Data Generator Tool

Purpose & User Intent: The tool generates structured synthetic datasets to enable testing, development, and QA workflows without exposing real user data. Primary users are software developers, data engineers, data scientists, and QA engineers who need repeatable samples that conform to a defined schema. It is used when validating data pipelines, benchmarking processing performance, or seeding test environments with realistic but non-production data. Assumptions include a provided target schema and a desired dataset size.\n\nCore Logic & Features: The tool accepts a schema (field name, data type, constraints) and a target size, then produces a dataset that respects type rules and distribution controls. Required features: schema parsing, field-level constraints, reproducible seeding, and export formats. Optional/advanced features: per-field custom generators, cross-field dependencies, privacy options (hashing, k-anonymity presets), distribution profiles (uniform, normal, categorical), and multi-format exports (CSV, JSON, SQL). It supports common data types and allows locale-aware formatting.\n\nInputs & Outputs: Inputs include schema (structured object), size (integer), distribution (predefined presets), seed (integer, optional), and output_format (CSV/JSON/SQL). Outputs are a dataset file and a metadata summary (field types, counts, distribution used). Sample rows can be emitted for quick validation.\n\nAlgorithms & Calculations: Generation uses seeded pseudorandom number generation to ensure reproducibility. For each field type, it applies: integers with min/max, floats with precision, strings via alphabets and length, dates within a range, booleans by probability, and enumerations by weighted probabilities. Privacy-preserving steps include deterministic hashing for identifiers when enabled and optional anonymization modes. Dependencies between fields are expressed through dependency rules if provided.\n\nError & Edge Cases: Invalid or missing schema, unsupported types, zero or negative size, conflicting constraints, or unavailable external templates are handled by clear data-model validation results. Large sizes trigger streaming or chunked writes to limit memory usage. Incompatible distributions or invalid seeds produce deterministic defaults.\n\nIndustry/Region & Localization: Supports locale-specific formats (date and number formats), time zones, and currency if included in data fields. Complies with privacy guidelines where applicable and allows regional constraints to be defined per schema.\n\nAssumptions & Exclusions: Assumes synthetic datasets only; excludes live production data, direct data derivation from real users, or guarantees of regulatory compliance. External data sources are optional; binary fields are not generated unless specified.