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        injection_tokens: vec![],
285        local_type_declarations: Vec::new(),
286        public_signature_type_references: Vec::new(),
287        namespace_object_aliases: Vec::new(),
288        iconify_prefixes: Vec::new(),
289        iconify_icon_names: Vec::new(),
290        auto_import_candidates: Vec::new(),
291        directives: Vec::new(),
292        client_only_dynamic_import_spans: Vec::new(),
293        security_sinks,
294        security_sinks_skipped: 0,
295        security_unresolved_callee_sites: Vec::new(),
296        tainted_bindings: Vec::new(),
297        sanitized_sink_args: Vec::new(),
298        security_control_sites: Vec::new(),
299        callee_uses: Vec::new(),
300        misplaced_directives: Vec::new(),
301        inline_server_action_exports: Vec::new(),
302        di_key_sites: Vec::new(),
303        has_dynamic_provide: false,
304        referenced_import_bindings: Vec::new(),
305        component_props: Vec::new(),
306        has_props_attrs_fallthrough: false,
307        has_define_expose: false,
308        has_define_model: false,
309        has_unharvestable_props: false,
310        component_emits: Vec::new(),
311        angular_inputs: Vec::new(),
312        angular_outputs: Vec::new(),
313        angular_component_selectors: Vec::new(),
314        registered_custom_elements: Vec::new(),
315        // Custom-element tags rendered in a standalone `.html` document (an app
316        // shell, demo, or dev page) feed the Lit `unrendered-component` arm's
317        // project-wide rendered-tag union, so an element rendered only from HTML
318        // (e.g. a root `<my-app>` in `index.html`) is not falsely flagged.
319        used_custom_element_tags: collect_custom_element_tags(source),
320        angular_used_selectors,
321        angular_entry_component_refs: Vec::new(),
322        has_dynamic_component_render,
323        has_unharvestable_emits: false,
324        has_dynamic_emit: false,
325        has_emit_whole_object_use: false,
326        load_return_keys: Vec::new(),
327        has_unharvestable_load: false,
328        has_load_data_whole_use: false,
329        has_page_data_store_whole_use: false,
330        component_functions: Vec::new(),
331        react_props: Vec::new(),
332        hook_uses: Vec::new(),
333        render_edges: Vec::new(),
334        svelte_dispatched_events: Vec::new(),
335        svelte_listened_events: Vec::new(),
336        has_dynamic_dispatch: false,
337    }
338}
339
340#[cfg(test)]
341mod tests {
342    use super::*;
343
344    #[test]
345    fn is_html_file_html() {
346        assert!(is_html_file(Path::new("index.html")));
347    }
348
349    #[test]
350    fn is_html_file_nested() {
351        assert!(is_html_file(Path::new("pages/about.html")));
352    }
353
354    #[test]
355    fn is_html_file_rejects_htm() {
356        assert!(!is_html_file(Path::new("index.htm")));
357    }
358
359    #[test]
360    fn is_html_file_rejects_js() {
361        assert!(!is_html_file(Path::new("app.js")));
362    }
363
364    #[test]
365    fn is_html_file_rejects_ts() {
366        assert!(!is_html_file(Path::new("app.ts")));
367    }
368
369    #[test]
370    fn is_html_file_rejects_vue() {
371        assert!(!is_html_file(Path::new("App.vue")));
372    }
373
374    #[test]
375    fn remote_url_http() {
376        assert!(is_remote_url("http://example.com/script.js"));
377    }
378
379    #[test]
380    fn remote_url_https() {
381        assert!(is_remote_url("https://cdn.example.com/style.css"));
382    }
383
384    #[test]
385    fn remote_url_protocol_relative() {
386        assert!(is_remote_url("//cdn.example.com/lib.js"));
387    }
388
389    #[test]
390    fn remote_url_data() {
391        assert!(is_remote_url("data:text/javascript;base64,abc"));
392    }
393
394    #[test]
395    fn local_relative_not_remote() {
396        assert!(!is_remote_url("./src/entry.js"));
397    }
398
399    #[test]
400    fn local_root_relative_not_remote() {
401        assert!(!is_remote_url("/src/entry.js"));
402    }
403
404    #[test]
405    fn extracts_module_script_src() {
406        let info = parse_html_to_module(
407            FileId(0),
408            r#"<script type="module" src="./src/entry.js"></script>"#,
409            0,
410        );
411        assert_eq!(info.imports.len(), 1);
412        assert_eq!(info.imports[0].source, "./src/entry.js");
