unicode font support

svn path=/trunk/; revision=5279

1 files changed, 151 insertions, 59 deletions

diff --git a/e-util/e-font.c b/e-util/e-font.c
index cfee9f0923..15fa868254 100644
--- a/e-util/e-font.c
+++ b/e-util/e-font.c
@@ -16,58 +16,169 @@
 #include <unicode.h>
 #include "e-font.h"
 
+#define FONT_TESTING
+
 struct _EFont {
-	GdkFont font;
+	gint refcount;
+	GdkFont *font;
+	gboolean bytes;
+	unicode_iconv_t to;
+	unicode_iconv_t from;
 };
 
 EFont *
 e_font_from_gdk_name (const gchar *name)
 {
-	GdkFont *font;
+	EFont * font;
+	GdkFont *gdkfont;
 
-	font = gdk_fontset_load (name);
+	gdkfont = gdk_fontset_load (name);
+	font = e_font_from_gdk_font (gdkfont);
+	gdk_font_unref (gdkfont);
 
-	return (EFont *) font;
+	return font;
 }
 
 EFont *
-e_font_from_gdk_font (GdkFont *font)
+e_font_from_gdk_font (GdkFont *gdkfont)
 {
-	gdk_font_ref (font);
+	EFont *font;
+	XFontStruct *xfs;
+	Atom font_atom, atom;
+	Bool status;
+
+	gdk_font_ref (gdkfont);
+
+	/* Try to find iso-10646-1 encoded font with same name */
+
+	font_atom = gdk_atom_intern ("FONT", FALSE);
+	if (gdkfont->type == GDK_FONT_FONTSET) {
+		XFontStruct **font_structs;
+		gint num_fonts;
+		gchar **font_names;
+		num_fonts = XFontsOfFontSet (GDK_FONT_XFONT (gdkfont),
+					     &font_structs,
+					     &font_names);
+		status = XGetFontProperty (font_structs[0],
+					   font_atom,
+					   &atom);
+	} else {
+		status = XGetFontProperty (GDK_FONT_XFONT (gdkfont),
+					   font_atom,
+					   &atom);
+	}
+	if (status) {
+		gchar *name, *p;
+		gchar *newname;
+		GdkFont *newfont;
+		gint i;
+
+		name = gdk_atom_name (atom);
+		newname = alloca (strlen (name) + 12);
+		strcpy (newname, name);
+		p = newname;
+		for (i = 0; i < 13; i++) {
+			/* Skip hyphen */
+			while (*p && (*p != '-')) p++;
+			if (*p) p++;
+		}
+		g_snprintf (p, 12, "ISO10646-1");
+		newfont = gdk_font_load (newname);
+		if (newfont) {
+			gdk_font_unref (gdkfont);
+			gdkfont = newfont;
+		}
+		g_free (name);
+	}
+
+	font = g_new (EFont, 1);
+
+	xfs = GDK_FONT_XFONT (gdkfont);
+
+	font->refcount = 1;
+	font->font = gdkfont;
+	font->bytes = ((xfs->min_byte1 != 0) || (xfs->max_byte1 != 0)) ? 2 : 1;
+	font->to = e_uiconv_to_gdk_font (font->font);
+	font->from = e_uiconv_from_gdk_font (font->font);
+
+	return font;
 
-	return (EFont *) font;
 }
 
 void
 e_font_ref (EFont *font)
 {
-	gdk_font_ref (&font->font);
+	font->refcount++;
 }
 
 void
 e_font_unref (EFont *font)
 {
-	gdk_font_unref (&font->font);
+	font->refcount--;
+
+	if (font->refcount < 1) {
+		gdk_font_unref (font->font);
+		g_free (font);
+	}
 }
 
 gint
 e_font_ascent (EFont * font)
 {
-	return font->font.ascent;
+	return font->font->ascent;
 }
 
 gint
 e_font_descent (EFont * font)
 {
-	return font->font.descent;
+	return font->font->descent;
+}
+
+static gint
+e_font_to_native (EFont *font, gchar *native, gchar *utf, gint bytes)
+{
+	char *ib, *ob;
+	size_t ibl, obl;
+
+	ib = utf;
+	ibl = bytes;
+	ob = native;
+	obl = bytes * 4;
+
+	while (ibl > 0) {
+		unicode_iconv (font->to, (const char **) &ib, &ibl, &ob, &obl);
+		if (ibl > 0) {
+			gint len;
+			if ((*ib & 0x80) == 0x00) len = 1;
+			else if ((*ib &0xe0) == 0xc0) len = 2;
+			else if ((*ib &0xf0) == 0xe0) len = 3;
+			else if ((*ib &0xf80) == 0xf0) len = 4;
+			else {
+				g_warning ("Invalid UTF-8 sequence");
+				return ob - native;
+			}
+			ib += len;
+			ibl = bytes - (ib - utf);
+			if (ibl > bytes) ibl = 0;
+			if (font->bytes == 1) {
+				*ob++ = '_';
+				obl--;
+			} else {
+				*((guint16 *) ob) = '_';
+				ob += 2;
+				obl -= 2;
+			}
+		}
+	}
+
+	return ob - native;
 }
 
 void
 e_font_draw_utf8_text (GdkDrawable *drawable, EFont *font, EFontStyle style, GdkGC *gc, gint x, gint y, gchar *text, gint numbytes)
 {
-	guchar *iso;
-	gchar *p;
-	gint uni, len;
+	gchar *native;
+	gint native_bytes;
 
 	g_return_if_fail (drawable != NULL);
 	g_return_if_fail (font != NULL);
@@ -76,57 +187,55 @@ e_font_draw_utf8_text (GdkDrawable *drawable, EFont *font, EFontStyle style, Gdk
 
 	if (numbytes < 1) return;
 
