Laboratoire ICAR » Plateforme TXM

// STANDARD DECLARATIONS

package org.txm.macro.urs.edit

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.rcp.editors.concordances.*

import org.txm.searchengine.cqp.corpus.CQPCorpus

import org.txm.searchengine.cqp.corpus.MainCorpus

import org.txm.functions.concordances.*

import org.txm.annotation.urs.*

import org.txm.concordance.core.functions.Concordance

import org.txm.concordance.rcp.editors.ConcordanceEditor

import visuAnalec.elements.Unite

// BEGINNING OF PARAMETERS

@Field @Option(name="unit_type", usage="The unit type to create", widget="String", required=true, def="MENTION")

def unit_type

@Field @Option(name="create_only_if_new", usage="Create the unit if not already annotated", widget="Boolean", required=true, def="true")

def create_only_if_new

@Field @Option(name="prop", usage="prop", widget="String", required=true, def="REF")

def prop

@Field @Option(name="value", usage="default value", widget="String", required=true, def="NAME")

def value

// END OF PARAMETERS

// get a Concordance from 1) current Concordance editor or 2) CorporaView selection

Concordance concordance

if (editor instanceof ConcordanceEditor) {

        concordance = editor.getConcordance()

} else if (corpusViewSelection instanceof Concordance) {

        concordance = corpusViewSelection

} else {

        println "You must select a concordance or open a concordance result to run this macro."

        return false

}

if (concordance == null) {

        println "You must compute a concordance before."

        return

}

// check the analec corpus is ready

MainCorpus corpus = concordance.getCorpus().getMainCorpus();

String name = corpus.getID()

if (!URSCorpora.isAnnotationStructureReady(name)) {

        println "Annotation structure is not ready."

        return

}

// Open the parameters input dialog box

if (!ParametersDialog.open(this)) return;

// check the corpus structure has the unit_type provided

def analecCorpus = URSCorpora.getCorpus(name)

if (!analecCorpus.getStructure().getUnites().contains(unit_type)) {

        //println "The corpus structure does not contains unit with type=$unit_type"

        //return;

        analecCorpus.getStructure().ajouterType(Unite.class, unit_type);

}

if (!analecCorpus.getStructure().getNomsProps(Unite.class, unit_type).contains(prop)) {

        //println "The corpus structure does not contains unit with type=$unit_type"

        //return;

        analecCorpus.getStructure().ajouterProp(Unite.class, unit_type, prop)

}

if (!analecCorpus.getStructure().getValeursProp(Unite.class, unit_type, prop).contains(value)) {

        //println "The corpus structure does not contains unit with type=$unit_type"

        //return;

        analecCorpus.getStructure().ajouterVal(Unite.class, unit_type, prop, value)

}

// browse lines and check

def units = analecCorpus.getUnites(unit_type)

def lines = concordance.getLines()

int iUnit = 0

int n = 0

for (int iLine = 0 ; iLine < lines.size() ; iLine++) {

        def line = lines[iLine]

        def m = line.getMatch()

        if (create_only_if_new && iUnit < units.size()) { // test only if create_only_if_new == true

                def unit = null

                while (iUnit < units.size() && units[iUnit].getDeb() < m.getStart()) {

                        if (iUnit < units.size()) {

                                unit = units[iUnit++]

                                if (unit.getDeb() == m.getStart() && unit.getFin() == m.getEnd()) { // skip and print the line

                                        println "skiping line: "+line.getViewRef().toString()+" "+line.keywordToString()

                                        continue

                                }

                        }

                }

        }

        n++

        def props = [:]

        props[prop] = value

        Unite u = analecCorpus.addUniteSaisie(unit_type, m.getStart(), m.getEnd(), props)

        //println "$props -> "+u.getProps()

}

println "$n $unit_type created."

if (n > 0) corpus.setIsModified(true);