Statistics
| Revision:

root / tmp / org.txm.analec.rcp / src / org / txm / macro / urs / edit / UnitsAnnotateMacro.groovy @ 1973

History | View | Annotate | Download (4.3 kB)

1
// Copyright © 2016 ENS de Lyon, CNRS, University of Franche-Comté
2
// Licensed under the terms of the GNU General Public License (http://www.gnu.org/licenses)
3
// @author mdecorde
4
// @author sheiden
5
// STANDARD DECLARATIONS
6
package org.txm.macro.urs.edit
7

    
8
import org.apache.commons.lang.StringUtils
9
import org.kohsuke.args4j.*
10

    
11
import groovy.transform.Field
12

    
13
import org.txm.Toolbox
14
import org.txm.rcp.swt.widget.parameters.*
15
import org.txm.annotation.urs.*
16
import org.txm.macro.urs.AnalecUtils
17
import org.txm.searchengine.cqp.AbstractCqiClient
18
import org.txm.searchengine.cqp.corpus.*
19
import org.txm.searchengine.cqp.corpus.query.CQLQuery
20

    
21
import visuAnalec.donnees.Corpus
22
import visuAnalec.donnees.Structure
23
import visuAnalec.elements.*
24
import visuAnalec.vue.Vue
25

    
26
def scriptName = this.class.getSimpleName()
27

    
28
def selection = []
29
for (def s : corpusViewSelections) {
30
        if (s instanceof CQPCorpus) selection << s
31
        else if (s instanceof Partition) selection.addAll(s.getParts())
32
}
33

    
34
if (selection.size() == 0) {
35
        println "** $scriptName: please select a Corpus or a Partition in the Corpus view: "+corpusViewSelections
36
        return false
37
}
38

    
39
// BEGINNING OF PARAMETERS
40
@Field @Option(name="schema_ursql", usage="TYPE@PROP=VALUE", widget="String", required=true, def="CHAINE")
41
                String schema_ursql
42
@Field @Option(name="minimum_schema_size", usage="Minimum size needed to consider a schema", widget="Integer", required=true, def="3")
43
                int minimum_schema_size
44
@Field @Option(name="unit_ursql", usage="TYPE@PROP=VALUE", widget="String", required=false, def="MENTION")
45
                String unit_ursql
46
@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")
47
                int limit_distance_in_schema
48
@Field @Option(name="limit_cql", usage="CQL to build structure limits", widget="Query", required=true, def="<div> [] expand to div")
49
                limit_cql
50
@Field @Option(name="strict_inclusion", usage="Units must be strictly included into corpus matches", widget="Boolean", required=true, def="true")
51
                boolean strict_inclusion
52
@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")
53
                int limit_distance
54

    
55
@Field @Option(name="unit_property_to_set", usage="PROP", widget="String", required=false, def="TESTPROP")
56
                String unit_property_to_set
57
@Field @Option(name="unit_property_value_to_set", usage="VALUE", widget="String", required=false, def="TESTVALUE")
58
                String unit_property_value_to_set
59
@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF        ON        ALL        REALLY ALL", required=true, def="OFF")
60
                debug
61
if (!ParametersDialog.open(this)) return
62
        if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3
63

    
64
if (unit_property_to_set.length() == 0) {
65
        println "unit_property_to_set not set: aborting."
66
}
67

    
68
//corpus = corpusViewSelection
69
def allResults = [:]
70
def errors = new HashSet()
71
for (def corpus : selection) {
72
        println "Annotate $corpus..."
73

    
74
        def word = corpus.getWordProperty()
75
        def analecCorpus = URSCorpora.getCorpus(corpus)
76
        Vue analecView = URSCorpora.getVue(corpus)
77
        Structure structure = analecCorpus.getStructure()
78

    
79
        AnalecUtils.defineProperty(Unite.class, analecCorpus, unit_ursql, unit_property_to_set)
80
        analecView.initVueParDefaut()
81

    
82
        def selectedUnits = AnalecUtils.selectUnitsInSchema(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, Integer.MAX_VALUE,
83
                        unit_ursql, limit_distance_in_schema, limit_cql, strict_inclusion, limit_distance);
84
                
85
        println " "+selectedUnits.size()+" units to annotate..."
86
        def n = 0
87
        def nerrors = 0
88
        for (Unite unit : selectedUnits) {
89
                
90
                if (!analecView.setValeurChamp(unit, unit_property_to_set, unit_property_value_to_set)) {
91
                        errors << ""+unit.getDeb()+"->"+unit.getFin()
92
                        nerrors++
93
                } else {
94
                        n++
95
                }
96
        }
97
        
98
        corpus.getMainCorpus().setIsModified(true)
99
        allResults[corpus] = selectedUnits
100
        println " $n valeurs affectées à $corpus"
101
        println " $nerrors erreurs lors de l'annotation de $corpus"
102
}
103
if (errors.size() > 0) {
104
        println " Errors while annotating $unit_property_to_set=$unit_property_value_to_set"
105
        println errors.join("\n")
106
}
107
return allResults