git-diff-explainer-cli/src/gdiffer/language_detector.py

"""Language detection for code files."""



class LanguageDetector:
    """Detects programming language from file extensions and content."""

    EXTENSION_MAP = {
        'py': 'python',
        'pyw': 'python',
        'pyx': 'python',
        'js': 'javascript',
        'mjs': 'javascript',
        'cjs': 'javascript',
        'jsx': 'javascript',
        'ts': 'typescript',
        'tsx': 'typescript',
        'mts': 'typescript',
        'cts': 'typescript',
        'java': 'java',
        'kt': 'kotlin',
        'kts': 'kotlin',
        'go': 'go',
        'rs': 'rust',
        'c': 'c',
        'h': 'c',
        'cpp': 'cpp',
        'cc': 'cpp',
        'cxx': 'cpp',
        'hpp': 'cpp',
        'hxx': 'cpp',
        'cs': 'csharp',
        'rb': 'ruby',
        'erb': 'ruby',
        'php': 'php',
        'swift': 'swift',
        'm': 'objective-c',
        'mm': 'objective-c',
        'scala': 'scala',
        'sc': 'scala',
        'jl': 'julia',
        'r': 'r',
        'R': 'r',
        'lua': 'lua',
        'pl': 'perl',
        'pm': 'perl',
        'sql': 'sql',
        'sh': 'bash',
        'bash': 'bash',
        'zsh': 'bash',
        'fish': 'bash',
        'yaml': 'yaml',
        'yml': 'yaml',
        'json': 'json',
        'xml': 'xml',
        'html': 'html',
        'htm': 'html',
        'css': 'css',
        'scss': 'scss',
        'sass': 'sass',
        'less': 'less',
        'md': 'markdown',
        'markdown': 'markdown',
        'txt': 'text',
        'dockerfile': 'dockerfile',
        'Dockerfile': 'dockerfile',
    }

    CONTENT_PATTERNS = {
        'python': [
            r'^import\s+\w+',
            r'^from\s+\w+\s+import',
            r'^def\s+\w+\s*\(',
            r'^class\s+\w+\s*[:\(]',
            r'^if\s+__name__\s*==\s*['"]__main__['"]',
        ],
        'javascript': [
            r'^const\s+\w+\s*=',
            r'^let\s+\w+\s*=',
            r'^var\s+\w+\s*=',
            r'^function\s+\w+\s*\(',
            r'=>\s*\{',
            r'import\s+.*\s+from',
            r'export\s+(default\s+)?',
        ],
        'typescript': [
            r'^interface\s+\w+\s*\{',
            r'^type\s+\w+\s*=',
            r':\s*(string|number|boolean|any|void|null|undefined)',
            r'<[A-Z]\w*>',
        ],
        'java': [
            r'^package\s+[\w.]+;',
            r'^import\s+[\w.]+;',
            r'^public\s+(class|interface|enum)\s+\w+',
            r'^private\s+(static\s+)?(final\s+)?\w+\s+\w+;',
        ],
        'go': [
            r'^package\s+\w+',
            r'^import\s*\(',
            r'func\s+\w+\s*\(',
            r':=',
            r'go\s+func',
        ],
        'rust': [
            r'^fn\s+\w+\s*\(',
            r'^impl\s+\w+',
            r'^struct\s+\w+',
            r'^enum\s+\w+',
            r'let\s+mut\s+\w+',
            r'->\s*\w+',
        ],
        'c': [
            r'#include\s*<',
            r'#include\s*"',
            r'int\s+main\s*\(',
            r'struct\s+\w+\s*\{',
            r'void\s+\?\s*\w+\s*\(',
        ],
        'cpp': [
            r'#include\s*<',
            r'#include\s*"',
            r'class\s+\w+\s*(:\s*public)?',
            r'std::\w+',
            r'using\s+namespace\s+std',
        ],
        'ruby': [
            r'^require\s+['"]',
            r'^class\s+\w+(\s*<\s*\w+)?',
            r'^module\s+\w+',
            r'def\s+\w+',
            r'puts\s+',
            r'puts!',
        ],
        'php': [
            r'\<\?php',
            r'\$\w+\s*=',
            r'function\s+\w+\s*\(',
            r'class\s+\w+\s*\{',
        ],
    }

    def __init__(self):
        self._tree_sitter_languages = {}

    def detect_from_filename(self, filename: str) -> str | None:
        """Detect language from file extension."""
        if '.' not in filename:
            return None

        ext = filename.rsplit('.', 1)[-1].lower()
        return self.EXTENSION_MAP.get(ext)

    def detect_from_content(self, content: str) -> str | None:
        """Detect language from file content patterns."""
        first_lines = '\n'.join(content.splitlines()[:50])

        scores: dict[str, int] = {}

        for lang, patterns in self.CONTENT_PATTERNS.items():
            score = 0
            for pattern in patterns:
                import re
                matches = len(re.findall(pattern, first_lines, re.MULTILINE))
                score += matches

            if score > 0:
                scores[lang] = score

        if scores:
            best_lang = max(scores, key=scores.get)
            return best_lang

        return None

    def detect(self, filename: str, content: str = "") -> str:
        """Detect language from filename and optionally content."""
        ext_lang = self.detect_from_filename(filename)

        if ext_lang and ext_lang not in ['text', 'markdown', 'json', 'yaml', 'xml', 'html', 'css', 'dockerfile']:
            if content:
                content_lang = self.detect_from_content(content)
                if content_lang and content_lang != ext_lang:
                    return content_lang
            return ext_lang

        if content:
            content_lang = self.detect_from_content(content)
            if content_lang:
                return content_lang

        return ext_lang or "text"

    def get_supported_languages(self) -> list[str]:
        """Return list of supported languages."""
        return sorted(set(self.EXTENSION_MAP.values()))

    def is_language_supported(self, language: str) -> bool:
        """Check if a language is supported."""
        return language in self.get_supported_languages()


def detect_language(filename: str, content: str = "") -> str:
    """Detect programming language from filename and content."""
    detector = LanguageDetector()
    return detector.detect(filename, content)