bigquery 0.10.0

Library provides a simple API for Google BigQuery based on the gRPC Storage Read/Write and v2 APIs
Documentation
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
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
//! The Storage Read RPCs: `CreateReadSession` and `ReadRows`.
//!
//! A session is served a read rule's rows when one matches its table and row restriction, and
//! the table's own rows otherwise, projected to its selected fields and spread over the
//! table's read streams. The fake evaluates no filter, so a session with a row restriction
//! needs a rule, and one with a sample percentage or a snapshot time is served the current
//! rows.

use crate::db::fake::wire::{IpcCompression, IpcMessages};
use crate::db::fake::FakeCall;
use crate::testing::rules::BigQueryFakeRpc;
use crate::testing::server::FakeShared;
use crate::testing::state::{fit_to_layout, FakeReadStream, TableKey, DEFAULT_LOCATION};
use arrow_array::RecordBatch;
use arrow_schema::Schema;
use gcloud_sdk::google::cloud::bigquery::storage::v1::read_session::table_read_options::OutputFormatSerializationOptions;
use gcloud_sdk::google::cloud::bigquery::storage::v1::read_session::{self, TableReadOptions};
use gcloud_sdk::google::cloud::bigquery::storage::v1::{
    read_rows_response, ArrowRecordBatch, ArrowSchema, CreateReadSessionRequest, DataFormat,
    ReadRowsRequest, ReadRowsResponse, ReadSession, ReadStream,
};
use gcloud_sdk::tonic::Code;
use std::sync::Arc;

/// How BigQuery's `CreateReadSession` starts its `InvalidArgument` for a `selected_fields`
/// entry the table does not have. The client tells that error apart by the words "selected
/// fields do not exist" in it.
const SELECTED_FIELDS_DO_NOT_EXIST: &str =
    "The following selected fields do not exist in the table schema";

impl FakeShared {
    pub(super) async fn serve_read(&self, call: FakeCall) {
        match call.method() {
            "CreateReadSession" => self.create_read_session(call).await,
            "ReadRows" => self.read_rows(call).await,
            method => {
                let described = method.to_string();
                self.unmatched(call, &described, &[]);
            }
        }
    }

