fix(scan): mask multiline block comments

src/scan.rs

@@ -243,6 +243,7 @@ impl ScanState {
         let mut brace_depth = 0_i32;
         let mut rust_raw_end: Option<String> = None;
         let mut javascript_template_open = false;
+        let mut block_comment_depth = 0_usize;
 
         for (index, raw_line) in content.lines().enumerate() {
             let line_number = index + 1;
@@ -269,8 +270,9 @@ impl ScanState {
             } else {
                 raw_line
             };
-            let line = strip_line_comment(code_line, language);
+            let block_masked = mask_block_comments(code_line, &mut block_comment_depth);
-            let call_line = mask_quoted_strings(line);
+            let line = strip_line_comment(&block_masked, language);
+            let call_line = mask_quoted_strings(&line);
             let indent = raw_line.len() - raw_line.trim_start().len();
             if language == "python" && !line.trim().is_empty() {
                 while scopes.last().is_some_and(|scope| scope.indent >= indent) {
@@ -287,7 +289,7 @@ impl ScanState {
 
             if let Some(pattern) = patterns.enclosing_scope.as_ref()
                 && let Some(name) = pattern
-                    .captures(line)
+                    .captures(&line)
                     .and_then(|captures| captures.name("name"))
                 && let Some(id) = self
                     .symbols_by_name
@@ -303,7 +305,7 @@ impl ScanState {
             }
 
             let mut type_definition_name = None;
-            if let Some(captures) = patterns.class.captures(line)
+            if let Some(captures) = patterns.class.captures(&line)
                 && let Some(name) = captures.name("name")
             {
                 type_definition_name = Some(name.as_str().to_owned());
@@ -319,7 +321,7 @@ impl ScanState {
                     .relationships
                     .as_ref()
                     .into_iter()
-                    .flat_map(|pattern| pattern.captures_iter(line))
+                    .flat_map(|pattern| pattern.captures_iter(&line))
                     .filter_map(|captures| captures.name("names"))
                     .flat_map(|names| names.as_str().split(','))
                     .map(str::trim)
@@ -353,7 +355,7 @@ impl ScanState {
             let mut definition_name = None;
             if let Some(name) = patterns.function.iter().find_map(|pattern| {
                 pattern
-                    .captures(line)
+                    .captures(&line)
                     .and_then(|captures| captures.name("name").map(|name| name.as_str().to_owned()))
             }) && !is_keyword(&name)
             {
@@ -389,7 +391,7 @@ impl ScanState {
             for captures in patterns
                 .import
                 .iter()
-                .flat_map(|pattern| pattern.captures_iter(line))
+                .flat_map(|pattern| pattern.captures_iter(&line))
             {
                 if let Some(target) = captures.name("name").or_else(|| captures.name("name2")) {
                     self.pending.push(PendingRelationship {
@@ -779,13 +781,93 @@ fn language_for(path: &Path) -> Option<&'static str> {
     }
 }
 
-fn strip_line_comment<'a>(line: &'a str, language: &str) -> &'a str {
+fn strip_line_comment(line: &str, language: &str) -> String {
-    let marker = if language == "python" { '#' } else { '/' };
+    let chars: Vec<_> = line.chars().collect();
-    if marker == '/' {
+    let mut quote = None;
-        line.split_once("//").map_or(line, |(before, _)| before)
+    let mut escaped = false;
-    } else {
+    let mut index = 0;
-        line.split_once('#').map_or(line, |(before, _)| before)
+    while index < chars.len() {
+        let ch = chars[index];
+        if let Some(active_quote) = quote {
+            if escaped {
+                escaped = false;
+            } else if ch == '\\' {
+                escaped = true;
+            } else if ch == active_quote {
+                quote = None;
             }
+        } else if matches!(ch, '\'' | '"') && has_closing_quote(&chars, index + 1, ch) {
+            quote = Some(ch);
+        } else if language == "python" && ch == '#'
+            || language != "python" && ch == '/' && chars.get(index + 1) == Some(&'/')
+        {
+            return chars[..index].iter().collect();
+        }
+        index += 1;
+    }
+    line.into()
+}
+
+fn mask_block_comments(line: &str, depth: &mut usize) -> String {
+    let chars: Vec<_> = line.chars().collect();
+    let mut result = String::with_capacity(line.len());
+    let mut quote = None;
+    let mut escaped = false;
+    let mut index = 0;
+    while index < chars.len() {
+        let ch = chars[index];
+        let next = chars.get(index + 1).copied();
+        if *depth > 0 {
+            if ch == '/' && next == Some('*') {
+                *depth += 1;
+                result.push_str("  ");
+                index += 2;
+            } else if ch == '*' && next == Some('/') {
+                *depth -= 1;
+                result.push_str("  ");
+                index += 2;
+            } else {
+                result.push(' ');
+                index += 1;
+            }
+        } else if let Some(active_quote) = quote {
+            result.push(ch);
+            if escaped {
+                escaped = false;
+            } else if ch == '\\' {
+                escaped = true;
+            } else if ch == active_quote {
+                quote = None;
+            }
+            index += 1;
+        } else if matches!(ch, '\'' | '"') && has_closing_quote(&chars, index + 1, ch) {
+            quote = Some(ch);
+            result.push(ch);
+            index += 1;
+        } else if ch == '/' && next == Some('*') {
+            *depth = 1;
+            result.push_str("  ");
+            index += 2;
+        } else {
+            result.push(ch);
+            index += 1;
+        }
+    }
+    result
+}
+
+fn has_closing_quote(chars: &[char], start: usize, quote: char) -> bool {
+    let mut escaped = false;
+    for ch in &chars[start..] {
+        if escaped {
+            escaped = false;
+        } else if *ch == '\\' {
+            escaped = true;
+        } else if *ch == quote {
+            return true;
+        }
+    }
+    false
 }
 
