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

fallow_extract/
html.rs
