mz_sql_lexer/
lexer.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
// Copyright 2018 sqlparser-rs contributors. All rights reserved.
// Copyright Materialize, Inc. and contributors. All rights reserved.
//
// This file is derived from the sqlparser-rs project, available at
// https://github.com/andygrove/sqlparser-rs. It was incorporated
// directly into Materialize on December 21, 2019.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License in the LICENSE file at the
// root of this repository, or online at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

//! SQL lexer.
//!
//! This module lexes SQL according to the rules described in the ["Lexical
//! Structure"] section of the PostgreSQL documentation. The description is
//! intentionally not replicated here. Please refer to that chapter as you
//! read the code in this module.
//!
//! Where the PostgreSQL documentation is unclear, refer to their flex source
//! instead, located in the [backend/parser/scan.l] file in the PostgreSQL
//! Git repository.
//!
//! ["Lexical Structure"]: https://www.postgresql.org/docs/current/sql-syntax-lexical.html
//! [backend/parser/scan.l]: https://github.com/postgres/postgres/blob/90851d1d26f54ccb4d7b1bc49449138113d6ec83/src/backend/parser/scan.l

extern crate alloc;

use std::error::Error;
use std::{char, fmt};

use mz_ore::lex::LexBuf;
use mz_ore::str::{MaxLenString, StrExt};
use serde::{Deserialize, Serialize};

use crate::keywords::Keyword;

/// Maximum allowed identifier length in bytes.
pub const MAX_IDENTIFIER_LENGTH: usize = 255;

/// Newtype that limits the length of identifiers.
pub type IdentString = MaxLenString<MAX_IDENTIFIER_LENGTH>;

#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct LexerError {
    /// The error message.
    pub message: String,
    /// The byte position with which the error is associated.
    pub pos: usize,
}

impl fmt::Display for LexerError {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.write_str(&self.message)
    }
}

impl Error for LexerError {}

impl LexerError {
    /// Constructs an error with the provided message at the provided position.
    pub(crate) fn new<S>(pos: usize, message: S) -> LexerError
    where
        S: Into<String>,
    {
        LexerError {
            pos,
            message: message.into(),
        }
    }
}

#[derive(Debug, Clone, PartialEq)]
pub enum Token {
    Keyword(Keyword),
    Ident(IdentString),
    String(String),
    HexString(String),
    Number(String),
    Parameter(usize),
    Op(String),
    Star,
    Eq,
    LParen,
    RParen,
    LBracket,
    RBracket,
    Dot,
    Comma,
    Colon,
    DoubleColon,
    Semicolon,
    Arrow,
}

impl fmt::Display for Token {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        match self {
            Token::Keyword(kw) => f.write_str(kw.as_str()),
            Token::Ident(id) => write!(f, "identifier {}", id.quoted()),
            Token::String(s) => write!(f, "string literal {}", s.quoted()),
            Token::HexString(s) => write!(f, "hex string literal {}", s.quoted()),
            Token::Number(n) => write!(f, "number \"{}\"", n),
            Token::Parameter(n) => write!(f, "parameter \"${}\"", n),
            Token::Op(op) => write!(f, "operator {}", op.quoted()),
            Token::Star => f.write_str("star"),
            Token::Eq => f.write_str("equals sign"),
            Token::LParen => f.write_str("left parenthesis"),
            Token::RParen => f.write_str("right parenthesis"),
            Token::LBracket => f.write_str("left square bracket"),
            Token::RBracket => f.write_str("right square bracket"),
            Token::Dot => f.write_str("dot"),
            Token::Comma => f.write_str("comma"),
            Token::Colon => f.write_str("colon"),
            Token::DoubleColon => f.write_str("double colon"),
            Token::Semicolon => f.write_str("semicolon"),
            Token::Arrow => f.write_str("arrow"),
        }
    }
}

pub struct PosToken {
    pub kind: Token,
    pub offset: usize,
}

macro_rules! bail {
    ($pos:expr, $($fmt:expr),*) => {
        return Err(LexerError::new($pos, format!($($fmt),*)))
    }
}

