gitks/sanitize.rs

//! Input sanitization for git subprocess arguments.
//!
//! Prevents command injection by validating user-supplied strings before
//! passing them to git commands.

use crate::error::GitError;
use crate::error::GitResult;

/// Characters that are never allowed in git ref names / revision strings.
///
/// Git disallows: space, `~`, `^`, `:`, `?`, `*`, `[`, `\`, and all ASCII
/// control characters (bytes 0–31 and 127). The control characters are
/// checked separately via `is_ascii_control()`.
const FORBIDDEN_REF_CHARS: &[char] = &['~', '^', ':', '?', '*', '[', '\\', ' '];

/// Returns true if `c` is an ASCII control character (bytes 0–31, 127).
fn is_ascii_control(c: char) -> bool {
    let b = c as u32;
    b <= 31 || b == 127
}

/// Validate a git reference name (branch, tag, etc.).
///
/// Git ref rules (from `git check-ref-format`):
/// - Cannot contain forbidden chars
/// - Cannot start or end with '.'
/// - Cannot end with '/'
/// - Cannot contain '..'
/// - Cannot contain '@{'
/// - Cannot be empty
pub fn validate_oid_hex(hex: &str) -> GitResult<()> {
    if hex.is_empty() {
        return Err(GitError::InvalidArgument("oid hex cannot be empty".into()));
    }
    if !(4..=64).contains(&hex.len()) {
        return Err(GitError::InvalidArgument(format!(
            "oid hex length must be 4..=64 chars: {}",
            hex.len()
        )));
    }
    if !hex.chars().all(|c| c.is_ascii_hexdigit()) {
        return Err(GitError::InvalidArgument(format!(
            "oid hex contains non-hex character: {hex}"
        )));
    }
    Ok(())
}

pub fn validate_ref_name(name: &str) -> GitResult<()> {
    if name.is_empty() {
        return Err(GitError::InvalidArgument("ref name cannot be empty".into()));
    }
    if name.starts_with('.') || name.ends_with('.') {
        return Err(GitError::InvalidArgument(format!(
            "ref name cannot start or end with '.': {name}"
        )));
    }
    if name.ends_with('/') {
        return Err(GitError::InvalidArgument(format!(
            "ref name cannot end with '/': {name}"
        )));
    }
    if name.contains("..") {
        return Err(GitError::InvalidArgument(format!(
            "ref name cannot contain '..': {name}"
        )));
    }
    if name.contains("@{") {
        return Err(GitError::InvalidArgument(format!(
            "ref name cannot contain '@{{': {name}"
        )));
    }
    if name.contains(|c: char| FORBIDDEN_REF_CHARS.contains(&c) || is_ascii_control(c)) {
        return Err(GitError::InvalidArgument(format!(
            "ref name contains forbidden character: {name}"
        )));
    }
    if name.len() > 255 {
        return Err(GitError::InvalidArgument(format!(
            "ref name too long (max 255 chars): {name}"
        )));
    }
    Ok(())
}

/// Validate a revision string (branch name, tag name, or short expression).
///
/// Allows OID hex strings, ref names, and a small set of revision operators
/// (HEAD, ^{tree}, ~N, ^N) that are safe when passed as a single argument.
pub fn validate_revision(rev: &str) -> GitResult<()> {
    if rev.is_empty() {
        return Err(GitError::InvalidArgument("revision cannot be empty".into()));
    }
    if rev.len() > 256 {
        return Err(GitError::InvalidArgument(format!(
            "revision too long (max 256 chars): {}",
            rev.len()
        )));
    }
    if rev.chars().all(|c| c.is_ascii_hexdigit()) && rev.len() >= 4 && rev.len() <= 64 {
        return Ok(());
    }
    if rev == "HEAD" {
        return Ok(());
    }
    // Allow ref:refs/heads/... (git internal format)
    if let Some(rest) = rev.strip_prefix("ref:") {
        return validate_ref_name(rest.trim());
    }

    const MAX_ANCESTRY_DEPTH: u32 = 10000;

    if let Some(tilde_pos) = rev.rfind('~') {
        let num_part = &rev[tilde_pos + 1..];
        if !num_part.is_empty() && num_part.chars().all(|c| c.is_ascii_digit()) {
            let depth: u32 = num_part
                .parse()
                .map_err(|_| GitError::InvalidArgument("invalid ~N syntax".into()))?;
            if depth > MAX_ANCESTRY_DEPTH {
                return Err(GitError::InvalidArgument(format!(
                    "~N depth too large: {} (max {})",
                    depth, MAX_ANCESTRY_DEPTH
                )));
            }
        }
    }

