[PR #580] [MERGED] fix(artifacts): retain relative paths in directory identities #2080

New issue

Closed

opened 2026-10-16 21:14:01 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-16 21:14:01 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/580
Author: @kaoru0822-kitauji
Created: 10/5/2026
Status: ✅ Merged
Merged: 10/5/2026
Merged by: @morluto

Base: main ← Head: fix/artifact-directory-relative-paths

📝 Commits (1)

39322fd fix(artifacts): retain relative paths in directory identities

📊 Changes

2 files changed (+180 additions, -1 deletions)

View changed files

📝 src/application/ArtifactGraphConstruction.ts (+2 -1)
➕ tests/boundary/filesystem/artifactDirectoryIdentity.test.ts (+178 -0)

📄 Description

Summary

Preserve the complete child path relative to its represented directory when computing virtual-directory identities.

No tracking issue: found during real ZIP inventory/comparison testing; no matching open issue or PR was found.

Problem and expected behavior

A ZIP containing pkg/ and pkg/a/data.txt, without an explicit pkg/a/ entry, attaches the file to the nearest represented directory pkg. The directory hash previously retained only data.txt:

Renaming the implicit component a to b with identical file bytes left pkg's verified artifact ID unchanged. compare_artifacts still reported the file removed/added and archive root changed, but incorrectly counted pkg unchanged.
With a/data.txt=A and b/data.txt=B, reversing the two file entries changed pkg's identity and produced a spurious directory change.

Change and scope

Use the observed path relative to the represented directory as the existing child-name hash/sort key. This retains missing intermediate path components without synthesizing directory occurrences.

The hash shape and immediate-child names are unchanged. Empty and fully represented directory trees retain their exact existing identities; moving a directory and its contents together remains identity-neutral. Affected directory identities and ancestors change where omitted path components were previously lost.

This is not a general archive-ordering or collation overhaul: the existing localeCompare comparator is retained, including pre-existing equal-collation behavior for distinct unusual names. Late-directory parenting is handled separately in #578. Nested-ASAR parent selection is also unchanged.

Contract and boundary impact

Semantic owner and earliest changed stage: application artifact graph construction, virtual-directory hash materialization.
CLI and MCP/tool-catalog contract: unchanged schemas/tools; corrected directory artifact IDs and comparison classifications flow through shared inventory.
Provider, bridge, target-format, or platform compatibility: real ZIP fixtures tested on Linux; no reader/protocol changes.
Evidence, artifact, provenance, or reconstruction contract: preserve observed relative child paths in content-addressed virtual-directory identity; file content digests and occurrences remain unchanged.
Process execution, authorization, cleanup, or containment impact: none; no extraction or new execution, and no path-validation change.
Generated metadata, package, or installation impact: none.

Evidence and regression coverage

Six real ZIP cases cover implicit-path rename, reversed equal-basename entries, direct/implicit directory relocation, exact pre-fix IDs for ordinary immediate children and empty directories, and nested explicit-directory compatibility.

On unchanged main b225b0f, the two defect regressions fail and four compatibility controls pass. The fixed branch passes all six.

Executed the built inspect-artifact <zip> --format json CLI for four fixtures and passed its unmodified inventory Evidence into the production comparison:

implicit-path rename: pkg is changed;
reversed equal-basename members: only the raw archive root changes, while pkg and both files remain unchanged.

Independent review found no blocker within this bounded scope.

Observed, derived, and inferred claims remain distinguishable.
Artifact identity, source provenance, and failed attempts remain preserved.
Unsupported, incomplete, unavailable, or uncertain outcomes remain visible.

Validation performed

Final branch rebased onto main b225b0f; focused regressions also reproduced on that unchanged base.
npm run check:pr: static checks passed; 1,851 tests passed, 1 skipped, 40 failed across seven native bridge/lease suites due to sandbox Unix-domain socket listen EPERM. The same seven suites on unchanged b225b0f reproduce 40 failures / 22 passes. This is not a full local test-suite pass.
npm run docs:check: passed separately because full-test failures short-circuit this step.
npm run check:ci: passed.
npm_config_cache=/tmp/rea-npm-cache npm run verify:package: passed, including packaged CLI/MCP checks.
npm pack --dry-run --cache /tmp/rea-npm-cache: passed.
Four real CLI inventory/comparison probes: passed on the final build.
git diff --check: passed.

Compatibility, safety, and release

No schema, hash-envelope, dependency, or release-metadata changes. Exact legacy identity controls guard unaffected directories. Corrected IDs for omitted-path cases are intentional; no migration command is required.

Real Hopper/Ghidra and native macOS/Windows verification was not run locally. No archive safety, process, authorization, or containment policy is weakened.

