gunnar-server 2.0.0

//! **The build runner** (roadmap phase 3).
//!
//! A push lands. The runner materialises exactly that commit, builds it with no
//! network, hashes what came out, and signs a [`BuildRecord`] naming the commit
//! it was made from — which is the join that lets `gunnar verify` walk from an
//! artifact to a line of source.
//!
//! ```text
//!   RefUpdated ──▶ checkout(commit) ──▶ sandbox(build) ──▶ artifacts ──▶ signed record ──▶ log
//! ```
//!
//! # The recipe is in the commit, on purpose
//!
//! `gunnar-build.toml` is read **out of the tree being built**, not off the
//! runner's disk. So the recipe is part of what was pushed, part of what was
//! signed, and part of what a verifier can go back and look at. A recipe kept
//! beside the runner would be a build input nobody attested to — the one input
//! that decides what every other input does.
//!
//! # `network = false` is a fact here, not a claim
//!
//! `crate::sandbox` puts the build in a network namespace of its own. The
//! runner **refuses to build** when it cannot, rather than building anyway and
//! recording `network: true`: a server that quietly downgrades is a server whose
//! records mean different things on different machines. `--allow-network` is the
//! explicit opt-out, and it is recorded in the signed record.
//!
//! # What it is not
//!
//! Not a scheduler, not a queue, not a fleet. One build at a time, in the
//! process that watched the event. Scaling it horizontally is the phase-5 item
//! and it is a different problem: this is the part that has to be *right*.

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

use anyhow::{Context, Result};
use gunnar_store::git_storage_trait::{GitServe, ObjType};

use crate::stores::ServerStores;

/// The file a repository carries to say how it is built.
pub const RECIPE_FILE: &str = "gunnar-build.toml";

/// How a repository is built.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Recipe {
    /// What this build is called, in the record.
    pub name: String,
    /// The program to run, in the checkout.
    pub command: String,
    pub args: Vec<String>,
    /// Paths, relative to the checkout, that the build produces.
    ///
    /// Empty means **discover**: every ordinary file directly under
    /// `target/release` that is executable. That is the default recipe's shape,
    /// and naming them explicitly is what a repository does when it wants
    /// something else.
    pub artifacts: Vec<String>,
}

impl Recipe {
    /// **The default recipe**, for a cargo workspace that carries none.
    ///
    /// `--locked` above all: a build that may re-resolve is a build whose
    /// inputs are not the ones in the commit, and the SBOM this pairs with is
    /// read from the same lockfile. Without it the two describe different
    /// builds and nothing says so.
    pub fn cargo_default() -> Recipe {
        Recipe {
            name: "cargo build --release --locked".to_owned(),
            command: "cargo".to_owned(),
            args: vec![
                "build".to_owned(),
                "--release".to_owned(),
                "--locked".to_owned(),
            ],
            artifacts: Vec::new(),
        }
    }

    /// Parse `gunnar-build.toml`.
    ///
    /// A **small, total** reader rather than a TOML dependency, for the reason
    /// `gunnar_core::sbom`'s lockfile reader is one: this crate is in the
    /// appliance graph. `key = "value"` and `key = ["a", "b"]`, and anything it
    /// does not understand is skipped rather than guessed at.
    pub fn parse(text: &str) -> Result<Recipe> {
        let mut recipe = Recipe {
            name: String::new(),
            command: String::new(),
            args: Vec::new(),
            artifacts: Vec::new(),
        };
        for line in text.lines() {
            let line = line.trim();
            if line.is_empty() || line.starts_with('#') || line.starts_with('[') {
                continue;
            }
            let Some((key, value)) = line.split_once('=') else {
                continue;
            };
            let value = value.trim();
            let list = |value: &str| -> Vec<String> {
                value
                    .trim_start_matches('[')
                    .trim_end_matches(']')
                    .split(',')
                    .map(|item| item.trim().trim_matches('"').to_owned())
                    .filter(|item| !item.is_empty())
                    .collect()
            };
            match key.trim() {
                "name" => recipe.name = value.trim_matches('"').to_owned(),
                "command" => recipe.command = value.trim_matches('"').to_owned(),
                "args" => recipe.args = list(value),
                "artifacts" => recipe.artifacts = list(value),
                _ => {}
            }
        }
        anyhow::ensure!(
            !recipe.command.is_empty(),
            "{RECIPE_FILE} names no `command`, so there is nothing to run. A recipe that builds \
             nothing is not a smaller recipe — it is a build that reports success having done \
             nothing."
        );
        if recipe.name.is_empty() {
            recipe.name = recipe.command.clone();
        }
        Ok(recipe)
    }
}

