polyglot_sql/dialects/
bigquery.rs

1//! BigQuery Dialect
2//!
3//! BigQuery-specific transformations based on sqlglot patterns.
4//! Key differences:
5//! - Uses backticks for identifiers (especially for project.dataset.table)
6//! - SAFE_ prefix for safe operations
7//! - Different date/time function names (DATE_DIFF, FORMAT_DATE, PARSE_DATE)
8//! - STRUCT and ARRAY syntax differences
9//! - No ILIKE support
10//! - QUALIFY clause support
11
12use super::{DialectImpl, DialectType};
13use crate::error::Result;
14use crate::expressions::{
15    Alias, BinaryOp, CeilFunc, Column, Exists, Expression, From, Function, FunctionBody,
16    Identifier, JsonExtractFunc, LikeOp, Literal, Select, SplitFunc, StringAggFunc, UnaryFunc,
17    UnnestFunc, VarArgFunc, Where,
18};
19#[cfg(feature = "generate")]
20use crate::generator::GeneratorConfig;
21use crate::tokens::TokenizerConfig;
22
23/// BigQuery dialect
24pub struct BigQueryDialect;
25
26impl DialectImpl for BigQueryDialect {
27    fn dialect_type(&self) -> DialectType {
28        DialectType::BigQuery
29    }
30
31    fn tokenizer_config(&self) -> TokenizerConfig {
32        let mut config = TokenizerConfig::default();
33        // BigQuery uses backticks for identifiers, NOT double quotes
34        // Remove double quote from identifiers (it's in the default config)
35        config.identifiers.remove(&'"');
36        config.identifiers.insert('`', '`');
37        // BigQuery supports double quotes for strings (in addition to single quotes)
38        config.quotes.insert("\"".to_string(), "\"".to_string());
39        // BigQuery supports triple-quoted strings
40        config.quotes.insert("'''".to_string(), "'''".to_string());
41        config
42            .quotes
43            .insert("\"\"\"".to_string(), "\"\"\"".to_string());
44        // BigQuery supports backslash escaping in strings
45        config.string_escapes = vec!['\'', '\\'];
46        // In BigQuery, b'...' is a byte string (bytes), not a bit string (binary digits)
47        config.b_prefix_is_byte_string = true;
48        // BigQuery supports hex number strings like 0xA, 0xFF
49        config.hex_number_strings = true;
50        // BigQuery: 0xA represents integer 10 (not binary/blob)
51        config.hex_string_is_integer_type = true;
52        // BigQuery supports # as single-line comments.
53        config.hash_comments = true;
54        config
55    }
56
57    #[cfg(feature = "generate")]
58
59    fn generator_config(&self) -> GeneratorConfig {
60        use crate::generator::{IdentifierQuoteStyle, NormalizeFunctions};
61        GeneratorConfig {
62            identifier_quote: '`',
63            identifier_quote_style: IdentifierQuoteStyle::BACKTICK,
64            dialect: Some(DialectType::BigQuery),
65            // BigQuery doesn't normalize function names (Python: NORMALIZE_FUNCTIONS = False)
66            normalize_functions: NormalizeFunctions::None,
67            // BigQuery-specific settings from Python sqlglot
68            interval_allows_plural_form: false,
69            join_hints: false,
70            query_hints: false,
71            table_hints: false,
72            limit_fetch_style: crate::generator::LimitFetchStyle::Limit,
73            rename_table_with_db: false,
74            nvl2_supported: false,
75            unnest_with_ordinality: false,
76            collate_is_func: true,
77            limit_only_literals: true,
78            supports_table_alias_columns: false,
79            unpivot_aliases_are_identifiers: false,
80            json_key_value_pair_sep: ",",
81            null_ordering_supported: false,
82            ignore_nulls_in_func: true,
83            json_path_single_quote_escape: true,
84            can_implement_array_any: true,
85            supports_to_number: false,
86            named_placeholder_token: "@",
87            hex_func: "TO_HEX",
88            with_properties_prefix: "OPTIONS",
89            supports_exploding_projections: false,
90            except_intersect_support_all_clause: false,
91            supports_unix_seconds: true,
92            // BigQuery uses SAFE_ prefix for safe operations
93            try_supported: true,
94            // BigQuery does not support SEMI/ANTI JOIN syntax
95            semi_anti_join_with_side: false,
96            ..Default::default()
97        }
98    }
99
100    #[cfg(feature = "transpile")]
101
102    fn transform_expr(&self, expr: Expression) -> Result<Expression> {
103        match expr {
104            // ===== Data Type Mappings =====
105            Expression::DataType(dt) => self.transform_data_type(dt),
106
107            Expression::Table(mut table)
108                if table.catalog.is_none()
109                    && table.alias.is_none()
110                    && table.schema.as_ref().map_or(false, |schema| {
111                        schema.quoted && schema.name.contains("INFORMATION_SCHEMA")
112                    }) =>
113            {
114                let schema = table.schema.take().expect("schema checked above");
115                let old_name = table.name.clone();
116                let alias = old_name.name.clone();
117                table.name = Identifier {
118                    name: format!("{}.{}", schema.name, old_name.name),
119                    quoted: true,
120                    trailing_comments: old_name.trailing_comments,
121                    span: old_name.span,
122                };
123                table.alias = Some(Identifier::new(alias));
124                Ok(Expression::Table(table))
125            }
126
127            // ===== Null handling =====
128            // IFNULL is native to BigQuery - keep as-is for identity
129            Expression::IfNull(f) => Ok(Expression::IfNull(f)),
130
131            // NVL -> IFNULL in BigQuery (BigQuery uses IFNULL, not NVL)
132            Expression::Nvl(f) => Ok(Expression::IfNull(f)),
133
134            // Coalesce stays as Coalesce
135            Expression::Coalesce(f) => Ok(Expression::Coalesce(f)),
136
137            // ===== String aggregation =====
138            // GROUP_CONCAT -> STRING_AGG in BigQuery
139            Expression::GroupConcat(f) => Ok(Expression::StringAgg(Box::new(StringAggFunc {
140                this: f.this,
141                separator: f.separator,
142                order_by: f.order_by,
143                distinct: f.distinct,
144                filter: f.filter,
145                limit: None,
146                inferred_type: None,
147            }))),
148
149            // ===== Cast operations =====
150            // Cast data types are transformed by transform_recursive in mod.rs
151            // which calls transform_data_type via transform_expr(Expression::DataType(...))
152
153            // TryCast -> SafeCast in BigQuery with type transformation
154            Expression::TryCast(c) => {
155                let transformed_type = match self.transform_data_type(c.to)? {
156                    Expression::DataType(dt) => dt,
157                    _ => return Err(crate::error::Error::parse("Expected DataType", 0, 0, 0, 0)),
158                };
159                Ok(Expression::SafeCast(Box::new(crate::expressions::Cast {
160                    this: c.this,
161                    to: transformed_type,
162                    trailing_comments: c.trailing_comments,
163                    double_colon_syntax: c.double_colon_syntax,
164                    format: c.format,
165                    default: c.default,
166                    inferred_type: None,
167                })))
168            }
169
170            // ===== Pattern matching =====
171            // ILIKE -> LOWER() LIKE LOWER() in BigQuery (no ILIKE support)
172            Expression::ILike(op) => {
173                let lower_left = Expression::Lower(Box::new(UnaryFunc::new(op.left)));
174                let lower_right = Expression::Lower(Box::new(UnaryFunc::new(op.right)));
175                Ok(Expression::Like(Box::new(LikeOp {
176                    left: lower_left,
177                    right: lower_right,
178                    escape: op.escape,
179                    quantifier: op.quantifier,
180                    inferred_type: None,
181                })))
182            }
183
184            // RegexpLike -> REGEXP_CONTAINS in BigQuery
185            Expression::RegexpLike(f) => Ok(Expression::Function(Box::new(Function::new(
186                "REGEXP_CONTAINS".to_string(),
187                vec![f.this, f.pattern],
188            )))),
189
190            // ===== Array operations =====
191            // EXPLODE -> UNNEST in BigQuery
192            Expression::Explode(f) => Ok(Expression::Unnest(Box::new(
193                crate::expressions::UnnestFunc {
194                    this: f.this,
195                    expressions: Vec::new(),
196                    with_ordinality: false,
197                    alias: None,
198                    offset_alias: None,
199                    inferred_type: None,
200                },
201            ))),
202
203            // ExplodeOuter -> UNNEST with LEFT JOIN semantics
204            Expression::ExplodeOuter(f) => Ok(Expression::Unnest(Box::new(
205                crate::expressions::UnnestFunc {
206                    this: f.this,
207                    expressions: Vec::new(),
208                    with_ordinality: false,
209                    alias: None,
210                    offset_alias: None,
211                    inferred_type: None,
212                },
213            ))),
214
215            // GenerateSeries -> GENERATE_ARRAY in BigQuery
216            Expression::GenerateSeries(f) => {
217                let mut args = Vec::new();
218                if let Some(start) = f.start {
219                    args.push(*start);
220                }
221                if let Some(end) = f.end {
222                    args.push(*end);
223                }
224                if let Some(step) = f.step {
225                    args.push(*step);
226                }
227                Ok(Expression::Function(Box::new(Function::new(
228                    "GENERATE_ARRAY".to_string(),
229                    args,
230                ))))
231            }
232
233            // ===== Bitwise operations =====
234            // BitwiseAndAgg -> BIT_AND
235            Expression::BitwiseAndAgg(f) => Ok(Expression::Function(Box::new(Function::new(
236                "BIT_AND".to_string(),
237                vec![f.this],
238            )))),
239
240            // BitwiseOrAgg -> BIT_OR
241            Expression::BitwiseOrAgg(f) => Ok(Expression::Function(Box::new(Function::new(
242                "BIT_OR".to_string(),
243                vec![f.this],
244            )))),
245
246            // BitwiseXorAgg -> BIT_XOR
247            Expression::BitwiseXorAgg(f) => Ok(Expression::Function(Box::new(Function::new(
248                "BIT_XOR".to_string(),
249                vec![f.this],
250            )))),
251
252            // BitwiseCount -> BIT_COUNT
253            Expression::BitwiseCount(f) => Ok(Expression::Function(Box::new(Function::new(
254                "BIT_COUNT".to_string(),
255                vec![f.this],
256            )))),
