GFQL/Cypher: ORDER BY on stringified-list columns uses Cypher list-orderability (#1359) - #1362

Summary

Closes pygraphistry-side semantic gap from #1359 (item #1 of meta-issue #1353 — Comparator + ORDER BY semantics tranche).

When a list-valued property is stored as a string column (e.g. round-tripped through CSV / Arrow string columns), ORDER BY <col> previously fell back to lex string sort, which mishandles negative numbers because '-' (0x2D) sorts before '2' (0x32) — so '[1, -20]' came before '[1, 2]', putting the wrong row in LIMIT N results.

What changed

graphistry/compute/gfql/row/ordering.py

• New helper order_detect_stringified_list_series(series) — returns True when every non-null entry is an actual string matching ^\[.*\]$. Complements order_detect_list_series (which only matches Python-list values).
• New helper parse_stringified_list_series(series) — parses the column via ast.literal_eval, returning a Python-list Series (or None on parse failure).

graphistry/compute/gfql/row/pipeline.py

• order_by: after the existing list_candidate check, if the column is a stringified-list series with no nulls, parse into a fresh __gfql_sort_listparsed_*__ aux column and re-enter the existing list-sort path (build_list_sort_columns).
• Track aux columns in aux_drop_cols and drop them after sort_values so they don't leak into downstream stages (regression-tested via existing test_string_cypher_supports_labels_projection_* and test_string_cypher_supports_graph_functions_on_list_wrapped_entities).

Python-list-typed columns continue through the existing path unchanged.

Tests

graphistry/tests/compute/gfql/cypher/test_lowering.py

• test_string_cypher_order_by_python_list_column_uses_list_orderability — pins ASC/DESC top-3 on a Python-list column.
• test_string_cypher_order_by_stringified_list_column_uses_list_orderability — same query, dtype="string" list column; pre-fix produced E ([2,-2,100]) in top-3 instead of A ([2,-2]); post-fix the SET matches openCypher.

Test plan

• ./bin/lint.sh clean
• ./bin/mypy.sh graphistry/compute/gfql/row/ordering.py graphistry/compute/gfql/row/pipeline.py clean
• pytest graphistry/tests/compute/gfql/cypher/ — 1224 passed, 90 skipped, 15 xfailed (no regressions)
• Direct cypher xfail-outcome verification for the 14 with-orderby keys: Cluster A with-orderby1-32-{1,2} (list ORDER BY DESC) flip to success_matches_expected; remaining 12 are pygraphistry-SET-correct, blocked at runner level
• Once tck-gfql#36 lands the matching fixture/runner fixes, all 14 keys flip to success_matches_expected

Out of scope (handled in tck-gfql#36)

• tests/cypher_tck/parse_cypher.py::_parse_properties doesn't parse list literals (lists become strings in fixtures).
• tests/cypher_tck/test_tck_runner.py::_rows_ordered enforces strict order even when the openCypher feature uses Then the result should be, in any order:.

Cross-reference

• Closes pygraphistry-side semantic debt for GFQL/Cypher: drive down WITH ORDER BY wrong-row tranche (14 cases, #1353 item #1) #1359 (meta Meta: Prioritized direct-Cypher wrong-row bugfix tranches (92 cases) #1353 item Replace NaNs with nulls since node cannot parse JSON with NaNs #1)
• Pairs with tck-gfql#36 to fully close all 14 with-orderby success_wrong_rows cases

🤖 Generated with Claude Code