New check: CFF top dict in ASCII range.

Checks if all strings in a font's CFF table top dict fit in the range of ASCII values.

EXPERIMENTAL - com.adobe.fonts/check/cff_ascii_strings
Added to the Universal profile.

(issue #4619)

CHANGELOG.md

@@ -3,6 +3,9 @@ A more detailed list of changes is available in the corresponding milestones for
 
 ## Upcoming release: 0.12.8 (2024-Jun-??)
 ### New checks
+#### Added to the OpenType profile
+  - **EXPERIMENTAL - [com.adobe.fonts/check/cff_ascii_strings]:** Checks if all strings in a font's CFF table top dict fit in the range of ASCII values (issue #4619)
+
 #### Added to the Universal profile
   - **EXPERIMENTAL - [com.google.fonts/check/gsub/smallcaps_before_ligatures]:** Ensure 'smcp' (small caps) lookups are defined before ligature lookups in the 'GSUB' table (issue #3020)
 

Lib/fontbakery/checks/opentype/cff.py

@@ -10,6 +10,7 @@ def __init__(self):
         self.glyphs_endchar_seac = []
         self.glyphs_exceed_max = []
         self.glyphs_recursion_errors = []
+        self.string_not_ascii = []
 
 
 def _get_subr_bias(count):
@@ -61,6 +62,14 @@ def _analyze_cff(analysis, top_dict, private_dict, fd_index=0):
     global_subrs = top_dict.GlobalSubrs
     gsubr_bias = _get_subr_bias(len(global_subrs))
 
+    if hasattr(top_dict, "rawDict"):
+        raw_dict = top_dict.rawDict
+        for key in ["Notice", "Copyright", "FontName", "FullName", "FamilyName"]:
+            for char in raw_dict.get(key, ""):
+                if ord(char) > 0x7F:
+                    analysis.string_not_ascii.append((key, raw_dict[key]))
+                    break
+
     if private_dict is not None and hasattr(private_dict, "Subrs"):
         subrs = private_dict.Subrs
         subr_bias = _get_subr_bias(len(subrs))
@@ -109,7 +118,11 @@ def cff_analysis(font):
         ttFont = TTFont(font.file)  # Use our own copy here since we are decompiling
 
     if "CFF " in ttFont:
-        cff = ttFont["CFF "].cff
+        try:
+            cff = ttFont["CFF "].cff
+        except UnicodeDecodeError:
+            analysis.string_not_ascii = None
+            return analysis
 
         for top_dict in cff.topDictIndex:
             if hasattr(top_dict, "FDArray"):
@@ -218,3 +231,35 @@ def com_adobe_fonts_check_cff_deprecated_operators(cff_analysis):
                 f'Glyph "{gn}" has deprecated use of "endchar"'
                 f" operator to build accented characters (seac).",
             )
+
+
+@check(
+    id="com.adobe.fonts/check/cff_ascii_strings",
+    conditions=["ttFont", "is_cff", "cff_analysis"],
+    rationale="""
+        All CFF Table top dict string chars should fit into the ASCII range.
+    """,
+    proposal="https://github.com/fonttools/fontbakery/issues/4619",
+    experimental="Since 2024/Jun/20",
+)
+def com_adobe_fonts_check_cff_ascii_strings(cff_analysis):
+    """Does the font's CFF table top dict strings fit into the ASCII range?"""
+
+    if cff_analysis.string_not_ascii is None:
+        yield FAIL, Message(
+            "cff-unable-to-decode",
+            "Unable to decode CFF table, possibly due to out"
+            " of ASCII range strings. Please check table strings.",
+        )
+    elif cff_analysis.string_not_ascii:
+        detailed_info = ""
+        for key, string in cff_analysis.string_not_ascii:
+            detailed_info += (
+                f"\n\n\t - {key}: {string.encode('latin-1').decode('utf-8')}"
+            )
+
+        yield FAIL, Message(
+            "cff-string-not-in-ascii-range",
+            f"The following CFF TopDict strings"
+            f" are not in the ASCII range: {detailed_info}",
+        )

Lib/fontbakery/profiles/adobefonts.py

@@ -15,6 +15,7 @@
             "com.adobe.fonts/check/cff2_call_depth",
             "com.adobe.fonts/check/cff_call_depth",
             "com.adobe.fonts/check/cff_deprecated_operators",
+            "com.adobe.fonts/check/cff_ascii_strings",
         ],
         "fontwerk": [
             "com.fontwerk/check/inconsistencies_between_fvar_stat",  # IS_OVERRIDDEN

Lib/fontbakery/profiles/opentype.py

@@ -23,9 +23,10 @@
             "com.google.fonts/check/varfont/wdth_valid_range",
             "com.google.fonts/check/varfont/stat_axis_record_for_each_axis",
             "com.google.fonts/check/loca/maxp_num_glyphs",
-            "com.adobe.fonts/check/cff2_call_depth",
+            "com.adobe.fonts/check/cff_ascii_strings",
             "com.adobe.fonts/check/cff_call_depth",
             "com.adobe.fonts/check/cff_deprecated_operators",
+            "com.adobe.fonts/check/cff2_call_depth",
             "com.google.fonts/check/font_version",
             "com.google.fonts/check/post_table_version",
             "com.google.fonts/check/monospace",

tests/checks/opentype/cff_test.py

@@ -1,3 +1,5 @@
+from fontTools.ttLib import TTFont
+
 from fontbakery.codetesting import (
     assert_PASS,
     assert_results_contain,
@@ -98,3 +100,41 @@ def test_check_cff_deprecated_operators():
             " to build accented characters (seac)."
         ),
     )
+
+
+def test_check_cff_strings():
+    check = CheckTester("com.adobe.fonts/check/cff_ascii_strings")
+
+    ttFont = TTFont(TEST_FILE("source-sans-pro/OTF/SourceSansPro-Regular.otf"))
+
+    # check that a healthy CFF font passes:
+    assert_PASS(check(ttFont))
+
+    # FIXME: The condition cff_analysis creates a new ttFont object
+    #        and, consequently, discards the modifications we made
+    #        here prior to invoking the check.
+    #
+    # # put an out of range char into FullName field:
+    # rawDict = ttFont["CFF "].cff.topDictIndex[0].rawDict
+    # rawDict["FullName"] = "S\u00F2urceSansPro-Regular"
+    # assert_results_contain(
+    #     check(ttFont),
+    #     FAIL,
+    #     "cff-string-not-in-ascii-range",
+    #     (
+    #         "The following CFF TopDict strings are not in the ASCII range:"
+    #         f"- FullName: {rawDict['FullName']}"
+    #     ),
+    # )
+
+    # Out-of-ascii-range char in the FontName field will cause decode issues:
+    ttFont = TTFont(TEST_FILE("unicode-decode-err/unicode-decode-err-cff.otf"))
+    assert_results_contain(
+        check(ttFont),
+        FAIL,
+        "cff-unable-to-decode",
+        (
+            "Unable to decode CFF table, possibly due to out "
+            "of ASCII range strings. Please check table strings."
+        ),
+    )