FrankChen021 commented on code in PR #20729: URL: https://github.com/apache/druid/pull/20729#discussion_r4377522869
########## extensions-contrib/tsfile-extensions/src/main/java/org/apache/druid/data/input/tsfile/TsFileInputEntityReader.java: ########## @@ -0,0 +1,355 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +package org.apache.druid.data.input.tsfile; + +import org.apache.druid.data.input.InputEntity; +import org.apache.druid.data.input.InputRow; +import org.apache.druid.data.input.InputRowSchema; +import org.apache.druid.data.input.IntermediateRowParsingReader; +import org.apache.druid.data.input.impl.MapInputRowParser; +import org.apache.druid.error.InvalidInput; +import org.apache.druid.java.util.common.CloseableIterators; +import org.apache.druid.java.util.common.parsers.CloseableIterator; +import org.apache.druid.java.util.common.parsers.ParseException; +import org.apache.tsfile.enums.TSDataType; +import org.apache.tsfile.exception.read.ReadProcessException; +import org.apache.tsfile.exception.write.NoMeasurementException; +import org.apache.tsfile.exception.write.NoTableException; +import org.apache.tsfile.file.metadata.TableSchema; +import org.apache.tsfile.read.query.dataset.ResultSet; +import org.apache.tsfile.read.v4.ITsFileReader; +import org.apache.tsfile.read.v4.TsFileReaderBuilder; +import org.apache.tsfile.write.schema.IMeasurementSchema; +import org.joda.time.Interval; + +import javax.annotation.Nullable; + +import java.io.File; +import java.io.IOException; +import java.io.UncheckedIOException; +import java.util.ArrayList; +import java.util.Base64; +import java.util.Collections; +import java.util.HashMap; +import java.util.Iterator; +import java.util.List; +import java.util.Locale; +import java.util.Map; +import java.util.NoSuchElementException; +import java.util.Set; +import java.util.stream.Collectors; + +/** + * Reads the rows of a TsFile file with the TsFile table-model reader and converts them into + * {@link MapBasedInputRow}-compatible maps. + * <p> + * TsFile stores its own timestamp, which is exposed to Druid under the name {@value TsFileInputFormat#TIME_COLUMN_NAME}. + * Tag, field, and attribute columns are exposed as-is. + */ +public class TsFileInputEntityReader extends IntermediateRowParsingReader<Map<String, Object>> +{ + private final InputRowSchema inputRowSchema; + private final InputEntity source; + @Nullable + private final File temporaryDirectory; + @Nullable + private final List<String> tables; + @Nullable + private final List<String> columns; + @Nullable + private final Interval interval; + + TsFileInputEntityReader( + InputRowSchema inputRowSchema, + InputEntity source, + @Nullable File temporaryDirectory, + @Nullable List<String> tables, + @Nullable List<String> columns, + @Nullable Interval interval + ) + { + this.inputRowSchema = inputRowSchema; + this.source = source; + this.temporaryDirectory = temporaryDirectory; + this.tables = tables; + this.columns = columns; + this.interval = interval; + } + + @Override + protected CloseableIterator<Map<String, Object>> intermediateRowIterator() throws IOException + { + final InputEntity.CleanableFile materialized = + source.fetch(temporaryDirectory, new byte[InputEntity.DEFAULT_FETCH_BUFFER_SIZE]); + final ITsFileReader reader; + final List<String> tableNames; + try { + reader = new TsFileReaderBuilder().file(materialized.file()).build(); + tableNames = resolveTableNames(reader); + } + catch (IOException | RuntimeException e) { + materialized.close(); + throw e; + } + + return CloseableIterators.wrap( + new RowIterator(reader, tableNames), + () -> { + reader.close(); + materialized.close(); + } + ); + } + + @Override + protected InputEntity source() + { + return source; + } + + @Override + protected List<InputRow> parseInputRows(Map<String, Object> intermediateRow) throws ParseException + { + return Collections.singletonList(MapInputRowParser.parse(inputRowSchema, intermediateRow)); + } + + @Override + protected List<Map<String, Object>> toMap(Map<String, Object> intermediateRow) + { + return Collections.singletonList(intermediateRow); + } + + private List<String> resolveTableNames(ITsFileReader reader) throws IOException + { + final List<TableSchema> allTables = reader.getAllTableSchema(); + if (tables == null || tables.isEmpty()) { + return allTables.stream().map(TableSchema::getTableName).collect(Collectors.toList()); Review Comment: [P1] Unsupported files can silently ingest no rows **Finding:** When tables is omitted, an empty result from getAllTableSchema() is treated as the valid all-tables