public static void parseOutputAggregation( AdvancedQueryPojo.QueryOutputPojo.AggregationOutputPojo aggregation, AliasLookupTable aliasLookup, boolean geoLowAccuracy, String[] entTypeFilterStrings, String[] assocVerbFilterStrings, SearchRequestBuilder searchSettings, BoolFilterBuilder parentFilterObj, String[] communityIdStrs) { // 1.] Go through aggregation list // 1.1] Apply "simple specifications" if necessary // Geo if ((null != aggregation) && (null != aggregation.geoNumReturn) && (aggregation.geoNumReturn > 0)) { CrossVersionFacetBuilder.TermsFacetBuilder fb = CrossVersionFacetBuilders.termsFacet("geo") .field(DocumentPojo.locs_) .size(aggregation.geoNumReturn); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); } // (TESTED) // Temporal if ((null != aggregation) && (null != aggregation.timesInterval)) { if (aggregation.timesInterval.contains("m")) { aggregation.timesInterval = "month"; } CrossVersionFacetBuilder.DateHistogramFacetBuilder fb = CrossVersionFacetBuilders.dateHistogramFacet("time") .field(DocumentPojo.publishedDate_) .interval(aggregation.timesInterval); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); // TODO (INF-2688): if using certain types of moments then don't want this? } // (TESTED) // Temporal Moments if ((null != aggregation) && (null != aggregation.moments)) { if (null == aggregation.moments.timesInterval) { if (null != aggregation.timesInterval) { aggregation.moments.timesInterval = aggregation.timesInterval; } else { aggregation.moments.timesInterval = "m"; } } if (aggregation.moments.timesInterval.contains("m")) { aggregation.moments.timesInterval = "month"; } // TODO (INF-2688): Other cross filter type things if (!geoLowAccuracy && (null != aggregation.moments.geoNumReturn) && (aggregation.moments.geoNumReturn > 0)) { DateHistogramBuilder timeAgg = AggregationBuilders.dateHistogram("moments") .field(DocumentPojo.publishedDate_) .interval(new Interval(aggregation.moments.timesInterval)); TermsBuilder geoAgg = AggregationBuilders.terms("geo") .field(DocumentPojo.locs_) .size(aggregation.moments.geoNumReturn); timeAgg.subAggregation(geoAgg); searchSettings.addAggregation(timeAgg); } // TODO (CORE-89) if (null != aggregation.moments.associationsNumReturn && aggregation.moments.associationsNumReturn >= 0) { // TODO need to check if indexes mapping use doc.associations.assoc_index == docValue // fail out or don't include those communities if they don't if (validateAssociationMapping(communityIdStrs)) { DateHistogramBuilder assocTimeAgg = AggregationBuilders.dateHistogram("moments.assoc") .field(DocumentPojo.publishedDate_) .interval(new Interval(aggregation.moments.timesInterval)); TermsBuilder assocAgg = AggregationBuilders.terms("assoc") .field(AssociationPojo.assoc_index_) .size(aggregation.moments.associationsNumReturn); NestedBuilder nested = AggregationBuilders.nested("moments.assoc.nested") .path(DocumentPojo.associations_) .subAggregation(assocAgg); assocTimeAgg.subAggregation(nested); searchSettings.addAggregation(assocTimeAgg); } } if (null != aggregation.moments.entityList) { for (String entIndex : aggregation.moments.entityList) { CrossVersionFacetBuilder.DateHistogramFacetBuilder fb = CrossVersionFacetBuilders.dateHistogramFacet("moments." + entIndex) .field(DocumentPojo.publishedDate_) .interval(aggregation.moments.timesInterval); EntityFeaturePojo alias = null; if (null != aliasLookup) { alias = aliasLookup.getAliases(entIndex); } if (null == alias) { // no alias fb = fb.facetFilter( FilterBuilders.nestedFilter( DocumentPojo.entities_, FilterBuilders.termFilter(EntityPojo.index_, entIndex))); } // TESTED else { QueryFilterBuilder qfb = null; if ((null != alias.getSemanticLinks()) && !alias.getSemanticLinks().isEmpty()) { BoolQueryBuilder qb = QueryBuilders.boolQuery(); for (String textAlias : alias.getSemanticLinks()) { qb = qb.should( CrossVersionQueryBuilders.matchPhraseQuery( DocumentPojo.fullText_, textAlias)); } qfb = FilterBuilders.queryFilter(qb); } // TESTED if (!alias.getAlias().isEmpty()) { NestedFilterBuilder nfb = FilterBuilders.nestedFilter( DocumentPojo.entities_, FilterBuilders.termsFilter(EntityPojo.index_, entIndex, alias.getAlias())); if (null == qfb) { fb = fb.facetFilter(nfb); } // TESTED else { BoolFilterBuilder bfb = FilterBuilders.boolFilter().should(nfb).should(qfb); fb = fb.facetFilter(bfb); } // TESTED } else if (null != qfb) { fb = fb.facetFilter(qfb); } // TESTED } // TESTED // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); } } // (end list over entities) } // TESTED // Entities - due to problems with significance, handled on a document by document basis, see // Significance helper class // Associations (Events/Facts) // Association verb category filter StringBuilder verbCatRegex = null; StringBuilder entTypeRegex = null; if (((null != aggregation) && (null != aggregation.eventsNumReturn) && (aggregation.eventsNumReturn > 0)) || ((null != aggregation) && (null != aggregation.factsNumReturn) && (aggregation.factsNumReturn > 0))) { if (null != entTypeFilterStrings) { boolean bNegative = false; if ('-' != entTypeFilterStrings[0].charAt(0)) { // positive filtering entTypeRegex = new StringBuilder("(?:"); } else { bNegative = true; entTypeRegex = new StringBuilder("(?!"); // (this is a lookahead but will be fine because of the .*/ in front of it) } for (String entType : entTypeFilterStrings) { if (bNegative && ('-' == entType.charAt(0))) { entType = entType.substring(1); } entType = entType.replace("|", "%7C"); entTypeRegex.append(".*?/").append(Pattern.quote(entType.toLowerCase())).append('|'); // (can't match greedily because of the 2nd instance of entity type) } entTypeRegex.setLength(entTypeRegex.length() - 1); // (remove trailing |) entTypeRegex.append(")"); if (bNegative) { entTypeRegex.append("[^|]*"); // (now the actual verb, if a -ve lookahead) } } // TESTED if (null != assocVerbFilterStrings) { boolean bNegative = false; if ('-' != assocVerbFilterStrings[0].charAt(0)) { // positive filtering verbCatRegex = new StringBuilder("\\|(?:"); } else { bNegative = true; verbCatRegex = new StringBuilder("\\|(?!"); // (this is a lookahead but will be fine because of the "^[^|]*\\" in front of it) // eg say I have -VERB then subject|VERB|object will match because if the } for (String assocVerbFilterString : assocVerbFilterStrings) { if (bNegative && ('-' == assocVerbFilterString.charAt(0))) { assocVerbFilterString = assocVerbFilterString.substring(1); } assocVerbFilterString = assocVerbFilterString.replace("|", "%7C"); verbCatRegex.append(Pattern.quote(assocVerbFilterString)).append('|'); } verbCatRegex.setLength(verbCatRegex.length() - 1); // (remove trailing |) verbCatRegex.append(")"); if (bNegative) { verbCatRegex.append("[^|]*"); // (now the actual verb, if a -ve lookahead) } } // TESTED } // TESTED (all combinations of 1/2 people, 1/2 verbs) if ((null != aggregation) && (null != aggregation.eventsNumReturn) && (aggregation.eventsNumReturn > 0)) { StringBuffer regex = new StringBuffer("^Event\\|"); if (null != entTypeRegex) { regex.append(entTypeRegex); } else { regex.append("[^|]*"); } if (null != verbCatRegex) { regex.append(verbCatRegex); } else if (null != entTypeRegex) { regex.append("\\|[^|]*"); } else { regex.append(".*"); } if (null != entTypeRegex) { regex.append("\\|").append(entTypeRegex); regex.append(".*"); } else { regex.append("\\|.*"); } // DEBUG // System.out.println("REGEX==" + regex.toString()); // TESTED (all combinations of 1/2 people, 1/2 verbs) CrossVersionFacetBuilder.TermsFacetBuilder fb = CrossVersionFacetBuilders.termsFacet("events") .field(AssociationPojo.assoc_index_) .size(aggregation.eventsNumReturn) .nested(DocumentPojo.associations_); fb.regex(regex.toString()); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); } if ((null != aggregation) && (null != aggregation.factsNumReturn) && (aggregation.factsNumReturn > 0)) { StringBuffer regex = new StringBuffer("^Fact\\|"); if (null != entTypeRegex) { regex.append(entTypeRegex); } else { regex.append("[^|]*"); } if (null != verbCatRegex) { regex.append(verbCatRegex); } else if (null != entTypeRegex) { regex.append("\\|[^|]*"); } else { regex.append(".