Skip to main content

sql_cli/sql/parser/expressions/
primary.rs

1// Primary expression parsing
2// Handles literals, identifiers, function calls, and parenthesized expressions
3
4use crate::sql::parser::ast::{ColumnRef, SqlExpression, WindowSpec};
5use crate::sql::parser::lexer::Token;
6use tracing::{debug, trace};
7
8use super::{log_parse_decision, trace_parse_entry, trace_parse_exit, ExpressionParser};
9
10/// Parser context for primary expressions
11pub struct PrimaryExpressionContext<'a> {
12    pub columns: &'a [String],
13    pub in_method_args: bool,
14}
15
16impl<'a> Default for PrimaryExpressionContext<'a> {
17    fn default() -> Self {
18        Self {
19            columns: &[],
20            in_method_args: false,
21        }
22    }
23}
24
25/// Parse a primary expression (literals, identifiers, functions, parentheses)
26/// This is the bottom of the expression hierarchy
27pub fn parse_primary<P>(
28    parser: &mut P,
29    ctx: &PrimaryExpressionContext,
30) -> Result<SqlExpression, String>
31where
32    P: ParsePrimary + ExpressionParser + ?Sized,
33{
34    trace_parse_entry("parse_primary", ExpressionParser::current_token(parser));
35
36    // Special case: check if a number literal could actually be a column name
37    // This handles cases where columns are named with pure numbers like "202204"
38    if let Token::NumberLiteral(num_str) = ExpressionParser::current_token(parser) {
39        if ctx.columns.iter().any(|col| col == num_str) {
40            log_parse_decision(
41                "parse_primary",
42                ExpressionParser::current_token(parser),
43                "Number literal matches column name, treating as column",
44            );
45            let expr = SqlExpression::Column(ColumnRef::unquoted(num_str.clone()));
46            ExpressionParser::advance(parser);
47            let result = Ok(expr);
48            trace_parse_exit("parse_primary", &result);
49            return result;
50        }
51    }
52
53    let result = match ExpressionParser::current_token(parser) {
54        Token::Case => {
55            debug!("Parsing CASE expression");
56            parser.parse_case_expression()
57        }
58
59        Token::DateTime => {
60            debug!("Parsing DateTime constructor");
61            parse_datetime_constructor(parser)
62        }
63
64        Token::Unnest => {
65            debug!("Parsing UNNEST expression");
66            parse_unnest(parser)
67        }
68
69        Token::Identifier(id) => {
70            let id_upper = id.to_uppercase();
71            let id_clone = id.clone();
72
73            // Check for boolean literals first
74            if id_upper == "TRUE" {
75                log_parse_decision(
76                    "parse_primary",
77                    ExpressionParser::current_token(parser),
78                    "Boolean literal TRUE",
79                );
80                ExpressionParser::advance(parser);
81                Ok(SqlExpression::BooleanLiteral(true))
82            } else if id_upper == "FALSE" {
83                log_parse_decision(
84                    "parse_primary",
85                    ExpressionParser::current_token(parser),
86                    "Boolean literal FALSE",
87                );
88                ExpressionParser::advance(parser);
89                Ok(SqlExpression::BooleanLiteral(false))
90            } else {
91                ExpressionParser::advance(parser);
92
93                // Check for table.column notation or method calls
94                if matches!(ExpressionParser::current_token(parser), Token::Dot) {
95                    ExpressionParser::advance(parser); // consume dot
96
97                    if let Token::Identifier(next_id) = ExpressionParser::current_token(parser) {
98                        let next_id = next_id.clone();
99                        ExpressionParser::advance(parser);
100
101                        // Check if this is a method call (followed by parentheses)
102                        if matches!(ExpressionParser::current_token(parser), Token::LeftParen) {
103                            debug!(object = %id_clone, method = %next_id, "Parsing method call");
104                            ExpressionParser::advance(parser); // consume (
105
106                            // Handle empty argument list
107                            let args = if matches!(
108                                ExpressionParser::current_token(parser),
109                                Token::RightParen
110                            ) {
111                                Vec::new()
112                            } else {
113                                parser.parse_expression_list()?
