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 typed semantic facts.
9
10use std::path::Path;
11use std::sync::LazyLock;
12
13use oxc_span::Span;
14
15use crate::asset_url::normalize_asset_url;
16use crate::sfc_template::angular;
17use crate::{
18    AngularTemplateMemberAccessFact, ImportInfo, ImportedName, MemberAccess, ModuleInfo,
19    SemanticFact,
20};
21use fallow_types::discover::FileId;
22
23/// Regex to match HTML comments (`<!-- ... -->`) for stripping before extraction.
24static HTML_COMMENT_RE: LazyLock<regex::Regex> =
25    LazyLock::new(|| crate::static_regex(r"(?s)<!--.*?-->"));
26
27/// Regex to extract `src` attribute from `<script>` tags.
28/// Matches both `<script src="...">` and `<script type="module" src="...">`.
29/// Uses `(?s)` so `.` matches newlines (multi-line attributes).
30static SCRIPT_SRC_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
31    crate::static_regex(r#"(?si)<script\b(?:[^>"']|"[^"]*"|'[^']*')*?\bsrc\s*=\s*["']([^"']+)["']"#)
32});
33
34/// Regex to extract `href` attribute from `<link>` tags with `rel="stylesheet"` or
35/// `rel="modulepreload"`.
36/// Handles attributes in any order (rel before or after href).
37static LINK_HREF_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
38    crate::static_regex(
39        r#"(?si)<link\b(?:[^>"']|"[^"]*"|'[^']*')*?\brel\s*=\s*["'](stylesheet|modulepreload)["'](?:[^>"']|"[^"]*"|'[^']*')*?\bhref\s*=\s*["']([^"']+)["']"#,
40    )
41});
42
43/// Regex for the reverse attribute order: href before rel.
44static LINK_HREF_REVERSE_RE: LazyLock<regex::Regex> = LazyLock::new(|| {
45    crate::static_regex(
46        r#"(?si)<link\b(?:[^>"']|"[^"]*"|'[^']*')*?\bhref\s*=\s*["']([^"']+)["'](?:[^>"']|"[^"]*"|'[^']*')*?\brel\s*=\s*["'](stylesheet|modulepreload)["']"#,
47    )
48});
49
50/// Check if a path is an HTML file.
51pub(crate) fn is_html_file(path: &Path) -> bool {
52    path.extension()
53        .and_then(|e| e.to_str())
54        .is_some_and(|ext| ext == "html")
55}
56
57/// Returns true if an HTML asset reference is a remote URL that should be skipped.
58pub(crate) fn is_remote_url(src: &str) -> bool {
59    src.starts_with("http://")
60        || src.starts_with("https://")
61        || src.starts_with("//")
62        || src.starts_with("data:")
63}
64
65/// Build-time template placeholders that aren't valid import specifiers and
66/// never resolve to a real file. Skip them at extraction time so they don't
67/// enter the import graph as unresolvable specifiers.
68///
69/// - `{{ ... }}` covers Handlebars (Ember `index.html`'s `{{rootURL}}`,
70///   `{{config.assetsPath}}`), Mustache (Jekyll, Hugo), Jinja2 (Pelican /
71///   11ty plugins), and pre-compiled Vue / Angular templates whose
72///   interpolation has leaked into a checked-in HTML scaffold.
73/// - `###...###` covers ember-cli blueprint scaffold placeholders
74///   (`###APPNAME###`, `###DUMMY###`) checked in as addon-fixture templates.
75///
76/// Neither shape is a legal URL or path character outside template engines,
77/// so the skip is generic across frameworks rather than gated on a plugin.
78/// Returns `true` for any `src` / `href` value that contains either marker.
79pub(crate) fn is_template_placeholder(value: &str) -> bool {
80    value.contains("{{") || value.contains("###")
81}
82
83/// Extract local (non-remote) asset references from HTML-like markup.
84///
85/// Returns the raw `src`/`href` strings (trimmed, remote URLs filtered). Shared
86/// between the HTML file parser and the JS/TS visitor's tagged template
87/// literal override so `` html`<script src="...">` `` in Hono/lit-html/htm
88/// layouts emits the same asset edges as a real `.html` file.
