A GitLab upgrade is scheduled for Monday, April 7, 2025. Service interruptions are expected in the morning. We'll keep you posted on the progress of the upgrade on the Mattermost channel: https://mattermost.inria.fr/devel/channels/gitlab. We recommend that you do not work on the platform until an announcement indicates that maintenance is complete.
#$s =~ s/([^\.][0-9\} ](?:\.(?: \.)*|[…。]))( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g; # STANDARD CASE
$s=~s/([^\.0-9][\} ](?:\.(?: \.)*|[…。]))( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 1
$s=~s/([^\.][0-9](?:\.(?: \.)+|[…。]))( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 2
$s=~s/^ ([^\.][0-9]\.)( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 3
$s=~s/([\.;\!\?…] [0-9]+ \.)( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 4
$s=~s/([^\s>0-9][0-9]+ (?:\.(?: \.)*|[…。]))( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 5
$s=~s/([^\.;\?\!…] [0-9]+ (?:\.(?: \.)*|[…。]))( +)($initialclass|$special_split)/$1<\/s>$2<s>$3/g;# STANDARD CASE 5
$s=~s/($maj$l+ $maj \.)<\/s> <s>($maj$l+ )/$1 $2/g;# oversegmentation correction by the standard case (case Prénom I. Nom)
$s=~s/\b((?:mr?s?|me?lle|[pd]r) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas Mr. et autres)
$s=~s/\b((?:mr?s?|me?lle|[pd]r|e \. +g) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas Mr. et autres)
if($langeq"pl"){
$s=~s/\b((?:np|m ?\. *in) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas np. et autres)
}elsif($langeq"ru"){
$s=~s/\b((?:Т \. +е) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas Т.е. et autres)
}elsif($langeq"es"){
$s=~s/\b((?:Ej) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas Ej. et autres)
}elsif($langeq"de"){
$s=~s/\b((?:bzw|z *\. *b|ggf) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas bzw. et autres)
}elsif($langeq"nl"){
$s=~s/\b((?:bv) \.)<\/s>( +)<s>((?:$maj \. )*$maj$l+ )/$1$2$3/gi;# oversegmentation correction by the standard case (cas bv. et autres)
}
$s=~s/( $strictmaj \.)<\/s> <s>($min(?:$l+| \.) )/$1 $2/g;# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (examples: A.F.D.S.F.G. est une entreprise connue. | We need to R.S.V.P. a.s.a.p., you know!)
$s=~s/( $min \. $min \.)<\/s> <s>($min(?:$l+| \.) )/$1 $2/g;# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (exemple: Please r.s.v.p. a.s.a.p. otherwise who know what will happen.)
$s=~s/( $min \.)<\/s> <s>($min \. )/$1 $2/g;# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (exemple: e. g.)
s=re.sub(rf"([^\.][0-9}} ](?:\.(?: \.)*|…))( +)({initialclass.pattern}|{special_split.pattern})",r"\1</s>\2<s>\3",s)# STANDARD CASE
s=re.sub(rf"({maj.pattern}{l.pattern}+ {maj.pattern} \.)</s> <s>({maj.pattern}{l.pattern}+ )",r"\1 \2",s)# oversegmentation correction by the standard case (case Prénom I. Nom)
<<<<<<<Updatedupstream
s=re.sub(rf"\b((?:mr?s?|me?lle|[pd]r) \.)</s>( +)<s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1\2\3",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas Mr. et autres)
=======
s=re.sub(rf"\b((?:mr?s?|me?lle|[pd]r) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas Mr. et autres)
iflang=="pl":
s=re.sub(rf"\b((?:np|m ?\. *in) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas np. et autres)
eliflang=="ru":
s=re.sub(rf"\b((?:Т \. +е) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas Т. е. et autres)
eliflang=="es":
s=re.sub(rf"\b((?:Ej) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction else if lang == "de":
s=re.sub(rf"\b((?:bzw|z *\. *b|ggf) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas bzw. et autres)
eliflang=="nl":
s=re.sub(rf"\b((?:bv) \.)</s> <s>((?:{maj.pattern} \. )*{maj.pattern}{l.pattern}+ )",r"\1 \2",s,flags=re.IGNORECASE)# oversegmentation correction by the standard case (cas bv. et autres)
>>>>>>>Stashedchanges
s=re.sub(rf"( {strictmaj.pattern} \.)</s> <s>({minus.pattern}(?:{l.pattern}+| \.) )",r"\1 \2",s)# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (examples: A.F.D.S.F.G. est une entreprise connue. | We need to R.S.V.P. a.s.a.p., you know!)
s=re.sub(rf"( {minus.pattern} \. {minus.pattern} \.)</s> <s>({minus.pattern}(?:{l.pattern}+| \.) )",r"\1 \2",s)# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (exemple: Please r.s.v.p. a.s.a.p. otherwise who know what will happen.)
s=re.sub(rf"( {minus.pattern} \.)</s> <s>({minus.pattern} \. )",r"\1 \2",s)# oversegmentation correction by the standard case in mode 'allow_sentence_initial_lowercase' (exemple: e. g.)
