sql/
statements.rs

1// Copyright 2023 Greptime Team
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7//     http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15pub mod admin;
16pub mod alter;
17pub mod comment;
18pub mod copy;
19pub mod create;
20pub mod cursor;
21pub mod delete;
22pub mod describe;
23pub mod drop;
24pub mod explain;
25pub mod insert;
26pub mod kill;
27mod option_map;
28pub mod query;
29pub mod set_variables;
30pub mod show;
31pub mod statement;
32pub mod tql;
33pub(crate) mod transform;
34pub mod truncate;
35
36use std::sync::Arc;
37
38use api::helper::ColumnDataTypeWrapper;
39use api::v1::SemanticType;
40use common_sql::default_constraint::parse_column_default_constraint;
41use common_time::timezone::Timezone;
42use datatypes::extension::json::{JsonExtensionType, JsonMetadata};
43use datatypes::json::JsonStructureSettings;
44use datatypes::prelude::ConcreteDataType;
45use datatypes::schema::{COMMENT_KEY, ColumnDefaultConstraint, ColumnSchema};
46use datatypes::types::json_type::JsonNativeType;
47use datatypes::types::{JsonFormat, JsonType, TimestampType};
48use datatypes::value::Value;
49use snafu::ResultExt;
50use sqlparser::ast::{ExactNumberInfo, Ident};
51
52use crate::ast::{
53    ColumnDef, ColumnOption, DataType as SqlDataType, ObjectNamePartExt, TimezoneInfo,
54    Value as SqlValue,
55};
56use crate::error::{
57    self, ConvertToGrpcDataTypeSnafu, ConvertValueSnafu, Result,
58    SerializeColumnDefaultConstraintSnafu, SetFulltextOptionSnafu, SetJsonStructureSettingsSnafu,
59    SetSkippingIndexOptionSnafu, SetVectorIndexOptionSnafu, SqlCommonSnafu,
60};
61use crate::statements::create::{Column, ColumnExtensions};
62pub use crate::statements::option_map::OptionMap;
63pub(crate) use crate::statements::transform::transform_statements;
64
65const VECTOR_TYPE_NAME: &str = "VECTOR";
66
67pub fn value_to_sql_value(val: &Value) -> Result<SqlValue> {
68    Ok(match val {
69        Value::Int8(v) => SqlValue::Number(v.to_string(), false),
70        Value::UInt8(v) => SqlValue::Number(v.to_string(), false),
71        Value::Int16(v) => SqlValue::Number(v.to_string(), false),
72        Value::UInt16(v) => SqlValue::Number(v.to_string(), false),
73        Value::Int32(v) => SqlValue::Number(v.to_string(), false),
74        Value::UInt32(v) => SqlValue::Number(v.to_string(), false),
75        Value::Int64(v) => SqlValue::Number(v.to_string(), false),
76        Value::UInt64(v) => SqlValue::Number(v.to_string(), false),
77        Value::Float32(v) => SqlValue::Number(v.to_string(), false),
78        Value::Float64(v) => SqlValue::Number(v.to_string(), false),
79        Value::Boolean(b) => SqlValue::Boolean(*b),
80        Value::Date(d) => SqlValue::SingleQuotedString(d.to_string()),
81        Value::Timestamp(ts) => SqlValue::SingleQuotedString(ts.to_iso8601_string()),
82        Value::String(s) => SqlValue::SingleQuotedString(s.as_utf8().to_string()),
83        Value::Null => SqlValue::Null,
84        // TODO(dennis): supports binary
85        _ => return ConvertValueSnafu { value: val.clone() }.fail(),
86    })
87}
88
89/// Return true when the `ColumnDef` options contain primary key
90pub fn has_primary_key_option(column_def: &ColumnDef) -> bool {
91    column_def
92        .options
93        .iter()
94        .any(|options| match options.option {
95            ColumnOption::Unique { is_primary, .. } => is_primary,
96            _ => false,
97        })
98}
99
100/// Create a `ColumnSchema` from `Column`.
