Skip to main content

hawdb_sql/pgq/
binder.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
15use std::collections::{BTreeMap, BTreeSet};
16use std::fmt;
17
18use hawdb_sql_syntax::{
19    BinaryOperatorSyntax, ExpressionKindSyntax, ExpressionSyntax, GraphElementPatternSyntax,
20    GraphPathPrimarySyntax, GraphPathSyntax, GraphTable, Identifier, LiteralSyntax,
21    PostgresFromItemSyntax, PostgresSelectSyntax, QualifiedName, Span, TableAlias,
22};
23
24use super::{
25    BoundPgqColumn, BoundPgqExpression, BoundPgqExpressionKind, BoundPgqGraphPattern,
26    BoundPgqGraphTable, BoundPgqLiteral, BoundPgqPath, BoundPgqPathFactor, BoundPgqPathPrimary,
27    BoundPgqVariable, PgqBindingContext, PgqCatalog, PgqDataType, PgqElementKind, PgqSlotId,
28    PropertyGraphElementSchema, PropertyGraphSchema,
29};
30
31#[derive(Debug, Clone, Copy, PartialEq, Eq)]
32pub enum PgqBindErrorCode {
33    UnknownGraph,
34    UnknownLabel,
35    DuplicateVariable,
36    ConflictingVariable,
37    InvalidPath,
38    UnknownReference,
39    UnknownProperty,
40    MissingColumnName,
41    DuplicateColumn,
42    ColumnAliasCountMismatch,
43    UnsupportedExpression,
44    InvalidLiteral,
45    TypeMismatch,
46}
47
48#[derive(Debug, Clone, PartialEq, Eq)]
49pub struct PgqBindError {
50    pub code: PgqBindErrorCode,
51    pub span: Span,
52    pub message: String,
53}
54
55impl PgqBindError {
56    fn new(code: PgqBindErrorCode, span: Span, message: impl Into<String>) -> Self {
57        Self {
58            code,
59            span,
60            message: message.into(),
61        }
62    }
63}
64
65impl fmt::Display for PgqBindError {
66    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
67        write!(
68            formatter,
69            "SQL/PGQ binding error {:?} at bytes {}..{}: {}",
70            self.code, self.span.start, self.span.end, self.message
71        )
72    }
73}
74
75impl std::error::Error for PgqBindError {}
76
77pub fn bind_postgres_graph_tables(
78    input: &str,
79    select: &PostgresSelectSyntax,
80    catalog: &dyn PgqCatalog,
81    context: &PgqBindingContext,
82) -> Result<Vec<BoundPgqGraphTable>, PgqBindError> {
83    let mut bound = Vec::new();
84    for source in &select.from {
85        bind_from_item(input, &source.relation, catalog, context, &mut bound)?;
86        for join in &source.joins {
87            bind_from_item(input, &join.relation, catalog, context, &mut bound)?;
88        }
89    }
90    Ok(bound)
91}
92
93fn bind_from_item(
94    input: &str,
95    item: &PostgresFromItemSyntax,
96    catalog: &dyn PgqCatalog,
97    context: &PgqBindingContext,
98    bound: &mut Vec<BoundPgqGraphTable>,
99) -> Result<(), PgqBindError> {
100    let PostgresFromItemSyntax::GraphTable(table) = item else {
101        return Ok(());
102    };
103    let graph_name = normalize_qualified_name(input, &table.graph);
104    let graph = catalog.property_graph(&graph_name).ok_or_else(|| {
105        PgqBindError::new(
106            PgqBindErrorCode::UnknownGraph,
107            table.graph.span,
108            format!("property graph {} does not exist", graph_name.join(".")),
109        )
110    })?;
111    bound.push(GraphTableBinder::new(input, graph, context).bind(table, graph_name)?);
112    Ok(())
113}
114
115struct GraphTableBinder<'a> {
116    input: &'a str,
117    graph: &'a PropertyGraphSchema,
118    context: &'a PgqBindingContext,
119    variables: Vec<BoundPgqVariable>,
120    named_slots: BTreeMap<String, PgqSlotId>,
121    slots_by_span: BTreeMap<(usize, usize), PgqSlotId>,
122}
123
124impl<'a> GraphTableBinder<'a> {
125    fn new(input: &'a str, graph: &'a PropertyGraphSchema, context: &'a PgqBindingContext) -> Self {
126        Self {
127            input,
128            graph,
129            context,
130            variables: Vec::new(),
131            named_slots: BTreeMap::new(),
132            slots_by_span: BTreeMap::new(),
133        }
134    }
135
136    fn bind(
137        mut self,
138        table: &GraphTable,
139        graph_name: Vec<String>,
140    ) -> Result<BoundPgqGraphTable, PgqBindError> {
141        if table.pattern.paths.len() != 1 {
142            return Err(PgqBindError::new(
143                PgqBindErrorCode::InvalidPath,
144                table.pattern.span,
145                "the qualified GRAPH_TABLE slice requires exactly one graph path",
146            ));
147        }
148        for path in &table.pattern.paths {
149            self.validate_path_shape(path)?;
150            self.register_path(path)?;
151        }
152
153        let paths = table
154            .pattern
155            .paths
156            .iter()
