Laboratoire ICAR » Plateforme TXM

package org.txm.macro.urs.democrat

import java.io.File

import java.nio.charset.Charset

import org.eclipse.core.runtime.IProgressMonitor

import org.kohsuke.args4j.*

import groovy.transform.Field

import visuAnalec.donnees.Corpus

import visuAnalec.vue.Vue

import org.txm.annotation.urs.URSCorpora

import org.txm.importer.StaxIdentityParser

import org.txm.objects.*

import org.txm.rcp.commands.workspace.LoadBinaryCorpus

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

import org.txm.rcp.utils.JobHandler

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

import org.txm.utils.CsvReader

import org.txm.utils.io.IOUtils

import org.txm.scripts.importer.XPathResult

import javax.xml.xpath.XPathConstants

import org.txm.annotation.urs.commands.*

@Field @Option(name="inputDirectory", usage="Directory with .txm files, metadata.xslx, metadata.properties...", widget="FileOpen", required=true, def="input directory path")

def inputDirectory

@Field @Option(name="outputDirectory", usage="the default teiHeader of texts", widget="FileOpen", required=true, def="output directory path")

def outputDirectory

@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 teiHeaderTemplateFile = new File(inputDirectory, "teiHeader.xml")

def xpathFile = new File(inputDirectory, "metadata.properties")

def metadataFile = new File(inputDirectory, "metadata.tsv")

if (!inputDirectory.exists()) {

        println "no inputDirectory found: $inputDirectory"

        return false;

}

outputDirectory.mkdirs()

if (!outputDirectory.exists()) {

        println "outputDirectory not created: $outputDirectory"

        return false;

}

if (!teiHeaderTemplateFile.exists()) {

        println "no teiHeader template found: $teiHeaderTemplateFile"

        return false;

}

if (!xpathFile.exists()) {

        println "no xpath properties found: $xpathFile"

        return false;

}

if (!metadataFile.exists()) {

        println "no metadata TSV found: $metadataFile"

        return false;

}

def xpathProperties = new Properties()

xpathProperties.load(IOUtils.getReader(xpathFile))

println xpathProperties

def csvReader = new CsvReader(metadataFile.getAbsolutePath(), "\t".charAt(0), Charset.forName("UTF-8"))

println csvReader

csvReader.readHeaders();

def header = csvReader.getHeaders()

if (!header.contains("id")) {

        println "** 'id' column not found in $metadataFile header=$header" 

        return;

}

if (!header.contains("corpus_id")) {

        println "** 'corpus_id' column not found in $metadataFile header=$header"

        return;

}

def corpora = [:]

for (Project project : Workspace.getInstance().getProjects()) {

        for (MainCorpus corpus : project.getChildren(MainCorpus.class)) {

                corpora[corpus.getID()] = corpus;

        }

}

while (csvReader.readRecord())        {

        String text_id = csvReader.get("id")

        if (text_id == null || text_id.length() == 0) {

                println "** @id not found for record="+csvReader.getRawRecord()

                continue;

        }

        String corpus_id = csvReader.get("corpus_id")

        if (corpus_id == null || corpus_id.length() == 0) {

                println "** @corpus_id not found for record="+csvReader.getRawRecord()

                continue;

        }

        MainCorpus corpus = corpora[corpus_id]

        Project project = null

        if (corpus == null) {

                File binCorpusFile = new File(inputDirectory, corpus_id+".txm")

                if (!binCorpusFile.exists()) {

                        println "** no corpus binary file found for ID=$corpus_id : $binCorpusFile"

                        continue;

                }

                println "CALL load corpus from $binCorpusFile"

                JobHandler job = LoadBinaryCorpus.loadBinaryCorpusArchive(binCorpusFile)

                job.join();

                project = job.getResultObject();

                corpus = project.getCorpusBuild(corpus_id);

                corpora[corpus.getID()] = corpus;

        } else {

                println "USING loaded corpus: "+corpus

                project = corpus.getProject()

        }

        def txmDir = new File(project.getProjectDirectory(), "txm/"+corpus.getID())

        if (!txmDir.exists()) {

                println "** the selected corpus has no XML-TXM files. Aborting."

                continue;

        }

        File txmFile = new File(txmDir, text_id+".xml")

        if (!txmFile.exists()) {

                println "** the selected corpus has no XML-TXM file: $txmFile"

                continue;

        }

        println "CALL Metadata2TeiHeaderMacro"

        Metadata2TEiHeaderMacro mthm = new Metadata2TEiHeaderMacro(debug);

        for (def h : header) data[h] = csvReader.get(h)

        String xmlteiHeaderContent = mthm.getCustomizedTEiHeader(teiHeaderTemplateFile, data, xpathProperties);

        if (xmlteiHeaderContent != null && xmlteiHeaderContent.length() > 0) {

                mthm.injecting(txmFile, xmlteiHeaderContent)

        } else {

                println "** Text header not updated: $txmFile"

        }

        println "CALL ComputeDemocratMeasureMetadataMacro"

        println "CALL URS Export"

        Corpus analecCorpus = URSCorpora.getCorpus(corpus);

        Vue vue = URSCorpora.getVue(corpus);

        if (!ExportTEICorpus.export(true, outputDirectory, true, null, corpus, analecCorpus, vue)) {

                println "FAIL TO EXPORT CORPUS$corpus"

        }

}