Laboratoire ICAR » Plateforme TXM

                  <dynamic

                        class="org.txm.annotation.urs.commands.URSToolsMenuContribution"

                        id="edit">

                  </dynamic>

                  <dynamic

                        class="org.txm.annotation.urs.commands.URSToolsMenuContribution"

                        id="check">

                  </dynamic>

                     label="Export">

                        id="export">

package org.txm.macro.urs.check

// 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

import groovy.transform.Field

import org.jfree.chart.JFreeChart

import org.kohsuke.args4j.*

import org.txm.Toolbox

import org.txm.annotation.urs.*

import org.txm.macro.urs.AnalecUtils

import org.txm.rcp.Application

import org.txm.rcp.IImageKeys

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

import org.txm.searchengine.cqp.CQPSearchEngine

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

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

import visuAnalec.elements.*

def scriptName = this.class.getSimpleName()

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "** $scriptName please select a Corpus to run the macro"

	return;

}

// BEGINNING OF PARAMETERS

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

		String schema_ursql

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

		String schema_property_display

@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 = CQPSearchEngine.getCqiClient()

def corpus = corpusViewSelection

def word = corpus.getWordProperty()

def analecCorpus = URSCorpora.getCorpus(corpus)

if (schema_property_display.length() > 0) {

	def errors = AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql, schema_property_display).size()

	if (errors > 0) {

		println "Error: some Schema types don't contain the $schema_property_display property: $errors"

		return

	}

}

def allUnits = [:]

def allSchemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus , schema_ursql, -1, Integer.MAX_VALUE)

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

	println "No schema match for '$schema_ursql' selection. Aborting"

	return

}

allUnits = AnalecUtils.groupAllUnitesInElements(debug, allSchemas)

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

	println "No unit selection. Aborting"

	return

}

if (debug) println "allUnits=${allUnits.size()}"

def duplicates = [:]

for (Schema schema : allSchemas) {

	def units = allUnits[schema];

	for (def unit : units) {

		if (!duplicates.containsKey(unit)) duplicates[unit] = []

		duplicates[unit] << schema

	}

}

def units = []

units.addAll(duplicates.keySet()) // remove non duplicates from hash

for (def unit : units) {

	if (duplicates[unit].size() < 2) duplicates.remove(unit)

}

if (duplicates.size() > 0) {

	println "Duplicates found"

	for (def unit : duplicates.keySet()) {

		println AnalecUtils.toString(CQI, word, unit)+" in: "

		for (Schema schema : duplicates[unit]) {

			println " '"+schema.getProp(schema_property_display)+"'\t"+schema.getProps()

		}

	}

} else {

	println "No duplicates found in $schema_ursql units"

}

return duplicates

// STANDARD DECLARATIONS

package org.txm.macro.urs.check

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.annotation.urs.*

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

import visuAnalec.elements.*

if (!(corpusViewSelection instanceof MainCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

@Field @Option(name="unit_type",usage="", widget="String", required=false, def="MENTION")

String unit_type

@Field @Option(name="unit_property_name", usage="", widget="String", required=false, def="REF")

String unit_property_name

@Field @Option(name="pruneUnusedValues", usage="", widget="Boolean", required=false, def="false")

boolean pruneUnusedValues

// Open the parameters input dialog box

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

// END OF PARAMETERS

MainCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus);

def structure = analecCorpus.getStructure()

def unit_types = structure.getTypes(Unite.class);

if (!unit_types.contains(unit_type)) {

	println "Missing unit type: $unit_type"

	return

}

def props = analecCorpus.getStructure().getUniteProperties(unit_type);

if (!props.contains(unit_property_name)) {

	println "No properties '$unit_property_name' in '$unit_type' unit."

	return

}

def prop  = unit_property_name

def tmpvalues = new HashSet()

tmpvalues.addAll(structure.getValeursProp(Unite.class, unit_type, prop));

println "Values stored in the structure: $tmpvalues"

def used_values = new HashSet();

def unites = analecCorpus.getUnites(unit_type);

for (Unite unite : unites) {

	used_values.add(unite.getProp(prop))

}

tmpvalues.removeAll(used_values);

if (tmpvalues.size() > 0) {

	println "The following values ("+tmpvalues.size()+") are not used: "+tmpvalues.join(", ")

	if (pruneUnusedValues) {

		println "Pruning the values..."

		for (String val : tmpvalues) {

			structure.supprimerVal(Unite.class, unit_type, unit_property_name, val);

		}

	} 

} else {

	println "All the values are used."