101pub fn column_to_schema(
102    column: &Column,
103    time_index: &str,
104    timezone: Option<&Timezone>,
105) -> Result<ColumnSchema> {
106    let is_time_index = column.name().value == time_index;
107
108    let is_nullable = column
109        .options()
110        .iter()
111        .all(|o| !matches!(o.option, ColumnOption::NotNull))
112        && !is_time_index;
113
114    let name = column.name().value.clone();
115    let data_type = sql_data_type_to_concrete_data_type(column.data_type(), &column.extensions)?;
116    let default_constraint =
117        parse_column_default_constraint(&name, &data_type, column.options(), timezone)
118            .context(SqlCommonSnafu)?;
119
120    let mut column_schema = ColumnSchema::new(name, data_type, is_nullable)
121        .with_time_index(is_time_index)
122        .with_default_constraint(default_constraint)
123        .context(error::InvalidDefaultSnafu {
124            column: &column.name().value,
125        })?;
126
127    if let Some(ColumnOption::Comment(c)) = column.options().iter().find_map(|o| {
128        if matches!(o.option, ColumnOption::Comment(_)) {
129            Some(&o.option)
130        } else {
131            None
132        }
133    }) {
134        let _ = column_schema
135            .mut_metadata()
136            .insert(COMMENT_KEY.to_string(), c.clone());
137    }
138
139    if let Some(options) = column.extensions.build_fulltext_options()? {
140        column_schema = column_schema
141            .with_fulltext_options(options)
142            .context(SetFulltextOptionSnafu)?;
143    }
144
145    if let Some(options) = column.extensions.build_skipping_index_options()? {
146        column_schema = column_schema
147            .with_skipping_options(options)
148            .context(SetSkippingIndexOptionSnafu)?;
149    }
150
151    if let Some(options) = column.extensions.build_vector_index_options()? {
152        column_schema = column_schema
153            .with_vector_index_options(&options)
154            .context(SetVectorIndexOptionSnafu)?;
155    }
156
157    column_schema.set_inverted_index(column.extensions.inverted_index_options.is_some());
158
159    if matches!(column.data_type(), SqlDataType::JSON) {
160        let settings = column
161            .extensions
162            .build_json_structure_settings()?
163            .unwrap_or_default();
164        let extension = JsonExtensionType::new(Arc::new(JsonMetadata {
165            json_structure_settings: Some(settings.clone()),
166        }));
167        column_schema
168            .with_extension_type(&extension)
169            .with_context(|_| SetJsonStructureSettingsSnafu {
170                value: format!("{settings:?}"),
171            })?;
172    }
173
174    Ok(column_schema)
175}
176
177/// Convert `ColumnDef` in sqlparser to `ColumnDef` in gRPC proto.
178pub fn sql_column_def_to_grpc_column_def(
179    col: &ColumnDef,
180    timezone: Option<&Timezone>,
181) -> Result<api::v1::ColumnDef> {
182    let name = col.name.value.clone();
183    let data_type = sql_data_type_to_concrete_data_type(&col.data_type, &Default::default())?;
184
185    let is_nullable = col
186        .options
187        .iter()
188        .all(|o| !matches!(o.option, ColumnOption::NotNull));
189
190    let default_constraint =
191        parse_column_default_constraint(&name, &data_type, &col.options, timezone)
192            .context(SqlCommonSnafu)?
193            .map(ColumnDefaultConstraint::try_into) // serialize default constraint to bytes
194            .transpose()
195            .context(SerializeColumnDefaultConstraintSnafu)?;
196    // convert ConcreteDataType to grpc ColumnDataTypeWrapper
197    let (datatype, datatype_ext) = ColumnDataTypeWrapper::try_from(data_type.clone())
198        .context(ConvertToGrpcDataTypeSnafu)?
199        .to_parts();
200
201    let is_primary_key = col.options.iter().any(|o| {
202        matches!(
203            o.option,
204            ColumnOption::Unique {
205                is_primary: true,
206                ..