257
258            // ByteLength -> BYTE_LENGTH
259            Expression::ByteLength(f) => Ok(Expression::Function(Box::new(Function::new(
260                "BYTE_LENGTH".to_string(),
261                vec![f.this],
262            )))),
263
264            // IntDiv -> DIV
265            Expression::IntDiv(f) => Ok(Expression::Function(Box::new(Function::new(
266                "DIV".to_string(),
267                vec![f.this, f.expression],
268            )))),
269
270            // Int64 -> INT64
271            Expression::Int64(f) => Ok(Expression::Function(Box::new(Function::new(
272                "INT64".to_string(),
273                vec![f.this],
274            )))),
275
276            // ===== Random =====
277            // RANDOM -> RAND in BigQuery
278            Expression::Random(_) => Ok(Expression::Rand(Box::new(crate::expressions::Rand {
279                seed: None,
280                lower: None,
281                upper: None,
282            }))),
283
284            // ===== UUID =====
285            // Uuid -> GENERATE_UUID in BigQuery
286            Expression::Uuid(_) => Ok(Expression::Function(Box::new(Function::new(
287                "GENERATE_UUID".to_string(),
288                vec![],
289            )))),
290
291            // ===== Approximate functions =====
292            // ApproxDistinct -> APPROX_COUNT_DISTINCT
293            Expression::ApproxDistinct(f) => Ok(Expression::Function(Box::new(Function::new(
294                "APPROX_COUNT_DISTINCT".to_string(),
295                vec![f.this],
296            )))),
297
298            // ArgMax -> MAX_BY in BigQuery
299            Expression::ArgMax(f) => Ok(Expression::Function(Box::new(Function::new(
300                "MAX_BY".to_string(),
301                vec![*f.this, *f.expression],
302            )))),
303
304            // ArgMin -> MIN_BY in BigQuery
305            Expression::ArgMin(f) => Ok(Expression::Function(Box::new(Function::new(
306                "MIN_BY".to_string(),
307                vec![*f.this, *f.expression],
308            )))),
309
310            // ===== Conditional =====
311            // CountIf -> COUNTIF in BigQuery
312            Expression::CountIf(f) => Ok(Expression::Function(Box::new(Function::new(
313                "COUNTIF".to_string(),
314                vec![f.this],
315            )))),
316
317            // ===== String functions =====
318            // StringAgg -> STRING_AGG in BigQuery - keep as-is to preserve ORDER BY
319            Expression::StringAgg(f) => Ok(Expression::StringAgg(f)),
320
321            // ===== Conversion =====
322            // Unhex -> FROM_HEX
323            Expression::Unhex(f) => Ok(Expression::Function(Box::new(Function::new(
324                "FROM_HEX".to_string(),
325                vec![*f.this],
326            )))),
327
328            // UnixToTime -> TIMESTAMP_SECONDS/MILLIS/MICROS based on scale
329            Expression::UnixToTime(f) => {
330                let scale = f.scale.unwrap_or(0);
331                match scale {
332                    0 => Ok(Expression::Function(Box::new(Function::new(
333                        "TIMESTAMP_SECONDS".to_string(),
334                        vec![*f.this],
335                    )))),
336                    3 => Ok(Expression::Function(Box::new(Function::new(
337                        "TIMESTAMP_MILLIS".to_string(),
338                        vec![*f.this],
339                    )))),
340                    6 => Ok(Expression::Function(Box::new(Function::new(
341                        "TIMESTAMP_MICROS".to_string(),
342                        vec![*f.this],
343                    )))),
344                    _ => {
345                        // TIMESTAMP_SECONDS(CAST(value / POWER(10, scale) AS INT64))
346                        let div_expr =
347                            Expression::Div(Box::new(crate::expressions::BinaryOp::new(
348                                *f.this,
349                                Expression::Function(Box::new(Function::new(
350                                    "POWER".to_string(),
351                                    vec![Expression::number(10), Expression::number(scale)],
352                                ))),
353                            )));
354                        let cast_expr = Expression::Cast(Box::new(crate::expressions::Cast {
355                            this: div_expr,
356                            to: crate::expressions::DataType::Custom {
357                                name: "INT64".to_string(),
358                            },
359                            double_colon_syntax: false,
360                            trailing_comments: vec![],
361                            format: None,
362                            default: None,
363                            inferred_type: None,
364                        }));
365                        Ok(Expression::Function(Box::new(Function::new(
366                            "TIMESTAMP_SECONDS".to_string(),
367                            vec![cast_expr],
368                        ))))
369                    }
370                }
371            }
372
373            // ===== Date/time =====
374            // DateDiff -> DATE_DIFF in BigQuery
375            Expression::DateDiff(f) => {
376                // BigQuery: DATE_DIFF(date1, date2, part)
377                let unit_str = match f.unit {
378                    Some(crate::expressions::IntervalUnit::Year) => "YEAR",
379                    Some(crate::expressions::IntervalUnit::Quarter) => "QUARTER",
380                    Some(crate::expressions::IntervalUnit::Month) => "MONTH",
381                    Some(crate::expressions::IntervalUnit::Week) => "WEEK",
382                    Some(crate::expressions::IntervalUnit::Day) => "DAY",
383                    Some(crate::expressions::IntervalUnit::Hour) => "HOUR",
384                    Some(crate::expressions::IntervalUnit::Minute) => "MINUTE",
385                    Some(crate::expressions::IntervalUnit::Second) => "SECOND",
386                    Some(crate::expressions::IntervalUnit::Millisecond) => "MILLISECOND",
387                    Some(crate::expressions::IntervalUnit::Microsecond) => "MICROSECOND",
388                    Some(crate::expressions::IntervalUnit::Nanosecond) => "NANOSECOND",
389                    None => "DAY",
390                };
391                let unit = Expression::Identifier(crate::expressions::Identifier {
392                    name: unit_str.to_string(),
393                    quoted: false,
394                    trailing_comments: Vec::new(),
395                    span: None,
396                });
397                Ok(Expression::Function(Box::new(Function::new(
398                    "DATE_DIFF".to_string(),
399                    vec![f.this, f.expression, unit],
400                ))))
401            }
402
403            // ===== Variance =====
404            // VarPop -> VAR_POP
405            Expression::VarPop(f) => Ok(Expression::Function(Box::new(Function::new(
406                "VAR_POP".to_string(),
407                vec![f.this],
408            )))),
409
410            // ===== Hash functions =====
411            // SHA -> SHA1
412            Expression::SHA(f) => Ok(Expression::Function(Box::new(Function::new(
413                "SHA1".to_string(),
414                vec![f.this],
415            )))),
416
417            // SHA1Digest -> SHA1
418            Expression::SHA1Digest(f) => Ok(Expression::Function(Box::new(Function::new(
419                "SHA1".to_string(),
420                vec![f.this],
421            )))),
422
423            // MD5Digest -> MD5
424            Expression::MD5Digest(f) => Ok(Expression::Function(Box::new(Function::new(
425                "MD5".to_string(),
426                vec![*f.this],
427            )))),
428
429            // ===== Type conversion =====
430            // JSONBool -> BOOL
431            Expression::JSONBool(f) => Ok(Expression::Function(Box::new(Function::new(
432                "BOOL".to_string(),
433                vec![f.this],
434            )))),
435
436            // StringFunc -> STRING
437            Expression::StringFunc(f) => Ok(Expression::Function(Box::new(Function::new(
438                "STRING".to_string(),
439                vec![*f.this],
440            )))),
441
442            // ===== Date/time from parts =====
443            // DateFromUnixDate -> DATE_FROM_UNIX_DATE
444            Expression::DateFromUnixDate(f) => Ok(Expression::Function(Box::new(Function::new(
445                "DATE_FROM_UNIX_DATE".to_string(),
446                vec![f.this],
447            )))),
448
449            // UnixDate -> UNIX_DATE
450            Expression::UnixDate(f) => Ok(Expression::Function(Box::new(Function::new(
451                "UNIX_DATE".to_string(),
452                vec![f.this],
453            )))),
454
455            // TimestampDiff -> TIMESTAMP_DIFF
456            Expression::TimestampDiff(f) => Ok(Expression::Function(Box::new(Function::new(
457                "TIMESTAMP_DIFF".to_string(),
458                vec![*f.this, *f.expression],
459            )))),
460
461            // FromTimeZone -> DATETIME
462            Expression::FromTimeZone(f) => Ok(Expression::Function(Box::new(Function::new(
463                "DATETIME".to_string(),
464                vec![*f.this],
465            )))),
466
467            // TsOrDsToDatetime -> DATETIME
468            Expression::TsOrDsToDatetime(f) => Ok(Expression::Function(Box::new(Function::new(
469                "DATETIME".to_string(),
470                vec![f.this],
471            )))),
472
473            // TsOrDsToTimestamp -> TIMESTAMP
474            Expression::TsOrDsToTimestamp(f) => Ok(Expression::Function(Box::new(Function::new(
475                "TIMESTAMP".to_string(),
476                vec![f.this],
477            )))),
478
479            // ===== IfFunc -> IF in BigQuery =====
480            Expression::IfFunc(f) => {
481                let mut args = vec![f.condition, f.true_value];
482                if let Some(false_val) = f.false_value {
483                    args.push(false_val);
484                } else {
485                    args.push(Expression::Null(crate::expressions::Null));
486                }
487                Ok(Expression::Function(Box::new(Function::new(
488                    "IF".to_string(),
489                    args,
490                ))))
491            }
492
493            // ===== HexString -> FROM_HEX =====
494            Expression::HexStringExpr(f) => Ok(Expression::Function(Box::new(Function::new(
495                "FROM_HEX".to_string(),
496                vec![*f.this],
497            )))),
498
499            // ===== Additional auto-generated transforms from Python sqlglot =====
500            // ApproxTopK -> APPROX_TOP_COUNT
501            Expression::ApproxTopK(f) => {
502                let mut args = vec![*f.this];
503                if let Some(expr) = f.expression {
504                    args.push(*expr);
505                }
506                Ok(Expression::Function(Box::new(Function::new(
507                    "APPROX_TOP_COUNT".to_string(),
