Revision 2167 tmp/org.txm.analec.rcp/src/org/txm/macro/urs/AnalecUtils.groovy
| AnalecUtils.groovy (revision 2167) | ||
|---|---|---|
| 11 | 11 |
|
| 12 | 12 |
|
| 13 | 13 |
static def isPropertyDefined(Class clazz, Corpus analecCorpus, String ursql) {
|
| 14 |
if (ursql == null || ursql.length() == 0) return new HashSet()
|
|
| 14 |
if (ursql == null || ursql.length() == 0) return new HashSet() |
|
| 15 | 15 |
def params = getFilterParameters(ursql) |
| 16 | 16 |
def typeRegexp = params[0] |
| 17 | 17 |
def propRegexp = params[1] |
| 18 |
println "params=$params" |
|
| 19 | 18 |
return isPropertyDefined(clazz, analecCorpus, typeRegexp, propRegexp) |
| 20 | 19 |
} |
| 21 | 20 |
|
| ... | ... | |
| 56 | 55 |
if (maximum_schema_size <= 0) maximum_schema_size = Integer.MAX_VALUE; |
| 57 | 56 |
if (minimum_schema_size < 0) minimum_schema_size = 0; |
| 58 | 57 |
def allSchemas = [] |
| 59 |
|
|
| 58 |
|
|
| 60 | 59 |
if (schema_ursql != null && schema_ursql.length() > 0) allSchemas = AnalecUtils.findAllInCorpus(debug, analecCorpus, Schema.class, schema_ursql) |
| 61 | 60 |
else allSchemas = analecCorpus.getTousSchemas() |
| 62 | 61 |
|
| 63 | 62 |
if (debug >= 2) println "allSchemas=${allSchemas.size()}"
|
| 64 | 63 |
allSchemas = AnalecUtils.filterBySize(allSchemas, minimum_schema_size, maximum_schema_size); |
| 65 |
|
|
| 64 |
|
|
| 66 | 65 |
return allSchemas |
| 67 | 66 |
} |
| 68 | 67 |
|
| 69 | 68 |
static def selectSchemasInCorpus(def debug, Corpus analecCorpus, org.txm.searchengine.cqp.corpus.CQPCorpus corpus, |
| 70 |
String schema_ursql, Integer minimum_schema_size, Integer maximum_schema_size, boolean strictInclusion) {
|
|
| 71 |
|
|
| 69 |
String schema_ursql, Integer minimum_schema_size, Integer maximum_schema_size, boolean strictInclusion) {
|
|
| 70 |
|
|
| 72 | 71 |
if (maximum_schema_size <= 0) maximum_schema_size = Integer.MAX_VALUE; |
| 73 | 72 |
if (minimum_schema_size < 0) minimum_schema_size = 0; |
| 74 |
|
|
| 73 |
|
|
| 75 | 74 |
def allSchemas = [] |
| 76 | 75 |
if (schema_ursql != null && schema_ursql.length() > 0) allSchemas = AnalecUtils.findAllInCorpus(debug, analecCorpus, Schema.class, schema_ursql) |
| 77 | 76 |
else allSchemas = analecCorpus.getTousSchemas() |
| 78 |
|
|
| 77 |
|
|
| 79 | 78 |
def selectedSchemas = [] |
| 80 | 79 |
for (Schema schema : allSchemas) {
|
| 81 | 80 |
def selectedUnits = AnalecUtils.filterUniteByInclusion(debug, schema.getUnitesSousjacentes(), corpus.getMatches(), strictInclusion, 0) |
| 82 |
|
|
| 81 |
|
|
| 83 | 82 |
if (minimum_schema_size <= selectedUnits.size() && selectedUnits.size() <= maximum_schema_size ) {
|
| 84 | 83 |
selectedSchemas << schema |
| 85 | 84 |
} |
| 86 | 85 |
} |
| 87 |
|
|
| 86 |
|
|
| 88 | 87 |
return selectedSchemas |
| 89 | 88 |
} |
| 90 | 89 |
|
| ... | ... | |
| 107 | 106 |
String schema_ursql, Integer minimum_schema_size, Integer maximum_schema_size, |
| 108 | 107 |
String unit_ursql, Integer position_in_schema, CQLQuery cql_limit, Boolean strict_inclusion, int position_in_matches) {
|
| 109 | 108 |
def groupedUnits = [] |
| 110 |
if (schema_ursql != null && schema_ursql.length() > 0 || minimum_schema_size > 1) {
|
|
| 109 |
