hawdb_sql/
ast.rs

1// Copyright 2026 Nowledge
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 use hawdb_expression::sql::{
16    Expr, ExprKind, SqlColumnRef, SqlComparisonOp, SqlExpression, SqlFunctionArgument,
17    SqlIndexColumn, SqlLikeEscape, SqlNullOrder, SqlOrderDirection, SqlOrderItem, SqlPredicate,
18    SqlSourceLocation, SqlSourceSpan, SqlValue,
19};
20
21use caseless::Caseless;
22use hawdb_core::{HawDBError, LogicalType, Result};
23
24#[derive(Debug, Clone, PartialEq, Eq)]
25pub enum SqlStatement {
26    Explain(SqlExplainStatement),
27    Select(SelectStatement),
28    Insert(InsertStatement),
29    Update(UpdateStatement),
30    Delete(DeleteStatement),
31    CreateTable(CreateTableStatement),
32    CreateIndex(CreateIndexStatement),
33    CreateMaterializedView(CreateMaterializedViewStatement),
34    DropMaterializedView(DropMaterializedViewStatement),
35    AlterTableAddColumn(AlterTableAddColumnStatement),
36    /// Project-scoped branch lifecycle commands. These are deliberately
37    /// distinct from relational DDL: their executor operates on the durable
38    /// branch catalog rather than a selected branch's schema.
39    Branch(BranchSqlStatement),
40}
41
42#[derive(Debug, Clone, PartialEq, Eq)]
43pub struct CreateMaterializedViewStatement {
44    pub name: SqlTableName,
45    pub if_not_exists: bool,
46    pub query: Box<SelectStatement>,
47}
48
49#[derive(Debug, Clone, PartialEq, Eq)]
50pub struct DropMaterializedViewStatement {
51    pub name: SqlTableName,
52    pub if_exists: bool,
53}
54
55/// A branch lifecycle command parsed by the PostgreSQL entrypoint.
56///
57/// Keeping this as an AST family makes branch operations visible to the query
58/// runtime and prevents a host integration from recognizing command strings
59/// itself. Lifecycle and selection commands dispatch outside ordinary implicit
60/// data transactions through the storage-owned admission/publication kernel.
61#[derive(Debug, Clone, PartialEq, Eq)]
62pub enum BranchSqlStatement {
63    ShowBranches(ShowBranchesStatement),
64    ShowBranch(ShowBranchStatement),
65    ShowCurrentBranch,
66    UseBranch(BranchSqlSelector),
67    CreateBranch(CreateBranchStatement),
68    DropBranch(DropBranchStatement),
69}
70
71impl BranchSqlStatement {
72    pub const fn changes_context_or_catalog(&self) -> bool {
73        matches!(
74            self,
75            Self::UseBranch(_) | Self::CreateBranch(_) | Self::DropBranch(_)
76        )
77    }
78
79    pub const fn command(&self) -> &'static str {
80        match self {
81            Self::ShowBranches(_) => "SHOW BRANCHES",
82            Self::ShowBranch(_) => "SHOW BRANCH",
83            Self::ShowCurrentBranch => "SHOW CURRENT BRANCH",
84            Self::UseBranch(_) => "USE BRANCH",
85            Self::CreateBranch(_) => "CREATE BRANCH",
86            Self::DropBranch(_) => "DROP BRANCH",
87        }
88    }
89}
90
91#[derive(Debug, Clone, PartialEq, Eq)]
92pub struct CreateBranchStatement {
93    pub name: Option<BranchSqlValue>,
94    pub source: BranchSqlSelector,
95    pub expected_source_revision: SqlBound,
96    pub request_key: BranchSqlValue,
97    pub owner: Option<BranchSqlValue>,
98}
99
100#[derive(Debug, Clone, PartialEq, Eq)]
101pub struct DropBranchStatement {
102    pub id: BranchSqlValue,
103    pub expected_metadata_revision: SqlBound,
104}
105
106/// A bounded page over the durable branch catalog.
107#[derive(Debug, Clone, PartialEq, Eq)]
108pub struct ShowBranchesStatement {
109    pub limit: SqlBound,
110    pub offset: Option<SqlBound>,
111}
112
113#[derive(Debug, Clone, PartialEq, Eq)]
114pub struct ShowBranchStatement {
115    pub selector: BranchSqlSelector,
116}
117
118/// A typed selector avoids treating UUID-looking names as identifiers. The
119/// runtime validates `Id` values as UUIDs only after parameter binding.
