Strabon

changeset 466:c63d7397f50d

merge
author Kostis Kyzirakos <kkyzir@di.uoa.gr>
date Thu Jul 19 20:25:20 2012 +0300 (2012-07-19)
parents 515413720e57 ebb1e9fb543b
children 88b4d2c21fb0
files scripts/v2.2/endpoint
line diff
     1.1 --- a/scripts/v2.2/endpoint	Thu Jul 19 20:25:03 2012 +0300
     1.2 +++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
     1.3 @@ -1,234 +0,0 @@
     1.4 -#!/bin/bash
     1.5 -#
     1.6 -# Script for executing SPARQL queries and SPARQL Update queries 
     1.7 -# as well as storing RDF triples on a Strabon Endpoint.
     1.8 -#
     1.9 -# Author: Charalampos (Babis) Nikolaou <charnik@di.uoa.gr>
    1.10 -#
    1.11 -
    1.12 -# this command
    1.13 -CMD="$(basename ${0})"
    1.14 -
    1.15 -function help() {
    1.16 -	echo "Usage: ${CMD} [OPTIONS] COMMAND ENDPOINT ARGS"
    1.17 -	echo
    1.18 -	echo "Execute SPARQL and SPARQL Update queries as well as store RDF triples on a Strabon endpoint."
    1.19 -	echo
    1.20 -	echo "	COMMAND	 : one of query, queryfile, update, store, or help"
    1.21 -	echo "	ENDPOINT : the URL of the Strabon Endpoint (e.g., http://localhost:8080/StrabonEndpoint)"
    1.22 -	echo "	ARGS	 : arguments according to selected command"
    1.23 -	echo 
    1.24 -	echo "OPTIONS can be any of the following"
    1.25 -	echo "	-d	 : don't run, just print what shall be executed"
    1.26 -}
    1.27 -
    1.28 -function help_query() {
    1.29 -	echo "Usage: ${CMD} query ENDPOINT SPARQL_QUERY [RESULT_FORMAT]"
    1.30 -	echo
    1.31 -	echo "	ENDPOINT	: the URL of Strabon Endpoint (e.g., http://localhost:8080/StrabonEndpoint/)"
    1.32 -	echo "	SPARQL_QUERY	: the SPARQL query to execute or the alias name corresponding to a"
    1.33 -	echo "			  predefined query such as:"
    1.34 -	echo "				\`size': SELECT (count(*) as ?c) WHERE {?s ?p ?o}"
    1.35 -	echo "	RESULT_FORMAT	: the format of the result. Possible values are \`KMLMAP\, \`GEOJSON', "
    1.36 -	echo "			  \`HTML', \`KMZMAP', \`XML' (default), or \`KML'."
    1.37 -}
    1.38 -
    1.39 -function help_queryfile() {
    1.40 -        echo "Usage: ${CMD} queryfile ENDPOINT SPARQL_QUERY_FILE [RESULT_FORMAT]"
    1.41 -        echo
    1.42 -        echo "  ENDPOINT          : the URL of Strabon Endpoint (e.g., http://localhost:8080/StrabonEndpoint/)"
    1.43 -        echo "  SPARQL_QUERY_FILE : the file that contain the SPARQL query to execute"
    1.44 -        echo "  RESULT_FORMAT     : the format of the result. Possible values are \`KMLMAP\, \`GEOJSON', "
    1.45 -        echo "                      \`HTML', \`KMZMAP', \`XML' (default), or \`KML'."
