Index: gc_core/js/text.js ================================================================== --- gc_core/js/text.js +++ gc_core/js/text.js @@ -7,11 +7,11 @@ "use strict"; var text = { - _zEndOfSentence: new RegExp ('[.?!:;…]+[»”’)]?[   ]+[»”’]?(?=[«"“‘–—   ]*[A-ZÀÂÉÈÊÎÔÇ])', "g"), + _zEndOfSentence: new RegExp ('[.?!…]+[»”’)]?[   ]+[»”’]?(?=[«"“‘–—   ]*[A-ZÀÂÉÈÊÎÔÇ])|[:;][   ]+', "g"), getSentenceBoundaries: function* (sText) { // generator: returns start and end of sentences found in let iStart = 0; let m; Index: gc_core/py/text.py ================================================================== --- gc_core/py/text.py +++ gc_core/py/text.py @@ -7,11 +7,11 @@ import re import textwrap from itertools import chain -_zEndOfSentence = re.compile(r'[.?!:;…]+[»”’)]?[   ]+[»”’]?(?=[«"“‘–—   ]*[A-ZÀÂÉÈÊÎÔÇ])') +_zEndOfSentence = re.compile(r'[.?!…]+[»”’)]?[   ]+[»”’]?(?=[«"“‘–—   ]*[A-ZÀÂÉÈÊÎÔÇ])|[:;][   ]+') def getSentenceBoundaries (sText): "generator: returns start and end of sentences found in " iStart = 0 for m in _zEndOfSentence.finditer(sText):