/// Lexes a SQL query.
///
/// Returns a list of tokens alongside their corresponding byte offset in the
/// input string. Returns an error if the SQL query is lexically invalid.
///
/// See the module documentation for more information about the lexical
/// structure of SQL.
pub fn lex(query: &str) -> Result<Vec<PosToken>, LexerError> {
    let buf = &mut LexBuf::new(query);
    let mut tokens = vec![];
    while let Some(ch) = buf.next() {
        let pos = buf.pos() - ch.len_utf8();
        let token = match ch {
            _ if ch.is_ascii_whitespace() => continue,
            '-' if buf.consume('-') => {
                lex_line_comment(buf);
                continue;
            }
            '/' if buf.consume('*') => {
                lex_multiline_comment(buf)?;
                continue;
            }
            '\'' => Token::String(lex_string(buf)?),
            'x' | 'X' if buf.consume('\'') => Token::HexString(lex_string(buf)?),
            'e' | 'E' if buf.consume('\'') => lex_extended_string(buf)?,
            'A'..='Z' | 'a'..='z' | '_' | '\u{80}'..=char::MAX => lex_ident(buf)?,
            '"' => lex_quoted_ident(buf)?,
            '0'..='9' => lex_number(buf)?,
            '.' if matches!(buf.peek(), Some('0'..='9')) => lex_number(buf)?,
            '$' if matches!(buf.peek(), Some('0'..='9')) => lex_parameter(buf)?,
            '$' => lex_dollar_string(buf)?,
            '(' => Token::LParen,
            ')' => Token::RParen,
            ',' => Token::Comma,
            '.' => Token::Dot,
            ':' if buf.consume(':') => Token::DoubleColon,
            ':' => Token::Colon,
            ';' => Token::Semicolon,
            '[' => Token::LBracket,
            ']' => Token::RBracket,
            #[rustfmt::skip]
            '+'|'-'|'*'|'/'|'<'|'>'|'='|'~'|'!'|'@'|'#'|'%'|'^'|'&'|'|'|'`'|'?' => lex_op(buf),
            _ => bail!(pos, "unexpected character in input: {}", ch),
        };
        tokens.push(PosToken {
            kind: token,
            offset: pos,
        })
    }

    #[cfg(debug_assertions)]
    for token in &tokens {
        assert!(query.is_char_boundary(token.offset));
    }

    Ok(tokens)
}

fn lex_line_comment(buf: &mut LexBuf) {
    buf.take_while(|ch| ch != '\n');
}

fn lex_multiline_comment(buf: &mut LexBuf) -> Result<(), LexerError> {
    let pos = buf.pos() - 2;
    let mut nesting = 0;
    while let Some(ch) = buf.next() {
        match ch {
            '*' if buf.consume('/') => {
                if nesting == 0 {
                    return Ok(());
                } else {
                    nesting -= 1;
                }
            }
            '/' if buf.consume('*') => nesting += 1,
            _ => (),
        }
    }
    bail!(pos, "unterminated multiline comment")
}

fn lex_ident(buf: &mut LexBuf) -> Result<Token, LexerError> {
    buf.prev();
    let pos: usize = buf.pos();
    let word = buf.take_while(
        |ch| matches!(ch, 'A'..='Z' | 'a'..='z' | '0'..='9' | '$' | '_' | '\u{80}'..=char::MAX),
    );
    match word.parse() {
        Ok(kw) => Ok(Token::Keyword(kw)),
        Err(_) => {
            let Ok(small) = IdentString::new(word.to_lowercase()) else {
                bail!(
                    pos,
                    "identifier length exceeds {MAX_IDENTIFIER_LENGTH} bytes"
                )
            };
            Ok(Token::Ident(small))
        }
    }
}