}

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.*

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

import org.txm.annotation.urs.*

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

import org.apache.commons.lang.StringUtils;

// BEGINNING OF PARAMETERS

@Field @Option(name="tsvFile",usage="", widget="FileSave", required=true, def="result.tsv")

File tsvFile

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

String default_schema_ursql

@Field @Option(name="default_minimum_schema_size", usage="", widget="Integer", required=true, def="3")

int default_minimum_schema_size

@Field @Option(name="schema_property_display_name",usage="", widget="String", required=false, def="REF")

String schema_property_display_name

@Field @Option(name="default_unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String default_unit_ursql

@Field @Option(name="default_word_property", usage="", widget="String", required=false, def="word")

String default_word_property

@Field @Option(name="default_pos_property", usage="", widget="String", required=false, def="CATEGORIE")

String default_pos_property

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

// END OF PARAMETERS

println "Corpora selections: "+corpusViewSelections

table = [] // contains all table lines

mesures = []

for (def corpus : corpusViewSelections) { // for each corpus selected in the corpora view

	if (!(corpus instanceof MainCorpus)) continue; // check if the corpus is a maincorpus

	def line = [] ; table << line // create and add a new table line

	line << corpus.getID() // add the corpus name in the first column

	println "*** Computing mesures for $corpus" // each macro return a "result" and a "data"

	params = [

		"unit_ursql":default_unit_ursql,

	]

	returnedValue = execMesure(ReferentialDensityMacro, line, corpus, params)

	line << returnedValue["result"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"unit_ursql":default_unit_ursql,

	]

	returnedValue = execMesure(MeanDistanceMacro, line, corpus, params)

	line << returnedValue["result"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"unit_ursql":default_unit_ursql,

	]

	returnedValue = execMesure(NumberOfSchemaMacro, line, corpus, params)

	line << returnedValue["result"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"schema_property_display_name":schema_property_display_name,

		"unit_ursql":default_unit_ursql+"@CATEGORIE=GN Défini|GN Démonstratif|Nom Propre",

		"word_property":default_word_property,

	]

	returnedValue = execMesure(StabilityScoreMacro, line, corpus, params)

	line << returnedValue["result"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"unit_ursql":default_unit_ursql,

	]

	returnedValue = execMesure(InterDistanceMacro, line, corpus, params)

	line << returnedValue["result"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"unit_ursql":default_unit_ursql,

		"word_property":default_pos_property,

	]

	returnedValue = execMesure(NatureOfTheFirstUnitMacro, line, corpus, params)

	line << returnedValue["data"]

	params = [

		"schema_ursql":default_schema_ursql,

		"minimum_schema_size":default_minimum_schema_size,

		"schema_property_display_name":schema_property_display_name,

		"unit_ursql":default_unit_ursql,

		"word_property":default_pos_property,

	]

	returnedValue = execMesure(GrammaticalCategoryMacro, line, corpus, params)

	line << returnedValue["data"]

}

// WRITE RESULTS IN THE TSV FILE

tsvFile.withWriter("UTF-8") { writer ->

	writer.println "\t"+mesures.join("\t")

	table.each { line -> writer.println line.join("\t")	}

}

println "Done. Results are saved in ${tsvFile.getAbsolutePath()} file."

// UTILITY FUNCTIONS

def execMesure(def mesure, def line, def corpus, def params) {

	def m = mesure.getSimpleName().substring(0, mesure.getSimpleName().indexOf("Macro"))

	mesures << m

	println "***** ${mesures.size()}- $m with parameters: $params"

	def r = gse.run(mesure, ["args":params, "corpusViewSelection":corpus, "monitor":monitor])

	if (r == null) throw new Exception("Null result");

	return r;

}

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.*

import org.txm.macro.urs.AnalecUtils

import visuAnalec.elements.*

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

import org.txm.annotation.urs.*

import org.txm.searchengine.cqp.*

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

import org.apache.commons.lang.StringUtils;

// 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="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="schema_display_property_name",usage="", widget="String", required=false, def="REF")

String schema_display_property_name

@Field @Option(name="unit_ursql", usage="TYPE@PROP=VALUE", widget="String", required=false, def="MENTION")

