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        exported_factory_return_object_shapes: Box::default(),
285        type_member_types: Box::default(),
286        injection_tokens: vec![],
287        local_type_declarations: Vec::new(),
288        public_signature_type_references: Vec::new(),
289        namespace_object_aliases: Vec::new(),
290        iconify_prefixes: Vec::new(),
291        iconify_icon_names: Vec::new(),
292        auto_import_candidates: Vec::new(),
293        directives: Vec::new(),
294        client_only_dynamic_import_spans: Vec::new(),
295        security_sinks,
296        security_sinks_skipped: 0,
297        security_unresolved_callee_sites: Vec::new(),
298        tainted_bindings: Vec::new(),
299        sanitized_sink_args: Vec::new(),
300        security_control_sites: Vec::new(),
301        callee_uses: Vec::new(),
302        misplaced_directives: Vec::new(),
303        inline_server_action_exports: Vec::new(),
304        di_key_sites: Vec::new(),
305        has_dynamic_provide: false,
306        referenced_import_bindings: Vec::new(),
307        component_props: Vec::new(),
308        has_props_attrs_fallthrough: false,
309        has_define_expose: false,
310        has_define_model: false,
311        has_unharvestable_props: false,
312        component_emits: Vec::new(),
313        angular_inputs: Vec::new(),
314        angular_outputs: Vec::new(),
315        angular_component_selectors: Vec::new(),
316        registered_custom_elements: Vec::new(),
317        // Custom-element tags rendered in a standalone `.html` document (an app
318        // shell, demo, or dev page) feed the Lit `unrendered-component` arm's
319        // project-wide rendered-tag union, so an element rendered only from HTML
320        // (e.g. a root `<my-app>` in `index.html`) is not falsely flagged.
321        used_custom_element_tags: collect_custom_element_tags(source),
322        angular_used_selectors,
323        angular_entry_component_refs: Vec::new(),
324        has_dynamic_component_render,
325        has_unharvestable_emits: false,
326        has_dynamic_emit: false,
327        has_emit_whole_object_use: false,
328        load_return_keys: Vec::new(),
329        has_unharvestable_load: false,
330        has_load_data_whole_use: false,
331        has_page_data_store_whole_use: false,
332        has_route_loader_data_whole_use: false,
333        component_functions: Vec::new(),
334        react_props: Vec::new(),
335        hook_uses: Vec::new(),
336        render_edges: Vec::new(),
337        svelte_dispatched_events: Vec::new(),
338        svelte_listened_events: Vec::new(),
339        has_dynamic_dispatch: false,
340    }
341}
342
343#[cfg(test)]
344mod tests {
345    use super::*;
346
347    #[test]
348    fn is_html_file_html() {
349        assert!(is_html_file(Path::new("index.html")));
350    }
351
352    #[test]
353    fn is_html_file_nested() {
354        assert!(is_html_file(Path::new("pages/about.html")));
355    }
356
357    #[test]
358    fn is_html_file_rejects_htm() {
359        assert!(!is_html_file(Path::new("index.htm")));
360    }
361
362    #[test]
363    fn is_html_file_rejects_js() {
364        assert!(!is_html_file(Path::new("app.js")));
365    }
366
367    #[test]
368    fn is_html_file_rejects_ts() {
369        assert!(!is_html_file(Path::new("app.ts")));
370    }
371
372    #[test]
373    fn is_html_file_rejects_vue() {
374        assert!(!is_html_file(Path::new("App.vue")));
375    }
376
377    #[test]
378    fn remote_url_http() {
379        assert!(is_remote_url("http://example.com/script.js"));
380    }
381
382    #[test]
383    fn remote_url_https() {
384        assert!(is_remote_url("https://cdn.example.com/style.css"));
385    }
386
387    #[test]
388    fn remote_url_protocol_relative() {
389        assert!(is_remote_url("//cdn.example.com/lib.js"));
390    }
391
392    #[test]
393    fn remote_url_data() {
394        assert!(is_remote_url("data:text/javascript;base64,abc"));
395    }
396
397    #[test]
398    fn local_relative_not_remote() {
399        assert!(!is_remote_url("./src/entry.js"));
400    }
401
402    #[test]
403    fn local_root_relative_not_remote() {
404        assert!(!is_remote_url("/src/entry.js"));
405    }
406
407    #[test]
408    fn extracts_module_script_src() {
409        let info = parse_html_to_module(
410            FileId(0),
411            r#"<script type="module" src="./src/entry.js"></script>"#,
