Révision 2717
tmp/org.txm.groovy.core/src/groovy/org/txm/macro/annotation/CQLList2WordPropertiesMacro.groovy (revision 2717) | ||
---|---|---|
25 | 25 |
@Field @Option(name="queries_file", usage="The CQL query to select tokens", widget="File", required=true, def="queries.tsv") |
26 | 26 |
File queries_file |
27 | 27 |
|
28 |
@Field @Option(name="annotation_type", usage="The annotation type", widget="String", required=true, def="type")
|
|
29 |
String annotation_type
|
|
28 |
@Field @Option(name="property", usage="The annotation type", widget="String", required=true, def="type")
|
|
29 |
String property
|
|
30 | 30 |
|
31 | 31 |
// Open the parameters input dialog box |
32 | 32 |
if (!ParametersDialog.open(this)) return |
... | ... | |
34 | 34 |
int total = 0; |
35 | 35 |
|
36 | 36 |
|
37 |
def hash = ["args": ["queries_file":queries_file, "annotation_type":annotation_type],
|
|
37 |
def hash = ["args": ["queries_file":queries_file, "property":property],
|
|
38 | 38 |
"corpusViewSelection":corpusViewSelection, "gse":gse |
39 | 39 |
] |
40 |
int n = gse.run(CQLList2AnnotationsMacro, hash) |
|
40 |
int n = gse.run(CQLList2WordAnnotationsMacro, hash)
|
|
41 | 41 |
|
42 | 42 |
if (n > 0) { |
43 | 43 |
println "Saving annotations..." |
tmp/org.txm.groovy.core/src/groovy/org/txm/macro/annotation/CQLList2WordAnnotationsMacro.groovy (revision 2717) | ||
---|---|---|
34 | 34 |
["query":split[1], "property":property, "annotation_value":split[0]], |
35 | 35 |
"corpusViewSelection":corpusViewSelection |
36 | 36 |
] |
37 |
int n = gse.run(CQL2AnnotationsMacro, hash) |
|
37 |
int n = gse.run(CQL2WordAnnotationsMacro, hash)
|
|
38 | 38 |
total += n |
39 | 39 |
} else { |
40 | 40 |
println "## ignoring "+line |
... | ... | |
43 | 43 |
|
44 | 44 |
println "$queries_file created $total annotations" |
45 | 45 |
|
46 |
return total |
|
47 |
package org.txm.macro.annotation |
|
48 |
|
|
49 |
// STANDARD DECLARATIONS |
|
50 |
import org.txm.rcpapplication.swt.widget.parameters.* |
|
51 |
import org.txm.searchengine.cqp.corpus.* |
|
52 |
import org.txm.rcp.views.corpora.CorporaView |
|
53 |
import org.txm.annotation.kr.core.KRAnnotationEngine |
|
54 |
import org.txm.annotation.kr.core.repository.* |
|
55 |
|
|
56 |
def scriptName = this.class.getSimpleName() |
|
57 |
|
|
58 |
if (!(corpusViewSelection instanceof CQPCorpus)) { |
|
59 |
println "** $scriptName: please select a Corpus in the Corpus view." |
|
60 |
return 0 |
|
61 |
} |
|
62 |
|
|
63 |
@Field @Option(name="queries_file", usage="The CQL query to select tokens", widget="File", required=true, def="queries.tsv") |
|
64 |
File queries_file |
|
65 |
|
|
66 |
@Field @Option(name="property", usage="The property to annotate", widget="String", required=true, def="type") |
|
67 |
String property |
|
68 |
|
|
69 |
// Open the parameters input dialog box |
|
70 |
if (!ParametersDialog.open(this)) return |
|
71 |
|
|
72 |
int total = 0; |
|
73 |
|
|
74 |
queries_file.eachLine("UTF-8") { line -> |
|
75 |
String[] split = line.split("\t", 2) |
|
76 |
if (!line.startsWith("#") && split.length == 2 && split[0].length() > 0 && split[1].length() >= 0) { |
|
77 |
|
|
78 |
def params = [ |
|
79 |
"query":split[1], "property":property, "value":split[0] |
|
80 |
] |
|
81 |
|
|
82 |
int n = gse.runMacro(CQL2AnnotationsMacro, params) |
|
83 |
total += n |
|
84 |
} else { |
|
85 |
println "## ignoring "+line |
|
86 |
} |
|
87 |
} |
|
88 |
|
|
89 |
println "$queries_file created $total annotations" |
|
90 |
|
|
91 | 46 |
return total |
tmp/org.txm.groovy.core/src/groovy/org/txm/macro/annotation/WordProperty2WordMacro.groovy (revision 2717) | ||
