magi-code 0.96.1

mod batch;

use super::*;
use crate::tools::process::{
    BoundedChildProcessLimits, BoundedChildProcessOutput, run_bounded_child_process,
};
use std::{
    collections::BTreeMap,
    io::Read,
    process::{Command, Stdio},
    time::Duration,
};

fn run(
    cwd: &Path,
    args: &[&str],
    limit: usize,
    options: &ReviewReadOptions<'_>,
) -> Result<BoundedChildProcessOutput, String> {
    run_with_input(cwd, args, limit, options, Stdio::null())
}

fn run_with_input(
    cwd: &Path,
    args: &[&str],
    limit: usize,
    options: &ReviewReadOptions<'_>,
    input: Stdio,
) -> Result<BoundedChildProcessOutput, String> {
    let timeout = options.remaining()?.min(Duration::from_secs(3));
    let mut command = Command::new("git");
    command
        .current_dir(cwd)
        .args([
            "--no-pager",
            "--literal-pathspecs",
            "-c",
            "core.fsmonitor=false",
            "-c",
            "diff.external=",
            "-c",
            "core.quotePath=false",
        ])
        .args(args)
        .env("GIT_OPTIONAL_LOCKS", "0")
        .env_remove("GIT_EXTERNAL_DIFF")
        .env("LC_ALL", "C")
        .env_remove("GIT_DIR")
        .env_remove("GIT_WORK_TREE")
        .env_remove("GIT_INDEX_FILE")
        .env_remove("GIT_COMMON_DIR")
        .stdin(input)
        .stdout(Stdio::piped())
        .stderr(Stdio::piped());
    #[cfg(unix)]
    {
        use std::os::unix::process::CommandExt;
        command.process_group(0);
    }
    let output = run_bounded_child_process(
        command.spawn().map_err(|e| e.to_string())?,
        BoundedChildProcessLimits {
            stdout_max_bytes: limit,
            stderr_max_bytes: 4096,
            timeout,
            poll_interval: Duration::from_millis(10),
        },
        options.cancellation,
    )
    .map_err(|e| e.to_string())?;
    options.check()?;
    if output.timed_out {
        return Err("Git review timed out".into());
    }
    Ok(output)
}
fn success(output: BoundedChildProcessOutput) -> Result<String, String> {
    if !output.status.is_some_and(|s| s.success()) {
        return Err(format!("Git review failed: {}", output.stderr.trim()));
    }
    if output.stdout_truncated {
        return Err("Git review path list exceeds size limit".into());
    }
    if output.stdout.contains('\u{fffd}') {
        return Err("Git review omitted non-UTF-8 data".into());
    }
    Ok(output.stdout)
}
pub(super) fn root(cwd: &Path, options: &ReviewReadOptions<'_>) -> Result<Option<PathBuf>, String> {
    let output = run(cwd, &["rev-parse", "--show-toplevel"], 16 * 1024, options)?;
    if !output.status.is_some_and(|s| s.success()) {
        if output.stderr.contains("not a git repository")
            || output.stderr.contains("must be run in a work tree")
        {
            return Ok(None);
        }
        return Err(format!(
            "Cannot find Git worktree: {}",
            output.stderr.trim()
        ));
    }
    let root = success(output)?;
    std::fs::canonicalize(root.trim_end_matches('\n'))
        .map(Some)
        .map_err(|e| e.to_string())
}

pub(super) fn files_cached(
    root: &Path,
    options: &ReviewReadOptions<'_>,
    cache: &mut SnapshotCache,
) -> Result<Vec<ReviewFile>, String> {
    let (captured, fingerprint) = capture_files(root, options, cache)?;
    let mut files = Vec::with_capacity(captured.len());
    for file in captured {
        let previous = cache
            .files
            .iter()
            .find(|previous| previous.path == file.path && previous.notice.is_none());
        files.push(prepare_file_rows(file, previous, options));
    }
    cache.files = files.clone();
    cache.fingerprint = fingerprint;
    Ok(files)
}