413    }
414
415    #[test]
416    fn extracts_plain_script_src() {
417        let info = parse_html_to_module(
418            FileId(0),
419            r#"<script src="./src/polyfills.js"></script>"#,
420            0,
421        );
422        assert_eq!(info.imports.len(), 1);
423        assert_eq!(info.imports[0].source, "./src/polyfills.js");
424    }
425
426    #[test]
427    fn extracts_multiple_scripts() {
428        let info = parse_html_to_module(
429            FileId(0),
430            r#"
431            <script type="module" src="./src/entry.js"></script>
432            <script src="./src/polyfills.js"></script>
433            "#,
434            0,
435        );
436        assert_eq!(info.imports.len(), 2);
437    }
438
439    #[test]
440    fn skips_inline_script() {
441        let info = parse_html_to_module(FileId(0), r#"<script>console.log("hello");</script>"#, 0);
442        assert!(info.imports.is_empty());
443    }
444
445    #[test]
446    fn skips_handlebars_placeholder_in_script_src() {
447        let info = parse_html_to_module(
448            FileId(0),
449            r#"<script src="{{rootURL}}assets/app.js"></script>
450               <script src="{{config.assetsPath}}vendor.js"></script>"#,
451            0,
452        );
453        assert!(
454            info.imports.is_empty(),
455            "Handlebars-placeholder script srcs should not enter the import graph; got {:?}",
456            info.imports
457        );
458    }
459
460    #[test]
461    fn skips_handlebars_placeholder_in_link_href() {
462        let info = parse_html_to_module(
463            FileId(0),
464            r#"<link rel="stylesheet" href="{{rootURL}}assets/app.css">"#,
465            0,
466        );
467        assert!(info.imports.is_empty());
468    }
469
470    #[test]
471    fn skips_ember_cli_blueprint_placeholder() {
472        let info = parse_html_to_module(
473            FileId(0),
474            r####"<script src="###APPNAME###/app.js"></script>"####,
475            0,
476        );
477        assert!(info.imports.is_empty());
478    }
479
480    #[test]
481    fn extracts_normal_specifier_alongside_placeholders() {
482        let info = parse_html_to_module(
483            FileId(0),
484            r#"<script src="{{rootURL}}assets/app.js"></script>
485               <script src="./src/main.ts"></script>"#,
486            0,
487        );
488        assert_eq!(info.imports.len(), 1);
489        assert_eq!(info.imports[0].source, "./src/main.ts");
490    }
491
492    #[test]
493    fn skips_remote_script() {
494        let info = parse_html_to_module(
495            FileId(0),
496            r#"<script src="https://cdn.example.com/lib.js"></script>"#,
497            0,
498        );
499        assert!(info.imports.is_empty());
500    }
501
502    #[test]
503    fn skips_protocol_relative_script() {
504        let info = parse_html_to_module(
505            FileId(0),
506            r#"<script src="//cdn.example.com/lib.js"></script>"#,
507            0,
508        );
509        assert!(info.imports.is_empty());
510    }
511
512    #[test]
513    fn extracts_stylesheet_link() {
514        let info = parse_html_to_module(
515            FileId(0),
516            r#"<link rel="stylesheet" href="./src/global.css" />"#,
517            0,
518        );
519        assert_eq!(info.imports.len(), 1);
520        assert_eq!(info.imports[0].source, "./src/global.css");
521    }
522
523    #[test]
524    fn extracts_modulepreload_link() {
525        let info = parse_html_to_module(
526            FileId(0),
527            r#"<link rel="modulepreload" href="./src/vendor.js" />"#,
528            0,
529        );
530        assert_eq!(info.imports.len(), 1);
531        assert_eq!(info.imports[0].source, "./src/vendor.js");
532    }
533
534    #[test]
535    fn extracts_link_with_reversed_attrs() {
536        let info = parse_html_to_module(
537            FileId(0),
538            r#"<link href="./src/global.css" rel="stylesheet" />"#,
539            0,
540        );
541        assert_eq!(info.imports.len(), 1);
542        assert_eq!(info.imports[0].source, "./src/global.css");
543    }
544
545    #[test]
546    fn bare_script_src_normalized_to_relative() {
547        let info = parse_html_to_module(FileId(0), r#"<script src="app.js"></script>"#, 0);
548        assert_eq!(info.imports.len(), 1);