207            }
208        )
209    });
210
211    let semantic_type = if is_primary_key {
212        SemanticType::Tag
213    } else {
214        SemanticType::Field
215    };
216
217    Ok(api::v1::ColumnDef {
218        name,
219        data_type: datatype as i32,
220        is_nullable,
221        default_constraint: default_constraint.unwrap_or_default(),
222        semantic_type: semantic_type as _,
223        comment: String::new(),
224        datatype_extension: datatype_ext,
225        options: None,
226    })
227}
228
229pub fn sql_data_type_to_concrete_data_type(
230    data_type: &SqlDataType,
231    column_extensions: &ColumnExtensions,
232) -> Result<ConcreteDataType> {
233    match data_type {
234        SqlDataType::BigInt(_) | SqlDataType::Int64 => Ok(ConcreteDataType::int64_datatype()),
235        SqlDataType::BigIntUnsigned(_) => Ok(ConcreteDataType::uint64_datatype()),
236        SqlDataType::Int(_) | SqlDataType::Integer(_) => Ok(ConcreteDataType::int32_datatype()),
237        SqlDataType::IntUnsigned(_) | SqlDataType::UnsignedInteger => {
238            Ok(ConcreteDataType::uint32_datatype())
239        }
240        SqlDataType::SmallInt(_) => Ok(ConcreteDataType::int16_datatype()),
241        SqlDataType::SmallIntUnsigned(_) => Ok(ConcreteDataType::uint16_datatype()),
242        SqlDataType::TinyInt(_) | SqlDataType::Int8(_) => Ok(ConcreteDataType::int8_datatype()),
243        SqlDataType::TinyIntUnsigned(_) | SqlDataType::Int8Unsigned(_) => {
244            Ok(ConcreteDataType::uint8_datatype())
245        }
246        SqlDataType::Char(_)
247        | SqlDataType::Varchar(_)
248        | SqlDataType::Text
249        | SqlDataType::TinyText
250        | SqlDataType::MediumText
251        | SqlDataType::LongText
252        | SqlDataType::String(_) => Ok(ConcreteDataType::string_datatype()),
253        SqlDataType::Float(_) => Ok(ConcreteDataType::float32_datatype()),
254        SqlDataType::Double(_) | SqlDataType::Float64 => Ok(ConcreteDataType::float64_datatype()),
255        SqlDataType::Boolean => Ok(ConcreteDataType::boolean_datatype()),
256        SqlDataType::Date => Ok(ConcreteDataType::date_datatype()),
257        SqlDataType::Binary(_)
258        | SqlDataType::Blob(_)
259        | SqlDataType::Bytea
260        | SqlDataType::Varbinary(_) => Ok(ConcreteDataType::binary_datatype()),
261        SqlDataType::Datetime(_) => Ok(ConcreteDataType::timestamp_microsecond_datatype()),
262        SqlDataType::Timestamp(precision, _) => Ok(precision
263            .as_ref()
264            .map(|v| TimestampType::try_from(*v))
265            .transpose()
266            .map_err(|_| {
267                error::SqlTypeNotSupportedSnafu {
268                    t: data_type.clone(),
269                }
270                .build()
271            })?
272            .map(|t| ConcreteDataType::timestamp_datatype(t.unit()))
273            .unwrap_or(ConcreteDataType::timestamp_millisecond_datatype())),
274        SqlDataType::Interval { .. } => Ok(ConcreteDataType::interval_month_day_nano_datatype()),
275        SqlDataType::Decimal(exact_info) => match exact_info {
276            ExactNumberInfo::None => Ok(ConcreteDataType::decimal128_default_datatype()),
277            // refer to https://dev.mysql.com/doc/refman/8.0/en/fixed-point-types.html
278            // In standard SQL, the syntax DECIMAL(M) is equivalent to DECIMAL(M,0).
