polyc-tools 2026.8.3

//! Seeding a delegated worker's workspace with named parent files (`#2295`).
//!
//! `#2286` fenced every delegated worker in its own workspace subtree and
//! re-rooted all the coding tools against it — reads included. That closed the
//! concurrent-write race, but it also ended a contract that had been implicit:
//! a worker could read the parent's workspace. After `#2286` a delegation like
//! "review the file I just wrote" hands the worker an empty scratch directory.
//!
//! This module restores that reach, narrowly. The target agent's manifest
//! declares a [`ShareInCeiling`] — glob patterns plus file-count and byte
//! bounds — and a `__delegate_to` call names literal paths inside it. Only the
//! intersection is copied, into the worker's own subtree at the same relative
//! paths, before its nested turn starts.
//!
//! # What this is not
//!
//! Copies, not a share: the worker owns its copy outright and its writes stay
//! its own, so seeding cannot reintroduce `#2286`'s clobbering. Nothing copies
//! back — a worker's output is still its final message alone (`#2295` tracks
//! the copy-out half separately, deliberately: a copy-out collision policy is
//! the same race in a new place).
//!
//! # Refusals are hard
//!
//! Every violation fails the delegation instead of seeding a subset. A worker
//! silently given three of the five files its task named would produce a
//! confidently wrong answer against a partial view; a failed delegation the
//! orchestrator can see and retry is strictly better.

use std::path::{Component, Path, PathBuf};

use polyc_agent::delegate::{ShareInCeiling, ShareInError};

use super::search::glob_to_regex;
use super::workspace::{WORKER_SUBDIR_PREFIX, resolve};

/// One parent file selected for seeding: where it lives now, where it lands,
/// and how big it is.
struct Selected {
    /// Absolute path in the parent workspace.
    source: PathBuf,
    /// Workspace-relative path, `/`-joined — the key the ceiling matches
    /// against and the path the copy lands at under the worker root.
    relative: String,
    /// Size in bytes, summed against [`ShareInCeiling::max_bytes`].
    bytes: u64,
}

/// Copy the parent files `requested` names into `worker_root`, enforcing
/// `ceiling`.
///
/// Resolves and bounds-checks the whole request before copying a single byte,
/// so a refused delegation never leaves a half-seeded worker tree behind.
/// Returns the workspace-relative paths actually seeded, in copy order, for
/// the delegation's forensic record.
///
/// # Errors
///
/// Returns [`ShareInError`] when a path escapes the workspace, names a
/// delegated worker's subtree, resolves to nothing, falls outside `ceiling`'s
/// patterns, exceeds its file or byte bounds, or cannot be read or written.
pub fn seed(
    parent_root: &Path,
    worker_root: &Path,
    requested: &[String],
    ceiling: &ShareInCeiling,
) -> Result<Vec<String>, ShareInError> {
    if requested.is_empty() {
        return Ok(Vec::new());
    }
    // A closed ceiling refuses before any filesystem work: an agent that
    // declares no share-in never seeds, which is `#2286`'s behavior exactly.
    if !ceiling.admits_anything() {
        return Err(ShareInError::OutsideCeiling {
            path: requested[0].clone(),
        });
    }

    let patterns = compile(ceiling);
    let mut selected = Vec::new();
    for entry in requested {
        collect(parent_root, entry, ceiling, &mut selected)?;
    }

    // Bounds are checked against the fully expanded set, not per entry — a
    // request is one unit, and two directories that are individually under the
    // cap must not add up to a worker tree over it.
    if selected.len() > ceiling.max_files {
        return Err(ShareInError::TooManyFiles {
            found: selected.len(),
            limit: ceiling.max_files,
        });
    }
    let total: u64 = selected.iter().map(|f| f.bytes).sum();
    if total > ceiling.max_bytes {
        return Err(ShareInError::TooManyBytes {
            found: total,
            limit: ceiling.max_bytes,
        });
    }
    for file in &selected {
        if !patterns.iter().any(|re| re.is_match(&file.relative)) {
            return Err(ShareInError::OutsideCeiling {
                path: file.relative.clone(),
            });
        }
    }

    let mut seeded = Vec::with_capacity(selected.len());
    for file in selected {
        let destination =
            resolve(worker_root, &file.relative).map_err(|reason| ShareInError::Escapes {
                path: file.relative.clone(),
                reason,
            })?;
        if let Some(parent) = destination.parent() {
            std::fs::create_dir_all(parent).map_err(|err| ShareInError::Io {
                path: file.relative.clone(),
                reason: err.to_string(),
            })?;
        }
        std::fs::copy(&file.source, &destination).map_err(|err| ShareInError::Io {
            path: file.relative.clone(),
            reason: err.to_string(),
        })?;
        seeded.push(file.relative);
    }
    Ok(seeded)
}

