sqlload 0.27.0

SQL data loading tooling. Pre-release, under development.
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
use crate::files_source::constants::LOADER_COLUMNS;
use crate::files_source::main::constant_columns::constant_columns;
use crate::files_source::main::path_template::path_template;
use crate::files_source::models::{RecordShape, SourceFormat};
use crate::project::_helpers::targets::{
    destination_kind_problems, is_files_target, stage_problems, target_problems,
};
use crate::project::constants::{
    FILES_CONNECTION_KINDS, MAX_POSTGRES_NAME_BYTES, MIN_BATCH_INTERVAL, STAGE_NAME_PARTS,
    TABLE_PLACEHOLDER,
};
use crate::project::models::{
    CdcDeletes, CdcMode, ConnectionConfig, ProjectConfig, SourceConfig, SourceKind, TableSettings,
    WriteStrategy,
};
use crate::store_paths::main::resolve_location::resolve_location;
use crate::store_paths::models::{Location, StoreKind};

/// Problems with names one section uses to refer to another.
pub(crate) fn reference_problems(project: &ProjectConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    if let Some(default_target) = &project.default_target
        && !project.targets.contains_key(default_target)
    {
        problems.push(format!(
            "default_target {default_target:?} is not a [targets] entry"
        ));
    }
    for (name, target) in &project.targets {
        problems.extend(target_problems(project, name, target));
    }
    for (name, connection) in &project.connections {
        let targeted = project
            .targets
            .values()
            .any(|target| target.connection == *name);
        problems.extend(stage_problems(name, connection, targeted));
        problems.extend(odbc_connection_problems(name, connection));
        if let ConnectionConfig::S3 {
            profile: Some(profile),
            ..
        } = connection
            && profile.is_empty()
        {
            problems.push(format!(
                "[connections.{name}] profile is empty; omit it to use AWS_PROFILE, else default"
            ));
        }
    }
    for (name, source) in &project.sources {
        problems.extend(source_problems(project, name, source));
    }
    problems
}

/// Why an `odbc` connection cannot connect: it needs exactly one of `connection_string` and `dsn`; `user` and `password` go with `dsn`; and `dsn` and `user` are written unbraced, as driver managers read them, so they cannot hold `;`, `{` or `}`.
fn odbc_connection_problems(name: &str, connection: &ConnectionConfig) -> Vec<String> {
    let ConnectionConfig::Odbc {
        connection_string,
        dsn,
        user,
        password,
    } = connection
    else {
        return Vec::new();
    };
    let place = format!("[connections.{name}]");
    let mut problems: Vec<String> = Vec::new();
    match (connection_string.is_some(), dsn.is_some()) {
        (false, false) => problems.push(format!("{place} needs connection_string, or dsn")),
        (true, true) => problems.push(format!(
            "{place} takes connection_string or dsn, not both"
        )),
        (true, false) if user.is_some() || password.is_some() => problems.push(format!(
            "{place} user and password go with dsn; with connection_string, put them in it (UID=...;PWD=...)"
        )),
        _ => {}
    }
    let unbraced = [dsn.as_deref(), user.as_deref()];
    if unbraced
        .into_iter()
        .flatten()
        .any(|value| value.contains([';', '{', '}']))
    {
        problems.push(format!(
            "{place} dsn and user cannot contain ';', '{{' or '}}'; use connection_string instead"
        ));
    }
    problems
}