    /// Answers `CreateReadSession` with a session over the rows of a read rule or of the
    /// table, and records its streams for `ReadRows`.
    async fn create_read_session(&self, call: FakeCall) {
        let Some((call, request)) = self.first_request::<CreateReadSessionRequest>(call).await
        else {
            return;
        };
        let session = request.read_session.unwrap_or_default();
        let key = match TableKey::from_path(&session.table) {
            Ok(key) => key,
            Err(err) => {
                let described = format!("CreateReadSession of an invalid table: {err}");
                self.unmatched(call, &described, &[]);
                return;
            }
        };
        let rpc = BigQueryFakeRpc::CreateReadSession;
        let Some(call) = self.unfaulted(call, rpc, Some(&key)).await else {
            return;
        };
        let options = session.read_options.clone().unwrap_or_default();
        if session.data_format() != DataFormat::Arrow {
            let described = format!("CreateReadSession of {key} in {:?}", session.data_format());
            self.unmatched(call, &described, &[]);
            return;
        }
        let table = self.state().tables.get(&key).map(|table| {
            (
                table.schema.clone(),
                table.arrow_schema.clone(),
                table.batches.clone(),
                table.read_streams,
            )
        });
        let Some((schema, arrow_schema, table_batches, read_streams)) = table else {
            let status = key.not_found();
            call.fail(status.code(), status.message());
            return;
        };
        let columns = match SelectedColumns::of(&arrow_schema, &options.selected_fields) {
            SelectedColumns::Indexes(columns) => columns,
            SelectedColumns::Unknown(names) => {
                let message = format!("{SELECTED_FIELDS_DO_NOT_EXIST}: {}", names.join(", "));
                call.fail(Code::InvalidArgument, &message);
                return;
            }
            SelectedColumns::Nested(name) => {
                let described =
                    format!("CreateReadSession of {key} selecting the sub-field {name:?}");
                self.unmatched(call, &described, &[]);
                return;
            }
        };
        let restriction = Some(options.row_restriction.as_str()).filter(|r| !r.is_empty());
        let ruled = self.rules().answer_read(&key, restriction);
        let batches = match (ruled, restriction) {
            (Some(rows), _) => match fit_to_layout(&rows, &schema, &arrow_schema) {
                Ok(rows) => vec![rows],
                Err(err) => {
                    self.internal(call, &format!("the read rule's rows of {key}: {err}"));
                    return;
                }
            },
            (None, None) => table_batches,
            (None, Some(restriction)) => {
                let described = format!("CreateReadSession of {key} where {restriction:?}");
                let rules = self.rules().describe_reads();
                self.unmatched(call, &described, &rules);
                return;
            }
        };
        let projected = batches
            .iter()
            .map(|batch| batch.project(&columns))
            .collect::<Result<Vec<_>, _>>()
            .and_then(|batches| Ok((Arc::new(arrow_schema.project(&columns)?), batches)));
        let (arrow_schema, batches) = match projected {
            Ok(projected) => projected,
            Err(err) => {
                self.internal(call, &format!("the rows of {key} do not project: {err}"));
                return;
            }
        };
        let compression = requested_compression(&options);
        let schema_message = match IpcMessages::encode(&arrow_schema, &[], compression) {
            Ok(messages) => messages.schema,
            Err(err) => {
                self.internal(call, &format!("the schema of {key} does not encode: {err}"));
                return;
            }
        };
        let stream_count = usize::try_from(request.max_stream_count)
            .ok()
            .filter(|&max| max > 0)
            .map_or(read_streams.get(), |max| max.min(read_streams.get()));
        let row_count: usize = batches.iter().map(RecordBatch::num_rows).sum();
        let mut state = self.state();
        let name = format!(
            "projects/{}/locations/{DEFAULT_LOCATION}/sessions/fake-session-{}",
            key.dataset.project,
            state.next_id()
        );
        let streams = (0..stream_count)
            .map(|index| {
                let stream = ReadStream {
                    name: format!("{name}/streams/{index}"),
                };
                let rows = rows_between(
                    &batches,
                    index * row_count / stream_count,
                    (index + 1) * row_count / stream_count,
                );
                state.read_streams.insert(
                    stream.name.clone(),
                    FakeReadStream {
                        table: key.clone(),
                        arrow_schema: arrow_schema.clone(),
                        batches: rows,
                        compression,
                    },
                );
                stream
            })
            .collect();
        drop(state);
        call.reply(&ReadSession {
            name,
            streams,
            estimated_row_count: i64::try_from(row_count).unwrap_or(i64::MAX),
            schema: Some(read_session::Schema::ArrowSchema(ArrowSchema {
                serialized_schema: schema_message,
            })),
            ..session
        });
    }