412            0,
413        );
414        assert_eq!(info.imports.len(), 1);
415        assert_eq!(info.imports[0].source, "./src/entry.js");
416    }
417
418    #[test]
419    fn extracts_plain_script_src() {
420        let info = parse_html_to_module(
421            FileId(0),
422            r#"<script src="./src/polyfills.js"></script>"#,
423            0,
424        );
425        assert_eq!(info.imports.len(), 1);
426        assert_eq!(info.imports[0].source, "./src/polyfills.js");
427    }
428
429    #[test]
430    fn extracts_multiple_scripts() {
431        let info = parse_html_to_module(
432            FileId(0),
433            r#"
434            <script type="module" src="./src/entry.js"></script>
435            <script src="./src/polyfills.js"></script>
436            "#,
437            0,
438        );
439        assert_eq!(info.imports.len(), 2);
440    }
441
442    #[test]
443    fn skips_inline_script() {
444        let info = parse_html_to_module(FileId(0), r#"<script>console.log("hello");</script>"#, 0);
445        assert!(info.imports.is_empty());
446    }
447
448    #[test]
449    fn skips_handlebars_placeholder_in_script_src() {
450        let info = parse_html_to_module(
451            FileId(0),
452            r#"<script src="{{rootURL}}assets/app.js"></script>
453               <script src="{{config.assetsPath}}vendor.js"></script>"#,
454            0,
455        );
456        assert!(
457            info.imports.is_empty(),
458            "Handlebars-placeholder script srcs should not enter the import graph; got {:?}",
459            info.imports
460        );
461    }
462
463    #[test]
464    fn skips_handlebars_placeholder_in_link_href() {
465        let info = parse_html_to_module(
466            FileId(0),
467            r#"<link rel="stylesheet" href="{{rootURL}}assets/app.css">"#,
468            0,
469        );
470        assert!(info.imports.is_empty());
471    }
472
473    #[test]
474    fn skips_ember_cli_blueprint_placeholder() {
475        let info = parse_html_to_module(
476            FileId(0),
477            r####"<script src="###APPNAME###/app.js"></script>"####,
478            0,
479        );
480        assert!(info.imports.is_empty());
481    }
482
483    #[test]
484    fn extracts_normal_specifier_alongside_placeholders() {
485        let info = parse_html_to_module(
486            FileId(0),
487            r#"<script src="{{rootURL}}assets/app.js"></script>
488               <script src="./src/main.ts"></script>"#,
489            0,
490        );
491        assert_eq!(info.imports.len(), 1);
492        assert_eq!(info.imports[0].source, "./src/main.ts");
493    }
494
495    #[test]
496    fn skips_remote_script() {
497        let info = parse_html_to_module(
498            FileId(0),
499            r#"<script src="https://cdn.example.com/lib.js"></script>"#,
500            0,
501        );
502        assert!(info.imports.is_empty());
503    }
504
505    #[test]
506    fn skips_protocol_relative_script() {
507        let info = parse_html_to_module(
508            FileId(0),
509            r#"<script src="//cdn.example.com/lib.js"></script>"#,
510            0,
511        );
512        assert!(info.imports.is_empty());
513    }
514
515    #[test]
516    fn extracts_stylesheet_link() {
517        let info = parse_html_to_module(
518            FileId(0),
519            r#"<link rel="stylesheet" href="./src/global.css" />"#,
520            0,
521        );
522        assert_eq!(info.imports.len(), 1);
523        assert_eq!(info.imports[0].source, "./src/global.css");
524    }
525
526    #[test]
527    fn extracts_modulepreload_link() {
528        let info = parse_html_to_module(
529            FileId(0),
530            r#"<link rel="modulepreload" href="./src/vendor.js" />"#,
531            0,
532        );
533        assert_eq!(info.imports.len(), 1);
534        assert_eq!(info.imports[0].source, "./src/vendor.js");
535    }
536
537    #[test]
538    fn extracts_link_with_reversed_attrs() {
539        let info = parse_html_to_module(
540            FileId(0),
541            r#"<link href="./src/global.css" rel="stylesheet" />"#,
542            0,
543        );
544        assert_eq!(info.imports.len(), 1);
545        assert_eq!(info.imports[0].source, "./src/global.css");
546    }
547
548    #[test]
549    fn bare_script_src_normalized_to_relative() {
550        let info = parse_html_to_module(FileId(0), r#"<script src="app.js"></script>"#, 0);
