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.
107#[cfg(test)]
108pub(crate) fn parse_html_to_module(file_id: FileId, source: &str, content_hash: u64) -> ModuleInfo {
109    parse_html_to_module_with_complexity(file_id, source, content_hash, false)
110}
111
112/// Parse an HTML file and optionally compute Angular template complexity.
113pub(crate) fn parse_html_to_module_with_complexity(
114    file_id: FileId,
115    source: &str,
116    content_hash: u64,
117    need_complexity: bool,
118) -> ModuleInfo {
119    let parsed_suppressions = crate::suppress::parse_suppressions_from_source(source);
120
121    // Bare filenames (e.g., `src="app.js"`) are normalized to `./app.js` so
122    // the resolver doesn't misclassify them as npm packages.
123    let mut imports: Vec<ImportInfo> = collect_asset_refs(source)
124        .into_iter()
125        .map(|raw| ImportInfo {
126            source: normalize_asset_url(&raw),
127            imported_name: ImportedName::SideEffect,
128            local_name: String::new(),
129            is_type_only: false,
130            from_style: false,
131            span: Span::default(),
132            source_span: Span::default(),
133        })
134        .collect();
135
136    // Deduplicate: the same asset may be referenced by both <script src> and
137    // <link rel="modulepreload" href> for the same path.
138    imports.sort_unstable_by(|a, b| a.source.cmp(&b.source));
139    imports.dedup_by(|a, b| a.source == b.source);
140
141    // Scan for Angular template syntax ({{ }}, [prop], (event), @if, etc.).
142    //
143    // Bare identifier refs (e.g. `title`, `dataService`, pipe names) are stored
144    // as `MemberAccess` entries with a sentinel object name so the analysis
145    // phase can credit them as members of the component class.
146    //
147    // Static member-access chains (`dataService.getTotal`) where `dataService`
148    // is an unresolved identifier are stored as regular (non-sentinel)
149    // `MemberAccess` entries. The analysis phase resolves these through the
150    // importing component's typed instance bindings (from
151    // `ClassHeritageInfo.instance_bindings`) to credit the target class's
152    // member as used.
153    let template_refs = angular::collect_angular_template_refs(source);
154    let mut member_accesses: Vec<MemberAccess> = template_refs
155        .identifiers
156        .into_iter()
157        .map(|name| MemberAccess {
158            object: ANGULAR_TPL_SENTINEL.to_string(),
159            member: name,
160        })
161        .collect();
162    member_accesses.extend(template_refs.member_accesses);
163
164    let complexity = if need_complexity {
165        crate::template_complexity::compute_angular_template_complexity(source)
166            .into_iter()
167            .collect()
168    } else {
169        Vec::new()
170    };
171
172    ModuleInfo {
173        file_id,
174        exports: Vec::new(),
175        imports,
176        re_exports: Vec::new(),
177        dynamic_imports: Vec::new(),
178        dynamic_import_patterns: Vec::new(),
179        require_calls: Vec::new(),
180        member_accesses,
181        whole_object_uses: Vec::new(),
182        has_cjs_exports: false,
183        has_angular_component_template_url: false,
184        content_hash,
185        suppressions: parsed_suppressions.suppressions,
186        unknown_suppression_kinds: parsed_suppressions.unknown_kinds,
187        unused_import_bindings: Vec::new(),
188        type_referenced_import_bindings: Vec::new(),
189        value_referenced_import_bindings: Vec::new(),
190        line_offsets: fallow_types::extract::compute_line_offsets(source),
191        complexity,
192        flag_uses: Vec::new(),
193        class_heritage: vec![],
194        local_type_declarations: Vec::new(),
195        public_signature_type_references: Vec::new(),
196        namespace_object_aliases: Vec::new(),
197    }
198}
199
200#[cfg(test)]
201mod tests {
202    use super::*;
203
204    // ── is_html_file ─────────────────────────────────────────────
205
206    #[test]
207    fn is_html_file_html() {
208        assert!(is_html_file(Path::new("index.html")));
209    }
210
211    #[test]
212    fn is_html_file_nested() {
213        assert!(is_html_file(Path::new("pages/about.html")));
214    }
215
216    #[test]
217    fn is_html_file_rejects_htm() {
218        assert!(!is_html_file(Path::new("index.htm")));
219    }
220
221    #[test]
222    fn is_html_file_rejects_js() {
223        assert!(!is_html_file(Path::new("app.js")));
