Statistics
| Revision:

root / tmp / org.txm.analec.rcp / src / org / txm / macro / urs / exploit / SchemasSummaryMacro.groovy @ 2105

History | View | Annotate | Download (3.4 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="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF        ON        ALL        REALLY ALL", required=true, def="OFF")
47
                debug
48
if (!ParametersDialog.open(this)) return
49
if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3
50

    
51
def CQI = CQPSearchEngine.getCqiClient()
52

    
53
def allresults = new LinkedHashMap()
54
def props = new HashSet()
55
for (def corpus : selection) {
56

    
57
        mainCorpus = corpus.getMainCorpus()
58

    
59
        def word = mainCorpus.getWordProperty()
60
        def analecCorpus = URSCorpora.getCorpus(mainCorpus)
61
        for (def type : analecCorpus.getStructure().getUnites())
62
                props.addAll(analecCorpus.getStructure().getUniteProperties(type));
63

    
64
        def selectedSchemas = AnalecUtils.selectSchemasInCorpus(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, maximum_schema_size);
65
        if (selectedSchemas.size() > 0)
66
                allresults[corpus] = selectedSchemas.sort(){it.getUnite0()};
67
}
68

    
69
// tabulate summary
70
def keys = allresults.keySet().sort();
71
println "Statistiques des schémas de "+corpusViewSelections.join(",")
72
println "\t"+keys.join("\t")
73

    
74
// number of schemas
75
print "Schemas"
76
for (def k : keys) {
77
        print "\t"+allresults[k].size();
78
}
79
println ""
80

    
81
// Average length
82
print "Average len"
83
for (def k : keys) {
84
        int total = 0;
85
        for (Schema s : allresults[k]) total+= s.contenu.size();
86
        
87
        print "\t"+total / allresults[k].size()
88
}
89
println ""
90

    
91
// median length
92
print "Median len"
93
for (def k : keys) {
94
        print "\t"+allresults[k][(int)(allresults[k].size() / 2)].contenu.size();
95
}
96
println ""
97

    
98
// number of properties values
99
for (def p : props) {
100
        print "$p"
101
        for (def k : keys) {
102
                def propsvalues = new HashSet()
103
                for (def u : allresults[k]) {
104
                        propsvalues.add(u.getProp(p))
105
                }
106
                if (debug) println "k @p values="+propsvalues
107
                print "\t"+propsvalues.size();
108
        }
109
        println ""
110
}
111