-	iso = alloca (numbytes);
+	native = alloca (numbytes * 4);
 
-	for (len = 0, p = text; p != NULL && p < (text + numbytes); len++, p = unicode_next_utf8 (p)) {
-		unicode_get_utf8 (p, &uni);
-		if ((uni < ' ') || (uni > 255)) uni = ' ';
-		iso[len] = uni;
-	}
+	native_bytes = e_font_to_native (font, native, text, numbytes);
 
-	gdk_draw_text (drawable, &font->font, gc, x, y, iso, len);
+	gdk_draw_text (drawable, font->font, gc, x, y, native, native_bytes);
 
 	if (style & E_FONT_BOLD)
-		gdk_draw_text (drawable, &font->font, gc, x + 1, y, iso, len);
+		gdk_draw_text (drawable, font->font, gc, x + 1, y, native, native_bytes);
 }
 
 gint
 e_font_utf8_text_width (EFont *font, EFontStyle style, char *text, int numbytes)
 {
-	guchar *iso;
-	gchar *p;
-	gint uni, len;
+	gchar *native;
+	gint native_bytes;
+	gint width;
 
 	g_return_val_if_fail (font != NULL, 0);
 	g_return_val_if_fail (text != NULL, 0);
 
-	iso = alloca (numbytes);
+	if (numbytes < 1) return 0;
 
-	for (len = 0, p = text; p != NULL && p < (text + numbytes); len++, p = unicode_next_utf8 (p)) {
-		unicode_get_utf8 (p, &uni);
-		if ((uni < ' ') || (uni > 255)) uni = ' ';
-		iso[len] = uni;
-	}
+	native = alloca (numbytes * 4);
 
-	return gdk_text_width (&font->font, iso, len);
+	native_bytes = e_font_to_native (font, native, text, numbytes);
+
+	width = gdk_text_width (font->font, native, native_bytes);
+
+	return width;
 }
 
 gint
 e_font_utf8_char_width (EFont *font, EFontStyle style, char *text)
 {
-	unicode_char_t uni;
-	guchar iso;
+	gint len;
 
 	g_return_val_if_fail (font != NULL, 0);
 	g_return_val_if_fail (text != NULL, 0);
 
-	if (!unicode_get_utf8 (text, &uni)) return 0;
-
-	if ((uni < ' ') || (uni > 255)) uni = ' ';
-
-	iso = uni;
+	if ((*text & 0x80) == 0x00) len = 1;
+	else if ((*text &0xe0) == 0xc0) len = 2;
+	else if ((*text &0xf0) == 0xe0) len = 3;
+	else if ((*text &0xf80) == 0xf0) len = 4;
+	else {
+		g_warning ("Invalid UTF-8 sequence");
+		return 0;
+	}
 
-	return gdk_text_width (&font->font, &iso, 1);
+	return e_font_utf8_text_width (font, style, text, len);
 }
 
 static const gchar *
@@ -151,7 +260,8 @@ translate_encoding (const gchar *encoding)
 		g_hash_table_insert (eh, "iso8859-13", "iso-8859-13");
 		g_hash_table_insert (eh, "iso8859-14", "iso-8859-14");
 		g_hash_table_insert (eh, "iso8859-15", "iso-8859-15");
-		g_hash_table_insert (eh, "iso10646-1", "UCS2");
+		g_hash_table_insert (eh, "iso10646-1", "UTF-16");
+		g_hash_table_insert (eh, "koi8-r", "koi8-r");
 	}
 
 	strncpy (e, encoding, 64);
@@ -200,27 +310,9 @@ e_gdk_font_encoding (GdkFont *font)
 		if (*p) p++;
 	}
 
-#if 0
-	p = strchr (name, '-'); /* Foundry */
-	p = strchr (p + 1, '-'); /* Family */
-	p = strchr (p + 1, '-'); /* Weight */
-	p = strchr (p + 1, '-'); /* Slant */
-	p = strchr (p + 1, '-'); /* Set Width */
-	p = strchr (p + 1, '-'); /* Add Style */
-	p = strchr (p + 1, '-'); /* Pixel Size */
-	p = strchr (p + 1, '-'); /* Point Size */
-	p = strchr (p + 1, '-'); /* Resolution X */
-	p = strchr (p + 1, '-'); /* Resolution Y */
-	p = strchr (p + 1, '-'); /* Spacing */
-	p = strchr (p + 1, '-'); /* Average Width */
-	p = strchr (p + 1, '-'); /* Charset */
-
-	encoding = translate_encoding (p + 1);
-#else
 	if (!*p) return NULL;
 
 	encoding = translate_encoding (p);
-#endif
 
 	g_free (name);