508                    args,
509                ))))
510            }
511
512            // SafeDivide -> SAFE_DIVIDE
513            Expression::SafeDivide(f) => Ok(Expression::Function(Box::new(Function::new(
514                "SAFE_DIVIDE".to_string(),
515                vec![*f.this, *f.expression],
516            )))),
517
518            // JSONKeysAtDepth -> JSON_KEYS
519            Expression::JSONKeysAtDepth(f) => Ok(Expression::Function(Box::new(Function::new(
520                "JSON_KEYS".to_string(),
521                vec![*f.this],
522            )))),
523
524            // JSONValueArray -> JSON_VALUE_ARRAY
525            Expression::JSONValueArray(f) => Ok(Expression::Function(Box::new(Function::new(
526                "JSON_VALUE_ARRAY".to_string(),
527                vec![*f.this],
528            )))),
529
530            // DateFromParts -> DATE
531            Expression::DateFromParts(f) => {
532                let mut args = Vec::new();
533                if let Some(y) = f.year {
534                    args.push(*y);
535                }
536                if let Some(m) = f.month {
537                    args.push(*m);
538                }
539                if let Some(d) = f.day {
540                    args.push(*d);
541                }
542                Ok(Expression::Function(Box::new(Function::new(
543                    "DATE".to_string(),
544                    args,
545                ))))
546            }
547
548            // SPLIT: BigQuery defaults to comma separator when none provided
549            // SPLIT(foo) -> SPLIT(foo, ',')
550            Expression::Split(f) => {
551                // Check if delimiter is empty or a placeholder - add default comma
552                let delimiter = match &f.delimiter {
553                    Expression::Literal(lit) if matches!(lit.as_ref(), Literal::String(s) if s.is_empty()) =>
554                    {
555                        let Literal::String(_) = lit.as_ref() else {
556                            unreachable!()
557                        };
558                        Expression::Literal(Box::new(Literal::String(",".to_string())))
559                    }
560                    _ => f.delimiter,
561                };
562                Ok(Expression::Split(Box::new(SplitFunc {
563                    this: f.this,
564                    delimiter,
565                })))
566            }
567
568            // Cast: Transform the target type according to BigQuery TYPE_MAPPING
569            // Special case: CAST to JSON -> PARSE_JSON in BigQuery
570            // Special case: CAST(x AS temporal FORMAT 'fmt') -> PARSE_DATE/PARSE_TIMESTAMP
571            Expression::Cast(c) => {
572                use crate::expressions::DataType;
573                // Check if casting to JSON - use PARSE_JSON instead
574                let is_json = matches!(c.to, DataType::Json | DataType::JsonB)
575                    || matches!(&c.to, DataType::Custom { name } if name.eq_ignore_ascii_case("JSON") || name.eq_ignore_ascii_case("JSONB"));
576                if is_json {
577                    return Ok(Expression::ParseJson(Box::new(UnaryFunc::new(c.this))));
578                }
579                // CAST(x AS temporal_type FORMAT 'fmt') -> PARSE_DATE/PARSE_TIMESTAMP(strftime_fmt, x)
580                if c.format.is_some() {
581                    let is_temporal = matches!(
582                        c.to,
583                        DataType::Date | DataType::Timestamp { .. } | DataType::Time { .. }
584                    ) || matches!(&c.to, DataType::Custom { name } if
585                        name.eq_ignore_ascii_case("TIMESTAMP") ||
586                        name.eq_ignore_ascii_case("DATE") ||
587                        name.eq_ignore_ascii_case("DATETIME") ||
588                        name.eq_ignore_ascii_case("TIME")
589                    );
590                    if is_temporal {
591                        let format_expr = c.format.as_ref().unwrap().as_ref();
592                        // Extract the actual format expr and timezone (if AT TIME ZONE is present)
593                        let (actual_format, timezone) = match format_expr {
594                            Expression::AtTimeZone(ref atz) => {
595                                (atz.this.clone(), Some(atz.zone.clone()))
596                            }
597                            _ => (format_expr.clone(), None),
598                        };
599                        let strftime_fmt = Self::bq_cast_format_to_strftime(&actual_format);
600                        let func_name = match &c.to {
601                            DataType::Date => "PARSE_DATE",
602                            DataType::Custom { name } if name.eq_ignore_ascii_case("DATE") => {
603                                "PARSE_DATE"
604                            }
605                            DataType::Custom { name } if name.eq_ignore_ascii_case("DATETIME") => {
606                                "PARSE_DATETIME"
607                            }
608                            _ => "PARSE_TIMESTAMP",
609                        };
610                        let mut func_args = vec![strftime_fmt, c.this];
611                        if let Some(tz) = timezone {
612                            func_args.push(tz);
613                        }
614                        return Ok(Expression::Function(Box::new(Function::new(
615                            func_name.to_string(),
616                            func_args,
617                        ))));
618                    }
619                }
620                let transformed_type = match self.transform_data_type(c.to)? {
621                    Expression::DataType(dt) => dt,
622                    _ => return Err(crate::error::Error::parse("Expected DataType", 0, 0, 0, 0)),
623                };
624                Ok(Expression::Cast(Box::new(crate::expressions::Cast {
625                    this: c.this,
626                    to: transformed_type,
627                    trailing_comments: c.trailing_comments,
628                    double_colon_syntax: c.double_colon_syntax,
629                    format: c.format,
630                    default: c.default,
631                    inferred_type: None,
632                })))
633            }
634
635            // SafeCast: Transform the target type according to BigQuery TYPE_MAPPING
636            Expression::SafeCast(c) => {
637                let transformed_type = match self.transform_data_type(c.to)? {
638                    Expression::DataType(dt) => dt,
639                    _ => return Err(crate::error::Error::parse("Expected DataType", 0, 0, 0, 0)),
640                };
641                Ok(Expression::SafeCast(Box::new(crate::expressions::Cast {
642                    this: c.this,
643                    to: transformed_type,
644                    trailing_comments: c.trailing_comments,
645                    double_colon_syntax: c.double_colon_syntax,
646                    format: c.format,
647                    default: c.default,
648                    inferred_type: None,
649                })))
650            }
651
652            // ===== SELECT-level transforms =====
653            // BigQuery: GROUP BY expression → alias when both GROUP BY and ORDER BY exist
654            Expression::Select(mut select) => {
655                if select.group_by.is_some() && select.order_by.is_some() {
656                    // Build map: expression → alias name for aliased projections
657                    let aliases: Vec<(Expression, Identifier)> = select
658                        .expressions
659                        .iter()
660                        .filter_map(|e| {
661                            if let Expression::Alias(a) = e {
662                                Some((a.this.clone(), a.alias.clone()))
663                            } else {
664                                None
665                            }
666                        })
667                        .collect();
668
669                    if let Some(ref mut group_by) = select.group_by {
670                        for grouped in group_by.expressions.iter_mut() {
671                            // Skip numeric indices (already aliased)
672                            if matches!(grouped, Expression::Literal(lit) if matches!(lit.as_ref(), Literal::Number(_)))
673                            {
674                                continue;
675                            }
676                            // Check if this GROUP BY expression matches a SELECT alias
677                            for (expr, alias_ident) in &aliases {
678                                if grouped == expr {
679                                    *grouped = Expression::boxed_column(Column {
680                                        name: alias_ident.clone(),
681                                        table: None,
682                                        join_mark: false,
683                                        trailing_comments: Vec::new(),
684                                        span: None,
685                                        inferred_type: None,
686                                    });
687                                    break;
688                                }
689                            }
690                        }
691                    }
692                }
693                Ok(Expression::Select(select))
694            }
695
696            // ===== ArrayContains → EXISTS(SELECT 1 FROM UNNEST(arr) AS _col WHERE _col = val) =====
697            Expression::ArrayContains(f) => {
698                let array_expr = f.this;
699                let value_expr = f.expression;
700
701                // Build: SELECT 1 FROM UNNEST(array) AS _col WHERE _col = value
702                let unnest = Expression::Unnest(Box::new(UnnestFunc {
703                    this: array_expr,
704                    expressions: Vec::new(),
705                    with_ordinality: false,
706                    alias: None,
707                    offset_alias: None,
708                    inferred_type: None,
709                }));
710                let aliased_unnest = Expression::Alias(Box::new(Alias {
711                    this: unnest,
712                    alias: Identifier::new("_col"),
713                    column_aliases: Vec::new(),
714                    alias_explicit_as: false,
715                    alias_keyword: None,
716                    pre_alias_comments: Vec::new(),
717                    trailing_comments: Vec::new(),
718                    inferred_type: None,
719                }));
720                let col_ref = Expression::boxed_column(Column {
721                    name: Identifier::new("_col"),
722                    table: None,
723                    join_mark: false,
724                    trailing_comments: Vec::new(),
725                    span: None,
726                    inferred_type: None,
727                });
728                let where_clause = Where {
729                    this: Expression::Eq(Box::new(BinaryOp {
730                        left: col_ref,