120#[derive(Debug, Clone, PartialEq, Eq)]
121pub enum BranchSqlSelector {
122    Name(BranchSqlValue),
123    Id(BranchSqlValue),
124}
125
126/// A branch name, UUID, or opaque selector parameter. Branch commands accept
127/// only string literals or PostgreSQL positional parameters, never SQL
128/// expression evaluation or interpolated text.
129#[derive(Debug, Clone, PartialEq, Eq)]
130pub enum BranchSqlValue {
131    Literal(String),
132    Parameter(usize),
133}
134
135#[derive(Debug, Clone, PartialEq, Eq)]
136pub struct SqlExplainStatement {
137    pub analyze: bool,
138    pub statement: Box<SqlStatement>,
139}
140
141#[derive(Debug, Clone, PartialEq, Eq)]
142pub struct SqlVectorIndexSearch {
143    pub index: SqlValue,
144    pub embedding: SqlValue,
145    pub mode: hawdb_core::VectorIndexQueryMode,
146    pub candidates: Option<SqlValue>,
147    pub max_work: Option<SqlValue>,
148    pub top_k: SqlValue,
149}
150
151#[derive(Debug, Clone, PartialEq, Eq)]
152pub struct SelectStatement {
153    pub vector_index: Option<Box<SqlVectorIndexSearch>>,
154    pub projection: Vec<SelectProjection>,
155    pub distinct: bool,
156    /// `None` for a FROM-less SELECT (e.g. `SELECT version();`): the
157    /// projection is evaluated exactly once against an implicit row with no
158    /// column bindings, so any column reference in it is rejected the same
159    /// way an unresolvable column against a real table would be.
160    pub from: Option<SqlTableName>,
161    pub from_alias: Option<String>,
162    pub joins: Vec<SqlJoin>,
163    pub selection: Option<SqlPredicate>,
164    pub group_by: Vec<SqlColumnRef>,
165    pub having: Option<SqlPredicate>,
166    pub order_by: Vec<SqlOrderItem>,
167    pub limit: Option<SqlBound>,
168    pub offset: Option<SqlBound>,
169    pub lock_strength: Option<SqlLockStrength>,
170}
171
172impl SelectStatement {
173    /// The base table, for every code path that only ever runs on a real
174    /// table (catalog resolution, access-path selection, join planning, row
175    /// scanning). FROM-less selects never reach these paths — they're
176    /// handled by a separate, earlier branch — so `None` here is a caller
177    /// bug, not a user-facing state.
178    pub fn from_table(&self) -> &SqlTableName {
179        self.from
180            .as_ref()
181            .expect("from_table called on a FROM-less SELECT; it must take the fromless path")
182    }
183}
184
185#[derive(Debug, Clone, Copy, PartialEq, Eq)]
186pub enum SqlLockStrength {
187    Share,
188    Update,
189}
190
191#[derive(Debug, Clone, PartialEq, Eq)]
192pub struct InsertStatement {
193    pub table: SqlTableName,
194    pub columns: Vec<String>,
195    pub rows: Vec<Vec<SqlValue>>,
196    pub on_conflict: Option<SqlOnConflict>,
197    pub returning: Vec<SqlColumnRef>,
198}
199
200#[derive(Debug, Clone, PartialEq, Eq)]
201pub struct UpdateStatement {
202    pub table: SqlTableName,
203    pub alias: Option<String>,
204    pub assignments: Vec<SqlAssignment>,
205    pub selection: Option<SqlPredicate>,
206}
207
208#[derive(Debug, Clone, PartialEq, Eq)]
209pub struct DeleteStatement {
210    pub table: SqlTableName,
211    pub alias: Option<String>,
212    pub selection: Option<SqlPredicate>,
213}
214
215#[derive(Debug, Clone, PartialEq, Eq)]
216pub struct SqlAssignment {
217    pub column: String,
218    pub value: SqlAssignmentValue,
219}
220
221#[derive(Debug, Clone, PartialEq, Eq)]
222pub enum SqlAssignmentValue {
223    Value(SqlValue),
224    Column(SqlColumnRef),
225    Arithmetic {
226        left: SqlArithmeticOperand,
227        operator: SqlArithmeticOperator,
228        right: SqlArithmeticOperand,
229    },
230}
231
232#[derive(Debug, Clone, PartialEq, Eq)]
233pub enum SqlArithmeticOperand {
234    Value(SqlValue),
235    Column(SqlColumnRef),
236}
237
238#[derive(Debug, Clone, Copy, PartialEq, Eq)]
239pub enum SqlArithmeticOperator {
240    Add,
241    Subtract,
242}
243
244#[derive(Debug, Clone, PartialEq, Eq)]