/// Compile the ceiling's globs into anchored regexes over `/`-joined paths.
///
/// A pattern that fails to compile is dropped rather than propagated: it can
/// only ever narrow what is admitted, so a malformed manifest entry fails
/// closed on its own without taking the whole ceiling down with it.
fn compile(ceiling: &ShareInCeiling) -> Vec<regex::Regex> {
    ceiling
        .allow
        .iter()
        .filter_map(|pattern| regex::Regex::new(&glob_to_regex(pattern)).ok())
        .collect()
}

/// Resolve one request entry and append every file it names to `out`.
fn collect(
    parent_root: &Path,
    entry: &str,
    ceiling: &ShareInCeiling,
    out: &mut Vec<Selected>,
) -> Result<(), ShareInError> {
    let source = resolve(parent_root, entry).map_err(|reason| ShareInError::Escapes {
        path: entry.to_owned(),
        reason,
    })?;
    if names_worker_subtree(parent_root, &source) {
        return Err(ShareInError::WorkerSubtree {
            path: entry.to_owned(),
        });
    }
    // `symlink_metadata` deliberately, not `metadata`: a symlink planted in
    // the parent workspace would otherwise be followed by the copy below,
    // reading a file outside the workspace into a worker that could never
    // have reached it. A symlink named directly is refused; one met while
    // walking a directory is skipped (see `walk`).
    let meta = std::fs::symlink_metadata(&source).map_err(|_| ShareInError::NotFound {
        path: entry.to_owned(),
    })?;
    if meta.is_symlink() {
        return Err(ShareInError::Escapes {
            path: entry.to_owned(),
            reason: "symbolic links are never seeded".to_owned(),
        });
    }
    if meta.is_dir() {
        walk(parent_root, &source, ceiling, out)
    } else {
        out.push(Selected {
            relative: relative_to(parent_root, &source),
            source,
            bytes: meta.len(),
        });
        Ok(())
    }
}

/// Whether `source` lands inside a delegated worker's subtree.
///
/// Tested against the RESOLVED path, not the request string: [`resolve`] has
/// already normalized away `.` components, and a raw-string check would miss
/// `./.worker-b/notes.md` — its first component is `CurDir`, so a
/// first-component test on the request would read the entry as ordinary and
/// hand over a sibling's files.
///
/// Only the first component of the normalized path matters: a worker root is
/// always a single component directly under the conversation root, and
/// [`resolve`] has already rejected anything that could climb back out.
fn names_worker_subtree(parent_root: &Path, source: &Path) -> bool {
    source
        .strip_prefix(parent_root)
        .ok()
        .and_then(|rel| rel.components().next())
        .is_some_and(|first| {
            matches!(first, Component::Normal(part)
            if part.to_string_lossy().starts_with(WORKER_SUBDIR_PREFIX))
        })
}

/// Recursively append every regular file under `dir` to `out`.
///
/// Bails as soon as the running count passes [`ShareInCeiling::max_files`], so
/// a request naming a huge tree is refused without first materializing it.
/// Symlinks and delegated workers' subtrees are skipped.
fn walk(
    parent_root: &Path,
    dir: &Path,
    ceiling: &ShareInCeiling,
    out: &mut Vec<Selected>,
) -> Result<(), ShareInError> {
    let entries = std::fs::read_dir(dir).map_err(|err| ShareInError::Io {
        path: relative_to(parent_root, dir),
        reason: err.to_string(),
    })?;
    for entry in entries.flatten() {
        let path = entry.path();
        let Ok(meta) = std::fs::symlink_metadata(&path) else {
            continue;
        };
        if meta.is_symlink() {
            continue;
        }
        if meta.is_dir() {
            let is_worker_tree = path
                .file_name()
                .is_some_and(|name| name.to_string_lossy().starts_with(WORKER_SUBDIR_PREFIX));
            if is_worker_tree {
                continue;
            }
            walk(parent_root, &path, ceiling, out)?;
        } else {
            out.push(Selected {
                relative: relative_to(parent_root, &path),
                source: path,
                bytes: meta.len(),
            });
        }
        if out.len() > ceiling.max_files {
            return Err(ShareInError::TooManyFiles {
                found: out.len(),
                limit: ceiling.max_files,
            });
        }
    }
    Ok(())
}