731                        right: value_expr,
732                        left_comments: Vec::new(),
733                        operator_comments: Vec::new(),
734                        trailing_comments: Vec::new(),
735                        inferred_type: None,
736                    })),
737                };
738                let inner_select = Expression::Select(Box::new(Select {
739                    expressions: vec![Expression::Literal(Box::new(Literal::Number(
740                        "1".to_string(),
741                    )))],
742                    from: Some(From {
743                        expressions: vec![aliased_unnest],
744                    }),
745                    where_clause: Some(where_clause),
746                    ..Default::default()
747                }));
748                Ok(Expression::Exists(Box::new(Exists {
749                    this: inner_select,
750                    not: false,
751                })))
752            }
753
754            // ===== JSON_OBJECT array form → key-value pairs =====
755            // BigQuery "signature 2": JSON_OBJECT(['a', 'b'], [10, NULL]) → JSON_OBJECT('a', 10, 'b', NULL)
756            Expression::JsonObject(mut f) => {
757                if f.pairs.len() == 1 {
758                    // Extract expressions from both Array and ArrayFunc variants
759                    let keys_exprs = match &f.pairs[0].0 {
760                        Expression::Array(arr) => Some(&arr.expressions),
761                        Expression::ArrayFunc(arr) => Some(&arr.expressions),
762                        _ => None,
763                    };
764                    let vals_exprs = match &f.pairs[0].1 {
765                        Expression::Array(arr) => Some(&arr.expressions),
766                        Expression::ArrayFunc(arr) => Some(&arr.expressions),
767                        _ => None,
768                    };
769                    if let (Some(keys), Some(vals)) = (keys_exprs, vals_exprs) {
770                        if keys.len() == vals.len() {
771                            let new_pairs: Vec<(Expression, Expression)> = keys
772                                .iter()
773                                .zip(vals.iter())
774                                .map(|(k, v)| (k.clone(), v.clone()))
775                                .collect();
776                            f.pairs = new_pairs;
777                        }
778                    }
779                }
780                Ok(Expression::JsonObject(f))
781            }
782
783            // ===== MOD function: unwrap unnecessary Paren from first argument =====
784            // BigQuery normalizes MOD((a + 1), b) -> MOD(a + 1, b)
785            Expression::ModFunc(mut f) => {
786                // Unwrap Paren from first argument if present
787                if let Expression::Paren(paren) = f.this {
788                    f.this = paren.this;
789                }
790                Ok(Expression::ModFunc(f))
791            }
792
793            // JSONExtract with variant_extract (Snowflake colon syntax) -> JSON_EXTRACT
794            Expression::JSONExtract(e) if e.variant_extract.is_some() => {
795                let path = match *e.expression {
796                    Expression::Literal(lit) if matches!(lit.as_ref(), Literal::String(_)) => {
797                        let Literal::String(s) = lit.as_ref() else {
798                            unreachable!()
799                        };
800                        let normalized = if s.starts_with('$') {
801                            s.clone()
802                        } else if s.starts_with('[') {
803                            format!("${}", s)
804                        } else {
805                            format!("$.{}", s)
806                        };
807                        Expression::Literal(Box::new(Literal::String(normalized)))
808                    }
809                    other => other,
810                };
811                Ok(Expression::Function(Box::new(Function::new(
812                    "JSON_EXTRACT".to_string(),
813                    vec![*e.this, path],
814                ))))
815            }
816
817            // Generic function transformations
818            Expression::Function(f) => self.transform_function(*f),
819
820            // Generic aggregate function transformations
821            Expression::AggregateFunction(f) => self.transform_aggregate_function(f),
822
823            // MethodCall: Handle SAFE.PARSE_DATE, SAFE.PARSE_DATETIME, SAFE.PARSE_TIMESTAMP
824            // These are parsed as MethodCall(this=SAFE, method=PARSE_DATE, args=[...])
825            Expression::MethodCall(mc) => self.transform_method_call(*mc),
826
827            // CreateFunction: Convert RETURNS TABLE(...) to RETURNS TABLE <...> for BigQuery
828            // and set is_table_function = true for TABLE FUNCTION syntax
829            Expression::CreateFunction(mut cf) => {
830                if let Some(ref mut rtb) = cf.returns_table_body {
831                    if rtb.starts_with("TABLE (") || rtb.starts_with("TABLE(") {
832                        // Convert TABLE (...) to TABLE <...> with BigQuery types
833                        let inner = if rtb.starts_with("TABLE (") {
834                            &rtb["TABLE (".len()..rtb.len() - 1]
835                        } else {
836                            &rtb["TABLE(".len()..rtb.len() - 1]
837                        };
838                        // Convert common types to BigQuery equivalents
839                        let converted = inner
840                            .replace(" INT,", " INT64,")
841                            .replace(" INT)", " INT64)")
842                            .replace(" INTEGER,", " INT64,")
843                            .replace(" INTEGER)", " INT64)")
844                            .replace(" FLOAT,", " FLOAT64,")
845                            .replace(" FLOAT)", " FLOAT64)")
846                            .replace(" BOOLEAN,", " BOOL,")
847                            .replace(" BOOLEAN)", " BOOL)")
848                            .replace(" VARCHAR", " STRING")
849                            .replace(" TEXT", " STRING");
850                        // Handle trailing type (no comma, no paren)
851                        let converted = if converted.ends_with(" INT") {
852                            format!("{}{}", &converted[..converted.len() - 4], " INT64")
853                        } else {
854                            converted
855                        };
856                        *rtb = format!("TABLE <{}>", converted);
857                        cf.is_table_function = true;
858                    }
859                }
860                // Convert string literal body to expression body for BigQuery TABLE FUNCTIONs only
861                if cf.is_table_function {
862                    if let Some(ref body) = cf.body {
863                        if matches!(body, FunctionBody::StringLiteral(_)) {
864                            if let Some(FunctionBody::StringLiteral(sql)) = cf.body.take() {
865                                // Parse the SQL string into an expression
866                                if let Ok(parsed) = crate::parser::Parser::parse_sql(&sql) {
867                                    if let Some(stmt) = parsed.into_iter().next() {
868                                        cf.body = Some(FunctionBody::Expression(stmt));
869                                    } else {
870                                        cf.body = Some(FunctionBody::StringLiteral(sql));
871                                    }
872                                } else {
873                                    cf.body = Some(FunctionBody::StringLiteral(sql));
874                                }
875                            }
876                        }
877                    }
878                }
879                Ok(Expression::CreateFunction(cf))
880            }
881
882            // Pass through everything else
883            _ => Ok(expr),
884        }
885    }
886}
887
888#[cfg(feature = "transpile")]
889impl BigQueryDialect {
890    /// Transform data types according to BigQuery TYPE_MAPPING
891    fn transform_data_type(&self, dt: crate::expressions::DataType) -> Result<Expression> {
892        use crate::expressions::DataType;
893        let transformed = match dt {
894            // BIGINT -> INT64
895            DataType::BigInt { .. } => DataType::Custom {
896                name: "INT64".to_string(),
897            },
898            // INT -> INT64
899            DataType::Int { .. } => DataType::Custom {
900                name: "INT64".to_string(),
901            },
902            // SMALLINT -> INT64
903            DataType::SmallInt { .. } => DataType::Custom {
904                name: "INT64".to_string(),
905            },
906            // TINYINT -> INT64
907            DataType::TinyInt { .. } => DataType::Custom {
908                name: "INT64".to_string(),
909            },
910            // FLOAT -> FLOAT64
911            DataType::Float { .. } => DataType::Custom {
912                name: "FLOAT64".to_string(),
913            },
914            // DOUBLE -> FLOAT64
915            DataType::Double { .. } => DataType::Custom {
916                name: "FLOAT64".to_string(),
917            },
918            // BOOLEAN -> BOOL
919            DataType::Boolean => DataType::Custom {
920                name: "BOOL".to_string(),
921            },
922            // CHAR -> STRING
923            DataType::Char { .. } => DataType::Custom {
924                name: "STRING".to_string(),
925            },
926            // VARCHAR -> STRING
927            DataType::VarChar { .. } => DataType::Custom {
928                name: "STRING".to_string(),
929            },
930            // TEXT -> STRING
931            DataType::Text => DataType::Custom {
932                name: "STRING".to_string(),
933            },
934            // STRING(n) -> STRING (BigQuery doesn't support length for STRING)
935            DataType::String { .. } => DataType::Custom {
936                name: "STRING".to_string(),
937            },
938            // BINARY -> BYTES
939            DataType::Binary { .. } => DataType::Custom {
940                name: "BYTES".to_string(),
941            },
942            // VARBINARY -> BYTES
943            DataType::VarBinary { .. } => DataType::Custom {
944                name: "BYTES".to_string(),
945            },
946            // BLOB -> BYTES
947            DataType::Blob => DataType::Custom {
948                name: "BYTES".to_string(),
949            },
950            // DECIMAL -> NUMERIC (BigQuery strips precision in CAST context)
951            DataType::Decimal { .. } => DataType::Custom {
952                name: "NUMERIC".to_string(),
953            },
954            // For BigQuery identity: preserve TIMESTAMP/DATETIME as Custom types
955            // This avoids the issue where parsed TIMESTAMP (timezone: false) would
956            // be converted to DATETIME by the generator
957            DataType::Timestamp {
958                timezone: false, ..
