fallow_extract/
html.rs

1//! HTML file parsing for script, stylesheet, and Angular template references.
2//!
3//! Extracts `<script src="...">` and `<link rel="stylesheet" href="...">` references
4//! from HTML files, creating graph edges so that referenced JS/CSS assets (and their
5//! transitive imports) are reachable from the HTML entry point.
6//!
7//! Also scans for Angular template syntax (`{{ }}`, `[prop]`, `(event)`, `@if`, etc.)
8//! and stores referenced identifiers as `MemberAccess` entries with a sentinel object,
9//! enabling the analysis phase to credit component class members used in external templates.
10
11use std::path::Path;
12use std::sync::LazyLock;
13
14use oxc_span::Span;
15
16use crate::asset_url::normalize_asset_url;
17use crate::sfc_template::angular::{self, ANGULAR_TPL_SENTINEL};
18use crate::{ImportInfo, ImportedName, MemberAccess, ModuleInfo};
19use fallow_types::discover::FileId;
20
21/// Regex to match HTML comments (`<!-- ... -->`) for stripping before extraction.
22static HTML_COMMENT_RE: LazyLock<regex::Regex> =
23    LazyLock::new(|| regex::Regex::new(r"(?s)<!--.*?-->").expect("valid regex"));
24
25/// Regex to extract `src` attribute from `<script>` tags.
26/// Matches both `<script src="...">` and `<script type="module" src="...">`.
27/// Uses `(?s)` so `.` matches newlines (multi-line attributes).
28static SCRIPT_SRC_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
29    regex::Regex::new(r#"(?si)<script\b(?:[^>"']|"[^"]*"|'[^']*')*?\bsrc\s*=\s*["']([^"']+)["']"#)
30        .expect("valid regex")
31});
32
33/// Regex to extract `href` attribute from `<link>` tags with `rel="stylesheet"` or
34/// `rel="modulepreload"`.
35/// Handles attributes in any order (rel before or after href).
36static LINK_HREF_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
37    regex::Regex::new(
38        r#"(?si)<link\b(?:[^>"']|"[^"]*"|'[^']*')*?\brel\s*=\s*["'](stylesheet|modulepreload)["'](?:[^>"']|"[^"]*"|'[^']*')*?\bhref\s*=\s*["']([^"']+)["']"#,
39    )
40    .expect("valid regex")
41});
42
43/// Regex for the reverse attribute order: href before rel.
44static LINK_HREF_REVERSE_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
45    regex::Regex::new(
46        r#"(?si)<link\b(?:[^>"']|"[^"]*"|'[^']*')*?\bhref\s*=\s*["']([^"']+)["'](?:[^>"']|"[^"]*"|'[^']*')*?\brel\s*=\s*["'](stylesheet|modulepreload)["']"#,
47    )
48    .expect("valid regex")
49});
50
51/// Check if a path is an HTML file.
52// Keep in sync with fallow_core::analyze::predicates::is_html_file (crate boundary prevents sharing)
53pub(crate) fn is_html_file(path: &Path) -> bool {
54    path.extension()
55        .and_then(|e| e.to_str())
56        .is_some_and(|ext| ext == "html")
57}
58
59/// Returns true if an HTML asset reference is a remote URL that should be skipped.
60pub(crate) fn is_remote_url(src: &str) -> bool {
61    src.starts_with("http://")
62        || src.starts_with("https://")
63        || src.starts_with("//")
64        || src.starts_with("data:")
65}
66
67/// Extract local (non-remote) asset references from HTML-like markup.
68///
69/// Returns the raw `src`/`href` strings (trimmed, remote URLs filtered). Shared
70/// between the HTML file parser and the JS/TS visitor's tagged template
71/// literal override so `` html`<script src="...">` `` in Hono/lit-html/htm
72/// layouts emits the same asset edges as a real `.html` file.