fn source_problems(project: &ProjectConfig, name: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    let place = source_place(project, name);
    let expected = match source.kind {
        SourceKind::Postgres | SourceKind::PostgresCdc => "postgres",
        SourceKind::Snowflake => "snowflake",
        SourceKind::Odbc => "odbc",
        SourceKind::Files => "s3 or filesystem",
        SourceKind::Python => "any",
        SourceKind::GoogleSheets => "google_sheets",
    };
    match project.connections.get(&source.connection) {
        None if source.connection.is_empty() && source.kind == SourceKind::Python => {}
        None if source.connection.is_empty() => {
            problems.push(format!("{place} needs a connection"));
        }
        None => problems.push(format!(
            "{place} connection {:?} is not a [connections] entry",
            source.connection
        )),
        Some(connection) if source.kind == SourceKind::Python => {
            problems.extend(shared_warehouse_problems(project, name, source, connection));
        }
        Some(connection) if !expected.split(" or ").any(|kind| kind == connection.kind()) => {
            problems.push(format!(
                "{place} has kind = {:?} but connection {:?} is a {} connection (expected {expected})",
                source_kind_name(source.kind),
                source.connection,
                connection.kind()
            ));
        }
        Some(_) => {}
    }
    let only_target = (project.targets.len() == 1)
        .then(|| project.targets.keys().next())
        .flatten();
    match source
        .destination
        .as_ref()
        .or(project.default_target.as_ref())
        .or(only_target)
    {
        Some(target) if !project.targets.contains_key(target) => {
            problems.push(format!("{place} destination {target:?} is not a [targets] entry"));
        }
        Some(target) => problems.extend(destination_kind_problems(
            &place,
            source,
            is_files_target(project, target),
        )),
        None => problems.push(format!(
            "{place} needs a destination: set destination, or default_target when there are several targets"
        )),
    }
    problems.extend(kind_problems(&place, source));
    if source.kind == SourceKind::PostgresCdc {
        problems.extend(cdc_name_problems(&place, name, source));
    }
    problems
}

fn source_kind_name(kind: SourceKind) -> &'static str {
    match kind {
        SourceKind::Postgres => "postgres",
        SourceKind::PostgresCdc => "postgres_cdc",
        SourceKind::Files => "files",
        SourceKind::Snowflake => "snowflake",
        SourceKind::Odbc => "odbc",
        SourceKind::Python => "python",
        SourceKind::GoogleSheets => "google_sheets",
    }
}

/// A Python source reading a connection a warehouse target loads into, whose credentials Python code must never get (a files target's s3 or filesystem connection may be shared).
fn shared_warehouse_problems(
    project: &ProjectConfig,
    name: &str,
    source: &SourceConfig,
    connection: &ConnectionConfig,
) -> Vec<String> {
    if FILES_CONNECTION_KINDS.contains(&connection.kind()) {
        return Vec::new();
    }
    project
        .targets
        .iter()
        .filter(|(_, target)| target.connection == source.connection)
        .map(|(target, _)| {
            format!(
                "Python source {name} uses connection {}, which target {target} loads into; Python code never gets warehouse credentials — give the source its own connection",
                source.connection
            )
        })
        .collect()
}

/// How problems name a source: its TOML section, or the Python file declaring it.
fn source_place(project: &ProjectConfig, name: &str) -> String {
    match project.python_modules.get(name) {
        Some(module) => format!("Python source {name} ({module})"),
        None => format!("[sources.{name}]"),
    }
}

