Laboratoire ICAR » Plateforme TXM

// Copyright © 2023 ENS de Lyon

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

// @author sheiden

// STANDARD DECLARATIONS

package org.txm.macro.cqp

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.macro.cqp.CQPUtils

import org.txm.searchengine.cqp.CQPSearchEngine

scriptName = this.class.getSimpleName()

// BEGINNING OF PARAMETERS

@Field @Option(name="struct_props", usage="struct@prop1,prop2...", widget="String", required=true, def="")

def struct_props

@Field @Option(name="Vmax", usage="maximum number of values displayed", widget="Integer", required=true, def="100")

def Vmax

@Field @Option(name="groupByValue", usage="group results by value before counting", widget="Boolean", required=false, def="true")

def groupByValue

// Open the parameters input dialog box

if (!ParametersDialog.open(this)) return

// END OF PARAMETERS

utils = new CQPUtils()

corpusEngine = CQPSearchEngine.getCqiClient()

// check for a corpus selection

corpora = utils.getCorpora(this)

if ((corpora == null) || corpora.size() > 1) {

	println "** $scriptName: please select a corpus in the Corpus view or provide a corpus name. Aborting."

	return false

}

if ((struct_props == null) || struct_props.size() == 0) {

	println "** $scriptName: please set the 'struct_props' parameter. Aborting."

	return false

}

corpus = corpora[0].getMainCorpus().getName()

globalVmax = Vmax

(struct, prop) = struct_props.tokenize('@')*.trim()

  if ((prop == null) || prop.size() == 0) {

    println "** $scriptName: please set @properties in the 'struct_props' parameter."

  } else {

    if (prop.indexOf(',') == -1) {

      prop = [prop]

    } else {

      prop = prop.tokenize(',')*.trim()

    }

    occs = (0..corpusEngine.attributeSize("$corpus.${struct}_${prop[0]}")-1).collect { cpos ->

      prop.collect { prop ->

        corpusEngine.struc2Str("$corpus.${struct}_$prop", [cpos] as int[])[0]

      }.join('_')

    }

    if (groupByValue) {

      occs = occs.sort().countBy { it }.sort { a,b -> b.value <=> a.value ?: a.key <=> b.key }

      if (globalVmax == 0) {

      	occs.each { println sprintf("%s\t%d", it.key, it.value) }

      } else {

      occs.take(globalVmax).each { println sprintf("%s\t%d", it.key, it.value) }

      }

    } else {

      if (globalVmax == 0) {

        occs.each { println it }

      } else {

        occs.take(globalVmax).each { println it }

      }

    }

  }

// Copyright © 2023 ENS de Lyon

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

// @author sheiden

// STANDARD DECLARATIONS

package org.txm.macro.cqp

import org.kohsuke.args4j.*

import groovy.transform.Field

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

import org.txm.macro.cqp.CQPUtils

import org.txm.searchengine.cqp.CQPSearchEngine

scriptName = this. class .getSimpleName()

// BEGINNING OF PARAMETERS

@Field @Option(name = "structs_props", usage = "struct1@prop1_prop2,struct2@prop3_prop4...", widget 

    = "String", required = true, def  = "")

def structs_props

@Field @Option(name = "Vmax", usage = "maximum number of values displayed (use '0' to display all)", widget = "Integer", required = true, 

    def  = "0")

def Vmax

// Open the parameters input dialog box

if (!ParametersDialog.open(this)) {

    return

}

// END OF PARAMETERS

utils = new CQPUtils()

corpusEngine = CQPSearchEngine.getCqiClient()

// check for a corpus selection

corpora = utils.getCorpora(this)

if ((corpora == null) || corpora.size() > 1) {

    println "** $scriptName: please select a corpus in the Corpus view or provide a corpus name. Aborting."

    return false

}

// check for structs_props parameter

if ((structs_props == null) || structs_props.size() == 0) {

    println "** $scriptName: please set the 'structs_props' parameter. Aborting."

    return false

}

corpus_name = corpora[0].getMainCorpus().getName()

corpus_size = corpora[0].getMainCorpus().getSize()

if (structs_props.indexOf(',') == -1) {

    struct_props = [structs_props]

} else {

    struct_props = structs_props.tokenize(',')*.trim()

}

globalVmax = Vmax

def printPropValues(start, end, struct_prop, output) {

    if (struct_prop.size() > 0) {

        curr_struct_prop = struct_prop[0]

        s_p = curr_struct_prop.tokenize('@')*.trim()

        def struct = s_p[0]

        def prop = s_p[1]

        if ((prop == null) || prop.size() == 0) {

            println "** $scriptName: please set @properties in the 'structs_props' parameter."

        } else {

            if (prop.indexOf('_') == -1) {

                prop = [prop]

            } else {

                prop = prop.tokenize('_')*.trim()

            }

            occs = (0..corpusEngine.attributeSize("$corpus_name.${struct}_${prop[0]}")-1).collect {

                idx -> curr_start = corpusEngine.struc2Cpos("$corpus_name.${struct}_${prop[0]}", idx as int)[0]

                curr_end = corpusEngine.struc2Cpos("$corpus_name.${struct}_${prop[0]}", idx as int)[1]

                if (curr_start >= start && curr_end <= end) {

                    [

                        curr_start,

                        curr_end,

                        prop.collect {

                            curr_prop -> corpusEngine.struc2Str("$corpus_name.${struct}_$curr_prop", [idx] as int [])[0]

                        }.join('_')

                    ]

                } else null

            }

            def n = 0

            occs.each {

                if (it != null && (globalVmax == 0 || n < globalVmax)) {

                    (start, end, value) = it

                    // println prefix+value // struct+"@"+prop[0]+"="+

                    if (output.length() > 0) {

                        printPropValues(start, end, struct_prop.tail(), output + "\t" + value)

                    } else {

                        printPropValues(start, end, struct_prop.tail(), value)

                    }

                    n++

                }

            }

        }

    } else println output

}

printPropValues(0, corpus_size, struct_props, "")