fn lex_quoted_ident(buf: &mut LexBuf) -> Result<Token, LexerError> {
    let mut s = String::new();
    let pos = buf.pos() - 1;
    loop {
        match buf.next() {
            Some('"') if buf.consume('"') => s.push('"'),
            Some('"') => break,
            Some('\0') => bail!(pos, "null character in quoted identifier"),
            Some(c) => s.push(c),
            None => bail!(pos, "unterminated quoted identifier"),
        }
    }
    let Ok(small) = IdentString::new(s) else {
        bail!(
            pos,
            "identifier length exceeds {MAX_IDENTIFIER_LENGTH} bytes"
        )
    };
    Ok(Token::Ident(small))
}

fn lex_string(buf: &mut LexBuf) -> Result<String, LexerError> {
    let mut s = String::new();
    loop {
        let pos = buf.pos() - 1;
        loop {
            match buf.next() {
                Some('\'') if buf.consume('\'') => s.push('\''),
                Some('\'') => break,
                Some(c) => s.push(c),
                None => bail!(pos, "unterminated quoted string"),
            }
        }
        if !lex_to_adjacent_string(buf) {
            return Ok(s);
        }
    }
}

fn lex_extended_string(buf: &mut LexBuf) -> Result<Token, LexerError> {
    fn lex_unicode_escape(buf: &mut LexBuf, n: usize) -> Result<char, LexerError> {
        let pos = buf.pos() - 2;
        buf.next_n(n)
            .and_then(|s| u32::from_str_radix(s, 16).ok())
            .and_then(|codepoint| char::try_from(codepoint).ok())
            .ok_or_else(|| LexerError::new(pos, "invalid unicode escape"))
    }

    // We do not support octal (\o) or hexadecimal (\x) escapes, since it is
    // possible to construct invalid UTF-8 with these escapes. We could check
    // for and reject invalid UTF-8, of course, but it is too annoying to be
    // worth doing right now. We still lex the escapes to produce nice error
    // messages.

    fn lex_octal_escape(buf: &mut LexBuf) -> LexerError {
        let pos = buf.pos() - 2;
        buf.take_while(|ch| matches!(ch, '0'..='7'));
        LexerError::new(pos, "octal escapes are not supported")
    }

    fn lex_hexadecimal_escape(buf: &mut LexBuf) -> LexerError {
        let pos = buf.pos() - 2;
        buf.take_while(|ch| matches!(ch, '0'..='9' | 'A'..='F' | 'a'..='f'));
        LexerError::new(pos, "hexadecimal escapes are not supported")
    }

    let mut s = String::new();
    loop {
        let pos = buf.pos() - 1;
        loop {
            match buf.next() {
                Some('\'') if buf.consume('\'') => s.push('\''),
                Some('\'') => break,
                Some('\\') => match buf.next() {
                    Some('b') => s.push('\x08'),
                    Some('f') => s.push('\x0c'),
                    Some('n') => s.push('\n'),
                    Some('r') => s.push('\r'),
                    Some('t') => s.push('\t'),
                    Some('u') => s.push(lex_unicode_escape(buf, 4)?),
                    Some('U') => s.push(lex_unicode_escape(buf, 8)?),
                    Some('0'..='7') => return Err(lex_octal_escape(buf)),
                    Some('x') => return Err(lex_hexadecimal_escape(buf)),
                    Some(c) => s.push(c),
                    None => bail!(pos, "unterminated quoted string"),
                },
                Some(c) => s.push(c),
                None => bail!(pos, "unterminated quoted string"),
            }
        }
        if !lex_to_adjacent_string(buf) {
            return Ok(Token::String(s));
        }
    }
}

fn lex_to_adjacent_string(buf: &mut LexBuf) -> bool {
    // Adjacent string literals that are separated by whitespace are
    // concatenated if and only if that whitespace contains at least one newline
    // character. This bizarre rule matches PostgreSQL and the SQL standard.
    let whitespace = buf.take_while(|ch| ch.is_ascii_whitespace());
    whitespace.contains(&['\n', '\r'][..]) && buf.consume('\'')
}

