use of org.graylog.shaded.elasticsearch7.org.elasticsearch.search.SearchHit in project topcom-cloud by 545314690.
the class ElasticSearchService method search.
/**
* 功能描述:查询
* @param index 索引名
* @param type 类型
* @param constructor 查询构造
*/
public Page search(String index, String type, ESQueryBuilderConstructor constructor) {
List<Map<String, Object>> result = new ArrayList<>();
SearchRequestBuilder searchRequestBuilder = client.prepareSearch(index).setTypes(type);
// 排序
if (StringUtils.isNotEmpty(constructor.getAsc()))
searchRequestBuilder.addSort(constructor.getAsc(), SortOrder.ASC);
if (StringUtils.isNotEmpty(constructor.getDesc()))
searchRequestBuilder.addSort(constructor.getDesc(), SortOrder.DESC);
// 设置查询体
searchRequestBuilder.setQuery(constructor.listBuilders());
// 返回条目数
int size = constructor.getSize();
if (size < 0) {
size = 0;
}
if (size > MAX) {
size = MAX;
}
// 返回条目数
searchRequestBuilder.setSize(size);
searchRequestBuilder.setFrom(constructor.getFrom() < 0 ? 0 : constructor.getFrom());
SearchResponse searchResponse = searchRequestBuilder.execute().actionGet();
SearchHits hits = searchResponse.getHits();
SearchHit[] searchHists = hits.getHits();
for (SearchHit sh : searchHists) {
Map<String, Object> source = sh.getSource();
source.put("_id", sh.getId());
result.add(sh.getSource());
}
return new Page(result, hits.getTotalHits());
}
use of org.graylog.shaded.elasticsearch7.org.elasticsearch.search.SearchHit in project topcom-cloud by 545314690.
the class EtlController method weiboEtl.
/**
* 微博id->weiboId
* md5(id)->微博id
* @param limit
* @return
*/
@ApiOperation("weiboEtl")
@RequestMapping(value = "/weiboEtl", method = RequestMethod.GET, produces = "application/json")
@ResponseBody
public Object weiboEtl(@RequestParam Integer limit) {
long count = 0;
SearchResponse searchResponse = scanWeibo("", limit);
while (searchResponse.getHits().getHits().length > 0) {
SearchHit[] hits = searchResponse.getHits().hits();
List<JSONObject> commentsList = new ArrayList<>();
for (int i = 0; i < hits.length; i++) {
JSONObject jsonObject = JSONObject.fromObject(hits[i].sourceAsMap());
jsonObject.put("weiboId", jsonObject.get("id"));
jsonObject.put("id", MD5Utils.md5(jsonObject.get("id").toString()));
commentsList.add(jsonObject);
}
updateJson("yuqing", "article", commentsList);
count = count + searchResponse.getHits().getHits().length;
logger.error(searchResponse.getHits().getTotalHits());
logger.error("处理微博条数:" + count + "\n 完成度" + (double) count * 100 / searchResponse.getHits().getTotalHits() + "%");
searchResponse = scanWeibo(searchResponse.getScrollId(), limit);
}
return "done";
}
use of org.graylog.shaded.elasticsearch7.org.elasticsearch.search.SearchHit in project topcom-cloud by 545314690.
the class EtlController method bbsEtl.
/**
* 微博id->weiboId
* md5(id)->微博id
* @param limit
* @return
*/
@ApiOperation("bbsEtl")
@RequestMapping(value = "/bbsEtl", method = RequestMethod.GET, produces = "application/json")
@ResponseBody
public Object bbsEtl(@RequestParam Integer limit) {
long count = 0;
long count1 = 0;
SearchResponse searchResponse = scanBbs("", limit);
while (searchResponse.getHits().getHits().length > 0) {
SearchHit[] hits = searchResponse.getHits().hits();
List<JSONObject> commentsList = new ArrayList<>();
for (int i = 0; i < hits.length; i++) {
JSONObject jsonObject = JSONObject.fromObject(hits[i].sourceAsMap());
String type = jsonObject.get("type").toString();
if (type.equals("Bbs")) {
Object object = jsonObject.get("floorNum");
if (object != null && Integer.valueOf(object.toString()) > 0) {
commentsList.add(jsonObject);
}
}
}
// String path = "D:\\data\\bbs\\"+"bbs"+(count1/50000)+".txt";
String path = "/home/yuqing/data/esFile/" + "bbs" + (count1 / 50000) + ".txt";
save2File(commentsList, path);
dleteEsDoc("yuqing", "article", commentsList);
count1 = count1 + commentsList.size();
count = count + searchResponse.getHits().getHits().length;
logger.error("删除BBS条数:" + count1);
logger.error("处理BBS条数:" + count + "\n 完成度" + (double) count * 100 / searchResponse.getHits().getTotalHits() + "%");
searchResponse = scanBbs(searchResponse.getScrollId(), limit);
}
return "done";
}
use of org.graylog.shaded.elasticsearch7.org.elasticsearch.search.SearchHit in project topcom-cloud by 545314690.
the class EtlController method commentsEtl.
