Merge pull request #7023 from diomed/patch-2

Update srp_OCRFixReplaceList.xml
This commit is contained in:
Nikolaj Olsson 2023-06-19 14:54:29 -04:00 committed by GitHub
commit cb410ce428
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -10,7 +10,7 @@
<Word from="ćmo" to="ćemo" /> <Word from="ćmo" to="ćemo" />
<Word from="ćš" to="ćeš" /> <Word from="ćš" to="ćeš" />
<Word from="ćte" to="ćete" /> <Word from="ćte" to="ćete" />
<Word from="djete" to="dijete" /> <Word from="djete" to="dete" />
<Word from="hey" to="hej" /> <Word from="hey" to="hej" />
<Word from="Hey" to="Hej" /> <Word from="Hey" to="Hej" />
<Word from="htjeo" to="htio" /> <Word from="htjeo" to="htio" />
@ -39,17 +39,17 @@
<Word from="samnom" to="sa mnom" /> <Word from="samnom" to="sa mnom" />
<Word from="Samnom" to="Sa mnom" /> <Word from="Samnom" to="Sa mnom" />
<Word from="smjeo" to="smio" /> <Word from="smjeo" to="smio" />
<Word from="umijesto" to="umjesto" /> <Word from="umijesto" to="ujesto" />
<Word from="Umijesto" to="Umjesto" /> <Word from="Umijesto" to="Umesto" />
<Word from="uopče" to="uopće" /> <Word from="uopče" to="uopće" />
<Word from="Uopče" to="Uopće" /> <Word from="Uopče" to="Uopće" />
<Word from="uspiješan" to="uspješan" /> <Word from="uspiješan" to="uspešan" />
<Word from="uvjek" to="uvijek" /> <Word from="uvjek" to="uvek" />
<Word from="Uvjek" to="Uvijek" /> <Word from="Uvjek" to="Uvek" />
<Word from="valda" to="valjda" /> <Word from="valda" to="valjda" />
<Word from="zamnom" to="za mnom" /> <Word from="zamnom" to="za mnom" />
<Word from="Zamnom" to="Za mnom" /> <Word from="Zamnom" to="Za mnom" />
<Word from="želila" to="željela" /> <Word from="želila" to="želela" />
</WholeWords> </WholeWords>
<PartialWordsAlways /> <PartialWordsAlways />
<PartialWords> <PartialWords>
@ -119,7 +119,7 @@
<RegEx find="Ĺ˝" replaceWith="Ž" /> <RegEx find="Ĺ˝" replaceWith="Ž" />
<RegEx find="Ž" replaceWith="Ž" /> <RegEx find="Ž" replaceWith="Ž" />
<RegEx find="đž" replaceWith="dž" /> <RegEx find="đž" replaceWith="dž" />
<RegEx find="ajsmiješnij" replaceWith="ajsmješnij" /> <RegEx find="ajsmiješnij" replaceWith="ajsmešnij" />
<RegEx find="boži[čć]([aeiu]|em|ima)?\b" replaceWith="Božić$1" /> <RegEx find="boži[čć]([aeiu]|em|ima)?\b" replaceWith="Božić$1" />
<RegEx find=" g-dine\.$" replaceWith=" gospodine." /> <RegEx find=" g-dine\.$" replaceWith=" gospodine." />
<RegEx find=" g-dine +(?=[A-ZČĐŠŽ])" replaceWith=" g. " /> <RegEx find=" g-dine +(?=[A-ZČĐŠŽ])" replaceWith=" g. " />
@ -138,24 +138,24 @@
<RegEx find="\b([nN])eč([ue]š?|emo|ete)\b" replaceWith="$1eć$2" /> <RegEx find="\b([nN])eč([ue]š?|emo|ete)\b" replaceWith="$1eć$2" />
<RegEx find="emo[zž]e" replaceWith="e može" /> <RegEx find="emo[zž]e" replaceWith="e može" />