    /// Answers `ReadRows` with one response per batch of the stream, from the request's
    /// offset on.
    async fn read_rows(&self, call: FakeCall) {
        let Some((call, request)) = self.first_request::<ReadRowsRequest>(call).await else {
            return;
        };
        let stream = self.state().read_streams.get(&request.read_stream).cloned();
        let table = stream.as_ref().map(|stream| &stream.table);
        let Some(mut call) = self.unfaulted(call, BigQueryFakeRpc::ReadRows, table).await else {
            return;
        };
        let Some(stream) = stream else {
            let described = format!("ReadRows of the unknown stream {:?}", request.read_stream);
            self.unmatched(call, &described, &[]);
            return;
        };
        let row_count: usize = stream.batches.iter().map(RecordBatch::num_rows).sum();
        let Some(offset) = usize::try_from(request.offset)
            .ok()
            .filter(|&offset| offset <= row_count)
        else {
            let described = format!(
                "ReadRows of {} at offset {} of {row_count} rows",
                request.read_stream, request.offset
            );
            self.unmatched(call, &described, &[]);
            return;
        };
        let batches = rows_between(&stream.batches, offset, row_count);
        let messages = match IpcMessages::encode(&stream.arrow_schema, &batches, stream.compression)
        {
            Ok(messages) => messages,
            Err(err) => {
                let failure = format!("the rows of {} do not encode: {err}", stream.table);
                self.internal(call, &failure);
                return;
            }
        };
        for (batch, message) in batches.iter().zip(messages.batches) {
            call.send(&ReadRowsResponse {
                row_count: i64::try_from(batch.num_rows()).unwrap_or(i64::MAX),
                rows: Some(read_rows_response::Rows::ArrowRecordBatch(
                    ArrowRecordBatch {
                        serialized_record_batch: message,
                        ..Default::default()
                    },
                )),
                ..Default::default()
            });
        }
        call.finish();
    }
}

/// The columns a session's `selected_fields` names, ignoring case as BigQuery does.
enum SelectedColumns {
    /// Their indexes in table order, as BigQuery sends them whatever order they were named in;
    /// every column when none is named.
    Indexes(Vec<usize>),
    /// The names the table does not have.
    Unknown(Vec<String>),
    /// A STRUCT sub-field such as `rec.field`, which the fake does not project.
    Nested(String),
}

impl SelectedColumns {
    fn of(schema: &Schema, selected: &[String]) -> Self {
        if selected.is_empty() {
            return Self::Indexes((0..schema.fields().len()).collect());
        }
        let has_column = |name: &str| {
            schema
                .fields()
                .iter()
                .any(|field| field.name().eq_ignore_ascii_case(name))
        };
        let unknown: Vec<String> = selected
            .iter()
            .filter(|name| !has_column(name))
            .cloned()
            .collect();
        if let Some(nested) = unknown.iter().find(|name| {
            name.split_once('.')
                .is_some_and(|(column, _)| has_column(column))
        }) {
            return Self::Nested(nested.clone());
        }
        if !unknown.is_empty() {
            return Self::Unknown(unknown);
        }
        Self::Indexes(
            (0..schema.fields().len())
                .filter(|&index| {
                    let column = schema.field(index).name();
                    selected
                        .iter()
                        .any(|name| name.eq_ignore_ascii_case(column))
                })
                .collect(),
        )
    }
}

/// The buffer compression a session's Arrow options ask for.
fn requested_compression(options: &TableReadOptions) -> IpcCompression {
    match &options.output_format_serialization_options {
        Some(OutputFormatSerializationOptions::ArrowSerializationOptions(arrow)) => {
            arrow.buffer_compression().into()
        }
        _ => IpcCompression::default(),
    }
}

/// The rows from `start` up to `end` of `batches` taken as one sequence, as slices of them.
fn rows_between(batches: &[RecordBatch], start: usize, end: usize) -> Vec<RecordBatch> {
    let mut first_row = 0;
    batches
        .iter()
        .filter_map(|batch| {
            let (from, to) = (first_row, first_row + batch.num_rows());
            first_row = to;
            let (low, high) = (start.max(from), end.min(to));
            (low < high).then(|| batch.slice(low - from, high - low))
        })
        .collect()
}

#[cfg(test)]
mod tests {
    use crate::errors::BigQueryError;
    use crate::read::BigQueryBatchRows;
    use crate::testing::{BigQueryFake, BigQueryFakeCode, BigQueryFakeFault, BigQueryFakeRpc};
    use crate::{
        BigQueryDatasetId, BigQueryReadOptions, BigQueryResult, BigQueryTableId, BigQueryTableRef,
    };
    use arrow_array::RecordBatch;
    use futures::TryStreamExt;
    use gcloud_sdk::google::cloud::bigquery::storage::v1::{
        CreateReadSessionRequest, DataFormat, ReadRowsRequest, ReadSession,
    };
    use gcloud_sdk::tonic::Code;
    use serde::{Deserialize, Serialize};
    use std::panic::{catch_unwind, AssertUnwindSafe};

    const SHOP: BigQueryDatasetId = BigQueryDatasetId::from_static("shop");
    const ORDERS: BigQueryTableId = BigQueryTableId::from_static("orders");

