Newer
Older
# TODO maybe remove after testing
set -e
HELP="description:
marvin_extraction_run.sh and databus-release.sh take one argument, which is the extraction group
selects download.\$GROUP.properties and extraction.\$GROUP.properties from extractionConfig dir and uses \$GROUP as a path.
usage:
./marvin_extraction_run.sh {test|generic|mappings|wikidata|text}
"
##############
# setup paths
##############
ROOT="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
CONFIGDIR="$ROOT/extractionConfiguration"
DIEFDIR="$ROOT/marvin-extraction/extraction-framework"
LOGDIR="$ROOT/marvin-extraction/logs/$(date +%Y-%m-%d)" && mkdir -p $LOGDIR
EXTRACTIONBASEDIR="$ROOT/marvin-extraction/wikidumps" && mkdir -p $EXTRACTIONBASEDIR
DATABUSDIR="$ROOT/marvin-extraction/databus-maven-plugin"
##############
# functions
##############
# exception for generic, 1. spark, 2. as English is big and has to be run separately
>&2 ../run sparkextraction $CONFIGDIR/extraction.generic.properties;
>&2 ../run sparkextraction $CONFIGDIR/extraction.generic.en.properties;
else
# run for all
>&2 ../run extraction $CONFIGDIR/extraction.$GROUP.properties;
fi
cd $DIEFDIR/scripts;
echo "post-processing of $GROUP"
>&2 ../run ResolveTransitiveLinks $EXTRACTIONBASEDIR redirects redirects_transitive .ttl.bz2 @downloaded;
>&2 ../run MapObjectUris $EXTRACTIONBASEDIR redirects_transitive .ttl.bz2 mappingbased-objects-uncleaned _redirected .ttl.bz2 @downloaded;
>&2 ../run TypeConsistencyCheck type.consistency.check.properties;
elif [ "$GROUP" = "wikidata" ]
then
>&2 ../run ResolveTransitiveLinks $EXTRACTIONBASEDIR redirects transitive-redirects .ttl.bz2 wikidata
>&2 ../run MapObjectUris $EXTRACTIONBASEDIR transitive-redirects .ttl.bz2 mappingbased-objects-uncleaned,raw -redirected .ttl.bz2 wikidata
>&2 ../run TypeConsistencyCheck type.consistency.check.properties;
elif [ "$GROUP" = "generic" ]
then
>&2 ../run ResolveTransitiveLinks $EXTRACTIONBASEDIR redirects redirects_transitive .ttl.bz2 @downloaded;
>&2 ../run MapObjectUris $EXTRACTIONBASEDIR redirects_transitive .ttl.bz2 disambiguations,infobox-properties,page-links,persondata,topical-concepts _redirected .ttl.bz2 @downloaded;
elif [ "$GROUP" = "test" ]
then
>&2 ../run ResolveTransitiveLinks $EXTRACTIONBASEDIR redirects redirects_transitive .ttl.bz2 @downloaded;
>&2 ../run MapObjectUris $EXTRACTIONBASEDIR redirects_transitive .ttl.bz2 mappingbased-objects-uncleaned _redirected .ttl.bz2 @downloaded;
>&2 ../run TypeConsistencyCheckManual mappingbased-objects instance-types ro;
# log files from same day get overwritten, only latest is kept
for f in $(find $LOGDIR -type f ); do lbzip2 -f $f; done;
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
##########################
# Databus Mapping
##########################
# switch case for some language exceptions
mapLangToContVar() {
lang=$(echo "$1" | sed 's|wiki||g')
case "$lang" in
"bat_smg") echo "_lang=batsmg";;
"zh_min_nan") echo "_lang=nan";;
"zh_yue") echo "_lang=yue";;
"wikidata") echo "";;
"commons" ) echo "_commons";;
*) echo "_lang=$lang";;
esac
}
mapNamesToDatabus() {
case "$1" in
# generic
"article-templates-nested") echo "article-templates_nested";;
"citation-data") echo "citations_data";;
"citation-links") echo "citations_links";;
"commons-page-links") echo "commons-sameas-links";;
"page-ids") echo "page_ids";;
"page-length") echo "page_length";;
"page-links") echo "wikilinks";;
"article-categories") echo "categories_articles";;
"category-labels") echo "categories_labels";;
"skos-categories") echo "categories_skos";;
"revision-ids") echo "revisions_ids";;
"revision-uris") echo "revisions_uris";;
# mappings
"instance-types-transitive") echo "instance-types_transitive";;
"mappingbased-objects-disjoint-domain") echo "mappingbased-objects_disjointDomain";;
"mappingbased-objects-disjoint-range") echo "mappingbased-objects_disjointRange";;
*) echo "$1";;
esac
}
mapAndCopy() {
# iterate all .ttl.bz2 files
for path in $(find "$EXTRACTIONBASEDIR" -name "*.ttl.bz2"); do
# split filename
# how to use ${string##/*}
# https://www.tldp.org/LDP/abs/html/string-manipulation.html#Substring%20Removal#Substring Removal
file="${path##*/}"
version="${file#*-}"
version="${version%%-*}"
version="${version:0:4}.${version:4:2}.${version:6:2}"
lang="${file%%-*}"
extraction="${file#*-*-}"
extraction="${extraction%%.*}"
extension="${file#*.}"
# map names and languages
mapped="$(mapNamesToDatabus $extraction)"
contVars="$(mapLangToContVar $lang)"
if [[ "$mapped" == *"_"* ]]; then
contVars="${contVars}_${mapped#*_}"
fi
artifact="${mapped%%_*}"
targetFolder="$DATABUSDIR/dbpedia/$GROUP/$artifact/$version"
targetFile="$artifact$contVars.$extension"
# remove "_redirected"
targetFile=$(echo -n "$targetFile" | sed 's/_redirected//g' )
# copy
echo "< $path
> $targetFolder/$targetFile"
# TODO enable after testing
#cp -vn "$path" "$DATABUSMVNPOMDIR/$targetArVe/$targetFile"
done
}