Use tests/common's binary() and git_sha256 in 40 more CLI tests (#824)

[Mikola Lysenko (mikolalysenko)]

LLM Description written by Claude Code:claude-opus-5-5

Refs #824 (children 2–3, slice 2; the tracker stays open for the files that open PRs change).

Summary

Forty more CLI test files drop their private binary() and git_sha256 and import the ones in tests/common instead. Standalone test binaries declare #[path = "common/mod.rs"] mod common;; directory binaries (apply/, cli/, get/, remove/, repair/, rollback/, scan/, e2e_vex_lockfile/) use crate::common. The two pnpm build suites keep their SOCKET_PATCH_PNPM_E2E_SOCKET_BIN override, renamed to socket_bin() so it no longer looks like a copy. The shared_helper_copies ratchet drops the 40 migrated files.

Why

• Issue: Tracking: share CLI test helpers through one test-support module instead of 100+ per-file copies #824 (register C30, register/20-audit-core.md; living document doc/08-tests-ci-docs.md, shared test support). It follows Use tests/common's binary() and git_sha256 in CLI tests (#824) #1124, which moved the first ~75 files.
• Leverage: B 0, U 0 (the test-binary merge, C31, waits on Mirror global flags into the environment once before dispatch, and read the lock timeout through one accessor #794), D 56 duplicate helper bodies collapsed (31 binary(), 25 git_sha256), S 0, R L (test-only). It ranked first among candidates whose files no open PR changes. Every production candidate with higher B or U (the manifest-load cluster An unparseable .socket/manifest.json is reported under five different --json error codes depending on the command #931/apply, apply --check and vendor report noManifest (exit 0) when .socket/manifest.json exists but can't be stat'd #998/scan reads a corrupt manifest as empty, so --prune skips GC and reports success #1063/A directory at .socket/manifest.json exits 2 with no --json output when --manifest-path spells the path, but exits 1 with manifest_unreadable under the default path #1123, Locate hosted Maven pom edits and their upstream restore through the formats::maven element scanner #717, Upstream restore sends every public-registry lookup at once; the registry_concurrency cap has no caller #1220, Chunk patch batch searches through one core helper with one set of batch limits #675, Mirror global flags into the environment once before dispatch, and read the lock timeout through one accessor #794) overlaps an open PR.

What changed

• 40 test files: private helpers deleted, shared ones imported; unused sha2/PathBuf imports removed.
• 9 standalone e2e suites take cache_env/hermetic through common instead of loading them twice (189dddb, from the final review).
• tests/cli/shared_helper_copies.rs: 40 entries removed from PENDING_PRIVATE_HELPERS (31 left: 27 files changed by open PRs, and 4 shared modules whose includers don't all declare mod common).

Deleted

git diff --stat origin/main: 41 files, +152/−421 (after 189dddb), all under crates/socket-patch-cli/tests. Production: 0 lines.

Behavior

None. Every removed git_sha256 was the Git-blob SHA-256 (SHA256("blob <len>\0" ++ bytes)) or a call to compute_git_sha256_from_bytes; tests/common's own tests pin both shapes as equal. Every removed binary() returned env!("CARGO_BIN_EXE_socket-patch") as a PathBuf or &'static str, and every caller compiles against the PathBuf version.

Test evidence

• cargo test -p socket-patch-cli --all-features --no-run on all 34 touched targets: no warnings.
• cargo test -p socket-patch-cli --all-features on the same targets plus spawn_env_hygiene: all pass (apply 119, cli 97 including the ratchet, get 81, remove 99, rollback 57, scan 120, e2e_vex_lockfile 344, vendor_jvm_cli 50, e2e_vendor_pnpm_build 49, e2e_vendor_pypi_build 52, …). The only failures are the 2 root-only repair tests (repair_exits_zero_and_stays_quiet_when_lock_file_unremovable, repair_cleanup_failure_is_reported_in_json_and_silent_modes). They fail the same way on main because the sandbox runs as root, and their files are not changed here.
• cargo clippy --workspace --all-features -- -D warnings: clean.
• rustfmt was run only on the touched files whose main copy was already rustfmt-clean (all 41).

Risk

Low. Test-only, mechanical, and the compiler checks every call site. The ignored e2e bodies (docker, toolchain builds) are compiled but not run here; they call the same helpers as before.

🤖 Generated with Claude Code

---

Note

Low Risk
Test-only mechanical deduplication; shared helpers are already pinned by tests/common and the ratchet test.

Overview
Forty CLI integration/e2e tests now import shared binary() and git_sha256() from tests/common instead of maintaining local copies. Standalone test binaries add #[path = "common/mod.rs"] mod common; nested suites under apply/, cli/, get/, etc. use crate::common. Several e2e suites also pull cache_env / hermetic through common so those modules are not loaded twice.