/// What one build did.
#[derive(Debug, Clone)]
pub struct Built {
    pub record: gunnar_core::BuildRecord,
    /// Where the checkout and its output are, for a caller that wants to look.
    pub workdir: PathBuf,
}

/// Materialise `tree` into `into`, from the store and nothing else.
///
/// No `git checkout`, no worktree, no gix: the tree is read through `GitServe`,
/// which is the one contract every engine has. That is also why the runner can
/// build a repository the archive engine stores, which has no working copy on
/// disk to check out from.
fn checkout(
    git: &dyn GitServe,
    hash: gix_hash::Kind,
    tree: gix_hash::ObjectId,
    into: &Path,
) -> Result<usize> {
    std::fs::create_dir_all(into).with_context(|| format!("creating {}", into.display()))?;
    let (kind, bytes) = git
        .read(tree.as_bytes())?
        .with_context(|| format!("tree {tree} is not in this repository"))?;
    anyhow::ensure!(kind == ObjType::Tree, "{tree} is not a tree");

    let mut files = 0usize;
    for entry in gunnar_store::object::TreeIter::from_bytes(&bytes, hash).named() {
        let entry = entry.with_context(|| format!("decoding tree {tree}"))?;
        let name = String::from_utf8_lossy(entry.name).into_owned();
        // The sandbox boundary starts here: a tree entry named `..` or with a
        // separator in it would write outside the checkout, and a repository is
        // something anybody with push access controls.
        anyhow::ensure!(
            name != "." && name != ".." && !name.contains('/'),
            "tree {tree} carries an entry named {name:?}, which would write outside the checkout"
        );
        let path = into.join(&name);
        match entry.kind {
            gunnar_store::object::EntryKind::Tree => {
                files += checkout(git, hash, entry.oid, &path)?;
            }
            gunnar_store::object::EntryKind::Commit => {
                // A gitlink names a commit in another repository. There is
                // nothing here to write, and writing an empty directory would
                // make the build look like it had a submodule checked out.
                continue;
            }
            kind => {
                let Some((ObjType::Blob, blob)) = git.read(entry.oid.as_bytes())? else {
                    anyhow::bail!("{} is not a blob in this repository", entry.oid);
                };
                if kind == gunnar_store::object::EntryKind::Link {
                    // A symlink's blob is its target. Recreated as a symlink so
                    // a build that follows one behaves as it did for the author.
                    let target = String::from_utf8_lossy(&blob).into_owned();
                    let _ = std::fs::remove_file(&path);
                    std::os::unix::fs::symlink(&target, &path)
                        .with_context(|| format!("linking {}", path.display()))?;
                } else {
                    std::fs::write(&path, &blob)
                        .with_context(|| format!("writing {}", path.display()))?;
                    if kind == gunnar_store::object::EntryKind::BlobExecutable {
                        use std::os::unix::fs::PermissionsExt as _;
                        std::fs::set_permissions(&path, std::fs::Permissions::from_mode(0o755))
                            .with_context(|| format!("chmod {}", path.display()))?;
                    }
                }
                files += 1;
            }
        }
    }
    Ok(files)
}

