/opentui

Code & Diff Components

Components for source code, line-number gutters, unified diffs, Markdown, and text tables.

Code Component

CodeRenderable displays plain text immediately and applies asynchronous Tree-sitter highlighting when filetype and a parser are available.

Basic Usage

// React / Solid
<code content={sourceCode} filetype="typescript" syntaxStyle={syntaxStyle} />

// Core
const code = new CodeRenderable(renderer, {
  content: sourceCode,
  filetype: "typescript",
  syntaxStyle,
  wrapMode: "none", // "none" | "char" | "word"
})

OpenTUI bundles parsers for JavaScript/JSX, TypeScript/TSX, Markdown, Markdown-inline, and Zig. Other grammars require Tree-sitter asset configuration. Without filetype, Code renders unhighlighted text.

Highlight Hooks

onHighlight can replace the syntax ranges before styling. It receives SimpleHighlight[] tuples and { content, filetype, syntaxStyle }; return an array or undefined, synchronously or asynchronously.

<code
  content={sourceCode}
  filetype="typescript"
  syntaxStyle={syntaxStyle}
  onHighlight={(highlights, context) =>
    highlights.filter((highlight) => highlight[2] !== "comment")
  }
/>

onChunks runs afterward and can replace the resolved TextChunk[]. Its context also includes highlights.

import { detectLinks } from "@opentui/core"

<code
  content={markdown}
  filetype="markdown"
  syntaxStyle={syntaxStyle}
  onChunks={(chunks, context) => detectLinks(chunks, context)}
/>

detectLinks applies links for recognized Markdown/URL highlight scopes.

TextTable Component

TextTableRenderable is Core-only. It displays styled chunk cells with borders, wrapping, width fitting, and selection.

import {
  TextTableRenderable,
  bold,
  fg,
  type TextChunk,
  type TextTableContent,
} from "@opentui/core"

const cell = (text: string): TextChunk[] => [{ __isChunk: true, text }]
const content: TextTableContent = [
  [[bold("Service")], [bold("Status")], [bold("Notes")]],
  [cell("api"), [fg("#00d4aa")("OK")], cell("latency 28ms")],
]

const table = new TextTableRenderable(renderer, {
  content,
  wrapMode: "word",
  columnWidthMode: "full",   // "content" | "full"
  columnFitter: "balanced",  // "proportional" | "balanced"
  cellPadding: 1,
  border: true,
  outerBorder: true,
  borderStyle: "rounded",
  selectable: true,
})

TextTableCellContent is TextChunk[] | null | undefined. Each literal chunk needs __isChunk: true; styled-text helpers such as bold() and fg() already return valid chunks. getSelectedText() and hasSelection() expose selection; vertical drags within one column retain columnar selection.

Line Number Component

LineNumberRenderable is a gutter for another renderable that implements LineInfoProvider; it does not accept source code itself.

// React (use <line_number> in Solid)
<line-number
  ref={lineNumbersRef}
  fg="#6b7280"
  bg="#161b22"
  minWidth={3}
  paddingRight={1}
  lineNumberOffset={0}
>
  <code content={sourceCode} filetype="typescript" syntaxStyle={syntaxStyle} />
</line-number>

// Core
const code = new CodeRenderable(renderer, {
  content: sourceCode,
  filetype: "typescript",
  syntaxStyle,
})
const lineNumbers = new LineNumberRenderable(renderer, {
  target: code,
  minWidth: 3,
  paddingRight: 1,
})

Use methods rather than nonexistent diagnostics, addedLines, removedLines, or highlightedLines props:

lineNumbers.setLineColor(4, "#1a4d1a")
lineNumbers.setLineSign(4, { after: " +", afterColor: "#22c55e" })
lineNumbers.highlightLines(9, 11, "#4d1a1a")
lineNumbers.clearHighlightLines(9, 11)

Other methods include clearLineColor(), setLineColors(), clearAllLineColors(), clearLineSign(), setLineSigns(), and clearAllLineSigns(). lineNumberOffset changes displayed numbering.

Diff Component

DiffRenderable accepts a unified diff string. It does not compute a diff from old and new source strings.

// React / Solid
<diff
  diff={unifiedPatch}
  filetype="typescript"
  syntaxStyle={syntaxStyle}
  view="split"
  syncScroll
  showLineNumbers
/>

// Core
const diff = new DiffRenderable(renderer, {
  diff: unifiedPatch,
  filetype: "typescript",
  syntaxStyle,
  view: "unified",
})

Options

Use view, not mode; use addedBg/removedBg/contextBg, not addedLineColor/removedLineColor/unchangedLineColor. Context lines are already encoded in the patch, so there is no context option. For multi-file input, Diff currently displays only the first parsed file patch.

Programmatic Line Highlighting

diff.setLineColor(10, "#FFFF0030")
diff.clearLineColor(10)
diff.setLineColors(new Map([
  [5, "#FF000030"],
  [10, { bg: "#00FF0030", fg: "#FFFFFF" }],
]))
diff.highlightLines(20, 25, "#0000FF30")
diff.clearHighlightLines(20, 25)
diff.clearAllLineColors()

getHunkRowOffsets() returns display-row offsets for parsed hunks. Re-read it after changing the patch, view, wrapping, or dimensions.

Markdown Component

MarkdownRenderable parses Markdown into styled renderables. Pass a SyntaxStyle for fenced code highlighting.

<markdown
  content={markdownText}
  syntaxStyle={syntaxStyle}
  conceal
  concealCode={false}
  streaming={false}
  internalBlockMode="coalesced"
/>

Custom Node Rendering

renderNode receives (token, context). Return a custom renderable, context.defaultRender() for the built-in representation, or null/ undefined as appropriate.

const markdown = new MarkdownRenderable(renderer, {
  content: "# Custom Heading",
  syntaxStyle,
  renderNode(token, context) {
    if (token.type === "heading") {
      return new TextRenderable(renderer, { content: `>> ${token.text} <<` })
    }
    return context.defaultRender()
  },
})

For fenced-code specialization, use createMarkdownCodeBlockRenderer() to dispatch normalized filetypes to custom renderers while retaining the default renderer for unmatched tokens.

Streaming Markdown

<markdown
  content={streamedContent}
  syntaxStyle={syntaxStyle}
  streaming={isStreaming}
  internalBlockMode="top-level"
/>

Keep streaming true while appending and set it false when complete so the final parse can settle. top-level mode exposes stable top-level blocks, which is useful for LLM output and incremental views.

Gotchas

• React uses <line-number>; Solid uses <line_number>.
• Code uses content and filetype, not code and language.
• Diff uses one unified diff string and view, not old/new strings and mode.
• Line Number wraps a target; it does not render source code by itself.
• Tree-sitter loading is asynchronous. Use OTUI_TREE_SITTER_WORKER_PATH when packaging requires a custom worker path.
• Put large Code/Line Number views inside a height-constrained ScrollBox.

See Also

• Text & Display - Styled text and image rendering
• Containers - ScrollBox for large content
• Core API - Imperative renderables
• Testing - Frame and snapshot tests