feat: add Perl language support and JCODEMUNCH_EXTRA_EXTENSIONS

.gitignore

@@ -55,3 +55,8 @@ dmypy.json
 
 # Local working notes (may contain sensitive info)
 prompt.md
+
+CLAUDE.md
+.vbw-planning
+
+.vbw-planning/

LANGUAGE_SUPPORT.md

@@ -13,7 +13,12 @@
 | PHP        | `.php`        | tree-sitter-php        | function, class, method, type (interface/trait/enum), constant | `#[Attribute]` | `/** */` PHPDoc | PHP 8+ attributes supported; language-file `<?php` tag required  |
 | Dart       | `.dart`       | tree-sitter-dart       | function, class (class/mixin/extension), method, type (enum/typedef) | `@annotation` | `///` doc comments | Constructors and top-level constants are not indexed               |
 | C#         | `.cs`         | tree-sitter-csharp     | class (class/record), method (method/constructor), type (interface/enum/struct/delegate) | `[Attribute]` | `/// <summary>` XML doc comments | Properties and `const` fields not indexed                          |
-| C          | `.c`, `.h`    | tree-sitter-c          | function, type (struct/enum/union), constant | —             | `/* */` and `//` comments | `#define` macros extracted as constants; no class/method hierarchy |
+| C          | `.c`          | tree-sitter-c          | function, type (struct/enum/union), constant | —             | `/* */` and `//` comments | `#define` macros extracted as constants; no class/method hierarchy |
+| C++        | `.cpp`, `.cc`, `.cxx`, `.hpp`, `.hh`, `.hxx`, `.h`* | tree-sitter-cpp | function, class, method, type (struct/enum/union/alias), constant | — | `/* */` and `//` comments | Namespace symbols are used for qualification but not emitted as standalone symbols |
+| Elixir     | `.ex`, `.exs` | tree-sitter-elixir | class (defmodule/defimpl), type (defprotocol/@type/@callback), method (def/defp/defmacro/defguard inside module), function (top-level def) | — | `@doc`/`@moduledoc` strings | Homoiconic grammar; custom walker required. `defstruct`, `use`, `import`, `alias` not indexed |
+| Ruby       | `.rb`, `.rake` | tree-sitter-ruby  | class, type (module), method (instance + `self.` singleton), function (top-level def) | — | `#` preceding comments | `attr_accessor`, constants, and `include`/`extend` not indexed |
+
+\* `.h` uses C++ parsing first, then falls back to C when no C++ symbols are extracted.
 