/// The recipe a checkout carries, or the default one it qualifies for.
fn recipe_for(checkout: &Path) -> Result<Recipe> {
    let path = checkout.join(RECIPE_FILE);
    if path.exists() {
        let text = std::fs::read_to_string(&path)
            .with_context(|| format!("reading {}", path.display()))?;
        return Recipe::parse(&text)
            .with_context(|| format!("{} is not a usable recipe", path.display()));
    }
    anyhow::ensure!(
        checkout.join("Cargo.toml").exists(),
        "this repository carries no {RECIPE_FILE} and is not a cargo workspace, so there is no \
         recipe and no default that fits. Add a {RECIPE_FILE} naming a `command`."
    );
    Ok(Recipe::cargo_default())
}

/// Every artifact the recipe produced.
fn collect(checkout: &Path, recipe: &Recipe) -> Result<Vec<PathBuf>> {
    if !recipe.artifacts.is_empty() {
        let mut out = Vec::new();
        for named in &recipe.artifacts {
            let path = checkout.join(named);
            anyhow::ensure!(
                path.is_file(),
                "the recipe names {named} and the build did not produce it. A build that \
                 reports success without its own declared output is a build nobody checked."
            );
            out.push(path);
        }
        return Ok(out);
    }
    // Discovery, for the default recipe: executables directly under
    // `target/release`. Not recursive — `target/release/deps` and `build/` are
    // cargo's own scratch, and shipping them as artifacts would attest to
    // hundreds of files nobody asked for.
    use std::os::unix::fs::PermissionsExt as _;
    let dir = checkout.join("target").join("release");
    let mut out = Vec::new();
    if let Ok(entries) = std::fs::read_dir(&dir) {
        for entry in entries.flatten() {
            let path = entry.path();
            let Ok(meta) = entry.metadata() else { continue };
            if meta.is_file() && meta.permissions().mode() & 0o111 != 0 {
                out.push(path);
            }
        }
    }
    out.sort();
    anyhow::ensure!(
        !out.is_empty(),
        "the build produced no executable in {}. Either it built nothing, or its output is \
         somewhere the default recipe does not look — name the artifacts in {RECIPE_FILE}.",
        dir.display()
    );
    Ok(out)
}

/// **Build one commit and attest to what came out.**
///
/// `allow_network` is the explicit opt-out from hermeticity, and it goes into
/// the signed record. Without it a runner that cannot enter the sandbox
/// **fails** rather than building: a record that says `network: false` must mean
/// it on every machine, or it means nothing on any of them.
pub fn build_commit(
    stores: &ServerStores,
    repo: &gunnar_core::RepoRef,
    commit: gix_hash::ObjectId,
    workspace: &Path,
    builder: &str,
    allow_network: bool,
) -> Result<Built> {
    let entry = stores
        .catalog()
        .get(repo)?
        .with_context(|| format!("no such repository: {repo}"))?;
    let store = stores.repo_store(&entry)?;
    let git = store.serve();
    let hash = store.hash_kind();

    let (kind, bytes) = git
        .read(commit.as_bytes())?
        .with_context(|| format!("commit {commit} is not in {repo}"))?;
    anyhow::ensure!(kind == ObjType::Commit, "{commit} is not a commit");
    let tree = gunnar_store::object::commit(&bytes, hash)
        .with_context(|| format!("decoding commit {commit}"))?
        .tree;

    let workdir = workspace
        .join(repo.namespace())
        .join(repo.name())
        .join(commit.to_string());
    // A fresh checkout every time. Reusing one would make the build depend on
    // what the *last* build left behind, which is the first thing that makes a
    // "reproducible" build not one.
    let _ = std::fs::remove_dir_all(&workdir);
    let files = checkout(git, hash, tree, &workdir)?;
    anyhow::ensure!(
        files > 0,
        "commit {commit} has an empty tree; there is nothing to build"
    );

    let recipe = recipe_for(&workdir)?;