224    }
225
226    #[test]
227    fn is_html_file_rejects_ts() {
228        assert!(!is_html_file(Path::new("app.ts")));
229    }
230
231    #[test]
232    fn is_html_file_rejects_vue() {
233        assert!(!is_html_file(Path::new("App.vue")));
234    }
235
236    // ── is_remote_url ────────────────────────────────────────────
237
238    #[test]
239    fn remote_url_http() {
240        assert!(is_remote_url("http://example.com/script.js"));
241    }
242
243    #[test]
244    fn remote_url_https() {
245        assert!(is_remote_url("https://cdn.example.com/style.css"));
246    }
247
248    #[test]
249    fn remote_url_protocol_relative() {
250        assert!(is_remote_url("//cdn.example.com/lib.js"));
251    }
252
253    #[test]
254    fn remote_url_data() {
255        assert!(is_remote_url("data:text/javascript;base64,abc"));
256    }
257
258    #[test]
259    fn local_relative_not_remote() {
260        assert!(!is_remote_url("./src/entry.js"));
261    }
262
263    #[test]
264    fn local_root_relative_not_remote() {
265        assert!(!is_remote_url("/src/entry.js"));
266    }
267
268    // ── parse_html_to_module: script src extraction ──────────────
269
270    #[test]
271    fn extracts_module_script_src() {
272        let info = parse_html_to_module(
273            FileId(0),
274            r#"<script type="module" src="./src/entry.js"></script>"#,
275            0,
276        );
277        assert_eq!(info.imports.len(), 1);
278        assert_eq!(info.imports[0].source, "./src/entry.js");
279    }
280
281    #[test]
282    fn extracts_plain_script_src() {
283        let info = parse_html_to_module(
284            FileId(0),
285            r#"<script src="./src/polyfills.js"></script>"#,
286            0,
287        );
288        assert_eq!(info.imports.len(), 1);
289        assert_eq!(info.imports[0].source, "./src/polyfills.js");
290    }
291
292    #[test]
293    fn extracts_multiple_scripts() {
294        let info = parse_html_to_module(
295            FileId(0),
296            r#"
297            <script type="module" src="./src/entry.js"></script>
298            <script src="./src/polyfills.js"></script>
299            "#,
300            0,
301        );
302        assert_eq!(info.imports.len(), 2);
303    }
304
305    #[test]
306    fn skips_inline_script() {
307        let info = parse_html_to_module(FileId(0), r#"<script>console.log("hello");</script>"#, 0);
308        assert!(info.imports.is_empty());
309    }
310
311    #[test]
312    fn skips_remote_script() {
313        let info = parse_html_to_module(
314            FileId(0),
315            r#"<script src="https://cdn.example.com/lib.js"></script>"#,
316            0,
317        );
318        assert!(info.imports.is_empty());
319    }
320
321    #[test]
322    fn skips_protocol_relative_script() {
323        let info = parse_html_to_module(
324            FileId(0),
325            r#"<script src="//cdn.example.com/lib.js"></script>"#,
326            0,
327        );
328        assert!(info.imports.is_empty());
329    }
330
331    // ── parse_html_to_module: link href extraction ───────────────
332
333    #[test]
334    fn extracts_stylesheet_link() {
335        let info = parse_html_to_module(
336            FileId(0),
337            r#"<link rel="stylesheet" href="./src/global.css" />"#,
338            0,
339        );
340        assert_eq!(info.imports.len(), 1);
341        assert_eq!(info.imports[0].source, "./src/global.css");
342    }
343
344    #[test]
345    fn extracts_modulepreload_link() {
346        let info = parse_html_to_module(
347            FileId(0),
348            r#"<link rel="modulepreload" href="./src/vendor.js" />"#,
349            0,
350        );
351        assert_eq!(info.imports.len(), 1);
352        assert_eq!(info.imports[0].source, "./src/vendor.js");
353    }
354
355    #[test]
356    fn extracts_link_with_reversed_attrs() {
357        let info = parse_html_to_module(
358            FileId(0),
359            r#"<link href="./src/global.css" rel="stylesheet" />"#,
360            0,
361        );
362        assert_eq!(info.imports.len(), 1);
363        assert_eq!(info.imports[0].source, "./src/global.css");
364    }
365
366    // ── Bare asset references normalized to relative paths ──────
367    // Regression tests for the same class of bug as #99 (Angular templateUrl).
368    // Browsers resolve `src="app.js"` and `href="styles.css"` relative to the
369    // HTML file, so emitting these as bare specifiers would misclassify them
370    // as unlisted npm packages.
