use of com.alibaba.otter.canal.client.adapter.es.core.config.ESSyncConfig.ESMapping in project canal by alibaba.
the class ESEtlService method importData.
public EtlResult importData(List<String> params) {
ESMapping mapping = config.getEsMapping();
logger.info("start etl to import data to index: {}", mapping.get_index());
String sql = mapping.getSql();
return importData(sql, params);
}
use of com.alibaba.otter.canal.client.adapter.es.core.config.ESSyncConfig.ESMapping in project canal by alibaba.
the class ESEtlService method executeSqlImport.
protected boolean executeSqlImport(DataSource ds, String sql, List<Object> values, AdapterConfig.AdapterMapping adapterMapping, AtomicLong impCount, List<String> errMsg) {
try {
ESMapping mapping = (ESMapping) adapterMapping;
Util.sqlRS(ds, sql, values, rs -> {
int count = 0;
try {
ESBulkRequest esBulkRequest = this.esConnection.new ES7xBulkRequest();
long batchBegin = System.currentTimeMillis();
while (rs.next()) {
Map<String, Object> esFieldData = new LinkedHashMap<>();
Object idVal = null;
for (FieldItem fieldItem : mapping.getSchemaItem().getSelectFields().values()) {
String fieldName = fieldItem.getFieldName();
if (mapping.getSkips().contains(fieldName)) {
continue;
}
// 如果是主键字段则不插入
if (fieldItem.getFieldName().equals(mapping.get_id())) {
idVal = esTemplate.getValFromRS(mapping, rs, fieldName, fieldName);
} else {
Object val = esTemplate.getValFromRS(mapping, rs, fieldName, fieldName);
esFieldData.put(Util.cleanColumn(fieldName), val);
}
}
if (!mapping.getRelations().isEmpty()) {
mapping.getRelations().forEach((relationField, relationMapping) -> {
Map<String, Object> relations = new HashMap<>();
relations.put("name", relationMapping.getName());
if (StringUtils.isNotEmpty(relationMapping.getParent())) {
FieldItem parentFieldItem = mapping.getSchemaItem().getSelectFields().get(relationMapping.getParent());
Object parentVal;
try {
parentVal = esTemplate.getValFromRS(mapping, rs, parentFieldItem.getFieldName(), parentFieldItem.getFieldName());
} catch (SQLException e) {
throw new RuntimeException(e);
}
if (parentVal != null) {
relations.put("parent", parentVal.toString());
esFieldData.put("$parent_routing", parentVal.toString());
}
}
esFieldData.put(Util.cleanColumn(relationField), relations);
});
}
if (idVal != null) {
String parentVal = (String) esFieldData.remove("$parent_routing");
if (mapping.isUpsert()) {
ESUpdateRequest esUpdateRequest = this.esConnection.new ES7xUpdateRequest(mapping.get_index(), idVal.toString()).setDoc(esFieldData).setDocAsUpsert(true);
if (StringUtils.isNotEmpty(parentVal)) {
esUpdateRequest.setRouting(parentVal);
}
esBulkRequest.add(esUpdateRequest);
} else {
ESIndexRequest esIndexRequest = this.esConnection.new ES7xIndexRequest(mapping.get_index(), idVal.toString()).setSource(esFieldData);
if (StringUtils.isNotEmpty(parentVal)) {
esIndexRequest.setRouting(parentVal);
}
esBulkRequest.add(esIndexRequest);
}
} else {
idVal = esFieldData.get(mapping.getPk());
ESSearchRequest esSearchRequest = this.esConnection.new ESSearchRequest(mapping.get_index()).setQuery(QueryBuilders.termQuery(mapping.getPk(), idVal)).size(10000);
SearchResponse response = esSearchRequest.getResponse();
for (SearchHit hit : response.getHits()) {
ESUpdateRequest esUpdateRequest = this.esConnection.new ES7xUpdateRequest(mapping.get_index(), hit.getId()).setDoc(esFieldData);
esBulkRequest.add(esUpdateRequest);
}
}
if (esBulkRequest.numberOfActions() % mapping.getCommitBatch() == 0 && esBulkRequest.numberOfActions() > 0) {
long esBatchBegin = System.currentTimeMillis();
ESBulkResponse rp = esBulkRequest.bulk();
if (rp.hasFailures()) {
rp.processFailBulkResponse("全量数据 etl 异常 ");
}
if (logger.isTraceEnabled()) {
logger.trace("全量数据批量导入批次耗时: {}, es执行时间: {}, 批次大小: {}, index; {}", (System.currentTimeMillis() - batchBegin), (System.currentTimeMillis() - esBatchBegin), esBulkRequest.numberOfActions(), mapping.get_index());
}
batchBegin = System.currentTimeMillis();
esBulkRequest.resetBulk();
}
count++;
impCount.incrementAndGet();
}
if (esBulkRequest.numberOfActions() > 0) {
long esBatchBegin = System.currentTimeMillis();
ESBulkResponse rp = esBulkRequest.bulk();
if (rp.hasFailures()) {
rp.processFailBulkResponse("全量数据 etl 异常 ");
}
if (logger.isTraceEnabled()) {
logger.trace("全量数据批量导入最后批次耗时: {}, es执行时间: {}, 批次大小: {}, index; {}", (System.currentTimeMillis() - batchBegin), (System.currentTimeMillis() - esBatchBegin), esBulkRequest.numberOfActions(), mapping.get_index());
}
}
} catch (Exception e) {
logger.error(e.getMessage(), e);
errMsg.add(mapping.get_index() + " etl failed! ==>" + e.getMessage());
throw new RuntimeException(e);
}
return count;
});
return true;
} catch (Exception e) {
logger.error(e.getMessage(), e);
return false;
}
}
use of com.alibaba.otter.canal.client.adapter.es.core.config.ESSyncConfig.ESMapping in project canal by alibaba.
