owlen/crates/core/agent/src/git.rs

//! Git integration module for detecting repository state and validating git commands.
//!
//! This module provides functionality to:
//! - Detect if the current directory is a git repository
//! - Capture git repository state (branch, status, uncommitted changes)
//! - Validate git commands for safety (read-only vs destructive operations)

use color_eyre::eyre::Result;
use std::path::Path;
use std::process::Command;

/// Status of a file in the git working tree
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum GitFileStatus {
    /// File has been modified
    Modified { path: String },
    /// File has been added (staged)
    Added { path: String },
    /// File has been deleted
    Deleted { path: String },
    /// File has been renamed
    Renamed { from: String, to: String },
    /// File is untracked
    Untracked { path: String },
}

impl GitFileStatus {
    /// Get the primary path associated with this status
    pub fn path(&self) -> &str {
        match self {
            Self::Modified { path } => path,
            Self::Added { path } => path,
            Self::Deleted { path } => path,
            Self::Renamed { to, .. } => to,
            Self::Untracked { path } => path,
        }
    }
}

/// Complete state of a git repository
#[derive(Debug, Clone)]
pub struct GitState {
    /// Whether the current directory is in a git repository
    pub is_git_repo: bool,
    /// Current branch name (None if not in a repo or detached HEAD)
    pub current_branch: Option<String>,
    /// Main branch name (main/master, None if not detected)
    pub main_branch: Option<String>,
    /// Status of files in the working tree
    pub status: Vec<GitFileStatus>,
    /// Whether there are any uncommitted changes
    pub has_uncommitted_changes: bool,
    /// Remote URL for the repository (None if no remote configured)
    pub remote_url: Option<String>,
}

impl GitState {
    /// Create a default GitState for non-git directories
    pub fn not_a_repo() -> Self {
        Self {
            is_git_repo: false,
            current_branch: None,
            main_branch: None,
            status: Vec::new(),
            has_uncommitted_changes: false,
            remote_url: None,
        }
    }
}

/// Detect the current git repository state
///
/// This function runs various git commands to gather information about the repository.
/// If git is not available or the directory is not a git repo, returns a default state.
pub fn detect_git_state(working_dir: &Path) -> Result<GitState> {
    // Check if this is a git repository
    let is_repo = Command::new("git")
        .arg("rev-parse")
        .arg("--git-dir")
        .current_dir(working_dir)
        .output()
        .map(|output| output.status.success())
        .unwrap_or(false);

    if !is_repo {
        return Ok(GitState::not_a_repo());
    }

    // Get current branch
    let current_branch = get_current_branch(working_dir)?;

    // Detect main branch (try main first, then master)
    let main_branch = detect_main_branch(working_dir)?;

    // Get file status
    let status = get_git_status(working_dir)?;

    // Check if there are uncommitted changes
    let has_uncommitted_changes = !status.is_empty();

    // Get remote URL
    let remote_url = get_remote_url(working_dir)?;

    Ok(GitState {
        is_git_repo: true,
        current_branch,
        main_branch,
        status,
        has_uncommitted_changes,
        remote_url,
    })
}

/// Get the current branch name
fn get_current_branch(working_dir: &Path) -> Result<Option<String>> {
    let output = Command::new("git")
        .arg("rev-parse")
        .arg("--abbrev-ref")
        .arg("HEAD")
        .current_dir(working_dir)
        .output()?;

    if !output.status.success() {
        return Ok(None);
    }

    let branch = String::from_utf8_lossy(&output.stdout).trim().to_string();

    // "HEAD" means detached HEAD state
    if branch == "HEAD" {
        Ok(None)
    } else {
        Ok(Some(branch))
    }
}

/// Detect the main branch (main or master)
fn detect_main_branch(working_dir: &Path) -> Result<Option<String>> {
    // Try to get all branches
    let output = Command::new("git")
        .arg("branch")
        .arg("-a")
        .current_dir(working_dir)
        .output()?;

    if !output.status.success() {
        return Ok(None);
    }

    let branches = String::from_utf8_lossy(&output.stdout);