73pub(crate) fn collect_asset_refs(source: &str) -> Vec<String> {
74    let stripped = HTML_COMMENT_RE.replace_all(source, "");
75    let mut refs: Vec<String> = Vec::new();
76
77    for cap in SCRIPT_SRC_RE.captures_iter(&stripped) {
78        if let Some(m) = cap.get(1) {
79            let src = m.as_str().trim();
80            if !src.is_empty() && !is_remote_url(src) {
81                refs.push(src.to_string());
82            }
83        }
84    }
85
86    for cap in LINK_HREF_RE.captures_iter(&stripped) {
87        if let Some(m) = cap.get(2) {
88            let href = m.as_str().trim();
89            if !href.is_empty() && !is_remote_url(href) {
90                refs.push(href.to_string());
91            }
92        }
93    }
94    for cap in LINK_HREF_REVERSE_RE.captures_iter(&stripped) {
95        if let Some(m) = cap.get(1) {
96            let href = m.as_str().trim();
97            if !href.is_empty() && !is_remote_url(href) {
98                refs.push(href.to_string());
99            }
100        }
101    }
102
103    refs
104}
105
106/// Parse an HTML file, extracting script and stylesheet references as imports.
107pub(crate) fn parse_html_to_module(file_id: FileId, source: &str, content_hash: u64) -> ModuleInfo {
108    let suppressions = crate::suppress::parse_suppressions_from_source(source);
109
110    // Bare filenames (e.g., `src="app.js"`) are normalized to `./app.js` so
111    // the resolver doesn't misclassify them as npm packages.
112    let mut imports: Vec<ImportInfo> = collect_asset_refs(source)
113        .into_iter()
114        .map(|raw| ImportInfo {
115            source: normalize_asset_url(&raw),
116            imported_name: ImportedName::SideEffect,
117            local_name: String::new(),
118            is_type_only: false,
119            span: Span::default(),
120            source_span: Span::default(),
121        })
122        .collect();
123
124    // Deduplicate: the same asset may be referenced by both <script src> and
125    // <link rel="modulepreload" href> for the same path.
126    imports.sort_unstable_by(|a, b| a.source.cmp(&b.source));
127    imports.dedup_by(|a, b| a.source == b.source);
128
129    // Scan for Angular template syntax ({{ }}, [prop], (event), @if, etc.).
130    // Referenced identifiers are stored as MemberAccess entries with a sentinel
131    // object name so the analysis phase can bridge them to the component class.
132    let template_refs = angular::collect_angular_template_refs(source);
133    let member_accesses: Vec<MemberAccess> = template_refs
134        .into_iter()
135        .map(|name| MemberAccess {
136            object: ANGULAR_TPL_SENTINEL.to_string(),
137            member: name,
138        })
139        .collect();
140
141    ModuleInfo {
142        file_id,
143        exports: Vec::new(),
144        imports,
145        re_exports: Vec::new(),
146        dynamic_imports: Vec::new(),
147        dynamic_import_patterns: Vec::new(),
148        require_calls: Vec::new(),
149        member_accesses,
150        whole_object_uses: Vec::new(),
151        has_cjs_exports: false,
152        content_hash,
153        suppressions,
154        unused_import_bindings: Vec::new(),
155        line_offsets: fallow_types::extract::compute_line_offsets(source),
156        complexity: Vec::new(),
157        flag_uses: Vec::new(),
158    }
159}
160
161#[cfg(test)]
162mod tests {
163    use super::*;
164
165    // ── is_html_file ─────────────────────────────────────────────
166
167    #[test]
168    fn is_html_file_html() {
169        assert!(is_html_file(Path::new("index.html")));
170    }
171
172    #[test]
173    fn is_html_file_nested() {
174        assert!(is_html_file(Path::new("pages/about.html")));
175    }
176
177    #[test]
178    fn is_html_file_rejects_htm() {
179        assert!(!is_html_file(Path::new("index.htm")));
180    }
181
182    #[test]
183    fn is_html_file_rejects_js() {
184        assert!(!is_html_file(Path::new("app.js")));
185    }
186
187    #[test]
188    fn is_html_file_rejects_ts() {
189        assert!(!is_html_file(Path::new("app.ts")));
190    }
191
192    #[test]
193    fn is_html_file_rejects_vue() {
194        assert!(!is_html_file(Path::new("App.vue")));
195    }
196
