Finish doc comment detection

crates/liyi/src/markers.rs.liyi.jsonc

@@ -43,7 +43,7 @@
       "source_anchor": "const ALIAS_TABLE: &[(&str, &str)] = &[",
       "related": {
         "marker-normalization": "sha256:f1786541ac9be4533a36924a80c578a749024a4980fbc323417c2d5640ed6306",
-        "quine-escape-in-source": null
+        "quine-escape-in-source": "sha256:b929750d0d3c9ce24a6f814d7a1db15f2fc03983efa2647cfcf6a7462d2c8b81"
       }
     },
     {
@@ -125,7 +125,7 @@
       "source_anchor": "fn is_fence_delimiter(line: &str) -> bool {",
       "confidence": 0.95,
       "related": {
-        "markdown-fenced-block-skip": null
+        "markdown-fenced-block-skip": "sha256:0bdd7d1ad747e2a826adfe2eca5652c7a9f382e1089ff12dd2c9b8d88f3920b7"
       }
     },
     {
@@ -164,8 +164,8 @@
       "source_hash": "sha256:882dc8ffe044cfd565123f41aa19ad94b6ec641f829f96a3c7b02565beb43de3",
       "source_anchor": "pub fn scan_markers(content: &str) -> Vec<SourceMarker> {",
       "related": {
-        "markdown-fenced-block-skip": null,
-        "quine-escape-in-source": null
+        "markdown-fenced-block-skip": "sha256:0bdd7d1ad747e2a826adfe2eca5652c7a9f382e1089ff12dd2c9b8d88f3920b7",
+        "quine-escape-in-source": "sha256:b929750d0d3c9ce24a6f814d7a1db15f2fc03983efa2647cfcf6a7462d2c8b81"
       }
     }
   ]

crates/liyi/src/prompt.rs.liyi.jsonc

@@ -51,7 +51,7 @@
       "source_hash": "sha256:4b58abb88e8267fb636eae6727230d6f4c90fdaab97f0125c916e5a55985851e",
       "source_anchor": "pub fn build_prompt_output(",
       "related": {
-        "quine-escape-in-source": null
+        "quine-escape-in-source": "sha256:b929750d0d3c9ce24a6f814d7a1db15f2fc03983efa2647cfcf6a7462d2c8b81"
       }
     },
     {

crates/liyi/src/tree_path/lang_c.rs

@@ -46,6 +46,26 @@ fn c_node_name(node: &Node, source: &str) -> Option<String> {
 }
 
 /// C language configuration.
+/// Detect C doc comments (`/** ... */` and `/// ...`).
+///
+/// C's tree-sitter grammar uses a uniform `comment` kind for all comments.
+/// We distinguish doc comments by prefix: `/**` for block doc and `///`
+/// for line doc. Previous siblings that are not comments are skipped.
+fn c_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        if s.kind() == "comment" {
+            let text = &source[s.byte_range()];
+            if text.starts_with("/**") || text.starts_with("///") {
+                return true;
+            }
+            sibling = s.prev_sibling();
+        } else {
+            break;
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_c::LANGUAGE.into(),
     extensions: &["c"],
@@ -59,7 +79,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body"],
     custom_name: Some(c_node_name),
-    doc_comment_detector: None,
+    doc_comment_detector: Some(c_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_c.rs.liyi.jsonc

@@ -44,11 +44,11 @@
       "reviewed": true,
       "intent": "Define the C language configuration for tree_path resolution: register tree-sitter-c grammar, .c extension only, kind mappings for fn/struct/enum/typedef, wire the custom c_node_name callback for name extraction, and no doc comment detector.",
       "source_span": [
-        49,
-        64
+        69,
+        84
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:896b0b51a031e70d135ee3c4f5c5fec8e24548e376d24f3551c1bcf865815c16",
+      "source_hash": "sha256:bc2b8ed39645ec4e16137afa87b145349dde8c1ff38d950953bfc843eedfedfe",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]

crates/liyi/src/tree_path/lang_cpp.rs

@@ -41,6 +41,30 @@ fn cpp_node_name(node: &Node, source: &str) -> Option<String> {
 }
 
 /// C++ language configuration.
+/// Detect C++ doc comments (`/** ... */` and `/// ...`).
+///
+/// C++'s tree-sitter grammar uses a uniform `comment` kind. We distinguish
+/// doc comments by prefix: `/**` for block doc and `///` for line doc.
+/// Skips `access_specifier` siblings (public/private/protected).
+fn cpp_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        match s.kind() {
+            "comment" => {
+                let text = &source[s.byte_range()];
+                if text.starts_with("/**") || text.starts_with("///") {
+                    return true;
+                }
+                sibling = s.prev_sibling();
+            }
+            "access_specifier" => {
+                sibling = s.prev_sibling();
+            }
+            _ => break,
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_cpp::LANGUAGE.into(),
     extensions: &["cpp", "cc", "cxx", "h", "hpp", "hh", "hxx", "h++", "c++"],
@@ -58,7 +82,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body", "declaration_list"],
     custom_name: Some(cpp_node_name),
-    doc_comment_detector: None,
+    doc_comment_detector: Some(cpp_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_cpp.rs.liyi.jsonc

