Force fRTL for each item returned from ScriptItemize() to the bidi

2005-03-20  Tor Lillqvist  <tml@novell.com>

	* modules/basic/basic-win32.c (itemize_shape_and_place): Force
	fRTL for each item returned from ScriptItemize() to the bidi
	embedding parity for the whole run. Pango passes us a run that it
	correctly has analyzed should be of a certain direction, but
	ScriptItemize() still might split it into items, some of which
	have neutral direction. (#170931)
	(uniscribe_shape, text_is_simple): Use g_utf8_to_utf16().

	Merge from HEAD:

	* pango/pangowin32-private.h
	* pango/pangowin32.c (pango_win32_font_get_metrics): Cache metrics
	by language, like the other backends do. Don't try to get the
	width of the 0..9 digit string if the font doesn't cover digits.

	* modules/basic/Makefile.am (INCLUDES): Remove extraneous slash
	that confuses gcc on Win32.

6 files changed, 155 insertions, 78 deletions

diff --git a/ChangeLog b/ChangeLog
index 375e1cd2..a05da6bd 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,23 @@
+2005-03-20  Tor Lillqvist  <tml@novell.com>
+
+	* modules/basic/basic-win32.c (itemize_shape_and_place): Force
+	fRTL for each item returned from ScriptItemize() to the bidi
+	embedding parity for the whole run. Pango passes us a run that it
+	correctly has analyzed should be of a certain direction, but
+	ScriptItemize() still might split it into items, some of which
+	have neutral direction. (#170931)
+	(uniscribe_shape, text_is_simple): Use g_utf8_to_utf16().
+
+	Merge from HEAD:
+
+	* pango/pangowin32-private.h
+	* pango/pangowin32.c (pango_win32_font_get_metrics): Cache metrics
+	by language, like the other backends do. Don't try to get the
+	width of the 0..9 digit string if the font doesn't cover digits.
+
+	* modules/basic/Makefile.am (INCLUDES): Remove extraneous slash
+	that confuses gcc on Win32.
+
 Sat Mar  5 13:51:04 2005  Manish Singh  <yosh@gimp.org>
 
 	* pango/mini-fribidi/fribidi_types.[ch]: const correctness for
diff --git a/ChangeLog.pre-1-10 b/ChangeLog.pre-1-10
index 375e1cd2..a05da6bd 100644
--- a/ChangeLog.pre-1-10
+++ b/ChangeLog.pre-1-10
@@ -1,3 +1,23 @@
+2005-03-20  Tor Lillqvist  <tml@novell.com>
+
+	* modules/basic/basic-win32.c (itemize_shape_and_place): Force
+	fRTL for each item returned from ScriptItemize() to the bidi
+	embedding parity for the whole run. Pango passes us a run that it
+	correctly has analyzed should be of a certain direction, but
+	ScriptItemize() still might split it into items, some of which
+	have neutral direction. (#170931)
+	(uniscribe_shape, text_is_simple): Use g_utf8_to_utf16().
+
+	Merge from HEAD:
+
+	* pango/pangowin32-private.h
+	* pango/pangowin32.c (pango_win32_font_get_metrics): Cache metrics
+	by language, like the other backends do. Don't try to get the
+	width of the 0..9 digit string if the font doesn't cover digits.
+
+	* modules/basic/Makefile.am (INCLUDES): Remove extraneous slash
+	that confuses gcc on Win32.
+
 Sat Mar  5 13:51:04 2005  Manish Singh  <yosh@gimp.org>
 
 	* pango/mini-fribidi/fribidi_types.[ch]: const correctness for
diff --git a/modules/basic/Makefile.am b/modules/basic/Makefile.am
index 54736bf1..4b9a30a1 100644
--- a/modules/basic/Makefile.am
+++ b/modules/basic/Makefile.am
@@ -13,7 +13,7 @@ INCLUDES = 				\
 	-DG_DISABLE_DEPRECATED		\
 	$(PANGO_DEBUG_FLAGS)		\
 	-I$(top_srcdir)			\
-	-I$(top_srcdir)/pango/		\
+	-I$(top_srcdir)/pango		\
 	$(GLIB_CFLAGS)
 
 if PLATFORM_WIN32
diff --git a/modules/basic/basic-win32.c b/modules/basic/basic-win32.c
index 236532c6..83d89bbf 100644
--- a/modules/basic/basic-win32.c
+++ b/modules/basic/basic-win32.c
@@ -435,32 +435,23 @@ make_langid (PangoLanguage *lang)
 #define CASEN(t,p) if (pango_language_matches (lang, t)) return MAKELANGID (LANG_##p, SUBLANG_NEUTRAL)
 
