Conversion Pipeline Overview

console2svg is far more than a basic screenshot utility that wraps terminal text in an image container. It launches child processes through OS-level pseudo-terminals (PTYs), processes raw output byte streams through an in-memory terminal emulator to maintain an accurate virtual display, and converts the resulting buffer into highly optimized SVG vectors, raster images, video streams, and real-time web broadcasts.

To achieve maximum visual fidelity alongside aggressive deduplication, console2svg decomposes the transformation into clearly demarcated pipeline stages:

[Child Process Launch / Replay Input]
         │ Raw VT byte stream
         ▼
[Stage 1: Input Capture and Stream Control]
         │ Coalesced event stream
         ▼
[Stage 2: Terminal Emulation and Sequence Parsing]
         │ Virtual Screen Buffer (ScreenBuffer)
         ▼
[Stage 3: Secret Detection and Redaction]
         │ Masked cell grid
         ▼
[Stage 4: SVG Layer Structuring and Optimization]
         │ Optimized vector markup
         ├─► [Stage 5A: Still Image / Animated SVG]
         ├─► [Stage 5B: Fast Rasterization / Video Encoding]
         └─► [Stage 5C: Real-Time Web Broadcast]

Pipeline Stages

The transformation process spans five foundational stages:

Stage 1: Input Capture and Stream Control

Accurate terminal reproduction begins by capturing raw child process output without byte corruption or timing jitter.

PTY Launch, Control, and Stream Capture: Bridges Windows ConPTY and Unix PTY implementations, guarantees multibyte UTF-8 boundary integrity, and coalesces closely spaced output events to minimize parser overhead.
Deterministic Input Replay: Normalizes recorded keystroke timestamps and synthesizes authentic VT escape byte sequences to recreate terminal interactions deterministically.

Stage 2: Terminal Emulation and Sequence Parsing

Raw terminal streams contain ANSI control codes for cursor repositioning, buffer clearing, and SGR attribute styling rather than flat text.

Escape Sequence Parsing and Buffer Updates: Employs a stateful parsing automaton to decode CSI, OSC, and SGR sequences, handling full-width character column widths, margins, and cursor state within a two-dimensional ScreenBuffer.
Special Characters, Box-Drawing, and Blocks: Replaces font-dependent box-drawing characters and block elements with crisp vector geometry paths, smoothing junction seams and rounded corners.

Stage 3: Secret Detection and Redaction

When capturing terminal sessions for documentation or public video, sensitive credentials must never leak into visual output.

QuickLeaks Secret Detection and Redaction: Accelerates regex scanning using precompiled literal anchors and maps detected character bounds across two coordinate planes to excise confidential tokens directly from SVG text nodes.

Stage 4: SVG Layer Structuring and Optimization

Converting virtual screen cells into web-standard SVG requires structural optimization to prevent massive document inflation.

SVG Layer Hierarchy and Coordinate Systems: Separates outer margins, window chrome, drop shadows, background fills, and foreground text into distinct layers under a unified character grid coordinate system.
Pipeline Optimizations: Leverages copy-on-write buffer snapshotting, XxHash3 row signatures, and horizontal span merging for rects and paths to achieve extreme file size reductions.

Stage 5: Target Output Generation

The structured, optimized screen representation is finally dispatched into specific export formats according to CLI commands:

SMIL Animated SVG Generation: Indexes distinct rows into a <defs> catalog and applies discrete SMIL <animate> visibility switching alongside partial row deltas to produce compact animated SVGs.
PNG Rasterization via resvg: Embeds the Rust-based resvg rendering engine in-process via C ABI, caching OS font scans across runs to render PNG frames in milliseconds.
Video Encoding via ffmpeg Pipeline: Powers deterministic video export through unidirectional emulation advances, a two-stage memory cache, multithreaded rasterization, and direct image2pipe streaming to ffmpeg.
live-server Streaming Architecture: Runs an embedded HTTP/SSE server featuring three-layer streaming, update settle delays, and client-side rAF coalescing for real-time terminal mirroring in web browsers.

For architectural details, algorithms, and design choices, explore the linked articles above.