157            .map(|path| self.bind_path(path))
158            .collect::<Result<Vec<_>, _>>()?;
159        let predicate = table
160            .pattern
161            .predicate
162            .as_ref()
163            .map(|predicate| self.bind_predicate(predicate))
164            .transpose()?;
165        let mut columns = self.bind_columns(&table.columns)?;
166        if let Some(alias) = &table.alias {
167            self.apply_table_column_aliases(alias, &mut columns)?;
168        }
169        let alias = table
170            .alias
171            .as_ref()
172            .map(|alias| normalize_table_alias(self.input, alias));
173
174        Ok(BoundPgqGraphTable {
175            graph_name,
176            variables: self.variables,
177            pattern: BoundPgqGraphPattern { paths, predicate },
178            columns,
179            alias,
180            span: table.span,
181        })
182    }
183
184    fn validate_path_shape(&self, path: &GraphPathSyntax) -> Result<(), PgqBindError> {
185        if path.factors.len() == 1
186            && matches!(
187                path.factors[0].primary,
188                GraphPathPrimarySyntax::Parenthesized { .. }
189            )
190        {
191            return Ok(());
192        }
193        let mut expect_vertex = true;
194        for factor in &path.factors {
195            let is_vertex = matches!(factor.primary, GraphPathPrimarySyntax::Vertex(_));
196            let is_edge = matches!(factor.primary, GraphPathPrimarySyntax::Edge(_));
197            if !is_vertex && !is_edge {
198                return Err(PgqBindError::new(
199                    PgqBindErrorCode::InvalidPath,
200                    factor.span,
201                    "a parenthesized path must be a standalone path factor in this slice",
202                ));
203            }
204            if is_vertex != expect_vertex {
205                return Err(PgqBindError::new(
206                    PgqBindErrorCode::InvalidPath,
207                    factor.span,
208                    "graph paths must alternate vertex and edge patterns and end at a vertex",
209                ));
210            }
211            if factor.quantifier.is_some() && is_vertex {
212                return Err(PgqBindError::new(
213                    PgqBindErrorCode::InvalidPath,
214                    factor.span,
215                    "vertex pattern quantifiers are not supported",
216                ));
217            }
218            if let Some(quantifier) = &factor.quantifier
219                && quantifier.min > quantifier.max
220            {
221                return Err(PgqBindError::new(
222                    PgqBindErrorCode::InvalidPath,
223                    quantifier.span,
224                    "graph quantifier lower bound exceeds upper bound",
225                ));
226            }
227            expect_vertex = !expect_vertex;
228        }
229        if expect_vertex {
230            return Err(PgqBindError::new(
231                PgqBindErrorCode::InvalidPath,
232                path.span,
233                "graph paths must end at a vertex pattern",
234            ));
235        }
236        Ok(())
237    }
238
239    fn register_path(&mut self, path: &GraphPathSyntax) -> Result<(), PgqBindError> {
240        for factor in &path.factors {
241            match &factor.primary {
242                GraphPathPrimarySyntax::Vertex(vertex) => {
243                    self.register_element(vertex, PgqElementKind::Vertex)?;
244                }
245                GraphPathPrimarySyntax::Edge(edge) => {
246                    self.register_element_fields(
247                        edge.variable.as_ref(),
248                        &edge.labels,
249                        PgqElementKind::Edge,
250                        edge.span,
251                    )?;
252                }
253                GraphPathPrimarySyntax::Parenthesized { path, .. } => {
254                    self.validate_path_shape(path)?;
255                    self.register_path(path)?;
256                }
257            }
258        }
259        Ok(())
260    }
261
262    fn register_element(
263        &mut self,
264        element: &GraphElementPatternSyntax,
265        kind: PgqElementKind,
266    ) -> Result<PgqSlotId, PgqBindError> {
267        self.register_element_fields(
268            element.variable.as_ref(),
269            &element.labels,
270            kind,
271            element.span,
272        )
273    }
274
275    fn register_element_fields(
276        &mut self,
277        variable: Option<&Identifier>,
278        labels: &[Identifier],
279        kind: PgqElementKind,
