feat: add optional parse error locations

.github/workflows/ci.yml

@@ -168,6 +168,7 @@ jobs:
         working-directory: rust
         run: cargo test
 
+
       - name: Test (no default features)
         working-directory: rust
         run: cargo test --no-default-features

README.md

@@ -132,6 +132,24 @@ const std::optional<lexer_error>& get_last_error();
 
 Returns the last parse error, if any.
 
+### `lexer::get_last_error_location`
+
+```cpp
+const std::optional<error_location>& get_last_error_location();
+```
+
+Returns the location of the last parse error, if available. Location tracking
+is best-effort and may be unavailable.
+
+### `lexer::error_location`
+
+```cpp
+struct error_location {
+  uint32_t line;    // 1-based
+  uint32_t column;  // 1-based (byte-oriented)
+};
+```
+
 ## C API
 
 merve provides a C API (`merve_c.h`) for use from C programs, FFI bindings, or any language that can call C functions. The C API is compiled into the merve library alongside the C++ implementation.
@@ -141,11 +159,13 @@ merve provides a C API (`merve_c.h`) for use from C programs, FFI bindings, or a
 ```c
 #include "merve_c.h"
 #include <stdio.h>
+#include <string.h>
 
 int main(void) {
   const char* source = "exports.foo = 1;\nexports.bar = 2;\n";
 
-  merve_analysis result = merve_parse_commonjs(source, strlen(source));
+  merve_error_loc err_loc = {0, 0};
+  merve_analysis result = merve_parse_commonjs(source, strlen(source), &err_loc);
 
   if (merve_is_valid(result)) {
     size_t count = merve_get_exports_count(result);
@@ -157,6 +177,9 @@ int main(void) {
     }
   } else {
     printf("Parse error: %d\n", merve_get_last_error());
+    if (err_loc.line != 0) {
+      printf("  at line %u, column %u\n", err_loc.line, err_loc.column);
+    }
   }
 
   merve_free(result);
@@ -180,12 +203,13 @@ Found 2 exports:
 | `merve_string` | Non-owning string reference (`data` + `length`). Not null-terminated. |
 | `merve_analysis` | Opaque handle to a parse result. Must be freed with `merve_free()`. |
 | `merve_version_components` | Struct with `major`, `minor`, `revision` fields. |
+| `merve_error_loc` | Error location (`line`, `column`). `{0,0}` means unavailable. |
 
 #### Functions
 
 | Function | Description |
 |----------|-------------|
-| `merve_parse_commonjs(input, length)` | Parse CommonJS source. Returns a handle (NULL only on OOM). |
+| `merve_parse_commonjs(input, length, out_err)` | Parse CommonJS source and optionally fill error location. Returns a handle (NULL only on OOM). |
 | `merve_is_valid(result)` | Check if parsing succeeded. NULL-safe. |
 | `merve_free(result)` | Free a parse result. NULL-safe. |
 | `merve_get_exports_count(result)` | Number of named exports found. |
@@ -198,6 +222,9 @@ Found 2 exports:
 | `merve_get_version()` | Version string (e.g. `"1.0.1"`). |
 | `merve_get_version_components()` | Version as `{major, minor, revision}`. |
 
+On parse failure, `merve_parse_commonjs` writes a non-zero location when
+`out_err` is non-NULL and the location is available.
+
 #### Error Constants
 
 | Constant | Value | Description |

include/merve/parser.h

@@ -3,6 +3,7 @@
 
 #include "merve/version.h"
 
+#include <cstddef>
 #include <cstdint>
 #include <optional>
 #include <string>
@@ -37,6 +38,17 @@ enum lexer_error {
   TEMPLATE_NEST_OVERFLOW, ///< Template literal nesting too deep
 };
 
+/**
+ * @brief Source location information for a parse error.
+ *
+ * - line and column are 1-based.
+ * - column is byte-oriented.
+ */
+struct error_location {
+  uint32_t line;
+  uint32_t column;
+};
+
 /**
  * @brief Type alias for export names.
  *
@@ -146,6 +158,18 @@ std::optional<lexer_analysis> parse_commonjs(std::string_view file_contents);
  */
 const std::optional<lexer_error>& get_last_error();
 
+/**
+ * @brief Get the location of the last failed parse operation.
+ *
+ * @return const std::optional<error_location>& The last error location, or
+ *         std::nullopt if unavailable.
+ *
+ * @note This is global state and may be overwritten by subsequent calls
+ *       to parse_commonjs().
+ * @note Location tracking is best-effort and may be unavailable.
+ */
+const std::optional<error_location>& get_last_error_location();
+
 }  // namespace lexer
 
 #endif  // MERVE_PARSER_H

include/merve_c.h

@@ -39,6 +39,19 @@ typedef struct {
   int revision;
 } merve_version_components;
 
+/**
+ * @brief Source location for a parse error.
+ *
+ * - line and column are 1-based.
+ * - column is byte-oriented.
+ *
+ * A zeroed location (`{0, 0}`) means the location is unavailable.
+ */
+typedef struct {
+  uint32_t line;
+  uint32_t column;
+} merve_error_loc;
+
 /* Error codes corresponding to lexer::lexer_error values. */
 #define MERVE_ERROR_TODO 0
 #define MERVE_ERROR_UNEXPECTED_PAREN 1
@@ -59,20 +72,32 @@ extern "C" {
 #endif
 
 /**
- * Parse CommonJS source code and extract export information.
+ * Parse CommonJS source code and optionally return error location.
  *
  * The source buffer must remain valid while accessing string_view-backed
  * export names from the returned handle.
  *
+ * If @p out_err is non-NULL, it is always written:
+ * - On success: set to {0, 0}.
+ * - On parse failure with known location: set to that location.
+ * - On parse failure without available location: set to {0, 0}.
+ *
  * You must call merve_free() on the returned handle when done.
  *
- * @param input  Pointer to the JavaScript source (need not be null-terminated).
- *               NULL is treated as an empty string.
- * @param length Length of the input in bytes.
+ * @param input   Pointer to the JavaScript source (need not be
+ *                null-terminated). NULL is treated as an empty string.
+ * @param length  Length of the input in bytes.
+ * @param out_err Optional output pointer for parse error location.
  * @return A handle to the parse result, or NULL on out-of-memory.
  *         Use merve_is_valid() to check if parsing succeeded.
  */
-merve_analysis merve_parse_commonjs(const char* input, size_t length);
+#ifdef __cplusplus
+merve_analysis merve_parse_commonjs(const char* input, size_t length,
+                                    merve_error_loc* out_err = nullptr);
+#else
+merve_analysis merve_parse_commonjs(const char* input, size_t length,
+                                    merve_error_loc* out_err);
+#endif
 
 /**
  * Check whether the parse result is valid (parsing succeeded).
@@ -165,7 +190,7 @@ const char* merve_get_version(void);
 merve_version_components merve_get_version_components(void);
 
 #ifdef __cplusplus
-}  /* extern "C" */
+} /* extern "C" */
 #endif
 
 #endif /* MERVE_C_H */

rust/README.md

@@ -51,7 +51,7 @@ merve = { version = "...", features = ["libcpp"] }
 ### `parse_commonjs`
 
