@@ -32,12 +32,16 @@ def unicode(s): return s
32
32
time_re = re .compile (br"(\d):(\d+):(\d+)" )
33
33
34
34
anticrash_res = {
35
- re .compile (br'\b(|\d+|\W+)(|un|anti|re)c(ae|\xe6)sur' , re .I ): br'\1\2seizur' ,
35
+ re .compile (br'\b(|\d+|\W+)? (|un|anti|re)c(ae|\xe6)sur' , re .I ): br'\1\2seizur' ,
36
36
re .compile (br"\b(|\d+|\W+)h'(r|v)[e]" , re .I ): br"\1h ' \2 e" ,
37
- # re.compile(r"\b(|\d+|\W+)wed[h]esday", re.I): r"\1wed hesday",
38
- re .compile (br'hesday' ): b' hesday' ,
39
- re .compile (br"\b(|\d+|\W+)tz[s]che" , re .I ): br"\1tz sche"
40
- }
37
+ re .compile (br"\b(\w+[bdflmnrvzqh])hes([bcdfgjklmnprtw]\w+)\b" , re .I ): br"\1 hes\2" ,
38
+ re .compile (br"(\d):(\d\d[snrt][tdh])" , re .I ): br"\1 \2" ,
39
+ re .compile (br"h'([bdfjkpstvx']+)'([rtv][aeiou]?)" , re .I ): br"h \1 \2" ,
40
+ re .compile (br"(re|un|non|anti)cosp" , re .I ): br"\1kosp" ,
41
+ re .compile (br"(anti|non|re|un)caesure" , re .I ): br"\1ceasure" ,
42
+ re .compile (br"(EUR[A-Z]+)(\d+)" , re .I ): br"\1 \2" ,
43
+ re .compile (br"\b(|\d+|\W+)?t+z[s]che" , re .I ): br"\1tz sche"
44
+ }
41
45
42
46
english_fixes = {
43
47
re .compile (r'(\w+)\.([a-zA-Z]+)' ): r'\1 dot \2' ,
0 commit comments