Skip to main content

cpd_tokenizer/
sfc.rs

1use std::collections::BTreeMap;
2
3use cpd_core::models::{DetectionToken, Token};
4
5use crate::embedded::{blank_ranges_preserve_newlines, html_only_maps, tokenize_blocks_shifted};
6use crate::line_index::LineIndex;
7use crate::markdown::{offset_detection_tokens, tokens_to_detection};
8use crate::tokenizer::{Mode, TokenMap, TokenizeOptions, tokenize_format_to_detection};
9
10#[derive(Debug, Clone)]
11pub struct Block {
12    pub block_format: String,
13    pub content: String,
14    pub start_offset: usize,
15    pub start_line: u32,
16}
17
18struct SfcBlock {
19    block_format: String,
20    block_start: usize,
21    inner_start: usize,
22    inner_end: usize,
23    block_end: usize,
24}
25
26pub fn tokenize_sfc_maps(
27    source: &str,
28    file_format: &str,
29    options: &TokenizeOptions,
30) -> Vec<TokenMap> {
31    if source.is_empty() {
32        return Vec::new();
33    }
34
35    let blocks = find_sfc_blocks(source, file_format);
36    if blocks.is_empty() {
37        return html_only_maps(source, options);
38    }
39
40    let blank_ranges: Vec<[usize; 2]> = blocks
41        .iter()
42        .filter_map(|b| {
43            if b.inner_start < b.inner_end {
44                Some([b.inner_start, b.inner_end])
45            } else {
46                None
47            }
48        })
49        .collect();
50
51    let sanitized = blank_ranges_preserve_newlines(source, &blank_ranges);
52    let line_index = LineIndex::new(source.as_bytes());
53
54    let mut grouped: BTreeMap<String, Vec<DetectionToken>> = BTreeMap::new();
55
56    // A Vue file is top-level blocks only: all markup lives inside
57    // `<template>`, so what remains outside the blocks is just the wrapper
58    // tags — `<template>`, `</template>`, `<script ...>`, `</script>`,
59    // `<style ...>`, `</style>`. Those tags are identical in every Vue file
60    // and sit at both ends of it; fed into the html stream they would stretch
61    // every template clone to the last `</style>` and count the script and
62    // style bodies as duplicated html lines. That holds with or without a
63    // template block (a script-only component has nothing but wrapper tags
64    // outside its blocks). Svelte and Astro have no template wrapper: their
65    // top-level markup is the skeleton and is kept.
66    let wrapper_tags_only = file_format == "vue";
67
68    let markup_tokens = crate::generic::tokenize_generic(&sanitized, "html");
69    let mut markup_detection = tokens_to_detection(markup_tokens, options);
70    markup_detection.retain(|t| t.range[0] < t.range[1]);
71    if !markup_detection.is_empty() && !wrapper_tags_only {
72        grouped
73            .entry("html".to_string())
74            .or_default()
75            .extend(markup_detection);
76    }
77
78    for block in &blocks {
79        if block.inner_start >= block.inner_end {
80            continue;
81        }
82        let inner = &source[block.inner_start..block.inner_end];
83        let inner_start_loc = line_index.location(block.inner_start);
84
85        let mut inner_tokens = tokenize_format_to_detection(&block.block_format, inner, options);
86        offset_detection_tokens(&mut inner_tokens, block.inner_start, &inner_start_loc);
87
88        grouped
89            .entry(block.block_format.clone())
90            .or_default()
91            .extend(inner_tokens);
92    }
93
94    // Skeleton and block-inner tokens are collected separately — restore
95    // source order so clone endpoints follow the file.