    1.46 -}
    1.47 -
    1.48 -function help_update() {
    1.49 -	echo "Usage: ${CMD} update ENDPOINT SPARQL_QUERY"
    1.50 -	echo
    1.51 -	echo "	ENDPOINT     : the URL of Strabon Endpoint (e.g., http://localhost:8080/StrabonEndpoint/)"
    1.52 -	echo "	SPARQL_QUERY : the SPARQL update query to execute or the alias name corresponding to a"
    1.53 -	echo "		       predefined query such as:"
    1.54 -	echo "				\`clear': DELETE {?s ?p ?o} WHERE {?s ?p ?o}"
    1.55 -}
    1.56 -
    1.57 -function help_store() {
    1.58 -	echo "Usage: ${CMD} store ENDPOINT FORMAT -t TRIPLES|-u TRIPLES_URL"
    1.59 -	echo
    1.60 -	echo "	ENDPOINT    : the URL of Strabon Endpoint (e.g., http://localhost:8080/StrabonEndpoint/)"
    1.61 -	echo "	FORMAT	    : the RDF format of the input (one of RDF/XML, N-Triples, Turtle, N3, TriX, TriG, or BinaryRDF)"
    1.62 -	echo "	TRIPLES	    : the RDF triples to store"
    1.63 -	echo "	TRIPLES_URL : the URL containing the RDF triples to store"
    1.64 -}
    1.65 -
    1.66 -CURL_OPTS="-w HTTP_CODE='%{http_code}\n' -H \"Content-Type:application/x-www-form-urlencoded\" -H \"Accept:text/xml\""
    1.67 -
    1.68 -# if set to 1, then only the command to be executed is printed
    1.69 -DEBUG=0
    1.70 -
    1.71 -case "${1}" in
    1.72 -	-d)
    1.73 -		shift
    1.74 -		DEBUG=1
    1.75 -		;;
    1.76 -esac
    1.77 -
    1.78 -case "${1}" in
    1.79 -	help)
    1.80 -		shift
    1.81 -		if test $# -eq 1; then
    1.82 -			case "${1}" in
    1.83 -				query)
    1.84 -					help_query
    1.85 -					;;
    1.86 -				queryfile)
    1.87 -					help_queryfile
    1.88 -					;;
    1.89 -				update)
    1.90 -					help_update
    1.91 -					;;
    1.92 -				store)
    1.93 -					help_store
    1.94 -					;;
    1.95 -				*)
    1.96 -					help
    1.97 -					;;
    1.98 -			esac
    1.99 -			exit 1
   1.100 -		fi
   1.101 -		help
   1.102 -		exit 0
   1.103 -		;;
   1.104 -	query)
   1.105 -		shift
   1.106 -		if ! test $# -ge 2; then
   1.107 -			help_query
   1.108 -			exit 1
   1.109 -		fi
   1.110 -		URL="${1}/Query"
   1.111 -		QUERY="${2}"
   1.112 -
   1.113 -		shift
   1.114 -		shift
   1.115 -		# predefined queries
   1.116 -		case "${QUERY}" in
   1.117 -			size)
   1.118 -				QUERY="SELECT (count(*) as ?c) WHERE {?s ?p ?o}"
   1.119 -				;;
   1.120 -		esac
   1.121 -
   1.122 -		# set default format
   1.123 -		FORMAT="XML"
   1.124 -		if test $# -eq 1; then
   1.125 -			FORMAT="${1}"
   1.126 -		fi
   1.127 -
   1.128 -		case "${FORMAT}" in
   1.129 -			[Kk][Mm][Ll][Mm][Aa][Pp]|[Gg][Ee][Oo][Jj][Ss][Oo][Nn]|[Hh][Tt][Mm][Ll]|[Kk][mM][Zz][Mm][Aa][Pp]|[Xx][Mm][Ll]|[Kk][Mm][Ll])
   1.130 -				shift
   1.131 -				;;
   1.132 -			*)
   1.133 -				echo "${CMD}: unknown format \"${FORMAT}\"."