selection. RowIterator then reaches EOF immediately, so a TsFile with no table schema—the legacy tree-model form this extension says it does not support—can make ingestion succeed while dropping every row. **Suggestion:** Reject an empty table-schema list with an InvalidInput error, or explicitly detect and reject the tree model before constructing the row iterator. ########## extensions-contrib/tsfile-extensions/src/main/java/org/apache/druid/data/input/tsfile/TsFileInputEntityReader.java: ########## @@ -0,0 +1,355 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +package org.apache.druid.data.input.tsfile; + +import org.apache.druid.data.input.InputEntity; +import org.apache.druid.data.input.InputRow; +import org.apache.druid.data.input.InputRowSchema; +import org.apache.druid.data.input.IntermediateRowParsingReader; +import org.apache.druid.data.input.impl.MapInputRowParser; +import org.apache.druid.error.InvalidInput; +import org.apache.druid.java.util.common.CloseableIterators; +import org.apache.druid.java.util.common.parsers.CloseableIterator; +import org.apache.druid.java.util.common.parsers.ParseException; +import org.apache.tsfile.enums.TSDataType; +import org.apache.tsfile.exception.read.ReadProcessException; +import org.apache.tsfile.exception.write.NoMeasurementException; +import org.apache.tsfile.exception.write.NoTableException; +import org.apache.tsfile.file.metadata.TableSchema; +import org.apache.tsfile.read.query.dataset.ResultSet; +import org.apache.tsfile.read.v4.ITsFileReader; +import org.apache.tsfile.read.v4.TsFileReaderBuilder; +import org.apache.tsfile.write.schema.IMeasurementSchema; +import org.joda.time.Interval; + +import javax.annotation.Nullable; + +import java.io.File; +import java.io.IOException; +import java.io.UncheckedIOException; +import java.util.ArrayList; +import java.util.Base64; +import java.util.Collections; +import java.util.HashMap; +import java.util.Iterator; +import java.util.List; +import java.util.Locale; +import java.util.Map; +import java.util.NoSuchElementException; +import java.util.Set; +import java.util.stream.Collectors; + +/** + * Reads the rows of a TsFile file with the TsFile table-model reader and converts them into + * {@link MapBasedInputRow}-compatible maps. + * <p> + * TsFile stores its own timestamp, which is exposed to Druid under the name {@value TsFileInputFormat#TIME_COLUMN_NAME}. + * Tag, field, and attribute columns are exposed as-is. + */ +public class TsFileInputEntityReader extends IntermediateRowParsingReader<Map<String, Object>> +{ + private final InputRowSchema inputRowSchema; + private final InputEntity source; + @Nullable + private final File temporaryDirectory; + @Nullable + private final List<String> tables; + @Nullable + private final List<String> columns; + @Nullable + private final Interval interval; + + TsFileInputEntityReader( + InputRowSchema inputRowSchema, + InputEntity source, + @Nullable File temporaryDirectory, + @Nullable List<String> tables, + @Nullable List<String> columns, + @Nullable Interval interval + ) + { + this.inputRowSchema = inputRowSchema; + this.source = source; + this.temporaryDirectory = temporaryDirectory; + this.tables = tables; + this.columns = columns; + this.interval = interval; + } + + @Override + protected CloseableIterator<Map<String, Object>> intermediateRowIterator() throws IOException + { + final InputEntity.CleanableFile materialized = + source.fetch(temporaryDirectory, new byte[InputEntity.DEFAULT_FETCH_BUFFER_SIZE]); + final ITsFileReader reader; + final List<String> tableNames; + try { + reader = new TsFileReaderBuilder().file(materialized.file()).build(); + tableNames = resolveTableNames(reader); + } + catch (IOException | RuntimeException e) { Review Comment: [P2] Reader leaks when table validation fails **Finding:** The TsFile reader is built before resolveTableNames. If requested table validation throws, as in the tested unknown-table case, this catch closes only the materialized file and rethrows, so no iterator cleanup callback runs and the ITsFileReader remains open. **Suggestion:** Own the reader and materialized file with a closer or try/finally block and close both on every failure before returning the iterator. ########## extensions-contrib/tsfile-extensions/src/main/java/org/apache/druid/data/input/tsfile/TsFileInputEntityReader.java: ########## @@ -0,0 +1,355 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +package org.apache.druid.data.input.tsfile; + +import org.apache.druid.data.input.InputEntity; +import org.apache.druid.data.input.InputRow; +import