diff options
Diffstat (limited to 'corpus')
| -rw-r--r-- | corpus/support/token_list | 48 | ||||
| -rw-r--r-- | corpus/support/token_patterns | 1 |
2 files changed, 49 insertions, 0 deletions
diff --git a/corpus/support/token_list b/corpus/support/token_list index d38638cf..69dbfdc2 100644 --- a/corpus/support/token_list +++ b/corpus/support/token_list @@ -1,6 +1,54 @@ ##################### hyphenated words added by Fei since 3/7/05 ##X-ray +# Finnish +eaa. +ap. +arv. +ay. +eKr. +em. +engl. +esim. +fil. +lis. +fil. +maist. +fil.toht. +harv. +ilt. +jatk. +jKr. +jms. +jne. +joht. +klo +ko. +ks. +leht. +lv. +lyh. +mm. +mon. +nim. +nro. +ns. +nti. +os. +oy. +pj. +pnä. +puh. +pvm. +rva. +tms. +ts. +vars. +vrt. +ym. +yms. +yo. + # hindi abbreviation patterns जन. फर. diff --git a/corpus/support/token_patterns b/corpus/support/token_patterns index de64fb2a..e51e5e72 100644 --- a/corpus/support/token_patterns +++ b/corpus/support/token_patterns @@ -1,5 +1,6 @@ /^(al|el|ul|e)\-[a-z]+$/ /^((а|А)(ль|ш)|уль)-\p{Cyrillic}+$/ +/^(юл)-\p{Cyrillic}+$/ /^\p{Cyrillic}\.\p{Cyrillic}\.$/ /^(\d|\d\d|\d\d\d)\.$/ |