Laboratoire ICAR » Plateforme TXM

// STANDARD DECLARATIONS

package org.txm.macro

import org.kohsuke.args4j.*

import groovy.transform.Field

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

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

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

import org.txm.cooccurrence.core.functions.*

// BEGINNING OF PARAMETERS

println "corpora selection: "+corpusViewSelection

if (!(corpusViewSelection instanceof CQPCorpus)) {

	println "Selection is not a corpus: $corpusViewSelection"

	return false;

}

@Field @Option(name="word", usage="an example query", widget="String", required=true, def='Life')

def word

@Field @Option(name="ensure_transitive", usage="an example query", widget="Boolean", required=true, def='true')

def ensure_transitive

@Field @Option(name="property", usage="an example query", widget="String", required=true, def='lemma')

def property

@Field @Option(name="distance_max", usage="an example query", widget="Integer", required=true, def='9')

def distance_max

@Field @Option(name="score_minimum", usage="an example query", widget="Float", required=true, def='2.5')

def score_minimum

@Field @Option(name="profondeur_max", usage="an example query", widget="Integer", required=true, def='3')

def profondeur_max

@Field @Option(name="debug", usage="an example query", widget="Boolean", required=true, def='false')

def debug

// Open the parameters input dialog box

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

lmax = ""

for (int i = 0 ; i < profondeur_max; i++) lmax += "\t"

def corpus = corpusViewSelection

wordProperty = corpus.getProperty(property)

if (wordProperty == null) {

	println "CQP property not found: $property"

	return false

}

def rightCooc = new Cooccurrence(corpus)

r = new HashSet()

def printCoocs(def pivots, def level, def cooc, def right_dir) {

	if (level.length() > lmax.length()) return;

	if (debug) println "#"+level+pivots

	def bouts = []

	for (def s : pivots) {

		bouts << "[$property=\"$s\"%cd]"

	}

	String qs = bouts.join(" []{0, ${distance_max}} ")

	def q = new CQLQuery(qs)

	if (debug) println level+q

	r.addAll(pivots)

	if (right_dir)

		cooc.setParameters(q, [wordProperty], null, 0, 0, 1, distance_max+1, 1, score_minimum, 1, false, false);

	else

		cooc.setParameters(q, [wordProperty], null, distance_max+1, 1, 0, 0, 1, score_minimum, 1, false, false);

	cooc.setDirty(true)

	cooc.compute()

	def lines = cooc.getLines()

	def words = [:]

	for (def line : lines) {

		if (!r.contains(line.props[0])) {

			words[line.props[0]] = line

		} 

	}

	if (debug) println level+lines.size()

	for (def word : words.keySet()) {

		int size = 10+words[word].score

		if (size > 30) size = 30

		def new_pivots = []

		if (right_dir) {

			def pivot = pivots[0]

			println "$level\"$pivot\" -> \"$word\"	[label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"

			if (ensure_transitive) new_pivots.addAll(pivots)

			new_pivots << word

		} else { 

			def pivot = pivots[-1]

			println "$level\"$word\" -> \"$pivot\"	[label=\"${words[word].score}\",fontsize=$size,weight=${words[word].score}];"

			new_pivots << word

			if (ensure_transitive) new_pivots.addAll(pivots)

		}

		printCoocs(new_pivots, level+" ", cooc, right_dir)

	}

}

println "digraph {"

println "rankdir=LR;"

println "\"$word\" [shape=box];"

printCoocs([word], "", rightCooc, false)

r = new HashSet()

printCoocs([word], "", rightCooc, true)

println "}"