   1.134 -				echo "${CMD}: possible values are \`KMLMAP\, \`GEOJSON' \`HTML', \`KMZMAP', \`XML' (default), or \`KML'"
   1.135 -				exit 2
   1.136 -				;;
   1.137 -		esac
   1.138 -
   1.139 -		EXEC="curl ${CURL_OPTS} -d format='${FORMAT}' --data-urlencode SPARQLQuery='${QUERY}' ${URL}"
   1.140 -		;;
   1.141 -	queryfile)
   1.142 -		shift
   1.143 -		if ! test $# -ge 2; then
   1.144 -			help_queryfile
   1.145 -			exit 1
   1.146 -		fi
   1.147 -
   1.148 -		if [ ! -f ${2} ];
   1.149 -		then
   1.150 -			echo "${CMD}: File not found."
   1.151 -			exit 2
   1.152 -		fi
   1.153 -
   1.154 -		URL="${1}/Query"
   1.155 -		QUERY=`cat ${2} | sed 's/\"/\\\"/g'`
   1.156 -
   1.157 -		shift
   1.158 -		shift
   1.159 -		
   1.160 -		# set default format
   1.161 -		FORMAT="XML"
   1.162 -		if test $# -eq 1; then
   1.163 -			FORMAT="${1}"
   1.164 -		fi
   1.165 -
   1.166 -		case "${FORMAT}" in
   1.167 -			[Kk][Mm][Ll][Mm][Aa][Pp]|[Gg][Ee][Oo][Jj][Ss][Oo][Nn]|[Hh][Tt][Mm][Ll]|[Kk][mM][Zz][Mm][Aa][Pp]|[Xx][Mm][Ll]|[Kk][Mm][Ll])
   1.168 -				shift
   1.169 -				;;
   1.170 -			*)
   1.171 -				echo "${CMD}: unknown format \"${FORMAT}\"."
   1.172 -				echo "${CMD}: possible values are \`KMLMAP\, \`GEOJSON' \`HTML', \`KMZMAP', \`XML' (default), or \`KML'"
   1.173 -				exit 2
   1.174 -				;;
   1.175 -		esac
   1.176 -
   1.177 -		EXEC="curl ${CURL_OPTS} -d format='${FORMAT}' --data-urlencode SPARQLQuery='${QUERY}' ${URL}"
   1.178 -		;;
   1.179 -
   1.180 -	update)
   1.181 -		shift
   1.182 -		if ! test $# -eq 2; then
   1.183 -			help_update
   1.184 -			exit 1
   1.185 -		fi
   1.186 -		URL="${1}/Update"
   1.187 -		QUERY="${2}"
   1.188 -		
   1.189 -		# predefined queries
   1.190 -		case "${QUERY}" in
   1.191 -			clear)
   1.192 -				QUERY="DELETE {?s ?p ?o} WHERE {?s ?p ?o}"
   1.193 -				;;
   1.194 -		esac
   1.195 -
   1.196 -		EXEC="curl ${CURL_OPTS} --data-urlencode SPARQLQuery='${QUERY}' ${URL}"
   1.197 -		;;
   1.198 -	store)
   1.199 -		shift
   1.200 -		if ! test $# -eq 4; then
   1.201 -			help_store
   1.202 -			exit 1
   1.203 -		fi
   1.204 -		URL="${1}/Store"
   1.205 -		FORMAT="${2}"
   1.206 -		case "${3}" in
   1.207 -			-t)
   1.208 -				TRIPLES="${4}"
   1.209 -				EXEC="curl ${CURL_OPTS} -d format='${FORMAT}' --data-urlencode data='${TRIPLES}' ${URL}"
   1.210 -				;;
   1.211 -			-u)
   1.212 -				URL_TRIPLES="${4}"
   1.213 -				EXEC="curl ${CURL_OPTS} -d format='${FORMAT}' --data-urlencode url='${URL_TRIPLES}' -d fromurl='' ${URL}"
   1.214 -				;;
   1.215 -			 *)
   1.216 -				help_store
   1.217 -				exit 1
   1.218 -				;;
   1.219 -		esac
   1.220 -		;;
   1.221 -	*) 
   1.222 -		help
   1.223 -		echo
   1.224 -		echo "${CMD}: unknown command \"${1}\"."