fn capture_files(
    root: &Path,
    options: &ReviewReadOptions<'_>,
    cache: &mut SnapshotCache,
) -> Result<(Vec<ReviewFile>, (usize, u64)), String> {
    use std::hash::{Hash, Hasher};
    let (head, paths) = changed_paths(root, options)?;
    if cache.root != root || cache.head != head {
        cache.files.clear();
        cache.root = root.to_owned();
        cache.head = head.clone();
    }
    let omitted = paths.len().saturating_sub(MAX_FILES);
    let originals = batch::read_originals(root, &head, &paths, cache, options)?;
    let mut budget = MAX_TOTAL_SOURCE;
    let mut files = Vec::new();
    let mut fingerprint = std::collections::hash_map::DefaultHasher::new();
    let count = paths.len();
    for (index, (path, change)) in paths.into_iter().enumerate() {
        options.check()?;
        path.hash(&mut fingerprint);
        change.hash(&mut fingerprint);
        if index >= MAX_FILES {
            continue;
        }
        let mut file = capture_file_sources(
            root,
            &path,
            change != ReviewFileChange::Written,
            (budget / 2).min(MAX_SOURCE),
            options,
            cache
                .files
                .iter()
                .find(|file| file.path == path && file.notice.is_none()),
            originals.get(&path).map(String::as_str),
        );
        file.change = Some(change);
        options.check()?;
        // Match snapshot storage costs, not the read ceiling of omitted sources.
        // Reads remain bounded independently by MAX_FILES and MAX_SOURCE.
        budget = budget.saturating_sub(file.original.len() + file.current.len());
        file.original.hash(&mut fingerprint);
        file.current.hash(&mut fingerprint);
        file.notice.hash(&mut fingerprint);
        files.push(file);
    }
    if omitted > 0 {
        files.push(notice_file(
            "[review limit]",
            format!("{omitted} files omitted; review limit is {MAX_FILES} files"),
        ));
    }
    let fingerprint = if count == 0 { 0 } else { fingerprint.finish() };
    Ok((files, (count, fingerprint)))
}

pub(super) fn changed_paths(
    root: &Path,
    options: &ReviewReadOptions<'_>,
) -> Result<(String, BTreeMap<String, ReviewFileChange>), String> {
    let head = run(
        root,
        &["rev-parse", "--verify", "--quiet", "HEAD"],
        1024,
        options,
    )?;
    let head = if head.status.is_some_and(|s| s.success()) {
        success(head)?
    } else {
        String::new()
    };
    let mut paths = BTreeMap::new();
    if !head.is_empty() {
        let names = success(run(
            root,
            &[
                "diff",
                "--no-ext-diff",
                "--no-textconv",
                "--no-renames",
                "--name-status",
                "-z",
                "HEAD",
                "--",
            ],
            MAX_SOURCE,
            options,
        )?)?;
        let dir = cap_std::fs::Dir::open_ambient_dir(root, cap_std::ambient_authority())
            .map_err(|e| e.to_string())?;
        let mut parts = names.split_terminator('\0');
        while let Some(status) = parts.next() {
            options.check()?;
            let path = parts.next().ok_or("Malformed Git diff path list")?;
            let change = match status {
                "A" => ReviewFileChange::Written,
                // Git reports index deletions even when a replacement exists in the
                // worktree. Check presence before hashing, including omitted sources
                // and dangling symlinks; source readability does not determine status.
                "D" => {
                    validate_path(path)?;
                    match dir.symlink_metadata(path) {
                        Ok(_) => ReviewFileChange::Edited,
                        Err(error)
                            if matches!(
                                error.kind(),
                                std::io::ErrorKind::NotFound | std::io::ErrorKind::NotADirectory
                            ) =>
                        {
                            ReviewFileChange::Deleted
                        }
                        Err(error) => return Err(error.to_string()),
                    }
                }
                _ => ReviewFileChange::Edited,
            };
            paths.insert(path.to_owned(), change);
        }
    }
    let args: &[&str] = if !head.is_empty() {
        &["ls-files", "--others", "--exclude-standard", "-z"]
    } else {
        &[
            "ls-files",
            "--cached",
            "--others",
            "--exclude-standard",
            "-z",
        ]
    };
    for path in success(run(root, args, MAX_SOURCE, options)?)?.split_terminator('\0') {
        options.check()?;
        paths
            .entry(path.to_owned())
            .or_insert(ReviewFileChange::Written);
    }
    Ok((head, paths))
}
pub(super) fn change_fingerprint(
    root: &Path,
    options: &ReviewReadOptions<'_>,
) -> Result<(usize, u64), String> {
    capture_files(root, options, &mut SnapshotCache::default()).map(|(_, fingerprint)| fingerprint)
}

pub(super) fn notice_file(path: &str, notice: String) -> ReviewFile {
    ReviewFile {
        path: path.into(),
        change: None,
        original: String::new(),
        current: String::new(),
        rows: Vec::new(),
        notice: Some(notice),
    }
}
pub(super) fn read_file(
    root: &Path,
    path: &str,
    has_original: bool,
    limit: usize,
    options: &ReviewReadOptions<'_>,
) -> ReviewFile {
    let file = capture_file_sources(root, path, has_original, limit, options, None, None);
    prepare_file_rows(file, None, options)
}

fn capture_file_sources(
    root: &Path,
    path: &str,
    has_original: bool,
    limit: usize,
    options: &ReviewReadOptions<'_>,
    previous: Option<&ReviewFile>,
    batch_original: Option<&str>,
) -> ReviewFile {
    let mut file = notice_file(path, String::new());
    file.notice = None;
    if limit == 0 {
        file.notice = Some("Source omitted: snapshot size limit".into());
        return file;
    }
    let original = previous
        .map(|file| file.original.as_str())
        .or(batch_original);
    match read_sources(root, path, has_original, limit, options, original) {
        Ok((original, current)) => {
            file.original = original;
            file.current = current;
        }
        Err(error) => {
            file.notice = Some(error);
            return file;
        }
    }
    file
}

