GitLab

nfd (commit #b6d54b57 to master) is currently producing incorrect output in the following cases (on both cs-osl-08 and cs-osl-11).

cameron@cs-osl-11:~/parabix-devel/build12$ bin/nfd ag62218-9 | hexdump -C
00000000  d7 a9 d6 bc e0 b7 81 0a  d7 a9 d6 bc e0 b7 82 0a  |................|
00000010

The correct output is produced by uconv:

cameron@cs-osl-11:~/parabix-devel/build12$ uconv -x any-nfd ag62218-9 | hexdump -C
00000000  d7 a9 d6 bc d7 81 0a d7  a9 d6 bc d7 82 0a        |..............|
0000000e

The issue appears to be incorrect computatation of a character class. The following trace shows that the computed NFD_Basis[8-11] is D in two cases when it should be 5.

cameron@cs-osl-11:~/parabix-devel/build12$ bin/nfd ag62218-9 --illustrator-width=100
שּශ
שּෂ
             ByteStream | ........
         BasisBits[4-7] | EAA0EAA0
         BasisBits[0-3] | FCCAFCDA
                u8index | ..11..11
             U21[20-20] | 0000
             U21[16-19] | 0000
             U21[12-15] | F0F0
              U21[8-11] | B0B0
               U21[4-7] | 2020
               U21[0-3] | CADA
       Insertion_BixNum | 2020
             SpreadMask | 1..11..1
   ExpandedBasis[20-20] | 00000000
   ExpandedBasis[16-19] | 00000000
   ExpandedBasis[12-15] | F000F000
    ExpandedBasis[8-11] | B000B000
     ExpandedBasis[4-7] | 20002000
     ExpandedBasis[0-3] | C00AD00A
                 LV_LVT | 00000000
      LIndexBixNum[4-4] | 00000000
      LIndexBixNum[0-3] | 00000000
      VIndexBixNum[4-4] | 00000000
      VIndexBixNum[0-3] | 00000000
      TIndexBixNum[4-4] | 00000000
      TIndexBixNum[0-3] | 00000000
Hangul_NFD_Basis[20-20] | 00000000
Hangul_NFD_Basis[16-19] | 00000000
Hangul_NFD_Basis[12-15] | F000F000
 Hangul_NFD_Basis[8-11] | B000B000
  Hangul_NFD_Basis[4-7] | 20002000
  Hangul_NFD_Basis[0-3] | C00AD00A
       NFD_Basis[20-20] | 00000000
       NFD_Basis[16-19] | 00000000
       NFD_Basis[12-15] | 00000000
        NFD_Basis[8-11] | 55D055D0
         NFD_Basis[4-7] | EBC0EBC0
         NFD_Basis[0-3] | 9C1A9C2A
       OutputBasis[4-7] | DADBEB80DADBEB80
       OutputBasis[0-3] | 796C071A796C072A

ag62218-9