114                            };
115                            ExpressionParser::consume(parser, Token::RightParen)?;
116
117                            log_parse_decision(
118                                "parse_primary",
119                                &Token::Identifier(next_id.clone()),
120                                "Method call",
121                            );
122                            Ok(SqlExpression::MethodCall {
123                                object: id_clone,
124                                method: next_id,
125                                args,
126                            })
127                        } else {
128                            // It's a qualified column reference
129                            let col_ref = ColumnRef::qualified(id_clone, next_id.clone());
130                            log_parse_decision(
131                                "parse_primary",
132                                &Token::Identifier(next_id),
133                                "Qualified column reference",
134                            );
135                            Ok(SqlExpression::Column(col_ref))
136                        }
137                    } else {
138                        Err("Expected identifier after '.'".to_string())
139                    }
140                // CAST(expr AS type) / TRY_CAST(expr AS type) — the `AS type`
141                // form is not a normal argument list, so intercept it here and
142                // lower it into a two-arg function call CAST(expr, 'TYPE').
143                } else if (id_upper == "CAST" || id_upper == "TRY_CAST")
144                    && matches!(ExpressionParser::current_token(parser), Token::LeftParen)
145                {
146                    parse_cast_expression(parser, &id_upper)
147                // Check if this is a function call
148                } else if matches!(ExpressionParser::current_token(parser), Token::LeftParen) {
149                    debug!(function = %id_upper, "Parsing function call");
150                    ExpressionParser::advance(parser); // consume (
151                    let (args, has_distinct) = parser.parse_function_args()?;
152                    ExpressionParser::consume(parser, Token::RightParen)?;
153
154                    // Check for OVER clause for window functions
155                    if matches!(ExpressionParser::current_token(parser), Token::Over) {
156                        debug!(function = %id_upper, "Window function detected");
157                        ExpressionParser::advance(parser); // consume OVER
158                        ExpressionParser::consume(parser, Token::LeftParen)?;
159                        let window_spec = parser.parse_window_spec()?;
160                        ExpressionParser::consume(parser, Token::RightParen)?;
161                        Ok(SqlExpression::WindowFunction {
162                            name: id_upper,
163                            args,
164                            window_spec,
165                        })
166                    } else {
167                        Ok(SqlExpression::FunctionCall {
168                            name: id_upper,
169                            args,
170                            distinct: has_distinct,
171                        })
172                    }
173                } else {
174                    // Otherwise treat as simple column
175                    log_parse_decision(
176                        "parse_primary",
177                        &Token::Identifier(id_clone.clone()),
178                        "Column reference",
179                    );
180                    Ok(SqlExpression::Column(ColumnRef::unquoted(id_clone)))
181                }
182            }
183        }
184
185        Token::QuotedIdentifier(id) => {
186            let expr = if ctx.in_method_args {
187                // In method arguments, treat quoted identifiers as string literals
188                log_parse_decision(
189                    "parse_primary",
190                    ExpressionParser::current_token(parser),
191                    "Quoted identifier in method args - treating as string",
192                );
193                SqlExpression::StringLiteral(id.clone())
194            } else {
195                // Otherwise it's a column name like "Customer Id"
196                log_parse_decision(
197                    "parse_primary",
198                    ExpressionParser::current_token(parser),
199                    "Quoted identifier as column name",
200                );
201                SqlExpression::Column(ColumnRef::quoted(id.clone()))
202            };
203            ExpressionParser::advance(parser);
204            Ok(expr)
205        }
206
207        Token::StringLiteral(s) => {
208            trace!("String literal: {}", s);
209            let expr = SqlExpression::StringLiteral(s.clone());
