This commit is contained in:
ldolse 2011-02-06 19:55:35 +08:00
parent 9088903f4c
commit 9b1ae4ba97

View File

@ -121,6 +121,7 @@ class TXTInput(InputFormatPlugin):
length = docanalysis.line_length(.5)
preprocessor = HeuristicProcessor(options, log=getattr(self, 'log', None))
txt = preprocessor.punctuation_unwrap(length, txt, 'txt')
txt = separate_paragraphs_single_line(txt)
if getattr(options, 'enable_heuristics', False) and getattr(options, 'dehyphenate', False):
docanalysis = DocAnalysis('txt', txt)