implement SID to glyph ID mapping with predefined Charset

Also fixes oss-fuzz 21769

src/hb-ot-cff1-table.cc

@@ -29,11 +29,24 @@
 #ifndef HB_NO_CFF
 
 #include "hb-draw.hh"
+#include "hb-algs.hh"
 #include "hb-ot-cff1-table.hh"
 #include "hb-cff1-interp-cs.hh"
 
 using namespace CFF;
 
+struct sid_to_gid_t
+{
+  uint16_t  sid;
+  uint8_t   gid;
+
+  int cmp (uint16_t a) const
+  {
+    if (a == sid) return 0;
+    return (a < sid) ? -1 : 1;
+  }
+};
+
 /* SID to code */
 static const uint8_t standard_encoding_to_code [] =
 {
@@ -105,6 +118,80 @@ static const uint16_t expert_subset_charset_to_sid [] =
   340,  341,  342,  343,  344,  345,  346
 };
 
+/* SID to glyph ID */
+static const sid_to_gid_t expert_charset_sid_to_gid [] =
+{
+    { 1, 1 },     { 13, 12 },   { 14, 13 },   { 15, 14 },
+    { 27, 26 },   { 28, 27 },   { 99, 15 },   { 109, 46 },
+    { 110, 47 },  { 150, 111 }, { 155, 101 }, { 158, 100 },
+    { 163, 102 }, { 164, 112 }, { 169, 113 }, { 229, 2 },
+    { 230, 3 },   { 231, 4 },   { 232, 5 },   { 233, 6 },
+    { 234, 7 },   { 235, 8 },   { 236, 9 },   { 237, 10 },
+    { 238, 11 },  { 239, 16 },  { 240, 17 },  { 241, 18 },
+    { 242, 19 },  { 243, 20 },  { 244, 21 },  { 245, 22 },
+    { 246, 23 },  { 247, 24 },  { 248, 25 },  { 249, 28 },
+    { 250, 29 },  { 251, 30 },  { 252, 31 },  { 253, 32 },
+    { 254, 33 },  { 255, 34 },  { 256, 35 },  { 257, 36 },
+    { 258, 37 },  { 259, 38 },  { 260, 39 },  { 261, 40 },
+    { 262, 41 },  { 263, 42 },  { 264, 43 },  { 265, 44 },
+    { 266, 45 },  { 267, 48 },  { 268, 49 },  { 269, 50 },
+    { 270, 51 },  { 271, 52 },  { 272, 53 },  { 273, 54 },
+    { 274, 55 },  { 275, 56 },  { 276, 57 },  { 277, 58 },
+    { 278, 59 },  { 279, 60 },  { 280, 61 },  { 281, 62 },
+    { 282, 63 },  { 283, 64 },  { 284, 65 },  { 285, 66 },
+    { 286, 67 },  { 287, 68 },  { 288, 69 },  { 289, 70 },
+    { 290, 71 },  { 291, 72 },  { 292, 73 },  { 293, 74 },
+    { 294, 75 },  { 295, 76 },  { 296, 77 },  { 297, 78 },
+    { 298, 79 },  { 299, 80 },  { 300, 81 },  { 301, 82 },
+    { 302, 83 },  { 303, 84 },  { 304, 85 },  { 305, 86 },
+    { 306, 87 },  { 307, 88 },  { 308, 89 },  { 309, 90 },
+    { 310, 91 },  { 311, 92 },  { 312, 93 },  { 313, 94 },
+    { 314, 95 },  { 315, 96 },  { 316, 97 },  { 317, 98 },
+    { 318, 99 },  { 319, 103 }, { 320, 104 }, { 321, 105 },
+    { 322, 106 }, { 323, 107 }, { 324, 108 }, { 325, 109 },
+    { 326, 110 }, { 327, 114 }, { 328, 115 }, { 329, 116 },
+    { 330, 117 }, { 331, 118 }, { 332, 119 }, { 333, 120 },
+    { 334, 121 }, { 335, 122 }, { 336, 123 }, { 337, 124 },
+    { 338, 125 }, { 339, 126 }, { 340, 127 }, { 341, 128 },
+    { 342, 129 }, { 343, 130 }, { 344, 131 }, { 345, 132 },
+    { 346, 133 }, { 347, 134 }, { 348, 135 }, { 349, 136 },
+    { 350, 137 }, { 351, 138 }, { 352, 139 }, { 353, 140 },
+    { 354, 141 }, { 355, 142 }, { 356, 143 }, { 357, 144 },
+    { 358, 145 }, { 359, 146 }, { 360, 147 }, { 361, 148 },
+    { 362, 149 }, { 363, 150 }, { 364, 151 }, { 365, 152 },
+    { 366, 153 }, { 367, 154 }, { 368, 155 }, { 369, 156 },
+    { 370, 157 }, { 371, 158 }, { 372, 159 }, { 373, 160 },
+    { 374, 161 }, { 375, 162 }, { 376, 163 }, { 377, 164 },
+    { 378, 165 }
+};
+
+/* SID to glyph ID */
+static const sid_to_gid_t expert_subset_charset_sid_to_gid [] =
+{
+  { 1, 1 },       { 13, 8 },      { 14, 9 },      { 15, 10 },
+  { 27, 22 },     { 28, 23 },     { 99, 11 },     { 109, 41 },
+  { 110, 42 },    { 150, 64 },    { 155, 55 },    { 158, 54 },
+  { 163, 56 },    { 164, 65 },    { 169, 66 },    { 231, 2 },
+  { 232, 3 },     { 235, 4 },     { 236, 5 },     { 237, 6 },
+  { 238, 7 },     { 239, 12 },    { 240, 13 },    { 241, 14 },
+  { 242, 15 },    { 243, 16 },    { 244, 17 },    { 245, 18 },
+  { 246, 19 },    { 247, 20 },    { 248, 21 },    { 249, 24 },
+  { 250, 25 },    { 251, 26 },    { 253, 27 },    { 254, 28 },
+  { 255, 29 },    { 256, 30 },    { 257, 31 },    { 258, 32 },
+  { 259, 33 },    { 260, 34 },    { 261, 35 },    { 262, 36 },
+  { 263, 37 },    { 264, 38 },    { 265, 39 },    { 266, 40 },
+  { 267, 43 },    { 268, 44 },    { 269, 45 },    { 270, 46 },
+  { 272, 47 },    { 300, 48 },    { 301, 49 },    { 302, 50 },
+  { 305, 51 },    { 314, 52 },    { 315, 53 },    { 320, 57 },
+  { 321, 58 },    { 322, 59 },    { 323, 60 },    { 324, 61 },
+  { 325, 62 },    { 326, 63 },    { 327, 67 },    { 328, 68 },
+  { 329, 69 },    { 330, 70 },    { 331, 71 },    { 332, 72 },
+  { 333, 73 },    { 334, 74 },    { 335, 75 },    { 336, 76 },
+  { 337, 77 },    { 338, 78 },    { 339, 79 },    { 340, 80 },
+  { 341, 81 },    { 342, 82 },    { 343, 83 },    { 344, 84 },
+  { 345, 85 },    { 346, 86 }
+};
+
 /* code to SID */
 static const uint8_t standard_encoding_to_sid [] =
 {
@@ -158,6 +245,18 @@ hb_codepoint_t OT::cff1::lookup_expert_subset_charset_for_sid (hb_codepoint_t gl
     return 0;
 }
 