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct Order {
        id: i64,
        customer: String,
        total: f64,
        paid: bool,
        receipt: Option<Vec<u8>>,
    }

    #[derive(Debug, PartialEq, Deserialize)]
    struct OrderCustomer {
        id: i64,
        customer: String,
    }

    fn order(id: i64, customer: &str) -> Order {
        Order {
            id,
            customer: customer.to_string(),
            total: id as f64 * 1.5,
            paid: id % 2 == 0,
            receipt: (id % 3 == 0).then(|| vec![0, 159, 255]),
        }
    }

    fn orders() -> BigQueryTableRef {
        SHOP.table(ORDERS)
    }

    /// A fake whose `shop.orders` holds `rows`, over `streams` read streams.
    async fn fake_with(rows: &[Order], streams: usize) -> BigQueryResult<BigQueryFake> {
        let fake = BigQueryFake::start().await?;
        fake.table(orders(), |columns| columns.from_type::<Order>())
            .rows(rows)
            .read_streams(streams)
            .create()?;
        Ok(fake)
    }

    async fn read_orders(fake: &BigQueryFake) -> BigQueryResult<Vec<Order>> {
        let mut rows: Vec<Order> = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .obj()
            .query()
            .await?;
        rows.sort_by_key(|row| row.id);
        Ok(rows)
    }

    /// Whether dropping `fake`, which verifies it, panics.
    fn drop_panics(fake: BigQueryFake) -> bool {
        catch_unwind(AssertUnwindSafe(move || drop(fake))).is_err()
    }

    #[tokio::test]
    async fn a_read_returns_the_table_rows() -> BigQueryResult<()> {
        let rows = vec![order(1, "Alice"), order(2, "Bob"), order(3, "Carol")];
        let fake = fake_with(&rows, 1).await?;

        assert_eq!(read_orders(&fake).await?, rows);
        Ok(())
    }

    #[tokio::test]
    async fn selected_fields_project_the_columns() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice"), order(2, "Bob")], 1).await?;

        let batches: Vec<RecordBatch> = fake
            .db()
            .fluent()
            .select()
            .fields(["customer", "id"])
            .from(orders())
            .record_batches()
            .await?
            .try_collect()
            .await?;

        let columns: Vec<&str> = batches[0]
            .schema_ref()
            .fields()
            .iter()
            .map(|field| field.name().as_str())
            .collect();
        assert_eq!(columns, ["id", "customer"]);
        let read: Vec<OrderCustomer> = batches
            .iter()
            .flat_map(BigQueryBatchRows::new)
            .collect::<BigQueryResult<_>>()?;
        let expected = vec![
            OrderCustomer {
                id: 1,
                customer: "Alice".into(),
            },
            OrderCustomer {
                id: 2,
                customer: "Bob".into(),
            },
        ];
        assert_eq!(read, expected);
        Ok(())
    }

    #[tokio::test]
    async fn an_unknown_selected_field_is_a_schema_mismatch() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice")], 1).await?;

        let read = fake
            .db()
            .fluent()
            .select()
            .fields(["id", "discount"])
            .from(orders())
            .obj::<Order>()
            .query()
            .await;

        assert!(
            matches!(read, Err(BigQueryError::SchemaMismatchError(_))),
            "{read:?}"
        );
        Ok(())
    }