    if let Some(caret_pos) = rev.rfind('^') {
        let after_caret = &rev[caret_pos + 1..];
        if !after_caret.starts_with('{')
            && !after_caret.is_empty()
            && let Some(first_char) = after_caret.chars().next()
            && first_char.is_ascii_digit()
        {
            let num_part: String = after_caret
                .chars()
                .take_while(|c| c.is_ascii_digit())
                .collect();
            if !num_part.is_empty() {
                let depth: u32 = num_part
                    .parse()
                    .map_err(|_| GitError::InvalidArgument("invalid ^N syntax".into()))?;
                if depth > MAX_ANCESTRY_DEPTH {
                    return Err(GitError::InvalidArgument(format!(
                        "^N depth too large: {} (max {})",
                        depth, MAX_ANCESTRY_DEPTH
                    )));
                }
            }
        }
    }

    let mut base = rev;

    base = base
        .trim_end_matches("^{tree}")
        .trim_end_matches("^{commit}")
        .trim_end_matches("^{object}");

    if let Some(tilde_pos) = base.rfind('~') {
        let after_tilde = &base[tilde_pos + 1..];
        if !after_tilde.is_empty() && after_tilde.chars().all(|c| c.is_ascii_digit()) {
            base = &base[..tilde_pos];
        }
    } else if let Some(caret_pos) = base.rfind('^') {
        let after_caret = &base[caret_pos + 1..];
        if !after_caret.starts_with('{')
            && !after_caret.is_empty()
            && after_caret.chars().all(|c| c.is_ascii_digit())
        {
            base = &base[..caret_pos];
        }
    }

    if base.is_empty() {
        return Ok(());
    }
    validate_ref_name(base)?;
    Ok(())
}

/// Validate a file path within a commit action.
///
/// Must be a relative path (no leading '/'), no '..' traversal,
/// no null bytes, no .git directory access, and reasonable length.
pub fn validate_file_path(path: &str) -> GitResult<()> {
    if path.is_empty() {
        return Err(GitError::InvalidArgument(
            "file path cannot be empty".into(),
        ));
    }
    if path.starts_with('/') {
        return Err(GitError::InvalidArgument(format!(
            "file path must be relative, not absolute: {path}"
        )));
    }
    if path.contains("..") {
        return Err(GitError::InvalidArgument(format!(
            "file path cannot contain '..': {path}"
        )));
    }
    if path.contains('\0') {
        return Err(GitError::InvalidArgument(format!(
            "file path cannot contain null byte: {path}"
        )));
    }
    if path.len() > 4096 {
        return Err(GitError::InvalidArgument(format!(
            "file path too long (max 4096 chars): {path}"
        )));
    }

    if path == ".git"
        || path.starts_with(".git/")
        || path.contains("/.git/")
        || path.ends_with("/.git")
    {
        return Err(GitError::InvalidArgument(format!(
            "cannot modify .git directory: {path}"
        )));
    }

    // Windows reserved names check
    #[cfg(target_os = "windows")]
    {
        const RESERVED_NAMES: &[&str] = &[
            "CON", "PRN", "AUX", "NUL", "COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7",
            "COM8", "COM9", "LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9",
        ];

        for component in path.split('/') {
            let name_part = component.split('.').next().unwrap_or(component);
            let name_upper = name_part.to_uppercase();

            if RESERVED_NAMES.contains(&name_upper.as_str()) {
                return Err(GitError::InvalidArgument(format!(
                    "Windows reserved device name: {component}"
                )));
            }
        }
    }

    Ok(())
}

/// Git config keys that are dangerous to set remotely.
/// Setting these could allow arbitrary command execution or bypass security.
const DANGEROUS_CONFIG_KEYS: &[&str] = &[
    "core.sshCommand",
    "core.gitProxy",
    "http.proxy",
    "https.proxy",
    "remote.*.url",
    "credential.*",
    "safe.directory",
    "core.hooksPath",
    "receive.fsckObjects",
    "receive.denyCurrentBranch",
    "receive.denyDeleteCurrent",
];