The shared_helper_copies ratchet drops the 40 migrated paths from PENDING_PRIVATE_HELPERS, leaving only files still pending or blocked by other PRs.

pnpm build e2e keeps the SOCKET_PATCH_PNPM_E2E_SOCKET_BIN override but renames the local helper to socket_bin() so it is not confused with the shared binary().

No production or CLI runtime behavior changes—duplicate helper bodies and unused sha2 imports are removed only.

^{Reviewed by Cursor Bugbot for commit 189dddb. Configure here.}

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

BugBot review

---

Generated by Claude Code

[cursor]

Stale Bugbot comment from a previous run.

[Mikola Lysenko (mikolalysenko)]

Labeled Ready for review by the burn-down agent.

• Head: bad5ac25b37969fbb56747cb7ed625d781db3790
• CI: all 293 check runs on this head completed success/skipped/neutral; no merge conflict with main.
• Bugbot: reviewed this head, no findings; no unresolved review threads.
• CHANGELOG.md untouched.

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

Final review brief (bad5ac25b)

What it does: 40 socket-patch-cli test files drop their private binary() and git_sha256 copies and use the shared ones from tests/common (more of #824). The shared_helper_copies ratchet loses the matching 40 entries. The two pnpm build suites keep their SOCKET_PATCH_PNPM_E2E_SOCKET_BIN override, now as socket_bin().

Risk: low. Test-only and mechanical: 41 files, +142/−401, no production code. Every removed copy matched the shared helper. git_sha256 was always SHA256("blob <len>\0" + content) or compute_git_sha256_from_bytes, and binary() was always CARGO_BIN_EXE_socket-patch. The compiler checks every call site, and no assertion was removed.

Look here:

• tests/common/mod.rs:46 binary and :126 git_sha256: the shared helpers. The module is #[allow(dead_code)] (:17).
• e2e_vendor_pnpm_build.rs:106: socket_bin() keeps the env override, and no bare binary() call is left in either pnpm suite.
• cli/shared_helper_copies.rs:16:`` the ratchet list.
• e2e_gem.rs:21-32: the wiring pattern; see the note below.

Verified: Compared every removed helper line by line with the shared one. cargo test -p socket-patch-cli --all-features --no-run builds all 29 touched targets with no warnings. Ran 10 of them, all passing: cli (incl. the ratchet), rollback, scan, e2e_vex_lockfile, e2e_vendor_pnpm_build, vendor_jvm_cli, cli_scan_silent, covgap_commands_scan_mod, mode_migration_cargo, e2e_gem. CHANGELOG.md untouched. CI 293/293 success/skipped (ci-ok, clippy green), Bugbot success, no review threads, mergeable.

Changes I made: none.

Open questions / non-blocking note: 9 standalone files now pull in common/mod.rs (which already has pub mod cache_env/hermetic) and also keep their own #[path = "common/cache_env.rs"]/hermetic.rs mods: e2e_gem, e2e_npm, e2e_pypi, e2e_redirect_{gem,gradle}build, e2e_scan, e2e_vendor{composer,pnpm,pypi}_build. That adds clippy::duplicate_mod hits, and 6 targets that were clean now fail under local cargo clippy --all-targets. CI's clippy runs without --all-targets (ci.yml:81), and main already has 312 such errors, so nothing gates on this. A later #824 pass can switch those files to use common::{cache_env, hermetic};. e2e_gem.rs also keeps a private git_sha256_file, which is outside this PR's scope.

Auto-merge is armed, so approving sends it straight to the merge queue.

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

[agent] Addressed the final-review note in 189dddb: the 9 standalone suites (e2e_gem, e2e_npm, e2e_pypi, e2e_redirect_{gem,gradle}build, e2e_scan, e2e_vendor{composer,pnpm,pypi}_build) now take use common::{cache_env, hermetic} instead of re-loading those files through their own #[path] mods, so the cache_env.rs/hermetic.rs duplicate_mod hits are gone. All 9 targets pass (cargo test -p socket-patch-cli --all-features).

One duplicate_mod hit of the same kind remains under --all-targets: common/jvm_env.rs is loaded by both common/mod.rs and the shared prebuilt_common/mod.rs in the four suites that include both modules. e2e_vex_lockfile/main.rs on main already has this pattern. Fixing it means changing prebuilt_common, whose includers don't all declare mod common, so it stays for a later #824 pass. e2e_gem.rs's private git_sha256_file is also left for that pass.

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

BugBot review

---

Generated by Claude Code

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 189dddb. Configure here.}

[Mikola Lysenko (mikolalysenko)]

[final reviewer] Disarmed auto-merge: the head moved from bad5ac25b (my brief) to 189dddb, a non-merge commit that addresses the brief note (shared cache_env/hermetic helpers in 9 standalone suites). ci-ok hasn't finished on the new head yet. I'll re-review the delta and re-arm once CI is green and the brief is updated.

---

Generated by Claude Code