   /* Languages that most probably don't affect Uniscribe have been
-   * left out. Presumably still many of those mentioned below don't
-   * have any effect either.
+   * left out. Uniscribe is documented to use
+   * SCRIPT_CONTROL::uDefaultLanguage only to select digit shapes, so
+   * just leave languages with own digits.
    */
 
-  CASEN ("sq", ALBANIAN);
   CASEN ("ar", ARABIC);
-  CASEN ("az", AZERI);
-  CASEN ("hy", ARMENIAN);
   CASEN ("as", ASSAMESE);
   CASEN ("az", AZERI);
-  CASEN ("eu", BASQUE);
-  CASEN ("be", BELARUSIAN);
   CASEN ("bn", BENGALI);
-  CASEN ("bg", BULGARIAN);
-  CASEN ("ca", CATALAN);
   CASE ("zh-tw", CHINESE, TRADITIONAL);
   CASE ("zh-cn", CHINESE, SIMPLIFIED);
   CASE ("zh-hk", CHINESE, HONGKONG);
   CASE ("zh-sg", CHINESE, SINGAPORE);
   CASE ("zh-mo", CHINESE, MACAU);
-  CASEN ("hr", CROATIAN);
-  CASE ("sr", SERBIAN, CYRILLIC);
   CASEN ("dib", DIVEHI);
   CASEN ("fa", FARSI);
   CASEN ("ka", GEORGIAN);
-  CASEN ("el", GREEK);
   CASEN ("gu", GUJARATI);
   CASEN ("he", HEBREW);
   CASEN ("hi", HINDI);
@@ -487,13 +478,10 @@ make_langid (PangoLanguage *lang)
   CASEN ("tt", TATAR);
   CASEN ("te", TELUGU);
   CASEN ("th", THAI);
-  CASEN ("tr", TURKISH);
-  CASEN ("uk", UKRAINIAN);
   CASE ("ur-pk", URDU, PAKISTAN);
   CASE ("ur-in", URDU, INDIA);
   CASEN ("ur", URDU);
   CASEN ("uz", UZBEK);
-  CASEN ("vi", VIETNAMESE);
 
 #undef CASE
 #undef CASEN
@@ -524,7 +512,8 @@ dump_glyphs_and_log_clusters (gboolean rtl,
     printf ("  log_clusters: ");
     for (j = 0; j < itemlen; j++)
       printf ("%d ", log_clusters[j]);
-    printf ("\n");    nclusters = 0;
+    printf ("\n");
+    nclusters = 0;
     for (j = 0; j < itemlen; j++)
       {
 	if (j == 0 || log_clusters[j-1] != log_clusters[j])
@@ -720,19 +709,24 @@ itemize_shape_and_place (PangoFont        *font,
   control.uDefaultLanguage = make_langid (analysis->language);
   state.uBidiLevel = analysis->level;
   
+#ifdef BASIC_WIN32_DEBUGGING
+  if (pango_win32_debug)
+    printf (G_STRLOC ": ScriptItemize: uDefaultLanguage:%04x uBidiLevel:%d\n",
+	    control.uDefaultLanguage, state.uBidiLevel);
+#endif
   if ((*script_itemize) (wtext, wlen, G_N_ELEMENTS (items), &control, NULL,
 			 items, &nitems))
     {
 #ifdef BASIC_WIN32_DEBUGGING
       if (pango_win32_debug)
-	printf ("pango-basic-win32: ScriptItemize failed\n");
+	printf ("ScriptItemize failed\n");
 #endif
       return FALSE;
     }
 
 #ifdef BASIC_WIN32_DEBUGGING
   if (pango_win32_debug)
-    printf (G_STRLOC ": ScriptItemize: %d items\n", nitems);
+    printf ("%d items:\n", nitems);
 #endif
 
   if (analysis->level % 2)
@@ -778,7 +772,9 @@ itemize_shape_and_place (PangoFont        *font,
 		items[item].iCharPos, items[item+1].iCharPos-1, itemlen);
 #endif
 
-      if ((*script_shape) (hdc, &script_cache[script], wtext + items[item].iCharPos, itemlen,
+      items[item].a.fRTL = analysis->level % 2;
+      if ((*script_shape) (hdc, &script_cache[script],
+			   wtext + items[item].iCharPos, itemlen,
 			   G_N_ELEMENTS (iglyphs),
 			   &items[item].a,
 			   iglyphs,
@@ -865,20 +861,18 @@ uniscribe_shape (PangoFont        *font,
 		 PangoGlyphString *glyphs)
 {
   wchar_t *wtext;
-  int wlen, i;
+  long wlen;
+  int i;
   gboolean retval = TRUE;
   HGDIOBJ old_font = NULL;
   HFONT hfont = NULL;
   LOGFONT *lf;
   SCRIPT_CACHE script_cache[100];
 