280        span: Span,
281    ) -> Result<PgqSlotId, PgqBindError> {
282        let labels = labels
283            .iter()
284            .map(|label| normalize_identifier(self.input, label))
285            .collect::<Vec<_>>();
286        self.validate_labels(kind, &labels, span)?;
287        let name = variable.map(|variable| normalize_identifier(self.input, variable));
288        let slot = if let Some(name) = &name {
289            if let Some(slot) = self.named_slots.get(name).copied() {
290                let existing = &mut self.variables[slot.0 as usize];
291                if existing.kind != kind {
292                    return Err(PgqBindError::new(
293                        PgqBindErrorCode::ConflictingVariable,
294                        span,
295                        format!("graph variable {name} is used as both vertex and edge"),
296                    ));
297                }
298                if !labels.is_empty() {
299                    if existing.labels.is_empty() {
300                        existing.labels.clone_from(&labels);
301                    } else if existing.labels != labels {
302                        return Err(PgqBindError::new(
303                            PgqBindErrorCode::ConflictingVariable,
304                            span,
305                            format!("graph variable {name} has conflicting label predicates"),
306                        ));
307                    }
308                }
309                slot
310            } else {
311                let slot = self.next_slot(span)?;
312                self.named_slots.insert(name.clone(), slot);
313                self.variables.push(BoundPgqVariable {
314                    slot,
315                    name: Some(name.clone()),
316                    kind,
317                    labels,
318                    span,
319                });
320                slot
321            }
322        } else {
323            let slot = self.next_slot(span)?;
324            self.variables.push(BoundPgqVariable {
325                slot,
326                name: None,
327                kind,
328                labels,
329                span,
330            });
331            slot
332        };
333        self.slots_by_span.insert((span.start, span.end), slot);
334        Ok(slot)
335    }
336
337    fn next_slot(&self, span: Span) -> Result<PgqSlotId, PgqBindError> {
338        u32::try_from(self.variables.len())
339            .map(PgqSlotId)
340            .map_err(|_| {
341                PgqBindError::new(
342                    PgqBindErrorCode::DuplicateVariable,
343                    span,
344                    "too many graph variables",
345                )
346            })
347    }
348
349    fn validate_labels(
350        &self,
351        kind: PgqElementKind,
352        labels: &[String],
353        span: Span,
354    ) -> Result<(), PgqBindError> {
355        let schemas = self.element_schemas(kind);
356        for label in labels {
357            if !schemas.contains_key(label) {
358                return Err(PgqBindError::new(
359                    PgqBindErrorCode::UnknownLabel,
360                    span,
361                    format!("unknown {} label {label}", element_kind_name(kind)),
362                ));
363            }
364        }
365        Ok(())
366    }
367
368    fn bind_path(&self, path: &GraphPathSyntax) -> Result<BoundPgqPath, PgqBindError> {
369        let factors = path
370            .factors
371            .iter()
372            .map(|factor| {
373                let primary = match &factor.primary {
374                    GraphPathPrimarySyntax::Vertex(vertex) => BoundPgqPathPrimary::Vertex {
375                        slot: self.slot_for_span(vertex.span)?,
376                        predicate: vertex
377                            .predicate
378                            .as_ref()
379                            .map(|predicate| self.bind_predicate(predicate))
380                            .transpose()?,
381                    },
382                    GraphPathPrimarySyntax::Edge(edge) => BoundPgqPathPrimary::Edge {
383                        slot: self.slot_for_span(edge.span)?,
384                        direction: edge.direction,
385                        predicate: edge
386                            .predicate
387                            .as_ref()
388                            .map(|predicate| self.bind_predicate(predicate))
389                            .transpose()?,
390                    },
391                    GraphPathPrimarySyntax::Parenthesized {
392                        path, predicate, ..