549        assert_eq!(info.imports[0].source, "./app.js");
550    }
551
552    #[test]
553    fn bare_module_script_src_normalized_to_relative() {
554        let info = parse_html_to_module(
555            FileId(0),
556            r#"<script type="module" src="main.ts"></script>"#,
557            0,
558        );
559        assert_eq!(info.imports.len(), 1);
560        assert_eq!(info.imports[0].source, "./main.ts");
561    }
562
563    #[test]
564    fn bare_stylesheet_link_href_normalized_to_relative() {
565        let info = parse_html_to_module(
566            FileId(0),
567            r#"<link rel="stylesheet" href="styles.css" />"#,
568            0,
569        );
570        assert_eq!(info.imports.len(), 1);
571        assert_eq!(info.imports[0].source, "./styles.css");
572    }
573
574    #[test]
575    fn bare_link_href_reversed_attrs_normalized_to_relative() {
576        let info = parse_html_to_module(
577            FileId(0),
578            r#"<link href="styles.css" rel="stylesheet" />"#,
579            0,
580        );
581        assert_eq!(info.imports.len(), 1);
582        assert_eq!(info.imports[0].source, "./styles.css");
583    }
584
585    #[test]
586    fn bare_modulepreload_link_href_normalized_to_relative() {
587        let info = parse_html_to_module(
588            FileId(0),
589            r#"<link rel="modulepreload" href="vendor.js" />"#,
590            0,
591        );
592        assert_eq!(info.imports.len(), 1);
593        assert_eq!(info.imports[0].source, "./vendor.js");
594    }
595
596    #[test]
597    fn bare_asset_with_subdir_normalized_to_relative() {
598        let info = parse_html_to_module(FileId(0), r#"<script src="assets/app.js"></script>"#, 0);
599        assert_eq!(info.imports.len(), 1);
600        assert_eq!(info.imports[0].source, "./assets/app.js");
601    }
602
603    #[test]
604    fn root_absolute_script_src_unchanged() {
605        let info = parse_html_to_module(FileId(0), r#"<script src="/src/main.ts"></script>"#, 0);
606        assert_eq!(info.imports.len(), 1);
607        assert_eq!(info.imports[0].source, "/src/main.ts");
608    }
609
610    #[test]
611    fn parent_relative_script_src_unchanged() {
612        let info = parse_html_to_module(
613            FileId(0),
614            r#"<script src="../shared/vendor.js"></script>"#,
615            0,
616        );
617        assert_eq!(info.imports.len(), 1);
618        assert_eq!(info.imports[0].source, "../shared/vendor.js");
619    }
620
621    #[test]
622    fn skips_preload_link() {
623        let info = parse_html_to_module(
624            FileId(0),
625            r#"<link rel="preload" href="./src/font.woff2" as="font" />"#,
626            0,
627        );
628        assert!(info.imports.is_empty());
629    }
630
631    #[test]
632    fn skips_icon_link() {
633        let info =
634            parse_html_to_module(FileId(0), r#"<link rel="icon" href="./favicon.ico" />"#, 0);
635        assert!(info.imports.is_empty());
636    }
637
638    #[test]
639    fn skips_remote_stylesheet() {
640        let info = parse_html_to_module(
641            FileId(0),
642            r#"<link rel="stylesheet" href="https://fonts.googleapis.com/css" />"#,
643            0,
644        );
645        assert!(info.imports.is_empty());
646    }
647
648    #[test]
649    fn skips_commented_out_script() {
650        let info = parse_html_to_module(
651            FileId(0),
652            r#"<!-- <script src="./old.js"></script> -->
653            <script src="./new.js"></script>"#,
654            0,
655        );
656        assert_eq!(info.imports.len(), 1);
657        assert_eq!(info.imports[0].source, "./new.js");
658    }
659
660    #[test]
661    fn skips_commented_out_link() {
662        let info = parse_html_to_module(
663            FileId(0),
664            r#"<!-- <link rel="stylesheet" href="./old.css" /> -->
665            <link rel="stylesheet" href="./new.css" />"#,
666            0,
667        );
668        assert_eq!(info.imports.len(), 1);
669        assert_eq!(info.imports[0].source, "./new.css");
670    }
671
672    #[test]
673    fn handles_multiline_script_tag() {
674        let info = parse_html_to_module(
675            FileId(0),
676            "<script\n  type=\"module\"\n  src=\"./src/entry.js\"\n></script>",
677            0,
678        );
679        assert_eq!(info.imports.len(), 1);
680        assert_eq!(info.imports[0].source, "./src/entry.js");