197    // ── is_remote_url ────────────────────────────────────────────
198
199    #[test]
200    fn remote_url_http() {
201        assert!(is_remote_url("http://example.com/script.js"));
202    }
203
204    #[test]
205    fn remote_url_https() {
206        assert!(is_remote_url("https://cdn.example.com/style.css"));
207    }
208
209    #[test]
210    fn remote_url_protocol_relative() {
211        assert!(is_remote_url("//cdn.example.com/lib.js"));
212    }
213
214    #[test]
215    fn remote_url_data() {
216        assert!(is_remote_url("data:text/javascript;base64,abc"));
217    }
218
219    #[test]
220    fn local_relative_not_remote() {
221        assert!(!is_remote_url("./src/entry.js"));
222    }
223
224    #[test]
225    fn local_root_relative_not_remote() {
226        assert!(!is_remote_url("/src/entry.js"));
227    }
228
229    // ── parse_html_to_module: script src extraction ──────────────
230
231    #[test]
232    fn extracts_module_script_src() {
233        let info = parse_html_to_module(
234            FileId(0),
235            r#"<script type="module" src="./src/entry.js"></script>"#,
236            0,
237        );
238        assert_eq!(info.imports.len(), 1);
239        assert_eq!(info.imports[0].source, "./src/entry.js");
240    }
241
242    #[test]
243    fn extracts_plain_script_src() {
244        let info = parse_html_to_module(
245            FileId(0),
246            r#"<script src="./src/polyfills.js"></script>"#,
247            0,
248        );
249        assert_eq!(info.imports.len(), 1);
250        assert_eq!(info.imports[0].source, "./src/polyfills.js");
251    }
252
253    #[test]
254    fn extracts_multiple_scripts() {
255        let info = parse_html_to_module(
256            FileId(0),
257            r#"
258            <script type="module" src="./src/entry.js"></script>
259            <script src="./src/polyfills.js"></script>
260            "#,
261            0,
262        );
263        assert_eq!(info.imports.len(), 2);
264    }
265
266    #[test]
267    fn skips_inline_script() {
268        let info = parse_html_to_module(FileId(0), r#"<script>console.log("hello");</script>"#, 0);
269        assert!(info.imports.is_empty());
270    }
271
272    #[test]
273    fn skips_remote_script() {
274        let info = parse_html_to_module(
275            FileId(0),
276            r#"<script src="https://cdn.example.com/lib.js"></script>"#,
277            0,
278        );
279        assert!(info.imports.is_empty());
280    }
281
282    #[test]
283    fn skips_protocol_relative_script() {
284        let info = parse_html_to_module(
285            FileId(0),
286            r#"<script src="//cdn.example.com/lib.js"></script>"#,
287            0,
288        );
289        assert!(info.imports.is_empty());
290    }
291
292    // ── parse_html_to_module: link href extraction ───────────────
293
294    #[test]
295    fn extracts_stylesheet_link() {
296        let info = parse_html_to_module(
297            FileId(0),
298            r#"<link rel="stylesheet" href="./src/global.css" />"#,
299            0,
300        );
301        assert_eq!(info.imports.len(), 1);
302        assert_eq!(info.imports[0].source, "./src/global.css");
303    }
304
305    #[test]
306    fn extracts_modulepreload_link() {
307        let info = parse_html_to_module(
308            FileId(0),
309            r#"<link rel="modulepreload" href="./src/vendor.js" />"#,
310            0,
311        );
312        assert_eq!(info.imports.len(), 1);
313        assert_eq!(info.imports[0].source, "./src/vendor.js");
314    }
315
316    #[test]
317    fn extracts_link_with_reversed_attrs() {
318        let info = parse_html_to_module(
319            FileId(0),
320            r#"<link href="./src/global.css" rel="stylesheet" />"#,
321            0,
322        );
323        assert_eq!(info.imports.len(), 1);
324        assert_eq!(info.imports[0].source, "./src/global.css");
325    }
326
327    // ── Bare asset references normalized to relative paths ──────
328    // Regression tests for the same class of bug as #99 (Angular templateUrl).
329    // Browsers resolve `src="app.js"` and `href="styles.css"` relative to the
330    // HTML file, so emitting these as bare specifiers would misclassify them
331    // as unlisted npm packages.