/// Render `path` relative to `root` as a `/`-joined string.
///
/// `path` always sits under `root` here — every caller derived it from
/// [`resolve`] or from walking `root` — so the strip cannot fail; the fallback
/// keeps the lossy rendering total rather than panicking on a future caller
/// that breaks that assumption.
fn relative_to(root: &Path, path: &Path) -> String {
    path.strip_prefix(root)
        .unwrap_or(path)
        .components()
        .filter_map(|c| match c {
            Component::Normal(part) => Some(part.to_string_lossy().into_owned()),
            _ => None,
        })
        .collect::<Vec<_>>()
        .join("/")
}

#[cfg(test)]
mod tests {
    #![allow(clippy::pedantic, clippy::nursery, missing_docs)]
    use super::*;

    fn ceiling(allow: &[&str], max_files: usize, max_bytes: u64) -> ShareInCeiling {
        ShareInCeiling {
            allow: allow.iter().map(|s| (*s).to_owned()).collect(),
            max_files,
            max_bytes,
        }
    }

    fn write(root: &Path, rel: &str, body: &str) {
        let path = root.join(rel);
        std::fs::create_dir_all(path.parent().unwrap()).unwrap();
        std::fs::write(path, body).unwrap();
    }

    /// A parent workspace with one worker subtree already under it, matching
    /// the layout `ToolRegistry::for_worker` creates. The caller leaves the
    /// directory behind; `tmp_dir` keys it per-process and per-call.
    fn roots() -> (PathBuf, PathBuf) {
        let parent = super::super::tmp_dir("share-in").join("workspace");
        let worker = parent.join(".worker-call-a-0000");
        std::fs::create_dir_all(&worker).unwrap();
        (parent, worker)
    }

    #[test]
    fn seeds_a_named_file_at_the_same_relative_path() {
        let (parent, worker) = roots();
        write(&parent, "src/parser.rs", "fn parse() {}");

        let seeded = seed(
            &parent,
            &worker,
            &["src/parser.rs".to_owned()],
            &ceiling(&["src/**"], 10, 1024),
        )
        .expect("a file inside the ceiling seeds");

        assert_eq!(seeded, vec!["src/parser.rs".to_owned()]);
        assert_eq!(
            std::fs::read_to_string(worker.join("src/parser.rs")).unwrap(),
            "fn parse() {}",
        );
    }

    #[test]
    fn refuses_a_file_outside_the_ceiling() {
        let (parent, worker) = roots();
        write(&parent, "secrets/token", "s3cret");

        let err = seed(
            &parent,
            &worker,
            &["secrets/token".to_owned()],
            &ceiling(&["src/**"], 10, 1024),
        )
        .expect_err("a file the ceiling does not admit must be refused");

        assert!(
            matches!(err, ShareInError::OutsideCeiling { .. }),
            "{err:?}"
        );
        assert!(
            !worker.join("secrets/token").exists(),
            "a refused request must copy nothing"
        );
    }

    #[test]
    fn refuses_traversal_out_of_the_parent_workspace() {
        let (parent, worker) = roots();

        let err = seed(
            &parent,
            &worker,
            &["../outside.txt".to_owned()],
            &ceiling(&["**"], 10, 1024),
        )
        .expect_err("`..` must never leave the parent workspace");

        assert!(matches!(err, ShareInError::Escapes { .. }), "{err:?}");
    }

    #[test]
    fn refuses_reaching_into_a_sibling_workers_subtree() {
        // The cross-worker read `#2286` fenced off: without this check a
        // worker could name a sibling's scratch directory and be handed its
        // files, since `resolve` treats `.worker-*` as an ordinary component.
        let (parent, worker) = roots();
        write(&parent, ".worker-call-b-1111/notes.md", "sibling's work");

        let err = seed(
            &parent,
            &worker,
            &[".worker-call-b-1111/notes.md".to_owned()],
            &ceiling(&["**"], 10, 1024),
        )
        .expect_err("a sibling worker's subtree is never seedable");

        assert!(matches!(err, ShareInError::WorkerSubtree { .. }), "{err:?}");
    }

    #[test]
    fn refuses_a_dot_prefixed_path_into_a_worker_subtree() {
        // `./` normalizes away, so a first-component test against the REQUEST
        // string reads this as an ordinary path and hands over the sibling's
        // file. The guard has to run against the resolved path.
        let (parent, worker) = roots();
        write(&parent, ".worker-call-b-1111/notes.md", "sibling's work");