    if !allow_network {
        if let Err(err) = crate::sandbox::available() {
            anyhow::bail!(
                "this build was to be hermetic and the sandbox cannot be entered: {}\n\n\
                 Refusing rather than building with the network up. A record that says \
                 `network: false` has to mean it on every machine. Pass --allow-network to \
                 build anyway; the record will say so.",
                err.explain()
            );
        }
    }

    let started = gunnar_grpc::bus::now_unix_ms();
    let mut command = hardened_command(&recipe.command, &workdir, allow_network);
    command
        .args(&recipe.args)
        .env("CARGO_NET_OFFLINE", "true")
        .env("GUNNAR_BUILD_COMMIT", commit.to_string())
        .env("GUNNAR_BUILD_REPO", repo.to_string());
    let output = command
        .output()
        .with_context(|| format!("running {}", recipe.command))?;
    let finished = gunnar_grpc::bus::now_unix_ms();

    if let Some(err) = output
        .status
        .code()
        .and_then(crate::sandbox::error_from_exit)
    {
        anyhow::bail!("the build never started: {}", err.explain());
    }
    anyhow::ensure!(
        output.status.success(),
        "the build failed ({}):\n{}",
        output.status,
        String::from_utf8_lossy(&output.stderr)
    );

    let artifacts = collect(&workdir, &recipe)?;
    let subjects = artifacts
        .iter()
        .map(|path| -> Result<gunnar_core::Subject> {
            Ok(gunnar_core::Subject {
                name: path
                    .strip_prefix(&workdir)
                    .unwrap_or(path)
                    .display()
                    .to_string(),
                sha256: gunnar_core::provenance::digest_file(path)
                    .with_context(|| format!("hashing {}", path.display()))?,
            })
        })
        .collect::<Result<Vec<_>>>()?;

    // The lockfile is a **material**: it is what decided every dependency, and
    // it is in the commit, so naming it here is what lets the SBOM beside this
    // record be checked against the same bytes.
    let mut materials = Vec::new();
    let lock = workdir.join("Cargo.lock");
    if lock.is_file() {
        materials.push(gunnar_core::Subject {
            name: "Cargo.lock".to_owned(),
            sha256: gunnar_core::provenance::digest_file(&lock)?,
        });
    }

    let record = gunnar_core::BuildRecord {
        builder: builder.to_owned(),
        source_repo: repo.to_string(),
        source_commit: commit.to_string(),
        recipe: recipe.name.clone(),
        toolchain: toolchain(&recipe.command, &workdir, allow_network),
        network: allow_network,
        materials,
        subjects,
        started_at_unix_ms: started,
        finished_at_unix_ms: finished,
    };

    let key = stores.provenance_key()?;
    let signed = gunnar_core::provenance::sign_build(&key, record.clone())
        .context("signing the build record")?;
    let mut line = serde_json::to_string(&signed)?;
    line.push('\n');
    let _framed = gunnar_core::txlog::sink::append_line(stores.build_log().records_path(), line.trim_end_matches('\n'))?;
    use std::io::Write as _;
    let mut file = std::fs::OpenOptions::new()
        .create(true)
        .append(true)
        .open(stores.build_log().records_path())?;
    file.write_all(line.as_bytes())?;