String unit_ursql

@Field @Option(name="property", usage="", widget="String", required=false, def="CATEGORIE")

String property

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

debug

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

// Open the parameters input dialog box

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

// END OF PARAMETERS

MainCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def CQI = CQPSearchEngine.getCqiClient()

def prop = corpus.getProperty(property)

if (prop == null) {

	analecCorpus.getStructure()

}

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

def allFreqs = [:]

def n = 0

for (def schema : schemas) {

	n++

	def freqs = [:]

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	for (def unit : units) { // no need to sort units

		String forme =  null;

		if (prop == null) { // property is the analec unit property to use

			forme = unit.getProp(property)

		} else {

			int[] pos = null;

			if (unit.getDeb() == unit.getFin()) pos = [unit.getDeb()]

			else pos = unit.getDeb()..unit.getFin()

			forme = StringUtils.join(CQI.cpos2Str(prop.getQualifiedName(), pos), " ") // ids is enough

		}

		if (!freqs.containsKey(forme)) freqs[forme] = 0;

		freqs[forme] = freqs[forme] + 1;

		if (!allFreqs.containsKey(forme)) allFreqs[forme] = 0;

		allFreqs[forme] = allFreqs[forme] + 1;

	}

	if (schema_display_property_name != null) {

		println "Index des natures de $unit_ursql de '"+schema.getProp(schema_display_property_name)+"' : "

	} else {

		println "Index des natures de $schema_ursql - $n : "

	}

	int max = 0;

	def result = "";

	for (def forme : freqs.sort() { a, b -> -a.value <=> -b.value ?: a.key <=> b.key }) {

		println forme.key+"\t"+forme.value

	}

}

int max = 0;

def result = "";

println "Index des natures de $schema_ursql : "

for (def forme : allFreqs.sort() { a, b -> -a.value <=> -b.value ?: a.key <=> b.key }) {

	println forme.key+"\t"+forme.value

	if (max < forme.value) {

		max = forme.value

		result = "$forme: "+forme.value

	}

}

return ["result":result, "data":allFreqs]

@Field @Option(name="limit_cql", usage="CQL to build structure limits", widget="Query", required=false, def="")

@Field @Option(name="strict_inclusion", usage="Units must be strictly included into corpus matches", widget="Boolean", required=false, def="true")

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.annotation.urs.*

import visuAnalec.elements.*

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

import org.txm.macro.urs.AnalecUtils

import org.txm.Toolbox

import org.txm.rcp.commands.*

import org.apache.commons.lang.StringUtils

import org.txm.searchengine.cqp.CQPSearchEngine

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpus view selection is not a Corpus"

	return;

}

// 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="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="unit_ursql", usage="TYPE@PROP=VALUE", widget="String", required=false, def="MENTION")

String unit_ursql

@Field @Option(name="word_property", usage="", widget="StringArray", metaVar="word	lemma	frlemma	frolemma	#forme#	id", required=false, def="word")

String word_property

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

String separator

@Field @Option(name="buildCQL", usage="générer la requête des unités", widget="Boolean", required=true, def='false')

def buildCQL

@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

CQPCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def CQI = CQPSearchEngine.getCqiClient()

if (buildCQL) {

	word_prop = corpus.getProperty("id")

} else {

	word_prop = corpus.getProperty(word_property)

}

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

schemas.sort() {it.getProps()}

def nSchemas = 0

def lens = [:]

for (def schema : schemas) {

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	print schema.getProps().toString()+ ": "

	def first = true

	for (def unit : units) {

		String forme =  null;

		if (buildCQL) {

			int[] pos = null

			if (unit.getDeb() == unit.getFin()) pos = [unit.getDeb()]

			else pos = (unit.getDeb()..unit.getFin())

			def first2= true

			q = ""

			pos.each {

				if (first2) { first2 = false } else { q = q+" " }

				int[] pos2 = [it]

				q = q+"["+word_prop+"=\""+CQI.cpos2Str(word_prop.getQualifiedName(), pos2)[0]+"\"]"

			}

			if (first) { first = false } else { print "|" }

			print "("+q+")"

		} else {

			if (word_prop == null) { // word_property is the analec unit property to use

				forme = unit.getProp(word_property)