if (schema_ursql != null && schema_ursql.length() > 0) {
|
|
| 111 | 110 |
def allSchema = null; |
| 112 | 111 |
|
| 113 | 112 |
if (schema_ursql != null && schema_ursql.length() > 0) allSchema = AnalecUtils.findAllInCorpus(debug, analecCorpus, Schema.class, schema_ursql) |
| 114 | 113 |
else allSchema = analecCorpus.getTousSchemas() |
| 115 | 114 |
if (debug >= 2) println "allSchema=${allSchema.size()}"
|
| 116 | 115 |
|
| 117 |
allSchema = AnalecUtils.filterBySize(allSchema, minimum_schema_size, maximum_schema_size); |
|
| 118 |
if (debug >= 2) println "allSchema=${allSchema.size()}"
|
|
| 119 |
|
|
| 120 | 116 |
groupedUnits = AnalecUtils.groupAllUnitesInElements(debug, allSchema, unit_ursql) |
| 117 |
if (debug >= 2) println "groupedUnits=${groupedUnits.size()}"
|
|
| 121 | 118 |
|
| 122 |
if (position_in_schema >= 0) groupedUnits = AnalecUtils.filterUniteByInclusionInSchema(debug, groupedUnits, position_in_schema) |
|
| 119 |
groupedUnits = AnalecUtils.filterUniteByInclusionInSchema(debug, groupedUnits, position_in_schema) |
|
| 120 |
if (debug >= 2) println "groupedUnits=${groupedUnits.size()}"
|
|
| 123 | 121 |
|
| 122 |
|
|
| 124 | 123 |
} else {
|
| 125 | 124 |
groupedUnits = ["all":AnalecUtils.findAllInCorpus(debug, analecCorpus, Unite.class, unit_ursql)] |
| 126 | 125 |
} |
| ... | ... | |
| 139 | 138 |
def allUnits = [] |
| 140 | 139 |
for (def k : groupedUnits.keySet()) {
|
| 141 | 140 |
def selectedUnits = AnalecUtils.filterUniteByInclusion(debug, groupedUnits[k], matches, strict_inclusion, position_in_matches) |
| 142 |
allUnits.addAll(selectedUnits) |
|
| 141 |
|
|
| 142 |
if (minimum_schema_size <= selectedUnits.size() && selectedUnits.size() <= maximum_schema_size ) {
|
|
| 143 |
allUnits.addAll(selectedUnits) |
|
| 144 |
} else {
|
|
| 145 |
|
|
| 146 |
} |
|
| 143 | 147 |
} |
| 144 | 148 |
if (debug >= 2) println "selectedUnits=${allUnits.size()}"
|
| 145 | 149 |
|
| 146 | 150 |
Collections.sort(allUnits) |
| 147 |
|
|
| 151 |
|
|
| 148 | 152 |
return allUnits |
| 149 | 153 |
} |
| 150 | 154 |
/** |
| ... | ... | |
| 156 | 160 |
* @return |
| 157 | 161 |
*/ |
| 158 | 162 |
static def filterUniteByInclusionInSchema(def debug, def groups, Integer distance) {
|
| 163 |
println "dist=$distance" |
|
| 159 | 164 |
if (distance == 0) return groups; |
| 160 | 165 |
if (distance > 0) distance = distance-1; |
| 161 | 166 |
def newGroups = [:] |
| ... | ... | |
| 166 | 171 |
continue; |
| 167 | 172 |
} |
| 168 | 173 |
def indexes = null |
| 169 |
if (distance > 0) {
|
|
| 174 |
if (distance >= 0) {
|
|
| 170 | 175 |
indexes = 0..Math.min(distance, group.size()) |
| 171 | 176 |
} else {
|
| 172 | 177 |
indexes = Math.max(distance, -group.size())..-1 |
| 173 | 178 |
} |
| 179 |
|
|
| 174 | 180 |
newGroups[k] = group[indexes]; |
| 175 | 181 |
} |
| 176 | 182 |
return newGroups |
| ... | ... | |
| 253 | 259 |
|
| 254 | 260 |
def filteredElements = [] |
| 255 | 261 |
for (Element e : elements) {
|
| 256 |
Unite[] units = e.getUnitesSousjacentes();
|
|
| 257 |
int size = units.length;
|
|
| 258 |
if (size < minimum_schema_size) continue;
|
|
| 259 |
if (size > maximum_schema_size) continue;
|
|
| 260 |
filteredElements << e;
|
|
| 262 |