    Ok(Built { record, workdir })
}

/// **The one constructor of every process a build runs.**
///
/// GUN-006. `build_commit` built its `Command` with `env_clear`,
/// `current_dir(&workdir)` and `pre_exec(enter_sandbox)`, and then
/// [`toolchain`] ran the SAME attacker-named program a second time with
/// `Command::new(command).arg("--version")` — no cleared environment, no
/// working directory, no sandbox. One hardened exec and one bare one, of a
/// `recipe.command` read out of the pushed tree and validated against nothing
/// but `!is_empty()`.
///
/// The escape was worth more than the build: the bare child ran as the server's
/// uid with the network up, so it could read `$GUNNAR_SECRET_KEK` or
/// `secrets.kek` from disk and decrypt the sealed store — the SSH host key and
/// the provenance signing key — and then sign `BuildRecord`s of its own.
///
/// So the hardening is a function both call sites must go through, and
/// `the_runner_has_exactly_one_process_constructor` holds a third from being
/// added beside them. What is NOT here is the build's own environment
/// (`CARGO_NET_OFFLINE`, `GUNNAR_BUILD_*`): those are facts about a build and
/// `toolchain` is not one, and adding them here would make the version probe
/// claim to be a build step.
fn hardened_command(program: &str, workdir: &Path, allow_network: bool) -> std::process::Command {
    // A32-14, a process by design: the build runner runs a repository's own declared build program, hardened (cleared environment, no network unless allowed); it is the feature, and no cargo or sh of Gunnar's own.
    let mut command = std::process::Command::new(program);
    command
        .current_dir(workdir)
        // A clean-ish environment: the build's inputs should be the commit, not
        // the operator's shell. `PATH` and `HOME` stay because a toolchain
        // needs to be findable and cargo needs somewhere to put its caches.
        .env_clear()
        .env("PATH", std::env::var_os("PATH").unwrap_or_default())
        .env("HOME", std::env::var_os("HOME").unwrap_or_default());
    if !allow_network {
        use std::os::unix::process::CommandExt as _;
        // The hook is bound OUTSIDE the `unsafe` block on purpose: an `unsafe`
        // block extends lexically into a closure written inside it, so nesting
        // the two made the inner one redundant (`unused_unsafe`) and its SAFETY
        // note unenforced. Bound here, each `unsafe` covers exactly one call.
        let enter_sandbox = || {
            // SAFETY: `pre_exec` runs in the forked child, single-threaded,
            // before `exec`.
            if let Err(err) = unsafe { crate::sandbox::enter() } {
                crate::sandbox::leave_child(crate::sandbox::exit_code(err));
            }
            Ok(())
        };
        // SAFETY: `sandbox::enter` allocates nothing and takes no lock, which
        // is the whole requirement `pre_exec` imposes.
        unsafe {
            command.pre_exec(enter_sandbox);
        }
    }
    command
}

/// What the toolchain says it is, or an honest blank.
///
/// Asked of the tool that ran, not read from a config: a `rust-toolchain.toml`
/// says what was *requested* and this says what answered. Empty when the tool
/// will not say, because a made-up version in a signed record is worse than no
/// version at all.
///
/// Through [`hardened_command`], because this runs the pushed tree's own
/// `recipe.command` — see GUN-006 — and its stdout goes verbatim into the
/// SIGNED `toolchain` field of the build record.
fn toolchain(command: &str, workdir: &Path, allow_network: bool) -> String {
    hardened_command(command, workdir, allow_network)
        .arg("--version")
        .output()
        .ok()
        .filter(|out| out.status.success())
        .map(|out| String::from_utf8_lossy(&out.stdout).trim().to_owned())
        .unwrap_or_default()
}

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

    /// ★ **A watch that ends is said** (A28 M18 follow-up: serve.rs dropped the result with
    /// `let _ =`). An error is an ALERT at ERROR with its cause; any other end is a WARN that
    /// says how many pushes were built and that no more are.
    #[test]
    fn a_watch_that_ends_is_said_and_an_error_is_an_alert() {
        let (level, words) = said_end(&Err(anyhow::anyhow!("building a runtime for the runner's subscription: no threads")));
        assert_eq!(level, tracing::Level::ERROR);
        assert!(words.starts_with("ALERT runner"), "{words}");
        assert!(words.contains("no threads") && words.contains("no longer built"), "{words}");
        let (level, words) = said_end(&Ok(3));
        assert_eq!(level, tracing::Level::WARN);
        assert!(words.contains("3 build") && words.contains("no longer built"), "{words}");
    }