332
333    #[test]
334    fn bare_script_src_normalized_to_relative() {
335        let info = parse_html_to_module(FileId(0), r#"<script src="app.js"></script>"#, 0);
336        assert_eq!(info.imports.len(), 1);
337        assert_eq!(info.imports[0].source, "./app.js");
338    }
339
340    #[test]
341    fn bare_module_script_src_normalized_to_relative() {
342        let info = parse_html_to_module(
343            FileId(0),
344            r#"<script type="module" src="main.ts"></script>"#,
345            0,
346        );
347        assert_eq!(info.imports.len(), 1);
348        assert_eq!(info.imports[0].source, "./main.ts");
349    }
350
351    #[test]
352    fn bare_stylesheet_link_href_normalized_to_relative() {
353        let info = parse_html_to_module(
354            FileId(0),
355            r#"<link rel="stylesheet" href="styles.css" />"#,
356            0,
357        );
358        assert_eq!(info.imports.len(), 1);
359        assert_eq!(info.imports[0].source, "./styles.css");
360    }
361
362    #[test]
363    fn bare_link_href_reversed_attrs_normalized_to_relative() {
364        let info = parse_html_to_module(
365            FileId(0),
366            r#"<link href="styles.css" rel="stylesheet" />"#,
367            0,
368        );
369        assert_eq!(info.imports.len(), 1);
370        assert_eq!(info.imports[0].source, "./styles.css");
371    }
372
373    #[test]
374    fn bare_modulepreload_link_href_normalized_to_relative() {
375        let info = parse_html_to_module(
376            FileId(0),
377            r#"<link rel="modulepreload" href="vendor.js" />"#,
378            0,
379        );
380        assert_eq!(info.imports.len(), 1);
381        assert_eq!(info.imports[0].source, "./vendor.js");
382    }
383
384    #[test]
385    fn bare_asset_with_subdir_normalized_to_relative() {
386        let info = parse_html_to_module(FileId(0), r#"<script src="assets/app.js"></script>"#, 0);
387        assert_eq!(info.imports.len(), 1);
388        assert_eq!(info.imports[0].source, "./assets/app.js");
389    }
390
391    #[test]
392    fn root_absolute_script_src_unchanged() {
393        // `/src/main.ts` is a web convention (Vite root-relative) and must
394        // stay absolute so the resolver's HTML special case applies.
395        let info = parse_html_to_module(FileId(0), r#"<script src="/src/main.ts"></script>"#, 0);
396        assert_eq!(info.imports.len(), 1);
397        assert_eq!(info.imports[0].source, "/src/main.ts");
398    }
399
400    #[test]
401    fn parent_relative_script_src_unchanged() {
402        let info = parse_html_to_module(
403            FileId(0),
404            r#"<script src="../shared/vendor.js"></script>"#,
405            0,
406        );
407        assert_eq!(info.imports.len(), 1);
408        assert_eq!(info.imports[0].source, "../shared/vendor.js");
409    }
410
411    #[test]
412    fn skips_preload_link() {
413        let info = parse_html_to_module(
414            FileId(0),
415            r#"<link rel="preload" href="./src/font.woff2" as="font" />"#,
416            0,
417        );
418        assert!(info.imports.is_empty());
419    }
420
421    #[test]
422    fn skips_icon_link() {
423        let info =
424            parse_html_to_module(FileId(0), r#"<link rel="icon" href="./favicon.ico" />"#, 0);
425        assert!(info.imports.is_empty());
426    }
427
428    #[test]
429    fn skips_remote_stylesheet() {
430        let info = parse_html_to_module(
431            FileId(0),
432            r#"<link rel="stylesheet" href="https://fonts.googleapis.com/css" />"#,
433            0,
434        );
435        assert!(info.imports.is_empty());
436    }
437
438    // ── HTML comment stripping ───────────────────────────────────
439
440    #[test]
441    fn skips_commented_out_script() {
442        let info = parse_html_to_module(
443            FileId(0),
444            r#"<!-- <script src="./old.js"></script> -->
445            <script src="./new.js"></script>"#,
446            0,
447        );
448        assert_eq!(info.imports.len(), 1);
449        assert_eq!(info.imports[0].source, "./new.js");
450    }
451
452    #[test]
453    fn skips_commented_out_link() {
454        let info = parse_html_to_module(
455            FileId(0),
456            r#"<!-- <link rel="stylesheet" href="./old.css" /> -->
457            <link rel="stylesheet" href="./new.css" />"#,
458            0,
459        );
460        assert_eq!(info.imports.len(), 1);
461        assert_eq!(info.imports[0].source, "./new.css");
462    }
463
464    // ── Multi-line attributes ────────────────────────────────────
465
466    #[test]
467    fn handles_multiline_script_tag() {
468        let info = parse_html_to_module(
