From 41bfbbc3867cb2dbb725caaee7081ad0e346e579 Mon Sep 17 00:00:00 2001
From: 7000pctAUTO <lukems3823@gmail.com>
Date: Tue, 3 Feb 2026 10:30:15 +0000
Subject: [PATCH] Initial upload of ai-code-audit-cli project

---
 src/utils/language_detector.py | 155 +++++++++++++++++++++++++++++++++
 1 file changed, 155 insertions(+)
 create mode 100644 src/utils/language_detector.py

diff --git a/src/utils/language_detector.py b/src/utils/language_detector.py
new file mode 100644
index 0000000..2416b79
--- /dev/null
+++ b/src/utils/language_detector.py
@@ -0,0 +1,155 @@
+"""Language detection utilities for AI Code Audit CLI."""
+
+from pathlib import Path
+from typing import Optional
+
+
+class LanguageDetector:
+    """Detect programming language of source files."""
+
+    LANGUAGE_MAP = {
+        ".py": "python",
+        ".pyw": "python",
+        ".pyx": "python",
+        ".js": "javascript",
+        ".mjs": "javascript",
+        ".jsx": "javascript",
+        ".ts": "typescript",
+        ".tsx": "typescript",
+        ".mts": "typescript",
+    }
+
+    PYTHON_SHEBANG = "#!"
+    JS_SHEBANG = "#!"
+
+    def __init__(self):
+        """Initialize the language detector."""
+        self.language_map = self.LANGUAGE_MAP
+
+    def detect(self, file_path: Path | str) -> "Language":
+        """Detect the language of a file."""
+        path = Path(file_path) if isinstance(file_path, str) else file_path
+
+        ext = path.suffix.lower()
+        if ext in self.language_map:
+            return Language(self.language_map[ext])
+
+        try:
+            content = path.read_text(encoding="utf-8", errors="replace")
+            return self.detect_from_content(content, path.name)
+        except Exception:
+            return Language("unknown")
+
+    def detect_from_content(
+        self, content: str, filename: str = ""
+    ) -> "Language":
+        """Detect language from file content."""
+        if not content:
+            return Language("unknown")
+
+        first_line = content.split('\n')[0].strip()
+
+        if first_line.startswith("#!"):
+            shebang = first_line[2:].strip().split()[0] if len(first_line) > 2 else ""
+            if "python" in shebang.lower():
+                return Language("python")
+            elif "node" in shebang.lower() or "js" in shebang.lower():
+                return Language("javascript")
+
+        if filename.endswith(".py") or filename.endswith(".pyw"):
+            return Language("python")
+
+        if filename.endswith((".ts", ".tsx")):
+            return Language("typescript")
+
+        if filename.endswith((".js", ".jsx")):
+            return Language("javascript")
+
+        keywords = self._detect_keywords(content)
+        if keywords:
+            return Language(keywords)
+
+        return Language("unknown")
+
+    def _detect_keywords(self, content: str) -> Optional[str]:
+        """Detect language from keywords in content."""
+        content_lower = content.lower()
+
+        python_indicators = [
+            "import ",
+            "from ",
+            "def ",
+            "class ",
+            "if __name__",
+            "print(",
+            "return ",
+            "except ",
+            "try:",
+            "with open",
+            "lambda ",
+        ]
+
+        js_indicators = [
+            "const ",
+            "let ",
+            "=>",
+            "function ",
+            "require(",
+            "module.exports",
+            "export ",
+            "import ",
+            "console.log",
+            "async function",
+            "await ",
+        ]
+
+        python_score = sum(1 for ind in python_indicators if ind in content_lower)
+        js_score = sum(1 for ind in js_indicators if ind in content_lower)
+
+        if python_score > js_score:
+            return "python"
+        elif js_score > python_score:
+            return "javascript"
+
+        return None
+
+    def get_supported_languages(self) -> list[str]:
+        """Get list of supported languages."""
+        return ["python", "javascript", "typescript"]
+
+    def is_supported(self, language: str) -> bool:
+        """Check if a language is supported."""
+        return language.lower() in self.get_supported_languages()
+
+
+class Language:
+    """Represents a detected programming language."""
+
+    def __init__(self, value: str):
+        """Initialize the language."""
+        self._value = value.lower()
+
+    @property
+    def value(self) -> str:
+        """Get the language value."""
+        return self._value
+
+    def __eq__(self, other: object) -> bool:
+        """Check equality with another Language or string."""
+        if isinstance(other, Language):
+            return self._value == other._value
+        if isinstance(other, str):
+            return self._value == other.lower()
+        return False
+
+    def __hash__(self) -> int:
+        """Hash for use in sets and dicts."""
+        return hash(self._value)
+
+    def __str__(self) -> str:
+        """String representation."""
+        return self._value
+
+    def __repr__(self) -> str:
+        """Repr representation."""
+        return f"Language('{self._value}')"