   1.225 -		exit 1
   1.226 -		;;
   1.227 -esac
   1.228 -
   1.229 -# execute or debug
   1.230 -if test $DEBUG -eq 1; then
   1.231 -	echo "${CMD}: Debug is ON"
   1.232 -	echo "${CMD}: Printing command for execution"	
   1.233 -	echo "	$EXEC"
   1.234 -else
   1.235 -	eval ${EXEC}
   1.236 -fi
   1.237 -
     2.1 --- a/scripts/v2.2/runNoaRefinementChain.sh	Thu Jul 19 20:25:03 2012 +0300
     2.2 +++ b/scripts/v2.2/runNoaRefinementChain.sh	Thu Jul 19 20:25:20 2012 +0300
     2.3 @@ -18,10 +18,9 @@
     2.4  #dataDir="http://kk.di.uoa.gr/out_triples/"
     2.5  #dataDir="http://godel.di.uoa.gr/allhot/"
     2.6  dataDir="http://jose.di.uoa.gr/rdf/hotspots/20"
     2.7 -name="HMSG2_IR_039_s7_"
     2.8 -suffix=".hotspots.nt"
     2.9  
    2.10 -HOTSPOTS_URL="http://jose.di.uoa.gr/rdf/hotspots"
    2.11 +#HOTSPOTS_URL="http://jose.di.uoa.gr/rdf/hotspots"
    2.12 +HOTSPOTS_URL="http://jose.di.uoa.gr/rdf/hotspots/msg1"
    2.13  
    2.14  logFile="chain.log"
    2.15  #countWTime="/usr/bin/time -p   %e"
    2.16 @@ -63,7 +62,7 @@
    2.17  	#elif test -s /etc/SuSE-release ; then
    2.18  	#elif test -s /etc/gentoo-release ; then
    2.19  	elif test -s /etc/lsb-release ; then # Ubuntu
    2.20 -			tomcat="tomcat7"
    2.21 +			tomcat=$(ls -1 /etc/init.d/| grep tomcat | head -1)
    2.22  	elif test -s /etc/debian_version ; then
    2.23  			tomcat="tomcat"
    2.24  	fi
    2.25 @@ -126,18 +125,20 @@
    2.26  #echo "Continue?"
    2.27  #read a
    2.28  
    2.29 -#for y in 2008; do
    2.30 -for y in 2007 2008 2010 2011 ;do
    2.31 +#for y in 2007 2008 2010 2011 ;do
    2.32 +for y in 2012 ;do
    2.33  	# get hotpost URLS
    2.34 -	for hot in $(curl -s ${HOTSPOTS_URL}/${y}/ | grep -o '>HMSG2.*\.nt' | colrm 1 1); do
    2.35 +	for hot in $(curl -s ${HOTSPOTS_URL}/${y}/ | grep -o '>HMSG.*\.nt' | colrm 1 1); do
    2.36  		file="${HOTSPOTS_URL}/${y}/${hot}"
    2.37  
    2.38 +		time_status=$(echo ${hot} | egrep -o '[[:digit:]]{6}_[[:digit:]]{4}')
    2.39 +
    2.40  		# get time information for acquisition
    2.41 -		year=${y}
    2.42 -		month=$(expr substr ${hot} 19 2)
    2.43 -		day=$(expr substr ${hot} 21 2)
    2.44 -		time2=$(expr substr ${hot} 24 2)
    2.45 -		time2="${time2}:$(expr substr ${hot} 26 2)"
    2.46 +		year="20$(expr substr ${time_status} 1 2)"
    2.47 +		month=$(expr substr ${time_status} 3 2)
    2.48 +		day=$(expr substr ${time_status} 5 2)
    2.49 +		time2=$(expr substr ${time_status} 8 2)
    2.50 +		time2="${time2}:$(expr substr ${time_status} 10 2)"
    2.51  
    2.52  		# store file
    2.53  		echo -n "storing " $file; echo; echo;