245pub struct SqlOnConflict {
246    pub columns: Vec<String>,
247    pub action: SqlConflictAction,
248}
249
250#[derive(Debug, Clone, PartialEq, Eq)]
251pub enum SqlConflictAction {
252    DoNothing,
253    DoUpdate(Vec<SqlAssignment>),
254}
255
256#[derive(Debug, Clone, PartialEq, Eq)]
257pub struct CreateTableStatement {
258    pub table: SqlTableName,
259    pub if_not_exists: bool,
260    pub columns: Vec<SqlColumnDefinition>,
261    pub constraints: Vec<SqlTableConstraint>,
262    pub storage: SqlTableStorage,
263}
264
265#[derive(Debug, Clone, PartialEq, Eq, Default)]
266pub enum SqlTableStorage {
267    #[default]
268    RowPage,
269    StrictAppend {
270        partition_key: Vec<String>,
271        order_key: Vec<String>,
272        generated_order: SqlGeneratedOrder,
273    },
274}
275
276#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
277pub enum SqlGeneratedOrder {
278    #[default]
279    CallerProvided,
280    CommitSequence,
281}
282
283#[derive(Debug, Clone, PartialEq, Eq)]
284pub struct SqlColumnDefinition {
285    pub name: String,
286    pub data_type: SqlDataType,
287    pub nullable: bool,
288    pub default: Option<SqlColumnDefault>,
289    pub primary_key: bool,
290    pub unique: bool,
291    pub references: Option<SqlForeignKeyReference>,
292}
293
294#[derive(Debug, Clone, PartialEq, Eq)]
295pub enum SqlColumnDefault {
296    Literal(SqlValue),
297    UuidV7,
298}
299
300#[derive(Debug, Clone, Copy, PartialEq, Eq)]
301pub enum SqlDataType {
302    Boolean,
303    BigInt,
304    DoublePrecision,
305    Text,
306    Bytea,
307    Uuid,
308    Vector(hawdb_core::VectorDimension),
309}
310
311impl SqlDataType {
312    pub const fn logical_type(self) -> LogicalType {
313        match self {
314            Self::Boolean => LogicalType::Boolean,
315            Self::BigInt => LogicalType::Int64,
316            Self::DoublePrecision => LogicalType::Float64,
317            Self::Text => LogicalType::Text,
318            Self::Bytea => LogicalType::Binary,
319            Self::Uuid => LogicalType::Uuid,
320            Self::Vector(dimension) => LogicalType::Vector(dimension),
321        }
322    }
323}
324
325#[derive(Debug, Clone, PartialEq, Eq)]
326pub enum SqlTableConstraint {
327    PrimaryKey(Vec<String>),
328    Unique(Vec<String>),
329    ForeignKey {
330        columns: Vec<String>,
331        reference: SqlForeignKeyReference,
332    },
333}
334
335#[derive(Debug, Clone, PartialEq, Eq)]
336pub struct SqlForeignKeyReference {
337    pub table: SqlTableName,
338    pub columns: Vec<String>,
339    pub on_delete: SqlReferentialAction,
340    pub on_update: SqlReferentialAction,
341}
342
343#[derive(Debug, Clone, Copy, PartialEq, Eq)]
344pub enum SqlReferentialAction {
345    NoAction,
346    Restrict,
347    Cascade,
348    SetNull,
349}
350
351#[derive(Debug, Clone, PartialEq, Eq)]
352pub struct CreateIndexStatement {
353    pub name: String,
354    pub table: SqlTableName,
355    pub columns: Vec<SqlIndexColumn>,
356    pub unique: bool,
357    pub if_not_exists: bool,
358}
359
360#[derive(Debug, Clone, PartialEq, Eq)]
361pub struct AlterTableAddColumnStatement {
362    pub table: SqlTableName,
363    pub if_not_exists: bool,
364    pub column: SqlColumnDefinition,
365}
366
367#[derive(Debug, Clone, Copy, PartialEq, Eq)]
368pub enum SqlBound {
369    Literal(u64),
370    Parameter(usize),
371}
372
373#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord)]
374pub struct SqlTableName {
375    pub schema: Option<String>,
376    pub name: String,
377}
378
379#[derive(Debug, Clone, PartialEq, Eq)]
380pub enum SelectProjection {
381    Wildcard,
382    Expression {
383        expression: SqlExpression,
384        alias: Option<String>,
385    },
386}
387
388#[derive(Debug, Clone, PartialEq, Eq)]
389pub struct SqlJoin {
390    pub kind: SqlJoinKind,
391    pub table: SqlTableName,
392    pub alias: Option<String>,
393    pub on: SqlPredicate,
394    /// First input visible to ON, with the SELECT base relation at ordinal zero.
