version 1.7.2.4, 2007/10/24 20:36:06
|
version 1.7.2.6, 2007/12/03 21:30:19
|
Line 26 separators=['']
|
Line 26 separators=['']
|
# kommas relevant for graphemes will not be deleted |
# kommas relevant for graphemes will not be deleted |
komma_exception="([^sStThH])," |
komma_exception="([^sStThH])," |
# grapheme boundaries |
# grapheme boundaries |
graphemeBounds="\{|\}|<|>|\(|\)|-|_|\#|,|\||\]|\[|\!|\?" |
#graphemeBounds="\{|\}|<|>|\(|\)|-|_|\#|,|\||\]|\[|\!|\?" |
|
graphemeBounds="\{|\}|<|>|-|_|\#|,|\]|\[|\!|\?|\"" |
# for words |
# for words |
wordBounds="<|>|\(|\)|_|\#|,|\||\]|\[|\!|\?" |
#wordBounds="<|>|\(|\)|_|\#|,|\||\]|\[|\!|\?" |
|
wordBounds="<|>|_|\#|,|\]|\[|\!|\?|\"" |
|
|
|
|
class cdliSplitter: |
class cdliSplitter: |