210            ExpressionParser::advance(parser);
211            Ok(expr)
212        }
213
214        Token::NumberLiteral(n) => {
215            trace!("Number literal: {}", n);
216            let expr = SqlExpression::NumberLiteral(n.clone());
217            ExpressionParser::advance(parser);
218            Ok(expr)
219        }
220
221        Token::Null => {
222            trace!("NULL literal");
223            ExpressionParser::advance(parser);
224            Ok(SqlExpression::Null)
225        }
226
227        // Handle LEFT and RIGHT as function names when followed by parentheses
228        Token::Left | Token::Right => {
229            let func_name = match ExpressionParser::current_token(parser) {
230                Token::Left => "LEFT".to_string(),
231                Token::Right => "RIGHT".to_string(),
232                _ => unreachable!(),
233            };
234
235            ExpressionParser::advance(parser);
236
237            // Check if this is a function call
238            if matches!(ExpressionParser::current_token(parser), Token::LeftParen) {
239                debug!(function = %func_name, "Parsing LEFT/RIGHT function call");
240                ExpressionParser::advance(parser); // consume (
241                let (args, _has_distinct) = parser.parse_function_args()?;
242                ExpressionParser::consume(parser, Token::RightParen)?;
243
244                Ok(SqlExpression::FunctionCall {
245                    name: func_name,
246                    args,
247                    distinct: false,
248                })
249            } else {
250                // If not followed by parenthesis, this is likely a JOIN keyword - error
251                Err(format!(
252                    "{} keyword unexpected in expression context",
253                    func_name
254                ))
255            }
256        }
257
258        Token::LeftParen => {
259            debug!("Parsing parenthesized expression or subquery");
260            ExpressionParser::advance(parser); // consume (
261
262            // Check if this is a subquery (starts with SELECT)
263            if matches!(ExpressionParser::current_token(parser), Token::Select) {
264                debug!("Detected subquery - parsing SELECT statement");
265                let subquery = parser.parse_subquery()?;
266                ExpressionParser::consume(parser, Token::RightParen)?;
267                Ok(SqlExpression::ScalarSubquery {
268                    query: Box::new(subquery),
269                })
270            } else {
271                // Parenthesized expression, possibly a tuple for tuple IN:
272                // (a, b) IN (SELECT x, y FROM ...)
273                let first = parser.parse_logical_or()?;
274
275                if matches!(ExpressionParser::current_token(parser), Token::Comma) {
276                    // Collect the remaining tuple elements
277                    let mut exprs = vec![first];
278                    while matches!(ExpressionParser::current_token(parser), Token::Comma) {
279                        ExpressionParser::advance(parser); // consume ,
280                        exprs.push(parser.parse_logical_or()?);
281                    }
282                    ExpressionParser::consume(parser, Token::RightParen)?;
283
284                    // Expect IN or NOT IN immediately after
285                    match ExpressionParser::current_token(parser) {
286                        Token::In => {
287                            ExpressionParser::advance(parser); // consume IN
288                            ExpressionParser::consume(parser, Token::LeftParen)?;
289                            if !matches!(ExpressionParser::current_token(parser), Token::Select) {
290                                return Err("Tuple IN requires a subquery on the right".to_string());
291                            }
292                            let subquery = parser.parse_subquery()?;
293                            ExpressionParser::consume(parser, Token::RightParen)?;
294                            Ok(SqlExpression::InSubqueryTuple {
295                                exprs,
296                                subquery: Box::new(subquery),
297                            })
298                        }
299                        Token::Not => {
300                            ExpressionParser::advance(parser); // consume NOT
301                            if !matches!(ExpressionParser::current_token(parser), Token::In) {
302                                return Err("Expected IN after NOT for tuple".to_string());
303                            }
304                            ExpressionParser::advance(parser); // consume IN
305                            ExpressionParser::consume(parser, Token::LeftParen)?;
306                            if !matches!(ExpressionParser::current_token(parser), Token::Select) {
307                                return Err(
