42 c project have list allow function per project (of course you can use your function),…
- User
42 c project have list allow function per project (of course you can use your function), libft is allow or not,.. i want write a robust linter to check it
Assistant
A robust linter for 42 projects sounds like a useful tool! You'll need to handle the varying allowed functions across different projects and the libft situation.
Here's a structure I'd recommend:
Core Architecture:
// Project configuration structure typedef struct s_project_config { char *name; char **allowed_functions; bool libft_allowed; char **forbidden_patterns; int max_functions_per_file; int max_lines_per_function; } t_project_config;Key Components:
Function Detection: Parse C files to extract function calls using regex or a simple parser. You'll need to handle function calls in various contexts (direct calls, function pointers, etc.).
Configuration System: Create JSON or YAML config files for each project:
{ "libft": { "allowed_functions": ["write", "malloc", "free"], "libft_allowed": false, "max_functions_per_file": 5 }, "get_next_line": { "allowed_functions": ["read", "malloc", "free"], "libft_allowed": true } }Libft Handling: When libft is allowed, you'll need to either parse the student's libft directory to see which functions they've implemented, or maintain a standard list of libft functions.
Detection Strategies:
- Parse
#includestatements to catch unauthorized headers - Use AST parsing or regex to find function calls
- Check for system calls that might be disguised
- Validate Makefile rules and linking
Implementation Tips:
- Start with a simple regex-based approach, then enhance with proper parsing
- Handle edge cases like function pointers and macros
- Consider using existing C parsers like libclang if you want robust AST analysis
- Make it extensible so adding new projects is just configuration
Would you like me to elaborate on any specific part, like the parsing logic or configuration management?
- Parse
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.