| commit | db0110475c0db5dc3cb1bb12de0b0c475880899e | |
| author | Christophe Rhodes <csr21@cantab.net> | |
| Sat, 18 May 2013 12:37:55 +0000 (18 13:37 +0100) | ||
| committer | Christophe Rhodes <csr21@cantab.net> | |
| Sat, 18 May 2013 19:04:05 +0000 (18 20:04 +0100) | ||
| tree | 5f293c9f7ebb2e32f7b4f1c7f4f0c79cae193263 | treesnapshot (tar.gz zip) |
| parent | d0895e12b8275044532ce9a6041b91043cd69d4a | commitdiff |
implement primary and canonical composition, and hence NFC/NFKC
Read in the non-algorithmically-specified composition exclusions from
Unicode's CompositionExclusions.txt file, and generate a hash table
using the concatenated 42 bits of code points. This is a bit of a
sucky hash-table key, particularly on 32-bit platforms; I have a plan
to reduce the key to 24 bits (using some auxiliary information in ucd)
but the advantage of getting this try in is...
... hook in NFC/NFKC into normalization tests, and check that tests
pass.
Read in the non-algorithmically-specified composition exclusions from
Unicode's CompositionExclusions.txt file, and generate a hash table
using the concatenated 42 bits of code points. This is a bit of a
sucky hash-table key, particularly on 32-bit platforms; I have a plan
to reduce the key to 24 bits (using some auxiliary information in ucd)
but the advantage of getting this try in is...
... hook in NFC/NFKC into normalization tests, and check that tests
pass.