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, find_ignore_ranges, in_ignore};
19    use cpd_core::models::{Token, TokenKind};
20
21    /// Simple word-split fallback tokenizer. Never panics.
22    pub fn tokenize(source: &str, _format: &str) -> Vec<Token> {
23        let ignore_ranges = find_ignore_ranges(source);
24        let bytes = source.as_bytes();
25        let line_index = LineIndex::new(bytes);
26        let mut tokens = Vec::new();
27        let mut i = 0;
28        while i < bytes.len() {
29            let ch = match source[i..].chars().next() {
30                Some(c) => c,
31                None => break,
32            };
33            if ch.is_whitespace() {
34                i += ch.len_utf8();
35                continue;
36            }
37            if ch.is_alphanumeric() || ch == '_' || ch == '$' {
38                let start = i;
39                while i < bytes.len() {
40                    let c = source[i..].chars().next().unwrap_or('\0');
41                    if c.is_alphanumeric() || c == '_' || c == '$' {
42                        i += c.len_utf8();
43                    } else {
44                        break;
45                    }
46                }
47                let kind = if in_ignore(start, i, &ignore_ranges) {
48                    TokenKind::Ignore
49                } else {
50                    TokenKind::Other
51                };
52                tokens.push(Token {
53                    kind,
54                    value: source[start..i].to_string(),
55                    start: line_index.location(start),
56                    end: line_index.location(i),
57                });
58            } else {
59                let start = i;
60                i += ch.len_utf8();
61                let kind = if in_ignore(start, i, &ignore_ranges) {
62                    TokenKind::Ignore
63                } else {
64                    TokenKind::Other
65                };
66                tokens.push(Token {
67                    kind,
68                    value: ch.to_string(),
69                    start: line_index.location(start),
70                    end: line_index.location(i),
71                });
72            }
73        }
74        tokens
75    }
76}
77
78// ── helpers ──────────────────────────────────────────────────────────────────
79
80fn find_ignore_ranges(source: &str) -> Vec<[usize; 2]> {
81    let mut ranges = Vec::new();
82    let mut start: Option<usize> = None;
83    let bytes = source.as_bytes();
84    let mut i = 0;
85    while i < bytes.len() {
86        if i + 1 < bytes.len() && bytes[i] == b'/' {
87            let end = if bytes[i + 1] == b'/' {
88                bytes[i..]
89                    .iter()
90                    .position(|&b| b == b'\n')
91                    .map(|p| i + p)
92                    .unwrap_or(bytes.len())
93            } else if bytes[i + 1] == b'*' {
94                bytes[i..]
95                    .windows(2)
96                    .position(|w| w == b"*/")
97                    .map(|p| i + p + 2)
98                    .unwrap_or(bytes.len())
99            } else {
100                i += 1;
101                continue;
102            };
103            let comment_text = &source[i..end];
104            if comment_text.contains("jscpd:ignore-start") {
105                start = Some(end);
106            } else if comment_text.contains("jscpd:ignore-end") {
107                if let Some(s) = start.take() {
108                    ranges.push([s, i]);
109                }
110            }
111            i = end;
112            continue;
113        }
114        i += 1;
115    }
116    ranges
117}
118
119fn in_ignore(offset: usize, end: usize, ranges: &[[usize; 2]]) -> bool {
120    ranges.iter().any(|[rs, re]| offset < *re && end > *rs)
121}
122
123const fn map_kind(kind: Kind) -> TokenKind {
124    if matches!(kind, Kind::Ident) {
125        return TokenKind::Identifier;
126    }
127    if kind.is_any_keyword() {
128        return TokenKind::Keyword;
129    }
130    if kind.is_literal() {
131        return TokenKind::Literal;
132    }
133    if kind.is_assignment_operator() {
134        return TokenKind::Operator;
135    }
136    if kind.is_binary_operator()
137        || kind.is_logical_operator()
138        || kind.is_unary_operator()
139        || kind.is_update_operator()
140    {
141        return TokenKind::Operator;
142    }
143    match kind {
144        Kind::Arrow => TokenKind::Operator,
145        Kind::Semicolon
146        | Kind::Comma
147        | Kind::Dot
148        | Kind::Dot3
149        | Kind::Colon
150        | Kind::LParen
151        | Kind::RParen
152        | Kind::LCurly
153        | Kind::RCurly
154        | Kind::LBrack
155        | Kind::RBrack
156        | Kind::At => TokenKind::Punctuation,
157        Kind::QuestionDot => TokenKind::Punctuation,
158        _ => TokenKind::Other,
159    }
160}
161
162fn source_type_for_format(format: &str) -> SourceType {
163    let filename = match format {
164        "typescript" => "input.ts",
165        "tsx" => "input.tsx",
166        _ => "input.jsx", // javascript + jsx both use jsx
167    };
168    SourceType::from_path(Path::new(filename)).unwrap_or_default()
169}
170
171// ── public API ───────────────────────────────────────────────────────────────
172
173/// Tokenize JS/TS/JSX/TSX source. Never panics.
174pub fn tokenize_js(source: &str, format: &str) -> Vec<Token> {
175    tokenize_js_impl(source, format, false)
176}
177
178/// Tokenize TS/TSX source with erasable TypeScript-only syntax stripped from
179/// the token stream (cross-format detection). Token locations still reference
180/// the original source. Never panics.