393                    } => BoundPgqPathPrimary::Parenthesized {
394                        path: Box::new(self.bind_path(path)?),
395                        predicate: predicate
396                            .as_ref()
397                            .map(|predicate| self.bind_predicate(predicate))
398                            .transpose()?,
399                    },
400                };
401                Ok(BoundPgqPathFactor {
402                    primary,
403                    quantifier: factor.quantifier.clone(),
404                })
405            })
406            .collect::<Result<Vec<_>, PgqBindError>>()?;
407        Ok(BoundPgqPath { factors })
408    }
409
410    fn slot_for_span(&self, span: Span) -> Result<PgqSlotId, PgqBindError> {
411        self.slots_by_span
412            .get(&(span.start, span.end))
413            .copied()
414            .ok_or_else(|| {
415                PgqBindError::new(
416                    PgqBindErrorCode::UnknownReference,
417                    span,
418                    "graph element has no registered slot",
419                )
420            })
421    }
422
423    fn bind_columns(
424        &self,
425        columns: &[hawdb_sql_syntax::GraphTableColumn],
426    ) -> Result<Vec<BoundPgqColumn>, PgqBindError> {
427        let mut names = BTreeSet::new();
428        columns
429            .iter()
430            .map(|column| {
431                let expression = self.bind_expression(&column.expression)?;
432                let name = if let Some(alias) = &column.alias {
433                    normalize_identifier(self.input, alias)
434                } else {
435                    derive_column_name(self.input, &column.expression).ok_or_else(|| {
436                        PgqBindError::new(
437                            PgqBindErrorCode::MissingColumnName,
438                            column.span,
439                            "complex GRAPH_TABLE output expressions require an explicit alias",
440                        )
441                    })?
442                };
443                if !names.insert(name.clone()) {
444                    return Err(PgqBindError::new(
445                        PgqBindErrorCode::DuplicateColumn,
446                        column.span,
447                        format!("duplicate GRAPH_TABLE output column {name}"),
448                    ));
449                }
450                Ok(BoundPgqColumn {
451                    name,
452                    data_type: expression.data_type,
453                    expression,
454                })
455            })
456            .collect()
457    }
458
459    fn apply_table_column_aliases(
460        &self,
461        alias: &TableAlias,
462        columns: &mut [BoundPgqColumn],
463    ) -> Result<(), PgqBindError> {
464        if alias.columns.len() > columns.len() {
465            return Err(PgqBindError::new(
466                PgqBindErrorCode::ColumnAliasCountMismatch,
467                alias.span,
468                "GRAPH_TABLE alias has more column names than its output schema",
469            ));
470        }
471        for (column, alias) in columns.iter_mut().zip(&alias.columns) {
472            column.name = normalize_identifier(self.input, alias);
473        }
474        let mut names = BTreeSet::new();
475        for column in columns {
476            if !names.insert(column.name.clone()) {
477                return Err(PgqBindError::new(
478                    PgqBindErrorCode::DuplicateColumn,
479                    alias.span,
480                    format!("duplicate GRAPH_TABLE output column {}", column.name),
481                ));
482            }
483        }
484        Ok(())
485    }
486
487    fn bind_predicate(
488        &self,
489        expression: &ExpressionSyntax,
490    ) -> Result<BoundPgqExpression, PgqBindError> {
491        let bound = self.bind_expression(expression)?;
492        if !matches!(bound.data_type, PgqDataType::Boolean | PgqDataType::Unknown) {
493            return Err(PgqBindError::new(
494                PgqBindErrorCode::TypeMismatch,
495                expression.span,
496                "graph predicates must have boolean type",
497            ));
498        }
499        Ok(bound)
500    }
501
502    fn bind_expression(
503        &self,
504        expression: &ExpressionSyntax,
505    ) -> Result<BoundPgqExpression, PgqBindError> {
506        let (kind, data_type) = match &expression.kind {
507            ExpressionKindSyntax::Like { .. } | ExpressionKindSyntax::Filtered { .. } => {
508                return Err(PgqBindError::new(
509                    PgqBindErrorCode::UnsupportedExpression,
510                    expression.span,
511                    "relational LIKE and FILTER are not qualified GRAPH_TABLE expressions",
512                ));
513            }
514            ExpressionKindSyntax::Column(name) => self.bind_column_reference(name)?,
515            ExpressionKindSyntax::Wildcard(_) => {
516                return Err(PgqBindError::new(
517                    PgqBindErrorCode::UnsupportedExpression,
