Skip to content
zit Artifact [be8e20a2c6]
Login

Artifact [be8e20a2c6]

Check-ins Using Download Hex

Artifact be8e20a2c6fc9f83a981611e99882e41bf0d9b520f48a32487f5512f72dfa9a9:

//! tidy: a dead-file guard over the src/ test surface.
//!
//! Zig compiles and runs only the tests reachable from a root the build wires.
//! A `.zig` file that no root imports still COMPILES — its tests simply stop
//! running, and the coverage evaporates with no error. This guard closes that
//! hazard: it walks `src/**/*.zig`, builds the `@import` graph, and asserts every
//! file is reachable from a root `zig build test` runs — the library, CLI, and
//! DST roots plus the separately compiled Git-import test roots. Keeping those
//! importer tests as build roots avoids path-importing their private modules
//! into the public `zit` module; dedicated process and benchmark roots are
//! instead reviewed in `allowlist`.
//!
//! The mechanism is borrowed from TigerBeetle's `src/tidy.zig` DeadFilesDetector
//! (github.com/tigerbeetle/tigerbeetle), which walks its own source tree and
//! fails the build when a file falls out of the import graph.
//!
//! It exists BEFORE the `main.zig` -> `src/cmd/` split precisely so that split
//! cannot silently orphan a file's tests: move a file, forget to re-root it, and
//! this test names the file and the one-line fix.
//!
//! Imports are extracted with the Zig tokenizer, not a byte scan, so an
//! `@import("...")` written inside a string literal or comment — as this very
//! file does in the remediation message below — is correctly NOT read as an
//! import edge.

const std = @import("std");

/// The repository root, wired in build.zig's `build_unit_test_module` (the same
/// handle src/packaging_test.zig uses). A test binary's working directory is not
/// the checkout, so the src tree is opened by this absolute path rather than a
/// cwd-relative one.
const repo_options = @import("repo_options");

/// The test roots `zig build test` executes, each of which aggregates other
/// files through `_ = @import("...")` lines that can be dropped by hand, plus
/// the independent importer test modules build.zig runs through the same test
/// step. Paths are relative to `src/`.
const roots = [_][]const u8{
    "root.zig",
    "main.zig",
    "dst.zig",
    "git/git_import_test.zig",
    "git/git_import_refs_test.zig",
    "git/git_import_cli_test.zig",
    "git/git_export_cli_test.zig",
    "git/git_export_e2e.zig",
};

/// Files legitimately unreachable from the test roots because build.zig wires
/// them as their OWN roots through dedicated steps. Each was verified against
/// build.zig — this is the reviewed exception set, not a convenience to silence
/// a real orphan. If one of these ever becomes reachable from a test root, the
/// hygiene check below fails and demands the entry be removed. Paths are
/// relative to `src/`.
const allowlist = [_][]const u8{
    // Root of `zig build e2e`. `zig build test` compiles it as a type-check gate
    // but never runs it (it spawns the real `zit` binary ~150x). Its own tests
    // run under the e2e step, so it needs no test-root import.
    "e2e.zig",
    // Root of `zig build smoke`. Compiled-but-not-run by `test`, run only by the
    // milestone-manual smoke step that drives the real `fossil` binary.
    "smoke.zig",
    // Root of `zig build muscle` (zit#162): the fossil-muscle-memory
    // differential transcript. Same shape as smoke — compiled by `test`
    // (its judge has unit tests there), run only by the manual step.
    "muscle.zig",
    // Root of `zig build bench`. build.zig ALSO compiles it into a test binary
    // (`bench_tests`) that joins `zig build test`, so its inline tests do run —
    // just not through a tidy root. It is its own root.
    "logstore_benchmark.zig",
    // Reached only from logstore_benchmark.zig (the bench root above); their
    // tests ride the same `bench_tests` binary into `zig build test`.
    "storage_benchmark.zig",
    "sync_benchmark.zig",
    "warmread_benchmark.zig",
    // The `serve` readiness probe (`zig build bench -- serve`), dispatched from
    // logstore_benchmark.zig; its pure tests ride the same `bench_tests` binary.
    "serve_benchmark.zig",
    // `git-import-e2e` is compile-gated by `test` but runs only by its explicit
    // process-heavy step, just like the general e2e root above.
    "git/git_import_e2e.zig",
    // The importer benchmark is out-of-band, and this support module is reached
    // only from that separate executable root.
    "git/git_import_benchmark.zig",
    // Build-only pipe helper used by the CLI transaction test's build option.
    // It is an executable root, not a module a test root can import.
    "git/git_import_cli_pipe_helper.zig",
};