*"); } if (null != entTypeRegex) { regex.append("\\|").append(entTypeRegex); regex.append(".*"); } else { regex.append("\\|.*"); } // DEBUG // System.out.println("REGEX==" + regex.toString()); // TESTED (all combinations of 1/2 people, 1/2 verbs) CrossVersionFacetBuilder.TermsFacetBuilder fb = CrossVersionFacetBuilders.termsFacet("facts") .field(AssociationPojo.assoc_index_) .size(aggregation.factsNumReturn) .nested(DocumentPojo.associations_); fb.regex(regex.toString()); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); } // Source management/monitoring if ((null != aggregation) && (null != aggregation.sourceMetadata) && (aggregation.sourceMetadata > 0)) { CrossVersionFacetBuilder.TermsFacetBuilder fb = CrossVersionFacetBuilders.termsFacet("sourceTags") .field(DocumentPojo.tags_) .size(aggregation.sourceMetadata) .facetFilter(parentFilterObj); CrossVersionFacetBuilder.TermsFacetBuilder fb1 = CrossVersionFacetBuilders.termsFacet("sourceTypes") .field(DocumentPojo.mediaType_) .size(aggregation.sourceMetadata) .facetFilter(parentFilterObj); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); fb1 = fb1.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); searchSettings.addFacet(fb1); } if ((null != aggregation) && (null != aggregation.sources) && (aggregation.sources > 0)) { CrossVersionFacetBuilder.TermsFacetBuilder fb = CrossVersionFacetBuilders.termsFacet("sourceKeys") .field(DocumentPojo.sourceKey_) .size(aggregation.sources); // Gross raw handling for facets if (null != parentFilterObj) { fb = fb.facetFilter(parentFilterObj); } searchSettings.addFacet(fb); } } // TESTED
/** * Updates the feature entries for the list of entities that was just extracted including changing * frequency, adding aliases etc * * <p>This method now has 3 steps: 1. Try to update alias 1.a If fail, create new gaz 2. Update * totalfreq and doccount * * @param ents List of entities to update in the entity feature */ public static void updateEntityFeatures( Map<String, Map<ObjectId, EntityFeaturePojo>> entFeatures) { DBCollection col = DbManager.getFeature().getEntity(); String savedSyncTime = null; for (Map<ObjectId, EntityFeaturePojo> entCommunity : entFeatures.values()) { Iterator<Map.Entry<ObjectId, EntityFeaturePojo>> it = entCommunity.entrySet().iterator(); while (it.hasNext()) { Map.Entry<ObjectId, EntityFeaturePojo> entFeatureKV = it.next(); try { EntityFeaturePojo entFeature = entFeatureKV.getValue(); long nSavedDocCount = entFeature.getDoccount(); long nSavedFreqCount = entFeature.getTotalfreq(); // (these should be constant across all communities but keep it here // so can assign it using entFeature, it's v cheap so no need to get once like for sync // vars) ObjectId communityID = entFeature.getCommunityId(); if (null != communityID) { // For each community, see if the entity feature already exists *for that community* BasicDBObject query = new BasicDBObject(EntityFeaturePojo.index_, entFeature.getIndex()); query.put(EntityFeaturePojo.communityId_, communityID); BasicDBObject updateOp = new BasicDBObject(); // Add aliases: BasicDBObject updateOpA = new BasicDBObject(); BasicDBObject multiopE = new BasicDBObject(MongoDbManager.each_, entFeature.getAlias()); updateOpA.put(EntityFeaturePojo.alias_, multiopE); // Add link data, if there is any: if ((null != entFeature.getSemanticLinks()) && !entFeature.getSemanticLinks().isEmpty()) { BasicDBObject multiopF = new BasicDBObject(MongoDbManager.each_, entFeature.getSemanticLinks()); updateOpA.put(EntityFeaturePojo.linkdata_, multiopF); } updateOp.put(MongoDbManager.addToSet_, updateOpA); // Update frequency: BasicDBObject updateOpB = new BasicDBObject(); updateOpB.put(EntityFeaturePojo.totalfreq_, nSavedFreqCount); updateOpB.put(EntityFeaturePojo.doccount_, nSavedDocCount); updateOp.put(MongoDbManager.inc_, updateOpB); // try to use find/modify to see if something comes back and set doc freq/totalfreq BasicDBObject fields = new BasicDBObject(EntityFeaturePojo.totalfreq_, 1); fields.put(EntityFeaturePojo.doccount_, 