Unite[] selectedUnits = e.getUnitesSousjacentes();
|
|
| 263 |
int size = selectedUnits.length;
|
|
| 264 |
if (minimum_schema_size <= selectedUnits.size() && selectedUnits.size() <= maximum_schema_size ) {
|
|
| 265 |
filteredElements << e
|
|
| 266 |
}
|
|
| 261 | 267 |
} |
| 262 | 268 |
return filteredElements |
| 263 | 269 |
} |
| ... | ... | |
| 288 | 294 |
selectedUnitsPerMatch[iCurrentMatch] = selectedUnits |
| 289 | 295 |
|
| 290 | 296 |
while (iCurrentMatch < matchesSize && iCurrentUnit < unitsSize) {
|
| 291 |
if (debug >= 2) println "** M $iCurrentMatch < $matchesSize && U $iCurrentUnit < $unitsSize"
|
|
| 297 |
if (debug >= 3) println "** M $iCurrentMatch < $matchesSize && U $iCurrentUnit < $unitsSize"
|
|
| 292 | 298 |
|
| 293 | 299 |
Unite unit = allUnites[iCurrentUnit] |
| 294 | 300 |
Match match = matches[iCurrentMatch] |
| 295 | 301 |
if (debug >= 3) println ""+unit.getDeb()+"->"+unit.getFin()+" "+match.getStart()+"->"+match.getEnd() |
| 296 | 302 |
if (unit.getFin() < match.getStart()) {
|
| 297 | 303 |
if (debug >= 3) "println next unit" |
| 298 |
|
|
| 299 |
iCurrentUnit++ |
|
| 304 |
|
|
| 305 |
iCurrentUnit++
|
|
| 300 | 306 |
} else if (unit.getDeb() > match.getEnd()) {
|
| 301 | 307 |
if (debug >= 3) "println next match" |
| 302 |
|
|
| 303 |
iCurrentMatch++ |
|
| 308 |
|
|
| 309 |
iCurrentMatch++
|
|
| 304 | 310 |
selectedUnits = [] |
| 305 | 311 |
selectedUnitsPerMatch[iCurrentMatch] = selectedUnits |
| 306 | 312 |
} else {
|
| ... | ... | |
| 407 | 413 |
if (!eq) {
|
| 408 | 414 |
equal_start_idx-- |
| 409 | 415 |
} |
| 410 |
|
|
| 416 |
|
|
| 411 | 417 |
if (atidx >= 0 && equal_start_idx >= 0 && atidx < equal_start_idx) { // TYPE@PROP=VALUE
|
| 412 | 418 |
type = URSQL.substring(0, atidx) |
| 413 | 419 |
prop = URSQL.substring(atidx+1, equal_start_idx) |
| ... | ... | |
| 432 | 438 |
} |
| 433 | 439 |
|
| 434 | 440 |
static def filterElements(def debug, def allElements, String typeRegex, String propName, boolean eq, String valueRegex) {
|
| 435 |
if (debug >= 2) println "filtering "+allElements.size()+" elements with typeRegex='$typeRegex' propName='$propName' and valueRegex='$valueRegex'"
|
|
| 441 |
if (debug >= 3) println "filtering "+allElements.size()+" elements with typeRegex='$typeRegex' propName='$propName' and valueRegex='$valueRegex'"
|
|
| 436 | 442 |
if (typeRegex != null && typeRegex.length() > 0) {
|
| 437 | 443 |
def filteredElements = [] |
| 438 | 444 |
def matcher = /$typeRegex/ |
| ... | ... | |
| 444 | 450 |
|
| 445 | 451 |
allElements = filteredElements; |
| 446 | 452 |
} |
| 447 |
if (debug >= 2) println " type step result: "+allElements.size()
|
|
| 453 |
if (debug >= 3) println " type step result: "+allElements.size()
|
|
| 448 | 454 |
|
| 449 | 455 |
if (propName != null && propName.length() > 0) {
|
| 450 | 456 |
def filteredElements = [] |
| ... | ... | |
| 468 | 474 |
|
| 469 | 475 |
allElements = filteredElements; |
| 470 | 476 |
} |
| 471 |
if (debug >= 2) println " prop&value step result: "+allElements.size()
|
|
| 477 |
if (debug >= 3) println " prop&value step result: "+allElements.size()
|
|
| 472 | 478 |
return allElements; |
| 473 | 479 |
} |
| 474 | 480 |
|
Also available in: Unified diff