551        assert_eq!(info.imports.len(), 1);
552        assert_eq!(info.imports[0].source, "./app.js");
553    }
554
555    #[test]
556    fn bare_module_script_src_normalized_to_relative() {
557        let info = parse_html_to_module(
558            FileId(0),
559            r#"<script type="module" src="main.ts"></script>"#,
560            0,
561        );
562        assert_eq!(info.imports.len(), 1);
563        assert_eq!(info.imports[0].source, "./main.ts");
564    }
565
566    #[test]
567    fn bare_stylesheet_link_href_normalized_to_relative() {
568        let info = parse_html_to_module(
569            FileId(0),
570            r#"<link rel="stylesheet" href="styles.css" />"#,
571            0,
572        );
573        assert_eq!(info.imports.len(), 1);
574        assert_eq!(info.imports[0].source, "./styles.css");
575    }
576
577    #[test]
578    fn bare_link_href_reversed_attrs_normalized_to_relative() {
579        let info = parse_html_to_module(
580            FileId(0),
581            r#"<link href="styles.css" rel="stylesheet" />"#,
582            0,
583        );
584        assert_eq!(info.imports.len(), 1);
585        assert_eq!(info.imports[0].source, "./styles.css");
586    }
587
588    #[test]
589    fn bare_modulepreload_link_href_normalized_to_relative() {
590        let info = parse_html_to_module(
591            FileId(0),
592            r#"<link rel="modulepreload" href="vendor.js" />"#,
593            0,
594        );
595        assert_eq!(info.imports.len(), 1);
596        assert_eq!(info.imports[0].source, "./vendor.js");
597    }
598
599    #[test]
600    fn bare_asset_with_subdir_normalized_to_relative() {
601        let info = parse_html_to_module(FileId(0), r#"<script src="assets/app.js"></script>"#, 0);
602        assert_eq!(info.imports.len(), 1);
603        assert_eq!(info.imports[0].source, "./assets/app.js");
604    }
605
606    #[test]
607    fn root_absolute_script_src_unchanged() {
608        let info = parse_html_to_module(FileId(0), r#"<script src="/src/main.ts"></script>"#, 0);
609        assert_eq!(info.imports.len(), 1);
610        assert_eq!(info.imports[0].source, "/src/main.ts");
611    }
612
613    #[test]
614    fn parent_relative_script_src_unchanged() {
615        let info = parse_html_to_module(
616            FileId(0),
617            r#"<script src="../shared/vendor.js"></script>"#,
618            0,
619        );
620        assert_eq!(info.imports.len(), 1);
621        assert_eq!(info.imports[0].source, "../shared/vendor.js");
622    }
623
624    #[test]
625    fn skips_preload_link() {
626        let info = parse_html_to_module(
627            FileId(0),
628            r#"<link rel="preload" href="./src/font.woff2" as="font" />"#,
629            0,
630        );
631        assert!(info.imports.is_empty());
632    }
633
634    #[test]
635    fn skips_icon_link() {
636        let info =
637            parse_html_to_module(FileId(0), r#"<link rel="icon" href="./favicon.ico" />"#, 0);
638        assert!(info.imports.is_empty());
639    }
640
641    #[test]
642    fn skips_remote_stylesheet() {
643        let info = parse_html_to_module(
644            FileId(0),
645            r#"<link rel="stylesheet" href="https://fonts.googleapis.com/css" />"#,
646            0,
647        );
648        assert!(info.imports.is_empty());
649    }
650
651    #[test]
652    fn skips_commented_out_script() {
653        let info = parse_html_to_module(
654            FileId(0),
655            r#"<!-- <script src="./old.js"></script> -->
656            <script src="./new.js"></script>"#,
657            0,
658        );
659        assert_eq!(info.imports.len(), 1);
660        assert_eq!(info.imports[0].source, "./new.js");
661    }
662
663    #[test]
664    fn skips_commented_out_link() {
665        let info = parse_html_to_module(
666            FileId(0),
667            r#"<!-- <link rel="stylesheet" href="./old.css" /> -->
668            <link rel="stylesheet" href="./new.css" />"#,
669            0,
670        );
671        assert_eq!(info.imports.len(), 1);
672        assert_eq!(info.imports[0].source, "./new.css");
673    }
674
675    #[test]
676    fn handles_multiline_script_tag() {
677        let info = parse_html_to_module(
678            FileId(0),
679            "<script\n  type=\"module\"\n  src=\"./src/entry.js\"\n></script>",
680            0,
681        );
682        assert_eq!(info.imports.len(), 1);