@ApiOperation("commentsEtl")
@RequestMapping(value = "/commentsEtl", method = RequestMethod.GET, produces = "application/json")
@ResponseBody
public Object commentsEtl(@RequestParam Integer limit) {
long count = 0;
long count1 = 0;
SearchResponse searchResponse = scanComments("", limit);
while (searchResponse.getHits().getHits().length > 0) {
SearchHit[] hits = searchResponse.getHits().hits();
List<JSONObject> commentsList = new ArrayList<>();
for (int i = 0; i < hits.length; i++) {
JSONObject jsonObject = JSONObject.fromObject(hits[i].sourceAsMap());
commentsList.add(jsonObject);
}
// String path = "D:\\data\\comments\\"+"comments"+(count1/50000)+".txt";
String path = "/home/yuqing/data/esFile/comments/" + "comments" + (count1 / 50000) + ".txt";
save2File(commentsList, path);
dleteCommentsDoc("yuqing", "comment", commentsList);
count1 = count1 + commentsList.size();
count = count + searchResponse.getHits().getHits().length;
logger.error("删除comments条数:" + count1);
logger.error("处理comments条数:" + count + "\n 完成度" + (double) count * 100 / searchResponse.getHits().getTotalHits() + "%");
searchResponse = scanComments(searchResponse.getScrollId(), limit);
}
return "done";
}
use of org.graylog.shaded.elasticsearch7.org.elasticsearch.search.SearchHit in project herd by FINRAOS.
the class IndexSearchDaoTest method testIndexSearch.
private void testIndexSearch(Set<String> fields, Set<String> match, List<IndexSearchFilter> searchFilters, List<String> facetList, boolean isHitHighlightingEnabled, boolean testExceptions, boolean setInvalidSearchResultIndexName, boolean isColumnFields) throws IOException {
// Build the mocks
SearchRequestBuilder searchRequestBuilder = mock(SearchRequestBuilder.class);
SearchRequestBuilder searchRequestBuilderWithSource = mock(SearchRequestBuilder.class);
SearchRequestBuilder searchRequestBuilderWithSize = mock(SearchRequestBuilder.class);
SearchRequestBuilder searchRequestBuilderWithSorting = mock(SearchRequestBuilder.class);
SearchRequestBuilder searchRequestBuilderWithHighlighting = mock(SearchRequestBuilder.class);
SearchResponse searchResponse = mock(SearchResponse.class);
SearchHits searchHits = mock(SearchHits.class);
SearchHit searchHit1 = mock(SearchHit.class);
SearchHit searchHit2 = mock(SearchHit.class);
SearchShardTarget searchShardTarget1 = mock(SearchShardTarget.class);
SearchShardTarget searchShardTarget2 = mock(SearchShardTarget.class);
SearchHit[] searchHitArray = new SearchHit[2];
searchHitArray[0] = searchHit1;
searchHitArray[1] = searchHit2;
HighlightField highlightField = mock(HighlightField.class);
when(highlightField.getName()).thenReturn("displayName");
Text[] value = { new Text("match <hlt>fragment</hlt class=\"highlight\">"), new Text("<hlt class=\"highlight\">match</hlt>") };
when(highlightField.getFragments()).thenReturn(value);
@SuppressWarnings("unchecked") ListenableActionFuture<SearchResponse> listenableActionFuture = mock(ListenableActionFuture.class);
final String highlightFieldsConfigValue = "{\"highlightFields\":[{\"fieldName\":\"displayName\",\"fragmentSize\":100,\"matchedFields\":[\"displayName\",\"displayName.stemmed\",\"displayName.ngrams\"],\"numOfFragments\":5}]}";
final String highlightFieldsColumnMatchConfigValue = "{\"highlightFields\":[{\"fieldName\":\"columnName\",\"fragmentSize\":100,\"matchedFields\":[\"columnName\",\"columnName.stemmed\",\"columnName.ngrams\"],\"numOfFragments\":5}]}";
// Mock the call to external methods
when(configurationHelper.getProperty(ConfigurationValue.TAG_SHORT_DESCRIPTION_LENGTH, Integer.class)).thenReturn(300);
when(configurationHelper.getProperty(ConfigurationValue.BUSINESS_OBJECT_DEFINITION_SHORT_DESCRIPTION_LENGTH, Integer.class)).thenReturn(300);
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_NGRAMS)).thenReturn("{\"displayName\":\"1.0\"}");
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_STEMMED)).thenReturn("{\"displayName\":\"1.0\"}");
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_SHINGLES)).thenReturn("{\"displayName\":\"1.0\"}");
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_HIGHLIGHT_PRETAGS)).thenReturn("<hlt class=\"highlight\">");