/// Vacuity floor: the walk must find at least this many `.zig` files or the tree
/// was not actually scanned (wrong path, empty handle) and every check below
/// would pass by finding nothing. The tree holds ~60 files today; 40 is a floor
/// with headroom, not a running count to keep in sync.
const min_expected_files = 40;

/// Read cap per source file. The largest src file is well under 1 MiB; 8 MiB is
/// a tripwire against a runaway read, not an expected size.
const max_source_bytes = 8 << 20;

test "every src/**/*.zig is reachable from a test root" {
    const io = std.testing.io;

    // One arena for every path string and graph structure: keys stay pointer
    // stable for the hash maps, and a single deinit reclaims all of it.
    var arena_state = std.heap.ArenaAllocator.init(std.testing.allocator);
    defer arena_state.deinit();
    const arena = arena_state.allocator();

    const src_abs = try std.fmt.allocPrint(arena, "{s}/src", .{repo_options.root});

    var src_dir = std.Io.Dir.cwd().openDir(io, src_abs, .{ .iterate = true }) catch |err| switch (err) {
        // The only legitimate skip: the wired root is not a directory here, so
        // the machine cannot answer the question — the same stance the
        // version-control pin in packaging_test.zig takes. Anything else is a
        // real fault and must fail loudly rather than skip into silence.
        error.FileNotFound, error.NotDir => return error.SkipZigTest,
        else => {
            std.debug.print("\ntidy: cannot open {s}: {s}\n", .{ src_abs, @errorName(err) });
            return error.TidyWalkFailed;
        },
    };
    defer src_dir.close(io);

    // Pass 1 — every src/**/*.zig, path relative to src/. On POSIX the walker's
    // separator is '/', matching both the import strings and the roots/allowlist
    // above; a '\\'-separated platform would make every lookup miss and fail this
    // guard loudly, never pass it wrongly.
    var files: std.ArrayList([]const u8) = .empty;
    var file_set: std.StringHashMapUnmanaged(void) = .empty;
    {
        var walker = try src_dir.walk(arena);
        defer walker.deinit();
        while (try walker.next(io)) |entry| {
            if (entry.kind != .file) continue;
            if (!std.mem.endsWith(u8, entry.path, ".zig")) continue;
            const rel = try arena.dupe(u8, entry.path);
            try files.append(arena, rel);
            try file_set.put(arena, rel, {});
        }
    }

    // Vacuity guard: a silent zero (wrong cwd, empty handle) would let every
    // assertion below pass by finding nothing to check.
    if (files.items.len < min_expected_files) {
        std.debug.print("\ntidy: walked {s} and found only {d} .zig files; expected at least {d}. " ++
            "The walk is not seeing the source tree — this guard would pass vacuously.\n", .{ src_abs, files.items.len, min_expected_files });
        return error.TidyWalkFailed;
    }

    // Each root must be a real file, or reachability starts from nothing and
    // waves everything through. A renamed or moved root has to fail here.
    for (roots) |root| {
        if (!file_set.contains(root)) {
            std.debug.print("\ntidy: test root \"{s}\" does not exist under src/. The reachability " ++
                "guard starts from the roots build.zig wires into `zig build test`; update this list when a root moves.\n", .{root});
            return error.TidyRootMissing;
        }
    }

    // Pass 2 — breadth-first over the @import graph from every test root.
    var reached: std.StringHashMapUnmanaged(void) = .empty;
    var queue: std.ArrayList([]const u8) = .empty;
    for (roots) |root| {
        const key = file_set.getKey(root).?; // just asserted present
        if ((try reached.getOrPut(arena, key)).found_existing) continue;
        try queue.append(arena, key);
    }
    var head: usize = 0;
    while (head < queue.items.len) : (head += 1) {
        const file = queue.items[head];
        const source = try src_dir.readFileAllocOptions(io, file, arena, .limited(max_source_bytes), .of(u8), 0);
        var imports: std.ArrayList([]const u8) = .empty;
        try collectImports(arena, source, parentDir(file), &imports);
        for (imports.items) |target| {
            // Imports of `std`, `zit`, `build_options`, `repo_options`, etc. carry
            // no `.zig` and are already filtered; a resolved path not in the file
            // set is a cross-module or generated import and is not our concern.
            const key = file_set.getKey(target) orelse continue;
            if ((try reached.getOrPut(arena, key)).found_existing) continue;
            try queue.append(arena, key);
        }
    }