518                    expression.span,
519                    "wildcards are not scalar GRAPH_TABLE expressions",
520                ));
521            }
522            ExpressionKindSyntax::Parameter(position) => (
523                BoundPgqExpressionKind::Parameter(*position),
524                PgqDataType::Unknown,
525            ),
526            ExpressionKindSyntax::Literal(literal) => {
527                let (literal, data_type) = bind_literal(self.input, literal)?;
528                (BoundPgqExpressionKind::Literal(literal), data_type)
529            }
530            ExpressionKindSyntax::TypedString { data_type, value } => {
531                let data_type_name = normalize_identifier(self.input, data_type);
532                let value = decode_string(self.input, *value).ok_or_else(|| {
533                    PgqBindError::new(
534                        PgqBindErrorCode::InvalidLiteral,
535                        *value,
536                        "invalid PostgreSQL string literal",
537                    )
538                })?;
539                (
540                    BoundPgqExpressionKind::Literal(BoundPgqLiteral::TypedString {
541                        data_type: data_type_name,
542                        value,
543                    }),
544                    PgqDataType::String,
545                )
546            }
547            ExpressionKindSyntax::Unary {
548                operator,
549                expression: inner,
550            } => {
551                let inner = self.bind_expression(inner)?;
552                let data_type = match operator {
553                    hawdb_sql_syntax::UnaryOperatorSyntax::Not => {
554                        require_boolean(inner.data_type, expression.span)?;
555                        PgqDataType::Boolean
556                    }
557                    hawdb_sql_syntax::UnaryOperatorSyntax::Plus
558                    | hawdb_sql_syntax::UnaryOperatorSyntax::Minus => {
559                        require_numeric(inner.data_type, expression.span)?;
560                        inner.data_type
561                    }
562                };
563                (
564                    BoundPgqExpressionKind::Unary {
565                        operator: *operator,
566                        expression: Box::new(inner),
567                    },
568                    data_type,
569                )
570            }
571            ExpressionKindSyntax::Binary {
572                left,
573                operator,
574                right,
575            } => {
576                let left = self.bind_expression(left)?;
577                let right = self.bind_expression(right)?;
578                validate_binary_operands(
579                    *operator,
580                    left.data_type,
581                    right.data_type,
582                    expression.span,
583                )?;
584                let data_type = binary_result_type(*operator, left.data_type, right.data_type);
585                (
586                    BoundPgqExpressionKind::Binary {
587                        left: Box::new(left),
588                        operator: *operator,
589                        right: Box::new(right),
590                    },
591                    data_type,
592                )
593            }
594            ExpressionKindSyntax::IsNull {
595                expression: inner,
596                negated,
597            } => (
598                BoundPgqExpressionKind::IsNull {
599                    expression: Box::new(self.bind_expression(inner)?),
600                    negated: *negated,
601                },
602                PgqDataType::Boolean,
603            ),
604            ExpressionKindSyntax::InList {
605                expression: inner,
606                values,
607                negated,
608            } => {
609                let inner = self.bind_expression(inner)?;
610                let values = values
611                    .iter()
612                    .map(|value| self.bind_expression(value))
613                    .collect::<Result<Vec<_>, _>>()?;
614                for value in &values {
615                    require_comparable(inner.data_type, value.data_type, expression.span)?;
616                }
617                (
618                    BoundPgqExpressionKind::InList {
619                        expression: Box::new(inner),
620                        values,
621                        negated: *negated,
622                    },
623                    PgqDataType::Boolean,
624                )
625            }
626            ExpressionKindSyntax::Between {
627                expression: inner,
628                low,
629                high,
630                negated,
631            } => {
632                let inner = self.bind_expression(inner)?;
633                let low = self.bind_expression(low)?;
634                let high = self.bind_expression(high)?;
635                require_comparable(inner.data_type, low.data_type, expression.span)?;
636                require_comparable(inner.data_type, high.data_type, expression.span)?;