			} else {

				int[] pos = null

				if (unit.getDeb() == unit.getFin()) pos = [unit.getDeb()]

				else pos = (unit.getDeb()..unit.getFin())

					forme = StringUtils.join(CQI.cpos2Str(word_prop.getQualifiedName(), pos), " ") // ids is enough

			}

			if (first) { first = false } else { print separator }

			print forme

		}

	}

	println ""

	nSchemas++

}

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.annotation.urs.*

import org.txm.macro.urs.AnalecUtils

import visuAnalec.elements.*

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

import org.txm.Toolbox

import org.txm.rcp.commands.*

import org.txm.statsengine.r.core.RWorkspace

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

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

String schema_ursql

@Field @Option(name="minimum_schema_size", usage="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="schema_property_display", usage="schema property to show", widget="String", required=true, def="REF")

String schema_property_display

@Field @Option(name="unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String unit_ursql

@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 corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

int nSchemas = 0;

def lens = [:]

def lensnames = [:]

for (def schema : schemas) {

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	int nUnites = units.size();

	if (!lens.containsKey(nUnites)) {

		lens[nUnites] = 0;

		lensnames[nUnites] = [];

	}

	lens[nUnites] = lens[nUnites] + 1;

	lensnames[nUnites] << schema.getProp(schema_property_display)

	nSchemas++;

}

//println "nSchemas=$nSchemas"

def freqs = lens.keySet();

freqs.sort();

int t = 0;

int n = 0;

//println "Fréquences ("+freqs.size()+")"

for (def f : freqs) {

	t += f * lens[f]

	n += lens[f]

}

coef = (t/n)

def slens = lens.sort { a, b -> -a.value <=> -b.value ?: -a.key <=> -b.key }

def flens = []

slens.each { key, value -> value.times { flens << key } }

def nbins = flens.size()*2

def cfreq = 0

println "Longueur moyenne des chaînes de référence : $t/$n = "+coef

println "Index hiérarchique des longueurs de chaînes :\nlen\tfreq\tcfreq"

slens.each { println it.key+"	"+it.value+"	"+(cfreq+=it.value)+"	"+lensnames[it.key] }

def slens2 = slens.sort { a, b -> -a.key <=> -b.key }

def r = RWorkspace.getRWorkspaceInstance()

r.addVectorToWorkspace("len", slens2.keySet() as int[])

r.addVectorToWorkspace("freq", slens2.values() as int[])

r.addVectorToWorkspace("flen", flens as int[])

def corpusName = corpus.getID()

def PNGFile = File.createTempFile("txm", ".png", new File(Toolbox.getTxmHomePath(), "results"))

def PNGFilePath = PNGFile.getAbsolutePath()

println "PNG file: "+PNGFilePath

def SVGFile = File.createTempFile("txm", ".svg", new File(Toolbox.getTxmHomePath(), "results"))

def SVGFilePath = SVGFile.getAbsolutePath()

println "SVG file: "+SVGFilePath

/// BEGINNING OF R SCRIPT

def script ="""

hist(flen, xaxt='n', col="gray", xlab="Length", breaks=$nbins, main="$corpusName Longueur des chaînes ($nbins bins)")

axis(side=1, at=len)

dev.off()

"""

/// END OF R SCRIPT

// execute R script

r.plot(PNGFile, "png(file = \"${PNGFilePath}\"); "+script)

r.plot(SVGFile, "svglite(file = \"${SVGFilePath}\"); "+script)

//display the SVG results graphic

monitor.syncExec(new Runnable() {

	@Override

	public void run() { OpenBrowser.openfile(SVGFilePath, corpusName+" Longueur des chaînes") }

})

return ["result":coef, "data":lens]

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.*

import org.txm.macro.urs.AnalecUtils

import visuAnalec.elements.*

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

import org.txm.annotation.urs.*

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

import org.apache.commons.lang.StringUtils;

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

// BEGINNING OF PARAMETERS

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

String schema_ursql

@Field @Option(name="minimum_schema_size", usage="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="schema_display_property_name",usage="", widget="String", required=false, def="REF")

String schema_display_property_name

@Field @Option(name="unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String unit_ursql

@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

CQPCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

def distances = [];

def nDistances = 0

def cadences = [];

for (def schema : schemas) {

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	Collections.sort(units)

	for (int i = 0 ; i < units.size() ; i++) {

		int d1 = 0;

		int d2 = 0;

		if (i < units.size()-1) d1 = units[i+1].getDeb() - units[i].getFin();

		if (d1 < 0) {

			//println "D1 "+units[i+1].getDeb()+" - "+units[i].getFin()+" = "+d1

			d1 = 0; // the first unit pass the next one ?