89pub(crate) fn collect_asset_refs(source: &str) -> Vec<String> {
90    let stripped = HTML_COMMENT_RE.replace_all(source, "");
91    let mut refs: Vec<String> = Vec::new();
92
93    for cap in SCRIPT_SRC_RE.captures_iter(&stripped) {
94        if let Some(m) = cap.get(1) {
95            let src = m.as_str().trim();
96            if !src.is_empty() && !is_remote_url(src) && !is_template_placeholder(src) {
97                refs.push(src.to_string());
98            }
99        }
100    }
101
102    for cap in LINK_HREF_RE.captures_iter(&stripped) {
103        if let Some(m) = cap.get(2) {
104            let href = m.as_str().trim();
105            if !href.is_empty() && !is_remote_url(href) && !is_template_placeholder(href) {
106                refs.push(href.to_string());
107            }
108        }
109    }
110    for cap in LINK_HREF_REVERSE_RE.captures_iter(&stripped) {
111        if let Some(m) = cap.get(1) {
112            let href = m.as_str().trim();
113            if !href.is_empty() && !is_remote_url(href) && !is_template_placeholder(href) {
114                refs.push(href.to_string());
115            }
116        }
117    }
118
119    refs
120}
121
122/// Regex matching an opening or closing custom-element tag. The HTML spec
123/// requires a custom-element name to contain a hyphen, so `[a-z][a-z0-9]*-...`
124/// captures `<x-foo>` / `<my-element>` while native tags (`div`, `span`) never
125/// match. The capture stops before attributes / `>` / `/`.
126static CUSTOM_ELEMENT_TAG_RE: std::sync::LazyLock<regex::Regex> =
127    std::sync::LazyLock::new(|| crate::static_regex(r"</?\s*([a-z][a-z0-9]*-[a-z0-9-]*)"));
128
129/// Collect the custom-element tag names rendered in an `html` template snippet
130/// (`<x-foo>` / `</x-foo>` -> `x-foo`). HTML comments are stripped first so a
131/// commented-out `<!-- <x-foo> -->` does not credit the element. Deduped; native
132/// HTML tags are excluded by the hyphen requirement. Feeds the Lit
133/// `unrendered-component` arm's project-wide rendered-tag union.
134pub(crate) fn collect_custom_element_tags(source: &str) -> Vec<String> {
135    let stripped = HTML_COMMENT_RE.replace_all(source, "");
136    let mut tags: Vec<String> = Vec::new();
137    for cap in CUSTOM_ELEMENT_TAG_RE.captures_iter(&stripped) {
138        if let Some(m) = cap.get(1) {
139            let tag = m.as_str();
140            if !tags.iter().any(|t| t == tag) {
141                tags.push(tag.to_string());
142            }
143        }
144    }
145    tags
146}
147
148/// Parse an HTML file, extracting script and stylesheet references as imports.
149#[cfg(test)]
150pub(crate) fn parse_html_to_module(file_id: FileId, source: &str, content_hash: u64) -> ModuleInfo {
151    parse_html_to_module_with_complexity(file_id, source, content_hash, false)
152}
153
154/// Computed building blocks for an HTML [`ModuleInfo`], gathered before the
155/// (irreducible) struct literal is assembled.
156struct HtmlModuleParts {
157    imports: Vec<ImportInfo>,
158    member_accesses: Vec<MemberAccess>,
159    semantic_facts: Vec<SemanticFact>,
160    security_sinks: Vec<fallow_types::extract::SinkSite>,
161    angular_used_selectors: Vec<String>,
162    has_dynamic_component_render: bool,
163    complexity: Vec<fallow_types::extract::FunctionComplexity>,
164}
165
166/// Collect the asset-reference imports, Angular template member accesses /
167/// security sinks / used selectors, and (optionally) template complexity for an
168/// HTML source.