96    if let Some(tokens) = grouped.get_mut("html") {
97        tokens.sort_by_key(|token| token.range[0]);
98    }
99
100    grouped
101        .into_iter()
102        .filter(|(_, tokens)| !tokens.is_empty())
103        .map(|(format, tokens)| TokenMap { format, tokens })
104        .collect()
105}
106
107fn sfc_tag_names(file_format: &str) -> &[&'static str] {
108    match file_format {
109        "svelte" | "astro" => &["script", "style"],
110        _ => &["template", "script", "style"],
111    }
112}
113
114fn find_sfc_blocks(source: &str, file_format: &str) -> Vec<SfcBlock> {
115    let source_lower = source.to_ascii_lowercase();
116    let mut blocks = Vec::new();
117
118    if file_format == "astro"
119        && let Some(fm) = astro_frontmatter_block(source)
120    {
121        blocks.push(fm);
122    }
123
124    for tag in sfc_tag_names(file_format) {
125        let mut search_from = 0usize;
126        while let Some(block) = find_sfc_tag_block(source, &source_lower, tag, search_from) {
127            search_from = block.block_end;
128            blocks.push(block);
129        }
130    }
131
132    blocks.sort_by_key(|b| b.block_start);
133    let mut deduped = Vec::new();
134    for block in blocks {
135        let nested = deduped.iter().any(|existing: &SfcBlock| {
136            block.block_start >= existing.block_start && block.block_start < existing.block_end
137        });
138        if !nested {
139            deduped.push(block);
140        }
141    }
142    deduped
143}
144
145fn find_tag_bounds(
146    _source: &str,
147    source_lower: &str,
148    tag: &str,
149    from: usize,
150) -> Option<(usize, usize, usize)> {
151    let open_needle = format!("<{}", tag);
152    let close_needle = format!("</{}>", tag);
153
154    let open_start = source_lower[from..].find(&open_needle)? + from;
155    let after_tag_name = open_start + 1 + tag.len();
156    if source_lower
157        .as_bytes()
158        .get(after_tag_name)
159        .is_some_and(|b| b.is_ascii_alphabetic())
160    {
161        return None;
162    }
163    let tag_end = source_lower[open_start..].find('>')? + open_start + 1;
164    let close_start = source_lower[tag_end..].find(&close_needle)? + tag_end;
165
166    Some((open_start, tag_end, close_start))
167}
168
169fn find_sfc_tag_block(
170    source: &str,
171    source_lower: &str,
172    tag: &str,
173    from: usize,
174) -> Option<SfcBlock> {
175    let (open_start, tag_end, close_start) = find_tag_bounds(source, source_lower, tag, from)?;
176    let close_needle = format!("</{}>", tag);
177
178    let attrs = &source[open_start + 1 + tag.len()..tag_end];
179    let inner_start = tag_end;
180    let inner_end = close_start;
181    let block_end = source_lower[close_start..]
182        .find('>')
183        .map(|i| close_start + i + 1)
184        .unwrap_or(close_start + close_needle.len());
185    let block_end = block_end.min(source.len());
186
187    let block_format = detect_sfc_block_format(attrs, tag);
188
189    Some(SfcBlock {
190        block_format,
191        block_start: open_start,
192        inner_start,
193        inner_end: inner_end.max(inner_start),
194        block_end,
195    })
196}
197
198fn detect_block_format(attrs: &str, tag: &str, strict: bool) -> String {
199    let lang = extract_lang_attr_value(attrs);
200    match tag {
201        "script" => match lang.as_deref() {
202            Some("ts" | "typescript") => "typescript".to_string(),
203            Some("js" | "javascript") => "javascript".to_string(),
204            Some(other) if strict => {
205                if crate::formats::get_format_by_extension(other).is_some()
206                    || crate::formats::SUPPORTED_FORMATS
207                        .iter()
208                        .any(|e| e.name == other)
209                {
210                    other.to_string()
211                } else {
212                    "javascript".to_string()
213                }
214            }
215            _ => "javascript".to_string(),
216        },
217        "style" => match lang.as_deref() {
218            Some("scss" | "sass") => "scss".to_string(),
219            Some("less") => "less".to_string(),
220            _ => "css".to_string(),
221        },
222        "template" => match lang.as_deref() {
223            Some(v) if v == "pug" || v == "jade" => "pug".to_string(),
224            _ => "html".to_string(),
225        },
226        _ => "html".to_string(),
227    }
228}
229
230fn detect_sfc_block_format(attrs: &str, tag: &str) -> String {
231    detect_block_format(attrs, tag, true)
232}
233
234fn detect_display_block_format(attrs: &str, tag: &str) -> String {
235    detect_block_format(attrs, tag, false)
236}
237
238fn astro_frontmatter_block(source: &str) -> Option<SfcBlock> {
239    if !(source.starts_with("---\n") || source.starts_with("---\r\n")) {
240        return None;
241    }
242    let lines = crate::markdown::line_spans(source);
243    let close_idx = lines
244        .iter()
245        .enumerate()
246        .skip(1)
247        .find(|(_, span)| source[span.start..span.end].trim() == "---")
248        .map(|(idx, _)| idx)?;
249    let inner_start = lines.get(1)?.start;
250    let inner_end = source[..lines[close_idx].start]
251        .strip_suffix('\n')
252        .map(|prefix: &str| prefix.len())
253        .unwrap_or(lines[close_idx].start);
254    let block_end = lines[close_idx].next_start.min(source.len());
255    Some(SfcBlock {
256        block_format: "typescript".to_string(),
257        block_start: 0,
258        inner_start,
259        inner_end: inner_end.max(inner_start),
260        block_end,
261    })
262}
263
264fn extract_lang_attr_value(attrs: &str) -> Option<String> {
265    let lower = attrs.to_ascii_lowercase();
266    let lang_pos = lower.find("lang=")?;
267    let rest = &attrs[lang_pos + 5..];
268    let quote = if rest.starts_with('"') {
269        '"'
270    } else if rest.starts_with('\'') {
271        '\''
272    } else {
273        return None;
274    };
275    let value_start = 1;
276    let value_end = rest[value_start..].find(quote)? + value_start;
277    Some(rest[value_start..value_end].to_ascii_lowercase())
278}
279
280/// Extract blocks from a Vue/Svelte/Astro file (display path).
