These patches add support for utf8 and seem to improve language detection accuracy, especially for non-latin scripts.
43 lines
643 B
Text
43 lines
643 B
Text
--- /dev/null 2011-02-04 21:03:13 +0000
|
|
+++ src/libtextcat.map 2011-02-04 20:49:38 +0000
|
|
@@ -0,0 +1,40 @@
|
|
+{
|
|
+ global:
|
|
+ charcopy
|
|
+ issame
|
|
+ nextcharstart
|
|
+ utfstrlen
|
|
+ wgmempool_Done
|
|
+ wgmempool_Init
|
|
+ wgmempool_Reset
|
|
+ wgmempool_alloc
|
|
+ wgmempool_getline
|
|
+ wgmempool_strdup
|
|
+ special_textcat_Init
|
|
+ textcat_Classify
|
|
+ textcat_Done
|
|
+ textcat_Init
|
|
+ textcat_Version
|
|
+ fp_Compare
|
|
+ fp_Create
|
|
+ fp_Debug
|
|
+ fp_Done
|
|
+ fp_Init
|
|
+ fp_Name
|
|
+ fp_Print
|
|
+ fp_Read
|
|
+ heapextract
|
|
+ wg_calloc
|
|
+ wg_free
|
|
+ wg_getline
|
|
+ wg_malloc
|
|
+ wg_split
|
|
+ wg_strdup
|
|
+ wg_strgmov
|
|
+ wg_trim
|
|
+ wg_zalloc
|
|
+ wgmem_error
|
|
+
|
|
+ local:
|
|
+ *;
|
|
+}
|