181///
182/// Sources that fail to parse fall back to the word-split tokenizer WITHOUT
183/// stripping — they simply won't cross-match JavaScript files.
184pub fn tokenize_js_stripped(source: &str, format: &str) -> Vec<Token> {
185    tokenize_js_impl(source, format, true)
186}
187
188fn tokenize_js_impl(source: &str, format: &str, strip_types: bool) -> Vec<Token> {
189    if source.is_empty() {
190        return Vec::new();
191    }
192
193    match catch_unwind(AssertUnwindSafe(|| {
194        parse_with_oxc(source, format, strip_types)
195    })) {
196        Ok(Some(tokens)) => tokens,
197        Ok(None) => {
198            log::debug!("cpd-tokenizer: OXC parse errors in {format} source, using fallback");
199            fallback::tokenize(source, format)
200        }
201        Err(_) => {
202            log::debug!("cpd-tokenizer: OXC panicked on {format} source, using fallback");
203            fallback::tokenize(source, format)
204        }
205    }
206}
207
208fn parse_with_oxc(source: &str, format: &str, strip_types: bool) -> Option<Vec<Token>> {
209    let allocator = Allocator::new();
210    let source_type = source_type_for_format(format);
211
212    let parser_return = Parser::new(&allocator, source, source_type)
213        .with_config(TokensParserConfig)
214        .parse();
215
216    if !parser_return.errors.is_empty() {
217        return None;
218    }
219
220    let erasable = if strip_types {
221        crate::ts_strip::collect_erasable_spans(&parser_return.program, source)
222    } else {
223        crate::ts_strip::ErasableSpans::default()
224    };
225
226    let ignore_ranges = find_ignore_ranges(source);
227    let bytes = source.as_bytes();
228    // Build LineIndex once — O(n) — then all location calls are O(log n).
229    let line_index = LineIndex::new(bytes);
230
231    let tokens = parser_return
232        .tokens
233        .into_iter()
234        .filter_map(|token| {
235            let start = (token.start() as usize).min(source.len());
236            let end = (token.end() as usize).min(source.len());
237            if start >= end {
238                return None;
239            }
240            let kind = token.kind();
241            if matches!(kind, Kind::Eof | Kind::Undetermined | Kind::Skip) {
242                return None;
243            }
244            let value = &source[start..end];
245            if !erasable.is_empty()
246                && (erasable.intersects_range(start as u32, end as u32)
247                    || erasable.is_modifier_token(start as u32, end as u32, value))
248            {
249                return None;
250            }
251            let token_kind = if in_ignore(start, end, &ignore_ranges) {
252                TokenKind::Ignore
253            } else {
254                map_kind(kind)
255            };
256
257            Some(Token {
258                kind: token_kind,
259                value: value.to_string(),
260                start: line_index.location(start),
261                end: line_index.location(end),
262            })
263        })
264        .collect::<Vec<Token>>();
265
266    Some(tokens)
267}
268
269// ── tests ─────────────────────────────────────────────────────────────────────
270#[cfg(test)]
271mod tests {
272    use super::*;
273
274    #[test]
275    fn valid_js_produces_tokens() {
276        let tokens = tokenize_js("function hello() { return 42; }", "javascript");
277        assert!(!tokens.is_empty(), "valid JS must produce tokens");
278    }
279
280    #[test]
281    fn typescript_produces_tokens() {
282        let tokens = tokenize_js("const x: number = 5;", "typescript");
283        assert!(!tokens.is_empty());
284    }
285
286    #[test]
287    fn malformed_js_does_not_panic() {
288        let result = std::panic::catch_unwind(|| tokenize_js("let x = {{{", "javascript"));
289        assert!(result.is_ok(), "malformed JS must not panic");
290    }
291
292    #[test]
293    fn empty_source_returns_empty() {
294        let tokens = tokenize_js("", "javascript");
295        drop(tokens);
296    }
297
298    #[test]
299    fn ignore_region_tokens_marked_as_ignore() {
300        let source = r#"
301const a = 1;
302// jscpd:ignore-start
303const b = 2;
304// jscpd:ignore-end
305const c = 3;
306"#;
307        let tokens = tokenize_js(source, "javascript");
308        let has_ignore = tokens
309            .iter()
310            .any(|t| t.kind == cpd_core::models::TokenKind::Ignore);
311        assert!(has_ignore, "tokens in ignore region must be marked Ignore");
312    }
313
314    #[test]
315    fn jsx_produces_tokens() {
316        let tokens = tokenize_js("const el = <div>hello</div>;", "jsx");
317        assert!(!tokens.is_empty());
318    }
319
320    #[test]
321    fn tsx_with_type_annotation() {
322        let tokens = tokenize_js("const fn = (x: React.FC): void => {};", "tsx");
323        assert!(!tokens.is_empty());
324    }
325
326    #[test]
327    fn multiline_location_uses_binary_search() {
328        let source = "const a = 1;\nconst b = 2;\nconst c = 3;";
329        let tokens = tokenize_js(source, "javascript");
330        // "b" is on line 2
331        let b_token = tokens.iter().find(|t| t.value == "b");
332        assert!(b_token.is_some(), "must find token b");
333        assert_eq!(b_token.unwrap().start.line, 2, "b must be on line 2");
334    }
335}
cpd_tokenizer/javascript.rs

cpd_tokenizer/
javascript.rs
