app/validators/code_checker.py

"""
Validate generated code for common issues.
"""

import ast
import re
from typing import Optional


class CodeChecker:
    """Validate generated Space code for correctness and common issues."""

    def check(self, files: dict, sdk: str) -> dict:
        """
        Validate all files in the repo.

        Returns dict with:
            valid: bool
            errors: list of error strings
            warnings: list of warning strings
            file_checks: dict of filename -> {valid, issues}
        """
        errors = []
        warnings = []
        file_checks = {}

        for filename, content in files.items():
            check = self._check_file(filename, content, sdk)
            file_checks[filename] = check
            errors.extend(check.get("errors", []))
            warnings.extend(check.get("warnings", []))

        # Cross-file checks
        cross_issues = self._cross_file_checks(files, sdk)
        errors.extend(cross_issues.get("errors", []))
        warnings.extend(cross_issues.get("warnings", []))

        return {
            "valid": len(errors) == 0,
            "errors": errors,
            "warnings": warnings,
            "file_checks": file_checks,
        }

    def _check_file(self, filename: str, content: str, sdk: str) -> dict:
        """Check a single file."""
        errors = []
        warnings = []

        if not content or not content.strip():
            errors.append(f"{filename}: File is empty")
            return {"valid": False, "errors": errors, "warnings": warnings}

        if filename.endswith(".py"):
            py_result = self._check_python(filename, content)
            errors.extend(py_result["errors"])
            warnings.extend(py_result["warnings"])
        elif filename == "Dockerfile":
            docker_result = self._check_dockerfile(content)
            errors.extend(docker_result["errors"])
            warnings.extend(docker_result["warnings"])
        elif filename == "requirements.txt":
            req_result = self._check_requirements(content)
            errors.extend(req_result["errors"])
            warnings.extend(req_result["warnings"])
        elif filename == "README.md":
            if "---" not in content:
                warnings.append("README.md: Missing YAML frontmatter")
        elif filename.endswith(".html"):
            html_result = self._check_html(filename, content)
            errors.extend(html_result["errors"])
            warnings.extend(html_result["warnings"])

        valid = len(errors) == 0
        return {"valid": valid, "errors": errors, "warnings": warnings}

    def _check_python(self, filename: str, content: str) -> dict:
        """Validate Python code syntax and common patterns."""
        errors = []
        warnings = []

        # Syntax check
        try:
            ast.parse(content)
        except SyntaxError as e:
            errors.append(f"{filename}: Python syntax error at line {e.lineno}: {e.msg}")
            return {"errors": errors, "warnings": warnings}

        # Check for common issues
        if "import " not in content and "from " not in content:
            warnings.append(f"{filename}: No imports found")

        # Check for dangerous patterns
        dangerous_patterns = [
            (r"os\.system\(", "os.system() call found - potential security risk"),
            (r"eval\(", "eval() call found - potential security risk"),
            (r"exec\(", "exec() call found - potential security risk"),
            (r"__import__\(", "__import__() call found - potential security risk"),
        ]
        for pattern, msg in dangerous_patterns:
            if re.search(pattern, content):
                warnings.append(f"{filename}: {msg}")

        # Check for hardcoded tokens/secrets
        secret_patterns = [
            (r'(?:token|key|secret|password)\s*=\s*["\'][^"\']{10,}["\']', "Possible hardcoded secret"),
        ]
        for pattern, msg in secret_patterns:
            if re.search(pattern, content, re.IGNORECASE):
                warnings.append(f"{filename}: {msg}")

        return {"errors": errors, "warnings": warnings}

    def _check_dockerfile(self, content: str) -> dict:
        """Validate Dockerfile content."""
        errors = []
        warnings = []

        if "FROM" not in content:
            errors.append("Dockerfile: Missing FROM instruction")
        if "EXPOSE" not in content:
            warnings.append("Dockerfile: Missing EXPOSE instruction")
        if "7860" not in content:
            warnings.append("Dockerfile: Port 7860 not found (required for HF Spaces)")
        if "CMD" not in content and "ENTRYPOINT" not in content:
            errors.append("Dockerfile: Missing CMD or ENTRYPOINT")

        return {"errors": errors, "warnings": warnings}

    def _check_requirements(self, content: str) -> dict:
        """Validate requirements.txt."""
        errors = []
        warnings = []

        lines = [l.strip() for l in content.strip().split("\n") if l.strip() and not l.strip().startswith("#")]
        if not lines:
            warnings.append("requirements.txt: No dependencies listed")

        for line in lines:
            # Basic format check
            if " " in line and ";" not in line and "#" not in line:
                warnings.append(f"requirements.txt: Suspicious line: '{line}'")

        return {"errors": errors, "warnings": warnings}

    def _check_html(self, filename: str, content: str) -> dict:
        """Basic HTML validation."""
        errors = []
        warnings = []

        if "<html" not in content.lower() and "<!doctype" not in content.lower():
            warnings.append(f"{filename}: Missing <html> tag or DOCTYPE")

        # Check for unclosed tags (very basic)
        for tag in ["html", "head", "body"]:
            open_count = len(re.findall(f"<{tag}[\\s>]", content, re.IGNORECASE))
            close_count = len(re.findall(f"</{tag}>", content, re.IGNORECASE))
            if open_count > close_count:
                warnings.append(f"{filename}: Unclosed <{tag}> tag")

        return {"errors": errors, "warnings": warnings}

    def _cross_file_checks(self, files: dict, sdk: str) -> dict:
        """Perform checks across multiple files."""
        errors = []
        warnings = []

        if sdk == "gradio":
            if "app.py" not in files:
                errors.append("Missing app.py (required for Gradio Spaces)")
            if "requirements.txt" not in files:
                warnings.append("Missing requirements.txt")
            if "README.md" not in files:
                warnings.append("Missing README.md")

            # Check that requirements includes gradio
            req = files.get("requirements.txt", "")
            if "gradio" not in req.lower():
                warnings.append("requirements.txt: 'gradio' not listed as dependency")

        elif sdk == "docker":
            if "Dockerfile" not in files:
                errors.append("Missing Dockerfile (required for Docker Spaces)")
            if "README.md" not in files:
                warnings.append("Missing README.md")

        elif sdk == "static":
            if "index.html" not in files:
                errors.append("Missing index.html (required for Static Spaces)")

        return {"errors": errors, "warnings": warnings}