@@ -20,11 +20,11 @@
       "reviewed": true,
       "intent": "Define the C++ language configuration for tree_path resolution: register tree-sitter-cpp grammar, C++ file extensions (.cpp/.cc/.cxx/.hpp etc.), kind mappings for fn/class/struct/namespace/enum/template/typedef/using, wire the custom cpp_node_name callback, and no doc comment detector.",
       "source_span": [
-        44,
-        63
+        68,
+        87
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:435f6f5356bbbd674a5b9c72860141710adb9e43fbdaaec09fe98584eaff1e34",
+      "source_hash": "sha256:9214c791323abe5ab620f8622089e719f4f0f8e7966cce114cc8a3936208c458",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]

crates/liyi/src/tree_path/lang_csharp.rs

@@ -1,6 +1,34 @@
 use super::LanguageConfig;
 
+use tree_sitter::Node;
+
 /// C# language configuration.
+/// Detect C# XML doc comments (`/// ...`).
+///
+/// C#'s tree-sitter grammar uses a uniform `comment` kind. The C# convention
+/// is `///` for XML documentation comments. We also check for `/**` as some
+/// projects use Javadoc-style. `attribute_list` and `modifier` siblings are
+/// skipped since attributes and access modifiers may appear between the doc
+/// comment and the declaration.
+fn csharp_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        match s.kind() {
+            "comment" => {
+                let text = &source[s.byte_range()];
+                if text.starts_with("///") || text.starts_with("/**") {
+                    return true;
+                }
+                sibling = s.prev_sibling();
+            }
+            "attribute_list" | "modifier" => {
+                sibling = s.prev_sibling();
+            }
+            _ => break,
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_c_sharp::LANGUAGE.into(),
     extensions: &["cs"],
@@ -20,7 +48,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body"],
     custom_name: None,
-    doc_comment_detector: None,
+    doc_comment_detector: Some(csharp_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_csharp.rs.liyi.jsonc

@@ -8,11 +8,11 @@
       "reviewed": true,
       "intent": "Define the C# language configuration for tree_path resolution: register tree-sitter-c-sharp grammar, .cs extension, kind mappings for fn/class/interface/enum/struct/namespace/constructor/property/record/delegate, using standard name field with no custom name extraction and no doc comment detector.",
       "source_span": [
-        4,
-        25
+        32,
+        53
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:6c0a47b592b0b8db93bcf44a875e7e2ba199ecbf1dbf767894797972b290398b",
+      "source_hash": "sha256:c95d3244ecb954bd82a2f3df065dda6f2d58c18a11c277fd9600990ef4127b48",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]

crates/liyi/src/tree_path/lang_kotlin.rs

@@ -39,6 +39,39 @@ fn kotlin_node_name(node: &Node, source: &str) -> Option<String> {
 }
 
 /// Kotlin language configuration.
+/// Detect Kotlin doc comments (`/** ... */` and `/// ...`).
+///
+/// Kotlin's tree-sitter-grammar distinguishes `block_comment` (for `/* */`
+/// and `/** */`) from `line_comment` (for `//` and `///`). We check for
+/// `/**` in block_comment and `///` in line_comment. `modifiers` and
+/// `annotation` siblings are skipped since they may appear between the
+/// doc comment and the declaration.
+fn kotlin_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        match s.kind() {
+            "block_comment" => {
+                let text = &source[s.byte_range()];
+                if text.starts_with("/**") {
+                    return true;
+                }
+                sibling = s.prev_sibling();
+            }
+            "line_comment" => {
+                let text = &source[s.byte_range()];
+                if text.starts_with("///") {
+                    return true;
+                }
+                sibling = s.prev_sibling();
+            }
+            "modifiers" | "annotation" | "annotation_list" => {
+                sibling = s.prev_sibling();
+            }
+            _ => break,
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_kotlin_ng::LANGUAGE.into(),
     extensions: &["kt", "kts"],
@@ -53,7 +86,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body", "class_body"],
     custom_name: Some(kotlin_node_name),
-    doc_comment_detector: None,
+    doc_comment_detector: Some(kotlin_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_kotlin.rs.liyi.jsonc