+hb_codepoint_t OT::cff1::lookup_expert_charset_for_glyph (hb_codepoint_t sid)
+{
+  const auto  *pair = hb_bsearch (sid, expert_charset_sid_to_gid, ARRAY_LENGTH (expert_charset_sid_to_gid));
+  return pair ? pair->gid : 0;
+}
+
+hb_codepoint_t OT::cff1::lookup_expert_subset_charset_for_glyph (hb_codepoint_t sid)
+{
+  const auto  *pair = hb_bsearch (sid, expert_subset_charset_sid_to_gid, ARRAY_LENGTH (expert_subset_charset_sid_to_gid));
+  return pair ? pair->gid : 0;
+}
+
 hb_codepoint_t OT::cff1::lookup_standard_encoding_for_sid (hb_codepoint_t code)
 {
   if (code < ARRAY_LENGTH (standard_encoding_to_sid))

src/hb-ot-cff1-table.hh

@@ -1217,6 +1217,31 @@ struct cff1
       }
     }
 
+    hb_codepoint_t sid_to_glyph (hb_codepoint_t sid) const
+    {
+      if (charset != &Null(Charset))
+        return charset->get_glyph (sid, num_glyphs);
+      else
+      {
+        hb_codepoint_t glyph = 0;
+        switch (topDict.CharsetOffset)
+        {
+          case  ISOAdobeCharset:
+            if (sid <= 228 /*zcaron*/) glyph = sid;
+            break;
+          case  ExpertCharset:
+            glyph = lookup_expert_charset_for_glyph (sid);
+            break;
+          case  ExpertSubsetCharset:
+            glyph = lookup_expert_subset_charset_for_glyph (sid);
+            break;
+          default:
+            break;
+        }
+        return glyph;
+      }
+    }
+
     protected:
     hb_blob_t	       *blob;
     hb_sanitize_context_t   sc;
@@ -1312,7 +1337,7 @@ struct cff1
       gname_t	key = { hb_bytes_t (name, len), 0 };
       const gname_t *gname = glyph_names.bsearch (key);
       if (gname == nullptr) return false;
-      hb_codepoint_t gid = charset->get_glyph (gname->sid, num_glyphs);
+      hb_codepoint_t gid = sid_to_glyph (gname->sid);
       if (!gid && gname->sid) return false;
       *glyph = gid;
       return true;
@@ -1357,6 +1382,8 @@ struct cff1
   HB_INTERNAL static hb_codepoint_t lookup_expert_encoding_for_code (hb_codepoint_t sid);
   HB_INTERNAL static hb_codepoint_t lookup_expert_charset_for_sid (hb_codepoint_t glyph);
   HB_INTERNAL static hb_codepoint_t lookup_expert_subset_charset_for_sid (hb_codepoint_t glyph);
+  HB_INTERNAL static hb_codepoint_t lookup_expert_charset_for_glyph (hb_codepoint_t sid);
+  HB_INTERNAL static hb_codepoint_t lookup_expert_subset_charset_for_glyph (hb_codepoint_t sid);
   HB_INTERNAL static hb_codepoint_t lookup_standard_encoding_for_sid (hb_codepoint_t code);
 
   public:

test/api/test-ot-glyphname.c

@@ -67,6 +67,20 @@ test_non_standard_names (void)
   hb_face_destroy (face);
 }
 
+static void
+test_predef_charset_names (void)
+{
+  hb_face_t *face = hb_test_open_font_file ("fonts/cff1_expert.otf");
+  hb_font_t *font = hb_font_create (face);
+
+  test_one_glyph (font, 0,   ".notdef");
+  test_one_glyph (font, 29,  "centsuperior");
+  test_one_glyph (font, 86,  "commainferior");
+
+  hb_font_destroy (font);
+  hb_face_destroy (face);
+}
+
 int
 main (int argc, char **argv)
 {
@@ -74,6 +88,7 @@ main (int argc, char **argv)
 
   hb_test_add (test_standard_names);
   hb_test_add (test_non_standard_names);
+  hb_test_add (test_predef_charset_names);
 
   return hb_test_run();
 }