308                                    "Tuple NOT IN requires a subquery on the right".to_string()
309                                );
310                            }
311                            let subquery = parser.parse_subquery()?;
312                            ExpressionParser::consume(parser, Token::RightParen)?;
313                            Ok(SqlExpression::NotInSubqueryTuple {
314                                exprs,
315                                subquery: Box::new(subquery),
316                            })
317                        }
318                        _ => Err(
319                            "A tuple (expr, expr, ...) may only appear as the left side of IN / NOT IN"
320                                .to_string(),
321                        ),
322                    }
323                } else {
324                    // Regular parenthesized expression
325                    debug!("Regular parenthesized expression");
326                    ExpressionParser::consume(parser, Token::RightParen)?;
327                    Ok(first)
328                }
329            }
330        }
331
332        Token::Not => {
333            debug!("Parsing NOT expression");
334            parse_not_expression(parser)
335        }
336
337        Token::Star => {
338            // Handle * as a literal (like in COUNT(*))
339            trace!("Star token as literal");
340            ExpressionParser::advance(parser);
341            Ok(SqlExpression::StringLiteral("*".to_string()))
342        }
343
344        // Handle window-related keywords that can also be column names
345        Token::Row => {
346            trace!("ROW token treated as identifier in expression context");
347            ExpressionParser::advance(parser);
348            Ok(SqlExpression::Column(ColumnRef::unquoted(
349                "row".to_string(),
350            )))
351        }
352
353        Token::Rows => {
354            trace!("ROWS token treated as identifier in expression context");
355            ExpressionParser::advance(parser);
356            Ok(SqlExpression::Column(ColumnRef::unquoted(
357                "rows".to_string(),
358            )))
359        }
360
361        Token::Range => {
362            trace!("RANGE token treated as identifier in expression context");
363            ExpressionParser::advance(parser);
364            Ok(SqlExpression::Column(ColumnRef::unquoted(
365                "range".to_string(),
366            )))
367        }
368
369        Token::Minus => {
370            // Unary minus: -expr is parsed as 0 - expr
371            debug!("Parsing unary minus expression");
372            ExpressionParser::advance(parser);
373            let operand = parse_primary(parser, ctx)?;
374            Ok(SqlExpression::BinaryOp {
375                left: Box::new(SqlExpression::NumberLiteral("0".to_string())),
376                op: "-".to_string(),
377                right: Box::new(operand),
378            })
379        }
380
381        _ => {
382            let err = format!(
383                "Unexpected token in primary expression: {:?}",
384                ExpressionParser::current_token(parser)
385            );
386            debug!(error = %err);
387            Err(err)
388        }
389    };
390
391    trace_parse_exit("parse_primary", &result);
392    result
393}
394
395/// Parse DateTime constructor
396fn parse_datetime_constructor<P>(parser: &mut P) -> Result<SqlExpression, String>
397where
398    P: ParsePrimary + ExpressionParser + ?Sized,
399{
400    ExpressionParser::advance(parser); // consume DateTime
401    ExpressionParser::consume(parser, Token::LeftParen)?;
402
403    // Check if empty parentheses for DateTime() - today's date
404    if matches!(ExpressionParser::current_token(parser), Token::RightParen) {
405        ExpressionParser::advance(parser); // consume )
406        debug!("DateTime() - today's date");
407        return Ok(SqlExpression::DateTimeToday {
408            hour: None,
409            minute: None,
410            second: None,
411        });
412    }
413
414    // Parse year
415    let year = if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
416        n.parse::<i32>().map_err(|_| "Invalid year")?
417    } else {
418        return Err("Expected year in DateTime constructor".to_string());
419    };
420    ExpressionParser::advance(parser);
421    ExpressionParser::consume(parser, Token::Comma)?;
422
423    // Parse month
424    let month = if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
425        n.parse::<u32>().map_err(|_| "Invalid month")?
426    } else {
427        return Err("Expected month in DateTime constructor".to_string());
428    };
429    ExpressionParser::advance(parser);
430    ExpressionParser::consume(parser, Token::Comma)?;
431
432    // Parse day
433    let day = if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
434        n.parse::<u32>().map_err(|_| "Invalid day")?