1); fields.put(EntityFeaturePojo.alias_, 1); fields.put(EntityFeaturePojo.linkdata_, 1); // (slightly annoying, since only want these 2 largish fields if updating freq but won't // know // until after i've got this object) fields.put(EntityFeaturePojo.db_sync_time_, 1); fields.put(EntityFeaturePojo.db_sync_doccount_, 1); DBObject dboUpdate = null; if (_diagnosticMode) { dboUpdate = col.findOne(query, fields); } else { dboUpdate = col.findAndModify( query, fields, new BasicDBObject(), false, updateOp, false, true); // (can use findAndModify because specify index, ie the shard key) // (returns entity before the changes above, update the feature object below) // (also atomically creates the object if it doesn't exist so is "distributed-safe") } if ((dboUpdate != null) && !dboUpdate.keySet().isEmpty()) { // (Update the entity feature to be correct so that it can be accurately synchronized // with the index) EntityFeaturePojo gp = EntityFeaturePojo.fromDb(dboUpdate, EntityFeaturePojo.class); entFeature.setTotalfreq(gp.getTotalfreq() + nSavedFreqCount); entFeature.setDoccount(gp.getDoccount() + nSavedDocCount); entFeature.setDbSyncDoccount(gp.getDbSyncDoccount()); entFeature.setDbSyncTime(gp.getDbSyncTime()); if (null != gp.getAlias()) { entFeature.addAllAlias(gp.getAlias()); } if (null != gp.getSemanticLinks()) { entFeature.addToSemanticLinks(gp.getSemanticLinks()); } if (_diagnosticMode) { System.out.println( "EntityAggregationUtils.updateEntityFeatures, found: " + ((BasicDBObject) gp.toDb()).toString()); System.out.println( "EntityAggregationUtils.updateEntityFeatures, ^^^ found from query: " + query.toString() + " / " + updateOp.toString()); } } else // (the object in memory is now an accurate representation of the database, minus // some fields we'll now add) { // Synchronization settings for the newly created object if (null == savedSyncTime) { savedSyncTime = Long.toString(System.currentTimeMillis()); } entFeature.setDbSyncDoccount(nSavedDocCount); entFeature.setDbSyncTime(savedSyncTime); // This is all "distributed safe" (apart from the db_syc_xxx and it doesn't matter if // that is // out of date, the update will just be slightly out-of-date at worst) since // (otherwise) these fields are // only set here, and the findAndModify is atomic // (Do in raw MongoDB for performance) BasicDBObject baseFields = new BasicDBObject(); baseFields.put(EntityFeaturePojo.dimension_, entFeature.getDimension().toString()); baseFields.put(EntityFeaturePojo.type_, entFeature.getType()); baseFields.put( EntityFeaturePojo.disambiguated_name_, entFeature.getDisambiguatedName()); baseFields.put(EntityFeaturePojo.db_sync_doccount_, entFeature.getDbSyncDoccount()); baseFields.put(EntityFeaturePojo.db_sync_time_, entFeature.getDbSyncTime()); if ((null != entFeature.getSemanticLinks()) && !entFeature.getSemanticLinks().isEmpty()) { baseFields.put(EntityFeaturePojo.linkdata_, entFeature.getSemanticLinks()); } // attempt to add geotag (makes necessary checks on util side) // also add ontology type if geotag is found EntityGeotagAggregationUtils.addEntityGeo(entFeature); if (entFeature.getGeotag() != null) { BasicDBObject geo = new BasicDBObject(GeoPojo.lat_, entFeature.getGeotag().lat); geo.put(GeoPojo.lon_, entFeature.getGeotag().lon); baseFields.put(EntityFeaturePojo.geotag_, geo); if (entFeature.getOntology_type() != null) { baseFields.put(EntityFeaturePojo.ontology_type_, entFeature.getOntology_type()); } } if (!_diagnosticMode) { // Store the object col.update(query, new BasicDBObject(MongoDbManager.set_, baseFields)); } else { System.out.println( "EntityAggregationUtils.updateEntityFeatures, not found: " + query.toString() + ": " + baseFields.toString()); } entFeature.setDbSyncTime(null); // (ensures that index re-sync will occur) } } } catch (Exception e) { // Exception, remove from feature list it.remove(); // If an exception occurs log the error logger.error("Exception Message: " + e.getMessage(), e); } } // (end loop over communities) } // (end loop over indexes) } // TESTED (just by eye - made few changes during re-factoring)