Updates in the build script to address acoli-repo#4 and acoli-repo#7

max-ionov · Apr 12, 2022 · b9bc488 · b9bc488
1 parent 2563003
commit b9bc488
Show file tree

Hide file tree

Showing 7 changed files with 54 additions and 14 deletions.
diff --git a/stable/apertium/build-fintan.sh b/stable/apertium/build-fintan.sh
@@ -11,7 +11,8 @@
 
 # Path to Fintan executable
 # TODO: download and install it if not set
-FINTAN=`~/Projects/LiODi/fintan-backend/run.sh`
+FINTAN=~/Projects/LiODi/fintan-backend/run.sh
+ARQ=arq
 
 #################
 # DO NOT CHANGE #
@@ -106,16 +107,26 @@ for file in langs/*; do
 		origfile=`echo $origfile | sed s/'\\'/'\\\/'/g;`
 	fi;
 
-	$FINTAN -c apertium-full.json -p $SRC $SRCLANG $TGTLANG $file;
+	CURDIR=`pwd`
+	FILEDIR=`dirname $file`
+	SRC=`cd $FILEDIR; git config --get remote.origin.url; cd $CURDIR`
+	SRC=${SRC/git@github.com:/https://github.com/}
+
+	$FINTAN -c fintan/apertium-arq.json -p $SRC $SRCLANG $TGTLANG $file;
 	mv src_noLexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$SRCLANG.ttl
 	mv tgt_noLexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$TGTLANG.ttl
 
 	mv src_Lexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$SRCLANG.ttl
 	mv tgt_Lexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$TGTLANG.ttl
 
 	mv trans.ttl $DIR/Apertium-$srclang-${tgtlang}_TranslationSet$SRCLANG-$TGTLANG.ttl
-	mv tiad.tsv $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
-	sed -i '1d' $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
+	#mv tiad.tsv $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
+	#sed -i '1d' $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
+	# Temporary until Fintan bug is not fixed
+	$ARQ --data=$DIR/Apertium-$srclang-${tgtlang}_Lexicon$SRCLANG.ttl \
+		--data=$DIR/Apertium-$srclang-${tgtlang}_Lexicon$TGTLANG.ttl \
+		--data=$DIR/Apertium-$srclang-${tgtlang}_TranslationSet$SRCLANG-$TGTLANG.ttl \
+		--query=ontolex2tsv.sparql --results=TSV | grep -v '^?' > $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
 
 	gzip $RELEASE/trans_$SRCLANG-$TGTLANG.tsv
 	cd $DIR;

diff --git a/stable/apertium/build-single-dict-fintan.sh b/stable/apertium/build-single-dict-fintan.sh
@@ -11,8 +11,12 @@ file=$1
 # TODO: download and install it if not set
 FINTAN=~/Projects/LiODi/fintan-backend/run.sh
 SRC=https://github.com/apertium/apertium-trunk.git;
+ARQ=arq
 
-echo $file
+CURDIR=`pwd`
+FILEDIR=`dirname $file`
+SRC=`cd $FILEDIR; git config --get remote.origin.url; cd $CURDIR`
+SRC=${SRC/git@github.com:/https://github.com/}
 
 RELEASE=apertium-rdf-`date +%F`;
 mkdir -p $RELEASE;
@@ -27,12 +31,17 @@ echo creating $DIR 1>&2;
 mkdir -p $DIR;
 origfile=`ls -l $file | sed s/'.*\-> *\.\.\/'//g;`;
 
-$FINTAN -c apertium-full.json -p $SRC $SRCLANG $TGTLANG $file;
+$FINTAN -c fintan/apertium-arq.json -p $SRC $SRCLANG $TGTLANG $file;
 mv src_noLexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$SRCLANG.ttl
 mv tgt_noLexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$TGTLANG.ttl
 
-mv src_Lexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$SRCLANG.ttl
-mv tgt_Lexinfo.ttl $DIR/Apertium-${srclang}-${tgtlang}_Lexicon$TGTLANG.ttl
+mv src_Lexinfo.ttl $DIR/Apertium-lexinfo-${srclang}-${tgtlang}_Lexicon$SRCLANG.ttl
+mv tgt_Lexinfo.ttl $DIR/Apertium-lexinfo-${srclang}-${tgtlang}_Lexicon$TGTLANG.ttl
 
 mv trans.ttl $DIR/Apertium-$srclang-${tgtlang}_TranslationSet$SRCLANG-$TGTLANG.ttl
-mv tiad.tsv $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
+# mv tiad.tsv $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
+# temporary fix until Fintan bug is not solved
+$ARQ --data=$DIR/Apertium-$srclang-${tgtlang}_Lexicon$SRCLANG.ttl \
+	--data=$DIR/Apertium-$srclang-${tgtlang}_Lexicon$TGTLANG.ttl \
+	--data=$DIR/Apertium-$srclang-${tgtlang}_TranslationSet$SRCLANG-$TGTLANG.ttl \
+	--query=ontolex2tsv.sparql --results=TSV | grep -v '^?' > $RELEASE/trans_$SRCLANG-$TGTLANG.tsv;
diff --git a/stable/apertium/build-single-dict.sh b/stable/apertium/build-single-dict.sh
@@ -11,7 +11,9 @@ SAXON=saxon;
 ARQ=arq;
 SRC=https://github.com/apertium/apertium-trunk.git;
 