        let err = seed(
            &parent,
            &worker,
            &["./.worker-call-b-1111/notes.md".to_owned()],
            &ceiling(&["**"], 10, 1024),
        )
        .expect_err("`./` must not smuggle a request into a worker subtree");

        assert!(matches!(err, ShareInError::WorkerSubtree { .. }), "{err:?}");
    }

    #[test]
    fn refuses_when_over_the_file_bound() {
        let (parent, worker) = roots();
        for i in 0..5 {
            write(&parent, &format!("src/f{i}.rs"), "x");
        }

        let err = seed(
            &parent,
            &worker,
            &["src".to_owned()],
            &ceiling(&["src/**"], 3, 1024),
        )
        .expect_err("five files must not seed under a three-file ceiling");

        assert!(matches!(err, ShareInError::TooManyFiles { .. }), "{err:?}");
    }

    #[test]
    fn refuses_when_over_the_byte_bound() {
        let (parent, worker) = roots();
        write(&parent, "src/big.rs", &"x".repeat(200));

        let err = seed(
            &parent,
            &worker,
            &["src/big.rs".to_owned()],
            &ceiling(&["src/**"], 10, 64),
        )
        .expect_err("a file over the byte ceiling must be refused");

        assert!(matches!(err, ShareInError::TooManyBytes { .. }), "{err:?}");
        assert!(
            !worker.join("src/big.rs").exists(),
            "a refused request must copy nothing"
        );
    }

    #[test]
    fn seeds_a_directory_recursively() {
        let (parent, worker) = roots();
        write(&parent, "src/a.rs", "a");
        write(&parent, "src/nested/b.rs", "b");

        let mut seeded = seed(
            &parent,
            &worker,
            &["src".to_owned()],
            &ceiling(&["src/**"], 10, 1024),
        )
        .expect("a directory seeds every file under it");
        seeded.sort();

        assert_eq!(
            seeded,
            vec!["src/a.rs".to_owned(), "src/nested/b.rs".to_owned()]
        );
        assert_eq!(
            std::fs::read_to_string(worker.join("src/nested/b.rs")).unwrap(),
            "b"
        );
    }

    #[test]
    fn a_closed_ceiling_seeds_nothing() {
        // The default an agent gets when its manifest says nothing about
        // share-in: every request is refused, so `#2286`'s fencing is intact.
        let (parent, worker) = roots();
        write(&parent, "src/parser.rs", "fn parse() {}");

        let err = seed(
            &parent,
            &worker,
            &["src/parser.rs".to_owned()],
            &ShareInCeiling::default(),
        )
        .expect_err("an unconfigured ceiling admits nothing");

        assert!(
            matches!(err, ShareInError::OutsideCeiling { .. }),
            "{err:?}"
        );
    }

    #[test]
    fn an_empty_request_seeds_nothing_and_succeeds() {
        let (parent, worker) = roots();
        write(&parent, "src/parser.rs", "fn parse() {}");

        let seeded = seed(&parent, &worker, &[], &ceiling(&["src/**"], 10, 1024))
            .expect("no request is not an error");

        assert!(seeded.is_empty());
        assert!(!worker.join("src/parser.rs").exists());
    }

    #[test]
    fn refuses_a_symlink_named_directly() {
        let (parent, worker) = roots();
        write(&parent, "src/real.rs", "real");
        #[cfg(unix)]
        std::os::unix::fs::symlink("/etc/hostname", parent.join("src/link.rs")).unwrap();

        #[cfg(unix)]
        {
            let err = seed(
                &parent,
                &worker,
                &["src/link.rs".to_owned()],
                &ceiling(&["src/**"], 10, 1024),
            )
            .expect_err("a symlink could read outside the workspace");
            assert!(matches!(err, ShareInError::Escapes { .. }), "{err:?}");
        }
    }

    #[test]
    fn refuses_a_path_that_names_nothing() {
        let (parent, worker) = roots();

        let err = seed(
            &parent,
            &worker,
            &["src/missing.rs".to_owned()],
            &ceiling(&["src/**"], 10, 1024),
        )
        .expect_err("a path naming nothing is an error, not an empty seed");

        assert!(matches!(err, ShareInError::NotFound { .. }), "{err:?}");
    }
}