435    } else {
436        return Err("Expected day in DateTime constructor".to_string());
437    };
438    ExpressionParser::advance(parser);
439
440    // Check for optional time components
441    let mut hour = None;
442    let mut minute = None;
443    let mut second = None;
444
445    if matches!(ExpressionParser::current_token(parser), Token::Comma) {
446        ExpressionParser::advance(parser); // consume comma
447
448        // Parse hour
449        if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
450            hour = Some(n.parse::<u32>().map_err(|_| "Invalid hour")?);
451            ExpressionParser::advance(parser);
452
453            // Check for minute
454            if matches!(ExpressionParser::current_token(parser), Token::Comma) {
455                ExpressionParser::advance(parser); // consume comma
456
457                if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
458                    minute = Some(n.parse::<u32>().map_err(|_| "Invalid minute")?);
459                    ExpressionParser::advance(parser);
460
461                    // Check for second
462                    if matches!(ExpressionParser::current_token(parser), Token::Comma) {
463                        ExpressionParser::advance(parser); // consume comma
464
465                        if let Token::NumberLiteral(n) = ExpressionParser::current_token(parser) {
466                            second = Some(n.parse::<u32>().map_err(|_| "Invalid second")?);
467                            ExpressionParser::advance(parser);
468                        }
469                    }
470                }
471            }
472        }
473    }
474
475    ExpressionParser::consume(parser, Token::RightParen)?;
476
477    debug!(year = year, month = month, day = day, hour = ?hour, minute = ?minute, second = ?second, "DateTime constructor parsed");
478
479    Ok(SqlExpression::DateTimeConstructor {
480        year,
481        month,
482        day,
483        hour,
484        minute,
485        second,
486    })
487}
488
489/// Parse NOT expression
490fn parse_not_expression<P>(parser: &mut P) -> Result<SqlExpression, String>
491where
492    P: ParsePrimary + ExpressionParser + ?Sized,
493{
494    ExpressionParser::advance(parser); // consume NOT
495
496    // Check if this is a NOT IN expression
497    if let Ok(inner_expr) = parser.parse_comparison() {
498        // After parsing the inner expression, check if we're followed by IN
499        if matches!(ExpressionParser::current_token(parser), Token::In) {
500            debug!("NOT IN expression detected");
501            ExpressionParser::advance(parser); // consume IN
502            ExpressionParser::consume(parser, Token::LeftParen)?;
503            let values = parser.parse_expression_list()?;
504            ExpressionParser::consume(parser, Token::RightParen)?;
505
506            Ok(SqlExpression::NotInList {
507                expr: Box::new(inner_expr),
508                values,
509            })
510        } else {
511            // Regular NOT expression
512            debug!("Regular NOT expression");
513            Ok(SqlExpression::Not {
514                expr: Box::new(inner_expr),
515            })
516        }
517    } else {
518        Err("Expected expression after NOT".to_string())
519    }
520}
521
522/// Parse UNNEST expression
523/// Syntax: UNNEST(column_expr, 'delimiter')
524fn parse_unnest<P>(parser: &mut P) -> Result<SqlExpression, String>
525where
526    P: ParsePrimary + ExpressionParser + ?Sized,
527{
528    debug!("parse_unnest: starting");
529    ExpressionParser::advance(parser); // consume UNNEST
530    ExpressionParser::consume(parser, Token::LeftParen)?;
531
532    // Parse the column expression (first argument)
533    let column = parser.parse_logical_or()?;
534    debug!("parse_unnest: parsed column expression");
535
536    // Expect comma
537    ExpressionParser::consume(parser, Token::Comma)?;
538
539    // Parse the delimiter (second argument - must be a string literal)
540    let delimiter = match ExpressionParser::current_token(parser) {
541        Token::StringLiteral(s) => {
542            let delim = s.clone();
543            ExpressionParser::advance(parser);
544            delim
545        }
546        _ => {
547            return Err("UNNEST delimiter must be a string literal".to_string());
548        }
549    };
550
551    debug!(delimiter = %delimiter, "parse_unnest: parsed delimiter");
552
553    ExpressionParser::consume(parser, Token::RightParen)?;
554
555    debug!("parse_unnest: complete");
556    Ok(SqlExpression::Unnest {
557        column: Box::new(column),
558        delimiter,
559    })
560}
561
562/// Parse a CAST / TRY_CAST expression.