395    /// The current join's right input is the inclusive end of this scope.
396    pub on_scope_start: usize,
397}
398
399#[derive(Debug, Clone, Copy, PartialEq, Eq)]
400pub enum SqlJoinKind {
401    Inner,
402    Left,
403}
404
405/// Matches SQL LIKE patterns with locale-independent Unicode default case folding for ILIKE.
406pub fn sql_like_matches(
407    value: &str,
408    pattern: &str,
409    escape: SqlLikeEscape,
410    case_insensitive: bool,
411) -> Result<bool> {
412    let layout = like_layout(pattern, escape, case_insensitive)?;
413    let value_chars = value.chars().count();
414    layout.scratch_bytes(value_chars)?;
415    let (tokens, literals) = tokenize_like_pattern(pattern, escape, case_insensitive, &layout)?;
416    let mut characters = Vec::with_capacity(value_chars);
417    characters.extend(value.chars());
418    let value = characters;
419
420    let mut value_index = 0;
421    let mut pattern_index = 0;
422    let mut wildcard_index = None;
423    let mut wildcard_value_index = 0;
424    while value_index < value.len() {
425        let matched = match tokens.get(pattern_index) {
426            Some(LikeToken::Literal(expected)) => {
427                if let Some(consumed) = match_like_literal(
428                    &value[value_index..],
429                    &literals[expected.clone()],
430                    case_insensitive,
431                ) {
432                    value_index += consumed;
433                    pattern_index += 1;
434                    true
435                } else {
436                    false
437                }
438            }
439            Some(LikeToken::One) => {
440                value_index += 1;
441                pattern_index += 1;
442                true
443            }
444            Some(LikeToken::Many) => {
445                wildcard_index = Some(pattern_index);
446                wildcard_value_index = value_index;
447                pattern_index += 1;
448                true
449            }
450            None => false,
451        };
452        if matched {
453            continue;
454        }
455        let Some(wildcard_index) = wildcard_index else {
456            return Ok(false);
457        };
458        wildcard_value_index += 1;
459        value_index = wildcard_value_index;
460        pattern_index = wildcard_index + 1;
461    }
462    while matches!(tokens.get(pattern_index), Some(LikeToken::Many)) {
463        pattern_index += 1;
464    }
465    Ok(pattern_index == tokens.len())
466}
467
468/// Requested heap bytes used by the LIKE matcher, before allocating its buffers.
469/// Escape parsing and Unicode folding share the matcher's exact layout pass.
470#[doc(hidden)]
471pub fn sql_like_scratch_bytes(
472    value: &str,
473    pattern: &str,
474    escape: SqlLikeEscape,
475    case_insensitive: bool,
476) -> Result<usize> {
477    like_layout(pattern, escape, case_insensitive)?.scratch_bytes(value.chars().count())
478}
479
480struct LikeLayout {
481    tokens: usize,
482    literal_bytes: usize,
483}
484
485impl LikeLayout {
486    fn scratch_bytes(&self, value_chars: usize) -> Result<usize> {
487        self.tokens
488            .checked_mul(std::mem::size_of::<LikeToken>())
489            .and_then(|bytes| bytes.checked_add(self.literal_bytes))
490            .and_then(|bytes| {
491                value_chars
492                    .checked_mul(std::mem::size_of::<char>())
493                    .and_then(|value| bytes.checked_add(value))
494            })
495            .ok_or_else(like_size_overflow)
496    }
497}
498
499#[derive(Debug, Clone)]
500enum LikeToken {
501    Literal(std::ops::Range<usize>),
502    One,
503    Many,
504}
505
506#[derive(Clone, Copy)]
507enum LikePart {
508    Literal(char),
509    One,
510    Many,
511}
512
513fn like_size_overflow() -> HawDBError {
514    HawDBError::Execution("LIKE scratch length overflow".into())
515}
516
517fn visit_like_pattern(
518    pattern: &str,
519    escape: SqlLikeEscape,
520    mut visit: impl FnMut(LikePart) -> Result<()>,
521) -> Result<()> {
522    let escape = match escape {
523        SqlLikeEscape::Character(character) => Some(character),
524        SqlLikeEscape::Disabled => None,
525    };
526    let mut characters = pattern.chars();
527    while let Some(character) = characters.next() {
528        let part = if Some(character) == escape {
529            LikePart::Literal(characters.next().ok_or_else(|| {
530                HawDBError::Semantic("LIKE pattern ends with its escape character".to_string())
531            })?)