    /// ★ The watch's own end goes through [`said_end`]: a watch bounded to no builds ends at
    /// once, and what the server says of it is the WARN, not nothing.
    #[test]
    fn the_watch_returns_its_end_for_the_server_to_say() {
        let dir = tempfile::tempdir().expect("a temp dir");
        let stores = ServerStores::open(dir.path()).expect("stores");
        let bus = gunnar_grpc::EventBus::new(4);
        let config = WatchConfig { workspace: dir.path().join("ws"), builder: "runner".into(), allow_network: true, refs: Vec::new(), max_builds: Some(0) };
        let end = watch(&stores, &bus, &config, |_| {});
        assert_eq!(said_end(&end).0, tracing::Level::WARN, "{:?}", said_end(&end));
    }

    #[test]
    fn a_recipe_parses_and_one_with_no_command_is_refused() {
        let recipe = Recipe::parse(
            r#"
            name = "the appliance"
            command = "make"
            args = ["-j", "4"]
            artifacts = ["out/gunnar"]
            "#,
        )
        .expect("a usable recipe");
        assert_eq!(recipe.name, "the appliance");
        assert_eq!(recipe.command, "make");
        assert_eq!(recipe.args, vec!["-j", "4"]);
        assert_eq!(recipe.artifacts, vec!["out/gunnar"]);

        // A recipe that builds nothing must be refused rather than treated as a
        // smaller one: it would report success having done nothing.
        let err = Recipe::parse("name = \"empty\"").expect_err("no command");
        assert!(format!("{err}").contains("command"), "{err}");
    }

    #[test]
    fn a_recipe_with_no_name_is_named_after_its_command() {
        let recipe = Recipe::parse("command = \"cargo\"").expect("usable");
        assert_eq!(recipe.name, "cargo");
    }

    /// **GUN-006: every process a build runs goes through one constructor.**
    ///
    /// The finding was not that `toolchain` was written carelessly — it was
    /// that there were TWO places that exec a `recipe.command` read out of the
    /// pushed tree, and only one of them was hardened. A fix that hardened the
    /// second one would leave a third free to forget again, which is why the
    /// hardening became a function and why this counts.
    ///
    /// Source-level on purpose. The property is "there is one constructor",
    /// and no behaviour of a single call can assert that; the file is the
    /// evidence. `include_str!` reads THIS file at compile time, so the guard
    /// cannot drift from what it guards.
    ///
    /// Mutation-proven 2026-09-11: put `Command::new` back in `toolchain` and
    /// the count is 2.
    #[test]
    fn the_runner_has_exactly_one_process_constructor() {
        let source = include_str!("runner.rs");
        let constructors = source
            .lines()
            // A doc comment quoting the old line is not a call site, and
            // neither is a string literal — including the needle on this very
            // line, which is why the quote character is spelled rather than
            // written.
            .filter(|line| !line.trim_start().starts_with("//"))
            .filter(|line| !line.contains(char::from(34)))
            .filter(|line| line.contains("std::process::Command::new("))
            .count();
        assert_eq!(
            constructors, 1,
            "every exec a build performs must go through `hardened_command`; \
             found {constructors} fully-qualified process constructors in runner.rs"
        );
        // …and the spelling cannot be dodged by importing the type, which
        // would make the count above read 1 while a second exec sat beside it.
        // Spelled in two halves so the needle is not in the file it searches.
        // A guard that trips on its own source proves nothing about the code
        // it guards.
        assert!(
            !source.contains(concat!("use std::process::", "Command;")),
            "runner.rs imports the process builder by name; the count above \
             only sees the fully-qualified spelling"
        );
    }

