task_charts__table__column_rank_label

Public Contract

1. Domain: charts
2. Scene: table
3. Source file: src/trace_tasks/tasks/charts/table/column_rank_label.py
4. Prompt assets: src/trace_tasks/resources/prompts/charts/table/charts_table_counting_v1.json, src/trace_tasks/resources/prompts/charts/table/charts_table_ranking_v1.json, src/trace_tasks/resources/prompts/charts/table/charts_table_statistics_v1.json, and src/trace_tasks/resources/prompts/charts/table/charts_table_temporal_v1.json
5. Query ids: highest_rank_in_column, lowest_rank_in_column

Program Contract

Program: rank(rows by value(column), direction, rank_k); output=row_label; annotation=bbox(answer_value_cell); scene=table; scope=column_rank_label

Candidate set: the visible table cells with row and column labels inside the column_rank_label objective scope. Operands: prompt-bound labels, categories, series names, thresholds, intervals, references, and encoded chart values, plus the active query id's comparator, direction, target role, or extremum focus when present. Operation: evaluate rank over the candidate set using the filters, comparisons, aggregations, rankings, projections, or counterfactual edits named in the program expression; generation enforces a unique final answer. Output binding: answer is the unspecified value bound by row_label. Annotation witnesses: unspecified witnesses bound by bbox(answer_value_cell). The Annotation Contract below defines the prompt-facing witnesses. Query ids: highest_rank_in_column, lowest_rank_in_column.

Reasoning Operations

Families: ranking

Contract Notes

This task uses the current source layout. Scene-local reusable code lives under src/trace_tasks/tasks/charts/table/shared/; public task files own objective logic, query selection, answer binding, and annotation binding.