532        } else {
533            match character {
534                '%' => LikePart::Many,
535                '_' => LikePart::One,
536                literal => LikePart::Literal(literal),
537            }
538        };
539        visit(part)?;
540    }
541    Ok(())
542}
543
544fn visit_like_literal(
545    literal: char,
546    case_insensitive: bool,
547    mut visit: impl FnMut(char) -> Result<()>,
548) -> Result<()> {
549    if case_insensitive {
550        for character in std::iter::once(literal).default_case_fold() {
551            visit(character)?;
552        }
553        Ok(())
554    } else {
555        visit(literal)
556    }
557}
558
559fn like_layout(pattern: &str, escape: SqlLikeEscape, case_insensitive: bool) -> Result<LikeLayout> {
560    let mut layout = LikeLayout {
561        tokens: 0,
562        literal_bytes: 0,
563    };
564    let mut previous = None;
565    visit_like_pattern(pattern, escape, |part| {
566        let merged = matches!(
567            (previous, part),
568            (Some(LikePart::Many), LikePart::Many)
569                | (Some(LikePart::Literal(_)), LikePart::Literal(_))
570        );
571        if !merged {
572            layout.tokens = layout
573                .tokens
574                .checked_add(1)
575                .ok_or_else(like_size_overflow)?;
576        }
577        if let LikePart::Literal(literal) = part {
578            visit_like_literal(literal, case_insensitive, |character| {
579                layout.literal_bytes = layout
580                    .literal_bytes
581                    .checked_add(character.len_utf8())
582                    .ok_or_else(like_size_overflow)?;
583                Ok(())
584            })?;
585        }
586        previous = Some(part);
587        Ok(())
588    })?;
589    Ok(layout)
590}
591
592fn tokenize_like_pattern(
593    pattern: &str,
594    escape: SqlLikeEscape,
595    case_insensitive: bool,
596    layout: &LikeLayout,
597) -> Result<(Vec<LikeToken>, String)> {
598    let mut tokens = Vec::with_capacity(layout.tokens);
599    let mut literals = String::with_capacity(layout.literal_bytes);
600    visit_like_pattern(pattern, escape, |part| {
601        match part {
602            LikePart::Literal(literal) => {
603                let start = literals.len();
604                visit_like_literal(literal, case_insensitive, |character| {
605                    literals.push(character);
606                    Ok(())
607                })?;
608                if let Some(LikeToken::Literal(range)) = tokens.last_mut() {
609                    range.end = literals.len();
610                } else {
611                    tokens.push(LikeToken::Literal(start..literals.len()));
612                }
613            }
614            LikePart::Many if !matches!(tokens.last(), Some(LikeToken::Many)) => {
615                tokens.push(LikeToken::Many)
616            }
617            LikePart::Many => {}
618            LikePart::One => tokens.push(LikeToken::One),
619        }
620        Ok(())
621    })?;
622    debug_assert_eq!(tokens.len(), layout.tokens);
623    debug_assert_eq!(literals.len(), layout.literal_bytes);
624    Ok((tokens, literals))
625}
626
627fn match_like_literal(value: &[char], expected: &str, case_insensitive: bool) -> Option<usize> {
628    let mut expected = expected.chars().peekable();
629    for (index, character) in value.iter().copied().enumerate() {
630        if case_insensitive {
631            for folded in std::iter::once(character).default_case_fold() {
632                if expected.next() != Some(folded) {
633                    return None;
634                }
635            }
636        } else if expected.next() != Some(character) {
637            return None;
638        }
639        // Wildcards and literal boundaries consume complete source characters,
640        // even when one character folds into several code points.
641        if expected.peek().is_none() {
642            return Some(index + 1);
643        }
644    }
645    None
646}
647
648#[cfg(test)]
649mod tests {
650    use super::*;
651
652    #[test]
653    fn postgres_data_types_use_shared_logical_types() {
654        assert_eq!(SqlDataType::Boolean.logical_type(), LogicalType::Boolean);
655        assert_eq!(SqlDataType::BigInt.logical_type(), LogicalType::Int64);
656        assert_eq!(
657            SqlDataType::DoublePrecision.logical_type(),
658            LogicalType::Float64
659        );
660        assert_eq!(SqlDataType::Text.logical_type(), LogicalType::Text);
661        assert_eq!(SqlDataType::Bytea.logical_type(), LogicalType::Binary);
662    }
663}
hawdb_sql/ast.rs

hawdb_sql/
ast.rs