-  wtext = (wchar_t *) g_convert (text, length, "UTF-16LE", "UTF-8",
-				 NULL, &wlen, NULL);
+  wtext = (wchar_t *) g_utf8_to_utf16 (text, length, NULL, &wlen, NULL);
   if (wtext == NULL)
     return FALSE;
 
-  wlen /= 2;
-
   lf = pango_win32_font_logfont (font);
   hfont = pango_win32_font_cache_load (font_cache, lf);
   g_free (lf);
@@ -921,7 +915,6 @@ uniscribe_shape (PangoFont        *font,
 	  printf ("\n");
 	}
 #endif
-
     }
 
   g_free (wtext);
@@ -940,23 +933,20 @@ text_is_simple (const char *text,
 {
   gboolean retval;
   wchar_t *wtext;
-  int wlen;
+  long wlen;
 
-  wtext = (wchar_t *) g_convert (text, length, "UTF-16LE", "UTF-8",
-				 NULL, &wlen, NULL);
+  wtext = (wchar_t *) g_utf8_to_utf16 (text, length, NULL, &wlen, NULL);
 
   if (wtext == NULL)
     return TRUE;
 
-  wlen /= 2;
-
   retval = ((*script_is_complex) (wtext, wlen, SIC_COMPLEX) == S_FALSE);
 
   g_free (wtext);
 
 #ifdef BASIC_WIN32_DEBUGGING
   if (pango_win32_debug)
-    printf ("text_is_simple: %.*s (%d chars): %s\n",
+    printf ("text_is_simple: %.*s (%ld chars): %s\n",
 	    MIN (length, 10), text, wlen, retval ? "YES" : "NO");
 #endif
 
diff --git a/pango/pangowin32-private.h b/pango/pangowin32-private.h
index b1014214..e51f2d55 100644
--- a/pango/pangowin32-private.h
+++ b/pango/pangowin32-private.h
@@ -75,6 +75,8 @@ struct _PangoWin32Font
   LOGFONT logfont;
   int size;
 
+  GSList *metrics_by_lang;
+
   PangoFontMap *fontmap;
 
   /* Written by pango_win32_get_hfont: */
diff --git a/pango/pangowin32.c b/pango/pangowin32.c
index 0a02dbc2..62af6bdd 100644
--- a/pango/pangowin32.c
+++ b/pango/pangowin32.c
@@ -45,6 +45,14 @@ HDC pango_win32_hdc;
 OSVERSIONINFO pango_win32_os_version_info;
 gboolean pango_win32_debug = FALSE;
 
+typedef struct _PangoWin32MetricsInfo PangoWin32MetricsInfo;
+
+struct _PangoWin32MetricsInfo
+{
+  const char *sample_str;
+  PangoFontMetrics *metrics;
+};
+
 typedef struct _PangoWin32FontClass   PangoWin32FontClass;
 
 struct _PangoWin32FontClass
@@ -145,6 +153,8 @@ pango_win32_font_init (PangoWin32Font *win32font)
 {
   win32font->size = -1;
 
+  win32font->metrics_by_lang = NULL;
+
   win32font->glyph_info = g_hash_table_new_full (NULL, NULL, NULL, g_free);
 }
 
@@ -535,59 +545,84 @@ static PangoFontMetrics *
 pango_win32_font_get_metrics (PangoFont     *font,
 			      PangoLanguage *language)
 {
-  HFONT hfont;
-  TEXTMETRIC tm;
-  PangoLayout *layout;
-  PangoRectangle extents;
-  PangoContext *context;
-  PangoFontMetrics *metrics;
-
-  metrics = pango_font_metrics_new ();
-  
-  metrics->ascent = 0;
-  metrics->descent = 0;
-  metrics->approximate_digit_width = 0;
-  metrics->approximate_char_width = 0;
+  PangoWin32MetricsInfo *info = NULL; /* Quiet gcc */
+  PangoWin32Font *win32font = (PangoWin32Font *)font;
+  GSList *tmp_list;
+      
+  const char *sample_str = pango_language_get_sample_string (language);
   
-  hfont = pango_win32_get_hfont (font);
+  tmp_list = win32font->metrics_by_lang;
+  while (tmp_list)
+    {
+      info = tmp_list->data;
+      
+      if (info->sample_str == sample_str)    /* We _don't_ need strcmp */
+	break;
 
-  if (hfont != NULL)
+      tmp_list = tmp_list->next;
+    }
+
+  if (!tmp_list)
     {
-      PangoFontDescription *font_desc;
+      HFONT hfont;
+      PangoFontMetrics *metrics;
+      
+      info = g_new (PangoWin32MetricsInfo, 1);
+      win32font->metrics_by_lang = g_slist_prepend (win32font->metrics_by_lang, info);
+
+      info->sample_str = sample_str;
+      info->metrics = metrics = pango_font_metrics_new ();
+
+      hfont = pango_win32_get_hfont (font);
+      if (hfont != NULL)
+	{
+	  PangoCoverage *coverage;
+	  TEXTMETRIC tm;
 	    