959            } => DataType::Custom {
960                name: "TIMESTAMP".to_string(),
961            },
962            DataType::Timestamp { timezone: true, .. } => DataType::Custom {
963                name: "TIMESTAMP".to_string(),
964            },
965            // UUID -> STRING (BigQuery doesn't have native UUID type)
966            DataType::Uuid => DataType::Custom {
967                name: "STRING".to_string(),
968            },
969            // RECORD -> STRUCT in BigQuery
970            DataType::Custom { ref name } if name.eq_ignore_ascii_case("RECORD") => {
971                DataType::Custom {
972                    name: "STRUCT".to_string(),
973                }
974            }
975            // TIMESTAMPTZ (custom) -> TIMESTAMP
976            DataType::Custom { ref name } if name.eq_ignore_ascii_case("TIMESTAMPTZ") => {
977                DataType::Custom {
978                    name: "TIMESTAMP".to_string(),
979                }
980            }
981            // BYTEINT (custom) -> INT64
982            DataType::Custom { ref name } if name.eq_ignore_ascii_case("BYTEINT") => {
983                DataType::Custom {
984                    name: "INT64".to_string(),
985                }
986            }
987            // Keep all other types as-is
988            other => other,
989        };
990        Ok(Expression::DataType(transformed))
991    }
992
993    fn transform_function(&self, f: Function) -> Result<Expression> {
994        let name_upper = f.name.to_uppercase();
995        match name_upper.as_str() {
996            // IFNULL -> COALESCE (both work in BigQuery)
997            "IFNULL" if f.args.len() == 2 => Ok(Expression::Coalesce(Box::new(VarArgFunc {
998                original_name: None,
999                expressions: f.args,
1000                inferred_type: None,
1001            }))),
1002
1003            // NVL -> COALESCE
1004            "NVL" if f.args.len() == 2 => Ok(Expression::Coalesce(Box::new(VarArgFunc {
1005                original_name: None,
1006                expressions: f.args,
1007                inferred_type: None,
1008            }))),
1009
1010            // ISNULL -> COALESCE
1011            "ISNULL" if f.args.len() == 2 => Ok(Expression::Coalesce(Box::new(VarArgFunc {
1012                original_name: None,
1013                expressions: f.args,
1014                inferred_type: None,
1015            }))),
1016
1017            // GROUP_CONCAT -> STRING_AGG in BigQuery
1018            "GROUP_CONCAT" if !f.args.is_empty() => Ok(Expression::Function(Box::new(
1019                Function::new("STRING_AGG".to_string(), f.args),
1020            ))),
1021
1022            // SUBSTR -> SUBSTRING (both work)
1023            "SUBSTR" => Ok(Expression::Function(Box::new(Function::new(
1024                "SUBSTRING".to_string(),
1025                f.args,
1026            )))),
1027
1028            // RANDOM -> RAND
1029            "RANDOM" => Ok(Expression::Rand(Box::new(crate::expressions::Rand {
1030                seed: None,
1031                lower: None,
1032                upper: None,
1033            }))),
1034
1035            // CURRENT_DATE -> CURRENT_DATE() in BigQuery
1036            // Keep as Function when it has args (e.g., CURRENT_DATE('UTC'))
1037            "CURRENT_DATE" if f.args.is_empty() => {
1038                Ok(Expression::CurrentDate(crate::expressions::CurrentDate))
1039            }
1040            "CURRENT_DATE" => Ok(Expression::Function(Box::new(Function {
1041                name: "CURRENT_DATE".to_string(),
1042                args: f.args,
1043                distinct: false,
1044                trailing_comments: Vec::new(),
1045                use_bracket_syntax: false,
1046                no_parens: false,
1047                quoted: false,
1048                span: None,
1049                inferred_type: None,
1050            }))),
1051
1052            // NOW -> CURRENT_TIMESTAMP in BigQuery
1053            "NOW" => Ok(Expression::CurrentTimestamp(
1054                crate::expressions::CurrentTimestamp {
1055                    precision: None,
1056                    sysdate: false,
1057                },
1058            )),
1059
1060            // TO_DATE -> PARSE_DATE in BigQuery
1061            "TO_DATE" => Ok(Expression::Function(Box::new(Function::new(
1062                "PARSE_DATE".to_string(),
1063                f.args,
1064            )))),
1065
1066            // TO_TIMESTAMP -> PARSE_TIMESTAMP in BigQuery
1067            "TO_TIMESTAMP" => Ok(Expression::Function(Box::new(Function::new(
1068                "PARSE_TIMESTAMP".to_string(),
1069                f.args,
1070            )))),
1071
1072            // TO_TIME -> TIME in BigQuery
1073            "TO_TIME" if f.args.len() == 1 => Ok(Expression::Function(Box::new(Function::new(
1074                "TIME".to_string(),
1075                f.args,
1076            )))),
1077
1078            // DATE_FORMAT -> FORMAT_DATE in BigQuery (argument order may differ)
1079            "DATE_FORMAT" => Ok(Expression::Function(Box::new(Function::new(
1080                "FORMAT_DATE".to_string(),
1081                f.args,
1082            )))),
1083
1084            // POSITION -> STRPOS in BigQuery
1085            // BigQuery uses STRPOS(string, substring)
1086            "POSITION" if f.args.len() == 2 => {
1087                let mut args = f.args;
1088                // Swap arguments: POSITION(sub IN str) -> STRPOS(str, sub)
1089                let first = args.remove(0);
1090                let second = args.remove(0);
1091                Ok(Expression::Function(Box::new(Function::new(
1092                    "STRPOS".to_string(),
1093                    vec![second, first],
1094                ))))
1095            }
1096
1097            // LEN -> LENGTH
1098            "LEN" if f.args.len() == 1 => Ok(Expression::Length(Box::new(UnaryFunc::new(
1099                f.args.into_iter().next().unwrap(),
1100            )))),
1101
1102            // CEILING -> CEIL (both work)
1103            "CEILING" if f.args.len() == 1 => Ok(Expression::Ceil(Box::new(CeilFunc {
1104                this: f.args.into_iter().next().unwrap(),
1105                decimals: None,
1106                to: None,
1107            }))),
1108
1109            // GETDATE -> CURRENT_TIMESTAMP
1110            "GETDATE" => Ok(Expression::CurrentTimestamp(
1111                crate::expressions::CurrentTimestamp {
1112                    precision: None,
1113                    sysdate: false,
1114                },
1115            )),
1116
1117            // ARRAY_LENGTH -> ARRAY_LENGTH (native)
1118            // CARDINALITY -> ARRAY_LENGTH
1119            "CARDINALITY" if f.args.len() == 1 => Ok(Expression::ArrayLength(Box::new(
1120                UnaryFunc::new(f.args.into_iter().next().unwrap()),
1121            ))),
1122
1123            // UNNEST is native to BigQuery
1124
1125            // GENERATE_SERIES -> GENERATE_ARRAY in BigQuery
1126            "GENERATE_SERIES" => Ok(Expression::Function(Box::new(Function::new(
1127                "GENERATE_ARRAY".to_string(),
1128                f.args,
1129            )))),
1130
1131            // APPROX_COUNT_DISTINCT -> APPROX_COUNT_DISTINCT (native)
1132            // APPROX_DISTINCT -> APPROX_COUNT_DISTINCT
1133            "APPROX_DISTINCT" => Ok(Expression::Function(Box::new(Function::new(
1134                "APPROX_COUNT_DISTINCT".to_string(),
1135                f.args,
1136            )))),
1137
1138            // COUNT_IF -> COUNTIF in BigQuery
1139            "COUNT_IF" => Ok(Expression::Function(Box::new(Function::new(
1140                "COUNTIF".to_string(),
1141                f.args,
1142            )))),
1143
1144            // SHA1 -> SHA1 (native), SHA -> SHA1
1145            "SHA" => Ok(Expression::Function(Box::new(Function::new(
1146                "SHA1".to_string(),
1147                f.args,
1148            )))),
1149
1150            // SHA256/SHA2 -> SHA256
1151            "SHA2" => Ok(Expression::Function(Box::new(Function::new(
1152                "SHA256".to_string(),
1153                f.args,
1154            )))),
1155
1156            // MD5 in BigQuery returns bytes, often combined with TO_HEX
1157            // TO_HEX(MD5(x)) pattern
1158            "MD5" => Ok(Expression::Function(Box::new(Function::new(
1159                "MD5".to_string(),
1160                f.args,
1161            )))),
1162
1163            // VARIANCE/VAR_SAMP -> VAR_SAMP (native)
1164            // VAR_POP -> VAR_POP (native)
