cpd_tokenizer/
javascript.rs

1// cpd-tokenizer: OXC-based tokenizer for JavaScript/TypeScript/JSX/TSX.
2// Dispatched for formats: javascript, typescript, jsx, tsx.
3
4use std::panic::{AssertUnwindSafe, catch_unwind};
5use std::path::Path;
6
7use oxc_allocator::Allocator;
8use oxc_parser::{Kind, Parser, config::TokensParserConfig};
9use oxc_span::SourceType;
10
11use cpd_core::models::{Token, TokenKind};
12
13use crate::line_index::LineIndex;
14
15// ── fallback tokenizer ────────────────────────────────────────────────────────
16
17mod fallback {
18    use super::LineIndex;
19    use cpd_core::models::{Token, TokenKind};
20
21    fn find_ignore_ranges(source: &str) -> Vec<[usize; 2]> {
22        let mut ranges = Vec::new();
23        let mut start: Option<usize> = None;
24        let bytes = source.as_bytes();
25        let mut i = 0;
26        while i < bytes.len() {
27            if i + 1 < bytes.len() && bytes[i] == b'/' {
28                let end = if bytes[i + 1] == b'/' {
29                    bytes[i..]
30                        .iter()
31                        .position(|&b| b == b'\n')
32                        .map(|p| i + p)
33                        .unwrap_or(bytes.len())
34                } else if bytes[i + 1] == b'*' {
35                    bytes[i..]
36                        .windows(2)
37                        .position(|w| w == b"*/")
38                        .map(|p| i + p + 2)
39                        .unwrap_or(bytes.len())
40                } else {
41                    i += 1;
42                    continue;
43                };
44                let comment_text = &source[i..end];
45                if comment_text.contains("jscpd:ignore-start") {
46                    start = Some(end);
47                } else if comment_text.contains("jscpd:ignore-end") {
48                    if let Some(s) = start.take() {
49                        ranges.push([s, i]);
50                    }
51                }
52                i = end;
53                continue;
54            }
55            i += 1;
56        }
57        ranges
58    }
59
60    fn in_ignore(offset: usize, end: usize, ranges: &[[usize; 2]]) -> bool {
61        ranges.iter().any(|[rs, re]| offset < *re && end > *rs)
62    }
63
64    /// Simple word-split fallback tokenizer. Never panics.
65    pub fn tokenize(source: &str, _format: &str) -> Vec<Token> {
66        let ignore_ranges = find_ignore_ranges(source);
67        let bytes = source.as_bytes();
68        let line_index = LineIndex::new(bytes);
69        let mut tokens = Vec::new();
70        let mut i = 0;
71        while i < bytes.len() {
72            let ch = match source[i..].chars().next() {
73                Some(c) => c,
74                None => break,
75            };
76            if ch.is_whitespace() {
77                i += ch.len_utf8();
78                continue;
79            }
80            if ch.is_alphanumeric() || ch == '_' || ch == '$' {
81                let start = i;
82                while i < bytes.len() {
83                    let c = source[i..].chars().next().unwrap_or('\0');
84                    if c.is_alphanumeric() || c == '_' || c == '$' {
85                        i += c.len_utf8();
86                    } else {
87                        break;
88                    }
89                }
90                let kind = if in_ignore(start, i, &ignore_ranges) {
91                    TokenKind::Ignore
92                } else {
93                    TokenKind::Other
94                };
95                tokens.push(Token {
96                    kind,
97                    value: source[start..i].to_string(),
98                    start: line_index.location(start),
99                    end: line_index.location(i),
100                });
101            } else {
102                let start = i;
103                i += ch.len_utf8();
104                let kind = if in_ignore(start, i, &ignore_ranges) {
105                    TokenKind::Ignore
106                } else {
107                    TokenKind::Other
108                };
109                tokens.push(Token {
110                    kind,
111                    value: ch.to_string(),
112                    start: line_index.location(start),
113                    end: line_index.location(i),
114                });
115            }
116        }
117        tokens
118    }
119}
120
121// ── helpers ──────────────────────────────────────────────────────────────────
122
123fn find_ignore_ranges(source: &str) -> Vec<[usize; 2]> {
124    let mut ranges = Vec::new();
125    let mut start: Option<usize> = None;
126    let bytes = source.as_bytes();
127    let mut i = 0;
128    while i < bytes.len() {
129        if i + 1 < bytes.len() && bytes[i] == b'/' {
130            let end = if bytes[i + 1] == b'/' {
131                bytes[i..]