279            ExactNumberInfo::Precision(p) => Ok(ConcreteDataType::decimal128_datatype(*p as u8, 0)),
280            ExactNumberInfo::PrecisionAndScale(p, s) => {
281                Ok(ConcreteDataType::decimal128_datatype(*p as u8, *s as i8))
282            }
283        },
284        SqlDataType::JSON => {
285            let format = if let Some(x) = column_extensions.build_json_structure_settings()? {
286                if let Some(fields) = match x {
287                    JsonStructureSettings::Structured(fields) => fields,
288                    JsonStructureSettings::UnstructuredRaw => None,
289                    JsonStructureSettings::PartialUnstructuredByKey { fields, .. } => fields,
290                } {
291                    let datatype = &ConcreteDataType::Struct(fields);
292                    JsonFormat::Native(Box::new(datatype.into()))
293                } else {
294                    JsonFormat::Native(Box::new(JsonNativeType::Null))
295                }
296            } else {
297                JsonFormat::Jsonb
298            };
299            Ok(ConcreteDataType::Json(JsonType::new(format)))
300        }
301        // Vector type
302        SqlDataType::Custom(name, d)
303            if name.0.as_slice().len() == 1
304                && name.0.as_slice()[0]
305                    .to_string_unquoted()
306                    .to_ascii_uppercase()
307                    == VECTOR_TYPE_NAME
308                && d.len() == 1 =>
309        {
310            let dim = d[0].parse().map_err(|e| {
311                error::ParseSqlValueSnafu {
312                    msg: format!("Failed to parse vector dimension: {}", e),
313                }
314                .build()
315            })?;
316            Ok(ConcreteDataType::vector_datatype(dim))
317        }
318        _ => error::SqlTypeNotSupportedSnafu {
319            t: data_type.clone(),
320        }
321        .fail(),
322    }
323}
324
325pub fn concrete_data_type_to_sql_data_type(data_type: &ConcreteDataType) -> Result<SqlDataType> {
326    match data_type {
327        ConcreteDataType::Int64(_) => Ok(SqlDataType::BigInt(None)),
328        ConcreteDataType::UInt64(_) => Ok(SqlDataType::BigIntUnsigned(None)),
329        ConcreteDataType::Int32(_) => Ok(SqlDataType::Int(None)),
330        ConcreteDataType::UInt32(_) => Ok(SqlDataType::IntUnsigned(None)),
331        ConcreteDataType::Int16(_) => Ok(SqlDataType::SmallInt(None)),
332        ConcreteDataType::UInt16(_) => Ok(SqlDataType::SmallIntUnsigned(None)),
333        ConcreteDataType::Int8(_) => Ok(SqlDataType::TinyInt(None)),
334        ConcreteDataType::UInt8(_) => Ok(SqlDataType::TinyIntUnsigned(None)),
335        ConcreteDataType::String(_) => Ok(SqlDataType::String(None)),
336        ConcreteDataType::Float32(_) => Ok(SqlDataType::Float(ExactNumberInfo::None)),
337        ConcreteDataType::Float64(_) => Ok(SqlDataType::Double(ExactNumberInfo::None)),
338        ConcreteDataType::Boolean(_) => Ok(SqlDataType::Boolean),
339        ConcreteDataType::Date(_) => Ok(SqlDataType::Date),
340        ConcreteDataType::Timestamp(ts_type) => Ok(SqlDataType::Timestamp(
341            Some(ts_type.precision()),
342            TimezoneInfo::None,
343        )),
344        ConcreteDataType::Time(time_type) => Ok(SqlDataType::Time(
345            Some(time_type.precision()),
346            TimezoneInfo::None,
347        )),
348        ConcreteDataType::Interval(_) => Ok(SqlDataType::Interval {
349            fields: None,
350            precision: None,
351        }),
352        ConcreteDataType::Binary(_) => Ok(SqlDataType::Varbinary(None)),
353        ConcreteDataType::Decimal128(d) => Ok(SqlDataType::Decimal(
354            ExactNumberInfo::PrecisionAndScale(d.precision() as u64, d.scale() as i64),
355        )),
356        ConcreteDataType::Json(_) => Ok(SqlDataType::JSON),
357        ConcreteDataType::Vector(v) => Ok(SqlDataType::Custom(
358            vec![Ident::new(VECTOR_TYPE_NAME)].into(),
359            vec![v.dim.to_string()],
360        )),
361        ConcreteDataType::Duration(_)
362        | ConcreteDataType::Null(_)
363        | ConcreteDataType::List(_)
364        | ConcreteDataType::Struct(_)
365        | ConcreteDataType::Dictionary(_) => error::ConcreteTypeNotSupportedSnafu {
366            t: data_type.clone(),
367        }
368        .fail(),
369    }
370}
371
372#[cfg(test)]
373mod tests {
374    use api::v1::ColumnDataType;
375    use datatypes::schema::{
