feat: ensure source maps are cached (#628)

fraidev · web-flow · commit 8701dcceb46f · 2025-11-26T11:11:15.000-03:00
diff --git a/src/analysis.rs b/src/analysis.rs
@@ -276,6 +276,9 @@ pub struct ModuleInfo {
   /// or `@import { SomeType } from "npm:some-module"`.
   #[serde(skip_serializing_if = "Vec::is_empty", default)]
   pub jsdoc_imports: Vec<JsDocImportInfo>,
+  /// Source map URL extracted from sourceMappingURL comment
+  #[serde(skip_serializing_if = "Option::is_none", default)]
+  pub source_map_url: Option<SpecifierWithRange>,
 }
 
 fn is_false(v: &bool) -> bool {
@@ -442,6 +445,14 @@ pub fn find_jsx_import_source_types(text: &str) -> Option<regex::Match<'_>> {
     .and_then(|c| c.get(1))
 }
 
+/// Matches the `sourceMappingURL` comment.
+pub fn find_source_mapping_url(text: &str) -> Option<regex::Match<'_>> {
+  static SOURCE_MAPPING_URL_RE: Lazy<Regex> = Lazy::new(|| {
+    Regex::new(r"(?i)^[#@]\s*sourceMappingURL\s*=\s*(\S+)").unwrap()
+  });
+  SOURCE_MAPPING_URL_RE.captures(text).and_then(|c| c.get(1))
+}
+
 /// Matches the `@ts-self-types` pragma.
 pub fn find_ts_self_types(text: &str) -> Option<regex::Match<'_>> {
   static TS_SELF_TYPES_RE: Lazy<Regex> = Lazy::new(|| {
@@ -505,6 +516,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(&module_info, json!({}));
   }
@@ -574,6 +586,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -659,6 +672,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -704,6 +718,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -734,6 +749,7 @@ mod test {
       }),
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -764,6 +780,7 @@ mod test {
         },
       }),
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -819,6 +836,7 @@ mod test {
           resolution_mode: Some(TypeScriptTypesResolutionMode::Require),
         },
       ]),
+      source_map_url: None,
     };
     run_serialization_test(
       &module_info,
@@ -1053,6 +1071,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     let json = json!({
       "dependencies": [{
@@ -1097,6 +1116,7 @@ mod test {
       jsx_import_source: None,
       jsx_import_source_types: None,
       jsdoc_imports: Vec::new(),
+      source_map_url: None,
     };
     let json = json!({
       "dependencies": [{
@@ -1109,6 +1129,36 @@ mod test {
     run_v1_deserialization_test(json, &expected);
   }
 
