LLVM IR Viewer

Purpose & User Intent: The LLVM IR Viewer solves the exact problem of revealing the structure and relationships inside LLVM intermediate representation code. Target users include compiler engineers, optimization researchers, and advanced educators who inspect IR snippets to validate passes, detect anomalies, and compare transformations. The tool accepts a block of LLVM IR text and returns a machine-readable graph alongside concise summaries that aid review and debugging.

Core Logic & Features: Domain capabilities include IR parsing, function and basic-block extraction, and construction of a control-flow graph (CFG) and data-flow graph (DFG). Required features are lexical and syntactic parsing, block linking via terminators, and per-instruction metadata mapping. Optional/advanced features include alias hints, inlining impact indicators, and cross-function data flow traces. Outputs include JSON schemas for CFG/DFG and a compact human-readable summary for quick inspection.

Inputs & Outputs: Inputs are LLVM IR text (required) and optional scope controls (e.g., function name, module). Outputs are structured data: CFG edges, basic blocks, instruction lists, and a compact JSON summary; plus a readable description. Outputs are designed for downstream tooling and editors needing quick visualization cues.

Algorithms & Calculations: The tool tokenizes IR, identifies function boundaries, assembles blocks, and links successors via terminator instructions. It computes graph structures for CFG and DFG, and derives metrics such as block counts, edge counts, and potential optimization hotspots. Conventions follow LLVM's textual syntax; no execution or dynamic analysis is performed.

Error & Edge Cases: If the IR is syntactically invalid, the tool reports precise error positions and partial graphs when possible. Missing functions or malformed blocks may produce incomplete graphs with warnings. Large modules may be processed in chunks; streaming outputs may be enabled in future releases.

Industry/Region & Localization: Follows LLVM project conventions; unit and terminology align with compiler engineering standards. Localization applies only to textual labels in outputs; numeric representations use LLVM's default formatting.

Assumptions & Exclusions: Assumes input is textual LLVM IR conforming to the standard. Excludes disassembly beyond IR, does not execute code, and does not emulate runtime behavior. It is not a decompiler or optimizer, but a visualization and inspection aid.