This commit is contained in:
Kovid Goyal 2022-01-26 08:47:15 +05:30
commit 800b191a63
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C

View File

@ -365,7 +365,7 @@ class HeuristicProcessor:
# define the pieces of the regex # define the pieces of the regex
# (?<!\&\w{4});) is a semicolon not part of an entity # (?<!\&\w{4});) is a semicolon not part of an entity
lookahead = "(?<=.{"+str(length)+r"}([a-zა-ჰäëïöüàèìòùáćéíĺóŕńśúýâêîôûçąężıãõñæøþðßěľščťžňďřů,:)\\IAß]|(?<!\&\w{4});))" lookahead = "(?<=.{"+str(length)+r"}([a-zა-ჰäëïöüàèìòùáćéíĺóŕńśúýźâêîôûçąężłıãõñæøþðßěľščťžňďřůёђєіїјљњћўџѣа-я,:)\\IAß]|(?<!\&\w{4});))"
em_en_lookahead = "(?<=.{"+str(length)+"}[\u2013\u2014])" em_en_lookahead = "(?<=.{"+str(length)+"}[\u2013\u2014])"
soft_hyphen = "\xad" soft_hyphen = "\xad"
line_ending = "\\s*(?P<style_close></(span|[iub])>)?\\s*(</(p|div)>)?" line_ending = "\\s*(?P<style_close></(span|[iub])>)?\\s*(</(p|div)>)?"