Merge branch 'data' into data-UA

clarin-eric · Sep 16, 2023 · b11cc46 · b11cc46
2 parents 5545d5f + ec9a10f
commit b11cc46
Show file tree

Hide file tree

Showing 47 changed files with 1,872 additions and 115,214 deletions.
diff --git a/.github/actions/ParlaMintValidate/validate.sh b/.github/actions/ParlaMintValidate/validate.sh
@@ -15,6 +15,11 @@ for parla in $(jq -r '.[]' <<< $1 ); do
   echo "Cleaning old sample files [$parla]"
   rm -f ${DATADIR}/ParlaMint-$parla/ParlaMint-*.{txt,tsv,conllu,vert}
 
+  echo "::warning:: TMP check whether are taxonomies translated"
+  make translateTaxonomies-$parla | sed "s/^\(.*\)\(\berror\b\)/::error::\1\2/i" | tee $DIR/taxonomies.log
+  make initTaxonomies4translation-$parla
+  make validateTaxonomies-$parla | sed "s/^\(.*\)\(\berror\b\)/::error:: incomplete taxonomy translation \1\2/i" | tee $DIR/taxonomies.log
+
   if [ -f "${DATADIR}/ParlaMint-$parla/ParlaMint-$parla.xml" ] ; then
 
     ( Scripts/validate-parlamint.pl Schema ${DATADIR}/ParlaMint-$parla 2>&1 || echo "ERROR: validate-parlamint.pl exited with <> 0" ) \
@@ -55,6 +60,10 @@ for parla in $(jq -r '.[]' <<< $1 ); do
     FAIL=1
     echo "::error:: ParlaMint-$parla validation failed"
   fi
+
+  echo "::warning:: TMP restore taxonomy"
+  git checkout Corpora/Taxonomies/ParlaMint-taxonomy*
+  git checkout ${DATADIR}/ParlaMint-$parla/ParlaMint-taxonomy*
 done
 
 if [ $FAIL -eq 1 ] ; then

diff --git a/Corpora/Taxonomies/ParlaMint-taxonomy-NER.ana.xml b/Corpora/Taxonomies/ParlaMint-taxonomy-NER.ana.xml
@@ -25,6 +25,9 @@
       <term>Entidades Nombradas</term>
    </desc>
    <!--Corpus ParlaMint-EE is missing et translation of Named entities-->
+   <desc n="ParlaMint-FI" xml:lang="fi">
+      <term>nimetyt entiteetit</term>
+   </desc>
    <!--Corpus ParlaMint-FR is missing fr translation of Named entities-->
    <desc n="ParlaMint-ES-GA" xml:lang="gl">
       <term>Entidades nomeadas</term>
@@ -87,6 +90,9 @@
       <catDesc n="ParlaMint-EE" xml:lang="et">
          <term>isik</term>
       </catDesc>
+      <catDesc n="ParlaMint-FI" xml:lang="fi">
+         <term>henkilö</term>
+      </catDesc>
       <catDesc n="ParlaMint-FR" xml:lang="fr">
          <term>personne</term>
       </catDesc>
@@ -158,6 +164,9 @@
       <catDesc n="ParlaMint-EE" xml:lang="et">
          <term>koht</term>
       </catDesc>
+      <catDesc n="ParlaMint-FI" xml:lang="fi">
+         <term>paikka</term>
+      </catDesc>
       <catDesc n="ParlaMint-FR" xml:lang="fr">
          <term>lieu</term>
       </catDesc>
@@ -229,6 +238,9 @@
       <catDesc n="ParlaMint-EE" xml:lang="et">
          <term>organisatsioon</term>
       </catDesc>
+      <catDesc n="ParlaMint-FI" xml:lang="fi">
+         <term>organisaatio</term>
+      </catDesc>
       <catDesc n="ParlaMint-FR" xml:lang="fr">
          <term>organisation</term>
       </catDesc>
@@ -300,6 +312,9 @@
       <catDesc n="ParlaMint-EE" xml:lang="et">
          <term>muu</term>
       </catDesc>
+      <catDesc n="ParlaMint-FI" xml:lang="fi">
+         <term>Muu</term>
+      </catDesc>
       <catDesc n="ParlaMint-FR" xml:lang="fr">
          <term>divers</term>
       </catDesc>