-      SelectObject (pango_win32_hdc, hfont);
-      GetTextMetrics (pango_win32_hdc, &tm);
+	  SelectObject (pango_win32_hdc, hfont);
+	  GetTextMetrics (pango_win32_hdc, &tm);
 
-      metrics->ascent = tm.tmAscent * PANGO_SCALE;
-      metrics->descent = tm.tmDescent * PANGO_SCALE;
-      metrics->approximate_char_width = tm.tmAveCharWidth * PANGO_SCALE;
+	  metrics->ascent = tm.tmAscent * PANGO_SCALE;
+	  metrics->descent = tm.tmDescent * PANGO_SCALE;
+	  metrics->approximate_char_width = tm.tmAveCharWidth * PANGO_SCALE;
 
-      /* lovely copy&paste programming (from pangox.c) FIXME: This is
-       * sort of a sledgehammer solution, if we cached the results,
-       * like the other backends did, it wouldn't big a huge performance
-       * problem.  Get the avg. width of the chars in "0123456789"
-       */
-      context = pango_win32_get_context ();
-      pango_context_set_language (context, language);
-      font_desc = pango_font_describe (font);
-      pango_context_set_font_description (context, font_desc);
-      layout = pango_layout_new (context);
-      pango_layout_set_text (layout, "0123456789", -1);
-      metrics->approximate_digit_width = max_glyph_width (layout);
-
-      /* FIXME: Should get the real values from the TrueType font file */
-      metrics->underline_position = -2 * PANGO_SCALE;
-      metrics->underline_thickness = 1 * PANGO_SCALE;
-      metrics->strikethrough_thickness = metrics->underline_thickness;
-      /* Really really wild guess */
-      metrics->strikethrough_position = metrics->ascent / 3;
-
-      pango_font_description_free (font_desc);
-      g_object_unref (layout);
-      g_object_unref (context);
-    }
+	  coverage = pango_win32_font_get_coverage (font, language);
+	  if (pango_coverage_get (coverage, '0') != PANGO_COVERAGE_NONE &&
+	      pango_coverage_get (coverage, '9') != PANGO_COVERAGE_NONE)
+	    {
+	      PangoContext *context;
+	      PangoFontDescription *font_desc;
+	      PangoLayout *layout;
+	      
+	      /*  Get the average width of the chars in "0123456789" */
+	      context = pango_win32_get_context ();
+	      pango_context_set_language (context, language);
+	      font_desc = pango_font_describe (font);
+	      pango_context_set_font_description (context, font_desc);
+	      layout = pango_layout_new (context);
+	      pango_layout_set_text (layout, "0123456789", -1);
+
+	      metrics->approximate_digit_width = max_glyph_width (layout);
+
+	      pango_font_description_free (font_desc);
+	      g_object_unref (layout);
+	      g_object_unref (context);
+	    }
+	  else
+	    metrics->approximate_digit_width = metrics->approximate_char_width;
+
+	  pango_coverage_unref (coverage);
 
-  return metrics;
+	  /* FIXME: Should get the real values from the TrueType font file */
+	  metrics->underline_position = -2 * PANGO_SCALE;
+	  metrics->underline_thickness = 1 * PANGO_SCALE;
+	  metrics->strikethrough_thickness = metrics->underline_thickness;
+	  /* Really really wild guess */
+	  metrics->strikethrough_position = metrics->ascent / 3;
+	}
+    }
+      
+  return pango_font_metrics_ref (info->metrics);
 }
 
 
@@ -631,6 +666,13 @@ pango_win32_font_dispose (GObject *object)
 }
 
 static void
+free_metrics_info (PangoWin32MetricsInfo *info)
+{
+  pango_font_metrics_unref (info->metrics);
+  g_free (info);
+}
+
+static void
 pango_win32_font_finalize (GObject *object)
 {
   PangoWin32Font *win32font = (PangoWin32Font *)object;
@@ -639,6 +681,9 @@ pango_win32_font_finalize (GObject *object)
   if (win32font->hfont != NULL)
     pango_win32_font_cache_unload (cache, win32font->hfont);
 
+  g_slist_foreach (win32font->metrics_by_lang, (GFunc)free_metrics_info, NULL);
+  g_slist_free (win32font->metrics_by_lang);
+  
   if (win32font->win32face)
     pango_win32_font_entry_remove (win32font->win32face, PANGO_FONT (win32font));