fix(core): list a distribution once, however many times sys.path finds it

Found by running, not by reading: `pipdock list` on a venv with one editable install
printed the package twice.

`importlib.metadata.distributions()` walks sys.path and yields one entry per metadata
directory. `pip install -e` puts the project's own src/<name>.egg-info on sys.path via the
.pth file, so the package is discovered both there and at the venv's .dist-info. `pip list`
shows it once; PipDock showed it twice.

Deduplicated by normalized name, first occurrence winning — sys.path precedence, which is
the copy an `import` would actually resolve to.

This matters more for S2 than for the CLI. The Installed table joins `pkg_list` against
`pkg_outdated` on the normalized name, so a duplicate is an ambiguous key, and it renders
as two rows a user cannot tell apart.

Covered by a new L2 step rather than a unit test: there is no Python test harness, and the
condition needs a real editable install to exist at all. The step asserts both of this
area's claims — listed once, and reporting no size. **Both halves were exercised**: with
the fix reverted the count assertion throws with "got 2", and with it restored the step
passes and leaves $LASTEXITCODE at 0, which is the trap the guard step hit in L2's first
runs. The pyproject is written as a string array rather than a here-string, because a
here-string's closing '@ has to land at column 0 after YAML strips the block indentation.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
poli0981andClaude Opus 5 committed 2026-08-04 15:46:48 +07:00
1 parent e7bc6a5d4d
commit bd05157a9d
2 files changed
+51 -1

No files matched your search

+36
View File
@@ -188,6 +188,42 @@ jobs:
}
.\target\debug\pipdock.exe list --env .itest --json | ConvertFrom-Json | Out-Null
# Two probe claims that only a real editable install can test: it must be listed once
# (its own .egg-info joins the venv's .dist-info on sys.path), and it must report no
# size (its RECORD lists the import shim, not the sources — a few hundred bytes for a
# project of any size, which is worse than reporting nothing).
- name: End-to-end — an editable install is listed once, and without a size
shell: pwsh
run: |
$ErrorActionPreference = 'Stop'
New-Item -ItemType Directory -Force .itest-editable\src\edmod | Out-Null
# An array rather than a here-string: a here-string's closing '@ must sit at column 0
# after YAML strips this block's indentation, which makes it break on a re-indent.
Set-Content .itest-editable\pyproject.toml @(
'[build-system]',
'requires = ["setuptools>=64"]',
'build-backend = "setuptools.build_meta"',
'[project]',
'name = "edmod"',
'version = "1.0.0"'
)
Set-Content .itest-editable\src\edmod\__init__.py '"""edmod."""'
# 2 MiB of real content, so a wrong size would be unmistakable.
Set-Content .itest-editable\src\edmod\payload.py ('DATA = "' + ('y' * 2MB) + '"')
.itest\Scripts\python.exe -m pip install --quiet -e .itest-editable
$dists = .\target\debug\pipdock.exe list --env .itest --json | ConvertFrom-Json
$ed = @($dists | Where-Object { $_.name -eq 'edmod' })
if ($ed.Count -ne 1) {
throw "edmod should be listed once, got $($ed.Count) — probe.py is not deduplicating"
}
if ($null -ne $ed[0].sizeBytes) {
throw "an editable install must report no size, got $($ed[0].sizeBytes) bytes"
}
$sized = @($dists | Where-Object { $null -ne $_.sizeBytes })
if ($sized.Count -lt 1) { throw "no distribution reported a size at all" }
Write-Host "$($dists.Count) distributions, $($sized.Count) with a size, edmod correctly without"
fixture-drift:
name: Re-capture against the latest engines
# Weekly rather than nightly: it hits PyPI hard enough that daily would be rude, and engine
+15 -1
View File
@@ -172,10 +172,20 @@ def _dists() -> list[dict[str, object]]:
A single unreadable distribution must not fail the whole probe -- broken
metadata in one package is common and the user still needs the rest of the
list. Such entries are reported with an `error` field instead.
Distributions are deduplicated by normalized name, **first occurrence winning**.
`importlib.metadata.distributions()` walks sys.path in order and yields one entry per
metadata directory it finds, so a package discoverable twice is reported twice -- most
commonly after `pip install -e`, where the project's own src/<name>.egg-info joins the
venv's .dist-info via the .pth file. `pip list` shows such a package once, and so must
PipDock: a duplicate row means an ambiguous key when the Installed table joins against
the outdated set, and two rows the user cannot tell apart. First-wins matches sys.path
precedence, which is the copy an `import` would actually resolve to.
"""
from importlib.metadata import distributions
out: list[dict[str, object]] = []
seen: set[str] = set()
for dist in distributions():
try:
meta = dist.metadata
@@ -188,9 +198,13 @@ def _dists() -> list[dict[str, object]]:
raw_name = meta.get("Name")
if not raw_name:
continue
name = _normalize(raw_name)
if name in seen:
continue
seen.add(name)
out.append(
{
"name": _normalize(raw_name),
"name": name,
"version": dist.version or "",
"requires_dist": list(dist.requires or []),
"requires_python": meta.get("Requires-Python"),