fn prepare_file_rows(
    mut file: ReviewFile,
    previous: Option<&ReviewFile>,
    options: &ReviewReadOptions<'_>,
) -> ReviewFile {
    if file.notice.is_some() {
        return file;
    }
    if let Some(previous) = previous
        && previous.original == file.original
        && previous.current == file.current
        && previous.change == file.change
    {
        return previous.clone();
    }
    let remaining = match options.remaining() {
        Ok(remaining) => remaining,
        Err(error) => {
            return ReviewFile {
                change: file.change,
                ..notice_file(&file.path, error)
            };
        }
    };
    let diff = similar::TextDiff::configure()
        .timeout(remaining.min(Duration::from_millis(30)))
        .diff_lines(&file.original, &file.current);
    for change in diff.iter_all_changes() {
        if let Err(error) = options.check() {
            return ReviewFile {
                change: file.change,
                ..notice_file(&file.path, error)
            };
        }
        file.rows.push(ReviewLine {
            old_line: change.old_index().map(|n| n + 1),
            new_line: change.new_index().map(|n| n + 1),
            text: change.value().trim_end_matches('\n').to_owned(),
            kind: match change.tag() {
                similar::ChangeTag::Equal => ReviewLineKind::Context,
                similar::ChangeTag::Insert => ReviewLineKind::Added,
                similar::ChangeTag::Delete => ReviewLineKind::Removed,
            },
        });
    }
    file
}
fn read_sources(
    root: &Path,
    path: &str,
    has_original: bool,
    limit: usize,
    options: &ReviewReadOptions<'_>,
    cached_original: Option<&str>,
) -> Result<(String, String), String> {
    options.check()?;
    validate_path(path)?;
    let original = if let Some(original) = cached_original.filter(|_| has_original) {
        if original.len() > limit {
            return Err("Source omitted: original exceeds size limit".into());
        }
        original.to_owned()
    } else if has_original {
        let output = run(
            root,
            &[
                "show",
                "--no-ext-diff",
                "--no-textconv",
                &format!("HEAD:{path}"),
            ],
            limit,
            options,
        )?;
        if output.stdout_truncated {
            return Err("Source omitted: original exceeds size limit".into());
        }
        if !output.status.is_some_and(|s| s.success())
            && (output.stderr.contains("does not exist in 'HEAD'")
                || output.stderr.contains("exists on disk, but not in 'HEAD'")
                || output.stderr.contains("invalid object name 'HEAD'"))
        {
            String::new()
        } else {
            success(output)?
        }
    } else {
        String::new()
    };
    options.check()?;
    let dir = cap_std::fs::Dir::open_ambient_dir(root, cap_std::ambient_authority())
        .map_err(|e| e.to_string())?;
    let current = match dir.symlink_metadata(path) {
        Err(e) if e.kind() == std::io::ErrorKind::NotFound => String::new(),
        Err(e) => return Err(e.to_string()),
        Ok(metadata) => {
            if !metadata.is_file() {
                return Err("Source omitted: symlink, submodule, or non-regular file".into());
            }
            let mut bytes = Vec::new();
            let mut open_options = cap_std::fs::OpenOptions::new();
            open_options.read(true);
            #[cfg(unix)]
            {
                use cap_std::fs::OpenOptionsExt;
                open_options.custom_flags(libc::O_NONBLOCK | libc::O_NOFOLLOW);
            }
            let handle = dir
                .open_with(path, &open_options)
                .map_err(|e| e.to_string())?;
            if !handle.metadata().map_err(|e| e.to_string())?.is_file() {
                return Err("Source omitted: non-regular file".into());
            }
            let mut handle = handle.take(limit as u64 + 1);
            let mut buffer = [0; 8192];
            loop {
                options.check()?;
                let count = handle.read(&mut buffer).map_err(|e| e.to_string())?;
                if count == 0 {
                    break;
                }
                bytes.extend_from_slice(&buffer[..count]);
            }
            if bytes.len() > limit {
                return Err("Source omitted: current exceeds size limit".into());
            }
            String::from_utf8(bytes).map_err(|_| "Source omitted: non-UTF-8 file".to_owned())?
        }
    };
    options.check()?;
    if original.contains('\0') || current.contains('\0') || original.contains('\u{fffd}') {
        return Err("Source omitted: binary or non-UTF-8 file".into());
    }
    Ok((original, current))
}
pub(super) fn validate_path(path: &str) -> Result<(), String> {
    if path.is_empty()
        || !Path::new(path)
            .components()
            .all(|c| matches!(c, std::path::Component::Normal(_)))
    {
        return Err("Invalid review file path".into());
    }
    Ok(())
}