13 Commits

Author SHA1 Message Date
Jean-Francois Dockes
06cd2bfd87 unac: exclude Tamil, move to Unicode 14.0.0, modernize autoxx, fix C build 2022-09-24 09:14:51 +02:00
Jean-Francois Dockes
560041cab9 cleared out errant tabs 2020-05-30 15:54:49 +02:00
Jean-Francois Dockes
c1fad4afc7 Replaced pthread with std:: thread and mutex 2016-07-12 18:08:21 +02:00
Jean-Francois Dockes
b6eb3589ba do not unaccent Bengali characters (process like the Hindi ones) 2014-07-16 12:47:30 +02:00
medoc
698affcfc8 Dont strip diacritics from Hindi Devanagari characters, they are determinant to word meaning 2013-10-26 18:56:25 +02:00
Jean-Francois Dockes
913dffc597 added code for unac to perform pure case-folding 2012-08-27 12:40:57 +02:00
Jean-Francois Dockes
a4c17941b1 Added a configuration parameter to set specific unaccenting/lowercasing for some characters to be handled differently than would result from using the Unicode database. Exemple: "a with ring above" could be set to be preserved by a Swedish locutor 2012-04-09 12:42:23 +02:00
Jean-Francois Dockes
0d24b5620b Make unac suppress combining accents found in input. Input in decomposed form was previously not unaccented 2011-11-04 21:06:48 +01:00
Jean-Francois Dockes
424e4173ba threading cleanup: add mutex protection around moronic change to transcode. Add mutex to equiv issue in unac. Rename const strings everywhere to cstr_xx to ease future detection of potentially problematic static variables. Most probably close issue #65 2011-09-28 15:01:14 +02:00
dockes
0fc81d26b6 new unac approach for japanese: dont decompose at all 2009-01-06 18:40:41 +00:00
dockes
869d75ee03 use unicode 5.1.0 + dont unaccent katakana/hiragana. Main change in unicode is that letters ae and o with stroke dont decompose anymore into a+e and o+e we may actually want to restore this if it proves a problem 2008-12-18 11:04:47 +00:00
dockes
00b954c4ef implemented additional case-folding 2006-01-06 13:10:08 +00:00
dockes
b396d2c39f initial import 2006-01-06 13:08:12 +00:00