Révision 3457
| TXM/trunk/org.txm.groovy.core/src/groovy/org/txm/macro/projects/antract/ImporterCorpusOkapiMacro.groovy (revision 3457) | ||
|---|---|---|
| 111 | 111 |
println "("+regex_identifiants_sujets.size()+")" + regex_identifiants_sujets
|
| 112 | 112 |
if (query.length() > maxCQLLimit) {
|
| 113 | 113 |
monitorShowError("Erreur : impossible de créer le sous-corpus, la requête CQL est trop longue (${query.length()}).")
|
| 114 |
return |
|
| 114 |
return
|
|
| 115 | 115 |
} |
| 116 |
|
|
| 116 | 117 |
} |
| 117 | 118 |
|
| 119 |
for (String id : liste_identifiants_sujets) {
|
|
| 120 |
if (!id.matches(regex_identifiants_sujets)) {
|
|
| 121 |
println "ERROR: $id not matching with regex" |
|
| 122 |
} else {
|
|
| 123 |
//println " OK: $id matching with regex" |
|
| 124 |
} |
|
| 125 |
} |
|
| 126 |
|
|
| 118 | 127 |
println "CQL = $query" |
| 119 | 128 |
|
| 120 | 129 |
Subcorpus sub = parentCorpus.createSubcorpus(new CQLQuery(query), nom_sous_corpus) |
| ... | ... | |
| 123 | 132 |
// println nom_sous_corpus+" est composé de "+nb_matches+" éléments." |
| 124 | 133 |
|
| 125 | 134 |
int s = sub.getSize() |
| 126 |
if (nb_matches == nb_identifiers) {
|
|
| 135 |
|
|
| 127 | 136 |
println "$nom_sous_corpus subcorpus is created." |
| 128 | 137 |
monitor.syncExec(new Runnable() {
|
| 129 | 138 |
public void run() {
|
| ... | ... | |
| 131 | 140 |
CorporaView.expand(parentCorpus); |
| 132 | 141 |
} |
| 133 | 142 |
}); |
| 134 |
} else {
|
|
| 135 |
println "Error: bad query, $nb_matches elements in sub-corpus instead of $nb_identifiers" |
|
| 136 |
monitorShowError("Error: bad query, $nb_matches elements in sub-corpus instead of $nb_identifiers")
|
|
| 143 |
|
|
| 144 |
if (nb_matches != nb_identifiers) {
|
|
| 145 |
println "Warning $nb_matches elements in sub-corpus instead of $nb_identifiers identifiers" |
|
| 146 |
// monitorShowError("Error: bad query, $nb_matches elements in sub-corpus instead of $nb_identifiers")
|
|
| 137 | 147 |
} |
| 138 | 148 |
|
| 139 | 149 |
// UTILS UTILS UTILS UTILS UTILS UTILS UTILS |
| ... | ... | |
| 143 | 153 |
breakNode(origtree, "", 0) |
| 144 | 154 |
def CQL1 = list.join("|")
|
| 145 | 155 |
def CQL2 = writeNode(origtree) |
| 156 |
//CQL2 = CQL2.replaceAll("\\(\\[()\\]\\)", "")
|
|
| 157 |
|
|
| 158 |
|
|
| 146 | 159 |
if (CQL1.length() < CQL2.length()) {
|
| 147 | 160 |
return CQL1; |
| 148 | 161 |
} else {
|
| 162 |
CQL2 = CQL2.replace("(\\d)", "\\d")
|
|
| 163 |
CQL2 = CQL2.replaceAll("\\(\\[([0-9])-([0-9])\\]\\)", "[\$1-\$2]") // ([0-9]) -> [0-9]
|
|
| 164 |
CQL2 = CQL2.replaceAll("\\(([0-9])\\|([0-9])\\)", "[\$1\$2]") // (1|2) -> [12]
|
|
| 149 | 165 |
return CQL2; |
| 150 | 166 |
} |
| 151 | 167 |
} |
| ... | ... | |
| 191 | 207 |
} |
| 192 | 208 |
|
| 193 | 209 |
if (ss.matches("([0-9]\\|){2,10}[0-9]")) ss = "["+ss.replace("|", "") +"]"
|
| 194 |
if (ss == "[0123456789]") ss = "[0-9]"
|
|
| 210 |
if (ss == "[0123456789]") ss = "\\d"
|
|
| 195 | 211 |
if (ss == "[012345678]") ss = "[0-8]" |
| 196 | 212 |
if (ss == "[01234567]") ss = "[0-7]" |
| 197 | 213 |
if (ss == "[0123456]") ss = "[0-6]" |
| 214 |
if (ss == "[012345]") ss = "[0-5]" |
|
| 198 | 215 |
if (ss == "[123456789]") ss = "[1-9]" |
| 199 | 216 |
if (ss == "[12345678]") ss = "[1-8]" |
| 200 | 217 |
if (ss == "[1234567]") ss = "[1-7]" |
| 201 | 218 |
if (ss == "[123456]") ss = "[1-6]" |
| 219 |
if (ss == "[12345]") ss = "[1-5]" |
|
| 202 | 220 |
if (ss == "[23456789]") ss = "[2-9]" |
| 203 | 221 |
if (ss == "[2345678]") ss = "[2-8]" |
| 204 | 222 |
if (ss == "[3456789]") ss = "[3-9]" |
| ... | ... | |
| 206 | 224 |
if (ss == "[456789]") ss = "[4-9]" |
| 207 | 225 |
if (ss == "[45678]") ss = "[4-8]" |
| 208 | 226 |
|
| 227 |
if (ss == "[012345689]") ss = "[0-689]" |
|
| 228 |
if (ss == "[012356789]") ss = "[0-35-9]"; |
|
| 229 |
if (ss == "[012456789]") ss = "[0-24-9]"; |
|
| 209 | 230 |
s += ss |
| 210 | 231 |
|
| 211 | 232 |
if (tree.size() > 1) s += ")" |
Formats disponibles : Unified diff