risingwave_connector/parser/
parquet_parser.rs

1// Copyright 2024 RisingWave Labs
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7//     http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15use std::sync::Arc;
16
17use futures_async_stream::try_stream;
18use prometheus::core::GenericCounter;
19use risingwave_common::array::arrow::arrow_array_iceberg::RecordBatch;
20use risingwave_common::array::arrow::{IcebergArrowConvert, is_parquet_schema_match_source_schema};
21use risingwave_common::array::{ArrayBuilderImpl, DataChunk, StreamChunk};
22use risingwave_common::metrics::LabelGuardedMetric;
23use risingwave_common::types::{Datum, ScalarImpl};
24use thiserror_ext::AsReport;
25
26use crate::parser::ConnectorResult;
27use crate::source::SourceColumnDesc;
28/// `ParquetParser` is responsible for converting the incoming `record_batch_stream`
29/// into a `streamChunk`.
30#[derive(Debug)]
31pub struct ParquetParser {
32    rw_columns: Vec<SourceColumnDesc>,
33    file_name: String,
34    offset: usize,
35}
36
37impl ParquetParser {
38    pub fn new(
39        rw_columns: Vec<SourceColumnDesc>,
40        file_name: String,
41        offset: usize,
42    ) -> ConnectorResult<Self> {
43        Ok(Self {
44            rw_columns,
45            file_name,
46            offset,
47        })
48    }
49
50    #[try_stream(boxed, ok = StreamChunk, error = crate::error::ConnectorError)]
51    pub async fn into_stream(
52        mut self,
53        record_batch_stream: parquet::arrow::async_reader::ParquetRecordBatchStream<
54            tokio_util::compat::Compat<opendal::FuturesAsyncReader>,
55        >,
56        file_source_input_row_count_metrics: Option<
57            LabelGuardedMetric<GenericCounter<prometheus::core::AtomicU64>>,
58        >,
59        parquet_source_skip_row_count_metrics: Option<
60            LabelGuardedMetric<GenericCounter<prometheus::core::AtomicU64>>,
61        >,
62    ) {
63        #[for_await]
64        for record_batch in record_batch_stream {
65            let record_batch: RecordBatch = record_batch?;
66            // Convert each record batch into a stream chunk according to user defined schema.
67            let chunk: StreamChunk = self.convert_record_batch_to_stream_chunk(
68                record_batch,
69                file_source_input_row_count_metrics.clone(),
70                parquet_source_skip_row_count_metrics.clone(),
71            )?;
72
73            yield chunk;
74        }
75    }
76
77    fn inc_offset(&mut self) {
78        self.offset += 1;
79    }
80
81    /// The function `convert_record_batch_to_stream_chunk` is designed to transform the given `RecordBatch` into a `StreamChunk`.
82    ///
83    /// For each column in the source column:
84    /// - If the column's schema matches a column in the `RecordBatch` (both the data type and column name are the same),
85    ///   the corresponding records are converted into a column of the `StreamChunk`.
86    /// - If the column's schema does not match, null values are inserted.
87    /// - Hidden columns are handled separately by filling in the appropriate fields to ensure the data chunk maintains the correct format.
88    /// - If a column in the Parquet file does not exist in the source schema, it is skipped.
89    ///
90    /// # Arguments
91    ///
92    /// * `record_batch` - The `RecordBatch` to be converted into a `StreamChunk`.
93    ///
94    /// # Returns
95    ///
96    /// A `StreamChunk` containing the converted data from the `RecordBatch`.
97    ///
98    /// The hidden columns that must be included here are `_rw_file` and `_rw_offset`.
99    /// Depending on whether the user specifies a primary key (pk), there may be an additional hidden column `row_id`.
100    /// Therefore, the maximum number of hidden columns is three.
