Statistics
| Revision:

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

History | View | Annotate | Download (3.9 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="unit_ursql", usage="TYPE@PROP=VALUE", widget="String", required=false, def="MENTION")
47
                String unit_ursql
48
@Field @Option(name="position_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")
49
                int position_in_schema
50
@Field @Option(name="cql_limit", usage="CQL to build structure limits", widget="Query", required=false, def="")
51
                cql_limit
52
@Field @Option(name="strict_inclusion", usage="Units must be strictly included into corpus matches", widget="Boolean", required=false, def="true")
53
                boolean strict_inclusion
54
@Field @Option(name="position_in_matches", usage="Unit distance to structure limit (0 = no selection, 1 = first after limit, -1 = last before limit, etc.)", widget="Integer", required=true, def="0")
55
                int position_in_matches
56
@Field @Option(name="debug", usage="Show internal variable content", widget="StringArray", metaVar="OFF        ON        ALL        REALLY ALL", required=true, def="OFF")
57
                debug
58
if (!ParametersDialog.open(this)) return
59
if (debug == "OFF") debug = 0; else if (debug == "ON") debug = 1; else if (debug == "ALL") debug = 2 else if (debug == "REALLY ALL") debug = 3
60

    
61
def CQI = CQPSearchEngine.getCqiClient()
62

    
63
def allresults = new LinkedHashMap()
64
def props = new HashSet()
65
for (def corpus : selection) {
66

    
67
        mainCorpus = corpus.getMainCorpus()
68

    
69
        def word = mainCorpus.getWordProperty()
70
        def analecCorpus = URSCorpora.getCorpus(mainCorpus)
71
        for (def type : analecCorpus.getStructure().getUnites())
72
                props.addAll(analecCorpus.getStructure().getUniteProperties(type));
73

    
74
        def selectedUnits = AnalecUtils.selectUnitsInSchema(debug, analecCorpus, corpus, schema_ursql, minimum_schema_size, maximum_schema_size, 
75
        unit_ursql, position_in_schema, cql_limit, strict_inclusion, position_in_matches);
76

    
77
        allresults[corpus] = selectedUnits;
78
}
79

    
80
// tabulate summary
81
def keys = allresults.keySet().sort();
82
println "Statistiques des unités de "+corpusViewSelections.join(",")
83
println "\t"+keys.join("\t")
84
print "Units"
85
for (def k : keys) {
86
        print "\t"+allresults[k].size();
87
}
88
println ""
89
for (def p : props) {
90
        print "$p"
91
        for (def k : keys) {
92
                def propsvalues = new HashSet()
93
                for (def u : allresults[k]) {
94
                        propsvalues.add(u.getProp(p))
95
                }
96
                if (debug) println "k @p values="+propsvalues
97
                print "\t"+propsvalues.size();
98
        }
99
        println ""
100
}
101