169fn collect_html_module_parts(source: &str, need_complexity: bool) -> HtmlModuleParts {
170    let mut imports: Vec<ImportInfo> = collect_asset_refs(source)
171        .into_iter()
172        .map(|raw| ImportInfo {
173            source: normalize_asset_url(&raw),
174            imported_name: ImportedName::SideEffect,
175            local_name: String::new(),
176            is_type_only: false,
177            from_style: false,
178            span: Span::default(),
179            source_span: Span::default(),
180        })
181        .collect();
182
183    imports.sort_unstable_by(|a, b| a.source.cmp(&b.source));
184    imports.dedup_by(|a, b| a.source == b.source);
185
186    let angular::AngularTemplateRefs {
187        identifiers,
188        member_accesses: template_member_accesses,
189        security_sinks,
190    } = angular::collect_angular_template_refs(source);
191    let identifiers: Vec<String> = identifiers.into_iter().collect();
192    let semantic_facts: Vec<SemanticFact> = identifiers
193        .iter()
194        .cloned()
195        .map(|member| {
196            SemanticFact::AngularTemplateMemberAccess(AngularTemplateMemberAccessFact { member })
197        })
198        .collect();
199    let member_accesses = template_member_accesses;
200
201    // Angular external template (`templateUrl`): harvest the custom element
202    // selector tags rendered here so the Angular `unrendered-component` detector
203    // unions them into the project-wide used-selector set, and flag the
204    // `*ngComponentOutlet` dynamic-render escape hatch (project-wide abstain).
205    let angular_used_selectors = angular::collect_angular_used_selectors(source);
206    let has_dynamic_component_render = source.contains("ngComponentOutlet");
207
208    let complexity = if need_complexity {
209        crate::template_complexity::compute_angular_template_complexity(source)
210            .into_iter()
211            .collect()
212    } else {
213        Vec::new()
214    };
215
216    HtmlModuleParts {
217        imports,
218        member_accesses,
219        semantic_facts,
220        security_sinks,
221        angular_used_selectors,
222        has_dynamic_component_render,
223        complexity,
224    }
225}
226
227/// Parse an HTML file and optionally compute Angular template complexity.
228pub(crate) fn parse_html_to_module_with_complexity(
229    file_id: FileId,
230    source: &str,
231    content_hash: u64,
232    need_complexity: bool,
233) -> ModuleInfo {
234    let parsed_suppressions = crate::suppress::parse_suppressions_from_source(source);
235    let parts = collect_html_module_parts(source, need_complexity);
236    html_module_info(file_id, content_hash, source, parsed_suppressions, parts)
237}
238
239/// Assemble the `ModuleInfo` for an HTML file from its computed parts; all
240/// JS-level fields stay empty since HTML carries no module structure. Pure
241/// plumbing struct literal.
242fn html_module_info(
243    file_id: FileId,
244    content_hash: u64,
245    source: &str,
246    parsed_suppressions: crate::suppress::ParsedSuppressions,
247    parts: HtmlModuleParts,
248) -> ModuleInfo {
249    let HtmlModuleParts {
250        imports,
251        member_accesses,
252        semantic_facts,
253        security_sinks,
254        angular_used_selectors,
255        has_dynamic_component_render,
256        complexity,
257    } = parts;
258
259    ModuleInfo {
260        file_id,
261        exports: Vec::new(),
262        imports,
263        re_exports: Vec::new(),
264        dynamic_imports: Vec::new(),
265        dynamic_import_patterns: Vec::new(),
266        require_calls: Vec::new(),
267        package_path_references: Box::default(),
268        member_accesses,
269        semantic_facts: semantic_facts.into(),
270        whole_object_uses: Box::default(),
271        has_cjs_exports: false,
272        has_angular_component_template_url: false,
273        content_hash,
274        suppressions: parsed_suppressions.suppressions,
275        unknown_suppression_kinds: parsed_suppressions.unknown_kinds,
276        unused_import_bindings: Vec::new(),
277        type_referenced_import_bindings: Vec::new(),
278        value_referenced_import_bindings: Vec::new(),
279        line_offsets: fallow_types::extract::compute_line_offsets(source),
280        complexity,
281        flag_uses: Vec::new(),
282        class_heritage: vec![],
283        exported_factory_returns: Box::default(),
284        type_member_types: Box::default(),
285        injection_tokens: vec![],
286        local_type_declarations: Vec::new(),
287        public_signature_type_references: Vec::new(),
288        namespace_object_aliases: Vec::new(),
289        iconify_prefixes: Vec::new(),
290        iconify_icon_names: Vec::new(),
291        auto_import_candidates: Vec::new(),
292        directives: Vec::new(),
293        client_only_dynamic_import_spans: Vec::new(),
294        security_sinks,
295        security_sinks_skipped: 0,
296        security_unresolved_callee_sites: Vec::new(),
297        tainted_bindings: Vec::new(),
298        sanitized_sink_args: Vec::new(),
299        security_control_sites: Vec::new(),
300        callee_uses: Vec::new(),
301        misplaced_directives: Vec::new(),
302        inline_server_action_exports: Vec::new(),
303        di_key_sites: Vec::new(),
304        has_dynamic_provide: false,
305        referenced_import_bindings: Vec::new(),
306        component_props: Vec::new(),
307        has_props_attrs_fallthrough: false,
308        has_define_expose: false,
309        has_define_model: false,
310        has_unharvestable_props: false,
311        component_emits: Vec::new(),
312        angular_inputs: Vec::new(),
313        angular_outputs: Vec::new(),
314        angular_component_selectors: Vec::new(),
315        registered_custom_elements: Vec::new(),
316        // Custom-element tags rendered in a standalone `.html` document (an app
317        // shell, demo, or dev page) feed the Lit `unrendered-component` arm's
318        // project-wide rendered-tag union, so an element rendered only from HTML
319        // (e.g. a root `<my-app>` in `index.html`) is not falsely flagged.