@@ -20,11 +20,11 @@
       "reviewed": true,
       "intent": "Define the Kotlin language configuration for tree_path resolution: register tree-sitter-kotlin-ng grammar, .kt/.kts extensions, kind mappings for fn/class/object/property/typealias, wire the custom kotlin_node_name callback, and no doc comment detector.",
       "source_span": [
-        42,
-        58
+        75,
+        91
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:c44502301bfedb32c49eb543da976a65e3dbe9b4c4d5fc7323a8c6541fe64423",
+      "source_hash": "sha256:868044c379a7f1e4d25d0ff81df7a26b4714d926b08b6cd708f8dbd0c292d325",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]

crates/liyi/src/tree_path/lang_objc.rs

@@ -68,6 +68,25 @@ fn objc_node_name(node: &Node, source: &str) -> Option<String> {
 }
 
 /// Objective-C language configuration.
+/// Detect Objective-C doc comments (`/** ... */` and `/// ...`).
+///
+/// ObjC's tree-sitter grammar uses a uniform `comment` kind. We check for
+/// `/**` (HeaderDoc/Javadoc) and `///` (Doxygen) prefixes.
+fn objc_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        if s.kind() == "comment" {
+            let text = &source[s.byte_range()];
+            if text.starts_with("/**") || text.starts_with("///") {
+                return true;
+            }
+            sibling = s.prev_sibling();
+        } else {
+            break;
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_objc::LANGUAGE.into(),
     extensions: &["m", "mm"],
@@ -86,7 +105,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body"],
     custom_name: Some(objc_node_name),
-    doc_comment_detector: None,
+    doc_comment_detector: Some(objc_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_objc.rs.liyi.jsonc

@@ -20,11 +20,11 @@
       "reviewed": true,
       "intent": "Define the Objective-C language configuration for tree_path resolution: register tree-sitter-objc grammar, .m/.mm extensions, kind mappings for fn/class/impl/protocol/method/method_decl/struct/enum/typedef, wire the custom objc_node_name callback, and no doc comment detector.",
       "source_span": [
-        71,
-        91
+        90,
+        110
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:34d6d1ac114bde885324b53fac1aaf04e217c62dac9f14c49c31582a51461770",
+      "source_hash": "sha256:036f2978cd0e40f1afde4dbca4940dac761e74bc316c96151dd1ead978b691ba",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]

crates/liyi/src/tree_path/lang_php.rs

@@ -20,6 +20,30 @@ fn php_node_name(node: &Node, source: &str) -> Option<String> {
 }
 
 /// PHP language configuration (PHP-only grammar, no HTML interleaving).
+/// Detect PHPDoc comments (`/** ... */`).
+///
+/// PHP's tree-sitter grammar uses a uniform `comment` kind. The PHP
+/// convention for documentation is `/**` (PHPDoc). `attribute_list`
+/// and `modifier` siblings are skipped.
+fn php_has_doc_comment(node: &Node, source: &str) -> bool {
+    let mut sibling = node.prev_sibling();
+    while let Some(s) = sibling {
+        match s.kind() {
+            "comment" => {
+                let text = &source[s.byte_range()];
+                if text.starts_with("/**") {
+                    return true;
+                }
+                sibling = s.prev_sibling();
+            }
+            "attribute_list" | "modifier" | "visibility_modifier" => {
+                sibling = s.prev_sibling();
+            }
+            _ => break,
+        }
+    }
+    false
+}
 pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     ts_language: || tree_sitter_php::LANGUAGE_PHP_ONLY.into(),
     extensions: &["php"],
@@ -37,7 +61,7 @@ pub(super) static CONFIG: LanguageConfig = LanguageConfig {
     name_overrides: &[],
     body_fields: &["body"],
     custom_name: Some(php_node_name),
-    doc_comment_detector: None,
+    doc_comment_detector: Some(php_has_doc_comment),
     transparent_kinds: &[],
 };
 

crates/liyi/src/tree_path/lang_php.rs.liyi.jsonc

@@ -20,11 +20,11 @@
       "reviewed": true,
       "intent": "Define the PHP language configuration for tree_path resolution: register tree-sitter-php PHP-only grammar (no HTML interleaving), .php extension, kind mappings for fn/class/method/interface/enum/trait/namespace/const, wire the custom php_node_name callback, and no doc comment detector.",
       "source_span": [
-        23,
-        42
+        47,
+        66
       ],
       "tree_path": "static.CONFIG",
-      "source_hash": "sha256:834be3bf9acb078c50e1628c4053e5bb63c524b36ca61a0fadee5ddbd2e75bc4",
+      "source_hash": "sha256:4d7ce38edb66173371fb985d94ec5b41be0eda73de3680b4cbbe36c3e7b2477d",
       "source_anchor": "pub(super) static CONFIG: LanguageConfig = LanguageConfig {"
     }
   ]