 ```rust
-pub fn parse_commonjs(source: &str) -> Result<Analysis<'_>, LexerError>
+pub fn parse_commonjs(source: &str) -> Result<Analysis<'_>, LocatedLexerError>
 ```
 
 Parse CommonJS source code and extract export information. The returned
@@ -100,6 +100,17 @@ Returned when the input contains ESM syntax or malformed constructs:
 
 `LexerError` implements `Display` and, with the `std` feature, `std::error::Error`.
 
+### `LocatedLexerError`
+
+```rust
+pub struct LocatedLexerError {
+    pub kind: LexerError,
+    pub location: Option<ErrorLocation>,
+}
+```
+
+`ErrorLocation` uses 1-based `line`/`column` (byte-oriented column).
+
 ### Versioning helpers
 
 ```rust

rust/build.rs

@@ -133,8 +133,8 @@ fn amalgamate_file(
     out.push_str(&format!("/* end file {filename} */\n"));
 }
 
-/// When building inside the merve repository, produce the three amalgamated
-/// files in `deps/`:  merve.h, merve.cpp, merve_c.h.
+/// When building inside the merve repository, produce the vendored
+/// sources in `deps/`: merve.h, merve.cpp, merve_c.cpp, merve_c.h.
 fn amalgamate_from_repo(project_root: &Path, deps: &Path) {
     let include_path = project_root.join("include");
     let source_path = project_root.join("src");
@@ -159,21 +159,23 @@ fn amalgamate_from_repo(project_root: &Path, deps: &Path) {
     );
     fs::write(deps.join("merve.h"), &header).expect("failed to write deps/merve.h");
 
-    // 2. Amalgamate merve.cpp (parser.cpp + merve_c.cpp with includes resolved).
+    // 2. Amalgamate merve.cpp (parser.cpp with includes resolved).
     let mut source = String::from("#include \"merve.h\"\n\n");
-    for cpp in &["parser.cpp", "merve_c.cpp"] {
-        amalgamate_file(
-            &include_path,
-            &source_path,
-            &source_path,
-            cpp,
-            &mut source,
-            &mut included,
-        );
-    }
+    amalgamate_file(
+        &include_path,
+        &source_path,
+        &source_path,
+        "parser.cpp",
+        &mut source,
+        &mut included,
+    );
     fs::write(deps.join("merve.cpp"), &source).expect("failed to write deps/merve.cpp");
 
-    // 3. Copy merve_c.h verbatim (standalone C header).
+    // 3. Copy merve_c.cpp verbatim (C API implementation).
+    fs::copy(source_path.join("merve_c.cpp"), deps.join("merve_c.cpp"))
+        .expect("failed to copy merve_c.cpp");
+
+    // 4. Copy merve_c.h verbatim (standalone C header).
     fs::copy(include_path.join("merve_c.h"), deps.join("merve_c.h"))
         .expect("failed to copy merve_c.h");
 }
@@ -207,15 +209,17 @@ fn main() {
         }
     }
 
-    // Both in-repo and published crate use the same layout: merve.cpp + merve.h + merve_c.h
+    // Both in-repo and published crate use the same layout:
+    // merve.cpp + merve_c.cpp + merve.h + merve_c.h
     assert!(
-        deps.join("merve.cpp").exists(),
+        deps.join("merve.cpp").exists() && deps.join("merve_c.cpp").exists(),
         "No C++ sources found in deps/. \
-         When building outside the repository, deps/ must contain the amalgamated sources."
+         When building outside the repository, deps/ must contain the vendored sources."
     );
 
     let mut build = cc::Build::new();
     build.file(deps.join("merve.cpp"));
+    build.file(deps.join("merve_c.cpp"));
     build.include(&deps);
     build.cpp(true).std("c++20").warnings(false);