    // Check for main branch first (modern convention)
    if branches.lines().any(|line| {
        let trimmed = line.trim_start_matches('*').trim();
        trimmed == "main" || trimmed.ends_with("/main")
    }) {
        return Ok(Some("main".to_string()));
    }

    // Fall back to master
    if branches.lines().any(|line| {
        let trimmed = line.trim_start_matches('*').trim();
        trimmed == "master" || trimmed.ends_with("/master")
    }) {
        return Ok(Some("master".to_string()));
    }

    Ok(None)
}

/// Get the git status for all files
fn get_git_status(working_dir: &Path) -> Result<Vec<GitFileStatus>> {
    let output = Command::new("git")
        .arg("status")
        .arg("--porcelain")
        .arg("-z") // Null-terminated for better parsing
        .current_dir(working_dir)
        .output()?;

    if !output.status.success() {
        return Ok(Vec::new());
    }

    let status_text = String::from_utf8_lossy(&output.stdout);
    let mut statuses = Vec::new();

    // Parse porcelain format with null termination
    // Format: XY filename\0 (where X is staged status, Y is unstaged status)
    for entry in status_text.split('\0').filter(|s| !s.is_empty()) {
        if entry.len() < 3 {
            continue;
        }

        let status_code = &entry[0..2];
        let path = entry[3..].to_string();

        // Parse status codes
        match status_code {
            "M " | " M" | "MM" => {
                statuses.push(GitFileStatus::Modified { path });
            }
            "A " | " A" | "AM" => {
                statuses.push(GitFileStatus::Added { path });
            }
            "D " | " D" | "AD" => {
                statuses.push(GitFileStatus::Deleted { path });
            }
            "??" => {
                statuses.push(GitFileStatus::Untracked { path });
            }
            s if s.starts_with('R') => {
                // Renamed files have format "R  old_name -> new_name"
                if let Some((from, to)) = path.split_once(" -> ") {
                    statuses.push(GitFileStatus::Renamed {
                        from: from.to_string(),
                        to: to.to_string(),
                    });
                } else {
                    // Fallback if parsing fails
                    statuses.push(GitFileStatus::Modified { path });
                }
            }
            _ => {
                // Unknown status code, treat as modified
                statuses.push(GitFileStatus::Modified { path });
            }
        }
    }

    Ok(statuses)
}

/// Get the remote URL for the repository
fn get_remote_url(working_dir: &Path) -> Result<Option<String>> {
    let output = Command::new("git")
        .arg("remote")
        .arg("get-url")
        .arg("origin")
        .current_dir(working_dir)
        .output()?;

    if !output.status.success() {
        return Ok(None);
    }

    let url = String::from_utf8_lossy(&output.stdout).trim().to_string();

    if url.is_empty() {
        Ok(None)
    } else {
        Ok(Some(url))
    }
}

/// Check if a git command is safe (read-only)
///
/// Safe commands include:
/// - status, log, show, diff, branch (without -D)
/// - remote (without add/remove)
/// - config --get
/// - rev-parse, ls-files, ls-tree
pub fn is_safe_git_command(command: &str) -> bool {
    let parts: Vec<&str> = command.split_whitespace().collect();

    if parts.is_empty() || parts[0] != "git" {
        return false;
    }

    if parts.len() < 2 {
        return false;
    }

    let subcommand = parts[1];

    // List of read-only git commands
    match subcommand {
        "status" | "log" | "show" | "diff" | "blame" | "reflog" => true,
        "ls-files" | "ls-tree" | "ls-remote" => true,
        "rev-parse" | "rev-list" => true,
        "describe" | "tag" if !command.contains("-d") && !command.contains("--delete") => true,
        "branch" if !command.contains("-D") && !command.contains("-d") && !command.contains("-m") => true,
        "remote" if command.contains("get-url") || command.contains("-v") || command.contains("show") => true,
        "config" if command.contains("--get") || command.contains("--list") => true,
        "grep" | "shortlog" | "whatchanged" => true,
        "fetch" if !command.contains("--prune") => true,
        _ => false,
    }
}