637                (
638                    BoundPgqExpressionKind::Between {
639                        expression: Box::new(inner),
640                        low: Box::new(low),
641                        high: Box::new(high),
642                        negated: *negated,
643                    },
644                    PgqDataType::Boolean,
645                )
646            }
647            ExpressionKindSyntax::Function {
648                name,
649                arguments,
650                distinct,
651            } => {
652                let name = normalize_qualified_name(self.input, name);
653                let function = name.last().map(String::as_str).unwrap_or_default();
654                if is_disallowed_graph_table_function(function) {
655                    return Err(PgqBindError::new(
656                        PgqBindErrorCode::UnsupportedExpression,
657                        expression.span,
658                        format!("function {function} is not allowed in GRAPH_TABLE expressions"),
659                    ));
660                }
661                let arguments = arguments
662                    .iter()
663                    .map(|argument| self.bind_expression(argument))
664                    .collect::<Result<Vec<_>, _>>()?;
665                let data_type = function_result_type(function, &arguments);
666                (
667                    BoundPgqExpressionKind::Function {
668                        name,
669                        arguments,
670                        distinct: *distinct,
671                    },
672                    data_type,
673                )
674            }
675            ExpressionKindSyntax::Cast {
676                expression: inner,
677                data_type,
678            } => {
679                let inner = self.bind_expression(inner)?;
680                let data_type_name = normalize_qualified_name(self.input, data_type);
681                let result_type = cast_result_type(&data_type_name);
682                (
683                    BoundPgqExpressionKind::Cast {
684                        expression: Box::new(inner),
685                        data_type: data_type_name,
686                    },
687                    result_type,
688                )
689            }
690            ExpressionKindSyntax::Collate {
691                expression: inner,
692                collation,
693            } => {
694                let inner = self.bind_expression(inner)?;
695                require_string(inner.data_type, expression.span)?;
696                let data_type = inner.data_type;
697                (
698                    BoundPgqExpressionKind::Collate {
699                        expression: Box::new(inner),
700                        collation: normalize_qualified_name(self.input, collation),
701                    },
702                    data_type,
703                )
704            }
705            ExpressionKindSyntax::Parenthesized(inner) => {
706                let inner = self.bind_expression(inner)?;
707                return Ok(BoundPgqExpression {
708                    span: expression.span,
709                    ..inner
710                });
711            }
712        };
713        Ok(BoundPgqExpression {
714            kind,
715            data_type,
716            span: expression.span,
717        })
718    }
719
720    fn bind_column_reference(
721        &self,
722        name: &QualifiedName,
723    ) -> Result<(BoundPgqExpressionKind, PgqDataType), PgqBindError> {
724        let parts = normalize_qualified_name(self.input, name);
725        if let Some(variable) = parts
726            .first()
727            .and_then(|variable| self.named_slots.get(variable))
728            .and_then(|slot| self.variables.get(slot.0 as usize))
729        {
730            return match parts.as_slice() {
731                [_] => Ok((
732                    BoundPgqExpressionKind::Variable(variable.slot),
733                    PgqDataType::GraphElement(variable.kind),
734                )),
735                [_, property] => Ok((
736                    BoundPgqExpressionKind::Property {
737                        slot: variable.slot,
738                        property: property.clone(),
739                    },
740                    self.property_type(variable, property, name.span)?,
741                )),
742                _ => Err(PgqBindError::new(
743                    PgqBindErrorCode::UnsupportedExpression,
744                    name.span,
745                    "nested graph property paths are not supported",
746                )),
747            };
748        }
749        if let Some(data_type) = self.context.outer_columns.get(&parts).copied() {
750            return Ok((BoundPgqExpressionKind::OuterColumn(parts), data_type));
751        }
752        Err(PgqBindError::new(
753            PgqBindErrorCode::UnknownReference,
754            name.span,
755            format!(
756                "unknown GRAPH_TABLE expression reference {}",
757                parts.join(".")
758            ),
759        ))
760    }
761