132                    .iter()
133                    .position(|&b| b == b'\n')
134                    .map(|p| i + p)
135                    .unwrap_or(bytes.len())
136            } else if bytes[i + 1] == b'*' {
137                bytes[i..]
138                    .windows(2)
139                    .position(|w| w == b"*/")
140                    .map(|p| i + p + 2)
141                    .unwrap_or(bytes.len())
142            } else {
143                i += 1;
144                continue;
145            };
146            let comment_text = &source[i..end];
147            if comment_text.contains("jscpd:ignore-start") {
148                start = Some(end);
149            } else if comment_text.contains("jscpd:ignore-end") {
150                if let Some(s) = start.take() {
151                    ranges.push([s, i]);
152                }
153            }
154            i = end;
155            continue;
156        }
157        i += 1;
158    }
159    ranges
160}
161
162fn in_ignore(offset: usize, end: usize, ranges: &[[usize; 2]]) -> bool {
163    ranges.iter().any(|[rs, re]| offset < *re && end > *rs)
164}
165
166fn map_kind(kind: Kind) -> TokenKind {
167    if kind == Kind::Ident {
168        return TokenKind::Identifier;
169    }
170    if kind.is_any_keyword() {
171        return TokenKind::Keyword;
172    }
173    if kind.is_literal() {
174        return TokenKind::Literal;
175    }
176    if kind.is_assignment_operator() {
177        return TokenKind::Operator;
178    }
179    if kind.is_binary_operator()
180        || kind.is_logical_operator()
181        || kind.is_unary_operator()
182        || kind.is_update_operator()
183    {
184        return TokenKind::Operator;
185    }
186    match kind {
187        Kind::Arrow => TokenKind::Operator,
188        Kind::Semicolon
189        | Kind::Comma
190        | Kind::Dot
191        | Kind::Dot3
192        | Kind::Colon
193        | Kind::LParen
194        | Kind::RParen
195        | Kind::LCurly
196        | Kind::RCurly
197        | Kind::LBrack
198        | Kind::RBrack
199        | Kind::At => TokenKind::Punctuation,
200        Kind::QuestionDot => TokenKind::Punctuation,
201        _ => TokenKind::Other,
202    }
203}
204
205fn source_type_for_format(format: &str) -> SourceType {
206    let filename = match format {
207        "typescript" => "input.ts",
208        "tsx" => "input.tsx",
209        _ => "input.jsx", // javascript + jsx both use jsx
210    };
211    SourceType::from_path(Path::new(filename)).unwrap_or_default()
212}
213
214// ── public API ───────────────────────────────────────────────────────────────
215
216/// Tokenize JS/TS/JSX/TSX source. Never panics.
217pub fn tokenize_js(source: &str, format: &str) -> Vec<Token> {
218    if source.is_empty() {
219        return Vec::new();
220    }
221
222    let owned = source.to_string();
223    let fmt = format.to_string();
224    match catch_unwind(AssertUnwindSafe(|| parse_with_oxc(&owned, &fmt))) {
225        Ok(Some(tokens)) => tokens,
226        Ok(None) => {
227            log::debug!("cpd-tokenizer: OXC parse errors in {format} source, using fallback");
228            fallback::tokenize(source, format)
229        }
230        Err(_) => {
231            log::debug!("cpd-tokenizer: OXC panicked on {format} source, using fallback");
232            fallback::tokenize(source, format)
233        }
234    }
235}
236
237fn parse_with_oxc(source: &str, format: &str) -> Option<Vec<Token>> {
238    let allocator = Allocator::new();
239    let source_type = source_type_for_format(format);
240
241    let parser_return = Parser::new(&allocator, source, source_type)
242        .with_config(TokensParserConfig)
243        .parse();
244
245    if !parser_return.errors.is_empty() {
246        return None;
247    }
248
249    let ignore_ranges = find_ignore_ranges(source);
250    let bytes = source.as_bytes();
251    // Build LineIndex once — O(n) — then all location calls are O(log n).
