Laboratoire ICAR » Plateforme TXM

// Copyright © 2016 ENS de Lyon, CNRS, University of Franche-Comté

// Licensed under the terms of the GNU General Public License (http://www.gnu.org/licenses)

// @author mdecorde

// @author sheiden

// STANDARD DECLARATIONS

package org.txm.macro.analec.exploit

import org.apache.commons.lang.StringUtils

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.Toolbox

import org.txm.macro.analec.AnalecUtils;

import org.txm.rcp.swt.widget.parameters.*

import org.txm.analec.*

import org.txm.searchengine.cqp.AbstractCqiClient

import org.txm.searchengine.cqp.corpus.*

import visuAnalec.donnees.Structure

import visuAnalec.elements.*

def scriptName = this.class.getSimpleName()

def selection = []

for (def s : corpusViewSelections) {

        if (s instanceof Corpus) selection << s

        else if (s instanceof Partition) selection.addAll(s.getParts())

}

if (selection.size() == 0) {

        println "** $scriptName: please select a Corpus or a Partition in the Corpus view: "+corpusViewSelections

        return false

}

// BEGINNING OF PARAMETERS

@Field @Option(name="schema_ursql", usage="TYPE@PROP=VALUE", widget="String", required=true, def="CHAINE")

                String schema_ursql

@Field @Option(name="minimum_schema_size", usage="Minimum size needed to consider a schema", widget="Integer", required=true, def="3")

                int minimum_schema_size

@Field @Option(name="maximum_schema_size", usage="Maximum size needed to consider a schema", widget="Integer", required=true, def="9999999")

                int maximum_schema_size

@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF        ON        ALL        REALLY ALL", required=true, def="OFF")

                debug

if (!ParametersDialog.open(this)) return

if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3

def CQI = CQPEngine.getCqiClient()

def allresults = new LinkedHashMap()

def props = new HashSet()

for (def corpus : selection) {

        mainCorpus = corpus.getMainCorpus()

        def word = mainCorpus.getWordProperty()

        def analecCorpus = AnalecCorpora.getCorpus(mainCorpus.getName())

        for (def type : analecCorpus.getStructure().getUnites())

                props.addAll(analecCorpus.getStructure().getUniteProperties(type));

        def selectedSchemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, maximum_schema_size);

        if (selectedSchemas.size() > 0)

                allresults[corpus] = selectedSchemas.sort(){it.getUnite0()};

}

// tabulate summary

def keys = allresults.keySet().sort();

println "Statistiques des schémas de "+corpusViewSelections.join(",")

println "\t"+keys.join("\t")

// number of schemas

print "Schemas"

for (def k : keys) {

        print "\t"+allresults[k].size();

}

println ""

// Average length

print "Average len"

for (def k : keys) {

        int total = 0;

        for (Schema s : allresults[k]) total+= s.contenu.size();

        print "\t"+total / allresults[k].size()

}

println ""

// median length

print "Median len"

for (def k : keys) {

        print "\t"+allresults[k][(int)(allresults[k].size() / 2)].contenu.size();

}

println ""

// number of properties values

for (def p : props) {

        print "$p"

        for (def k : keys) {

                def propsvalues = new HashSet()

                for (def u : allresults[k]) {

                        propsvalues.add(u.getProp(p))

                }

                if (debug) println "k @p values="+propsvalues

                print "\t"+propsvalues.size();

        }

        println ""

}