101    fn convert_record_batch_to_stream_chunk(
102        &mut self,
103        record_batch: RecordBatch,
104        file_source_input_row_count_metrics: Option<
105            LabelGuardedMetric<GenericCounter<prometheus::core::AtomicU64>>,
106        >,
107        parquet_source_skip_row_count_metrics: Option<
108            LabelGuardedMetric<GenericCounter<prometheus::core::AtomicU64>>,
109        >,
110    ) -> Result<StreamChunk, crate::error::ConnectorError> {
111        const MAX_HIDDEN_COLUMN_NUMS: usize = 3;
112        let column_size = self.rw_columns.len();
113        let mut chunk_columns = Vec::with_capacity(self.rw_columns.len() + MAX_HIDDEN_COLUMN_NUMS);
114
115        for source_column in self.rw_columns.clone() {
116            match source_column.column_type {
117                crate::source::SourceColumnType::Normal => {
118                    let rw_data_type: &risingwave_common::types::DataType =
119                        &source_column.data_type;
120                    let rw_column_name = &source_column.name;
121                    if let Some(parquet_column) = record_batch.column_by_name(rw_column_name)
122                        && is_parquet_schema_match_source_schema(
123                            parquet_column.data_type(),
124                            rw_data_type,
125                        )
126                    {
127                        let arrow_field = IcebergArrowConvert
128                            .to_arrow_field(rw_column_name, rw_data_type)
129                            .map_err(|e| {
130                                crate::parser::AccessError::ParquetParser {
131                                    message: format!(
132                                        "to_arrow_field failed, column='{}', rw_type='{}', offset={}, error={}",
133                                        rw_column_name, rw_data_type, self.offset, e.as_report()
134                                    )
135                                }
136                            })?;
137                        let array_impl = IcebergArrowConvert
138                            .array_from_arrow_array(&arrow_field, parquet_column)
139                            .map_err(|e| {
140                                crate::parser::AccessError::ParquetParser {
141                                    message: format!(
142                                        "array_from_arrow_array failed, column='{}', rw_type='{}', arrow_field='{}', parquet_type='{}', offset={}, error={}",
143                                        rw_column_name,
144                                        rw_data_type,
145                                        arrow_field.data_type(),
146                                        parquet_column.data_type(),
147                                        self.offset,
148                                        e.as_report()
149                                    )
150                                }
151                            })?;
152                        chunk_columns.push(Arc::new(array_impl));
153                    } else {
154                        // Handle additional columns, for file source, the additional columns are offset and file name;
155                        // for columns defined in the user schema but not present in the parquet file, fill with null.
156                        let column = if let Some(additional_column_type) =
157                            &source_column.additional_column.column_type
158                        {
159                            match additional_column_type {
160                                risingwave_pb::plan_common::additional_column::ColumnType::Offset(_) => {
161                                    let mut array_builder = ArrayBuilderImpl::with_type(column_size, source_column.data_type.clone());
162                                    for _ in 0..record_batch.num_rows() {
163                                        let datum: Datum = Some(ScalarImpl::Utf8((self.offset).to_string().into()));
164                                        self.inc_offset();
165                                        array_builder.append(datum);
166                                    }
167                                    Arc::new(array_builder.finish())
168                                }
169                                risingwave_pb::plan_common::additional_column::ColumnType::Filename(_) => {
170                                    let mut array_builder = ArrayBuilderImpl::with_type(column_size, source_column.data_type.clone());
171                                    let datum: Datum = Some(ScalarImpl::Utf8(self.file_name.clone().into()));
172                                    array_builder.append_n(record_batch.num_rows(), datum);
173                                    Arc::new(array_builder.finish())
174                                }
175                                _ => unreachable!(),
176                            }
177                        } else {
178                            // For columns defined in the source schema but not present in the Parquet file, null values are filled in.
179                            let mut array_builder =
180                                ArrayBuilderImpl::with_type(column_size, rw_data_type.clone());
181                            array_builder.append_n_null(record_batch.num_rows());
182                            if let Some(metrics) = parquet_source_skip_row_count_metrics.clone() {
183                                metrics.inc_by(record_batch.num_rows() as u64);
184                            }
185                            Arc::new(array_builder.finish())
186                        };
187                        chunk_columns.push(column);
188                    }
189                }
190                crate::source::SourceColumnType::RowId => {
191                    let mut array_builder =
192                        ArrayBuilderImpl::with_type(column_size, source_column.data_type.clone());
193                    let datum: Datum = None;
194                    array_builder.append_n(record_batch.num_rows(), datum);
195                    let res = array_builder.finish();
196                    let column = Arc::new(res);
197                    chunk_columns.push(column);
198                }
199                // The following fields are only used in CDC source
200                crate::source::SourceColumnType::Offset | crate::source::SourceColumnType::Meta => {
201                    unreachable!()
202                }
203            }
204        }
205        if let Some(metrics) = file_source_input_row_count_metrics {
206            metrics.inc_by(record_batch.num_rows() as u64);
207        }
208
209        let data_chunk = DataChunk::new(chunk_columns.clone(), record_batch.num_rows());
210        Ok(data_chunk.into())
211    }
212}
risingwave_connector/parser/parquet_parser.rs

risingwave_connector/parser/
parquet_parser.rs
