New check: allah_ligature

Added to the Universal profile.
allah_ligature: "Check correct formation of allah ligature in presence of explicit tashkeel."

(issue #4727)

CHANGELOG.md

@@ -2,6 +2,10 @@ Below are the noteworthy changes from each release.
 A more detailed list of changes is available in the corresponding milestones for each release in the Github issue tracker (https://github.com/googlefonts/fontbakery/milestones?state=closed).
 
 ##  Upcoming release: 0.13.0 (2024-Sep-??)
+### New checks
+#### Added to the Universal profile
+  - **[allah_ligature]:** Check correct formation of allah ligature in presence of explicit tashkeel. (issue #4727)
+
 ### Changes to existing checks
 #### On the Universal profile
   - **[name/family_and_style_max_length"]:** Use nameID 16 (Typographic family name) to determine name length if it exists. (PR #4811)
@@ -135,6 +139,7 @@ A more detailed list of changes is available in the corresponding milestones for
   - **[com.google.fonts/check/case_mapping]:** Only check letters. As per David Corbett's suggestion, we now check for unicode category = "L*". (issue #4733)
   - **EXPERIMENTAL - [com.google.fonts/check/tabular_kerning]:** Revised check another time to remove even further false positives (issue #4640)
 
+### Changes to existing checks
 #### On the Outline profile
   - **[com.google.fonts/check/outline_direction]:** fixed an error where the outermost path was not correctly detected. (issue #4719)
   - **[com.google.fonts/check/outline_direction]:** fix wording in rational, it stated the opposite of what the check does. (issue #4743)

Lib/fontbakery/checks/arabic.py

@@ -123,3 +123,54 @@ def check_arabic_high_hamza(ttFont):
             " the same size the arabic letter hamza (U+0621),"
             " but a different glyph outline area was detected.",
         )
+
+
+@check(
+    id="allah_ligature",
+    rationale="""
+        Ensure that the allah ligature, if present in a font, is correctly
+        formed in the presence of manually placed tashkeel marks.
+    """,
+    proposal="https://github.com/fonttools/fontbakery/issues/4727",
+)
+def com_google_fonts_check_allah_ligature(ttFont):
+    """Ensure correct formation of allah ligature in the presence of tashkeel marks."""
+
+    cmap = ttFont["cmap"].getBestCmap()
+    allah_unicode = 0xFDF2
+
+    if allah_unicode not in cmap.keys():
+        yield SKIP, "Font does not contain the allah ligature."
+        return
+
+    allah_glyphname = cmap[allah_unicode]
+
+    from vharfbuzz import Vharfbuzz
+
+    vharfbuzz = Vharfbuzz(ttFont.reader.file.name)
+
+    def shape(text):
+        buf = vharfbuzz.shape(text)
+        return vharfbuzz.serialize_buf(buf, glyphsonly=True)
+
+    # Even though the font contains the allah ligature, it's generally not used
+    if shape("الله") != allah_glyphname:
+        yield PASS, (
+            "The allah ligature is present but not substituted for "
+            "plain 'allah' input string."
+        )
+        return
+
+    # The allah ligature is used, so we need to check if it's correctly formed
+    plain_allah = shape("الله")
+    allah_with_tashkeel = shape("اللَّهُ")
+
+    if plain_allah in allah_with_tashkeel.split("|"):
+        yield FAIL, Message(
+            "wrong-allah-with-tashkeel",
+            "The basic allah ligature (U+FDF2) is used in the presence of tashkeel marks. ",
+        )
+
+    # TODO:
+    # add check that checks whether the damma is placed on the heh, not the lam
+    # see IBM Plex Sans Arabic v1.004

Lib/fontbakery/profiles/adobefonts.py

@@ -43,6 +43,7 @@
         "opentype/varfont/family_axis_ranges",
         "opentype/varfont/ital_range",
         "opentype/vendor_id",
+        "allah_ligature",
         "alt_caron",
         "arabic_high_hamza",
         "arabic_spacing_symbols",

Lib/fontbakery/profiles/fontbureau.py

@@ -4,6 +4,7 @@
     "pending_review": [
         "opentype/weight_class_fvar",
         "opentype/slant_direction",
+        "allah_ligature",
         "cjk_not_enough_glyphs",
         "cmap/format_12",
         "color_cpal_brightness",

Lib/fontbakery/profiles/fontwerk.py

@@ -17,6 +17,7 @@
         "fontdata_namecheck",
     ],
     "pending_review": [
+        "allah_ligature",
         "typographic_family_name",
         "vtt_volt_data",  # very similar to vttclean, may be a good idea to merge them.
         #

Lib/fontbakery/profiles/microsoft.py

@@ -10,6 +10,7 @@
         "opentype/cff_ascii_strings",
         "opentype/slant_direction",
         "opentype/weight_class_fvar",
+        "allah_ligature",
         "cjk_not_enough_glyphs",
         "cmap/format_12",
         "color_cpal_brightness",

Lib/fontbakery/profiles/typenetwork.py

@@ -32,6 +32,7 @@
         "googlefonts/negative_advance_width",
         "googlefonts/STAT/axis_order",
         #
+        "allah_ligature",
         "caps_vertically_centered",  # Disabled: issue #4274
         "case_mapping",
         "cmap/format_12",

Lib/fontbakery/profiles/universal.py

@@ -21,6 +21,7 @@
             "ufo_unnecessary_fields",
         ],
         "Universal Profile Checks": [
+            "allah_ligature",
             "alt_caron",
             "arabic_high_hamza",
             "arabic_spacing_symbols",

tests/test_checks_arabic.py

@@ -0,0 +1,31 @@
+from fontTools.ttLib import TTFont
+
+from fontbakery.status import FAIL
+from fontbakery.codetesting import (
+    assert_PASS,
+    assert_SKIP,
+    assert_results_contain,
+    CheckTester,
+    TEST_FILE,
+)
+
+
+def test_check_allah_ligature():
+    """Ensure correct formation of allah ligature in the presence of tashkeel marks."""
+    check = CheckTester("com.google.fonts/check/allah_ligature")
+
+    # font has no allah ligature
+    ttFont = TTFont(TEST_FILE("ibmplexsans-vf/IBMPlexSansVar-Roman.ttf"))
+    assert_SKIP(check(ttFont))
+
+    # allah ligature is present but generally unused
+    ttFont = TTFont(TEST_FILE("fustat/Fustat[wght].ttf"))
+    assert_PASS(check(ttFont))
+
+    # classic mistake: tashkeel are placed on top of the allah ligature, colliding
+    ttFont = TTFont(TEST_FILE("ibmplexsansarabic_v1_004/IBMPlexSansArabic-Regular.ttf"))
+    assert_results_contain(check(ttFont), FAIL, "wrong-allah-with-tashkeel")
+
+    # ligature is present and generally used, and correctly formed with tashkeel
+    ttFont = TTFont(TEST_FILE("ibmplexsansarabic_v1_005/IBMPlexSansArabic-Regular.ttf"))
+    assert_PASS(check(ttFont))