-echo $file
+CURDIR=`pwd`
+FILEDIR=`dirname $file`
+SRC=`cd $FILEDIR; git config --get remote.origin.url; cd $CURDIR`
 
 RELEASE=apertium-rdf-`date +%F`;
 mkdir -p $RELEASE;

diff --git a/stable/apertium/build.sh b/stable/apertium/build.sh
@@ -121,6 +121,10 @@ RELEASE=apertium-rdf-`date +%F`;
 mkdir -p $RELEASE;
 
 for file in langs/*; do
+	CURDIR=`pwd`
+	FILEDIR=`dirname $file`
+	SRC=`cd $FILEDIR; git config --get remote.origin.url; cd $CURDIR`
+
 	srclang=`echo $file | sed s/'.*\/apertium-\([^\-\.]*\)-.*'/'\1'/g;`;
 	tgtlang=`echo $file | sed s/'.*\/apertium-[^\-\.]*-\([^\-\.]*\)\..*'/'\1'/g;`;
 	SRCLANG=`echo $srclang | tr [a-z] [A-Z]`; # sed s/'.*'/'\U&'/g;`; # OS X compatibility

diff --git a/stable/apertium/dix2src-ttl.xsl b/stable/apertium/dix2src-ttl.xsl
@@ -34,8 +34,15 @@
         <xsl:text> a lime:Lexicon</xsl:text>
         <xsl:text>;&#10; lime:language "</xsl:text>
         <xsl:value-of select="lower-case($LANG)"/>
-        <xsl:text>".&#10; </xsl:text>
-
+        <xsl:text>";&#10; </xsl:text>
+
+        <xsl:text> rdfs:label "Apertium </xsl:text>
+        <xsl:value-of select="upper-case($LANG)"/>
+        <xsl:text> Lexicon"@en;&#10; </xsl:text>
+        <xsl:text> rdfs:comment "This is an RDF OntoLex-lemon </xsl:text>
+        <xsl:value-of select="upper-case($LANG)"/>
+        <xsl:text> lexicon that comes from the original Apertium bilingual dictionaries."@en .&#10; </xsl:text>
+
         <xsl:for-each select="//e/p/l">
             <xsl:variable name="l">
                 <xsl:variable name="tmp">

diff --git a/stable/apertium/dix2tgt-ttl.xsl b/stable/apertium/dix2tgt-ttl.xsl
@@ -34,7 +34,14 @@
         <xsl:text> a lime:Lexicon</xsl:text>
         <xsl:text>;&#10; lime:language "</xsl:text>
         <xsl:value-of select="lower-case($LANG)"/>
-        <xsl:text>".&#10; </xsl:text>
+        <xsl:text>";&#10; </xsl:text>
+
+        <xsl:text> rdfs:label "Apertium </xsl:text>
+        <xsl:value-of select="upper-case($LANG)"/>
+        <xsl:text> Lexicon"@en;&#10; </xsl:text>
+        <xsl:text> rdfs:comment "This is an RDF OntoLex-lemon </xsl:text>
+        <xsl:value-of select="upper-case($LANG)"/>
+        <xsl:text> lexicon that comes from the original Apertium bilingual dictionaries."@en .&#10; </xsl:text>
 
         <xsl:for-each select="//e/p/r">
             <xsl:variable name="l">

diff --git a/stable/apertium/ontolex2tsv.sparql b/stable/apertium/ontolex2tsv.sparql
@@ -21,7 +21,7 @@ PREFIX ontolex: <http://www.w3.org/ns/lemon/ontolex#>
 PREFIX lime: <http://www.w3.org/ns/lemon/lime#> 
 PREFIX vartrans: <http://www.w3.org/ns/lemon/vartrans#> 
 
-SELECT ?srep ?slex ?ssense ?trans ?tsense ?tlex ?turi ?trep ?pos
+SELECT ?srep ?slex ?ssense ?trans ?tsense ?tlex ?trep ?pos
 WHERE {
 	?trans vartrans:source ?ssense; vartrans:target ?tsense.
 	?ssense ontolex:isSenseOf ?slex.