    /// The hardening is applied, and it is applied to a process that can be
    /// watched: a cleared environment and the working directory the build was
    /// checked out into.
    ///
    /// `/usr/bin/env` with no arguments prints the environment it was given,
    /// and `/bin/pwd` prints the directory it was started in. Both are
    /// coreutils and both are already assumed by the suites that drive `git`
    /// and `ssh`; neither needs a shell.
    #[test]
    fn the_hardened_command_clears_the_environment_and_sets_the_directory() {
        let dir = tempfile::tempdir().expect("a temp dir");
        // A marker the caller's environment carries and the child must not.
        // Set on the CHILD-to-be's parent — this process — through the same
        // `Command` API, so no global state is mutated under cargo's test
        // threads.
        let seen = |program: &str| -> Option<String> {
            let out = hardened_command(program, dir.path(), true)
                .env("GUNNAR_TEST_LEAK", "this must not cross")
                .output()
                .ok()?;
            out.status
                .success()
                .then(|| String::from_utf8_lossy(&out.stdout).into_owned())
        };
        // `.env()` after `env_clear()` puts the marker back deliberately: it
        // proves the child is readable at all, so an empty answer below cannot
        // be mistaken for a cleared environment.
        let Some(environment) = seen("/usr/bin/env") else {
            return; // no coreutils on this box; nothing to measure
        };
        assert!(
            environment.contains("GUNNAR_TEST_LEAK=this must not cross"),
            "the fixture cannot see the child's environment at all: {environment:?}"
        );
        // …and nothing else of the caller's. `PATH` and `HOME` are the two
        // `hardened_command` puts back by name.
        for line in environment.lines().filter(|l| !l.is_empty()) {
            let name = line.split('=').next().unwrap_or_default();
            assert!(
                matches!(name, "PATH" | "HOME" | "GUNNAR_TEST_LEAK"),
                "the build child inherited {name} from the operator's shell"
            );
        }

        if let Some(cwd) = seen("/bin/pwd") {
            assert_eq!(
                std::fs::canonicalize(cwd.trim()).ok(),
                std::fs::canonicalize(dir.path()).ok(),
                "the build child did not start in the checkout"
            );
        }
    }

    #[test]
    fn the_default_recipe_locks_its_dependencies() {
        // `--locked` is the whole point of the default: without it the build
        // may re-resolve, and the SBOM read from the lockfile then describes a
        // different build with nothing saying so.
        assert!(
            Recipe::cargo_default()
                .args
                .contains(&"--locked".to_owned()),
            "the default cargo recipe does not pass --locked"
        );
    }
}

// ───────────────────────────────────────────────────────────────────────────
// Watching
// ───────────────────────────────────────────────────────────────────────────

/// How the runner is configured when it watches rather than being told.
#[derive(Debug, Clone)]
pub struct WatchConfig {
    /// Where checkouts and build output go.
    pub workspace: PathBuf,
    /// The ACL principal every record is attributed to.
    pub builder: String,
    /// Build with the network up. Off by default; recorded when on.
    pub allow_network: bool,
    /// Only build these refs. Empty means every branch.
    ///
    /// A list and not a pattern: a pattern language here would be a second
    /// ref-matching implementation beside `gunnar-policy`'s, and two of those
    /// can disagree about which branch is protected.
    pub refs: Vec<String>,
    /// Stop after this many builds. `None` watches for ever; a bound is what a
    /// test uses, and what an operator uses to drain one queue and stop.
    pub max_builds: Option<usize>,
}

/// What one watched build turned into.
#[derive(Debug)]
pub enum Watched {
    /// A build ran and attested.
    Built(Box<Built>),
    /// A ref moved and the runner did not build it.
    Skipped {
        repo: String,
        r#ref: String,
        why: String,
    },
    /// A build was attempted and failed. **Not fatal to the watch**: one
    /// repository's broken build must not stop every other repository's.
    Failed {
        repo: String,
        commit: String,
        why: String,
    },
}

/// Should this ref update be built?
fn wanted(config: &WatchConfig, ref_name: &str) -> bool {
    // Branches only. A tag move is not a new state of the code — the commit it
    // points at was already pushed and already built — and a runner that built
    // both would attest twice to one commit under two recipes' worth of noise.
    if !ref_name.starts_with("refs/heads/") {
        return false;
    }
    config.refs.is_empty() || config.refs.iter().any(|want| want == ref_name)
}

