Merge branch 'arabic-format' into 'main'

Fix Arabic format chars

Closes #642

See merge request GNOME/pango!553

5 files changed, 114 insertions, 18 deletions

diff --git a/pango/itemize.c b/pango/itemize.c
index 1d71b5c6..6c07e4c2 100644
--- a/pango/itemize.c
+++ b/pango/itemize.c
@@ -507,8 +507,7 @@ itemize_state_init (ItemizeState               *state,
 
   if (!PANGO_GRAVITY_IS_VERTICAL (state->context->resolved_gravity))
     state->width_iter.end = state->end;
-  else
-  if (state->emoji_iter.is_emoji)
+  else if (state->emoji_iter.is_emoji)
     state->width_iter.end = MAX (state->width_iter.end, state->emoji_iter.end);
 
   update_end (state);
@@ -903,27 +902,32 @@ itemize_state_update_for_new_run (ItemizeState *state)
 }
 
 /* We don't want space characters to affect font selection; in general,
-* it's always wrong to select a font just to render a space.
-* We assume that all fonts have the ASCII space, and for other space
-* characters if they don't, HarfBuzz will compatibility-decompose them
-* to ASCII space...
-* See bugs #355987 and #701652.
-*
-* We don't want to change fonts just for variation selectors.
-* See bug #781123.
-*
-* Finally, don't change fonts for line or paragraph separators.
-*
-* Note that we want spaces to use the 'better' font, comparing
-* the font that is used before and after the space. This is handled
-* in itemize_state_add_character().
-*/
+ * it's always wrong to select a font just to render a space.
+ *
+ * We assume that all fonts have the ASCII space, and for other space
+ * characters if they don't, HarfBuzz will compatibility-decompose them
+ * to ASCII space...
+ * See bugs #355987 and #701652.
+ *
+ * We don't want to change fonts just for variation selectors.
+ * See bug #781123.
+ *
+ * We don't want to change fonts for default ignorables such as Cf chars.
+ * Note that Cf chars in the Arabic block are visible and need to have
+ * a font, so we exclude.
+ *
+ * Finally, don't change fonts for line or paragraph separators.
+ *
+ * Note that we want spaces to use the 'better' font, comparing
+ * the font that is used before and after the space. This is handled
+ * in itemize_state_add_character().
+ */
 static gboolean
 consider_as_space (gunichar wc)
 {
   GUnicodeType type = g_unichar_type (wc);
   return type == G_UNICODE_CONTROL ||
-         type == G_UNICODE_FORMAT ||
+         (type == G_UNICODE_FORMAT && !((wc >= 0x600 && wc <= 0x06ff) || wc == 0x70f || wc == 0x8e2)) ||
          type == G_UNICODE_SURROGATE ||
          type == G_UNICODE_LINE_SEPARATOR ||
          type == G_UNICODE_PARAGRAPH_SEPARATOR ||
diff --git a/tests/fonts/README b/tests/fonts/README
index fb33f8f2..9cc88ca7 100644
--- a/tests/fonts/README
+++ b/tests/fonts/README
@@ -3,5 +3,7 @@ test-layout tests.
 
 Cantarell-VF.otf: Cantarell
 DejaVSans.ttf: DejaVu Sans
+DejaVSansMono.ttf: DejaVu Sans Mono
 emoji-subset.ttf: subset of Noto Color Emoji
 droid-sans-subset.ttf: subset of Droid Sans Japanese
+amiri-06dd.ttf: subset of Amiri containing U+06dd
diff --git a/tests/fonts/amiri-06dd.ttf b/tests/fonts/amiri-06dd.ttf
new file mode 100644
index 00000000..e60637d1
--- /dev/null
+++ b/tests/fonts/amiri-06dd.ttf
diff --git a/tests/layouts/arabic-format.layout b/tests/layouts/arabic-format.layout
new file mode 100644
index 00000000..630a0f03
--- /dev/null
+++ b/tests/layouts/arabic-format.layout
@@ -0,0 +1,89 @@
+{
+  "context" : {
+    "font" : "serif 12",
+    "base-gravity" : "south",
+    "gravity-hint" : "natural",
+    "base-dir" : "weak-ltr",
+    "round-glyph-positions" : true,
+    "transform" : [
+      1,
+      0,
+      0,
+      1,
+      0,
+      0
+    ]
+  },
+  "comment" : "Test that font fallback works\nfor visible format characters\nin the Arabic block.",
+  "text" : "۝",
+  "font" : "Sans 32",
+  "output" : {
+    "is-wrapped" : false,
+    "is-ellipsized" : false,
+    "unknown-glyphs" : 0,
+    "width" : 56320,
+    "height" : 56320,
+    "log-attrs" : [
+      {
+        "char-break" : true,
+        "cursor-position" : true,
+        "sentence-boundary" : true,
+        "backspace-deletes-character" : true,
+        "word-boundary" : true
+      },
+      {
+        "line-break" : true,
+        "mandatory-break" : true,
+        "char-break" : true,
+        "white" : true,
+        "cursor-position" : true,
+        "sentence-boundary" : true,
+        "backspace-deletes-character" : true,
+        "word-boundary" : true
+      }
+    ],
+    "lines" : [
+      {
+        "start-index" : 0,
+        "length" : 2,
+        "paragraph-start" : true,
+        "direction" : "ltr",
+        "runs" : [
+          {
+            "offset" : 0,
+            "length" : 2,
+            "text" : "۝",
+            "bidi-level" : 2,
+            "gravity" : "south",
+            "language" : "en-us",
+            "script" : "common",
+            "font" : {
+              "description" : "Amiri 32",
+              "checksum" : "d5e6374130d6df09e66f80f0bbcb7262112584ebadb861f0a412a21cc742a5dd",
+              "matrix" : [
+                1,
+                -0,
+                -0,
+                1,
+                0,
+                0
+              ]
+            },
+            "flags" : 0,
+            "y-offset" : 0,
+            "start-x-offset" : 0,
+            "end-x-offset" : 0,
+            "glyphs" : [
+              {
+                "glyph" : 1,
+                "width" : 56320,
+                "is-cluster-start" : true,
+                "log-cluster" : 0
+              }
+            ]
+          }
+        ]
+      }
+    ]
+  }
+}
diff --git a/tests/meson.build b/tests/meson.build
index 54592668..79510fb6 100644
--- a/tests/meson.build
+++ b/tests/meson.build
@@ -121,6 +121,7 @@ installed_test_layouts_data = [
   'layouts/valid-22.layout',
   'layouts/tab-crash.layout',
   'layouts/tab-assert.layout',
+  'layouts/arabic-format.layout',
 ]
 
 test_markups_data = [