RebaseArchiveEntries: fix archive path rebasing

[thaJeztah]

• follow-up to RebaseArchiveEntries: use POSIX / Unix paths #42 (comment)

Handle rebasing from the archive root explicitly so that both relative
and absolute entry names are joined to newBase with exactly one path
separator.

Also restrict rebasing to complete path prefixes. The previous use of
strings.Replace could replace oldBase outside the beginning of the
name or as part of another path component, such as rebasing origin in
original/file.

Apply the same handling to hardlink targets, while preserving the
remainder of archive paths without cleaning or canonicalizing them.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 65.16%. Comparing base (216738e) to head (8829a25).
⚠️ Report is 61 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #43      +/-   ##
==========================================
- Coverage   65.81%   65.16%   -0.65%     
==========================================
  Files          42       44       +2     
  Lines        2039     2271     +232     
==========================================
+ Hits         1342     1480     +138     
- Misses        519      592      +73     
- Partials      178      199      +21     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[vvoland]

What about

  {
  	name:       "rebase absolute path from root",
  	oldBase:    "/",
  	newBase:    "prefix",
  	headerName: "/foo/bar",
  	wantName:   "prefix/foo/bar",
  },

?

IIUC this would produce prefix//foo/bar

[thaJeztah]

Hm; right; don't think headers would have trailing slashes (or could they have?)

[vvoland]

Generall they should be relative but archive/tar accepts absolute header names such as /foo/bar AFAIK.

[Copilot]

Pull request overview

This PR refines RebaseArchiveEntries so archive entry names (and hardlink targets) are rebased by matching only full path prefixes, with special handling for rebasing from the archive root while preserving the remainder of entry paths verbatim (no cleaning/canonicalization).

Changes:

• Introduces rebaseName to perform prefix-only rebasing and handle oldBase == "/" explicitly.
• Replaces strings.Replace-based rebasing for hdr.Name and hardlink hdr.Linkname with the new prefix-safe logic.
• Expands test coverage for root rebasing, prefix matching, and preservation of unclean paths.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
copy.go	Adds rebaseName and uses it to rebase tar header names and hardlink targets safely (prefix-only, root-aware).
copy_test.go	Adds test cases covering root rebasing, exact-prefix behavior, and preservation of unclean archive paths.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

copy_test.go:56

• This test case duplicates the previous "rebase absolute name from root to empty" case (same name and same headerName), and it also omits newBase (implicitly ""). This makes the table harder to read and likely misses the intended coverage (e.g., rebasing a relative name from root to empty).

			name:       "rebase absolute name from root to empty",
			oldBase:    "/",
			headerName: "/foo/bar",
			wantName:   "foo/bar",
		},

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (3)

copy.go:352

• The doc comment says “replacing an occurrence of oldBase with newBase at the beginning of entry names”, but the implementation now only rebases when oldBase matches a complete path prefix (whole name or followed by "/"), and has special handling for archive-root rebases. Updating the comment would prevent callers from assuming substring replacement semantics.

// RebaseArchiveEntries rewrites the given srcContent archive replacing
// an occurrence of oldBase with newBase at the beginning of entry names.
func RebaseArchiveEntries(srcContent io.Reader, oldBase, newBase string) io.ReadCloser {
	rebase := newNameRebaser(oldBase, newBase)

copy.go:328

• strings.TrimSuffix(..., "/") removes at most one trailing slash, but the comment says trailing slashes are trimmed and root rebasing should treat base paths like "///" as root. With the current code, oldBase="///" becomes "//" and won’t hit the oldBase == "" root case, and newBase="prefix//" becomes "prefix/", potentially producing double separators. Use strings.TrimRight(..., "/") to remove all trailing slashes.

This issue also appears on line 349 of the same file.

	oldBase = strings.TrimSuffix(filepath.ToSlash(oldBase), "/")
	newBase = strings.TrimSuffix(filepath.ToSlash(newBase), "/")

copy_test.go:56

• This test case is a duplicate of the previous one (same name and effectively the same inputs/expectations since newBase defaults to ""). It makes the table harder to read and doesn’t add coverage. Consider repurposing it to cover the missing scenario (relative name from root rebased to empty) or remove it.

			name:       "rebase absolute name from root to empty",
			oldBase:    "/",
			headerName: "/foo/bar",
			wantName:   "foo/bar",
		},