the class ES7xTemplate method updateByQuery.
@Override
public void updateByQuery(ESSyncConfig config, Map<String, Object> paramsTmp, Map<String, Object> esFieldData) {
if (paramsTmp.isEmpty()) {
return;
}
ESMapping mapping = config.getEsMapping();
BoolQueryBuilder queryBuilder = QueryBuilders.boolQuery();
paramsTmp.forEach((fieldName, value) -> queryBuilder.must(QueryBuilders.termsQuery(fieldName, value)));
// 查询sql批量更新
DataSource ds = DatasourceConfig.DATA_SOURCES.get(config.getDataSourceKey());
StringBuilder sql = new StringBuilder("SELECT * FROM (" + mapping.getSql() + ") _v WHERE ");
List<Object> values = new ArrayList<>();
paramsTmp.forEach((fieldName, value) -> {
sql.append("_v.").append(fieldName).append("=? AND ");
values.add(value);
});
// TODO 直接外部包裹sql会导致全表扫描性能低, 待优化拼接内部where条件
int len = sql.length();
sql.delete(len - 4, len);
Integer syncCount = (Integer) Util.sqlRS(ds, sql.toString(), values, rs -> {
int count = 0;
try {
while (rs.next()) {
Object idVal = getIdValFromRS(mapping, rs);
append4Update(mapping, idVal, esFieldData);
commitBulk();
count++;
}
} catch (Exception e) {
throw new RuntimeException(e);
}
return count;
});
if (logger.isTraceEnabled()) {
logger.trace("Update ES by query affected {} records", syncCount);
}
}
use of com.alibaba.otter.canal.client.adapter.es.core.config.ESSyncConfig.ESMapping in project canal by alibaba.
the class ESEtlService method executeSqlImport.
protected boolean executeSqlImport(DataSource ds, String sql, List<Object> values, AdapterConfig.AdapterMapping adapterMapping, AtomicLong impCount, List<String> errMsg) {
try {
ESMapping mapping = (ESMapping) adapterMapping;
Util.sqlRS(ds, sql, values, rs -> {
int count = 0;
try {
ESBulkRequest esBulkRequest = this.esConnection.new ES6xBulkRequest();
long batchBegin = System.currentTimeMillis();
while (rs.next()) {
Map<String, Object> esFieldData = new LinkedHashMap<>();
Object idVal = null;
for (FieldItem fieldItem : mapping.getSchemaItem().getSelectFields().values()) {
String fieldName = fieldItem.getFieldName();
if (mapping.getSkips().contains(fieldName)) {
continue;
}
// 如果是主键字段则不插入
if (fieldItem.getFieldName().equals(mapping.get_id())) {
idVal = esTemplate.getValFromRS(mapping, rs, fieldName, fieldName);
} else {
Object val = esTemplate.getValFromRS(mapping, rs, fieldName, fieldName);
esFieldData.put(Util.cleanColumn(fieldName), val);
}
}
if (!mapping.getRelations().isEmpty()) {
mapping.getRelations().forEach((relationField, relationMapping) -> {
Map<String, Object> relations = new HashMap<>();
relations.put("name", relationMapping.getName());
if (StringUtils.isNotEmpty(relationMapping.getParent())) {
FieldItem parentFieldItem = mapping.getSchemaItem().getSelectFields().get(relationMapping.getParent());
Object parentVal;
try {
parentVal = esTemplate.getValFromRS(mapping, rs, parentFieldItem.getFieldName(), parentFieldItem.getFieldName());
} catch (SQLException e) {
throw new RuntimeException(e);
}
if (parentVal != null) {
relations.put("parent", parentVal.toString());
esFieldData.put("$parent_routing", parentVal.toString());
}
}
esFieldData.put(Util.cleanColumn(relationField), relations);
});
}
if (idVal != null) {
String parentVal = (String) esFieldData.remove("$parent_routing");
if (mapping.isUpsert()) {
ESUpdateRequest esUpdateRequest = this.esConnection.new ES6xUpdateRequest(mapping.get_index(), mapping.get_type(), idVal.toString()).setDoc(esFieldData).setDocAsUpsert(true);
if (StringUtils.isNotEmpty(parentVal)) {
esUpdateRequest.setRouting(parentVal);
}