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_HIGHLIGHT_POSTTAGS)).thenReturn("</hlt>");
if (match != null && match.contains(MATCH_COLUMN)) {
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_COLUMN_MATCH_HIGHLIGHT_FIELDS)).thenReturn(highlightFieldsColumnMatchConfigValue);
} else {
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_HIGHLIGHT_FIELDS)).thenReturn(highlightFieldsConfigValue);
}
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_BEST_FIELDS_QUERY_BOOST, Float.class)).thenReturn(1f);
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_PREFIX_QUERY_BOOST, Float.class)).thenReturn(1f);
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_QUERY_BOOST, Float.class)).thenReturn(1f);
when(configurationHelper.getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_QUERY_SLOP, Integer.class)).thenReturn(5);
Map<String, String> fieldsBoostMap = new HashMap<>();
fieldsBoostMap.put("displayName", "1.0");
if (isColumnFields) {
fieldsBoostMap.put(COLUMNS_NAME_FIELD, "1.0");
fieldsBoostMap.put(SCHEMA_COLUMNS_NAME_FIELD, "1.0");
}
if (testExceptions) {
when(jsonHelper.unmarshallJsonToObject(Map.class, "{\"displayName\":\"1.0\"}")).thenThrow(new IOException());
} else {
when(jsonHelper.unmarshallJsonToObject(Map.class, "{\"displayName\":\"1.0\"}")).thenReturn(fieldsBoostMap);
}
IndexSearchHighlightField indexSearchHighlightField = new IndexSearchHighlightField("displayName", 100, Arrays.asList("displayName", "displayName.stemmed", "displayName.ngrams"), 5);
IndexSearchHighlightFields highlightFields = new IndexSearchHighlightFields(Collections.singletonList(indexSearchHighlightField));
IndexSearchHighlightField indexSearchHighlightFieldColumnOnly = new IndexSearchHighlightField("columnName", 100, Arrays.asList("columnName", "columnName.stemmed", "columnName.ngrams"), 5);
IndexSearchHighlightFields highlightFieldsColumnOnly = new IndexSearchHighlightFields(Collections.singletonList(indexSearchHighlightFieldColumnOnly));
if (testExceptions) {
when(jsonHelper.unmarshallJsonToObject(IndexSearchHighlightFields.class, highlightFieldsConfigValue)).thenThrow(new IOException());
} else {
if (match != null && match.contains(MATCH_COLUMN)) {
when(jsonHelper.unmarshallJsonToObject(IndexSearchHighlightFields.class, highlightFieldsColumnMatchConfigValue)).thenReturn(highlightFieldsColumnOnly);
} else {
when(jsonHelper.unmarshallJsonToObject(IndexSearchHighlightFields.class, highlightFieldsConfigValue)).thenReturn(highlightFields);
}
}
when(searchRequestBuilder.setSource(any(SearchSourceBuilder.class))).thenReturn(searchRequestBuilderWithSource);
when(searchRequestBuilderWithSource.setSize(SEARCH_RESULT_SIZE)).thenReturn(searchRequestBuilderWithSize);
when(searchRequestBuilderWithSize.addSort(any(SortBuilder.class))).thenReturn(searchRequestBuilderWithSorting);
when(searchRequestBuilderWithSorting.highlighter(any(HighlightBuilder.class))).thenReturn(searchRequestBuilderWithHighlighting);
when(searchRequestBuilder.execute()).thenReturn(listenableActionFuture);
when(listenableActionFuture.actionGet()).thenReturn(searchResponse);
when(searchResponse.getHits()).thenReturn(searchHits);
when(searchHits.hits()).thenReturn(searchHitArray);
Map<String, Object> sourceMap1 = new HashMap<>();
Map<String, Object> tagTypeMap = new HashMap<>();
tagTypeMap.put(CODE, TAG_TYPE_CODE);
sourceMap1.put(TAG_TYPE, tagTypeMap);
when(searchHit1.sourceAsMap()).thenReturn(sourceMap1);
Map<String, Object> sourceMap2 = new HashMap<>();
Map<String, Object> businessObjectDefinitionMap = new HashMap<>();
businessObjectDefinitionMap.put(CODE, NAMESPACE_CODE);
sourceMap2.put(NAMESPACE, businessObjectDefinitionMap);
when(searchHit2.sourceAsMap()).thenReturn(sourceMap2);
when(searchHit1.getShard()).thenReturn(searchShardTarget1);
when(searchHit2.getShard()).thenReturn(searchShardTarget2);
when(searchShardTarget1.getIndex()).thenReturn(TAG_SEARCH_INDEX_NAME);
when(searchShardTarget2.getIndex()).thenReturn(BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME);
when(searchHits.getTotalHits()).thenReturn(200L);
Map<String, HighlightField> highlightFieldMap = new HashMap<>();