 ---
 
@@ -105,3 +110,21 @@ print_tree(tree.root_node)
 ```
 
 This inspection process helps identify the correct `symbol_node_types`, `name_fields`, and extraction rules when adding support for a new language.
+
+
+## Configuration
+
+### `JCODEMUNCH_EXTRA_EXTENSIONS`
+
+Map additional file extensions to languages at startup without modifying source:
+
+```
+JCODEMUNCH_EXTRA_EXTENSIONS=".cgi:perl,.psgi:perl,.mjs:javascript"
+```
+
+- Comma-separated `.ext:lang` pairs
+- Overrides built-in mappings on collision
+- Unknown languages and malformed entries are skipped with a warning
+- Valid language names: `python`, `javascript`, `typescript`, `go`, `rust`, `java`, `php`, `dart`, `csharp`, `c`, `cpp`, `swift`, `elixir`, `ruby`, `perl`
+
+Set via `.mcp.json` `env` block or any environment mechanism supported by your MCP client.

README.md

@@ -295,7 +295,12 @@ Every tool response includes a `_meta` envelope with timing, token savings, and
 | PHP        | `.php`        | function, class, method, type, constant |
 | Dart       | `.dart`       | function, class, method, type           |
 | C#         | `.cs`         | class, method, type, record             |
-| C          | `.c`, `.h`    | function, type, constant                |
+| C          | `.c`          | function, type, constant                |
+| C++        | `.cpp`, `.cc`, `.cxx`, `.hpp`, `.hh`, `.hxx`, `.h`* | function, class, method, type, constant |
+| Elixir     | `.ex`, `.exs` | class (module/impl), type (protocol/@type/@callback), method, function |
+| Ruby       | `.rb`, `.rake` | class, type (module), method, function |
+
+\* `.h` is parsed as C++ first, then falls back to C when no C++ symbols are extracted.
 
 See LANGUAGE_SUPPORT.md for full semantics.
 
@@ -373,6 +378,7 @@ For **LM Studio**, ensure the Local Server is running (usually on port 1234):
 | `OPENAI_MODEL`              | Model name for local LLMs (default: `qwen3-coder`) | No |
 | `OPENAI_TIMEOUT`            | Timeout in seconds for local requests (default: `60.0`) | No |
 | `CODE_INDEX_PATH`           | Custom cache path         | No       |
+| `JCODEMUNCH_MAX_INDEX_FILES`| Maximum files to index per repo/folder (default: `500`) | No |
 | `JCODEMUNCH_SHARE_SAVINGS`  | Set to `0` to disable anonymous community token savings reporting | No       |
 | `JCODEMUNCH_LOG_LEVEL`      | Log level: `DEBUG`, `INFO`, `WARNING`, `ERROR` (default: `WARNING`) | No       |
 | `JCODEMUNCH_LOG_FILE`       | Path to log file. If unset, logs go to stderr. Use a file to avoid polluting MCP stdio. | No       |

SECURITY.md

@@ -56,7 +56,7 @@ When a secret file is detected, a warning is included in the indexing response.
 
 * **Default maximum:** 500 KB per file (configurable via `max_file_size`).
 * Files exceeding the limit are skipped during discovery.
-* A configurable **file count limit** (default: 500 files) prevents runaway indexing of extremely large repositories.
+* A configurable **file count limit** (default: 500 files) prevents runaway indexing of extremely large repositories. Can be overridden using the `JCODEMUNCH_MAX_INDEX_FILES` environment variable.
 
 ---
 

SPEC.md

@@ -166,7 +166,7 @@ class Symbol:
     name: str                # Symbol name
     qualified_name: str      # Dot-separated with parent context
     kind: str                # function | class | method | constant | type
-    language: str            # python | javascript | typescript | go | rust | java | php | c
+    language: str            # python | javascript | typescript | go | rust | java | php | dart | csharp | c | cpp
     signature: str           # Full signature line(s)
     content_hash: str = ""   # SHA-256 of source bytes (drift detection)
     docstring: str = ""
@@ -212,7 +212,7 @@ Recursive directory walk with the full security pipeline.
 
 ### Filtering Pipeline (Both Paths)
 
-1. **Extension filter** — must be in `LANGUAGE_EXTENSIONS` (.py, .js, .jsx, .ts, .tsx, .go, .rs, .java, .php, .c, .h)
+1. **Extension filter** — must be in `LANGUAGE_EXTENSIONS` (.py, .js, .jsx, .ts, .tsx, .go, .rs, .java, .php, .c, .h, .cpp, .cc, .cxx, .hpp, .hh, .hxx)
 2. **Skip patterns** — `node_modules/`, `vendor/`, `.git/`, `build/`, `dist/`, lock files, minified files, etc.
 3. **`.gitignore`** — respected via the `pathspec` library
 4. **Secret detection** — `.env`, `*.pem`, `*.key`, `*.p12`, credentials files excluded

USER_GUIDE.md

@@ -244,7 +244,7 @@ To disable, set `JCODEMUNCH_SHARE_SAVINGS=0` in your MCP server env:
 Check the URL format (`owner/repo` or full GitHub URL). For private repositories, set `GITHUB_TOKEN`.
 
 **"No source files found"**
-The repository may not contain supported language files (`.py`, `.js`, `.ts`, `.go`, `.rs`, `.java`, `.c`, `.h`), or files may be excluded by skip patterns.
+The repository may not contain supported language files (`.py`, `.js`, `.ts`, `.go`, `.rs`, `.java`, `.c`, `.h`, `.cpp`, `.cc`, `.cxx`, `.hpp`, `.hh`, `.hxx`), or files may be excluded by skip patterns.
 
 **Rate limiting**
 Set `GITHUB_TOKEN` to increase GitHub API limits (5,000 requests/hour vs 60 unauthenticated).

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "jcodemunch-mcp"
-version = "0.2.13"
+version = "0.2.20"
 description = "Token-efficient MCP server for source code exploration via tree-sitter AST parsing"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -33,3 +33,10 @@ exclude = [".claude/"]
 [tool.pytest.ini_options]
 testpaths = ["tests"]
 asyncio_mode = "auto"
+
+[dependency-groups]
+dev = [
+    "pytest>=9.0.2",
+    "pytest-asyncio>=1.3.0",
+    "pytest-cov>=7.0.0",
+]

src/jcodemunch_mcp/__init__.py

@@ -1,3 +1,8 @@
 """github-codemunch-mcp - Token-efficient MCP server for GitHub source code exploration."""
 
-__version__ = "0.1.0"
+from importlib.metadata import PackageNotFoundError, version
+
+try:
+    __version__ = version("jcodemunch-mcp")
+except PackageNotFoundError:
+    __version__ = "unknown"