org.apache.druid.data.input.InputRowSchema; +import org.apache.druid.data.input.IntermediateRowParsingReader; +import org.apache.druid.data.input.impl.MapInputRowParser; +import org.apache.druid.error.InvalidInput; +import org.apache.druid.java.util.common.CloseableIterators; +import org.apache.druid.java.util.common.parsers.CloseableIterator; +import org.apache.druid.java.util.common.parsers.ParseException; +import org.apache.tsfile.enums.TSDataType; +import org.apache.tsfile.exception.read.ReadProcessException; +import org.apache.tsfile.exception.write.NoMeasurementException; +import org.apache.tsfile.exception.write.NoTableException; +import org.apache.tsfile.file.metadata.TableSchema; +import org.apache.tsfile.read.query.dataset.ResultSet; +import org.apache.tsfile.read.v4.ITsFileReader; +import org.apache.tsfile.read.v4.TsFileReaderBuilder; +import org.apache.tsfile.write.schema.IMeasurementSchema; +import org.joda.time.Interval; + +import javax.annotation.Nullable; + +import java.io.File; +import java.io.IOException; +import java.io.UncheckedIOException; +import java.util.ArrayList; +import java.util.Base64; +import java.util.Collections; +import java.util.HashMap; +import java.util.Iterator; +import java.util.List; +import java.util.Locale; +import java.util.Map; +import java.util.NoSuchElementException; +import java.util.Set; +import java.util.stream.Collectors; + +/** + * Reads the rows of a TsFile file with the TsFile table-model reader and converts them into + * {@link MapBasedInputRow}-compatible maps. + * <p> + * TsFile stores its own timestamp, which is exposed to Druid under the name {@value TsFileInputFormat#TIME_COLUMN_NAME}. + * Tag, field, and attribute columns are exposed as-is. + */ +public class TsFileInputEntityReader extends IntermediateRowParsingReader<Map<String, Object>> +{ + private final InputRowSchema inputRowSchema; + private final InputEntity source; + @Nullable + private final File temporaryDirectory; + @Nullable + private final List<String> tables; + @Nullable + private final List<String> columns; + @Nullable + private final Interval interval; + + TsFileInputEntityReader( + InputRowSchema inputRowSchema, + InputEntity source, + @Nullable File temporaryDirectory, + @Nullable List<String> tables, + @Nullable List<String> columns, + @Nullable Interval interval + ) + { + this.inputRowSchema = inputRowSchema; + this.source = source; + this.temporaryDirectory = temporaryDirectory; + this.tables = tables; + this.columns = columns; + this.interval = interval; + } + + @Override + protected CloseableIterator<Map<String, Object>> intermediateRowIterator() throws IOException + { + final InputEntity.CleanableFile materialized = + source.fetch(temporaryDirectory, new byte[InputEntity.DEFAULT_FETCH_BUFFER_SIZE]); + final ITsFileReader reader; + final List<String> tableNames; + try { + reader = new TsFileReaderBuilder().file(materialized.file()).build(); + tableNames = resolveTableNames(reader); + } + catch (IOException | RuntimeException e) { + materialized.close(); + throw e; + } + + return CloseableIterators.wrap( + new RowIterator(reader, tableNames), + () -> { + reader.close(); + materialized.close(); + } + ); + } + + @Override + protected InputEntity source() + { + return source; + } + + @Override + protected List<InputRow> parseInputRows(Map<String, Object> intermediateRow) throws ParseException + { + return Collections.singletonList(MapInputRowParser.parse(inputRowSchema, intermediateRow)); + } + + @Override + protected List<Map<String, Object>> toMap(Map<String, Object> intermediateRow) + { + return Collections.singletonList(intermediateRow); + } + + private List<String> resolveTableNames(ITsFileReader reader) throws IOException + { + final List<TableSchema> allTables = reader.getAllTableSchema(); + if (tables == null || tables.isEmpty()) { + return allTables.stream().map(TableSchema::getTableName).collect(Collectors.toList()); + } + + final Set<String> available = + allTables.stream().map(schema -> schema.getTableName().toLowerCase(Locale.ROOT)).collect(Collectors.toSet()); + for (String table : tables) { + if (!available.contains(table.toLowerCase(Locale.ROOT))) { + throw InvalidInput.exception( + "Table[%s] does not exist in the file. Available tables: %s", + table, + allTables.stream().map(TableSchema::getTableName).collect(Collectors.toList()) + ); + } + } + return tables; + } + + private List<String> selectColumns(TableSchema tableSchema) + { + if (columns != null && !columns.isEmpty()) { + return columns; + } + + final List<String> selected = new ArrayList<>(); + for (IMeasurementSchema