376        COLUMN_FULLTEXT_OPT_KEY_ANALYZER, COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, FulltextAnalyzer,
377    };
378    use sqlparser::ast::{ColumnOptionDef, Expr};
379
380    use super::*;
381    use crate::ast::TimezoneInfo;
382    use crate::statements::ColumnOption;
383    use crate::statements::create::ColumnExtensions;
384
385    fn check_type(sql_type: SqlDataType, data_type: ConcreteDataType) {
386        assert_eq!(
387            data_type,
388            sql_data_type_to_concrete_data_type(&sql_type, &Default::default()).unwrap()
389        );
390    }
391
392    #[test]
393    pub fn test_sql_data_type_to_concrete_data_type() {
394        check_type(
395            SqlDataType::BigInt(None),
396            ConcreteDataType::int64_datatype(),
397        );
398        check_type(SqlDataType::Int(None), ConcreteDataType::int32_datatype());
399        check_type(
400            SqlDataType::Integer(None),
401            ConcreteDataType::int32_datatype(),
402        );
403        check_type(
404            SqlDataType::SmallInt(None),
405            ConcreteDataType::int16_datatype(),
406        );
407        check_type(SqlDataType::Char(None), ConcreteDataType::string_datatype());
408        check_type(
409            SqlDataType::Varchar(None),
410            ConcreteDataType::string_datatype(),
411        );
412        check_type(SqlDataType::Text, ConcreteDataType::string_datatype());
413        check_type(
414            SqlDataType::String(None),
415            ConcreteDataType::string_datatype(),
416        );
417        check_type(
418            SqlDataType::Float(ExactNumberInfo::None),
419            ConcreteDataType::float32_datatype(),
420        );
421        check_type(
422            SqlDataType::Double(ExactNumberInfo::None),
423            ConcreteDataType::float64_datatype(),
424        );
425        check_type(SqlDataType::Boolean, ConcreteDataType::boolean_datatype());
426        check_type(SqlDataType::Date, ConcreteDataType::date_datatype());
427        check_type(
428            SqlDataType::Timestamp(None, TimezoneInfo::None),
429            ConcreteDataType::timestamp_millisecond_datatype(),
430        );
431        check_type(
432            SqlDataType::Varbinary(None),
433            ConcreteDataType::binary_datatype(),
434        );
435        check_type(
436            SqlDataType::BigIntUnsigned(None),
437            ConcreteDataType::uint64_datatype(),
438        );
439        check_type(
440            SqlDataType::IntUnsigned(None),
441            ConcreteDataType::uint32_datatype(),
442        );
443        check_type(
444            SqlDataType::SmallIntUnsigned(None),
445            ConcreteDataType::uint16_datatype(),
446        );
447        check_type(
448            SqlDataType::TinyIntUnsigned(None),
449            ConcreteDataType::uint8_datatype(),
450        );
451        check_type(
452            SqlDataType::Datetime(None),
453            ConcreteDataType::timestamp_microsecond_datatype(),
454        );
455        check_type(
456            SqlDataType::Interval {
457                fields: None,
458                precision: None,
459            },
460            ConcreteDataType::interval_month_day_nano_datatype(),
461        );
462        check_type(SqlDataType::JSON, ConcreteDataType::json_datatype());
463        check_type(
464            SqlDataType::Custom(
465                vec![Ident::new(VECTOR_TYPE_NAME)].into(),
466                vec!["3".to_string()],
467            ),
468            ConcreteDataType::vector_datatype(3),
469        );
470    }
471
472    #[test]
473    pub fn test_sql_column_def_to_grpc_column_def() {
474        // test basic
475        let column_def = ColumnDef {
476            name: "col".into(),
477            data_type: SqlDataType::Double(ExactNumberInfo::None),
478            options: vec![],
479        };
480
481        let grpc_column_def = sql_column_def_to_grpc_column_def(&column_def, None).unwrap();
482
483        assert_eq!("col", grpc_column_def.name);
484        assert!(grpc_column_def.is_nullable); // nullable when options are empty
485        assert_eq!(ColumnDataType::Float64 as i32, grpc_column_def.data_type);
486        assert!(grpc_column_def.default_constraint.is_empty());
487        assert_eq!(grpc_column_def.semantic_type, SemanticType::Field as i32);
488
489        // test not null
490        let column_def = ColumnDef {
491            name: "col".into(),
492            data_type: SqlDataType::Double(ExactNumberInfo::None),