563/// Syntax: `CAST(expr AS type)`.
564/// The current token on entry is the opening `(`. The result is lowered into a
565/// `FunctionCall` so it flows through the existing evaluator and AST machinery:
566/// `CAST(expr, 'TYPE')` where the type name is carried as a string literal.
567fn parse_cast_expression<P>(parser: &mut P, func_name: &str) -> Result<SqlExpression, String>
568where
569    P: ParsePrimary + ExpressionParser + ?Sized,
570{
571    ExpressionParser::advance(parser); // consume (
572
573    let inner = parser.parse_logical_or()?;
574
575    ExpressionParser::consume(parser, Token::As)?;
576
577    let type_name = parse_cast_type_name(parser)?;
578
579    ExpressionParser::consume(parser, Token::RightParen)?;
580
581    let name = if func_name.eq_ignore_ascii_case("TRY_CAST") {
582        "TRY_CAST"
583    } else {
584        "CAST"
585    };
586
587    debug!(target = %type_name, "Parsed CAST expression");
588    Ok(SqlExpression::FunctionCall {
589        name: name.to_string(),
590        args: vec![inner, SqlExpression::StringLiteral(type_name)],
591        distinct: false,
592    })
593}
594
595/// Read a SQL type name for CAST, e.g. `INTEGER`, `VARCHAR`, `DOUBLE`,
596/// `TIMESTAMP`. An optional precision/scale specifier such as `DECIMAL(10, 2)`
597/// or `VARCHAR(50)` is consumed and discarded — we coerce within our own type
598/// confines and do not honour width or scale.
599fn parse_cast_type_name<P>(parser: &mut P) -> Result<String, String>
600where
601    P: ParsePrimary + ExpressionParser + ?Sized,
602{
603    let type_name = match ExpressionParser::current_token(parser) {
604        Token::Identifier(id) => id.clone(),
605        // DATETIME is the one type spelling the lexer reserves as a keyword.
606        Token::DateTime => "DATETIME".to_string(),
607        other => {
608            return Err(format!(
609                "Expected a type name after AS in CAST, got {other:?}"
610            ))
611        }
612    };
613    ExpressionParser::advance(parser);
614
615    // Skip an optional (precision) or (precision, scale) specifier.
616    if matches!(ExpressionParser::current_token(parser), Token::LeftParen) {
617        ExpressionParser::advance(parser); // consume (
618        while !matches!(ExpressionParser::current_token(parser), Token::RightParen) {
619            if matches!(ExpressionParser::current_token(parser), Token::Eof) {
620                return Err("Unterminated type specifier in CAST".to_string());
621            }
622            ExpressionParser::advance(parser);
623        }
624        ExpressionParser::consume(parser, Token::RightParen)?;
625    }
626
627    Ok(type_name)
628}
629
630/// Trait that parsers must implement to use primary expression parsing
631pub trait ParsePrimary {
632    fn current_token(&self) -> &Token;
633    fn advance(&mut self);
634    fn consume(&mut self, expected: Token) -> Result<(), String>;
635
636    // These methods are called from parse_primary
637    fn parse_case_expression(&mut self) -> Result<SqlExpression, String>;
638    fn parse_function_args(&mut self) -> Result<(Vec<SqlExpression>, bool), String>;
639    fn parse_window_spec(&mut self) -> Result<WindowSpec, String>;
640    fn parse_logical_or(&mut self) -> Result<SqlExpression, String>;
641    fn parse_comparison(&mut self) -> Result<SqlExpression, String>;
642    fn parse_expression_list(&mut self) -> Result<Vec<SqlExpression>, String>;
643
644    // For subquery parsing (without parenthesis balance validation)
645    fn parse_subquery(&mut self) -> Result<crate::sql::parser::ast::SelectStatement, String>;
646}