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.misc

import org.apache.commons.lang.StringUtils

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

import groovy.transform.Field

import org.kohsuke.args4j.*

import org.txm.Toolbox

import org.txm.analec.*

import org.txm.lexicaltable.core.statsengine.r.data.LexicalTableImpl

import org.txm.macro.analec.*

import org.txm.rcp.commands.*

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

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

import org.txm.searchengine.cqp.corpus.query.Query;

import visuAnalec.donnees.*

import visuAnalec.elements.*

import cern.colt.matrix.DoubleFactory2D

import cern.colt.matrix.DoubleMatrix2D

def scriptName = this.class.getSimpleName()

def parent

def selection = []

if (!(corpusViewSelection instanceof Corpus)) {

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

}

// BEGINNING OF PARAMETERS

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

                String unit_ursql

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

                int limit_distance_in_schema

@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")

                boolean 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")

                int limit_distance

@Field @Option(name="unit_prop1", usage="PROP1", widget="String", required=false, def="PROP1")

                String unit_prop1

@Field @Option(name="unit_prop2", usage="PROP2", widget="String", required=false, def="PROP2")

                String unit_prop2

@Field @Option(name="corr_method", usage="try them all", widget="StringArray", metaVar="pearson        spearman        kendall", required=false, def="pearson")

                String corr_method

@Field @Option(name="corr_style", usage="try them all", widget="StringArray", metaVar="circle        square        ellipse        number        shade        color        pie", required=false, def="number")

String corr_style

@Field @Option(name="corr_layout", usage="try them all", widget="StringArray", metaVar="full        lower        upper", required=false, def="upper")

String corr_layout

@Field @Option(name="corr_order", usage="try them all", widget="StringArray", metaVar="AOE        FPC        hclust        alphabet", required=false, def="hclust")

String corr_order

@Field @Option(name="output_lexicaltable", usage="create or not a lexical table with the result", widget="Boolean", required=true, def="false")

                output_lexicaltable

@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 correlations = [:]

def values1 = new HashSet()

def values2 = new HashSet()

def corpus = corpusViewSelection

        mainCorpus = corpus.getMainCorpus()

        def word = mainCorpus.getWordProperty()

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

        def selectedUnits = AnalecUtils.selectUnitsInSchema(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, maximum_schema_size, 

        unit_ursql, limit_distance_in_schema, limit_cql, strict_inclusion, limit_distance);

for (def unit : selectedUnits) {

        def value1 = unit.getProp(unit_prop1)

        if (value1 == null) value1 = "<null>"

        if (value1.length() == 0) value1 = "<empty>"

        def value2 = unit.getProp(unit_prop2)

        if (value2 == null) value2 = "<null>"

        if (value2.length() == 0) value2 = "<empty>"

        values1 << value1

        values2 << value2

        if (!correlations.containsKey(value1)) correlations[value1] = [:]

        def line = correlations[value1]

        if (!line.containsKey(value2)) line[value2] = 0

        line[value2] += 1

}

def matrix = new int[values1.size()][values2.size()];

println "\t"+values2.join("\t")

int i = 0;

for (def value1 : values1) {

        print value1

        int j = 0;

        for (def value2 : values2) {

        if (correlations[value1][value2] == null) correlations[value1][value2] = 0;

                print "\t"+correlations[value1][value2]

                matrix[i][j] = correlations[value1][value2]

                j++

        }

        println ""

        i++

}

def r = RWorkspace.getRWorkspaceInstance()

r.addVectorToWorkspace("corrlines", values1 as String[])

r.addVectorToWorkspace("corrcols", values2 as String[])

r.addMatrixToWorkspace("corrmatrix", matrix)

r.eval("rownames(corrmatrix) = corrlines")

r.eval("colnames(corrmatrix) = corrcols")

def resultsDir = new File(Toolbox.getTxmHomePath(), "results")

resultsDir.mkdirs()

file = File.createTempFile("txm_corr_pairs_", ".svg", resultsDir)

def title = "${corpus.getMainCorpus()}.${corpus}\n${unit_ursql}"

if (limit_distance > 1) title += "[${limit_distance}]."

if (limit_cql != null && !limit_cql.getQueryString().equals("\"\"")) title += "\n(${limit_cql} limits)"

title += "\t P1=$unit_prop1 P2=unit_prop2"

def plotScript = """

r1 = cor(corrmatrix, use="complete.obs", method="$corr_method");

r2 = cov(corrmatrix, use="complete.obs") ;

library(corrplot)

corrplot(r1, type="$corr_layout", order="$corr_order", method="$corr_style")

"""

// execute R script

if (!output_lexicaltable) {

        r.plot(file, plotScript)

}

title = "$unit_prop1 $corr_method correlations"

def lt = null;

if (output_lexicaltable) {

        mFactory = DoubleFactory2D.dense

        dmatrix = mFactory.make(values1.size(), values2.size())

        for (int ii = 0 ; ii < values1.size() ; ii++) {

                for (int jj = 0 ; jj < values2.size() ; jj++) {

                        dmatrix.set(ii, jj, matrix[ii][jj])

                }

        }

        if (corpusViewSelection instanceof Partition) {

                lt = new LexicalTableImpl(dmatrix, corpusViewSelection, corpusViewSelection.getCorpus().getProperty("word"),

                                values1 as String[], values2 as String[])

                lt.setCorpus(corpusViewSelection.getCorpus());

                corpusViewSelection.storeResult(lt)

        } else {

                lt = new LexicalTableImpl(dmatrix, corpus.getProperty("word"),

                                values1 as String[], values2 as String[])

                lt.setCorpus(corpus);

                corpus.storeResult(lt)

        }

}

        monitor.syncExec(new Runnable() {

                                @Override

                                public void run() { try {

                                                if (UnitsCorrelationMacro.this.output_lexicaltable) {

                                                        CorporaView.refreshObject(corpus)

                                                        CorporaView.expand(lt)

                                                } else {

                                                        OpenSVGGraph.OpenSVGFile(UnitsCorrelationMacro.this.file.getAbsolutePath(), "Correlations Units")

                                                }

                                        } catch (e) { e.printStackTrace() }}

                        })

return correlations