DTL Viewer

DTL Viewer parses decision tree logic definitions from structured sources (JSON, YAML, or inline definitions) and renders a readable data model for analysis without executing domain code. It performs structural validation on nodes, edges, and conditional expressions, and flags syntax anomalies before deeper analysis. The tool emphasizes reproducibility by maintaining deterministic parses of definitions, enabling auditors to compare versions and verify that changes propagate as intended. Core Logic & Features: The tool builds an in-memory graph from JSON, YAML, or inline definitions; it validates node types, edges, conditions, and default paths; it detects cycles and unreachable segments; it provides an explorable view of evaluation paths and supports filtering by node type, criteria, or confidence score. It offers export options to JSON and to lightweight reports suitable for integration with documentation or CI pipelines. Inputs & Outputs: Inputs include DTL definitions in supported formats, optional version markers, and sample input payloads for path tracing. Outputs are a normalized graph model, comprehensive validation reports, path traces, and lineage data that can be serialized as JSON, CSV, or human-readable summaries. Algorithms & Calculations: Parsing uses schema-aware deserialization, graph construction with cycle detection and reachability analysis, and path evaluation via depth-first traversal with memoization and memoized path caches. Error & Edge Cases: It reports syntax errors, missing nodes, ambiguous or conflicting rules, cycles, and unreachable paths; handles large trees via chunked processing, and yields deterministic results for reproducibility; returns precise error codes for automation. Industry/Region & Localization: Aligns with data engineering standards, supports region-specific terminology, and can incorporate locale-aware naming and currency or unit hints when path justification requires it. Assumptions & Exclusions: Assumes DTL definitions contain explicit node identifiers and conditions; excludes runtime execution of domain logic, UI rendering details, performance benchmarks, and code generation.