+  #[test]
+  fn module_info_serialization_source_map_url() {
+    let module_info = ModuleInfo {
+      is_script: false,
+      dependencies: Vec::new(),
+      ts_references: Vec::new(),
+      self_types_specifier: None,
+      jsx_import_source: None,
+      jsx_import_source_types: None,
+      jsdoc_imports: Vec::new(),
+      source_map_url: Some(SpecifierWithRange {
+        text: "file.js.map".to_string(),
+        range: PositionRange {
+          start: Position::zeroed(),
+          end: Position::zeroed(),
+        },
+      }),
+    };
+
+    run_serialization_test(
+      &module_info,
+      json!({
+        "sourceMapUrl": {
+          "text": "file.js.map",
+          "range": [[0, 0], [0, 0]],
+        }
+      }),
+    );
+  }
+
   #[track_caller]
   fn run_serialization_test<
     T: DeserializeOwned + Serialize + std::fmt::Debug + PartialEq + Eq,
@@ -1133,4 +1183,59 @@ mod test {
     let deserialized_value = serde_json::from_value::<T>(json).unwrap();
     assert_eq!(deserialized_value, *value);
   }
+
+  #[test]
+  fn test_find_source_mapping_url() {
+    // Test with # prefix
+    let m = find_source_mapping_url("# sourceMappingURL=file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map");
+
+    // Test with @ prefix
+    let m = find_source_mapping_url("@ sourceMappingURL=file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map");
+
+    // Test case insensitivity
+    let m = find_source_mapping_url("# SOURCEMAPPINGURL=file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map");
+
+    // Test with no spaces
+    let m = find_source_mapping_url("#sourceMappingURL=file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map");
+
+    // Test with multiple spaces
+    let m = find_source_mapping_url("#  sourceMappingURL  =  file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map");
+
+    // Test with URL containing query params
+    let m = find_source_mapping_url("# sourceMappingURL=file.js.map?v=123");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "file.js.map?v=123");
+
+    // Test with data URL
+    let m = find_source_mapping_url(
+      "# sourceMappingURL=data:application/json;base64,eyJ2ZXJzaW9uIjozfQ==",
+    );
+    assert!(m.is_some());
+    assert_eq!(
+      m.unwrap().as_str(),
+      "data:application/json;base64,eyJ2ZXJzaW9uIjozfQ=="
+    );
+
+    // Test with relative path
+    let m = find_source_mapping_url("# sourceMappingURL=../maps/file.js.map");
+    assert!(m.is_some());
+    assert_eq!(m.unwrap().as_str(), "../maps/file.js.map");
+
+    // Test invalid formats
+    assert!(find_source_mapping_url("sourceMappingURL=file.js.map").is_none());
+    assert!(
+      find_source_mapping_url("// sourceMappingURL=file.js.map").is_none()
+    );
+    assert!(find_source_mapping_url("# sourceMappingURL").is_none());
+  }
 }
diff --git a/src/ast/mod.rs b/src/ast/mod.rs
@@ -16,6 +16,7 @@ use crate::analysis::find_jsx_import_source;
 use crate::analysis::find_jsx_import_source_types;
 use crate::analysis::find_path_reference;
 use crate::analysis::find_resolution_mode;
+use crate::analysis::find_source_mapping_url;
 use crate::analysis::find_ts_self_types;
 use crate::analysis::find_ts_types;
 use crate::analysis::find_types_reference;
@@ -268,6 +269,9 @@ impl<'a> ParserModuleAnalyzer<'a> {
         None => comments.get_leading(program.start()),
       },
     };
+    // Get trailing comments from the program end to extract sourceMappingURL
+    // which is typically at the very end of the file
+    let trailing_comments = comments.get_trailing(program.end());
     ModuleInfo {
       is_script: program.compute_is_script(),
       dependencies: analyze_dependencies(program, text_info, comments),
@@ -288,6 +292,7 @@ impl<'a> ParserModuleAnalyzer<'a> {
         leading_comments,
       ),
       jsdoc_imports: analyze_jsdoc_imports(media_type, text_info, comments),
+      source_map_url: analyze_source_map_url(text_info, trailing_comments),
     }
   }
 