fn lex_dollar_string(buf: &mut LexBuf) -> Result<Token, LexerError> {
    let pos = buf.pos() - 1;
    let tag = format!("${}$", buf.take_while(|ch| ch != '$'));
    let _ = buf.next();
    if let Some(s) = buf.take_to_delimiter(&tag) {
        Ok(Token::String(s.into()))
    } else {
        Err(LexerError::new(pos, "unterminated dollar-quoted string"))
    }
}

fn lex_parameter(buf: &mut LexBuf) -> Result<Token, LexerError> {
    let pos = buf.pos() - 1;
    let n = buf
        .take_while(|ch| matches!(ch, '0'..='9'))
        .parse()
        .map_err(|_| LexerError::new(pos, "invalid parameter number"))?;
    Ok(Token::Parameter(n))
}

fn lex_number(buf: &mut LexBuf) -> Result<Token, LexerError> {
    buf.prev();
    let mut s = buf.take_while(|ch| matches!(ch, '0'..='9')).to_owned();

    // Optional decimal component.
    if buf.consume('.') {
        s.push('.');
        s.push_str(buf.take_while(|ch| matches!(ch, '0'..='9')));
    }

    // Optional exponent.
    if buf.consume('e') || buf.consume('E') {
        s.push('E');
        let require_exp = if buf.consume('-') {
            s.push('-');
            true
        } else {
            buf.consume('+')
        };
        let exp = buf.take_while(|ch| matches!(ch, '0'..='9'));
        if require_exp && exp.is_empty() {
            return Err(LexerError::new(buf.pos() - 1, "missing required exponent"));
        } else if exp.is_empty() {
            // Put back consumed E.
            buf.prev();
            s.pop();
        } else {
            s.push_str(exp);
        }
    }

    Ok(Token::Number(s))
}

fn lex_op(buf: &mut LexBuf) -> Token {
    buf.prev();

    // Materialize special case: `=>` is lexed as an arrow token, rather than
    // an operator.
    if buf.consume_str("=>") {
        return Token::Arrow;
    }

    let mut s = String::new();

    // In PostgreSQL, operators might be composed of any of the characters in
    // the set below...
    while let Some(ch) = buf.next() {
        match ch {
            // ...except the sequences `--` and `/*` start comments, even within
            // what would otherwise be an operator...
            '-' if buf.peek() == Some('-') => {
                buf.prev();
                break;
            }
            '/' if buf.peek() == Some('*') => {
                buf.prev();
                break;
            }
            #[rustfmt::skip]
            '+'|'-'|'*'|'/'|'<'|'>'|'='|'~'|'!'|'@'|'#'|'%'|'^'|'&'|'|'|'`'|'?' => s.push(ch),
            _ => {
                buf.prev();
                break;
            }
        }
    }

    // ...and a multi-character operator that ends with `-` or `+` must also
    // contain at least one nonstandard operator character. This is so that e.g.
    // `1+-2` is lexed as `1 + (-2)` as required by the SQL standard, but `1@+2`
    // is lexed as `1 @+ 2`, as `@+` is meant to be a user-definable operator.
    if s.len() > 1
        && s.ends_with(&['-', '+'][..])
        && !s.contains(&['~', '!', '@', '#', '%', '^', '&', '|', '`', '?'][..])
    {
        while s.len() > 1 && s.ends_with(&['-', '+'][..]) {
            buf.prev();
            s.pop();
        }
    }

    match s.as_str() {
        // `*` and `=` are not just expression operators in SQL, so give them
        // dedicated tokens to simplify the parser.
        "*" => Token::Star,
        "=" => Token::Eq,
        // Normalize the two forms of the not-equals operator.
        "!=" => Token::Op("<>".into()),
        // Emit all other operators as is.
        _ => Token::Op(s),
    }
}