762    fn property_type(
763        &self,
764        variable: &BoundPgqVariable,
765        property: &str,
766        span: Span,
767    ) -> Result<PgqDataType, PgqBindError> {
768        let schemas = self.element_schemas(variable.kind);
769        let candidates = if variable.labels.is_empty() {
770            schemas.values().collect::<Vec<_>>()
771        } else {
772            variable
773                .labels
774                .iter()
775                .filter_map(|label| schemas.get(label))
776                .collect::<Vec<_>>()
777        };
778        let mut types = candidates
779            .into_iter()
780            .filter_map(|schema| schema.properties.get(property).copied());
781        let Some(first) = types.next() else {
782            return Err(PgqBindError::new(
783                PgqBindErrorCode::UnknownProperty,
784                span,
785                format!("unknown property {property} for graph variable"),
786            ));
787        };
788        Ok(if types.all(|data_type| data_type == first) {
789            first
790        } else {
791            PgqDataType::Unknown
792        })
793    }
794
795    fn element_schemas(
796        &self,
797        kind: PgqElementKind,
798    ) -> &BTreeMap<String, PropertyGraphElementSchema> {
799        match kind {
800            PgqElementKind::Vertex => &self.graph.vertex_labels,
801            PgqElementKind::Edge => &self.graph.edge_labels,
802        }
803    }
804}
805
806fn normalize_table_alias(input: &str, alias: &TableAlias) -> String {
807    normalize_identifier(input, &alias.name)
808}
809
810fn normalize_qualified_name(input: &str, name: &QualifiedName) -> Vec<String> {
811    name.parts
812        .iter()
813        .map(|part| normalize_identifier(input, part))
814        .collect()
815}
816
817fn normalize_identifier(input: &str, identifier: &Identifier) -> String {
818    let text = &input[identifier.span.start..identifier.span.end];
819    if identifier.quoted {
820        text[1..text.len() - 1].replace("\"\"", "\"")
821    } else {
822        text.to_ascii_lowercase()
823    }
824}
825
826fn derive_column_name(input: &str, expression: &ExpressionSyntax) -> Option<String> {
827    let ExpressionKindSyntax::Column(name) = &expression.kind else {
828        return None;
829    };
830    name.parts
831        .last()
832        .map(|identifier| normalize_identifier(input, identifier))
833}
834
835fn bind_literal(
836    input: &str,
837    literal: &LiteralSyntax,
838) -> Result<(BoundPgqLiteral, PgqDataType), PgqBindError> {
839    match literal {
840        LiteralSyntax::Null => Ok((BoundPgqLiteral::Null, PgqDataType::Null)),
841        LiteralSyntax::Boolean(value) => {
842            Ok((BoundPgqLiteral::Boolean(*value), PgqDataType::Boolean))
843        }
844        LiteralSyntax::Number(span) => {
845            let text = &input[span.start..span.end];
846            if let Ok(value) = text.parse::<i64>() {
847                Ok((BoundPgqLiteral::Int64(value), PgqDataType::Int64))
848            } else if let Ok(value) = text.parse::<f64>() {
849                Ok((BoundPgqLiteral::Float64(value), PgqDataType::Float64))
850            } else {
851                Err(PgqBindError::new(
852                    PgqBindErrorCode::InvalidLiteral,
853                    *span,
854                    "numeric literal is outside the supported range",
855                ))
856            }
857        }
858        LiteralSyntax::String(span) => decode_string(input, *span)
859            .map(|value| (BoundPgqLiteral::String(value), PgqDataType::String))
860            .ok_or_else(|| {
861                PgqBindError::new(
862                    PgqBindErrorCode::InvalidLiteral,
863                    *span,
864                    "invalid PostgreSQL string literal",
865                )
866            }),
867    }
868}
869
870fn decode_string(input: &str, span: Span) -> Option<String> {
871    let text = input.get(span.start..span.end)?;
872    if text.starts_with('\'') && text.ends_with('\'') && text.len() >= 2 {
873        return Some(text[1..text.len() - 1].replace("''", "'"));
874    }
875    if let Some(without_prefix) = text.strip_prefix('$') {
876        let delimiter_end = without_prefix.find('$')? + 1;
877        let delimiter = &text[..=delimiter_end];
878        if text.ends_with(delimiter) && text.len() >= delimiter.len() * 2 {
879            return Some(text[delimiter.len()..text.len() - delimiter.len()].to_string());
880        }
881    }
882    None
883}
884
885fn binary_result_type(
886    operator: BinaryOperatorSyntax,
887    left: PgqDataType,
888    right: PgqDataType,
889) -> PgqDataType {
890    match operator {
891        BinaryOperatorSyntax::Or
892        | BinaryOperatorSyntax::And
893        | BinaryOperatorSyntax::Equal
894        | BinaryOperatorSyntax::NotEqual
895        | BinaryOperatorSyntax::Less
896        | BinaryOperatorSyntax::LessOrEqual
897        | BinaryOperatorSyntax::Greater
898        | BinaryOperatorSyntax::GreaterOrEqual => PgqDataType::Boolean,
899        BinaryOperatorSyntax::Concat => PgqDataType::String,