esBulkRequest.add(esUpdateRequest);
} else {
ESIndexRequest esIndexRequest = this.esConnection.new ES6xIndexRequest(mapping.get_index(), mapping.get_type(), idVal.toString()).setSource(esFieldData);
if (StringUtils.isNotEmpty(parentVal)) {
esIndexRequest.setRouting(parentVal);
}
esBulkRequest.add(esIndexRequest);
}
} else {
idVal = esFieldData.get(mapping.getPk());
ESSearchRequest esSearchRequest = this.esConnection.new ESSearchRequest(mapping.get_index(), mapping.get_type()).setQuery(QueryBuilders.termQuery(mapping.getPk(), idVal)).size(10000);
SearchResponse response = esSearchRequest.getResponse();
for (SearchHit hit : response.getHits()) {
ESUpdateRequest esUpdateRequest = this.esConnection.new ES6xUpdateRequest(mapping.get_index(), mapping.get_type(), hit.getId()).setDoc(esFieldData);
esBulkRequest.add(esUpdateRequest);
}
}
if (esBulkRequest.numberOfActions() % mapping.getCommitBatch() == 0 && esBulkRequest.numberOfActions() > 0) {
long esBatchBegin = System.currentTimeMillis();
ESBulkResponse rp = esBulkRequest.bulk();
if (rp.hasFailures()) {
rp.processFailBulkResponse("全量数据 etl 异常 ");
}
if (logger.isTraceEnabled()) {
logger.trace("全量数据批量导入批次耗时: {}, es执行时间: {}, 批次大小: {}, index; {}", (System.currentTimeMillis() - batchBegin), (System.currentTimeMillis() - esBatchBegin), esBulkRequest.numberOfActions(), mapping.get_index());
}
batchBegin = System.currentTimeMillis();
esBulkRequest.resetBulk();
}
count++;
impCount.incrementAndGet();
}
if (esBulkRequest.numberOfActions() > 0) {
long esBatchBegin = System.currentTimeMillis();
ESBulkResponse rp = esBulkRequest.bulk();
if (rp.hasFailures()) {
rp.processFailBulkResponse("全量数据 etl 异常 ");
}
if (logger.isTraceEnabled()) {
logger.trace("全量数据批量导入最后批次耗时: {}, es执行时间: {}, 批次大小: {}, index; {}", (System.currentTimeMillis() - batchBegin), (System.currentTimeMillis() - esBatchBegin), esBulkRequest.numberOfActions(), mapping.get_index());
}
}
} catch (Exception e) {
logger.error(e.getMessage(), e);
errMsg.add(mapping.get_index() + " etl failed! ==>" + e.getMessage());
throw new RuntimeException(e);
}
return count;
});
return true;
} catch (Exception e) {
logger.error(e.getMessage(), e);
return false;
}
}
use of com.alibaba.otter.canal.client.adapter.es.core.config.ESSyncConfig.ESMapping in project canal by alibaba.
the class ESSyncService method mainTableInsert.
/**
* 主表(单表)复杂字段insert
*
* @param config es配置
* @param dml dml信息
* @param data 单行dml数据
*/
private void mainTableInsert(ESSyncConfig config, Dml dml, Map<String, Object> data) {
ESMapping mapping = config.getEsMapping();
String sql = mapping.getSql();
String condition = ESSyncUtil.pkConditionSql(mapping, data);
sql = ESSyncUtil.appendCondition(sql, condition);
DataSource ds = DatasourceConfig.DATA_SOURCES.get(config.getDataSourceKey());
if (logger.isTraceEnabled()) {
logger.trace("Main table insert to es index by query sql, destination:{}, table: {}, index: {}, sql: {}", config.getDestination(), dml.getTable(), mapping.get_index(), sql.replace("\n", " "));
}
Util.sqlRS(ds, sql, rs -> {
try {
while (rs.next()) {
Map<String, Object> esFieldData = new LinkedHashMap<>();
Object idVal = esTemplate.getESDataFromRS(mapping, rs, esFieldData);
if (logger.isTraceEnabled()) {
logger.trace("Main table insert to es index by query sql, destination:{}, table: {}, index: {}, id: {}", config.getDestination(), dml.getTable(), mapping.get_index(), idVal);
}
esTemplate.insert(mapping, idVal, esFieldData);
}
} catch (Exception e) {
throw new RuntimeException(e);
}
return 0;
});
}
Aggregations