highlightFieldMap.put("displayName", highlightField);
when(searchHit1.getHighlightFields()).thenReturn(highlightFieldMap);
when(searchHit2.getHighlightFields()).thenReturn(highlightFieldMap);
// Create index search request
final IndexSearchRequest indexSearchRequest = new IndexSearchRequest(SEARCH_TERM, searchFilters, facetList, isHitHighlightingEnabled);
List<TagTypeIndexSearchResponseDto> tagTypeIndexSearchResponseDtos = Collections.singletonList(new TagTypeIndexSearchResponseDto("code", Collections.singletonList(new TagIndexSearchResponseDto("tag1", 1, null)), null));
List<ResultTypeIndexSearchResponseDto> resultTypeIndexSearchResponseDto = Collections.singletonList(new ResultTypeIndexSearchResponseDto("type", 1, null));
when(elasticsearchHelper.getNestedTagTagIndexSearchResponseDto(searchResponse)).thenReturn(tagTypeIndexSearchResponseDtos);
when(elasticsearchHelper.getResultTypeIndexSearchResponseDto(searchResponse)).thenReturn(resultTypeIndexSearchResponseDto);
when(elasticsearchHelper.getFacetsResponse(any(ElasticsearchResponseDto.class), eq(BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME), eq(TAG_SEARCH_INDEX_NAME))).thenCallRealMethod();
when(elasticsearchHelper.addIndexSearchFilterBooleanClause(any(), any(), any())).thenCallRealMethod();
when(elasticsearchHelper.addFacetFieldAggregations(any(), any(SearchRequestBuilder.class))).thenReturn(searchRequestBuilder);
when(herdSearchQueryHelper.determineNegationTermsPresent(any(IndexSearchRequest.class))).thenCallRealMethod();
when(herdSearchQueryHelper.extractNegationTerms(any(IndexSearchRequest.class))).thenCallRealMethod();
when(herdSearchQueryHelper.extractSearchPhrase(any(IndexSearchRequest.class))).thenCallRealMethod();
SearchResult searchResult = mock(SearchResult.class);
when(jestClientHelper.searchExecute(any())).thenReturn(searchResult);
List<SearchResult.Hit<Map, Void>> searchHitList = new ArrayList<>();
Gson gson = new Gson();
Map<String, Object> map = new HashMap<>();
map.put(DISPLAY_NAME_SOURCE, "Display Name");
Map<String, Object> codeMap = new HashMap<>();
codeMap.put(CODE, "Code");
map.put(TAG_TYPE, codeMap);
map.put(NAMESPACE, codeMap);
JsonElement element = gson.toJsonTree(map);
List<String> highlightList = new ArrayList<>();
highlightList.add("Highlight 1");
Map<String, List<String>> highlightMap = new HashMap<>();
highlightMap.put("field", highlightList);
SearchResult.Hit<Map, Void> hit1 = new SearchResult(gson).new Hit(HashMap.class, element, HashMap.class, null, highlightMap, null, setInvalidSearchResultIndexName ? INVALID_VALUE : TAG_SEARCH_INDEX_NAME, TAG_SEARCH_INDEX_NAME, "type", 1.0);
SearchResult.Hit<Map, Void> hit2 = new SearchResult(gson).new Hit(HashMap.class, element, HashMap.class, null, highlightMap, null, BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME, BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME, "type", 2.0);
searchHitList.add(hit1);
searchHitList.add(hit2);
when(searchResult.getHits(Map.class)).thenReturn(searchHitList);
when(searchResult.getTotal()).thenReturn(200L);
MetricAggregation metricAggregation = mock(MetricAggregation.class);
TermsAggregation termsAggregation = mock(TermsAggregation.class);
when(searchResult.getAggregations()).thenReturn(metricAggregation);
when(metricAggregation.getTermsAggregation(TAG_TYPE_FACET_AGGS)).thenReturn(termsAggregation);
List<TermsAggregation.Entry> buckets = new ArrayList<>();
TermsAggregation.Entry entry1 = mock(TermsAggregation.Entry.class);
TermsAggregation.Entry entry2 = mock(TermsAggregation.Entry.class);
buckets.add(entry1);
buckets.add(entry2);
when(termsAggregation.getBuckets()).thenReturn(buckets);
// Call the method under test
IndexSearchResponse indexSearchResponse = indexSearchDao.indexSearch(indexSearchRequest, fields, match, BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME, TAG_SEARCH_INDEX_NAME);
List<IndexSearchResult> indexSearchResults = indexSearchResponse.getIndexSearchResults();
assertThat("Index search results list is null.", indexSearchResults, not(nullValue()));
assertThat(indexSearchResponse.getTotalIndexSearchResults(), is(200L));
// Verify external method calls.