320        used_custom_element_tags: collect_custom_element_tags(source),
321        angular_used_selectors,
322        angular_entry_component_refs: Vec::new(),
323        has_dynamic_component_render,
324        has_unharvestable_emits: false,
325        has_dynamic_emit: false,
326        has_emit_whole_object_use: false,
327        load_return_keys: Vec::new(),
328        has_unharvestable_load: false,
329        has_load_data_whole_use: false,
330        has_page_data_store_whole_use: false,
331        has_route_loader_data_whole_use: false,
332        component_functions: Vec::new(),
333        react_props: Vec::new(),
334        hook_uses: Vec::new(),
335        render_edges: Vec::new(),
336        svelte_dispatched_events: Vec::new(),
337        svelte_listened_events: Vec::new(),
338        has_dynamic_dispatch: false,
339    }
340}
341
342#[cfg(test)]
343mod tests {
344    use super::*;
345
346    #[test]
347    fn is_html_file_html() {
348        assert!(is_html_file(Path::new("index.html")));
349    }
350
351    #[test]
352    fn is_html_file_nested() {
353        assert!(is_html_file(Path::new("pages/about.html")));
354    }
355
356    #[test]
357    fn is_html_file_rejects_htm() {
358        assert!(!is_html_file(Path::new("index.htm")));
359    }
360
361    #[test]
362    fn is_html_file_rejects_js() {
363        assert!(!is_html_file(Path::new("app.js")));
364    }
365
366    #[test]
367    fn is_html_file_rejects_ts() {
368        assert!(!is_html_file(Path::new("app.ts")));
369    }
370
371    #[test]
372    fn is_html_file_rejects_vue() {
373        assert!(!is_html_file(Path::new("App.vue")));
374    }
375
376    #[test]
377    fn remote_url_http() {
378        assert!(is_remote_url("http://example.com/script.js"));
379    }
380
381    #[test]
382    fn remote_url_https() {
383        assert!(is_remote_url("https://cdn.example.com/style.css"));
384    }
385
386    #[test]
387    fn remote_url_protocol_relative() {
388        assert!(is_remote_url("//cdn.example.com/lib.js"));
389    }
390
391    #[test]
392    fn remote_url_data() {
393        assert!(is_remote_url("data:text/javascript;base64,abc"));
394    }
395
396    #[test]
397    fn local_relative_not_remote() {
398        assert!(!is_remote_url("./src/entry.js"));
399    }
400
401    #[test]
402    fn local_root_relative_not_remote() {
403        assert!(!is_remote_url("/src/entry.js"));
404    }
405
406    #[test]
407    fn extracts_module_script_src() {
408        let info = parse_html_to_module(
409            FileId(0),
410            r#"<script type="module" src="./src/entry.js"></script>"#,
411            0,
412        );
413        assert_eq!(info.imports.len(), 1);
414        assert_eq!(info.imports[0].source, "./src/entry.js");
415    }
416
417    #[test]
418    fn extracts_plain_script_src() {
419        let info = parse_html_to_module(
420            FileId(0),
421            r#"<script src="./src/polyfills.js"></script>"#,
422            0,
423        );
424        assert_eq!(info.imports.len(), 1);
425        assert_eq!(info.imports[0].source, "./src/polyfills.js");
426    }
427
428    #[test]
429    fn extracts_multiple_scripts() {
430        let info = parse_html_to_module(
431            FileId(0),
432            r#"
433            <script type="module" src="./src/entry.js"></script>
434            <script src="./src/polyfills.js"></script>
435            "#,
436            0,
437        );
438        assert_eq!(info.imports.len(), 2);
439    }
440
441    #[test]
442    fn skips_inline_script() {
443        let info = parse_html_to_module(FileId(0), r#"<script>console.log("hello");</script>"#, 0);
444        assert!(info.imports.is_empty());
445    }
446
447    #[test]
448    fn skips_handlebars_placeholder_in_script_src() {
449        let info = parse_html_to_module(
450            FileId(0),
451            r#"<script src="{{rootURL}}assets/app.js"></script>
452               <script src="{{config.assetsPath}}vendor.js"></script>"#,