/// Check if a git command is destructive
///
/// Returns (is_destructive, warning_message) tuple.
/// Destructive commands include:
/// - push --force, reset --hard, clean -fd
/// - rebase, amend, filter-branch
/// - branch -D, tag -d
pub fn is_destructive_git_command(command: &str) -> (bool, &'static str) {
    let cmd_lower = command.to_lowercase();

    // Check for force push
    if cmd_lower.contains("push") && (cmd_lower.contains("--force") || cmd_lower.contains("-f")) {
        return (true, "Force push can overwrite remote history and affect other collaborators");
    }

    // Check for hard reset
    if cmd_lower.contains("reset") && cmd_lower.contains("--hard") {
        return (true, "Hard reset will discard uncommitted changes permanently");
    }

    // Check for git clean
    if cmd_lower.contains("clean") && (cmd_lower.contains("-f") || cmd_lower.contains("-d")) {
        return (true, "Git clean will permanently delete untracked files");
    }

    // Check for rebase
    if cmd_lower.contains("rebase") {
        return (true, "Rebase rewrites commit history and can cause conflicts");
    }

    // Check for amend
    if cmd_lower.contains("commit") && cmd_lower.contains("--amend") {
        return (true, "Amending rewrites the last commit and changes its hash");
    }

    // Check for filter-branch or filter-repo
    if cmd_lower.contains("filter-branch") || cmd_lower.contains("filter-repo") {
        return (true, "Filter operations rewrite repository history");
    }

    // Check for branch/tag deletion
    if (cmd_lower.contains("branch") && (cmd_lower.contains("-D") || cmd_lower.contains("-d")))
        || (cmd_lower.contains("tag") && (cmd_lower.contains("-d") || cmd_lower.contains("--delete")))
    {
        return (true, "This will delete a branch or tag");
    }

    // Check for reflog expire
    if cmd_lower.contains("reflog") && cmd_lower.contains("expire") {
        return (true, "Expiring reflog removes recovery points for lost commits");
    }

    // Check for gc with aggressive or prune
    if cmd_lower.contains("gc") && (cmd_lower.contains("--aggressive") || cmd_lower.contains("--prune")) {
        return (true, "Aggressive garbage collection can make recovery difficult");
    }

    (false, "")
}

/// Format git state for human-readable display
///
/// Example output:
/// ```text
/// Git Repository: yes
/// Current branch: feature-branch
/// Main branch: main
/// Status: 3 modified, 1 untracked
/// Remote: https://github.com/user/repo.git
/// ```
pub fn format_git_status(state: &GitState) -> String {
    if !state.is_git_repo {
        return "Not a git repository".to_string();
    }

    let mut lines = Vec::new();

    lines.push("Git Repository: yes".to_string());

    if let Some(branch) = &state.current_branch {
        lines.push(format!("Current branch: {}", branch));
    } else {
        lines.push("Current branch: (detached HEAD)".to_string());
    }

    if let Some(main) = &state.main_branch {
        lines.push(format!("Main branch: {}", main));
    }

    // Summarize status
    if state.status.is_empty() {
        lines.push("Status: clean working tree".to_string());
    } else {
        let mut modified = 0;
        let mut added = 0;
        let mut deleted = 0;
        let mut renamed = 0;
        let mut untracked = 0;

        for status in &state.status {
            match status {
                GitFileStatus::Modified { .. } => modified += 1,
                GitFileStatus::Added { .. } => added += 1,
                GitFileStatus::Deleted { .. } => deleted += 1,
                GitFileStatus::Renamed { .. } => renamed += 1,
                GitFileStatus::Untracked { .. } => untracked += 1,
            }
        }

        let mut status_parts = Vec::new();
        if modified > 0 {
            status_parts.push(format!("{} modified", modified));
        }
        if added > 0 {
            status_parts.push(format!("{} added", added));
        }
        if deleted > 0 {
            status_parts.push(format!("{} deleted", deleted));
        }
        if renamed > 0 {
            status_parts.push(format!("{} renamed", renamed));
        }
        if untracked > 0 {
            status_parts.push(format!("{} untracked", untracked));
        }

        lines.push(format!("Status: {}", status_parts.join(", ")));
    }

    if let Some(url) = &state.remote_url {
        lines.push(format!("Remote: {}", url));
    } else {
        lines.push("Remote: (none)".to_string());
    }

    lines.join("\n")
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_is_safe_git_command() {
        // Safe commands
        assert!(is_safe_git_command("git status"));
        assert!(is_safe_git_command("git log --oneline"));
        assert!(is_safe_git_command("git diff HEAD"));
        assert!(is_safe_git_command("git branch -v"));
        assert!(is_safe_git_command("git remote -v"));
        assert!(is_safe_git_command("git config --get user.name"));