493            options: vec![ColumnOptionDef {
494                name: None,
495                option: ColumnOption::NotNull,
496            }],
497        };
498
499        let grpc_column_def = sql_column_def_to_grpc_column_def(&column_def, None).unwrap();
500        assert!(!grpc_column_def.is_nullable);
501
502        // test primary key
503        let column_def = ColumnDef {
504            name: "col".into(),
505            data_type: SqlDataType::Double(ExactNumberInfo::None),
506            options: vec![ColumnOptionDef {
507                name: None,
508                option: ColumnOption::Unique {
509                    is_primary: true,
510                    characteristics: None,
511                },
512            }],
513        };
514
515        let grpc_column_def = sql_column_def_to_grpc_column_def(&column_def, None).unwrap();
516        assert_eq!(grpc_column_def.semantic_type, SemanticType::Tag as i32);
517    }
518
519    #[test]
520    pub fn test_sql_column_def_to_grpc_column_def_with_timezone() {
521        let column_def = ColumnDef {
522            name: "col".into(),
523            // MILLISECOND
524            data_type: SqlDataType::Timestamp(Some(3), TimezoneInfo::None),
525            options: vec![ColumnOptionDef {
526                name: None,
527                option: ColumnOption::Default(Expr::Value(
528                    SqlValue::SingleQuotedString("2024-01-30T00:01:01".to_string()).into(),
529                )),
530            }],
531        };
532
533        // with timezone "Asia/Shanghai"
534        let grpc_column_def = sql_column_def_to_grpc_column_def(
535            &column_def,
536            Some(&Timezone::from_tz_string("Asia/Shanghai").unwrap()),
537        )
538        .unwrap();
539        assert_eq!("col", grpc_column_def.name);
540        assert!(grpc_column_def.is_nullable); // nullable when options are empty
541        assert_eq!(
542            ColumnDataType::TimestampMillisecond as i32,
543            grpc_column_def.data_type
544        );
545        assert!(!grpc_column_def.default_constraint.is_empty());
546
547        let constraint =
548            ColumnDefaultConstraint::try_from(&grpc_column_def.default_constraint[..]).unwrap();
549        assert!(
550            matches!(constraint, ColumnDefaultConstraint::Value(Value::Timestamp(ts))
551                         if ts.to_iso8601_string() == "2024-01-29 16:01:01+0000")
552        );
553
554        // without timezone
555        let grpc_column_def = sql_column_def_to_grpc_column_def(&column_def, None).unwrap();
556        assert_eq!("col", grpc_column_def.name);
557        assert!(grpc_column_def.is_nullable); // nullable when options are empty
558        assert_eq!(
559            ColumnDataType::TimestampMillisecond as i32,
560            grpc_column_def.data_type
561        );
562        assert!(!grpc_column_def.default_constraint.is_empty());
563
564        let constraint =
565            ColumnDefaultConstraint::try_from(&grpc_column_def.default_constraint[..]).unwrap();
566        assert!(
567            matches!(constraint, ColumnDefaultConstraint::Value(Value::Timestamp(ts))
568                         if ts.to_iso8601_string() == "2024-01-30 00:01:01+0000")
569        );
570    }
571
572    #[test]
573    pub fn test_has_primary_key_option() {
574        let column_def = ColumnDef {
575            name: "col".into(),
576            data_type: SqlDataType::Double(ExactNumberInfo::None),
577            options: vec![],
578        };
579        assert!(!has_primary_key_option(&column_def));
580
581        let column_def = ColumnDef {
582            name: "col".into(),
583            data_type: SqlDataType::Double(ExactNumberInfo::None),
584            options: vec![ColumnOptionDef {
585                name: None,
586                option: ColumnOption::Unique {
587                    is_primary: true,
588                    characteristics: None,
589                },
590            }],
591        };
592        assert!(has_primary_key_option(&column_def));
593    }
594
595    #[test]
596    pub fn test_column_to_schema() {
597        let column_def = Column {
598            column_def: ColumnDef {
599                name: "col".into(),
600                data_type: SqlDataType::Double(ExactNumberInfo::None),
601                options: vec![],
602            },
603            extensions: ColumnExtensions::default(),
604        };
605