281pub fn extract_blocks(source: &str, file_format: &str) -> Vec<Block> {
282    let source_lower = source.to_ascii_lowercase();
283    let mut blocks = Vec::new();
284    for tag in sfc_tag_names(file_format) {
285        let mut search_from = 0;
286        while let Some((block, next_from)) =
287            find_display_block(source, &source_lower, tag, search_from)
288        {
289            search_from = next_from;
290            blocks.push(block);
291        }
292    }
293    blocks.sort_by_key(|b: &Block| b.start_offset);
294    blocks
295}
296
297fn find_display_block(
298    source: &str,
299    source_lower: &str,
300    tag: &str,
301    from: usize,
302) -> Option<(Block, usize)> {
303    let (open_start, tag_end, close_start) = find_tag_bounds(source, source_lower, tag, from)?;
304
305    let attrs = &source[open_start + 1 + tag.len()..tag_end];
306    let content = source[tag_end..close_start].to_string();
307    let content_len = content.len();
308    let start_line = source[..tag_end].lines().count() as u32 + 1;
309    let block_format = detect_display_block_format(attrs, tag);
310
311    Some((
312        Block {
313            block_format,
314            content,
315            start_offset: tag_end,
316            start_line,
317        },
318        tag_end + content_len,
319    ))
320}
321
322pub fn tokenize_sfc(source: &str, file_format: &str, mode: Mode) -> Vec<Token> {
323    let blocks = extract_blocks(source, file_format);
324    tokenize_blocks_shifted(
325        blocks
326            .iter()
327            .map(|b| (b.block_format.as_str(), b.content.as_str(), b.start_line)),
328        mode,
329    )
330}
331
332#[cfg(test)]
333mod tests {
334    use super::*;
335
336    const VUE_FILE: &str = r#"<template>
337  <div>Hello</div>
338</template>
339
340<script>
341export default { name: 'Foo' }
342</script>
343
344<style>
345.foo { color: red; }
346</style>
347"#;
348
349    const VUE_TS_FILE: &str = r#"<template>
350  <div>Hello</div>
351</template>
352
353<script lang="ts">
354const x: number = 5;
355</script>
356
357<style lang="scss">
358.foo { color: red; }
359</style>
360"#;
361
362    #[test]
363    fn vue_file_extracts_three_blocks() {
364        let blocks = extract_blocks(VUE_FILE, "vue");
365        assert_eq!(blocks.len(), 3, "must find template, script, style blocks");
366    }
367
368    #[test]
369    fn script_block_default_format_is_javascript() {
370        let blocks = extract_blocks(VUE_FILE, "vue");
371        let script = blocks.iter().find(|b| b.block_format == "javascript");
372        assert!(script.is_some(), "plain <script> must be javascript format");
373    }
374
375    #[test]
376    fn script_lang_ts_produces_typescript_format() {
377        let blocks = extract_blocks(VUE_TS_FILE, "vue");
378        let ts_block = blocks.iter().find(|b| b.block_format == "typescript");
379        assert!(
380            ts_block.is_some(),
381            "<script lang=\"ts\"> must produce typescript format"
382        );
383    }
384
385    #[test]
386    fn unknown_lang_does_not_panic() {
387        let source = "<script lang=\"unknownlang123\">\nconst x = 1;\n</script>\n";
388        let result = std::panic::catch_unwind(|| extract_blocks(source, "vue"));
389        assert!(result.is_ok(), "unknown lang must not panic");
390    }
391
392    #[test]
393    fn no_blocks_returns_empty() {
394        let source = "just plain text no tags";
395        let blocks = extract_blocks(source, "vue");
396        assert!(blocks.is_empty());
397    }
398
399    #[test]
400    fn start_offset_is_after_opening_tag() {
401        let blocks = extract_blocks(VUE_FILE, "vue");
402        for block in &blocks {
403            assert!(block.start_offset > 0);
404        }
405    }
406
407    #[test]
408    fn vue_sfc_maps_produces_multiple_formats() {
409        let options = TokenizeOptions::new(Mode::Mild);
410        let maps = tokenize_sfc_maps(VUE_FILE, "vue", &options);
411        let formats: Vec<&str> = maps.iter().map(|m| m.format.as_str()).collect();
412        assert!(formats.contains(&"javascript"), "must have javascript map");
413        assert!(formats.contains(&"css"), "must have css map");
414        assert!(formats.contains(&"html"), "must have html map");
415    }
416
417    #[test]
418    fn vue_html_map_tokens_remain_in_source_order() {
419        let maps = tokenize_sfc_maps(VUE_FILE, "vue", &TokenizeOptions::new(Mode::Mild));
420        let html = maps.iter().find(|map| map.format == "html").unwrap();
421