453            0,
454        );
455        assert!(
456            info.imports.is_empty(),
457            "Handlebars-placeholder script srcs should not enter the import graph; got {:?}",
458            info.imports
459        );
460    }
461
462    #[test]
463    fn skips_handlebars_placeholder_in_link_href() {
464        let info = parse_html_to_module(
465            FileId(0),
466            r#"<link rel="stylesheet" href="{{rootURL}}assets/app.css">"#,
467            0,
468        );
469        assert!(info.imports.is_empty());
470    }
471
472    #[test]
473    fn skips_ember_cli_blueprint_placeholder() {
474        let info = parse_html_to_module(
475            FileId(0),
476            r####"<script src="###APPNAME###/app.js"></script>"####,
477            0,
478        );
479        assert!(info.imports.is_empty());
480    }
481
482    #[test]
483    fn extracts_normal_specifier_alongside_placeholders() {
484        let info = parse_html_to_module(
485            FileId(0),
486            r#"<script src="{{rootURL}}assets/app.js"></script>
487               <script src="./src/main.ts"></script>"#,
488            0,
489        );
490        assert_eq!(info.imports.len(), 1);
491        assert_eq!(info.imports[0].source, "./src/main.ts");
492    }
493
494    #[test]
495    fn skips_remote_script() {
496        let info = parse_html_to_module(
497            FileId(0),
498            r#"<script src="https://cdn.example.com/lib.js"></script>"#,
499            0,
500        );
501        assert!(info.imports.is_empty());
502    }
503
504    #[test]
505    fn skips_protocol_relative_script() {
506        let info = parse_html_to_module(
507            FileId(0),
508            r#"<script src="//cdn.example.com/lib.js"></script>"#,
509            0,
510        );
511        assert!(info.imports.is_empty());
512    }
513
514    #[test]
515    fn extracts_stylesheet_link() {
516        let info = parse_html_to_module(
517            FileId(0),
518            r#"<link rel="stylesheet" href="./src/global.css" />"#,
519            0,
520        );
521        assert_eq!(info.imports.len(), 1);
522        assert_eq!(info.imports[0].source, "./src/global.css");
523    }
524
525    #[test]
526    fn extracts_modulepreload_link() {
527        let info = parse_html_to_module(
528            FileId(0),
529            r#"<link rel="modulepreload" href="./src/vendor.js" />"#,
530            0,
531        );
532        assert_eq!(info.imports.len(), 1);
533        assert_eq!(info.imports[0].source, "./src/vendor.js");
534    }
535
536    #[test]
537    fn extracts_link_with_reversed_attrs() {
538        let info = parse_html_to_module(
539            FileId(0),
540            r#"<link href="./src/global.css" rel="stylesheet" />"#,
541            0,
542        );
543        assert_eq!(info.imports.len(), 1);
544        assert_eq!(info.imports[0].source, "./src/global.css");
545    }
546
547    #[test]
548    fn bare_script_src_normalized_to_relative() {
549        let info = parse_html_to_module(FileId(0), r#"<script src="app.js"></script>"#, 0);
550        assert_eq!(info.imports.len(), 1);
551        assert_eq!(info.imports[0].source, "./app.js");
552    }
553
554    #[test]
555    fn bare_module_script_src_normalized_to_relative() {
556        let info = parse_html_to_module(
557            FileId(0),
558            r#"<script type="module" src="main.ts"></script>"#,
559            0,
560        );
561        assert_eq!(info.imports.len(), 1);
562        assert_eq!(info.imports[0].source, "./main.ts");
563    }
564
565    #[test]
566    fn bare_stylesheet_link_href_normalized_to_relative() {
567        let info = parse_html_to_module(
568            FileId(0),
569            r#"<link rel="stylesheet" href="styles.css" />"#,
570            0,
571        );
572        assert_eq!(info.imports.len(), 1);
573        assert_eq!(info.imports[0].source, "./styles.css");
574    }
575
576    #[test]
577    fn bare_link_href_reversed_attrs_normalized_to_relative() {