		}

		if (i > 0) d2 = units[i].getDeb() - units[i-1].getFin();

		if (d2 < 0) {

			//println "D2 "+units[i].getDeb()+" - "+units[i-1].getFin()+" = "+d2

			d2 = 0; // the first unit pass the next one ?

		}

		distances << d1

		if (d1 < d2) cadences << d1 else cadences << d2

		nDistances++

	}

}

distances = distances.sort()

cadences = cadences.sort()

int distances_total = distances.sum()

int cadences_total = cadences.sum()

coef = (distances_total / nDistances)

cadence = (cadences_total / nDistances)

println "distances $distances"

println "distance moyenne inter-mayonnaise : $distances_total / $nDistances = $coef"

println "distance medianne inter-mayonnaise : "+distances[(int)(distances.size() / 2)]

println "distance quartils : "+distances[0]+" "+distances[(int)(distances.size() / 4)] + " "+distances[(int)(distances.size() / 2)]+" "+distances[(int)(3*distances.size() / 4)]+" "+distances[(int)(distances.size() -1)]

println "cadences $cadences"

println "cadence moyenne : $cadences_total / $nDistances = $cadence"

println "cadence medianne : "+cadences[(int)(cadences.size() / 2)]

println "cadence quartils : "+cadences[0]+" "+cadences[(int)(cadences.size() / 4)] + " "+cadences[(int)(cadences.size() / 2)]+" "+cadences[(int)(3*cadences.size() / 4)]+" "+cadences[(int)(cadences.size() -1)]

return ["result":coef, "result2":cadence, "data":["distances":distances, "nDistances":nDistances, "cadences":cadences]]

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.*

import org.txm.macro.urs.AnalecUtils

import visuAnalec.elements.*

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

import org.txm.annotation.urs.*

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

import org.txm.searchengine.cqp.CQPSearchEngine

import org.apache.commons.lang.StringUtils;

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

// BEGINNING OF PARAMETERS

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

String schema_ursql

@Field @Option(name="minimum_schema_size", usage="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String unit_ursql

@Field @Option(name="word_property", usage="", widget="String", required=false, def="CATEGORIE")

String word_property

@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

CQPCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def CQI = CQPSearchEngine.getCqiClient()

def prop = corpus.getProperty(word_property)

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

def freqs = [:]

def distances = 0;

def nDistances = 0

for (def schema : schemas) {

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	int nUnites = units.size();

	if (units.size() == 0) continue;

	def unit = units[0]

	String forme =  null;

	if (prop == null) { // word_property is the analec unit property to use

		forme = unit.getProp(word_property)

	} else {

		int[] pos = null;

		if (unit.getDeb() == unit.getFin()) pos = [unit.getDeb()]

		else pos = unit.getDeb()..unit.getFin()

		forme = StringUtils.join(CQI.cpos2Str(prop.getQualifiedName(), pos), " ") // ids is enough

	}

	if (!freqs.containsKey(forme)) freqs[forme] = 0;

	freqs[forme] = freqs[forme] + 1;

}

println "Index des natures de premier maillon :"

int max = 0;

def result = "";

for (def forme : freqs.keySet().sort() {it -> -freqs[it]}) {

	println "$forme\t"+freqs[forme]

	if (max < freqs[forme]) {

		max = freqs[forme]

		result = "$forme: "+freqs[forme]

	}

}

["result": result, "data": freqs]

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.annotation.urs.*

import visuAnalec.elements.*

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

import org.txm.macro.urs.AnalecUtils

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

// BEGINNING OF PARAMETERS

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

String schema_ursql

@Field @Option(name="minimum_schema_size", usage="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String unit_ursql

@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

CQPCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

// check Schema parameters

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

int nSchemas = schemas.size();

println "Nombre de chaînes de référence d'un texte : $nSchemas"

["result":nSchemas, "data":schemas]

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.annotation.urs.*

import visuAnalec.elements.*

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

import org.txm.macro.urs.AnalecUtils

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

@Field @Option(name="unit_ursql", usage="TYPE@PROP=VALUE", widget="String", required=false, def="MENTION")