columnSchema : tableSchema.getColumnSchemas()) { + final String columnName = columnSchema.getMeasurementName(); + if (!TsFileInputFormat.TIME_COLUMN_NAME.equalsIgnoreCase(columnName)) { + selected.add(columnName); + } + } + return selected; + } + + /** + * Iterates over the rows of every selected table, keeping at most one open {@link ResultSet} at a time. + */ + private class RowIterator implements Iterator<Map<String, Object>> + { + private final ITsFileReader reader; + private final Iterator<String> tableIterator; + + @Nullable + private ResultSet resultSet = null; + private List<String> currentColumns = null; + private List<TSDataType> currentTypes = null; + @Nullable + private Map<String, Object> bufferedRow = null; + + RowIterator(ITsFileReader reader, List<String> tableNames) + { + this.reader = reader; + this.tableIterator = tableNames.iterator(); + } + + @Override + public boolean hasNext() + { + if (bufferedRow != null) { + return true; + } + + while (true) { + if (resultSet != null) { + try { + if (resultSet.next()) { + bufferedRow = readCurrentRow(); + return true; + } + } + catch (IOException e) { + throw new UncheckedIOException(e); + } + closeCurrentResultSet(); + } + + if (!tableIterator.hasNext()) { + closeCurrentResultSet(); + return false; + } + openTable(tableIterator.next()); + } + } + + @Override + public Map<String, Object> next() + { + if (!hasNext()) { + throw new NoSuchElementException(); + } + final Map<String, Object> row = bufferedRow; + bufferedRow = null; + return row; + } + + private void openTable(String tableName) + { + final long queryStartTime; + final long queryEndTime; + if (interval == null) { + queryStartTime = Long.MIN_VALUE; + queryEndTime = Long.MAX_VALUE; + } else { + if (interval.getEndMillis() <= interval.getStartMillis()) { + // An empty interval matches no row, so there is nothing to read. + return; + } + // TsFile folds both time bounds into the comparison, while a Druid interval is end-exclusive. + queryStartTime = interval.getStartMillis(); + queryEndTime = interval.getEndMillis() - 1; + } + + final TableSchema tableSchema; + try { + tableSchema = reader.getTableSchemas(tableName) + .orElseThrow( + () -> InvalidInput.exception( + "Table[%s] does not exist in the file", + tableName + ) + ); + } + catch (IOException e) { + throw new UncheckedIOException(e); + } + final List<String> selectedColumns = selectColumns(tableSchema); + final List<TSDataType> types = new ArrayList<>(selectedColumns.size()); + for (String column : selectedColumns) { + final int columnIndex = tableSchema.findColumnIndex(column); + if (columnIndex < 0) { + throw InvalidInput.exception( + "Column[%s] does not exist in table[%s]", + column, + tableName + ); + } + types.add(tableSchema.getColumnSchemas().get(columnIndex).getType()); + } + + try { + resultSet = reader.query(tableName, selectedColumns, queryStartTime, queryEndTime); + } + catch (IOException e) { + throw new UncheckedIOException(e); + } + catch (ReadProcessException | NoTableException | NoMeasurementException e) { + throw new RuntimeException(e); + } + currentColumns = selectedColumns; + currentTypes = types; + } + + private void closeCurrentResultSet() + { + if (resultSet != null) { + resultSet.close(); + resultSet = null; + } + } + + private Map<String, Object> readCurrentRow() + { + final Map<String, Object> row = new HashMap<>(); + + // Column index 1 of every TsFile result set is the implicit time column. + final long timestamp = resultSet.getLong(1); + row.put(TsFileInputFormat.TIME_COLUMN_NAME, timestamp); + + final String timestampColumn = inputRowSchema.getTimestampSpec().getTimestampColumn(); + if (!TsFileInputFormat.TIME_COLUMN_NAME.equals(timestampColumn)) { + row.put(timestampColumn, timestamp); + } + + for (int i = 0; i < currentColumns.size(); i++) { + final int resultSetIndex = i + 2; + if (!resultSet.isNull(resultSetIndex)) { + row.put(currentColumns.get(i), readValue(resultSet, currentTypes.get(i), resultSetIndex)); Review Comment: [P2] Timestamp alias is overwritten by a same-named column **Finding:** For a custom timestampSpec column that is also present in the selected TsFile columns, the timestamp alias is added first but this loop overwrites it with the measurement value. The documented shadowing behavior is not honored; ingestion can parse the measurement as the timestamp or assign incorrect timestamps. **Suggestion:** Write the timestamp alias after projecting measurement columns, or skip the measurement when its name equals the timestamp column. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