578        let info = parse_html_to_module(
579            FileId(0),
580            r#"<link href="styles.css" rel="stylesheet" />"#,
581            0,
582        );
583        assert_eq!(info.imports.len(), 1);
584        assert_eq!(info.imports[0].source, "./styles.css");
585    }
586
587    #[test]
588    fn bare_modulepreload_link_href_normalized_to_relative() {
589        let info = parse_html_to_module(
590            FileId(0),
591            r#"<link rel="modulepreload" href="vendor.js" />"#,
592            0,
593        );
594        assert_eq!(info.imports.len(), 1);
595        assert_eq!(info.imports[0].source, "./vendor.js");
596    }
597
598    #[test]
599    fn bare_asset_with_subdir_normalized_to_relative() {
600        let info = parse_html_to_module(FileId(0), r#"<script src="assets/app.js"></script>"#, 0);
601        assert_eq!(info.imports.len(), 1);
602        assert_eq!(info.imports[0].source, "./assets/app.js");
603    }
604
605    #[test]
606    fn root_absolute_script_src_unchanged() {
607        let info = parse_html_to_module(FileId(0), r#"<script src="/src/main.ts"></script>"#, 0);
608        assert_eq!(info.imports.len(), 1);
609        assert_eq!(info.imports[0].source, "/src/main.ts");
610    }
611
612    #[test]
613    fn parent_relative_script_src_unchanged() {
614        let info = parse_html_to_module(
615            FileId(0),
616            r#"<script src="../shared/vendor.js"></script>"#,
617            0,
618        );
619        assert_eq!(info.imports.len(), 1);
620        assert_eq!(info.imports[0].source, "../shared/vendor.js");
621    }
622
623    #[test]
624    fn skips_preload_link() {
625        let info = parse_html_to_module(
626            FileId(0),
627            r#"<link rel="preload" href="./src/font.woff2" as="font" />"#,
628            0,
629        );
630        assert!(info.imports.is_empty());
631    }
632
633    #[test]
634    fn skips_icon_link() {
635        let info =
636            parse_html_to_module(FileId(0), r#"<link rel="icon" href="./favicon.ico" />"#, 0);
637        assert!(info.imports.is_empty());
638    }
639
640    #[test]
641    fn skips_remote_stylesheet() {
642        let info = parse_html_to_module(
643            FileId(0),
644            r#"<link rel="stylesheet" href="https://fonts.googleapis.com/css" />"#,
645            0,
646        );
647        assert!(info.imports.is_empty());
648    }
649
650    #[test]
651    fn skips_commented_out_script() {
652        let info = parse_html_to_module(
653            FileId(0),
654            r#"<!-- <script src="./old.js"></script> -->
655            <script src="./new.js"></script>"#,
656            0,
657        );
658        assert_eq!(info.imports.len(), 1);
659        assert_eq!(info.imports[0].source, "./new.js");
660    }
661
662    #[test]
663    fn skips_commented_out_link() {
664        let info = parse_html_to_module(
665            FileId(0),
666            r#"<!-- <link rel="stylesheet" href="./old.css" /> -->
667            <link rel="stylesheet" href="./new.css" />"#,
668            0,
669        );
670        assert_eq!(info.imports.len(), 1);
671        assert_eq!(info.imports[0].source, "./new.css");
672    }
673
674    #[test]
675    fn handles_multiline_script_tag() {
676        let info = parse_html_to_module(
677            FileId(0),
678            "<script\n  type=\"module\"\n  src=\"./src/entry.js\"\n></script>",
679            0,
680        );
681        assert_eq!(info.imports.len(), 1);
682        assert_eq!(info.imports[0].source, "./src/entry.js");
683    }
684
685    #[test]
686    fn handles_multiline_link_tag() {
687        let info = parse_html_to_module(
688            FileId(0),
689            "<link\n  rel=\"stylesheet\"\n  href=\"./src/global.css\"\n/>",
690            0,
691        );
692        assert_eq!(info.imports.len(), 1);
693        assert_eq!(info.imports[0].source, "./src/global.css");
694    }
695
696    #[test]
697    fn full_vite_html() {