900        BinaryOperatorSyntax::Add
901        | BinaryOperatorSyntax::Subtract
902        | BinaryOperatorSyntax::Multiply
903        | BinaryOperatorSyntax::Divide
904        | BinaryOperatorSyntax::Modulo => numeric_common_type(left, right),
905    }
906}
907
908fn validate_binary_operands(
909    operator: BinaryOperatorSyntax,
910    left: PgqDataType,
911    right: PgqDataType,
912    span: Span,
913) -> Result<(), PgqBindError> {
914    match operator {
915        BinaryOperatorSyntax::Or | BinaryOperatorSyntax::And => {
916            require_boolean(left, span)?;
917            require_boolean(right, span)
918        }
919        BinaryOperatorSyntax::Concat => {
920            require_string(left, span)?;
921            require_string(right, span)
922        }
923        BinaryOperatorSyntax::Add
924        | BinaryOperatorSyntax::Subtract
925        | BinaryOperatorSyntax::Multiply
926        | BinaryOperatorSyntax::Divide
927        | BinaryOperatorSyntax::Modulo => {
928            require_numeric(left, span)?;
929            require_numeric(right, span)
930        }
931        BinaryOperatorSyntax::Equal
932        | BinaryOperatorSyntax::NotEqual
933        | BinaryOperatorSyntax::Less
934        | BinaryOperatorSyntax::LessOrEqual
935        | BinaryOperatorSyntax::Greater
936        | BinaryOperatorSyntax::GreaterOrEqual => require_comparable(left, right, span),
937    }
938}
939
940fn require_comparable(
941    left: PgqDataType,
942    right: PgqDataType,
943    span: Span,
944) -> Result<(), PgqBindError> {
945    if matches!(left, PgqDataType::Unknown | PgqDataType::Null)
946        || matches!(right, PgqDataType::Unknown | PgqDataType::Null)
947        || left == right
948        || matches!(
949            (left, right),
950            (PgqDataType::Int64, PgqDataType::Float64) | (PgqDataType::Float64, PgqDataType::Int64)
951        )
952    {
953        return Ok(());
954    }
955    Err(PgqBindError::new(
956        PgqBindErrorCode::TypeMismatch,
957        span,
958        format!("cannot compare {left:?} with {right:?}"),
959    ))
960}
961
962fn require_boolean(data_type: PgqDataType, span: Span) -> Result<(), PgqBindError> {
963    require_type(
964        data_type,
965        &[PgqDataType::Boolean],
966        span,
967        "boolean expression",
968    )
969}
970
971fn require_numeric(data_type: PgqDataType, span: Span) -> Result<(), PgqBindError> {
972    require_type(
973        data_type,
974        &[PgqDataType::Int64, PgqDataType::Float64],
975        span,
976        "numeric expression",
977    )
978}
979
980fn require_string(data_type: PgqDataType, span: Span) -> Result<(), PgqBindError> {
981    require_type(data_type, &[PgqDataType::String], span, "string expression")
982}
983
984fn require_type(
985    data_type: PgqDataType,
986    accepted: &[PgqDataType],
987    span: Span,
988    expected: &'static str,
989) -> Result<(), PgqBindError> {
990    if matches!(data_type, PgqDataType::Unknown | PgqDataType::Null)
991        || accepted.contains(&data_type)
992    {
993        return Ok(());
994    }
995    Err(PgqBindError::new(
996        PgqBindErrorCode::TypeMismatch,
997        span,
998        format!("expected {expected}, found {data_type:?}"),
999    ))
1000}
1001
1002fn numeric_common_type(left: PgqDataType, right: PgqDataType) -> PgqDataType {
1003    match (left, right) {
1004        (PgqDataType::Float64, PgqDataType::Int64 | PgqDataType::Float64)
1005        | (PgqDataType::Int64, PgqDataType::Float64) => PgqDataType::Float64,
1006        (PgqDataType::Int64, PgqDataType::Int64) => PgqDataType::Int64,
1007        _ => PgqDataType::Unknown,
1008    }
1009}
1010
1011fn function_result_type(function: &str, arguments: &[BoundPgqExpression]) -> PgqDataType {
1012    if ["lower", "upper", "pg_typeof"]
1013        .into_iter()
1014        .any(|candidate| function.eq_ignore_ascii_case(candidate))
1015    {
1016        PgqDataType::String
1017    } else {
1018        arguments
1019            .first()
1020            .map_or(PgqDataType::Unknown, |argument| argument.data_type)
1021    }
1022}
1023
1024fn cast_result_type(name: &[String]) -> PgqDataType {
1025    match name.last().map(String::as_str) {
1026        Some("bool" | "boolean") => PgqDataType::Boolean,
1027        Some("bigint" | "int8" | "integer" | "int4") => PgqDataType::Int64,
1028        Some("float8" | "double" | "real" | "float4") => PgqDataType::Float64,
1029        Some("text" | "varchar" | "character") => PgqDataType::String,
1030        _ => PgqDataType::Unknown,
1031    }
1032}
1033
1034fn is_disallowed_graph_table_function(function: &str) -> bool {
1035    [
1036        "avg",
1037        "count",
1038        "generate_series",
1039        "max",
1040        "min",
1041        "row_number",
1042        "sum",
1043    ]
1044    .into_iter()
1045    .any(|candidate| function.eq_ignore_ascii_case(candidate))
1046}
1047
1048fn element_kind_name(kind: PgqElementKind) -> &'static str {
1049    match kind {
1050        PgqElementKind::Vertex => "vertex",
1051        PgqElementKind::Edge => "edge",
1052    }
1053}