606        let column_schema = column_to_schema(&column_def, "ts", None).unwrap();
607
608        assert_eq!("col", column_schema.name);
609        assert_eq!(
610            ConcreteDataType::float64_datatype(),
611            column_schema.data_type
612        );
613        assert!(column_schema.is_nullable());
614        assert!(!column_schema.is_time_index());
615
616        let column_schema = column_to_schema(&column_def, "col", None).unwrap();
617
618        assert_eq!("col", column_schema.name);
619        assert_eq!(
620            ConcreteDataType::float64_datatype(),
621            column_schema.data_type
622        );
623        assert!(!column_schema.is_nullable());
624        assert!(column_schema.is_time_index());
625
626        let column_def = Column {
627            column_def: ColumnDef {
628                name: "col2".into(),
629                data_type: SqlDataType::String(None),
630                options: vec![
631                    ColumnOptionDef {
632                        name: None,
633                        option: ColumnOption::NotNull,
634                    },
635                    ColumnOptionDef {
636                        name: None,
637                        option: ColumnOption::Comment("test comment".to_string()),
638                    },
639                ],
640            },
641            extensions: ColumnExtensions::default(),
642        };
643
644        let column_schema = column_to_schema(&column_def, "ts", None).unwrap();
645
646        assert_eq!("col2", column_schema.name);
647        assert_eq!(ConcreteDataType::string_datatype(), column_schema.data_type);
648        assert!(!column_schema.is_nullable());
649        assert!(!column_schema.is_time_index());
650        assert_eq!(
651            column_schema.metadata().get(COMMENT_KEY),
652            Some(&"test comment".to_string())
653        );
654    }
655
656    #[test]
657    pub fn test_column_to_schema_timestamp_with_timezone() {
658        let column = Column {
659            column_def: ColumnDef {
660                name: "col".into(),
661                // MILLISECOND
662                data_type: SqlDataType::Timestamp(Some(3), TimezoneInfo::None),
663                options: vec![ColumnOptionDef {
664                    name: None,
665                    option: ColumnOption::Default(Expr::Value(
666                        SqlValue::SingleQuotedString("2024-01-30T00:01:01".to_string()).into(),
667                    )),
668                }],
669            },
670            extensions: ColumnExtensions::default(),
671        };
672
673        // with timezone "Asia/Shanghai"
674
675        let column_schema = column_to_schema(
676            &column,
677            "ts",
678            Some(&Timezone::from_tz_string("Asia/Shanghai").unwrap()),
679        )
680        .unwrap();
681
682        assert_eq!("col", column_schema.name);
683        assert_eq!(
684            ConcreteDataType::timestamp_millisecond_datatype(),
685            column_schema.data_type
686        );
687        assert!(column_schema.is_nullable());
688
689        let constraint = column_schema.default_constraint().unwrap();
690        assert!(
691            matches!(constraint, ColumnDefaultConstraint::Value(Value::Timestamp(ts))
692                         if ts.to_iso8601_string() == "2024-01-29 16:01:01+0000")
693        );
694
695        // without timezone
696        let column_schema = column_to_schema(&column, "ts", None).unwrap();
697
698        assert_eq!("col", column_schema.name);
699        assert_eq!(
700            ConcreteDataType::timestamp_millisecond_datatype(),
701            column_schema.data_type
702        );
703        assert!(column_schema.is_nullable());
704
705        let constraint = column_schema.default_constraint().unwrap();
706        assert!(
707            matches!(constraint, ColumnDefaultConstraint::Value(Value::Timestamp(ts))
708                         if ts.to_iso8601_string() == "2024-01-30 00:01:01+0000")
709        );
710    }
711
712    #[test]
713    fn test_column_to_schema_with_fulltext() {
714        let column = Column {
715            column_def: ColumnDef {
716                name: "col".into(),
717                data_type: SqlDataType::Text,
718                options: vec![],
719            },
720            extensions: ColumnExtensions {
721                fulltext_index_options: Some(OptionMap::from([
722                    (
723                        COLUMN_FULLTEXT_OPT_KEY_ANALYZER.to_string(),