698        let info = parse_html_to_module(
699            FileId(0),
700            r#"<!doctype html>
701<html>
702  <head>
703    <link rel="stylesheet" href="./src/global.css" />
704    <link rel="icon" href="/favicon.ico" />
705  </head>
706  <body>
707    <div id="app"></div>
708    <script type="module" src="./src/entry.js"></script>
709  </body>
710</html>"#,
711            0,
712        );
713        assert_eq!(info.imports.len(), 2);
714        let sources: Vec<&str> = info.imports.iter().map(|i| i.source.as_str()).collect();
715        assert!(sources.contains(&"./src/global.css"));
716        assert!(sources.contains(&"./src/entry.js"));
717    }
718
719    #[test]
720    fn empty_html() {
721        let info = parse_html_to_module(FileId(0), "", 0);
722        assert!(info.imports.is_empty());
723    }
724
725    #[test]
726    fn html_with_no_assets() {
727        let info = parse_html_to_module(
728            FileId(0),
729            r"<!doctype html><html><body><h1>Hello</h1></body></html>",
730            0,
731        );
732        assert!(info.imports.is_empty());
733    }
734
735    #[test]
736    fn single_quoted_attributes() {
737        let info = parse_html_to_module(FileId(0), r"<script src='./src/entry.js'></script>", 0);
738        assert_eq!(info.imports.len(), 1);
739        assert_eq!(info.imports[0].source, "./src/entry.js");
740    }
741
742    #[test]
743    fn all_imports_are_side_effect() {
744        let info = parse_html_to_module(
745            FileId(0),
746            r#"<script src="./entry.js"></script>
747            <link rel="stylesheet" href="./style.css" />"#,
748            0,
749        );
750        for imp in &info.imports {
751            assert!(matches!(imp.imported_name, ImportedName::SideEffect));
752            assert!(imp.local_name.is_empty());
753            assert!(!imp.is_type_only);
754        }
755    }
756
757    #[test]
758    fn suppression_comments_extracted() {
759        let info = parse_html_to_module(
760            FileId(0),
761            "<!-- fallow-ignore-file -->\n<script src=\"./entry.js\"></script>",
762            0,
763        );
764        assert_eq!(info.imports.len(), 1);
765    }
766
767    #[test]
768    fn angular_template_extracts_member_refs() {
769        let info = parse_html_to_module(
770            FileId(0),
771            "<h1>{{ title() }}</h1>\n\
772             <p [class.highlighted]=\"isHighlighted\">{{ greeting() }}</p>\n\
773             <button (click)=\"onButtonClick()\">Toggle</button>",
774            0,
775        );
776        let fact_names: rustc_hash::FxHashSet<&str> = info
777            .semantic_facts
778            .iter()
779            .filter_map(|fact| {
780                if let SemanticFact::AngularTemplateMemberAccess(access) = fact {
781                    Some(access.member.as_str())
782                } else {
783                    None
784                }
785            })
786            .collect();
787        assert!(fact_names.contains("title"), "should contain 'title'");
788        assert!(
789            fact_names.contains("isHighlighted"),
790            "should contain 'isHighlighted'"
791        );
792        assert!(fact_names.contains("greeting"), "should contain 'greeting'");
793        assert!(
794            fact_names.contains("onButtonClick"),
795            "should contain 'onButtonClick'"
796        );
797        assert!(
798            info.member_accesses.is_empty(),
799            "Angular template refs should emit typed facts instead of member accesses: {:?}",
800            info.member_accesses
801        );
802    }
803
804    #[test]
805    fn plain_html_no_angular_refs() {
806        let info = parse_html_to_module(
807            FileId(0),
808            "<!doctype html><html><body><h1>Hello</h1></body></html>",
809            0,
810        );
811        assert!(info.member_accesses.is_empty());
812    }
813}
fallow_extract/html.rs

fallow_extract/
html.rs
