1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
// 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.

//! In-memory data source for presenting a Vec<RecordBatch> as a data source that can be
//! queried by DataFusion. This allows data to be pre-loaded into memory and then repeatedly
//! queried without incurring additional file I/O overhead.

use std::sync::{Arc, Mutex};

use arrow::datatypes::{Field, Schema};
use arrow::record_batch::RecordBatch;

use crate::datasource::{RecordBatchIterator, ScanResult, TableProvider};
use crate::error::{ExecutionError, Result};

/// In-memory table
pub struct MemTable {
    schema: Arc<Schema>,
    batches: Vec<RecordBatch>,
}

impl MemTable {
    /// Create a new in-memory table from the provided schema and record batches
    pub fn new(schema: Arc<Schema>, batches: Vec<RecordBatch>) -> Result<Self> {
        if batches
            .iter()
            .all(|batch| batch.schema().as_ref() == schema.as_ref())
        {
            Ok(Self { schema, batches })
        } else {
            Err(ExecutionError::General(
                "Mismatch between schema and batches".to_string(),
            ))
        }
    }

    /// Create a mem table by reading from another data source
    pub fn load(t: &TableProvider) -> Result<Self> {
        let schema = t.schema();
        let partitions = t.scan(&None, 1024 * 1024)?;

        let mut data: Vec<RecordBatch> = vec![];
        for it in &partitions {
            while let Ok(Some(batch)) = it.lock().unwrap().next() {
                data.push(batch);
            }
        }

        MemTable::new(schema.clone(), data)
    }
}

impl TableProvider for MemTable {
    fn schema(&self) -> &Arc<Schema> {
        &self.schema
    }

    fn scan(
        &self,
        projection: &Option<Vec<usize>>,
        _batch_size: usize,
    ) -> Result<Vec<ScanResult>> {
        let columns: Vec<usize> = match projection {
            Some(p) => p.clone(),
            None => {
                let l = self.schema.fields().len();
                let mut v = Vec::with_capacity(l);
                for i in 0..l {
                    v.push(i);
                }
                v
            }
        };

        let projected_columns: Result<Vec<Field>> = columns
            .iter()
            .map(|i| {
                if *i < self.schema.fields().len() {
                    Ok(self.schema.field(*i).clone())
                } else {
                    Err(ExecutionError::General(
                        "Projection index out of range".to_string(),
                    ))
                }
            })
            .collect();

        let projected_schema = Arc::new(Schema::new(projected_columns?));

        let batches = self
            .batches
            .iter()
            .map(|batch| {
                RecordBatch::try_new(
                    projected_schema.clone(),
                    columns.iter().map(|i| batch.column(*i).clone()).collect(),
                )
            })
            .collect();

        match batches {
            Ok(batches) => Ok(vec![Arc::new(Mutex::new(MemBatchIterator {
                schema: projected_schema.clone(),
                index: 0,
                batches,
            }))]),
            Err(e) => Err(ExecutionError::ArrowError(e)),
        }
    }
}

/// Iterator over an in-memory table
pub struct MemBatchIterator {
    schema: Arc<Schema>,
    index: usize,
    batches: Vec<RecordBatch>,
}

impl RecordBatchIterator for MemBatchIterator {
    fn schema(&self) -> &Arc<Schema> {
        &self.schema
    }

    fn next(&mut self) -> Result<Option<RecordBatch>> {
        if self.index < self.batches.len() {
            self.index += 1;
            Ok(Some(self.batches[self.index - 1].clone()))
        } else {
            Ok(None)
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use arrow::array::Int32Array;
    use arrow::datatypes::{DataType, Field, Schema};

    #[test]
    fn test_with_projection() {
        let schema = Arc::new(Schema::new(vec![
            Field::new("a", DataType::Int32, false),
            Field::new("b", DataType::Int32, false),
            Field::new("c", DataType::Int32, false),
        ]));

        let batch = RecordBatch::try_new(
            schema.clone(),
            vec![
                Arc::new(Int32Array::from(vec![1, 2, 3])),
                Arc::new(Int32Array::from(vec![4, 5, 6])),
                Arc::new(Int32Array::from(vec![7, 8, 9])),
            ],
        )
        .unwrap();

        let provider = MemTable::new(schema, vec![batch]).unwrap();

        // scan with projection
        let partitions = provider.scan(&Some(vec![2, 1]), 1024).unwrap();
        let batch2 = partitions[0].lock().unwrap().next().unwrap().unwrap();
        assert_eq!(2, batch2.schema().fields().len());
        assert_eq!("c", batch2.schema().field(0).name());
        assert_eq!("b", batch2.schema().field(1).name());
        assert_eq!(2, batch2.num_columns());
    }

    #[test]
    fn test_without_projection() {
        let schema = Arc::new(Schema::new(vec![
            Field::new("a", DataType::Int32, false),
            Field::new("b", DataType::Int32, false),
            Field::new("c", DataType::Int32, false),
        ]));

        let batch = RecordBatch::try_new(
            schema.clone(),
            vec![
                Arc::new(Int32Array::from(vec![1, 2, 3])),
                Arc::new(Int32Array::from(vec![4, 5, 6])),
                Arc::new(Int32Array::from(vec![7, 8, 9])),
            ],
        )
        .unwrap();

        let provider = MemTable::new(schema, vec![batch]).unwrap();

        let partitions = provider.scan(&None, 1024).unwrap();
        let batch1 = partitions[0].lock().unwrap().next().unwrap().unwrap();
        assert_eq!(3, batch1.schema().fields().len());
        assert_eq!(3, batch1.num_columns());
    }

    #[test]
    fn test_invalid_projection() {
        let schema = Arc::new(Schema::new(vec![
            Field::new("a", DataType::Int32, false),
            Field::new("b", DataType::Int32, false),
            Field::new("c", DataType::Int32, false),
        ]));

        let batch = RecordBatch::try_new(
            schema.clone(),
            vec![
                Arc::new(Int32Array::from(vec![1, 2, 3])),
                Arc::new(Int32Array::from(vec![4, 5, 6])),
                Arc::new(Int32Array::from(vec![7, 8, 9])),
            ],
        )
        .unwrap();

        let provider = MemTable::new(schema, vec![batch]).unwrap();

        let projection: Vec<usize> = vec![0, 4];

        match provider.scan(&Some(projection), 1024) {
            Err(ExecutionError::General(e)) => {
                assert_eq!("\"Projection index out of range\"", format!("{:?}", e))
            }
            _ => assert!(false, "Scan should failed on invalid projection"),
        };
    }

    #[test]
    fn test_schema_validation() {
        let schema1 = Arc::new(Schema::new(vec![
            Field::new("a", DataType::Int32, false),
            Field::new("b", DataType::Int32, false),
            Field::new("c", DataType::Int32, false),
        ]));

        let schema2 = Arc::new(Schema::new(vec![
            Field::new("a", DataType::Int32, false),
            Field::new("b", DataType::Float64, false),
            Field::new("c", DataType::Int32, false),
        ]));

        let batch = RecordBatch::try_new(
            schema1.clone(),
            vec![
                Arc::new(Int32Array::from(vec![1, 2, 3])),
                Arc::new(Int32Array::from(vec![4, 5, 6])),
                Arc::new(Int32Array::from(vec![7, 8, 9])),
            ],
        )
        .unwrap();

        match MemTable::new(schema2, vec![batch]) {
            Err(ExecutionError::General(e)) => assert_eq!(
                "\"Mismatch between schema and batches\"",
                format!("{:?}", e)
            ),
            _ => assert!(
                false,
                "MemTable::new should have failed due to schema mismatch"
            ),
        }
    }
}