Laboratoire ICAR » Plateforme TXM

// STANDARD DECLARATIONS

package org.txm.macroproto.partition

import groovy.transform.Field

import org.kohsuke.args4j.*

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

import org.txm.rcpapplication.views.*

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

//BEGINNING OF PARAMETERS

def corpus = corpusViewSelection

if (!(corpus instanceof Corpus)) {

        println "Error: this macro should be run with a Corpus selected"

        return

}

@Field @Option(name="property", usage="the structural Unit properties list separated with commas", widget="String", required=true, def="sadnesst1")

def property = "sadnesst1"

@Field @Option(name="values_file", usage="txt file", widget="File", required=true, def="C:\\Users\\Fanny\\Desktop\\values_num.txt")

def values_file

@Field @Option(name="balance", usage="the structural Unit properties list separated with commas", widget="Boolean", required=true, def="true")

def balance = false

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

def expand_struct = "question"

// Open the parameters input dialog box

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

def property_values = values_file.getText("UTF-8").replaceAll("\n", ",").replaceAll("\r", "")

println "Balance de la métadonnée '$property' du corpus '$corpus'"

def domain = []

for (def s : property_values.split(",")) {

        s = s.trim()

        try {

                domain << Integer.parseInt(s)

        } catch(Exception e) {

                domain << Float.parseFloat(s)

        }

}

def uniq_domain = new HashSet(domain).sort() 

println "domain size "+domain.size()

def names = []

def queries = []

def p = 2

String values = ""

if (!balance) {

for (def d : uniq_domain) {

        if (d < p) {

                values += "|$d"

        } else {

                if (values.startsWith("|")) values = values.substring(1)

                queries <<  "[_.text_"+property+"=\"$values\"] expand to $expand_struct"

                names << values

                values = d

                p++

        }

}

queries <<  "[_.text_"+property+"=\"$values\"] expand to $expand_struct"

names << values

corpus.createPartition(property+"_non_balanced", queries, names);

} else {

//String v = domain.toString()

//v = "c("+v.substring(1, v.length() -1) + ")"

String v  = "c($property_values)"

println "=========================="

//RWorkspace.getRWorkspaceInstance().addVectorToWorkspace("domain", domain.toArray(new double[domain.size()]))

def res = RWorkspace.getRWorkspaceInstance().eval("quantile($v)")

def ranges = res.asDoubles()

println "quantiles : $ranges"

values = ""

p = ranges[1]

i = 1

names = []

queries = []

for (def d : uniq_domain) {

        //println "d=$d, p=$p"

        if (d < p) {

                values += "|$d"

        } else {

                if (values.startsWith("|")) values = values.substring(1)

                println """[_.text_${property}="$values"]"""

                queries <<  "[_.text_"+property+"=\"$values\"] expand to $expand_struct"

                names << values

                values = d

                i++

                p = ranges[i]

        }

}

queries <<  "[_.text_"+property+"=\"$values\"] expand to $expand_struct"

names << values

corpus.createPartition(property+"_balanced", queries, names);

}

println queries

println names

//display the graphic

monitor.syncExec(new Runnable() {

        @Override

        public void run() {        

        org.txm.rcpapplication.commands.RestartTXM.reloadViews();

        }

});

/*

int[] counts = new int[51]

for (def d : domain) {

        counts[(int)(10*d)]++

}

println counts

*/