validation: external-agent literal reference rewrite

codex-rs/external-agent-migration/src/rewrite.rs

@@ -65,7 +65,7 @@ fn replace_with_boundaries(input: &str, needle: &str, replacement: &str) -> Stri
         let boundary_before = start == 0 || !is_word_byte(bytes[start - 1]);
         let boundary_after = end == bytes.len() || !is_word_byte(bytes[end]);
 
-        if boundary_before && boundary_after {
+        if boundary_before && boundary_after && !is_literal_reference_match(bytes, start, end) {
             output.push_str(&input[last_emitted..start]);
             output.push_str(replacement);
             last_emitted = end;
@@ -104,7 +104,7 @@ fn replace_case_insensitive_with_boundaries(
         let boundary_before = start == 0 || !is_word_byte(bytes[start - 1]);
         let boundary_after = end == bytes.len() || !is_word_byte(bytes[end]);
 
-        if boundary_before && boundary_after {
+        if boundary_before && boundary_after && !is_literal_reference_match(bytes, start, end) {
             output.push_str(&input[last_emitted..start]);
             output.push_str(replacement);
             last_emitted = end;
@@ -121,6 +121,32 @@ fn replace_case_insensitive_with_boundaries(
     output
 }
 
+/// Product names embedded in filesystem paths, URLs, or dotted identifiers are
+/// literal references to the source tool, not prose that should be retargeted.
+fn is_literal_reference_match(bytes: &[u8], start: usize, end: usize) -> bool {
+    if let Some(before) = start.checked_sub(1).and_then(|idx| bytes.get(idx))
+        && matches!(*before, b'/' | b'\\' | b'.')
+    {
+        return true;
+    }
+
+    let Some(after) = bytes.get(end) else {
+        return false;
+    };
+    if matches!(*after, b'/' | b'\\') {
+        return true;
+    }
+
+    *after == b'.'
+        && bytes
+            .get(end + 1)
+            .is_some_and(|byte| is_reference_suffix_byte(*byte))
+}
+
+fn is_reference_suffix_byte(byte: u8) -> bool {
+    byte.is_ascii_alphanumeric() || matches!(byte, b'_' | b'-')
+}
+
 fn is_word_byte(byte: u8) -> bool {
     byte.is_ascii_alphanumeric() || byte == b'_'
 }

codex-rs/external-agent-migration/src/rewrite_tests.rs

@@ -4,10 +4,65 @@ use pretty_assertions::assert_eq;
 const PROFILE: RewriteProfile = RewriteProfile::new("SOURCE.md", &["source agent"])
     .with_case_sensitive_term_variants(&["Source"]);
 
+const CLAUDE_PROFILE: RewriteProfile = RewriteProfile::new(
+    "CLAUDE.md",
+    &[
+        "claude code",
+        "claude-code",
+        "claude_code",
+        "claudecode",
+        "claude",
+    ],
+);
+
 #[test]
 fn rewrites_terms_only_at_word_boundaries() {
     assert_eq!(
         PROFILE.rewrite("SOURCE.md Source source agent source_agent"),
         "AGENTS.md Codex Codex source_agent"
     );
 }
+
+#[test]
+fn preserves_source_specific_filesystem_references() {
+    assert_eq!(
+        PROFILE.rewrite(
+            "Read ./SOURCE.md, C:\\repo\\SOURCE.md, ~/.Source/config, and .Source before using Source."
+        ),
+        "Read ./SOURCE.md, C:\\repo\\SOURCE.md, ~/.Source/config, and .Source before using Codex."
+    );
+}
+
+#[test]
+fn preserves_source_specific_urls_and_dotted_identifiers() {
+    assert_eq!(
+        PROFILE.rewrite(
+            "See https://Source.dev/docs, docs.Source.dev, and plugin.Source.config. Source."
+        ),
+        "See https://Source.dev/docs, docs.Source.dev, and plugin.Source.config. Codex."
+    );
+}
+
+#[test]
+fn rewrites_standalone_doc_name_but_not_path_qualified_doc_name() {
+    assert_eq!(
+        PROFILE.rewrite("SOURCE.md lives beside ../SOURCE.md and $HOME/SOURCE.md."),
+        "AGENTS.md lives beside ../SOURCE.md and $HOME/SOURCE.md."
+    );
+}
+
+#[test]
+fn preserves_claude_skill_paths_while_retargeting_prose() {
+    let input = concat!(
+        "Session transcripts live in ~/.claude/projects/<slug>/.\n",
+        "See also .claude/commands/example.md and C:\\Users\\me\\.claude\\plans\\today.md.\n",
+        "Use Claude Code UI and read CLAUDE.md."
+    );
+    let expected = concat!(
+        "Session transcripts live in ~/.claude/projects/<slug>/.\n",
+        "See also .claude/commands/example.md and C:\\Users\\me\\.claude\\plans\\today.md.\n",
+        "Use Codex UI and read AGENTS.md."
+    );
+
+    assert_eq!(CLAUDE_PROFILE.rewrite(input), expected);
+}