    // Allowlist hygiene: every entry must be a real file that is genuinely
    // unreachable. A stale entry (file gone) or a dead one (now reachable) means
    // the list papers over nothing — fail so it cannot rot into cover for a
    // future real orphan.
    for (allowlist) |allowed| {
        if (!file_set.contains(allowed)) {
            std.debug.print("\ntidy: allowlist names \"{s}\", which no longer exists under src/. " ++
                "Remove the stale entry.\n", .{allowed});
            return error.TidyAllowlistStale;
        }
        if (reached.contains(allowed)) {
            std.debug.print("\ntidy: allowlist names \"{s}\", but it is now reachable from a test root. " ++
                "Remove it from the allowlist — the exemption is dead.\n", .{allowed});
            return error.TidyAllowlistDead;
        }
    }

    // The assertion: every src file is reached or explicitly allowlisted.
    var orphans: usize = 0;
    for (files.items) |file| {
        if (reached.contains(file)) continue;
        if (isAllowlisted(file)) continue;
        orphans += 1;
        std.debug.print(
            \\
            \\tidy: src/{s} is not reachable from any test root.
            \\      Add `_ = @import("{s}");` to the test root that owns its module, or
            \\      add a reviewed build-only exception when an import would duplicate a module.
            \\      A file no test root imports still compiles — its tests just never run.
            \\
        , .{ file, file });
    }
    if (orphans != 0) return error.TidyDeadFile;
}

fn isAllowlisted(file: []const u8) bool {
    for (allowlist) |a| {
        if (std.mem.eql(u8, a, file)) return true;
    }
    return false;
}

/// The directory portion of a src-relative path: `sim/net.zig` -> `sim`, and a
/// file directly under src/ -> `""`.
fn parentDir(path: []const u8) []const u8 {
    const slash = std.mem.lastIndexOfScalar(u8, path, '/') orelse return "";
    return path[0..slash];
}

/// Extract every `@import("...zig")` from `source`, resolve each against the
/// importing file's directory, and append the src-relative results to `out`.
///
/// The Zig tokenizer drives this so imports inside comments or string literals
/// are not mistaken for edges. Import path literals carry no escapes, so the slice
/// between the string_literal token's quotes is the literal path.
fn collectImports(
    arena: std.mem.Allocator,
    source: [:0]const u8,
    importer_dir: []const u8,
    out: *std.ArrayList([]const u8),
) !void {
    var tokenizer = std.zig.Tokenizer.init(source);
    while (true) {
        const builtin = tokenizer.next();
        if (builtin.tag == .eof) break;
        if (builtin.tag != .builtin) continue;
        if (!std.mem.eql(u8, source[builtin.loc.start..builtin.loc.end], "@import")) continue;

        const lparen = tokenizer.next();
        if (lparen.tag != .l_paren) continue;
        const literal = tokenizer.next();
        if (literal.tag != .string_literal) continue;

        std.debug.assert(literal.loc.end - literal.loc.start >= 2);
        const raw = source[literal.loc.start + 1 .. literal.loc.end - 1];
        if (!std.mem.endsWith(u8, raw, ".zig")) continue;

        try out.append(arena, try resolveImport(arena, importer_dir, raw));
    }
}

/// Join an import target onto the importing file's directory and collapse `.`
/// and `..` components, yielding an arena-owned src-relative path. No real import
/// escapes src/ today; the `..` handling keeps the resolver correct if one ever
/// does.
fn resolveImport(arena: std.mem.Allocator, dir: []const u8, target: []const u8) ![]const u8 {
    var components: std.ArrayList([]const u8) = .empty;
    defer components.deinit(arena);

    var dir_it = std.mem.splitScalar(u8, dir, '/');
    while (dir_it.next()) |component| try pushComponent(&components, arena, component);
    var target_it = std.mem.splitScalar(u8, target, '/');
    while (target_it.next()) |component| try pushComponent(&components, arena, component);

    var out: std.ArrayList(u8) = .empty;
    defer out.deinit(arena);
    for (components.items, 0..) |component, i| {
        if (i != 0) try out.append(arena, '/');
        try out.appendSlice(arena, component);
    }
    return arena.dupe(u8, out.items);
}

fn pushComponent(components: *std.ArrayList([]const u8), arena: std.mem.Allocator, component: []const u8) !void {
    if (component.len == 0 or std.mem.eql(u8, component, ".")) return;
    if (std.mem.eql(u8, component, "..")) {
        if (components.items.len != 0) _ = components.pop();
        return;
    }
    try components.append(arena, component);
}