1165
1166            // DATEADD(unit, amount, date) → DATE_ADD(date, INTERVAL amount unit) for BigQuery
1167            "DATEADD" if f.args.len() == 3 => {
1168                let mut args = f.args;
1169                let unit_expr = args.remove(0);
1170                let amount = args.remove(0);
1171                let date = args.remove(0);
1172                // Convert unit identifier to IntervalUnit
1173                let unit_name = match &unit_expr {
1174                    Expression::Identifier(id) => id.name.to_uppercase(),
1175                    Expression::Var(v) => v.this.to_uppercase(),
1176                    Expression::Column(col) if col.table.is_none() => col.name.name.to_uppercase(),
1177                    _ => "DAY".to_string(),
1178                };
1179                let unit = match unit_name.as_str() {
1180                    "YEAR" | "YEARS" | "YY" | "YYYY" => crate::expressions::IntervalUnit::Year,
1181                    "QUARTER" | "QUARTERS" | "QQ" | "Q" => {
1182                        crate::expressions::IntervalUnit::Quarter
1183                    }
1184                    "MONTH" | "MONTHS" | "MM" | "M" => crate::expressions::IntervalUnit::Month,
1185                    "WEEK" | "WEEKS" | "WK" | "WW" => crate::expressions::IntervalUnit::Week,
1186                    "DAY" | "DAYS" | "DD" | "D" | "DAYOFMONTH" => {
1187                        crate::expressions::IntervalUnit::Day
1188                    }
1189                    "HOUR" | "HOURS" | "HH" => crate::expressions::IntervalUnit::Hour,
1190                    "MINUTE" | "MINUTES" | "MI" | "N" => crate::expressions::IntervalUnit::Minute,
1191                    "SECOND" | "SECONDS" | "SS" | "S" => crate::expressions::IntervalUnit::Second,
1192                    "MILLISECOND" | "MILLISECONDS" | "MS" => {
1193                        crate::expressions::IntervalUnit::Millisecond
1194                    }
1195                    "MICROSECOND" | "MICROSECONDS" | "US" => {
1196                        crate::expressions::IntervalUnit::Microsecond
1197                    }
1198                    _ => crate::expressions::IntervalUnit::Day,
1199                };
1200                Ok(Expression::DateAdd(Box::new(
1201                    crate::expressions::DateAddFunc {
1202                        this: date,
1203                        interval: amount,
1204                        unit,
1205                    },
1206                )))
1207            }
1208            "DATE_ADD" => Ok(Expression::Function(Box::new(Function::new(
1209                "DATE_ADD".to_string(),
1210                f.args,
1211            )))),
1212
1213            // DATE_DIFF in BigQuery (native)
1214            "DATEDIFF" => Ok(Expression::Function(Box::new(Function::new(
1215                "DATE_DIFF".to_string(),
1216                f.args,
1217            )))),
1218
1219            // TIMESTAMP_DIFF in BigQuery
1220            "TIMESTAMPDIFF" => Ok(Expression::Function(Box::new(Function::new(
1221                "TIMESTAMP_DIFF".to_string(),
1222                f.args,
1223            )))),
1224
1225            // TIME -> TIME (native)
1226            // DATETIME -> DATETIME (native)
1227
1228            // SAFE_DIVIDE -> SAFE_DIVIDE (native)
1229
1230            // NEWID/UUID -> GENERATE_UUID
1231            "NEWID" | "UUID" => Ok(Expression::Function(Box::new(Function::new(
1232                "GENERATE_UUID".to_string(),
1233                vec![],
1234            )))),
1235
1236            // LEVENSHTEIN -> EDIT_DISTANCE (BigQuery naming)
1237            "LEVENSHTEIN" => Ok(Expression::Function(Box::new(Function::new(
1238                "EDIT_DISTANCE".to_string(),
1239                f.args,
1240            )))),
1241
1242            // UNIX_TIMESTAMP -> UNIX_SECONDS
1243            "UNIX_TIMESTAMP" => Ok(Expression::Function(Box::new(Function::new(
1244                "UNIX_SECONDS".to_string(),
1245                f.args,
1246            )))),
1247
1248            // FROM_UNIXTIME -> TIMESTAMP_SECONDS
1249            "FROM_UNIXTIME" => Ok(Expression::Function(Box::new(Function::new(
1250                "TIMESTAMP_SECONDS".to_string(),
1251                f.args,
1252            )))),
1253
1254            // CHAR_LENGTH / CHARACTER_LENGTH -> LENGTH
1255            "CHAR_LENGTH" | "CHARACTER_LENGTH" => Ok(Expression::Function(Box::new(
1256                Function::new("LENGTH".to_string(), f.args),
1257            ))),
1258
1259            // OCTET_LENGTH -> BYTE_LENGTH in BigQuery
1260            "OCTET_LENGTH" => Ok(Expression::Function(Box::new(Function::new(
1261                "BYTE_LENGTH".to_string(),
1262                f.args,
1263            )))),
1264
1265            // JSON_EXTRACT_STRING_ARRAY -> JSON_VALUE_ARRAY in BigQuery
1266            "JSON_EXTRACT_STRING_ARRAY" => Ok(Expression::Function(Box::new(Function::new(
1267                "JSON_VALUE_ARRAY".to_string(),
1268                f.args,
1269            )))),
1270
1271            // INSTR is native to BigQuery
1272
1273            // SPLIT: BigQuery defaults to comma separator when none provided
1274            // SPLIT(foo) -> SPLIT(foo, ',')
1275            "SPLIT" if f.args.len() == 1 => {
1276                let mut args = f.args;
1277                args.push(Expression::Literal(Box::new(Literal::String(
1278                    ",".to_string(),
1279                ))));
1280                Ok(Expression::Split(Box::new(SplitFunc {
1281                    this: args.remove(0),
1282                    delimiter: args.remove(0),
1283                })))
1284            }
1285
1286            // SPLIT with two args - convert to Split expression
1287            "SPLIT" if f.args.len() == 2 => {
1288                let mut args = f.args;
1289                Ok(Expression::Split(Box::new(SplitFunc {
1290                    this: args.remove(0),
1291                    delimiter: args.remove(0),
1292                })))
1293            }
1294
1295            // REGEXP_SUBSTR -> REGEXP_EXTRACT in BigQuery (strip extra Snowflake args)
1296            "REGEXP_SUBSTR" if f.args.len() >= 2 => {
1297                // BigQuery REGEXP_EXTRACT supports (subject, pattern, pos, occ) max 4 args
1298                let args = if f.args.len() > 4 {
1299                    f.args[..4].to_vec()
1300                } else {
1301                    f.args
1302                };
1303                Ok(Expression::Function(Box::new(Function::new(
1304                    "REGEXP_EXTRACT".to_string(),
1305                    args,
1306                ))))
1307            }
1308            "REGEXP_SUBSTR" => Ok(Expression::Function(Box::new(Function::new(
1309                "REGEXP_EXTRACT".to_string(),
1310                f.args,
1311            )))),
1312
1313            // REGEXP_REPLACE - strip extra Snowflake-specific args
1314            "REGEXP_REPLACE" if f.args.len() > 3 => {
1315                let args = f.args[..3].to_vec();
1316                Ok(Expression::Function(Box::new(Function::new(
1317                    "REGEXP_REPLACE".to_string(),
1318                    args,
1319                ))))
1320            }
1321
1322            // OBJECT_CONSTRUCT_KEEP_NULL -> JSON_OBJECT
1323            "OBJECT_CONSTRUCT_KEEP_NULL" => Ok(Expression::Function(Box::new(Function::new(
1324                "JSON_OBJECT".to_string(),
1325                f.args,
1326            )))),
1327
1328            // EDITDISTANCE -> EDIT_DISTANCE with named max_distance parameter
1329            "EDITDISTANCE" if f.args.len() == 3 => {
1330                let col1 = f.args[0].clone();
1331                let col2 = f.args[1].clone();
1332                let max_dist = f.args[2].clone();
1333                Ok(Expression::Function(Box::new(Function::new(
1334                    "EDIT_DISTANCE".to_string(),
1335                    vec![
1336                        col1,
1337                        col2,
1338                        Expression::NamedArgument(Box::new(crate::expressions::NamedArgument {
1339                            name: crate::expressions::Identifier::new("max_distance".to_string()),
1340                            value: max_dist,
1341                            separator: crate::expressions::NamedArgSeparator::DArrow,
1342                        })),
1343                    ],
1344                ))))
1345            }
1346            "EDITDISTANCE" if f.args.len() == 2 => Ok(Expression::Function(Box::new(