        // Unsafe commands
        assert!(!is_safe_git_command("git commit -m test"));
        assert!(!is_safe_git_command("git push origin main"));
        assert!(!is_safe_git_command("git branch -D feature"));
        assert!(!is_safe_git_command("git remote add origin url"));
    }

    #[test]
    fn test_is_destructive_git_command() {
        // Destructive commands
        let (is_dest, msg) = is_destructive_git_command("git push --force origin main");
        assert!(is_dest);
        assert!(msg.contains("Force push"));

        let (is_dest, msg) = is_destructive_git_command("git reset --hard HEAD~1");
        assert!(is_dest);
        assert!(msg.contains("Hard reset"));

        let (is_dest, msg) = is_destructive_git_command("git clean -fd");
        assert!(is_dest);
        assert!(msg.contains("clean"));

        let (is_dest, msg) = is_destructive_git_command("git rebase main");
        assert!(is_dest);
        assert!(msg.contains("Rebase"));

        let (is_dest, msg) = is_destructive_git_command("git commit --amend");
        assert!(is_dest);
        assert!(msg.contains("Amending"));

        // Non-destructive commands
        let (is_dest, _) = is_destructive_git_command("git status");
        assert!(!is_dest);

        let (is_dest, _) = is_destructive_git_command("git log");
        assert!(!is_dest);

        let (is_dest, _) = is_destructive_git_command("git diff");
        assert!(!is_dest);
    }

    #[test]
    fn test_git_state_not_a_repo() {
        let state = GitState::not_a_repo();
        assert!(!state.is_git_repo);
        assert!(state.current_branch.is_none());
        assert!(state.main_branch.is_none());
        assert!(state.status.is_empty());
        assert!(!state.has_uncommitted_changes);
        assert!(state.remote_url.is_none());
    }

    #[test]
    fn test_git_file_status_path() {
        let status = GitFileStatus::Modified {
            path: "test.rs".to_string(),
        };
        assert_eq!(status.path(), "test.rs");

        let status = GitFileStatus::Renamed {
            from: "old.rs".to_string(),
            to: "new.rs".to_string(),
        };
        assert_eq!(status.path(), "new.rs");
    }

    #[test]
    fn test_format_git_status_not_repo() {
        let state = GitState::not_a_repo();
        let formatted = format_git_status(&state);
        assert_eq!(formatted, "Not a git repository");
    }

    #[test]
    fn test_format_git_status_clean() {
        let state = GitState {
            is_git_repo: true,
            current_branch: Some("main".to_string()),
            main_branch: Some("main".to_string()),
            status: Vec::new(),
            has_uncommitted_changes: false,
            remote_url: Some("https://github.com/user/repo.git".to_string()),
        };

        let formatted = format_git_status(&state);
        assert!(formatted.contains("Git Repository: yes"));
        assert!(formatted.contains("Current branch: main"));
        assert!(formatted.contains("clean working tree"));
    }

    #[test]
    fn test_format_git_status_with_changes() {
        let state = GitState {
            is_git_repo: true,
            current_branch: Some("feature".to_string()),
            main_branch: Some("main".to_string()),
            status: vec![
                GitFileStatus::Modified {
                    path: "file1.rs".to_string(),
                },
                GitFileStatus::Modified {
                    path: "file2.rs".to_string(),
                },
                GitFileStatus::Untracked {
                    path: "new.rs".to_string(),
                },
            ],
            has_uncommitted_changes: true,
            remote_url: None,
        };

        let formatted = format_git_status(&state);
        assert!(formatted.contains("2 modified"));
        assert!(formatted.contains("1 untracked"));
    }
}