Files
minishell/AGENTS.md
2026-02-12 20:32:12 +01:00

73 lines
5.1 KiB
Markdown

# Repository Guidelines
## Project Structure & Module Organization
- `src/` holds the shell implementation, with submodules for `builtins/`, `executor/`, `parser/`, and `variables/`.
- `src/builtins/` currently includes: `cd`, `echo`, `env`, `exit`, `export`, `pwd`, `unset`.
- `src/executor/` includes pipeline/process orchestration plus file redirections in `redirections.c`.
- `include/` contains public headers used across modules.
- `lib/` is populated at build time with third-party 42 libraries (libft, get_next_line, ft_printf, ft_args).
- `docs/` stores project references and manual test notes (see `docs/tests.md`).
- `build/` is the object output directory generated by `make`.
## Build, Test, and Development Commands
- `make` or `make all`: build `minishell` (auto-clones required libs into `lib/`).
- `make clean`: remove objects in `build/`.
- `make fclean`: remove objects and the `minishell` binary (also cleans libs).
- `make re`: full rebuild.
- `./minishell`: run locally after a build.
- `make DEBUG=lldb` or `make DEBUG=valgrind` or `make DEBUG=address`: rebuild with debug/ASan-friendly flags.
## Coding Style & Naming Conventions
- The codebase follows 42 **Norminette v4** rules. Run `norminette *.c *.h` (or on specific files) before submitting changes.
- Keep file names lowercase with underscores (e.g., `src/builtins/echo/echo.c`).
- Keep headers in `include/` and expose only what modules need.
- Before adding or changing code, check `allowed.txt` to know which functions are permitted.
- Any function not listed in `allowed.txt` is not allowed in this project.
## Parser & Lexer Functionality (Current `src/parser`)
- Runtime entrypoint is `parse(line, minishell)` from `src/minishell.c` (`readline -> parse -> execute`).
- `parse` calls `lex(line)` and then converts token lists to `t_command` nodes with `parse_tokens`.
- `command_new` builds one command from tokens up to `TOKEN_PIPE`.
- `words_add` stores consecutive `TOKEN_WORD` tokens in `command->argv` and increments `argc`.
- `expand_envs` is currently a TODO (no `$VAR` expansion is applied in parser stage).
- Redirection tokens are converted into `t_redirection` and stored in `t_command.redirections`; heredocs are stored in `t_command.heredocs`.
- Path resolution is handled in executor (`executor_resolve_command_path`) before `execve`.
- `src/parser/lexer.c` provides a separate lexer (`lex`) that tokenizes into `TOKEN_WORD`, `TOKEN_PIPE`, `TOKEN_REDIRECT_IN`, `TOKEN_REDIRECT_OUT`, `TOKEN_APPEND`, and `TOKEN_HEREDOC`.
- The lexer tracks single/double quote context so metacharacters inside quotes remain part of words.
- Meta runs are read as contiguous chunks in `read_token` (for example, repeated `|`/`<`/`>` are captured as one token value).
- Current parser flow consumes lexer output directly.
## Executor Redirections (Current `src/executor`)
- File redirections are applied in `src/executor/redirections.c` via `open` + `dup2` before command execution.
- Supported file redirections: input `<`, output truncate `>`, output append `>>`.
- Redirections are applied for both forked commands (child path) and single builtins executed in parent.
- Parent-builtin redirections save/restore `STDIN_FILENO` and `STDOUT_FILENO` after builtin execution.
## Parser & Lexer Known Gaps
- Heredoc tokens are parsed and stored, but runtime heredoc execution/input feeding is still pending in executor.
- No explicit unmatched-quote syntax error handling is implemented in parser/lexer path.
## Testing Guidelines
- There is no automated test runner. Use manual checks in `docs/tests.md` and basic shell behavior checks (pipes, redirects, builtins).
- A local builtin edge-case script exists at `tests/builtins_edge_cases.sh` (expects a compiled `./minishell`).
- When debugging memory issues, run under valgrind and use the suppression file in `valgrind/readline.supp`.
## Builtins Status
- `cd`: handles `HOME` fallback, `cd -` via `OLDPWD`, updates `PWD`/`OLDPWD`, and returns failure on invalid usage (`too many arguments`, missing `HOME`/`OLDPWD`).
- `echo`: supports repeated `-n` flags (`-n`, `-nnn`) and prints remaining args preserving spaces between tokens.
- `env`: prints current environment as `KEY=VALUE`; returns failure when called with extra arguments.
- `exit`: validates numeric argument (with overflow checks), returns `1` on `too many arguments` without exiting, and exits with `2` on non-numeric argument.
- `export`: supports `NAME=VALUE` and `NAME` (stored as empty value), validates identifiers, and returns failure when any identifier is invalid.
- `pwd`: prints working directory using dynamic `getcwd` and returns failure if `getcwd` fails.
- `unset`: validates identifiers and removes matching variables from the environment map.
## Commit & Pull Request Guidelines
- Commit messages in this repo use a simple `type: summary` format (examples: `update: ...`, `fix: ...`). Keep summaries short and specific.
- For PRs, include:
1. What changed and why.
2. How to test (commands or manual steps).
3. Notes on any parser/executor/builtin behavior changes.
## Configuration Tips
- The project depends on `readline`; ensure your system has `libreadline-dev` or equivalent before building.