1347                Function::new("EDIT_DISTANCE".to_string(), f.args),
1348            ))),
1349
1350            // HEX_DECODE_BINARY -> FROM_HEX
1351            "HEX_DECODE_BINARY" => Ok(Expression::Function(Box::new(Function::new(
1352                "FROM_HEX".to_string(),
1353                f.args,
1354            )))),
1355
1356            // BigQuery format string normalization for PARSE_DATE/DATETIME/TIMESTAMP functions
1357            // %Y-%m-%d -> %F and %H:%M:%S -> %T
1358            "PARSE_DATE"
1359            | "PARSE_DATETIME"
1360            | "PARSE_TIMESTAMP"
1361            | "SAFE.PARSE_DATE"
1362            | "SAFE.PARSE_DATETIME"
1363            | "SAFE.PARSE_TIMESTAMP" => {
1364                let args = self.normalize_time_format_args(f.args);
1365                Ok(Expression::Function(Box::new(Function {
1366                    name: f.name,
1367                    args,
1368                    distinct: f.distinct,
1369                    no_parens: f.no_parens,
1370                    trailing_comments: f.trailing_comments,
1371                    quoted: f.quoted,
1372                    use_bracket_syntax: f.use_bracket_syntax,
1373                    span: None,
1374                    inferred_type: None,
1375                })))
1376            }
1377
1378            // GET_PATH(obj, path) -> JSON_EXTRACT(obj, json_path) in BigQuery
1379            "GET_PATH" if f.args.len() == 2 => {
1380                let mut args = f.args;
1381                let this = args.remove(0);
1382                let path = args.remove(0);
1383                let json_path = match &path {
1384                    Expression::Literal(lit) if matches!(lit.as_ref(), Literal::String(_)) => {
1385                        let Literal::String(s) = lit.as_ref() else {
1386                            unreachable!()
1387                        };
1388                        let normalized = if s.starts_with('$') {
1389                            s.clone()
1390                        } else if s.starts_with('[') {
1391                            format!("${}", s)
1392                        } else {
1393                            format!("$.{}", s)
1394                        };
1395                        Expression::Literal(Box::new(Literal::String(normalized)))
1396                    }
1397                    _ => path,
1398                };
1399                Ok(Expression::JsonExtract(Box::new(JsonExtractFunc {
1400                    this,
1401                    path: json_path,
1402                    returning: None,
1403                    arrow_syntax: false,
1404                    hash_arrow_syntax: false,
1405                    wrapper_option: None,
1406                    quotes_option: None,
1407                    on_scalar_string: false,
1408                    on_error: None,
1409                })))
1410            }
1411
1412            // Pass through everything else
1413            _ => Ok(Expression::Function(Box::new(f))),
1414        }
1415    }
1416
1417    fn transform_aggregate_function(
1418        &self,
1419        f: Box<crate::expressions::AggregateFunction>,
1420    ) -> Result<Expression> {
1421        let name_upper = f.name.to_uppercase();
1422        match name_upper.as_str() {
1423            // GROUP_CONCAT -> STRING_AGG
1424            "GROUP_CONCAT" if !f.args.is_empty() => Ok(Expression::Function(Box::new(
1425                Function::new("STRING_AGG".to_string(), f.args),
1426            ))),
1427
1428            // Pass through everything else
1429            _ => Ok(Expression::AggregateFunction(f)),
1430        }
1431    }
1432
1433    /// Transform MethodCall expressions
1434    /// Handles SAFE.PARSE_DATE, SAFE.PARSE_DATETIME, SAFE.PARSE_TIMESTAMP format normalization
1435    fn transform_method_call(&self, mc: crate::expressions::MethodCall) -> Result<Expression> {
1436        use crate::expressions::MethodCall;
1437
1438        // Check if this is SAFE.PARSE_DATE/DATETIME/TIMESTAMP
1439        if let Expression::Column(ref col) = mc.this {
1440            if col.name.name.eq_ignore_ascii_case("SAFE") {
1441                let method_upper = mc.method.name.to_uppercase();
1442                if method_upper == "PARSE_DATE"
1443                    || method_upper == "PARSE_DATETIME"
1444                    || method_upper == "PARSE_TIMESTAMP"
1445                {
1446                    // Normalize the format string in the first argument
1447                    let args = self.normalize_time_format_args(mc.args);
1448                    return Ok(Expression::MethodCall(Box::new(MethodCall {
1449                        this: mc.this,
1450                        method: mc.method,
1451                        args,
1452                    })));
1453                }
1454            }
1455        }
1456
1457        // Pass through all other method calls
1458        Ok(Expression::MethodCall(Box::new(mc)))
1459    }
1460
1461    /// Normalize time format strings in function arguments
1462    /// BigQuery normalizes: %Y-%m-%d -> %F, %H:%M:%S -> %T
1463    fn normalize_time_format_args(&self, args: Vec<Expression>) -> Vec<Expression> {
1464        args.into_iter()
1465            .enumerate()
1466            .map(|(i, arg)| {
1467                // Only transform the first argument (the format string)
1468                if i == 0 {
1469                    if let Expression::Literal(ref lit) = arg {
1470                        if let Literal::String(s) = lit.as_ref() {
1471                            let normalized = self.normalize_time_format(&s);
1472                            return Expression::Literal(Box::new(Literal::String(normalized)));
1473                        }
1474                    }
1475                }
1476                arg
1477            })
1478            .collect()
1479    }
1480
1481    /// Normalize a time format string according to BigQuery conventions
1482    /// %Y-%m-%d -> %F (ISO date)
1483    /// %H:%M:%S -> %T (time)
1484    fn normalize_time_format(&self, format: &str) -> String {
1485        format.replace("%Y-%m-%d", "%F").replace("%H:%M:%S", "%T")
1486    }
1487
1488    /// Convert BigQuery CAST FORMAT elements to strftime equivalents,
1489    /// then normalize BigQuery shorthand forms (%Y-%m-%d -> %F, %H:%M:%S -> %T)
1490    fn bq_cast_format_to_strftime(format_expr: &Expression) -> Expression {
1491        use crate::expressions::Literal;
1492        if let Expression::Literal(lit) = format_expr {
1493            if let Literal::String(s) = lit.as_ref() {
1494                let result = s
1495                    .replace("YYYYMMDD", "%Y%m%d")
1496                    .replace("YYYY", "%Y")
1497                    .replace("YY", "%y")
1498                    .replace("MONTH", "%B")
1499                    .replace("MON", "%b")
1500                    .replace("MM", "%m")
1501                    .replace("DD", "%d")
1502                    .replace("HH24", "%H")
1503                    .replace("HH12", "%I")
1504                    .replace("HH", "%I")
1505                    .replace("MI", "%M")
1506                    .replace("SSTZH", "%S%z")
1507                    .replace("SS", "%S")
1508                    .replace("TZH", "%z");
1509                // Normalize: %Y-%m-%d -> %F, %H:%M:%S -> %T
1510                let normalized = result.replace("%Y-%m-%d", "%F").replace("%H:%M:%S", "%T");
1511                return Expression::Literal(Box::new(Literal::String(normalized)));
1512            }
1513        }
1514        format_expr.clone()
1515    }
1516}
1517
1518#[cfg(test)]
1519mod tests {
1520    use super::*;
1521    use crate::dialects::Dialect;
1522    use crate::parse_one;
1523
1524    fn transpile_to_bigquery(sql: &str) -> String {
1525        let dialect = Dialect::get(DialectType::Generic);
1526        let result = dialect
1527            .transpile(sql, DialectType::BigQuery)
1528            .expect("Transpile failed");
1529        result[0].clone()
1530    }
1531
1532    #[test]
1533    fn test_ifnull_identity() {
1534        // Generic -> BigQuery: IFNULL is normalized to COALESCE (matching sqlglot behavior)
1535        let result = transpile_to_bigquery("SELECT IFNULL(a, b)");
1536        assert!(
1537            result.contains("COALESCE"),
1538            "Expected COALESCE, got: {}",
1539            result
1540        );
1541    }
1542
1543    #[test]
1544    fn test_nvl_to_ifnull() {
1545        // NVL is converted to IFNULL in BigQuery
1546        let result = transpile_to_bigquery("SELECT NVL(a, b)");