@@ -608,6 +613,27 @@ fn analyze_ts_self_types(
   })
 }
 
+// Search source map URL from trailing comments
+fn analyze_source_map_url(
+  text_info: &SourceTextInfo,
+  trailing_comments: Option<&Vec<deno_ast::swc::common::comments::Comment>>,
+) -> Option<SpecifierWithRange> {
+  trailing_comments.and_then(|c| {
+    c.iter().rev().find_map(|comment| {
+      let source_mapping_url = find_source_mapping_url(&comment.text)?;
+      Some(SpecifierWithRange {
+        text: source_mapping_url.as_str().to_string(),
+        range: comment_source_to_position_range(
+          comment.start(),
+          source_mapping_url.range(),
+          text_info,
+          true,
+        ),
+      })
+    })
+  })
+}
+
 /// Searches comments for any `@ts-types` or `@deno-types` compiler hints.
 pub fn analyze_ts_or_deno_types(
   text_info: &SourceTextInfo,
@@ -1459,6 +1485,7 @@ export {};
         }),
         jsx_import_source_types: None,
         jsdoc_imports: vec![],
+        source_map_url: None,
       },
     );
   }
@@ -1539,4 +1566,105 @@ export {};
       Some(TypeScriptTypesResolutionMode::Import)
     );
   }
+
+  #[tokio::test]
+  async fn test_source_mapping_url_extraction() {
+    let specifier =
+      ModuleSpecifier::parse("file:///test.js").expect("bad specifier");
+    let source = r#"
+export function test() {
+  return "hello";
+}
+//# sourceMappingURL=test.js.map
+
+// comment after
+
+"#;
+    let module_info = DefaultModuleAnalyzer
+      .analyze(&specifier, source.into(), MediaType::JavaScript)
+      .await
+      .unwrap();
+    assert!(module_info.source_map_url.is_some());
+    let source_map = module_info.source_map_url.unwrap();
+    assert_eq!(source_map.text, "test.js.map");
+    // Verify the range points to the correct location
+    assert_eq!(source_map.range.start.line, 4);
+    assert_eq!(source_map.range.start.character, 21);
+    assert_eq!(source_map.range.end.line, 4);
+    assert_eq!(source_map.range.end.character, 32);
+  }
+
+  #[tokio::test]
+  async fn test_source_mapping_url_with_at_prefix() {
+    let specifier =
+      ModuleSpecifier::parse("file:///test.js").expect("bad specifier");
+    let source = r#"
+const x = 1;
+//@ sourceMappingURL=bundle.js.map
+"#;
+    let module_info = DefaultModuleAnalyzer
+      .analyze(&specifier, source.into(), MediaType::JavaScript)
+      .await
+      .unwrap();
+    assert!(module_info.source_map_url.is_some());
+    assert_eq!(module_info.source_map_url.unwrap().text, "bundle.js.map");
+  }
+
+  #[tokio::test]
+  async fn test_source_mapping_url_data_uri() {
+    let specifier =
+      ModuleSpecifier::parse("file:///test.js").expect("bad specifier");
+    let source = r#"
+console.log("minified");
+//# sourceMappingURL=data:application/json;base64,eyJ2ZXJzaW9uIjozfQ==
+"#;
+    let module_info = DefaultModuleAnalyzer
+      .analyze(&specifier, source.into(), MediaType::JavaScript)
+      .await
+      .unwrap();
+    assert!(module_info.source_map_url.is_some());
+    assert_eq!(
+      module_info.source_map_url.unwrap().text,
+      "data:application/json;base64,eyJ2ZXJzaW9uIjozfQ=="
+    );
+  }
+
+  #[tokio::test]
+  async fn test_no_source_mapping_url() {
+    let specifier =
+      ModuleSpecifier::parse("file:///test.js").expect("bad specifier");
+    let source = r#"
+export function test() {
+  return "hello";
+}
+"#;
+    let module_info = DefaultModuleAnalyzer
+      .analyze(&specifier, source.into(), MediaType::JavaScript)
+      .await
+      .unwrap();
+    assert!(module_info.source_map_url.is_none());
+  }
+
+  #[tokio::test]
+  async fn test_source_mapping_url_only_at_end() {
+    let specifier =
+      ModuleSpecifier::parse("file:///test.js").expect("bad specifier");
+    // sourceMappingURL in the middle of file should NOT be extracted
+    // since we only look for trailing comments at program.end()
+    let source = r#"
+export function test() {
+  // This is not a sourceMappingURL comment
+  return "hello";
+}
+//# sourceMappingURL=test.js.map
+console.log("more code");
+"#;
+    let module_info = DefaultModuleAnalyzer
+      .analyze(&specifier, source.into(), MediaType::JavaScript)
+      .await
+      .unwrap();
+    // Should NOT extract the sourceMappingURL if there's code after it
+    // because it's not at the end of the program
+    assert!(module_info.source_map_url.is_none());
+  }
 }
diff --git a/src/graph.rs b/src/graph.rs