    #[tokio::test]
    async fn rows_are_spread_over_the_read_streams() -> BigQueryResult<()> {
        let rows: Vec<Order> = (1..=7).map(|id| order(id, "Alice")).collect();
        let fake = fake_with(&rows, 3).await?;

        let batches: Vec<RecordBatch> = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .options(BigQueryReadOptions::new().with_max_stream_count(8))
            .record_batches()
            .await?
            .try_collect()
            .await?;

        let mut sizes: Vec<usize> = batches.iter().map(RecordBatch::num_rows).collect();
        sizes.sort_unstable();
        assert_eq!(sizes, [2, 2, 3]);
        assert_eq!(read_orders(&fake).await?, rows);
        Ok(())
    }

    #[tokio::test]
    async fn a_read_rule_answers_a_row_restriction() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice"), order(2, "Bob")], 1).await?;
        let bobs = vec![order(2, "Bob")];
        let rule = fake
            .when_read(orders())
            .row_restriction("customer = 'Bob'")
            .returns_rows(&bobs)?;

        let read: Vec<Order> = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .filter_sql("customer = 'Bob'")
            .obj()
            .query()
            .await?;

        assert_eq!(read, bobs);
        assert_eq!(rule.calls(), 1);
        Ok(())
    }

    #[tokio::test]
    async fn an_unmatched_row_restriction_fails_verify() -> BigQueryResult<()> {
        let rows = vec![order(1, "Alice")];
        let fake = fake_with(&rows, 1).await?;
        assert_eq!(read_orders(&fake).await?, rows);

        let read = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .filter_sql("customer = 'Bob'")
            .obj::<Order>()
            .query()
            .await;

        match &read {
            Err(err) => assert!(err.has_code(Code::Unimplemented), "{err:?}"),
            Ok(rows) => panic!("expected the unmatched read to fail, got {rows:?}"),
        }
        assert!(drop_panics(fake));
        Ok(())
    }

    #[tokio::test]
    async fn a_fault_refuses_the_session() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice")], 1).await?;
        let fault = fake
            .when_fault(BigQueryFakeRpc::CreateReadSession)
            .on_table(orders())
            .fails(BigQueryFakeFault::status(
                BigQueryFakeCode::PermissionDenied,
                "no access",
            ))?;

        let read = read_orders(&fake).await;

        match &read {
            Err(err) => assert!(err.has_code(Code::PermissionDenied), "{err:?}"),
            Ok(rows) => panic!("expected the session to be refused, got {rows:?}"),
        }
        assert_eq!(fault.calls(), 1);
        Ok(())
    }

    #[tokio::test]
    async fn a_read_rows_fault_is_retried() -> BigQueryResult<()> {
        let rows = vec![order(1, "Alice"), order(2, "Bob")];
        let fake = fake_with(&rows, 1).await?;
        let fault = fake
            .when_fault(BigQueryFakeRpc::ReadRows)
            .on_table(orders())
            .times(1)
            .fails(BigQueryFakeFault::status(
                BigQueryFakeCode::Unavailable,
                "try again",
            ))?;

        assert_eq!(read_orders(&fake).await?, rows);
        assert_eq!(fault.calls(), 1);
        Ok(())
    }

    #[tokio::test]
    async fn read_rows_starts_at_its_offset() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice"), order(2, "Bob"), order(3, "Carol")], 1).await?;
        let project = fake.db().options().google_project_id.clone();
        let session = fake
            .db()
            .read_client()
            .create_read_session(CreateReadSessionRequest {
                parent: format!("projects/{project}"),
                read_session: Some(ReadSession {
                    table: orders().table_path(&project),
                    data_format: DataFormat::Arrow.into(),
                    ..Default::default()
                }),
                max_stream_count: 1,
                ..Default::default()
            })
            .await
            .map_err(BigQueryError::from)?
            .into_inner();

        let mut responses = fake
            .db()
            .read_client()
            .read_rows(ReadRowsRequest {
                read_stream: session.streams[0].name.clone(),
                offset: 2,
                ..Default::default()
            })
            .await
            .map_err(BigQueryError::from)?
            .into_inner();
        let mut row_count = 0;
        while let Some(response) = responses.message().await.map_err(BigQueryError::from)? {
            row_count += response.row_count;
        }