String unit_ursql

@Field @Option(name="limit_cql", usage="CQL to build structure limits", widget="Query", required=true, def="<div> [] expand to div")

limit_cql

@Field @Option(name="strict_inclusion", usage="Units must be strictly included into corpus matches", widget="Boolean", required=true, def="true")

strict_inclusion

@Field @Option(name="limit_distance", usage="Unit distance to structure limit (0 = no selection, 1 = first after limit, -1 = last before limit, etc.)", widget="Integer", required=true, def="0")

limit_distance

@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

CQPCorpus corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus);

int nMots = corpus.getSize();

def units = AnalecUtils.selectUnitsInSchema(debug, analecCorpus, corpus, "", 0, 0,

	unit_ursql, 0, limit_cql, strict_inclusion, limit_distance);

int nUnites = units.size();

coef = (nUnites /nMots)

println "Densité référentielle : nUnites/nMots = $nUnites/$nMots = $coef = ${coef*100}%"

if (nUnites >= nMots) {

	println "WARNING: possible encoding error. Number of units ($nUnites) is greater than number of words ($nMots)"

}

return ["result":coef, "data":["nUnites":nUnites, "nMots":nMots]]

// 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.urs.exploit

import org.kohsuke.args4j.*

import groovy.transform.Field

import org.txm.*

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

import org.txm.annotation.urs.*

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

import org.apache.commons.lang.StringUtils;

import org.txm.macro.urs.AnalecUtils

import visuAnalec.elements.*

import org.txm.searchengine.cqp.CQPSearchEngine

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Corpora selection is not a Corpus"

	return;

}

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

String schema_ursql

@Field @Option(name="minimum_schema_size", usage="minimal schema size", widget="Integer", required=true, def="3")

int minimum_schema_size

@Field @Option(name="schema_display_property_name",usage="", widget="String", required=false, def="REF")

String schema_display_property_name

@Field @Option(name="unit_ursql", usage="TYPE@PROP=REGEX", widget="String", required=false, def="MENTION")

String unit_ursql

@Field @Option(name="word_property", usage="", widget="String", required=false, def="word")

String word_property

@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 corpus = corpusViewSelection

def analecCorpus = URSCorpora.getCorpus(corpus)

if (!AnalecUtils.isPropertyDefined(Schema.class, analecCorpus, schema_ursql)) {

	println "** The $schema_ursql schema URSQL cannot be computed in the corpus."

	return;

}

if (!AnalecUtils.isPropertyDefined(Unite.class, analecCorpus, unit_ursql)) {

	println "** $unit_ursql unit URSQL cannot be computed in the corpus."

	return;

}

def CQI = CQPSearchEngine.getCqiClient()

def prop = corpus.getProperty(word_property)

def schemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, 999999);

allFormesSet = new HashSet();

nUnitesGrandTotal = 0;

def coefs = []

int n = 1

for (def schema : schemas) {

	def formesSet = new HashSet(); // contient toutes les formes du CR courant

	nUnitesTotal = 0;

	def allUnites = schema.getUnitesSousjacentesNonTriees()

	def units = AnalecUtils.filterElements(debug, allUnites, unit_ursql)

	def nUnites = units.size()

	for (def unit : units) {

		String forme =  null;

		if (prop == null) { // word_property is the analec unit property to use

			forme = unit.getProp(word_property)

		} else {

			int[] pos = null;

			if (unit.getDeb() == unit.getFin()) pos = [unit.getDeb()]

			else pos = unit.getDeb()..unit.getFin()

			forme = StringUtils.join(CQI.cpos2Str(prop.getQualifiedName(), pos), " ") // ids is enough

		}

		formesSet.add(forme)

		nUnitesTotal++

	}

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

		coef = -1

	} else {

		coef = (nUnitesTotal/formesSet.size())

	}

	coefs << coef

	nUnitesGrandTotal += nUnitesTotal;

	allFormesSet.addAll(formesSet)

	if (schema_display_property_name != null) {

		print schema.getProp(schema_display_property_name)

	} else {

		print schema_ursql+"-"+n+" : "

	}

	println " ($nUnites units) : $nUnitesTotal selected units / ${formesSet.size()} forms = $coef"

	n++

}

return ["result":coefs, "data":["nUnitesTotal":nUnitesGrandTotal, "allFormesSet":allFormesSet], "coef":(nUnitesGrandTotal/allFormesSet.size())]