Statistics
| Revision:

root / tmp / org.txm.analec.rcp / src / org / txm / macro / urs / exploit / SchemasSummaryMacro.groovy @ 2144

History | View | Annotate | Download (3.6 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.exploit
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.CQPSearchEngine
20

    
21
import visuAnalec.donnees.Structure
22
import visuAnalec.elements.*
23

    
24
def scriptName = this.class.getSimpleName()
25

    
26
def selection = []
27
for (def s : corpusViewSelections) {
28
        if (s instanceof CQPCorpus) selection << s
29
        else if (s instanceof Partition) selection.addAll(s.getParts())
30
}
31

    
32
if (selection.size() == 0) {
33
        println "** $scriptName: please select a Corpus or a Partition in the Corpus view: "+corpusViewSelections
34
        return false
35
} else {
36
        for (def c : selection) c.compute(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="maximum_schema_size", usage="Maximum size needed to consider a schema", widget="Integer", required=true, def="9999999")
45
                int maximum_schema_size
46
@Field @Option(name="strict_inclusion", usage="Units must be strictly included into corpus matches", widget="Boolean", required=true, def="true")
47
                boolean strict_inclusion
48
@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF        ON        ALL        REALLY ALL", required=true, def="OFF")
49
                debug
50
if (!ParametersDialog.open(this)) return
51
if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3
52

    
53
def CQI = CQPSearchEngine.getCqiClient()
54

    
55
def allresults = new LinkedHashMap()
56
def props = new HashSet()
57
for (def corpus : selection) {
58

    
59
        mainCorpus = corpus.getMainCorpus()
60

    
61
        def word = mainCorpus.getWordProperty()
62
        def analecCorpus = URSCorpora.getCorpus(mainCorpus)
63
        for (def type : analecCorpus.getStructure().getUnites())
64
                props.addAll(analecCorpus.getStructure().getUniteProperties(type));
65

    
66
        def selectedSchemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, maximum_schema_size, strict_inclusion);
67
        if (selectedSchemas.size() > 0)
68
                allresults[corpus] = selectedSchemas.sort(){it.getUnite0()};
69
}
70

    
71
// tabulate summary
72
def keys = allresults.keySet().sort();
73
println "Statistiques des schémas de "+corpusViewSelections.join(",")
74
println "\t"+keys.join("\t")
75

    
76
// number of schemas
77
print "Schemas"
78
for (def k : keys) {
79
        print "\t"+allresults[k].size();
80
}
81
println ""
82

    
83
// Average length
84
print "Average len"
85
for (def k : keys) {
86
        int total = 0;
87
        for (Schema s : allresults[k]) total+= s.contenu.size();
88
        
89
        print "\t"+total / allresults[k].size()
90
}
91
println ""
92

    
93
// median length
94
print "Median len"
95
for (def k : keys) {
96
        print "\t"+allresults[k][(int)(allresults[k].size() / 2)].contenu.size();
97
}
98
println ""
99

    
100
// number of properties values
101
for (def p : props) {
102
        print "$p"
103
        for (def k : keys) {
104
                def propsvalues = new HashSet()
105
                for (def u : allresults[k]) {
106
                        propsvalues.add(u.getProp(p))
107
                }
108
                if (debug) println "k @p values="+propsvalues
109
                print "\t"+propsvalues.size();
110
        }
111
        println ""
112
}
113