/// Check if a git config key is safe to set remotely.
pub fn validate_config_key(key: &str) -> GitResult<()> {
    if key.is_empty() {
        return Err(GitError::InvalidArgument(
            "config key cannot be empty".into(),
        ));
    }
    for pattern in DANGEROUS_CONFIG_KEYS {
        if pattern.contains('*') {
            // e.g. "remote.*.url" — match any "remote.<something>.url"
            if let Some((prefix, suffix)) = pattern.split_once('*')
                && key.starts_with(prefix)
                && key.ends_with(suffix)
            {
                return Err(GitError::InvalidArgument(format!(
                    "config key '{key}' matches dangerous pattern '{pattern}'"
                )));
            }
        } else if key == *pattern {
            return Err(GitError::InvalidArgument(format!(
                "config key '{key}' is not allowed to be set remotely"
            )));
        }
    }
    if !key
        .chars()
        .all(|c| c.is_ascii_alphanumeric() || c == '.' || c == '-' || c == '_')
    {
        return Err(GitError::InvalidArgument(format!(
            "config key contains invalid characters: {key}"
        )));
    }
    Ok(())
}

/// Allowed URL schemes for git remotes.
const ALLOWED_REMOTE_SCHEMES: &[&str] = &["http://", "https://", "ssh://", "git://", "git+ssh://"];

/// Validate a remote URL for git operations.
///
/// Only allows standard transport protocols. Rejects `file://`, `ext::`,
/// and other schemes that could access local resources or execute commands.
pub fn validate_remote_url(url: &str) -> GitResult<()> {
    if url.is_empty() {
        return Err(GitError::InvalidArgument(
            "remote URL cannot be empty".into(),
        ));
    }
    if url.len() > 4096 {
        return Err(GitError::InvalidArgument(
            "remote URL too long (max 4096 chars)".into(),
        ));
    }
    if url.contains('\0') || url.contains('\n') || url.contains('\r') {
        return Err(GitError::InvalidArgument(
            "remote URL contains invalid characters".into(),
        ));
    }
    if !ALLOWED_REMOTE_SCHEMES.iter().any(|s| url.starts_with(s)) {
        return Err(GitError::InvalidArgument(format!(
            "remote URL must start with one of: {}. Got: {url}",
            ALLOWED_REMOTE_SCHEMES.join(", ")
        )));
    }
    Ok(())
}

/// Validate a git refspec string (e.g. `+refs/heads/*:refs/heads/*`).
///
/// Refspecs must not contain null bytes, newlines, or shell metacharacters.
pub fn validate_refspec(refspec: &str) -> GitResult<()> {
    if refspec.is_empty() {
        return Err(GitError::InvalidArgument("refspec cannot be empty".into()));
    }
    if refspec.contains('\0') || refspec.contains('\n') || refspec.contains('\r') {
        return Err(GitError::InvalidArgument(
            "refspec contains invalid characters".into(),
        ));
    }
    if refspec.contains(['$', '`', '(', ')', '{', '}', '|', ';', '&', '<', '>']) {
        return Err(GitError::InvalidArgument(format!(
            "refspec contains shell metacharacter: {refspec}"
        )));
    }
    if refspec.len() > 1024 {
        return Err(GitError::InvalidArgument(
            "refspec too long (max 1024 chars)".into(),
        ));
    }
    Ok(())
}

/// Validate a storage-relative path (used in resolve_for_init and from_repository_header).
///
/// Must not contain path traversal, must be a simple relative path.
pub fn validate_relative_path(path: &str) -> GitResult<()> {
    if path.is_empty() {
        return Err(GitError::InvalidArgument(
            "relative_path cannot be empty".into(),
        ));
    }
    if path.starts_with('/') {
        return Err(GitError::InvalidArgument(
            "relative_path must be relative, not absolute".into(),
        ));
    }
    if path.contains('\0') {
        return Err(GitError::InvalidArgument(
            "relative_path cannot contain null byte".into(),
        ));
    }
    if path.len() > 4096 {
        return Err(GitError::InvalidArgument(
            "relative_path too long (max 4096 chars)".into(),
        ));
    }
    if path.contains("..") {
        return Err(GitError::InvalidArgument(format!(
            "path traversal detected: relative_path contains '..': {path}"
        )));
    }
    Ok(())
}