<RegEx find="\b([nN])ezna([šm]o?|t[ei]|ju|jući|vši)?\b" replaceWith="$1e zna$2" /> <RegEx find="\b([nN])ezna([šm]o?|t[ei]|ju|jući|vši)?\b" replaceWith="$1e zna$2" />
<RegEx find="najcijenjen" replaceWith="najcjenjen" /> <RegEx find="najcijenjen" replaceWith="najcenjen" />
<RegEx find="N[jJ]u Jork" replaceWith="Njujork" /> <RegEx find="N[jJ]u Jork" replaceWith="Njujork" />
<RegEx find="([oO])d([kp])" replaceWith="$1t$2" /> <RegEx find="([oO])d([kp])" replaceWith="$1t$2" />
<RegEx find="ružij" replaceWith="ružj" /> <RegEx find="ružij" replaceWith="ružj" />
<RegEx find="([oO])sječa" replaceWith="$1sjeća" /> <RegEx find="([oO])sječa" replaceWith="$1seća" />
<RegEx find="([pPdD])onje([lt])" replaceWith="$1onije$2" /> <RegEx find="([pPdD])onje([lt])" replaceWith="$1onije$2" />
<RegEx find="([pP])objedi([mšto])" replaceWith="$1obijedi$2" /> <RegEx find="([pP])objedi([mšto])" replaceWith="$1obijedi$2" />
<RegEx find="ed([ph])" replaceWith="et$1" /> <RegEx find="ed([ph])" replaceWith="et$1" />
<RegEx find="rimjeti" replaceWith="rimijeti" /> <RegEx find="rimjeti" replaceWith="rimijeti" />
<RegEx find="romjeni([mštol])" replaceWith="romijeni$1" /> <RegEx find="romjeni([mštol])" replaceWith="romijeni$1" />
<RegEx find="azumijeć" replaceWith="azumjeć" /> <RegEx find="azumijeć" replaceWith="azumeć" />
<RegEx find="([Cc])jepljen" replaceWith="$1ijepljen" /> <RegEx find="([Cc])jepljen" replaceWith="$1epljen" />
<RegEx find="rimjenjen" replaceWith="rimijenjen" /> <RegEx find="rimjenjen" replaceWith="rimenjen" />
<RegEx find="([^d])rješit" replaceWith="$1riješit" /> <RegEx find="([^d])rješit" replaceWith="$1riješit" />
<RegEx find="lijede[čć]([aeiu]|e[mg])" replaceWith="ljedeć$1" /> <RegEx find="lijede[čć]([aeiu]|e[mg])" replaceWith="ledeć$1" />
<RegEx find="([sS])mješno" replaceWith="$1miješno" /> <RegEx find="([sS])mješno" replaceWith="$1mešno" />
<RegEx find="spijeh" replaceWith="spjeh" /> <RegEx find="spijeh" replaceWith="speh" />
<RegEx find="spiješn" replaceWith="spješn" /> <RegEx find="spiješn" replaceWith="spešn" />
<RegEx find="\b([vV])eč([aiu]|[ei][mg]|ih|ima|in[iu]|uom|o[mj])?\b" replaceWith="$1eć$2" /> <RegEx find="\b([vV])eč([aiu]|[ei][mg]|ih|ima|in[iu]|uom|o[mj])?\b" replaceWith="$1eć$2" />
<RegEx find="([zZ])ahtjeva([ojlmšt])" replaceWith="$1ahtijeva$2" /> <RegEx find="([zZ])ahtjeva([ojlmšt])" replaceWith="$1ahtijeva$2" />
<RegEx find="([ks]ao)\.:" replaceWith="$1:" /> <RegEx find="([ks]ao)\.:" replaceWith="$1:" />
@ -266,4 +266,4 @@
<!-- <RegEx find="^\.{3}([a-zčđšž&quot;&lt;])" replaceWith="$1" /> --> <!-- <RegEx find="^\.{3}([a-zčđšž&quot;&lt;])" replaceWith="$1" /> -->
<!-- <RegEx find=" +([.?!])" replaceWith="$1" /> --> <!-- <RegEx find=" +([.?!])" replaceWith="$1" /> -->
</RegularExpressions> </RegularExpressions>
</OCRFixReplaceList> </OCRFixReplaceList>