683        assert_eq!(info.imports[0].source, "./src/entry.js");
684    }
685
686    #[test]
687    fn handles_multiline_link_tag() {
688        let info = parse_html_to_module(
689            FileId(0),
690            "<link\n  rel=\"stylesheet\"\n  href=\"./src/global.css\"\n/>",
691            0,
692        );
693        assert_eq!(info.imports.len(), 1);
694        assert_eq!(info.imports[0].source, "./src/global.css");
695    }
696
697    #[test]
698    fn full_vite_html() {
699        let info = parse_html_to_module(
700            FileId(0),
701            r#"<!doctype html>
702<html>
703  <head>
704    <link rel="stylesheet" href="./src/global.css" />
705    <link rel="icon" href="/favicon.ico" />
706  </head>
707  <body>
708    <div id="app"></div>
709    <script type="module" src="./src/entry.js"></script>
710  </body>
711</html>"#,
712            0,
713        );
714        assert_eq!(info.imports.len(), 2);
715        let sources: Vec<&str> = info.imports.iter().map(|i| i.source.as_str()).collect();
716        assert!(sources.contains(&"./src/global.css"));
717        assert!(sources.contains(&"./src/entry.js"));
718    }
719
720    #[test]
721    fn empty_html() {
722        let info = parse_html_to_module(FileId(0), "", 0);
723        assert!(info.imports.is_empty());
724    }
725
726    #[test]
727    fn html_with_no_assets() {
728        let info = parse_html_to_module(
729            FileId(0),
730            r"<!doctype html><html><body><h1>Hello</h1></body></html>",
731            0,
732        );
733        assert!(info.imports.is_empty());
734    }
735
736    #[test]
737    fn single_quoted_attributes() {
738        let info = parse_html_to_module(FileId(0), r"<script src='./src/entry.js'></script>", 0);
739        assert_eq!(info.imports.len(), 1);
740        assert_eq!(info.imports[0].source, "./src/entry.js");
741    }
742
743    #[test]
744    fn all_imports_are_side_effect() {
745        let info = parse_html_to_module(
746            FileId(0),
747            r#"<script src="./entry.js"></script>
748            <link rel="stylesheet" href="./style.css" />"#,
749            0,
750        );
751        for imp in &info.imports {
752            assert!(matches!(imp.imported_name, ImportedName::SideEffect));
753            assert!(imp.local_name.is_empty());
754            assert!(!imp.is_type_only);
755        }
756    }
757
758    #[test]
759    fn suppression_comments_extracted() {
760        let info = parse_html_to_module(
761            FileId(0),
762            "<!-- fallow-ignore-file -->\n<script src=\"./entry.js\"></script>",
763            0,
764        );
765        assert_eq!(info.imports.len(), 1);
766    }
767
768    #[test]
769    fn angular_template_extracts_member_refs() {
770        let info = parse_html_to_module(
771            FileId(0),
772            "<h1>{{ title() }}</h1>\n\
773             <p [class.highlighted]=\"isHighlighted\">{{ greeting() }}</p>\n\
774             <button (click)=\"onButtonClick()\">Toggle</button>",
775            0,
776        );
777        let fact_names: rustc_hash::FxHashSet<&str> = info
778            .semantic_facts
779            .iter()
780            .filter_map(|fact| {
781                if let SemanticFact::AngularTemplateMemberAccess(access) = fact {
782                    Some(access.member.as_str())
783                } else {
784                    None
785                }
786            })
787            .collect();
788        assert!(fact_names.contains("title"), "should contain 'title'");
789        assert!(
790            fact_names.contains("isHighlighted"),
791            "should contain 'isHighlighted'"
792        );
793        assert!(fact_names.contains("greeting"), "should contain 'greeting'");
794        assert!(
795            fact_names.contains("onButtonClick"),
796            "should contain 'onButtonClick'"
797        );
798        assert!(
799            info.member_accesses.is_empty(),
800            "Angular template refs should emit typed facts instead of member accesses: {:?}",
801            info.member_accesses
802        );
803    }
804
805    #[test]
806    fn plain_html_no_angular_refs() {
807        let info = parse_html_to_module(
808            FileId(0),
809            "<!doctype html><html><body><h1>Hello</h1></body></html>",
810            0,
811        );
812        assert!(info.member_accesses.is_empty());
813    }
814}
fallow_extract/html.rs

fallow_extract/
html.rs