/// Settings each kind requires or rejects.
fn kind_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    let template = source.table_name_template();
    let cdc_only = source.publication.is_some()
        || source.slot.is_some()
        || source.cdc_mode != CdcMode::Merge
        || source.deletes != CdcDeletes::Soft
        || source.batch_interval.is_some();
    if cdc_only && source.kind != SourceKind::PostgresCdc {
        problems.push(format!(
            "{place} publication, slot, cdc_mode, deletes and batch_interval are for kind = \"postgres_cdc\""
        ));
    }
    if source.kind == SourceKind::PostgresCdc {
        problems.extend(cdc_problems(place, source));
    }
    let queries = source.tables.values().any(|table| table.sql.is_some());
    if !matches!(source.kind, SourceKind::Snowflake | SourceKind::Odbc) && queries {
        problems.push(format!(
            "{place} sql is for kind = \"snowflake\" and kind = \"odbc\""
        ));
    }
    if source.kind != SourceKind::Snowflake
        && (source.warehouse.is_some() || source.stage.is_some())
    {
        problems.push(format!(
            "{place} warehouse and stage are for kind = \"snowflake\""
        ));
    }
    if source.kind != SourceKind::Odbc
        && (source.max_text_size.is_some() || source.max_binary_size.is_some())
    {
        problems.push(format!(
            "{place} max_text_size and max_binary_size are for kind = \"odbc\""
        ));
    }
    if source.kind == SourceKind::Snowflake {
        problems.extend(unload_stage_problems(place, source));
    }
    let sheets_only = source.spreadsheet_id.is_some()
        || !source.ranges.is_empty()
        || sheet_settings(source)
            .any(|settings| settings.header.is_some() || settings.header_row.is_some());
    if sheets_only && source.kind != SourceKind::GoogleSheets {
        problems.push(format!(
            "{place} spreadsheet_id, ranges, header and header_row are for kind = \"google_sheets\""
        ));
    }
    match source.kind {
        SourceKind::Postgres
        | SourceKind::PostgresCdc
        | SourceKind::Snowflake
        | SourceKind::Odbc => {
            if source.include.is_empty() && !queries {
                let tables = match source.kind {
                    SourceKind::Snowflake | SourceKind::Odbc => {
                        "; list tables in include, or give a table a sql query"
                    }
                    _ => "",
                };
                problems.push(format!(
                    "{place} include is empty, so it selects no tables{tables}"
                ));
            }
            if !template.contains(TABLE_PLACEHOLDER) {
                problems.push(format!(
                    "{place} table_name {template:?} must contain {TABLE_PLACEHOLDER}, or every table gets the same name"
                ));
            }
            problems.extend(files_only_problems(place, source));
        }
        SourceKind::Files => {
            if source.path.is_none() {
                problems.push(format!("{place} kind = \"files\" needs a path"));
            }
            if !source.include.is_empty() || !source.exclude.is_empty() {
                problems.push(format!("{place} include and exclude are for database sources; a files source is one table"));
            }
            problems.extend(encoding_problems(place, source));
            problems.extend(path_value_problems(place, source));
        }
        SourceKind::GoogleSheets => {
            if source.ranges.len() != 1 && !template.contains(TABLE_PLACEHOLDER) {
                problems.push(format!(
                    "{place} table_name {template:?} must contain {TABLE_PLACEHOLDER}, or every table gets the same name"
                ));
            }
            problems.extend(files_only_problems(place, source));
            problems.extend(sheets_problems(place, source));
        }
        SourceKind::Python => {
            if source.tables.len() > 1 && !template.contains(TABLE_PLACEHOLDER) {
                problems.push(format!(
                    "{place} table_name {template:?} must contain {TABLE_PLACEHOLDER}, or every table gets the same name"
                ));
            }
        }
    }
    problems
}

/// Problems with a `snowflake` source's `stage`: an S3 `url` with its `storage_integration`, or an existing stage's `name`, not both.
fn unload_stage_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let Some(stage) = &source.stage else {
        return Vec::new();
    };
    let place = format!("{place} stage");
    let mut problems: Vec<String> = Vec::new();
    match (&stage.url, &stage.storage_integration, &stage.name) {
        (Some(url), Some(_), None) => match resolve_location(None, url) {
            Ok(Location {
                store: StoreKind::S3 { .. },
                ..
            }) => {}
            Ok(_) => problems.push(format!("{place} url {url:?} must be an s3:// URL")),
            Err(error) => problems.push(format!("{place} url {url:?}: {error}")),
        },
        (Some(_), None, None) => problems.push(format!(
            "{place} url needs the storage_integration Snowflake writes to it with"
        )),
        (None, None, Some(name))
            if name.split('.').filter(|part| !part.is_empty()).count() != STAGE_NAME_PARTS =>
        {
            problems.push(format!("{place} name {name:?} must be DATABASE.SCHEMA.STAGE"));
        }
        (None, None, Some(_)) => {}
        _ => problems.push(format!(
            "{place} takes either url and storage_integration (an S3 folder) or name (an existing stage)"
        )),
    }
    problems
}

