foundry

//! The shell's parser: command line -> tokens -> pipeline, and the drive
//! letter mapping (`D:/notes` <-> `/mnt/disk0/notes`).
//!
//! Pure functions with no syscalls, so the same file is compiled into the
//! shell (user/sh.rs) *and* into the kernel's test suite (tests/shell.rs),
//! where it's tested without starting any processes.
//!
//! Grammar (a small subset of what a Unix shell accepts):
//!
//!   line     = pipeline [ "&" ] [ "#" comment ]
//!   pipeline = command { "|" command }
//!   command  = { word | "<" word | ">" word | ">>" word }
//!   word     = { plain-char | 'single quoted' | "double quoted" }
//!
//! `|`, `<`, `>`, `>>` and `&` are operators even without spaces around
//! them (`ls|wc` works). Quotes just group: no `$variables` or escapes.

#![allow(dead_code)]

use alloc::{format, string::String, vec::Vec};

/// One word of a command. `quoted` words are never drive-translated, so
/// `grep "C:" file` searches for the text C:.
#[derive(Debug, Clone, PartialEq)]
pub struct Word {
    pub text: String,
    pub quoted: bool,
}

#[derive(Debug, Clone, PartialEq)]
pub enum Token {
    Word(Word),
    Pipe,       // |
    In,         // <
    Out,        // >
    Append,     // >>
    Background, // &
}

/// Split a line into tokens. Errors on an unterminated quote.
pub fn tokenize(line: &str) -> Result<Vec<Token>, String> {
    let mut tokens = Vec::new();
    let mut chars = line.chars().peekable();
    // The word being built, if any (a word can mix plain and quoted parts:
    // a"b c"d is the single word `ab cd`).
    let mut word: Option<Word> = None;

    while let Some(c) = chars.next() {
        let op = match c {
            '|' => Some(Token::Pipe),
            '<' => Some(Token::In),
            '&' => Some(Token::Background),
            '>' if chars.peek() == Some(&'>') => {
                chars.next();
                Some(Token::Append)
            }
            '>' => Some(Token::Out),
            _ => None,
        };
        if op.is_some() || c.is_whitespace() || (c == '#' && word.is_none()) {
            // Anything here ends the current word.
            if let Some(w) = word.take() {
                tokens.push(Token::Word(w));
            }
            if let Some(op) = op {
                tokens.push(op);
            } else if c == '#' {
                break; // comment: ignore the rest of the line
            }
            continue;
        }
        let w = word.get_or_insert(Word { text: String::new(), quoted: false });
        if c == '\'' || c == '"' {
            w.quoted = true;
            loop {
                match chars.next() {
                    Some(q) if q == c => break,
                    Some(other) => w.text.push(other),
                    None => return Err(format!("unterminated {} quote", c)),
                }
            }
        } else {
            w.text.push(c);
        }
    }
    if let Some(w) = word {
        tokens.push(Token::Word(w));
    }
    Ok(tokens)
}

/// One program (or builtin) with its arguments and redirections.
#[derive(Debug, Clone, Default, PartialEq)]
pub struct Command {
    pub words: Vec<Word>,           // words[0] is the command name
    pub stdin: Option<Word>,        // < file
    pub stdout: Option<(Word, bool)>, // > file (false) or >> file (true)
}

/// `a | b | c`, maybe with a trailing `&` (run in the background).
#[derive(Debug, Clone, PartialEq)]
pub struct Pipeline {
    pub commands: Vec<Command>,
    pub background: bool,
}

/// Parse a line. `Ok(None)` for an empty line or a comment.
pub fn parse(line: &str) -> Result<Option<Pipeline>, String> {
    let tokens = tokenize(line)?;
    if tokens.is_empty() {
        return Ok(None);
    }
    let mut commands = Vec::new();
    let mut current = Command::default();
    let mut background = false;
    let mut iter = tokens.into_iter().peekable();