469            FileId(0),
470            "<script\n  type=\"module\"\n  src=\"./src/entry.js\"\n></script>",
471            0,
472        );
473        assert_eq!(info.imports.len(), 1);
474        assert_eq!(info.imports[0].source, "./src/entry.js");
475    }
476
477    #[test]
478    fn handles_multiline_link_tag() {
479        let info = parse_html_to_module(
480            FileId(0),
481            "<link\n  rel=\"stylesheet\"\n  href=\"./src/global.css\"\n/>",
482            0,
483        );
484        assert_eq!(info.imports.len(), 1);
485        assert_eq!(info.imports[0].source, "./src/global.css");
486    }
487
488    // ── Full HTML document ───────────────────────────────────────
489
490    #[test]
491    fn full_vite_html() {
492        let info = parse_html_to_module(
493            FileId(0),
494            r#"<!doctype html>
495<html>
496  <head>
497    <link rel="stylesheet" href="./src/global.css" />
498    <link rel="icon" href="/favicon.ico" />
499  </head>
500  <body>
501    <div id="app"></div>
502    <script type="module" src="./src/entry.js"></script>
503  </body>
504</html>"#,
505            0,
506        );
507        assert_eq!(info.imports.len(), 2);
508        let sources: Vec<&str> = info.imports.iter().map(|i| i.source.as_str()).collect();
509        assert!(sources.contains(&"./src/global.css"));
510        assert!(sources.contains(&"./src/entry.js"));
511    }
512
513    // ── Edge cases ───────────────────────────────────────────────
514
515    #[test]
516    fn empty_html() {
517        let info = parse_html_to_module(FileId(0), "", 0);
518        assert!(info.imports.is_empty());
519    }
520
521    #[test]
522    fn html_with_no_assets() {
523        let info = parse_html_to_module(
524            FileId(0),
525            r"<!doctype html><html><body><h1>Hello</h1></body></html>",
526            0,
527        );
528        assert!(info.imports.is_empty());
529    }
530
531    #[test]
532    fn single_quoted_attributes() {
533        let info = parse_html_to_module(FileId(0), r"<script src='./src/entry.js'></script>", 0);
534        assert_eq!(info.imports.len(), 1);
535        assert_eq!(info.imports[0].source, "./src/entry.js");
536    }
537
538    #[test]
539    fn all_imports_are_side_effect() {
540        let info = parse_html_to_module(
541            FileId(0),
542            r#"<script src="./entry.js"></script>
543            <link rel="stylesheet" href="./style.css" />"#,
544            0,
545        );
546        for imp in &info.imports {
547            assert!(matches!(imp.imported_name, ImportedName::SideEffect));
548            assert!(imp.local_name.is_empty());
549            assert!(!imp.is_type_only);
550        }
551    }
552
553    #[test]
554    fn suppression_comments_extracted() {
555        let info = parse_html_to_module(
556            FileId(0),
557            "<!-- fallow-ignore-file -->\n<script src=\"./entry.js\"></script>",
558            0,
559        );
560        // HTML comments use <!-- --> not //, so suppression parsing
561        // from source text won't find standard JS-style comments.
562        // This is expected — HTML suppression is not supported.
563        assert_eq!(info.imports.len(), 1);
564    }
565
566    // ── Angular template scanning ──────────────────────────────
567
568    #[test]
569    fn angular_template_extracts_member_refs() {
570        let info = parse_html_to_module(
571            FileId(0),
572            "<h1>{{ title() }}</h1>\n\
573             <p [class.highlighted]=\"isHighlighted\">{{ greeting() }}</p>\n\
574             <button (click)=\"onButtonClick()\">Toggle</button>",
575            0,
576        );
577        let names: rustc_hash::FxHashSet<&str> = info
578            .member_accesses
579            .iter()
580            .filter(|a| a.object == ANGULAR_TPL_SENTINEL)
581            .map(|a| a.member.as_str())
582            .collect();
583        assert!(names.contains("title"), "should contain 'title'");
584        assert!(
585            names.contains("isHighlighted"),
586            "should contain 'isHighlighted'"
587        );
588        assert!(names.contains("greeting"), "should contain 'greeting'");
589        assert!(
590            names.contains("onButtonClick"),
591            "should contain 'onButtonClick'"
592        );
593    }
594
595    #[test]
596    fn plain_html_no_angular_refs() {
597        let info = parse_html_to_module(
598            FileId(0),
599            "<!doctype html><html><body><h1>Hello</h1></body></html>",
600            0,
601        );
602        assert!(info.member_accesses.is_empty());
603    }
604}
fallow_extract/html.rs

fallow_extract/
html.rs