422        assert!(
423            html.tokens
424                .windows(2)
425                .all(|pair| pair[0].range[0] <= pair[1].range[0]),
426            "HTML tokens must remain in source order"
427        );
428    }
429
430    #[test]
431    fn vue_html_map_holds_only_the_template_body() {
432        let maps = tokenize_sfc_maps(VUE_FILE, "vue", &TokenizeOptions::new(Mode::Mild));
433        let html = maps.iter().find(|map| map.format == "html").unwrap();
434        let template_start = VUE_FILE.find("<template>").unwrap() + "<template>".len();
435        let template_end = VUE_FILE.find("</template>").unwrap();
436        assert!(
437            html.tokens
438                .iter()
439                .all(|t| t.range[0] >= template_start && t.range[1] <= template_end),
440            "wrapper tags must not enter the html stream: {:?}",
441            html.tokens.iter().map(|t| t.range).collect::<Vec<_>>()
442        );
443        assert_eq!(html.tokens.first().unwrap().start.line, 2);
444        assert_eq!(html.tokens.last().unwrap().end.line, 2);
445    }
446
447    #[test]
448    fn vue_without_template_has_no_html_map() {
449        let source = "<script>\nexport default { render: (h) => h('div', 'x') }\n</script>\n<style>\n.x { margin: 0; }\n</style>\n";
450        let maps = tokenize_sfc_maps(source, "vue", &TokenizeOptions::new(Mode::Mild));
451        assert!(
452            maps.iter().all(|map| map.format != "html"),
453            "wrapper tags alone must not form an html map: {:?}",
454            maps.iter().map(|m| m.format.as_str()).collect::<Vec<_>>()
455        );
456    }
457
458    #[test]
459    fn svelte_keeps_top_level_markup() {
460        let source = "<script>\nlet n = 1;\n</script>\n<h1>Count {n}</h1>\n<button on:click={() => n++}>add</button>\n";
461        let maps = tokenize_sfc_maps(source, "svelte", &TokenizeOptions::new(Mode::Mild));
462        let html = maps.iter().find(|map| map.format == "html").unwrap();
463        assert!(
464            html.tokens.iter().any(|t| t.start.line == 4),
465            "top-level markup is the html map"
466        );
467    }
468
469    #[test]
470    fn vue_ts_maps_produces_typescript() {
471        let options = TokenizeOptions::new(Mode::Mild);
472        let maps = tokenize_sfc_maps(VUE_TS_FILE, "vue", &options);
473        let formats: Vec<&str> = maps.iter().map(|m| m.format.as_str()).collect();
474        assert!(formats.contains(&"typescript"), "must have typescript map");
475        assert!(formats.contains(&"scss"), "must have scss map");
476    }
477
478    #[test]
479    fn empty_sfc_returns_empty() {
480        let options = TokenizeOptions::new(Mode::Mild);
481        let maps = tokenize_sfc_maps("", "vue", &options);
482        assert!(maps.is_empty());
483    }
484
485    #[test]
486    fn svelte_sfc_maps_produces_multiple_formats() {
487        let source = r#"<script>
488  let count = 0;
489</script>
490
491<style>
492  .count { color: blue; }
493</style>
494"#;
495        let options = TokenizeOptions::new(Mode::Mild);
496        let maps = tokenize_sfc_maps(source, "svelte", &options);
497        let formats: Vec<&str> = maps.iter().map(|m| m.format.as_str()).collect();
498        assert!(
499            formats.contains(&"javascript"),
500            "svelte must have javascript map"
501        );
502        assert!(formats.contains(&"css"), "svelte must have css map");
503        assert!(
504            formats.contains(&"html"),
505            "svelte must have html markup map"
506        );
507    }
508
509    #[test]
510    fn svelte_script_containing_style_text_no_panic() {
511        let source = r#"<script>
512  const x = "<style>.red{color:red}</style>";
513</script>
514
515<style>
516  .blue { color: blue; }
517</style>
518"#;
519        let result = std::panic::catch_unwind(|| {
520            let options = TokenizeOptions::new(Mode::Mild);
521            tokenize_sfc_maps(source, "svelte", &options)
522        });
523        assert!(
524            result.is_ok(),
525            "must not panic when <style> text appears inside <script>"
526        );
527        let maps = result.unwrap();
528        let formats: Vec<&str> = maps.iter().map(|m| m.format.as_str()).collect();
529        assert!(
530            formats.contains(&"javascript"),
531            "must have javascript block"
532        );
533        assert!(formats.contains(&"css"), "must have real css block");
534        assert!(formats.contains(&"html"), "must have html markup");
535    }
536}