verify(herdSearchQueryHelper).determineNegationTermsPresent(indexSearchRequest);
verify(herdSearchQueryHelper).extractNegationTerms(indexSearchRequest);
verify(herdSearchQueryHelper).extractSearchPhrase(indexSearchRequest);
verify(configurationHelper).getProperty(ConfigurationValue.TAG_SHORT_DESCRIPTION_LENGTH, Integer.class);
verify(configurationHelper).getProperty(ConfigurationValue.BUSINESS_OBJECT_DEFINITION_SHORT_DESCRIPTION_LENGTH, Integer.class);
verify(configurationHelper, times(3)).getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_STEMMED);
verify(configurationHelper).getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_NGRAMS);
verify(configurationHelper).getProperty(ConfigurationValue.ELASTICSEARCH_SEARCHABLE_FIELDS_SHINGLES);
verify(configurationHelper).getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_PREFIX_QUERY_BOOST, Float.class);
verify(configurationHelper).getProperty(ConfigurationValue.ELASTICSEARCH_BEST_FIELDS_QUERY_BOOST, Float.class);
verify(configurationHelper, times(2)).getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_QUERY_BOOST, Float.class);
verify(configurationHelper, times(5)).getProperty(ConfigurationValue.ELASTICSEARCH_PHRASE_QUERY_SLOP, Integer.class);
verify(jsonHelper, times(5)).unmarshallJsonToObject(Map.class, "{\"displayName\":\"1.0\"}");
if (CollectionUtils.isNotEmpty(indexSearchRequest.getIndexSearchFilters())) {
verify(elasticsearchHelper).addIndexSearchFilterBooleanClause(indexSearchRequest.getIndexSearchFilters(), BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME, TAG_SEARCH_INDEX_NAME);
}
if (indexSearchRequest.isEnableHitHighlighting() != null) {
verifyHitHighlightingInteractions(searchRequestBuilder, indexSearchRequest.isEnableHitHighlighting(), match);
}
if (CollectionUtils.isNotEmpty(indexSearchRequest.getFacetFields())) {
verify(elasticsearchHelper).addFacetFieldAggregations(any(), any(SearchRequestBuilder.class));
if (indexSearchRequest.getFacetFields().contains(ElasticsearchHelper.TAG_FACET)) {
verify(elasticsearchHelper).getNestedTagTagIndexSearchResponseDto(searchResult);
verify(elasticsearchHelper).getTagTagIndexSearchResponseDto(searchResult);
}
if (indexSearchRequest.getFacetFields().contains(ElasticsearchHelper.RESULT_TYPE_FACET)) {
verify(elasticsearchHelper).getResultTypeIndexSearchResponseDto(searchResult);
}
verify(elasticsearchHelper).getFacetsResponse(any(ElasticsearchResponseDto.class), eq(BUSINESS_OBJECT_DEFINITION_SEARCH_INDEX_NAME), eq(TAG_SEARCH_INDEX_NAME));
}
verify(jestClientHelper).searchExecute(any());
verify(searchResult).getTotal();
verify(searchResult).getHits(Map.class);
verifyNoMoreInteractions(searchRequestBuilder, searchRequestBuilderWithSource, searchRequestBuilderWithSize, searchRequestBuilderWithSorting, searchRequestBuilderWithHighlighting, searchResponse, searchHits, searchHit1, searchHit2, searchShardTarget1, searchShardTarget2, highlightField, listenableActionFuture, searchResult, metricAggregation, termsAggregation, entry1, entry2);
verifyNoMoreInteractionsHelper();
}
Aggregations