        assert_eq!(row_count, 1);
        Ok(())
    }

    #[tokio::test]
    async fn a_sampled_read_is_served_the_current_rows() -> BigQueryResult<()> {
        let rows = vec![order(1, "Alice"), order(2, "Bob")];
        let fake = fake_with(&rows, 1).await?;

        let mut read: Vec<Order> = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .sample_percentage(50.0)
            .obj()
            .query()
            .await?;

        read.sort_by_key(|row| row.id);
        assert_eq!(read, rows);
        Ok(())
    }

    #[tokio::test]
    async fn selected_fields_match_columns_ignoring_case() -> BigQueryResult<()> {
        let fake = fake_with(&[order(1, "Alice")], 1).await?;

        let read: Vec<OrderCustomer> = fake
            .db()
            .fluent()
            .select()
            .fields(["CUSTOMER", "Id"])
            .from(orders())
            .obj()
            .query()
            .await?;

        let expected = vec![OrderCustomer {
            id: 1,
            customer: "Alice".into(),
        }];
        assert_eq!(read, expected);
        Ok(())
    }

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct NotedOrder {
        id: i64,
        customer: String,
        total: f64,
        paid: bool,
        receipt: Option<Vec<u8>>,
        note: Option<String>,
    }

    #[tokio::test]
    async fn a_read_rule_takes_a_column_added_after_it() -> BigQueryResult<()> {
        let fake = fake_with(&[], 1).await?;
        let bob = order(2, "Bob");
        fake.when_read(orders())
            .row_restriction("customer = 'Bob'")
            .returns_rows([&bob])?;
        fake.db()
            .fluent()
            .schema()
            .table(orders())
            .columns(|columns| columns.from_type::<NotedOrder>())
            .sync()
            .await?;

        let read: Vec<NotedOrder> = fake
            .db()
            .fluent()
            .select()
            .from(orders())
            .filter_sql("customer = 'Bob'")
            .obj()
            .query()
            .await?;

        let expected = NotedOrder {
            id: bob.id,
            customer: bob.customer,
            total: bob.total,
            paid: bob.paid,
            receipt: bob.receipt,
            note: None,
        };
        assert_eq!(read, vec![expected]);
        Ok(())
    }

    const DELIVERIES: BigQueryTableId = BigQueryTableId::from_static("deliveries");

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct Address {
        city: String,
    }

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct Delivery {
        id: i64,
        address: Address,
    }

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct ZipAddress {
        city: String,
        zip: Option<String>,
    }

    #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
    struct ZipDelivery {
        id: i64,
        address: ZipAddress,
    }

    #[tokio::test]
    async fn a_read_rule_takes_a_nested_field_added_after_it() -> BigQueryResult<()> {
        let fake = BigQueryFake::start().await?;
        fake.table(SHOP.table(DELIVERIES), |columns| {
            columns.from_type::<Delivery>()
        })
        .create()?;
        let paris = Delivery {
            id: 1,
            address: Address {
                city: "Paris".to_string(),
            },
        };
        fake.when_read(SHOP.table(DELIVERIES))
            .row_restriction("address.city = 'Paris'")
            .returns_rows([&paris])?;
        fake.db()
            .fluent()
            .schema()
            .table(SHOP.table(DELIVERIES))
            .columns(|columns| columns.from_type::<ZipDelivery>())
            .sync()
            .await?;

        let read: Vec<ZipDelivery> = fake
            .db()
            .fluent()
            .select()
            .from(SHOP.table(DELIVERIES))
            .filter_sql("address.city = 'Paris'")
            .obj()
            .query()
            .await?;

        let expected = ZipDelivery {
            id: paris.id,
            address: ZipAddress {
                city: paris.address.city,
                zip: None,
            },
        };
        assert_eq!(read, vec![expected]);
        Ok(())
    }
}