Problem Dataset Structure

Problems are sourced from an existing Kaggle dataset (imported via kagglehub), not scraped — this avoids the copyright/ToS concerns of pulling from Codeforces directly.

Dataset Columns

Column	Type	Notes
id	integer	Primary key, maps directly.
title	text	Direct column.
description	text	Direct column; full problem statement.
difficulty_level	text (Easy/Medium/Hard)	Maps to the difficulty tiers used in scoring.
created_at / updated_at	timestamp	Carried over as-is.
examples	JSON array	Sample input/output pairs shown to the student on the problem page.
constraints	JSON array of strings	Displayed as-is under the problem statement.
test_cases	JSON array of {input, expected_output}	Used by the judge harness; inputs are structured objects, not raw stdin text.

Function-Call vs. stdin/stdout

The dataset's test cases are LeetCode-style: input is a structured JSON object (e.g. a tree, a nested config, an array with named parameters), and expected_output is a value to compare — not a stdin/stdout text stream.

Standard Piston usage assumes a program reading stdin and writing stdout. This mismatch is resolved by an adapter/harness layer (see the Real-Time Contests / Execution section) — it's called out here because it affects the required function signature students must write.

Confirmed test_cases.input shapes

All test_cases.input values are named-key JSON objects matching function parameter names — never positional arrays or raw stdin:

• Scalar: {"n": 4}
• 1D array + scalar: {"nums": [...], "target": 9}
• 2D array: {"cost": [[...]]}
• Tree: {"root": {"val": 1, ...}, "target": 2} (root can be null)