unicode: introduce code for UTF-8 normalization
authorOlaf Weber <olaf@sgi.com>
Thu, 25 Apr 2019 17:45:46 +0000 (13:45 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Thu, 25 Apr 2019 17:45:46 +0000 (13:45 -0400)
commit44594c2fbf42528001dfb1597d26adb40ba6d178
tree08f72b2f0b6413988dd8785f06468fbcc2d8e8b0
parent955405d1174eebcd1b89ab335f720adc27d52b67
unicode: introduce code for UTF-8 normalization

Supporting functions for UTF-8 normalization are in utf8norm.c with the
header utf8norm.h. Two normalization forms are supported: nfdi and
nfdicf.

  nfdi:
   - Apply unicode normalization form NFD.
   - Remove any Default_Ignorable_Code_Point.

  nfdicf:
   - Apply unicode normalization form NFD.
   - Remove any Default_Ignorable_Code_Point.
   - Apply a full casefold (C + F).

For the purposes of the code, a string is valid UTF-8 if:

 - The values encoded are 0x1..0x10FFFF.
 - The surrogate codepoints 0xD800..0xDFFFF are not encoded.
 - The shortest possible encoding is used for all values.

The supporting functions work on null-terminated strings (utf8 prefix)
and on length-limited strings (utf8n prefix).

From the original SGI patch and for conformity with coding standards,
the utf8data_t typedef was dropped, since it was just masking the struct
keyword.  On other occasions, namely utf8leaf_t and utf8trie_t, I
decided to keep it, since they are simple pointers to memory buffers,
and using uchars here wouldn't provide any more meaningful information.

From the original submission, we also converted from the compatibility
form to canonical.

Changes made by Gabriel:
  Rebase to Mainline
  Fix up checkpatch.pl warnings
  Drop typedefs
  move out of libxfs
  Convert from NFKD to NFD

Signed-off-by: Olaf Weber <olaf@sgi.com>
Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/unicode/Makefile
fs/unicode/utf8-norm.c [new file with mode: 0644]
fs/unicode/utf8n.h [new file with mode: 0644]