---|---|---|
31 | 31 |
@Field @Option(name="outputDirectory", usage="The CQL query to select tokens", widget="Folder", required=true, def="result_directory") |
32 | 32 |
File outputDirectory |
33 | 33 |
|
34 |
@Field @Option(name="property", usage="The annotation type", widget="String", required=true, def="type")
|
|
34 |
@Field @Option(name="property", usage="The property to move", widget="String", required=true, def="type")
|
|
35 | 35 |
String property |
36 | 36 |
|
37 | 37 |
//@Field @Option(name="property_value_to_ignore", usage="The annotation type", widget="String", required=true, def="__undef__") |
... | ... | |
84 | 84 |
} |
85 | 85 |
} |
86 | 86 |
xp.process(result) |
87 |
} |
|
88 |
package org.txm.macro.annotation |
|
89 |
|
|
90 |
// STANDARD DECLARATIONS |
|
91 |
import org.txm.rcpapplication.swt.widget.parameters.* |
|
92 |
import org.txm.searchengine.cqp.corpus.* |
|
93 |
import org.txm.xml.DOMIdentityHook |
|
94 |
import org.txm.xml.LocalNameHookActivator |
|
95 |
import org.txm.xml.XMLProcessor |
|
96 |
import org.txm.objects.* |
|
97 |
import org.txm.rcp.utils.JobHandler |
|
98 |
import org.txm.rcp.views.corpora.CorporaView |
|
99 |
import org.eclipse.core.runtime.Status |
|
100 |
import org.txm.annotation.kr.core.KRAnnotationEngine |
|
101 |
import org.txm.annotation.kr.core.repository.* |
|
102 |
import org.txm.annotation.kr.rcp.commands.SaveAnnotations |
|
103 |
import org.txm.annotation.kr.rcp.concordance.WordAnnotationToolbar |
|
104 |
import org.txm.importer.StaxIdentityParser |
|
105 |
|
|
106 |
def scriptName = this.class.getSimpleName() |
|
107 |
|
|
108 |
if (!(corpusViewSelection instanceof MainCorpus)) { |
|
109 |
println "** $scriptName: please select a Corpus in the Corpus view." |
|
110 |
return 0 |
|
111 |
} |
|
112 |
|
|
113 |
MainCorpus mcorpus = corpusViewSelection |
|
114 |
Project project = mcorpus.getProject() |
|
115 |
File txmDir = new File(project.getProjectDirectory(), "txm/"+mcorpus.getName()) |
|
116 |
|
|
117 |
@Field @Option(name="outputDirectory", usage="The CQL query to select tokens", widget="Folder", required=true, def="result_directory") |
|
118 |
File outputDirectory |
|
119 |
|
|
120 |
@Field @Option(name="property", usage="The property used to create words", widget="String", required=true, def="type") |
|
121 |
String property |
|
122 |
|
|
123 |
// Open the parameters input dialog box |
|
124 |
if (!ParametersDialog.open(this)) return |
|
125 |
|
|
126 |
if (!outputDirectory.exists()) { |
|
127 |
println "Error: output directory not found: "+outputDirectory |
|
128 |
return false |
|
129 |
} |
|
130 |
|
|
131 |
for (Text text : project.getChildren(Text.class)) { |
|
132 |
println "Text: "+text |
|
133 |
|
|
134 |
File orig = text.getXMLTXMFile() |
|
135 |
if (!orig.exists()) { |
|
136 |
println "Warning: no XML-TXM file found for "+text |
|
137 |
continue // next ! |
|
138 |
} |
|
139 |
|
|
140 |
File result = new File(outputDirectory, orig.getName()) |
|
141 |
|
|
142 |
if (result.lastModified() > orig.lastModified()) { |
|
143 |
println "Skipping $text: result file is more recent: $result" |
|
144 |
continue // next ! |
|
145 |
} |
|
146 |
|
|
147 |
XMLProcessor xp = new XMLProcessor(orig); |
|
148 |
LocalNameHookActivator activator = new LocalNameHookActivator("w"); |
|
149 |
new DOMIdentityHook("word", activator, xp) { |
|
150 |
@Override |
|
151 |
public void processDom() { |
|
152 |
println dom.ana.size() |
|
153 |
} |
|
154 |
} |
|
155 |
xp.process(result) |
|
156 | 87 |
} |
Formats disponibles : Unified diff