/// The settings only a files source takes, set on a source of another kind.
fn files_only_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    if source.path.is_some()
        || !source.partition_columns.is_empty()
        || source.path_columns.is_some()
        || source.unmatched_paths.is_some()
        || !source.constant_columns.is_empty()
        || source.copy_to_internal_stage
    {
        problems.push(format!(
            "{place} path, partition_columns, path_columns, unmatched_paths, constant_columns and copy_to_internal_stage are for kind = \"files\""
        ));
    }
    if source.format != SourceFormat::Parquet
        || source.compression.is_some()
        || source.csv.is_some()
        || source.shape != RecordShape::Columns
        || source.record_column.is_some()
    {
        problems.push(format!(
            "{place} format, compression, csv, shape and record_column are for kind = \"files\""
        ));
    }
    problems
}

/// A source's settings and each of its tables'.
fn sheet_settings(source: &SourceConfig) -> impl Iterator<Item = &TableSettings> {
    std::iter::once(&source.defaults).chain(source.tables.values().map(|table| &table.settings))
}

/// What a `google_sheets` source needs and cannot take: a spreadsheet, and no cursor, since a sheet has no column to read after.
fn sheets_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    if source
        .spreadsheet_id
        .as_deref()
        .is_none_or(|id| id.trim().is_empty())
    {
        problems.push(format!("{place} kind = \"google_sheets\" needs a spreadsheet_id, the ID in the spreadsheet's URL"));
    }
    if !source.include.is_empty() || !source.exclude.is_empty() {
        problems.push(format!("{place} include and exclude are not for kind = \"google_sheets\"; name its tables in ranges"));
    }
    for settings in sheet_settings(source) {
        if settings.cursor_column.is_some()
            || settings.batch_size.is_some()
            || settings.write_strategy == Some(WriteStrategy::DeleteInsert)
        {
            problems.push(format!(
                "{place} a sheet has no cursor, so cursor_column, batch_size and write_strategy = \"delete_insert\" do not apply; use replace (the default), merge on a unique_key, or append"
            ));
            break;
        }
    }
    if sheet_settings(source).any(|settings| settings.header_row == Some(0)) {
        problems.push(format!("{place} header_row counts from 1"));
    }
    problems
}

/// A files source's `path_columns` and `constant_columns` that do not parse, and a column any two of them and `partition_columns` both fill.
fn path_value_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    let mut names: Vec<(String, &str)> = source
        .partition_columns
        .keys()
        .map(|name| (name.clone(), "partition_columns"))
        .collect();
    match source
        .path_columns
        .as_deref()
        .map(|text| path_template(text, &[]))
    {
        None if source.unmatched_paths.is_some() => problems.push(format!(
            "{place} unmatched_paths is for path_columns, which is not set"
        )),
        None => {}
        Some(Ok(template)) => names.extend(
            template
                .captures
                .into_iter()
                .map(|(name, _)| (name, "path_columns")),
        ),
        Some(Err(problem)) => problems.push(format!("{place} {problem}")),
    }
    if let Some(record) = &source.files_encoding().record {
        names.push((record.clone(), "record_column"));
    }
    match constant_columns(&source.constant_columns, &[]) {
        Ok(constants) => names.extend(
            constants
                .into_iter()
                .map(|(name, _, _)| (name, "constant_columns")),
        ),
        Err(problem) => problems.push(format!("{place} {problem}")),
    }
    for (index, (name, setting)) in names.iter().enumerate() {
        if let Some((_, first)) = names[..index]
            .iter()
            .find(|(seen, _)| seen.eq_ignore_ascii_case(name))
        {
            problems.push(format!(
                "{place} column {name} is in both {first} and {setting}; give it one source"
            ));
        }
    }
    problems
}