/// **How the watch ended, as the server says it** (A28 M18 follow-up): the server runs
/// [`watch`] for its whole life, so ANY end while the server runs means pushes are no longer
/// built. Never silent: an error is an `ALERT runner` at ERROR, any other end a WARN.
pub fn said_end(end: &Result<usize>) -> (tracing::Level, String) {
    match end {
        Err(err) => (tracing::Level::ERROR, format!("ALERT runner: the build runner stopped watching this server's pushes: {err:#}; pushes are no longer built until the server restarts")),
        Ok(built) => (tracing::Level::WARN, format!("the build runner stopped watching this server's pushes after {built} build(s); pushes are no longer built until the server restarts")),
    }
}

/// **Watch the event bus and build what lands.**
///
/// Blocking, and deliberately so: the runner is one build at a time in the
/// process that watched the event. A queue and a pool are the phase-5 problem;
/// this is the part that has to be right, and "right" is easier to see when
/// there is one of it.
///
/// Every outcome — built, skipped, failed — is handed to `report` rather than
/// logged and dropped, so a caller can assert on what happened. A build that
/// fails does **not** stop the watch: one repository's broken recipe must not
/// take every other repository's builds with it.
pub fn watch(
    stores: &ServerStores,
    bus: &gunnar_grpc::EventBus,
    config: &WatchConfig,
    mut report: impl FnMut(Watched),
) -> Result<usize> {
    // `tokio_stream`, not `futures_util`: the stream is a `BroadcastStream` and
    // tokio-stream is already resolved here. One stream crate, not two.
    use tokio_stream::StreamExt as _;

    let filter = gunnar_grpc::EventFilter {
        namespaces: Vec::new(),
        repos: Vec::new(),
        kinds: vec![gunnar_proto::EventKind::RefUpdated],
        include_attestation_object_ids: false,
    };
    let mut stream = bus.subscribe(filter);
    let runtime = tokio::runtime::Builder::new_current_thread()
        .enable_all()
        .build()
        .context("building a runtime for the runner's subscription")?;

    let mut built = 0usize;
    while config.max_builds.is_none_or(|max| built < max) {
        let Some(event) = runtime.block_on(stream.next()) else {
            break;
        };
        // The stream yields a `Result` because a gRPC subscriber's does. An
        // error here is the subscription itself ending, not one event going
        // wrong, so it stops the watch rather than being skipped past.
        let Ok(event) = event else { break };
        let Some(gunnar_proto::event::Payload::RefUpdated(update)) = &event.payload else {
            continue;
        };
        let Some(repo) = update.repo.as_ref() else {
            continue;
        };
        let key = format!("{}/{}", repo.namespace, repo.repo);
        if !wanted(config, &update.ref_name) {
            report(Watched::Skipped {
                repo: key,
                r#ref: update.ref_name.clone(),
                why: "not a branch this runner builds".to_owned(),
            });
            continue;
        }
        let Ok(commit) = gix_hash::ObjectId::from_hex(update.new_oid.as_bytes()) else {
            report(Watched::Skipped {
                repo: key,
                r#ref: update.ref_name.clone(),
                why: format!("{} is not an object id", update.new_oid),
            });
            continue;
        };
        let Ok(repo_ref) = gunnar_core::RepoRef::new(&repo.namespace, &repo.repo) else {
            continue;
        };
        match build_commit(
            stores,
            &repo_ref,
            commit,
            &config.workspace,
            &config.builder,
            config.allow_network,
        ) {
            Ok(outcome) => {
                built += 1;
                report(Watched::Built(Box::new(outcome)));
            }
            Err(err) => report(Watched::Failed {
                repo: key,
                commit: commit.to_string(),
                why: format!("{err:#}"),
            }),
        }
    }
    Ok(built)
}