Laboratoire ICAR » Plateforme TXM

// Copyright © 2021 MY_INSTITUTION

// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html)

// @author mdecorde

// STANDARD DECLARATIONS

package org.txm.macro.projects.antract

import org.txm.macro.projects.antract.OkapiSaphirAPI

import org.txm.searchengine.cqp.CQPSearchEngine

import org.kohsuke.args4j.*

import groovy.transform.Field

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

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

import org.txm.searchengine.cqp.corpus.Subcorpus

import org.txm.searchengine.cqp.corpus.query.CQLQuery

import org.txm.rcp.views.corpora.CorporaView

import org.txm.backtomedia.preferences.BackToMediaPreferences

// BEGINNING OF PARAMETERS

@Field @Option(name="liste_identifiants_sujets", usage="Liste des identifiants de notice séparés par des '|'", widget="Text", required=true, def="")

def liste_identifiants_sujets

@Field @Option(name="nom_sous_corpus", usage="Nom du sous-corpus à créer", widget="String", required=true, def="")

def nom_sous_corpus

//@Field @Option(name="identifiant_corpus_okapi", usage="Identifiant du corpus Okapi à importer", widget="String", required=true, def="")

//def identifiant_corpus_okapi

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Erreur: la sélection n'est pas un corpus."

	return false

} 

CQPCorpus corpus = corpusViewSelection

corpus.compute()

CQPCorpus parentCorpus = corpus.getMainCorpus()

String parentCorpusName = parentCorpus.getCqpId()

if (parentCorpusName.startsWith("AF-VOIX-OFF-")) {

} else if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {

} else {

	println "Erreur: le corpus n'est pas un corpus d'AF-VOIX-OFF ni AFNOTICE"

	return false

}

// Open the parameters input dialog box

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

//monitor.syncExec(new Runnable() {

//	public void run() {

//		if (!OkapiSaphirAPI.initializeCredentials(org.eclipse.swt.widgets.Display.getCurrent().getActiveShell(), BackToMediaPreferences.MEDIA_AUTH_LOGIN, BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "OKAPI - Antract", "S'identifier sur la plateforme Okapi.", "OKAPI - Antract", monitor)) {

//			return null;

//		}

//	}

//});

//

//String user = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN)

//String password = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD)

//if (user == null || user.length() == 0 || password == null || password.length() == 0) {

//	println "Annulation."

//}

//

//String sessionID = OkapiSaphirAPI.login(user, password)

//if (sessionID.length() == 0) {

//	println "Erreur : La session Okapi n'a pu être démarrée."

//	System.setProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN, "")

//	System.setProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "")

//}

//

//String corpora = OkapiSaphirAPI.get_user_corpus(sessionID, user, null, null, null)

//if (corpora.length() > 0 ) {

//	println "CORPORA"

//	println corpora

//} else {

//	println "Erreur : la liste des corpus n'a pas pu être récupérée."

//}

liste_identifiants_sujets = liste_identifiants_sujets.replace("\n", "|").replace("\t", "|").replace(" ", "|").replaceAll("(\\|)+", "|").trim()

println "Création du sous-corpus avec la liste: $liste_identifiants_sujets"

String prop = "div_id"

if (parentCorpusName.startsWith("AFNOTICES") || parentCorpusName.startsWith("AF-NOTICES")) {

	prop = "notice_identifiantdelanotice"

}

liste_identifiants_sujets = (liste_identifiants_sujets.split("\\|") as List)

liste_identifiants_sujets.remove("")

liste_identifiants_sujets.unique()

liste_identifiants_sujets = getShortCQL(liste_identifiants_sujets)

String query = "[_.$prop=\"$liste_identifiants_sujets\"] expand to $prop"

println "CQL=$query"

if (query.length() > 15500) {

	println "Erreur : la requête CQL est trop longue (${query.length()})."

	return

}

Subcorpus sub = parentCorpus.createSubcorpus(new CQLQuery(query), nom_sous_corpus)

int s = sub.getSize()

if (s > 0) {

	println "$nom_sous_corpus subcorpus is created."

	monitor.syncExec(new Runnable() {

					public void run() {

						CorporaView.refresh();

						CorporaView.expand(parentCorpus);

					}

				});

} else {

	println "Error: the created corpus is empty!"