/// Settings that do not fit a files source's format.
fn encoding_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    match (source.shape, source.format, &source.record_column) {
        (RecordShape::Record, SourceFormat::Parquet, _) => problems.push(format!(
            "{place} shape = \"record\" is for ndjson, json and csv files; Parquet files already have typed columns"
        )),
        (RecordShape::Columns, _, Some(_)) => problems.push(format!(
            "{place} record_column is for shape = \"record\", which is not set"
        )),
        (_, _, Some(name)) if name.trim().is_empty() => {
            problems.push(format!("{place} record_column must not be empty"));
        }
        (_, _, Some(name))
            if LOADER_COLUMNS
                .iter()
                .any(|loader| loader.eq_ignore_ascii_case(name)) =>
        {
            problems.push(format!(
                "{place} record_column {name} is a column SQLLoad adds itself; choose another name"
            ));
        }
        _ => {}
    }
    if let Some(record) = &source.files_encoding().record {
        for (table, overrides) in &source.tables {
            let pinned: Vec<&str> = overrides
                .columns
                .keys()
                .map(String::as_str)
                .filter(|name| name != record)
                .collect();
            if !pinned.is_empty() {
                problems.push(format!(
                    "{place} shape = \"record\" loads each record into one json column, {record}, so [tables.{table}] columns cannot pin {}",
                    pinned.join(", ")
                ));
            }
        }
    }
    if source.format == SourceFormat::Parquet && source.compression.is_some() {
        problems.push(format!(
            "{place} compression is for ndjson, json and csv files; Parquet files compress inside the file"
        ));
    }
    match (&source.csv, source.format) {
        (None, _) => {}
        (Some(csv), SourceFormat::Csv) => {
            for (setting, character) in [("delimiter", csv.delimiter), ("quote", csv.quote)] {
                if !character.is_ascii() || character == '\n' || character == '\r' {
                    problems.push(format!(
                        "{place} csv {setting} must be one ASCII character other than a line break, not {character:?}"
                    ));
                }
            }
            if csv.delimiter == csv.quote {
                problems.push(format!("{place} csv delimiter and quote must differ"));
            }
        }
        (Some(_), format) => problems.push(format!(
            "{place} csv is for format = \"csv\", not {format:?}",
            format = format.to_string()
        )),
    }
    problems
}

/// Settings a `postgres_cdc` source cannot take: the change stream decides how rows are written, so no strategy or cursor; `deletes = "hard"` needs `merge`.
fn cdc_problems(place: &str, source: &SourceConfig) -> Vec<String> {
    let mut problems: Vec<String> = Vec::new();
    let settings = std::iter::once(&source.defaults)
        .chain(source.tables.values().map(|table| &table.settings));
    for settings in settings {
        if settings.write_strategy.is_some()
            || settings.cursor_column.is_some()
            || settings.batch_size.is_some()
            || settings.state.is_some()
            || settings.checks.is_some()
        {
            problems.push(format!(
                "{place} kind = \"postgres_cdc\" writes every captured change, so it takes no write_strategy, cursor_column, batch_size, state or checks; choose cdc_mode = \"merge\" or \"changelog\""
            ));
            break;
        }
    }
    if source
        .batch_interval
        .is_some_and(|interval| interval < MIN_BATCH_INTERVAL)
    {
        problems.push(format!(
            "{place} batch_interval must be at least 1s, or `load --follow` would write a batch for almost every change"
        ));
    }
    if source.deletes == CdcDeletes::Hard && source.cdc_mode == CdcMode::Changelog {
        problems.push(format!(
            "{place} deletes = \"hard\" applies to cdc_mode = \"merge\"; a changelog keeps every delete as a row"
        ));
    }
    problems
}

/// A `postgres_cdc` source's slot and publication names, as set or by default (`sqlload_<source>`), must be plain Postgres names.
fn cdc_name_problems(place: &str, name: &str, source: &SourceConfig) -> Vec<String> {
    let default_name = format!("sqlload_{}", name.to_lowercase());
    let mut problems: Vec<String> = Vec::new();
    for (setting, chosen) in [("slot", &source.slot), ("publication", &source.publication)] {
        let effective = chosen.as_deref().unwrap_or(&default_name);
        let valid = !effective.is_empty()
            && effective.len() <= MAX_POSTGRES_NAME_BYTES
            && effective
                .chars()
                .all(|c| c.is_ascii_lowercase() || c.is_ascii_digit() || c == '_');
        if !valid {
            problems.push(format!(
                "{place} {setting} {effective:?} must be lower-case letters, digits and underscores, at most {MAX_POSTGRES_NAME_BYTES} characters; set {setting} = \"...\""
            ));
        }
    }
    problems
}