 fn brace_delta(line: &str) -> i32 {
@@ -962,10 +1044,32 @@ fn is_keyword(name: &str) -> bool {
 
 #[cfg(test)]
 mod tests {
-    use super::{ScanOptions, scan, scan_with_options};
+    use super::{ScanOptions, mask_block_comments, scan, scan_with_options, strip_line_comment};
     use crate::model::{EdgeKind, NodeKind};
     use std::fs;
 
+    #[test]
+    fn masks_nested_block_comments_without_truncating_quoted_urls() {
+        let mut depth = 0;
+        assert_eq!(
+            strip_line_comment(
+                r#"import value from "https://example.invalid/value.js"; // trailing"#,
+                "javascript"
+            ),
+            r#"import value from "https://example.invalid/value.js"; "#
+        );
+        assert_eq!(
+            mask_block_comments("code /* outer", &mut depth),
+            "code         "
+        );
+        assert_eq!(depth, 1);
+        let nested = "/* inner */ end */";
+        let masked = mask_block_comments(nested, &mut depth);
+        assert_eq!(masked.len(), nested.len());
+        assert!(masked.trim().is_empty());
+        assert_eq!(depth, 0);
+    }
+
     #[test]
     fn scans_functions_calls_and_cycles() {
         let dir = tempfile::tempdir().unwrap();

tests/fixtures/languages/c/basic.c

@@ -20,3 +20,8 @@ int start() {
 int helper() {
     return 0;
 }
+
+/*
+struct CommentGhost {};
+int comment_call() { return phantom_call(); }
+*/

tests/fixtures/languages/cpp/basic.cpp

@@ -19,3 +19,8 @@ public:
 int helper() {
     return 0;
 }
+
+/*
+class CommentGhost {};
+int comment_call() { return phantom_call(); }
+*/

tests/fixtures/languages/dart/basic.dart

@@ -17,3 +17,8 @@ abstract class Controller extends Base with Logging implements Runnable {
 
   void helper() {}
 }
+
+/*
+class CommentGhost {}
+void commentCall() { phantomCall(); }
+*/

tests/fixtures/languages/javascript/basic.js

@@ -20,3 +20,8 @@ const ignored = `
 class Phantom extends Ghost {}
 function fake() { phantomCall(); }
 `;
+
+/*
+class CommentGhost {}
+function commentCall() { phantomCall(); }
+*/

tests/fixtures/languages/rust/basic.rs

@@ -21,3 +21,8 @@ pub const fn constant() -> usize {
 pub unsafe fn unsafe_op() {}
 
 pub extern "C" fn ffi_entry() {}
+
+/*
+struct CommentGhost;
+fn comment_call() { phantom_call(); }
+*/

tests/fixtures/languages/typescript/basic.ts

@@ -25,3 +25,8 @@ const ignored = `
 interface Phantom {}
 function fake(): void { phantomCall(); }
 `;
+
+/*
+interface CommentGhost {}
+function commentCall(): void { phantomCall(); }
+*/

tests/language_parity.rs

@@ -266,6 +266,14 @@ fn assert_fixture(case: FixtureCase) {
         "{} call baseline",
         case.directory
     );
+    assert!(
+        !building.nodes.iter().any(|node| matches!(
+            node.name.as_str(),
+            "CommentGhost" | "commentCall" | "comment_call" | "phantom_call"
+        )),
+        "{} block-comment clean negative",
+        case.directory
+    );
 }
 
 fn fixture_root(directory: &str) -> PathBuf {