}

// UTILS UTILS UTILS UTILS UTILS UTILS UTILS

def getShortCQL(def list) {

	def origtree = ["":list]

	breakNode(origtree, "", 0)

	def CQL1 = list.join("|")

	def CQL2 = writeNode(origtree)

	if (CQL1.length() < CQL2.length()) {

		return CQL1;

	} else {

		return CQL2;

	}

}

def breakNode(def tree, def key, def p) {

	def space = (" "*p)

	//println space + "breaking: $key of $tree at $p"

	def children = new LinkedHashMap()

	def ids = tree[key]

	tree[key] = children

	for (def child : ids) {

		//println space+"child="+child

		if (child.length() <= p) continue;

		def letter = child[p]

		if (!children.containsKey(letter)) {

			children[letter] = []

		}

		children[letter].add(child)

	}

	for (def k : children.keySet()) {

		breakNode(children, k , p+1)

	}

}

def writeNode(def tree) {

	def s = ""

	//println "tree: "+tree

	int n = 0;

	if (tree.size() > 1) s += "("

	String ss = ""

	for (def child : tree.keySet()) {

		if (n > 0) ss += "|"

		ss += child

		ss += writeNode(tree[child])

		n++

	}

	if (ss.matches("([0-9]\\|){2,10}[0-9]")) ss = "["+ss.replace("|", "") +"]"

	if (ss == "[0123456789]") ss = "[0-9]"

	if (ss == "[123456789]") ss = "[1-9]"

	if (ss == "[23456789]") ss = "[2-9]"

	if (ss == "[3456789]") ss = "[3-9]"

	if (ss == "[0123456789]") ss = "[0-9]"

	if (ss == "[012345678]") ss = "[0-8]"

	if (ss == "[01234567]") ss = "[0-7]"

	if (ss == "[0123456]") ss = "[0-6]"

	if (ss == "[12345678]") ss = "[1-8]"

	if (ss == "[1234567]") ss = "[1-7]"

	s += ss

	if (tree.size() > 1) s += ")"

	return s

}

// Copyright © 2021 MY_INSTITUTION

// Licensed under the terms of the GNU General Public License version 3 (http://www.gnu.org/licenses/gpl-3.0.html)

// @author mdecorde

// STANDARD DECLARATIONS

package org.txm.macro.projects.antract

import org.txm.macro.projects.antract.OkapiSaphirAPI

import org.txm.searchengine.cqp.CQPSearchEngine

import org.kohsuke.args4j.*

import groovy.transform.Field

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

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

import org.txm.searchengine.cqp.corpus.Subcorpus

import org.txm.searchengine.cqp.corpus.query.CQLQuery

import org.txm.rcp.views.corpora.CorporaView

import org.txm.backtomedia.preferences.BackToMediaPreferences

// BEGINNING OF PARAMETERS

monitor.syncExec(new Runnable() {

	public void run() {

		if (!OkapiSaphirAPI.initializeCredentials(org.eclipse.swt.widgets.Display.getCurrent().getActiveShell(), BackToMediaPreferences.MEDIA_AUTH_LOGIN, BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "OKAPI - Antract", "S'identifier sur la plateforme Okapi.", "OKAPI - Antract", monitor)) {

			return null;

		}

	}

});

String user = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN)

String password = System.getProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD)

if (user == null || user.length() == 0 || password == null || password.length() == 0) {

	println "Annulation."

}

String sessionID = OkapiSaphirAPI.login(user, password)

if (sessionID.length() == 0) {

	println "Erreur : La session Okapi n'a pu être démarrée."

	System.setProperty(BackToMediaPreferences.MEDIA_AUTH_LOGIN, "")

	System.setProperty(BackToMediaPreferences.MEDIA_AUTH_PASSWORD, "")

}

String corpora = OkapiSaphirAPI.get_user_corpus(sessionID, user, null, null, null)

if (corpora.length() > 0 ) {

	println "CORPORA"

	println corpora

} else {

	println "Erreur : la liste des corpus n'a pas pu être récupérée."

}