371
372    #[test]
373    fn bare_script_src_normalized_to_relative() {
374        let info = parse_html_to_module(FileId(0), r#"<script src="app.js"></script>"#, 0);
375        assert_eq!(info.imports.len(), 1);
376        assert_eq!(info.imports[0].source, "./app.js");
377    }
378
379    #[test]
380    fn bare_module_script_src_normalized_to_relative() {
381        let info = parse_html_to_module(
382            FileId(0),
383            r#"<script type="module" src="main.ts"></script>"#,
384            0,
385        );
386        assert_eq!(info.imports.len(), 1);
387        assert_eq!(info.imports[0].source, "./main.ts");
388    }
389
390    #[test]
391    fn bare_stylesheet_link_href_normalized_to_relative() {
392        let info = parse_html_to_module(
393            FileId(0),
394            r#"<link rel="stylesheet" href="styles.css" />"#,
395            0,
396        );
397        assert_eq!(info.imports.len(), 1);
398        assert_eq!(info.imports[0].source, "./styles.css");
399    }
400
401    #[test]
402    fn bare_link_href_reversed_attrs_normalized_to_relative() {
403        let info = parse_html_to_module(
404            FileId(0),
405            r#"<link href="styles.css" rel="stylesheet" />"#,
406            0,
407        );
408        assert_eq!(info.imports.len(), 1);
409        assert_eq!(info.imports[0].source, "./styles.css");
410    }
411
412    #[test]
413    fn bare_modulepreload_link_href_normalized_to_relative() {
414        let info = parse_html_to_module(
415            FileId(0),
416            r#"<link rel="modulepreload" href="vendor.js" />"#,
417            0,
418        );
419        assert_eq!(info.imports.len(), 1);
420        assert_eq!(info.imports[0].source, "./vendor.js");
421    }
422
423    #[test]
424    fn bare_asset_with_subdir_normalized_to_relative() {
425        let info = parse_html_to_module(FileId(0), r#"<script src="assets/app.js"></script>"#, 0);
426        assert_eq!(info.imports.len(), 1);
427        assert_eq!(info.imports[0].source, "./assets/app.js");
428    }
429
430    #[test]
431    fn root_absolute_script_src_unchanged() {
432        // `/src/main.ts` is a web convention (Vite root-relative) and must
433        // stay absolute so the resolver's HTML special case applies.
434        let info = parse_html_to_module(FileId(0), r#"<script src="/src/main.ts"></script>"#, 0);
435        assert_eq!(info.imports.len(), 1);
436        assert_eq!(info.imports[0].source, "/src/main.ts");
437    }
438
439    #[test]
440    fn parent_relative_script_src_unchanged() {
441        let info = parse_html_to_module(
442            FileId(0),
443            r#"<script src="../shared/vendor.js"></script>"#,
444            0,
445        );
446        assert_eq!(info.imports.len(), 1);
447        assert_eq!(info.imports[0].source, "../shared/vendor.js");
448    }
449
450    #[test]
451    fn skips_preload_link() {
452        let info = parse_html_to_module(
453            FileId(0),
454            r#"<link rel="preload" href="./src/font.woff2" as="font" />"#,
455            0,
456        );
457        assert!(info.imports.is_empty());
458    }
459
460    #[test]
461    fn skips_icon_link() {
462        let info =
463            parse_html_to_module(FileId(0), r#"<link rel="icon" href="./favicon.ico" />"#, 0);
464        assert!(info.imports.is_empty());
465    }
466
467    #[test]
468    fn skips_remote_stylesheet() {
469        let info = parse_html_to_module(
470            FileId(0),
471            r#"<link rel="stylesheet" href="https://fonts.googleapis.com/css" />"#,
472            0,
473        );
474        assert!(info.imports.is_empty());
475    }
476
477    // ── HTML comment stripping ───────────────────────────────────
478
479    #[test]
480    fn skips_commented_out_script() {
481        let info = parse_html_to_module(
482            FileId(0),
483            r#"<!-- <script src="./old.js"></script> -->
484            <script src="./new.js"></script>"#,
485            0,
486        );
487        assert_eq!(info.imports.len(), 1);
488        assert_eq!(info.imports[0].source, "./new.js");
489    }
490
491    #[test]
492    fn skips_commented_out_link() {
493        let info = parse_html_to_module(
494            FileId(0),
495            r#"<!-- <link rel="stylesheet" href="./old.css" /> -->
496            <link rel="stylesheet" href="./new.css" />"#,
497            0,
498        );
499        assert_eq!(info.imports.len(), 1);
500        assert_eq!(info.imports[0].source, "./new.css");
501    }
502
503    // ── Multi-line attributes ────────────────────────────────────
504
505    #[test]
506    fn handles_multiline_script_tag() {
507        let info = parse_html_to_module(