1547        assert!(
1548            result.contains("IFNULL"),
1549            "Expected IFNULL, got: {}",
1550            result
1551        );
1552    }
1553
1554    #[test]
1555    fn test_try_cast_to_safe_cast() {
1556        let result = transpile_to_bigquery("SELECT TRY_CAST(a AS INT)");
1557        assert!(
1558            result.contains("SAFE_CAST"),
1559            "Expected SAFE_CAST, got: {}",
1560            result
1561        );
1562    }
1563
1564    #[test]
1565    fn test_random_to_rand() {
1566        let result = transpile_to_bigquery("SELECT RANDOM()");
1567        assert!(result.contains("RAND"), "Expected RAND, got: {}", result);
1568    }
1569
1570    #[test]
1571    fn test_basic_select() {
1572        let result = transpile_to_bigquery("SELECT a, b FROM users WHERE id = 1");
1573        assert!(result.contains("SELECT"));
1574        assert!(result.contains("FROM users"));
1575    }
1576
1577    #[test]
1578    fn test_group_concat_to_string_agg() {
1579        let result = transpile_to_bigquery("SELECT GROUP_CONCAT(name)");
1580        assert!(
1581            result.contains("STRING_AGG"),
1582            "Expected STRING_AGG, got: {}",
1583            result
1584        );
1585    }
1586
1587    #[test]
1588    fn test_generate_series_to_generate_array() {
1589        let result = transpile_to_bigquery("SELECT GENERATE_SERIES(1, 10)");
1590        assert!(
1591            result.contains("GENERATE_ARRAY"),
1592            "Expected GENERATE_ARRAY, got: {}",
1593            result
1594        );
1595    }
1596
1597    #[test]
1598    fn test_backtick_identifiers() {
1599        // BigQuery uses backticks for identifiers
1600        let dialect = BigQueryDialect;
1601        let config = dialect.generator_config();
1602        assert_eq!(config.identifier_quote, '`');
1603    }
1604
1605    fn bigquery_identity(sql: &str, expected: &str) {
1606        let dialect = Dialect::get(DialectType::BigQuery);
1607        let ast = dialect.parse(sql).expect("Parse failed");
1608        let transformed = dialect.transform(ast[0].clone()).expect("Transform failed");
1609        let result = dialect.generate(&transformed).expect("Generate failed");
1610        assert_eq!(result, expected, "SQL: {}", sql);
1611    }
1612
1613    #[test]
1614    fn test_safe_namespace_parses_as_function() {
1615        let expr = parse_one(
1616            "SELECT SAFE.PARSE_JSON(data) AS json_data FROM t",
1617            DialectType::BigQuery,
1618        )
1619        .expect("parse");
1620
1621        let Expression::Select(select) = expr else {
1622            panic!("expected SELECT");
1623        };
1624        let Expression::Alias(alias) = &select.expressions[0] else {
1625            panic!("expected alias");
1626        };
1627        let Expression::Function(function) = &alias.this else {
1628            panic!("expected SAFE namespace call to parse as Function");
1629        };
1630
1631        assert_eq!(function.name, "SAFE.PARSE_JSON");
1632        assert_eq!(function.args.len(), 1);
1633    }
1634
1635    #[test]
1636    fn test_safe_namespace_identity() {
1637        bigquery_identity("SAFE.PARSE_JSON(data)", "SAFE.PARSE_JSON(data)");
1638        bigquery_identity(
1639            "SAFE.PARSE_DATE('%Y-%m-%d', date_col)",
1640            "SAFE.PARSE_DATE('%F', date_col)",
1641        );
1642        bigquery_identity("SAFE.DIVIDE(a, b)", "SAFE.DIVIDE(a, b)");
1643    }
1644
1645    #[test]
1646    fn test_cast_char_to_string() {
1647        bigquery_identity("CAST(x AS CHAR)", "CAST(x AS STRING)");
1648    }
1649
1650    #[test]
1651    fn test_cast_varchar_to_string() {
1652        bigquery_identity("CAST(x AS VARCHAR)", "CAST(x AS STRING)");
1653    }
1654
1655    #[test]
1656    fn test_cast_nchar_to_string() {
1657        bigquery_identity("CAST(x AS NCHAR)", "CAST(x AS STRING)");
1658    }
1659
1660    #[test]
1661    fn test_cast_nvarchar_to_string() {
1662        bigquery_identity("CAST(x AS NVARCHAR)", "CAST(x AS STRING)");
1663    }
1664
1665    #[test]
1666    fn test_cast_timestamptz_to_timestamp() {
1667        bigquery_identity("CAST(x AS TIMESTAMPTZ)", "CAST(x AS TIMESTAMP)");
1668    }
1669
1670    #[test]
1671    fn test_cast_record_to_struct() {
1672        bigquery_identity("CAST(x AS RECORD)", "CAST(x AS STRUCT)");
1673    }
1674
1675    #[test]
1676    fn test_json_literal_to_parse_json() {
1677        // JSON 'string' literal syntax should be converted to PARSE_JSON()
1678        bigquery_identity(
1679            "SELECT JSON '\"foo\"' AS json_data",
1680            "SELECT PARSE_JSON('\"foo\"') AS json_data",
1681        );
1682    }
1683
1684    #[test]
1685    fn test_grant_as_alias_not_quoted() {
1686        // GRANT is not a reserved keyword in BigQuery, should not be backtick-quoted
1687        bigquery_identity(
1688            "SELECT GRANT FROM (SELECT 'input' AS GRANT)",
1689            "SELECT GRANT FROM (SELECT 'input' AS GRANT)",
1690        );
1691    }
1692
1693    #[test]
1694    fn test_timestamp_literal_to_cast() {
1695        // TIMESTAMP 'value' literal should be converted to CAST('value' AS TIMESTAMP)
1696        bigquery_identity(
1697            "CREATE VIEW `d.v` OPTIONS (expiration_timestamp=TIMESTAMP '2020-01-02T04:05:06.007Z') AS SELECT 1 AS c",
1698            "CREATE VIEW `d.v` OPTIONS (expiration_timestamp=CAST('2020-01-02T04:05:06.007Z' AS TIMESTAMP)) AS SELECT 1 AS c"
1699        );
1700    }
1701
1702    #[test]
1703    fn test_date_literal_to_cast_in_extract() {
1704        // Issue 1: DATE literal should become CAST syntax in BigQuery
1705        bigquery_identity(
1706            "EXTRACT(WEEK(THURSDAY) FROM DATE '2013-12-25')",
1707            "EXTRACT(WEEK(THURSDAY) FROM CAST('2013-12-25' AS DATE))",
1708        );
1709    }
1710
1711    #[test]
1712    fn test_json_object_with_json_literals() {
1713        // Issue 2: JSON literals in JSON_OBJECT should use PARSE_JSON, not CAST AS JSON
1714        bigquery_identity(
1715            "SELECT JSON_OBJECT('a', JSON '10') AS json_data",
1716            "SELECT JSON_OBJECT('a', PARSE_JSON('10')) AS json_data",
1717        );
1718    }
1719
1720    // NOTE: MOD paren unwrapping is tested in the conformance tests (sqlglot_dialect_identity).
1721    // The unit test version was removed due to stack overflow in debug builds (deep recursion).
1722    // Test case: MOD((a + 1), b) -> MOD(a + 1, b)
1723
1724    #[test]
1725    fn test_safe_parse_date_format_normalization() {
1726        // SAFE.PARSE_DATE format string normalization: %Y-%m-%d -> %F
1727        bigquery_identity(
1728            "SAFE.PARSE_DATE('%Y-%m-%d', '2024-01-15')",
1729            "SAFE.PARSE_DATE('%F', '2024-01-15')",
1730        );
1731    }
1732
1733    #[test]
1734    fn test_safe_parse_datetime_format_normalization() {
1735        // SAFE.PARSE_DATETIME format string normalization: %Y-%m-%d %H:%M:%S -> %F %T
1736        bigquery_identity(
1737            "SAFE.PARSE_DATETIME('%Y-%m-%d %H:%M:%S', '2024-01-15 10:30:00')",
1738            "SAFE.PARSE_DATETIME('%F %T', '2024-01-15 10:30:00')",
1739        );
1740    }
1741
1742    #[test]
1743    fn test_safe_parse_timestamp_format_normalization() {
1744        // SAFE.PARSE_TIMESTAMP format string normalization: %Y-%m-%d %H:%M:%S -> %F %T
1745        bigquery_identity(
1746            "SAFE.PARSE_TIMESTAMP('%Y-%m-%d %H:%M:%S', '2024-01-15 10:30:00')",
1747            "SAFE.PARSE_TIMESTAMP('%F %T', '2024-01-15 10:30:00')",
1748        );
1749    }
1750
1751    #[test]
1752    fn test_datetime_literal_to_cast() {
1753        // DATETIME 'value' literal should be converted to CAST('value' AS DATETIME)
1754        bigquery_identity(
1755            "LAST_DAY(DATETIME '2008-11-10 15:30:00', WEEK(SUNDAY))",
1756            "LAST_DAY(CAST('2008-11-10 15:30:00' AS DATETIME), WEEK)",
1757        );
1758    }
1759
1760    #[test]
1761    fn test_last_day_week_modifier_stripped() {
1762        // WEEK(SUNDAY) should become WEEK in BigQuery LAST_DAY function
1763        bigquery_identity("LAST_DAY(col, WEEK(MONDAY))", "LAST_DAY(col, WEEK)");
1764    }
1765
1766    #[test]
1767    fn test_hash_line_comment_parses() {
1768        // Regression test for issue #38:
1769        // BigQuery should accept # as a single-line comment.
1770        let result = parse_one("SELECT 1 as a #hello world", DialectType::BigQuery);
1771        assert!(result.is_ok(), "Expected parse to succeed, got: {result:?}");
1772    }
1773}
polyglot_sql/dialects/bigquery.rs

polyglot_sql/dialects/
bigquery.rs