252    let line_index = LineIndex::new(bytes);
253    let mut tokens = Vec::new();
254
255    for token in parser_return.tokens.iter() {
256        let start = (token.start() as usize).min(source.len());
257        let end = (token.end() as usize).min(source.len());
258        if start >= end {
259            continue;
260        }
261        let kind = token.kind();
262        if matches!(kind, Kind::Eof | Kind::Undetermined | Kind::Skip) {
263            continue;
264        }
265        let value = &source[start..end];
266        let token_kind = if in_ignore(start, end, &ignore_ranges) {
267            TokenKind::Ignore
268        } else {
269            map_kind(kind)
270        };
271        tokens.push(Token {
272            kind: token_kind,
273            value: value.to_string(),
274            start: line_index.location(start),
275            end: line_index.location(end),
276        });
277    }
278
279    Some(tokens)
280}
281
282// ── tests ─────────────────────────────────────────────────────────────────────
283#[cfg(test)]
284mod tests {
285    use super::*;
286
287    #[test]
288    fn valid_js_produces_tokens() {
289        let tokens = tokenize_js("function hello() { return 42; }", "javascript");
290        assert!(!tokens.is_empty(), "valid JS must produce tokens");
291    }
292
293    #[test]
294    fn typescript_produces_tokens() {
295        let tokens = tokenize_js("const x: number = 5;", "typescript");
296        assert!(!tokens.is_empty());
297    }
298
299    #[test]
300    fn malformed_js_does_not_panic() {
301        let result = std::panic::catch_unwind(|| tokenize_js("let x = {{{", "javascript"));
302        assert!(result.is_ok(), "malformed JS must not panic");
303    }
304
305    #[test]
306    fn empty_source_returns_empty() {
307        let tokens = tokenize_js("", "javascript");
308        drop(tokens);
309    }
310
311    #[test]
312    fn ignore_region_tokens_marked_as_ignore() {
313        let source = r#"
314const a = 1;
315// jscpd:ignore-start
316const b = 2;
317// jscpd:ignore-end
318const c = 3;
319"#;
320        let tokens = tokenize_js(source, "javascript");
321        let has_ignore = tokens
322            .iter()
323            .any(|t| t.kind == cpd_core::models::TokenKind::Ignore);
324        assert!(has_ignore, "tokens in ignore region must be marked Ignore");
325    }
326
327    #[test]
328    fn jsx_produces_tokens() {
329        let tokens = tokenize_js("const el = <div>hello</div>;", "jsx");
330        assert!(!tokens.is_empty());
331    }
332
333    #[test]
334    fn tsx_with_type_annotation() {
335        let tokens = tokenize_js("const fn = (x: React.FC): void => {};", "tsx");
336        assert!(!tokens.is_empty());
337    }
338
339    #[test]
340    fn multiline_location_uses_binary_search() {
341        let source = "const a = 1;\nconst b = 2;\nconst c = 3;";
342        let tokens = tokenize_js(source, "javascript");
343        // "b" is on line 2
344        let b_token = tokens.iter().find(|t| t.value == "b");
345        assert!(b_token.is_some(), "must find token b");
346        assert_eq!(b_token.unwrap().start.line, 2, "b must be on line 2");
347    }
348}
cpd_tokenizer/javascript.rs

cpd_tokenizer/
javascript.rs