724                        "English".to_string(),
725                    ),
726                    (
727                        COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE.to_string(),
728                        "true".to_string(),
729                    ),
730                ])),
731                vector_options: None,
732                skipping_index_options: None,
733                inverted_index_options: None,
734                json_datatype_options: None,
735                vector_index_options: None,
736            },
737        };
738
739        let column_schema = column_to_schema(&column, "ts", None).unwrap();
740        assert_eq!("col", column_schema.name);
741        assert_eq!(ConcreteDataType::string_datatype(), column_schema.data_type);
742        let fulltext_options = column_schema.fulltext_options().unwrap().unwrap();
743        assert_eq!(fulltext_options.analyzer, FulltextAnalyzer::English);
744        assert!(fulltext_options.case_sensitive);
745    }
746
747    #[test]
748    fn test_column_to_schema_with_vector_index() {
749        use datatypes::schema::{VectorDistanceMetric, VectorIndexEngineType};
750
751        // Test with custom metric and parameters
752        let column = Column {
753            column_def: ColumnDef {
754                name: "embedding".into(),
755                data_type: SqlDataType::Custom(
756                    vec![Ident::new(VECTOR_TYPE_NAME)].into(),
757                    vec!["128".to_string()],
758                ),
759                options: vec![],
760            },
761            extensions: ColumnExtensions {
762                fulltext_index_options: None,
763                vector_options: None,
764                skipping_index_options: None,
765                inverted_index_options: None,
766                json_datatype_options: None,
767                vector_index_options: Some(OptionMap::from([
768                    ("metric".to_string(), "cosine".to_string()),
769                    ("connectivity".to_string(), "32".to_string()),
770                    ("expansion_add".to_string(), "200".to_string()),
771                    ("expansion_search".to_string(), "100".to_string()),
772                ])),
773            },
774        };
775
776        let column_schema = column_to_schema(&column, "ts", None).unwrap();
777        assert_eq!("embedding", column_schema.name);
778        assert!(column_schema.is_vector_indexed());
779
780        let vector_options = column_schema.vector_index_options().unwrap().unwrap();
781        assert_eq!(vector_options.engine, VectorIndexEngineType::Usearch);
782        assert_eq!(vector_options.metric, VectorDistanceMetric::Cosine);
783        assert_eq!(vector_options.connectivity, 32);
784        assert_eq!(vector_options.expansion_add, 200);
785        assert_eq!(vector_options.expansion_search, 100);
786    }
787
788    #[test]
789    fn test_column_to_schema_with_vector_index_defaults() {
790        use datatypes::schema::{VectorDistanceMetric, VectorIndexEngineType};
791
792        // Test with default values (empty options map)
793        let column = Column {
794            column_def: ColumnDef {
795                name: "vec".into(),
796                data_type: SqlDataType::Custom(
797                    vec![Ident::new(VECTOR_TYPE_NAME)].into(),
798                    vec!["64".to_string()],
799                ),
800                options: vec![],
801            },
802            extensions: ColumnExtensions {
803                fulltext_index_options: None,
804                vector_options: None,
805                skipping_index_options: None,
806                inverted_index_options: None,
807                json_datatype_options: None,
808                vector_index_options: Some(OptionMap::default()),
809            },
810        };
811
812        let column_schema = column_to_schema(&column, "ts", None).unwrap();
813        assert_eq!("vec", column_schema.name);
814        assert!(column_schema.is_vector_indexed());
815
816        let vector_options = column_schema.vector_index_options().unwrap().unwrap();
817        // Verify defaults
818        assert_eq!(vector_options.engine, VectorIndexEngineType::Usearch);
819        assert_eq!(vector_options.metric, VectorDistanceMetric::L2sq);
820        assert_eq!(vector_options.connectivity, 16);
821        assert_eq!(vector_options.expansion_add, 128);
822        assert_eq!(vector_options.expansion_search, 64);
823    }
824}