dictionaries: fixed hun_OCRFixReplaceList.xml RegEx rules

This commit is contained in:
Waldi Ravens 2016-06-30 18:48:33 +02:00
parent 7b598c85c5
commit 9cc9aa65a9

View File

@ -9,21 +9,17 @@
<WholeLines />
<RegularExpressions>
<!-- nagy I-l javítások -->
<RegEx find="[\x41-\x5a\x61-\x7a\xc1-\xfc](II)" replaceWith="ll" />
<RegEx find="(II)[\x61-\x7a\xe1-\xfc]" replaceWith="ll" />
<RegEx find="[\x61-\x7a\xe1-\xfc](I)" replaceWith="l" />
<RegEx find="[\x20](I)[^aeou\x41-\x5a\xc1-\xdc]" replaceWith="l" />
<RegEx find="\b(l)[bcdfghjklmnpqrstvwxz]" replaceWith="I" />
<RegEx find="\W(l)[bcdfghjklmnpqrstvwxz]" replaceWith="I" />
<RegEx find="[\x41-\x5a\xc1-\xdc](I)[\x61-\x7a\xe1-\xfc]" replaceWith="l" />
<RegEx find="[\x61-\x7a\xe1-\xfc][\-](I)[\x61-\x7a\xe1-\xfc]" replaceWith="l" />
<RegEx find="[\x41-\x5a\xc1-\xdc](I)[\-][\x41-\x5a\xc1-\xdc][\x61-\x7a\xe1-\xfc]" replaceWith="l" />
<RegEx find="^[AEÜÓ](I)[^\x41-\x5a\xc1-\xdc]" replaceWith="l" />
<RegEx find="\W[AEÜÓ](I)[^\x41-\x5a\xc1-\xdc]" replaceWith="l" />
<RegEx find="\b(I)[aáeéiíoóöuúüy\xf5\xfb]" replaceWith="l" />
<RegEx find="\W(I)[aáeéiíoóöuúüy\xf5\xfb]" replaceWith="l" />
<RegEx find="\b(II)" replaceWith="Il" />
<RegEx find="\b(ll)" replaceWith="Il" />
<RegEx find="[\xf5\xfb](I)" replaceWith="l" />
<RegEx find="([\x41-\x5a\x61-\x7a\xc1-\xfc])II" replaceWith="$1ll" />
<RegEx find="II([\x61-\x7a\xe1-\xfc])" replaceWith="ll$1" />
<RegEx find="([\x61-\x7a\xe1-\xfc])I" replaceWith="$1l" />
<RegEx find="([\x20])I([^aeou\x41-\x5a\xc1-\xdc])" replaceWith="$1l$2" />
<RegEx find="\bl([bcdfghjklmnpqrstvwxz])" replaceWith="I$1" />
<RegEx find="([\x41-\x5a\xc1-\xdc])I([\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
<RegEx find="([\x61-\x7a\xe1-\xfc][\-])I([\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
<RegEx find="([\x41-\x5a\xc1-\xdc])I([\-][\x41-\x5a\xc1-\xdc][\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
<RegEx find="\b([AEÜÓ])I([^\x41-\x5a\xc1-\xdc])" replaceWith="$1l$2" />
<RegEx find="\bI([aáeéiíoóöuúüy\xf5\xfb])" replaceWith="l$1" />
<RegEx find="\b(?:II|ll)" replaceWith="Il" />
<RegEx find="([\xf5\xfb])I" replaceWith="$1l" />
</RegularExpressions>
</OCRFixReplaceList>