fix: guard against IndexError in decompress_list on empty string

[prabindersinghh]

Summary

Fixes #3017

compress_list([]) legitimately returns "" (joining an empty list).
However, decompress_list("") immediately crashed with IndexError
on lststr[0] because there was no guard for empty input, making the
round-trip decompress_list(compress_list([])) crash instead of
returning [].

Fix: add a two-line early-return guard before the index access.

Changes

• metaflow/util.py — 2 lines added (early return guard)
• test/unit/test_util_compress.py — new file, 6 unit tests

Reproducer

from metaflow.util import compress_list, decompress_list

compress_list([])            # returns ""
decompress_list(compress_list([]))  # IndexError before fix

Testing

6 new unit tests in test/unit/test_util_compress.py:

Test	What it verifies
test_compress_empty_list_returns_empty_string	compress_list([]) returns ""
test_decompress_empty_string_returns_empty_list	Core regression: no crash on ""
test_roundtrip_empty_list	Full round-trip returns []
test_roundtrip_single_item	Non-empty round-trip unaffected
test_roundtrip_multiple_items_with_common_prefix	Prefix-encoding path unaffected
test_roundtrip_items_without_common_prefix	Plain separator path unaffected

All 6 pass. Zero regressions.

AI Disclosure

• I used AI tools (Claude) to help identify the bug and draft
  the test file. I have reviewed every line, understand the root
  cause, and can explain the fix and failure modes independently.

[greptile-apps]

Greptile Summary

This PR fixes a one-line IndexError in decompress_list when called with an empty string — the output produced by compress_list([]) — by adding a two-line early-return guard before the lststr[0] index access. The fix is minimal, correct, and has no effect on any non-empty input path.

Key changes:

• metaflow/util.py: if not lststr: return [] added at the top of decompress_list, resolving the crash on the empty-list round-trip.
• test/unit/test_util_compress.py: 6 new unit tests covering the regression case and the three encoding paths (plain, prefix-encoded, zlib).

Notable design consideration:
Both compress_list([]) and compress_list([""]) produce "" (a pre-existing collision in the encoding scheme). The fix resolves this ambiguity in favour of [], but no test documents this behaviour, which could surprise callers who pass lists containing empty strings.

Confidence Score: 5/5

• This PR is safe to merge; the change is a minimal guard that fixes a crash with no impact on existing non-empty code paths.
• The fix is two lines, targets only the empty-string edge case, and is fully covered by new regression tests. No existing behaviour is altered for non-empty inputs.
• No files require special attention beyond the minor test-coverage gap noted for the [""] edge case.

Important Files Changed

Filename	Overview
metaflow/util.py	Adds a 2-line early-return guard (if not lststr: return []) to decompress_list to prevent the IndexError raised when an empty string is passed; the fix is minimal, correct, and does not affect non-empty inputs.
test/unit/test_util_compress.py	New test file with 6 unit tests covering the empty-list round-trip regression and common non-empty paths; missing a test that documents the known [""]/[] collision in the compress/decompress encoding scheme.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["decompress_list(lststr)"] --> B{"lststr is empty?"}
    B -->|"yes — NEW guard"| C["return []"]
    B -->|"no"| D{"first char == zlibmarker?"}
    D -->|"yes"| E["base64-decode\nthen zlib decompress twice\n→ decoded string"]
    D -->|"no"| F["decoded = lststr"]
    E --> G{"rangedelim in decoded?"}
    F --> G
    G -->|"yes"| H["split on rangedelim\nreturn prefix+suffix for each suffix"]
    G -->|"no"| I["return decoded.split(separator)"]

Loading

_{Last reviewed commit: 4bea37a}

[greptile-apps]

Missing edge-case test: [""] round-trip collision

compress_list([""]) produces "" (joining a list of one empty string), which is the same output as compress_list([]). After this fix, decompress_list("") returns [], meaning decompress_list(compress_list([""])) returns [] instead of [""].

This is a pre-existing design-level ambiguity (both an empty list and a list containing one empty string compress to the same token), but the fix silently resolves it in favor of []. Adding a test documenting the current behavior would make this intent explicit and prevent future surprises:

def test_roundtrip_single_empty_string():
    # compress_list([""]) == "" == compress_list([]),
    # so decompress cannot distinguish the two; [] is the chosen result.
    assert decompress_list(compress_list([""])) == []

If [""] inputs are genuinely expected in the codebase, the collision should be addressed at the compress_list level (e.g., encoding the count) rather than relying on callers to avoid empty-string elements.