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

fallow_extract/
html.rs