    while let Some(token) = iter.next() {
        if background {
            return Err(String::from("`&` must be at the end of the line"));
        }
        match token {
            Token::Word(w) => current.words.push(w),
            Token::Pipe => {
                if current.words.is_empty() {
                    return Err(String::from("missing command before `|`"));
                }
                commands.push(core::mem::take(&mut current));
            }
            Token::Background => background = true,
            Token::In | Token::Out | Token::Append => {
                let target = match iter.next() {
                    Some(Token::Word(w)) => w,
                    _ => return Err(String::from("missing file name after redirection")),
                };
                match token {
                    Token::In => current.stdin = Some(target),
                    Token::Out => current.stdout = Some((target, false)),
                    _ => current.stdout = Some((target, true)),
                }
            }
        }
    }
    if current.words.is_empty() {
        // e.g. `> file`, `&` or `ls |`: nothing to run.
        return Err(String::from(if commands.is_empty() { "missing command" } else { "missing command after `|`" }));
    }
    commands.push(current);
    Ok(Some(Pipeline { commands, background }))
}

// --- drive letters -----------------------------------------------------------

/// A drive letter and the directory it stands for.
#[derive(Debug, Clone, PartialEq)]
pub struct Drive {
    pub letter: char, // always upper case
    pub path: String, // absolute, no trailing slash except "/"
    pub fs: String,   // file system type, e.g. "fat32"
}

/// Assign letters from the kernel's mount table (the text from the
/// `mounts` syscall: one "path fstype" per line, in mount order).
/// C: is always the root; D:, E:, ... go to the mounts under /mnt (the
/// disks), in the order they were mounted. Other mounts such as /tmp are
/// reachable through C: (C:/tmp).
pub fn assign_drives(mount_table: &str) -> Vec<Drive> {
    let mut drives = Vec::new();
    let root_fs = mount_table
        .lines()
        .find_map(|l| l.split_once(' ').filter(|(p, _)| *p == "/").map(|(_, fs)| fs))
        .unwrap_or("ramfs");
    drives.push(Drive { letter: 'C', path: String::from("/"), fs: String::from(root_fs.trim()) });
    let mut letter = b'D';
    for line in mount_table.lines() {
        if let Some((path, fs)) = line.split_once(' ') {
            if path.starts_with("/mnt/") && letter <= b'Z' {
                drives.push(Drive { letter: letter as char, path: String::from(path), fs: String::from(fs.trim()) });
                letter += 1;
            }
        }
    }
    drives
}

/// Does `word` start with a drive letter ("D:", "d:/x", "E:notes")?
pub fn drive_letter(word: &str) -> Option<char> {
    let b = word.as_bytes();
    if b.len() >= 2 && b[0].is_ascii_alphabetic() && b[1] == b':' {
        Some((b[0] as char).to_ascii_uppercase())
    } else {
        None
    }
}

/// `D:/notes/todo.txt` -> `/mnt/disk0/notes/todo.txt`. Letters are case
/// insensitive; `D:` and `D:/` are the drive's root, and `D:notes` means
/// `D:/notes` (drives have no current directory of their own). `None` if
/// `word` doesn't start with a known drive letter.
pub fn translate(word: &str, drives: &[Drive]) -> Option<String> {
    let letter = drive_letter(word)?;
    let drive = drives.iter().find(|d| d.letter == letter)?;
    let rest = word[2..].trim_start_matches('/');
    Some(if rest.is_empty() {
        drive.path.clone()
    } else if drive.path == "/" {
        format!("/{}", rest)
    } else {
        format!("{}/{}", drive.path, rest)
    })
}

/// The other direction, for the prompt: `/mnt/disk0/bin` -> `D:/bin`,
/// `/tmp` -> `C:/tmp`. Picks the drive with the longest matching path.
pub fn display(path: &str, drives: &[Drive]) -> String {
    let mut best: Option<(&Drive, &str)> = None;
    for d in drives {
        let rest = if d.path == "/" {
            path.strip_prefix('/')
        } else {
            // "/mnt/disk0" matches "/mnt/disk0" and "/mnt/disk0/x", not "/mnt/disk01".
            path.strip_prefix(d.path.as_str())
                .filter(|r| r.is_empty() || r.starts_with('/'))
                .map(|r| r.trim_start_matches('/'))
        };
        if let Some(rest) = rest {
            if best.map_or(true, |(b, _)| d.path.len() > b.path.len()) {
                best = Some((d, rest));
            }
        }
    }
    match best {
        Some((d, rest)) => format!("{}:/{}", d.letter, rest),
        None => String::from(path),
    }
}