508            FileId(0),
509            "<script\n  type=\"module\"\n  src=\"./src/entry.js\"\n></script>",
510            0,
511        );
512        assert_eq!(info.imports.len(), 1);
513        assert_eq!(info.imports[0].source, "./src/entry.js");
514    }
515
516    #[test]
517    fn handles_multiline_link_tag() {
518        let info = parse_html_to_module(
519            FileId(0),
520            "<link\n  rel=\"stylesheet\"\n  href=\"./src/global.css\"\n/>",
521            0,
522        );
523        assert_eq!(info.imports.len(), 1);
524        assert_eq!(info.imports[0].source, "./src/global.css");
525    }
526
527    // ── Full HTML document ───────────────────────────────────────
528
529    #[test]
530    fn full_vite_html() {
531        let info = parse_html_to_module(
532            FileId(0),
533            r#"<!doctype html>
534<html>
535  <head>
536    <link rel="stylesheet" href="./src/global.css" />
537    <link rel="icon" href="/favicon.ico" />
538  </head>
539  <body>
540    <div id="app"></div>
541    <script type="module" src="./src/entry.js"></script>
542  </body>
543</html>"#,
544            0,
545        );
546        assert_eq!(info.imports.len(), 2);
547        let sources: Vec<&str> = info.imports.iter().map(|i| i.source.as_str()).collect();
548        assert!(sources.contains(&"./src/global.css"));
549        assert!(sources.contains(&"./src/entry.js"));
550    }
551
552    // ── Edge cases ───────────────────────────────────────────────
553
554    #[test]
555    fn empty_html() {
556        let info = parse_html_to_module(FileId(0), "", 0);
557        assert!(info.imports.is_empty());
558    }
559
560    #[test]
561    fn html_with_no_assets() {
562        let info = parse_html_to_module(
563            FileId(0),
564            r"<!doctype html><html><body><h1>Hello</h1></body></html>",
565            0,
566        );
567        assert!(info.imports.is_empty());
568    }
569
570    #[test]
571    fn single_quoted_attributes() {
572        let info = parse_html_to_module(FileId(0), r"<script src='./src/entry.js'></script>", 0);
573        assert_eq!(info.imports.len(), 1);
574        assert_eq!(info.imports[0].source, "./src/entry.js");
575    }
576
577    #[test]
578    fn all_imports_are_side_effect() {
579        let info = parse_html_to_module(
580            FileId(0),
581            r#"<script src="./entry.js"></script>
582            <link rel="stylesheet" href="./style.css" />"#,
583            0,
584        );
585        for imp in &info.imports {
586            assert!(matches!(imp.imported_name, ImportedName::SideEffect));
587            assert!(imp.local_name.is_empty());
588            assert!(!imp.is_type_only);
589        }
590    }
591
592    #[test]
593    fn suppression_comments_extracted() {
594        let info = parse_html_to_module(
595            FileId(0),
596            "<!-- fallow-ignore-file -->\n<script src=\"./entry.js\"></script>",
597            0,
598        );
599        // HTML comments use <!-- --> not //, so suppression parsing
600        // from source text won't find standard JS-style comments.
601        // This is expected — HTML suppression is not supported.
602        assert_eq!(info.imports.len(), 1);
603    }
604
605    // ── Angular template scanning ──────────────────────────────
606
607    #[test]
608    fn angular_template_extracts_member_refs() {
609        let info = parse_html_to_module(
610            FileId(0),
611            "<h1>{{ title() }}</h1>\n\
612             <p [class.highlighted]=\"isHighlighted\">{{ greeting() }}</p>\n\
613             <button (click)=\"onButtonClick()\">Toggle</button>",
614            0,
615        );
616        let names: rustc_hash::FxHashSet<&str> = info
617            .member_accesses
618            .iter()
619            .filter(|a| a.object == ANGULAR_TPL_SENTINEL)
620            .map(|a| a.member.as_str())
621            .collect();
622        assert!(names.contains("title"), "should contain 'title'");
623        assert!(
624            names.contains("isHighlighted"),
625            "should contain 'isHighlighted'"
626        );
627        assert!(names.contains("greeting"), "should contain 'greeting'");
628        assert!(
629            names.contains("onButtonClick"),
630            "should contain 'onButtonClick'"
631        );
632    }
633
634    #[test]
635    fn plain_html_no_angular_refs() {
636        let info = parse_html_to_module(
637            FileId(0),
638            "<!doctype html><html><body><h1>Hello</h1></body></html>",
639            0,
640        );
641        assert!(info.member_accesses.is_empty());
642    }
643}
fallow_extract/html.rs

fallow_extract/
html.rs