Review checklist

One focused outcome with a Conventional Commit title.
Reason for no linked issue stated.
Regression and compatibility controls included.
Owning contracts/docs remain accurate; generated checks run.
Representative CLI/comparison behavior executed.
Final diff checked for secrets, unrelated cleanup, and unsupported claims.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/580 **Author:** [@kaoru0822-kitauji](https://github.com/kaoru0822-kitauji) **Created:** 10/5/2026 **Status:** ✅ Merged **Merged:** 10/5/2026 **Merged by:** [@morluto](https://github.com/morluto) **Base:** `main` ← **Head:** `fix/artifact-directory-relative-paths` --- ### 📝 Commits (1) - [`39322fd`](https://github.com/morluto/rea/commit/39322fd15b63dd5e03d1446835fa7e57f09d861b) fix(artifacts): retain relative paths in directory identities ### 📊 Changes **2 files changed** (+180 additions, -1 deletions) <details> <summary>View changed files</summary> 📝 `src/application/ArtifactGraphConstruction.ts` (+2 -1) ➕ `tests/boundary/filesystem/artifactDirectoryIdentity.test.ts` (+178 -0) </details> ### 📄 Description ## Summary Preserve the complete child path relative to its represented directory when computing virtual-directory identities. No tracking issue: found during real ZIP inventory/comparison testing; no matching open issue or PR was found. ## Problem and expected behavior A ZIP containing `pkg/` and `pkg/a/data.txt`, without an explicit `pkg/a/` entry, attaches the file to the nearest represented directory `pkg`. The directory hash previously retained only `data.txt`: - Renaming the implicit component `a` to `b` with identical file bytes left `pkg`'s verified artifact ID unchanged. `compare_artifacts` still reported the file removed/added and archive root changed, but incorrectly counted `pkg` unchanged. - With `a/data.txt=A` and `b/data.txt=B`, reversing the two file entries changed `pkg`'s identity and produced a spurious directory change. ## Change and scope Use the observed path relative to the represented directory as the existing child-name hash/sort key. This retains missing intermediate path components without synthesizing directory occurrences. The hash shape and immediate-child names are unchanged. Empty and fully represented directory trees retain their exact existing identities; moving a directory and its contents together remains identity-neutral. Affected directory identities and ancestors change where omitted path components were previously lost. This is not a general archive-ordering or collation overhaul: the existing `localeCompare` comparator is retained, including pre-existing equal-collation behavior for distinct unusual names. Late-directory parenting is handled separately in #578. Nested-ASAR parent selection is also unchanged. ## Contract and boundary impact - Semantic owner and earliest changed stage: application artifact graph construction, virtual-directory hash materialization. - CLI and MCP/tool-catalog contract: unchanged schemas/tools; corrected directory artifact IDs and comparison classifications flow through shared inventory. - Provider, bridge, target-format, or platform compatibility: real ZIP fixtures tested on Linux; no reader/protocol changes. - Evidence, artifact, provenance, or reconstruction contract: preserve observed relative child paths in content-addressed virtual-directory identity; file content digests and occurrences remain unchanged. - Process execution, authorization, cleanup, or containment impact: none; no extraction or new execution, and no path-validation change. - Generated metadata, package, or installation impact: none. ## Evidence and regression coverage Six real ZIP cases cover implicit-path rename, reversed equal-basename entries, direct/implicit directory relocation, exact pre-fix IDs for ordinary immediate children and empty directories, and nested explicit-directory compatibility. On unchanged main `b225b0f`, the two defect regressions fail and four compatibility controls pass. The fixed branch passes all six. Executed the built `inspect-artifact <zip> --format json` CLI for four fixtures and passed its unmodified inventory Evidence into the production comparison: - implicit-path rename: `pkg` is changed; - reversed equal-basename members: only the raw archive root changes, while `pkg` and both files remain unchanged. Independent review found no blocker within this bounded scope. - [x] Observed, derived, and inferred claims remain distinguishable. - [x] Artifact identity, source provenance, and failed attempts remain preserved. - [x] Unsupported, incomplete, unavailable, or uncertain outcomes remain visible. ## Validation performed - Final branch rebased onto main `b225b0f`; focused regressions also reproduced on that unchanged base. - `npm run check:pr`: static checks passed; **1,851 tests passed, 1 skipped, 40 failed across seven native bridge/lease suites** due to sandbox Unix-domain socket `listen EPERM`. The same seven suites on unchanged `b225b0f` reproduce **40 failures / 22 passes**. This is not a full local test-suite pass. - `npm run docs:check`: passed separately because full-test failures short-circuit this step. - `npm run check:ci`: passed. - `npm_config_cache=/tmp/rea-npm-cache npm run verify:package`: passed, including packaged CLI/MCP checks. - `npm pack --dry-run --cache /tmp/rea-npm-cache`: passed. - Four real CLI inventory/comparison probes: passed on the final build. - `git diff --check`: passed. ## Compatibility, safety, and release No schema, hash-envelope, dependency, or release-metadata changes. Exact legacy identity controls guard unaffected directories. Corrected IDs for omitted-path cases are intentional; no migration command is required. Real Hopper/Ghidra and native macOS/Windows verification was not run locally. No archive safety, process, authorization, or containment policy is weakened. ## Review checklist - [x] One focused outcome with a Conventional Commit title. - [x] Reason for no linked issue stated. - [x] Regression and compatibility controls included. - [x] Owning contracts/docs remain accurate; generated checks run. - [x] Representative CLI/comparison behavior executed. - [x] Final diff checked for secrets, unrelated cleanup, and unsupported claims. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-16 21:14:01 +00:00