681    }
682
683    #[test]
684    fn handles_multiline_link_tag() {
685        let info = parse_html_to_module(
686            FileId(0),
687            "<link\n  rel=\"stylesheet\"\n  href=\"./src/global.css\"\n/>",
688            0,
689        );
690        assert_eq!(info.imports.len(), 1);
691        assert_eq!(info.imports[0].source, "./src/global.css");
692    }
693
694    #[test]
695    fn full_vite_html() {
696        let info = parse_html_to_module(
697            FileId(0),
698            r#"<!doctype html>
699<html>
700  <head>
701    <link rel="stylesheet" href="./src/global.css" />
702    <link rel="icon" href="/favicon.ico" />
703  </head>
704  <body>
705    <div id="app"></div>
706    <script type="module" src="./src/entry.js"></script>
707  </body>
708</html>"#,
709            0,
710        );
711        assert_eq!(info.imports.len(), 2);
712        let sources: Vec<&str> = info.imports.iter().map(|i| i.source.as_str()).collect();
713        assert!(sources.contains(&"./src/global.css"));
714        assert!(sources.contains(&"./src/entry.js"));
715    }
716
717    #[test]
718    fn empty_html() {
719        let info = parse_html_to_module(FileId(0), "", 0);
720        assert!(info.imports.is_empty());
721    }
722
723    #[test]
724    fn html_with_no_assets() {
725        let info = parse_html_to_module(
726            FileId(0),
727            r"<!doctype html><html><body><h1>Hello</h1></body></html>",
728            0,
729        );
730        assert!(info.imports.is_empty());
731    }
732
733    #[test]
734    fn single_quoted_attributes() {
735        let info = parse_html_to_module(FileId(0), r"<script src='./src/entry.js'></script>", 0);
736        assert_eq!(info.imports.len(), 1);
737        assert_eq!(info.imports[0].source, "./src/entry.js");
738    }
739
740    #[test]
741    fn all_imports_are_side_effect() {
742        let info = parse_html_to_module(
743            FileId(0),
744            r#"<script src="./entry.js"></script>
745            <link rel="stylesheet" href="./style.css" />"#,
746            0,
747        );
748        for imp in &info.imports {
749            assert!(matches!(imp.imported_name, ImportedName::SideEffect));
750            assert!(imp.local_name.is_empty());
751            assert!(!imp.is_type_only);
752        }
753    }
754
755    #[test]
756    fn suppression_comments_extracted() {
757        let info = parse_html_to_module(
758            FileId(0),
759            "<!-- fallow-ignore-file -->\n<script src=\"./entry.js\"></script>",
760            0,
761        );
762        assert_eq!(info.imports.len(), 1);
763    }
764
765    #[test]
766    fn angular_template_extracts_member_refs() {
767        let info = parse_html_to_module(
768            FileId(0),
769            "<h1>{{ title() }}</h1>\n\
770             <p [class.highlighted]=\"isHighlighted\">{{ greeting() }}</p>\n\
771             <button (click)=\"onButtonClick()\">Toggle</button>",
772            0,
773        );
774        let fact_names: rustc_hash::FxHashSet<&str> = info
775            .semantic_facts
776            .iter()
777            .filter_map(|fact| {
778                if let SemanticFact::AngularTemplateMemberAccess(access) = fact {
779                    Some(access.member.as_str())
780                } else {
781                    None
782                }
783            })
784            .collect();
785        assert!(fact_names.contains("title"), "should contain 'title'");
786        assert!(
787            fact_names.contains("isHighlighted"),
788            "should contain 'isHighlighted'"
789        );
790        assert!(fact_names.contains("greeting"), "should contain 'greeting'");
791        assert!(
792            fact_names.contains("onButtonClick"),
793            "should contain 'onButtonClick'"
794        );
795        assert!(
796            info.member_accesses.is_empty(),
797            "Angular template refs should emit typed facts instead of member accesses: {:?}",
798            info.member_accesses
799        );
800    }
801
802    #[test]
803    fn plain_html_no_angular_refs() {
804        let info = parse_html_to_module(
805            FileId(0),
806            "<!doctype html><html><body><h1>Hello</h1></body></html>",
807            0,
808        );
809        assert!(info.member_accesses.is_empty());
810    }
811}
fallow_extract/html.rs

fallow_extract/
html.rs
