LLMStream

https://github.com/synth-inc/llmstream
Admin
LLMStream is an iOS/macOS Swift package for displaying customizable, streamed responses from Large...
Tokens:5,802
Snippets:28
Trust Score:7.4
Update:1 year ago
Show doc for...
Tokens:
Raw
### SwiftUI Usage of LLMStreamView

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Demonstrates how to integrate the LLMStreamView into a SwiftUI view. This example shows initializing the view with sample Markdown and LaTeX content.

```swift
import LLMStream

struct ContentView: View {
    @State var height: CGFloat = 0

    var body: some View {
        LLMStreamView(
            text: "# Hello **Markdown** & $\\LaTeX$"
        )
    }
}
```

--------------------------------

### Markdown: Inline and Display Math with LaTeX

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Markdown example shows how to integrate LaTeX for mathematical expressions. It includes an inline equation using '$' delimiters and a display math equation using '\[\]' delimiters, demonstrating rendering capabilities for scientific content.

```markdown
# Integration Example

This is a **bold** text with an inline equation: $E = mc^2$

## Display Math

\[
\int_{-\infty}^{\infty} e^{-x^2} dx = \sqrt{\pi}
\]
```

--------------------------------

### Swift Package Manager Installation for LLMStream

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Instructions for adding the LLMStream Swift package to your project using Swift Package Manager. This includes updating your Package.swift file and adding necessary security entitlements.

```swift
dependencies: [
    .package(url: "https://github.com/synth-inc/LLMStream", branch: "main")
]
```

```xml
<key>com.apple.security.network.client</key>
<true/>
<key>com.apple.security.web</key>
<true/>
```

--------------------------------

### Install LLMStream using Swift Package Manager in Swift

Source: https://context7.com/synth-inc/llmstream/llms.txt

Instructions for adding the LLMStream library to a Swift project using Swift Package Manager. This involves specifying the repository URL and branch in the Package.swift file.

```swift
// Package.swift
// swift-tools-version: 5.7
import PackageDescription

let package = Package(
    name: "MyApp",
    platforms: [
        .macOS(.v13),
        .iOS(.v16)
    ],
    dependencies: [
        .package(url: "https://github.com/synth-inc/LLMStream", branch: "main")
    ],
    targets: [
        .target(
            name: "MyApp",
            dependencies: ["LLMStream"])
    ]
)
```

--------------------------------

### Swift: Apply Custom Configurations to LLMStreamView

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Swift code demonstrates how to create a comprehensive `LLMStreamConfiguration` by combining various sub-configurations (font, colors, layout, thought, animation, code block, table) and applying it to an LLMStreamView. This allows for a highly customized user experience.

```swift
let config = LLMStreamConfiguration(
    font: fontConfig,
    colors: colorConfig,
    layout: layoutConfig,
    thought: thoughtConfig,
    animation: animationConfig,
    codeBlock: codeConfig,
    table: tableConfig
)

LLMStreamView(
    text: "Your content here",
    configuration: config
)
```

--------------------------------

### Display Streamed LLM Responses with LLMStreamView (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

The main SwiftUI view component for rendering streamed LLM responses. It supports Markdown, LaTeX, and code highlighting. It takes the streamed text, a configuration object, and callbacks for URL clicks and code actions as input.

```swift
import SwiftUI
import LLMStream

struct ContentView: View {
    @State private var streamedText = """\n    # Mathematical Formula\n
    The quadratic formula is: $x = \\frac{-b \\pm \\sqrt{b^2-4ac}}{2a}$\n
    ## Code Example\n
    ```swift
    func fibonacci(_ n: Int) -> Int {
        guard n > 1 else { return n }
        return fibonacci(n-1) + fibonacci(n-2)
    }
    ```

    <think>
    Let me analyze this problem step by step...
    </think>

    The solution involves **recursive** calls with `O(2^n)` complexity.
    """

    var body: some View {
        ScrollView {
            LLMStreamView(
                text: streamedText,
                configuration: .default,
                onUrlClicked: { url in
                    print("URL clicked: \(url)")
                },
                onCodeAction: { code in
                    print("Execute code: \(code)")
                }
            )
            .padding()
        }
    }
}
```

--------------------------------

### Configure Table Rendering with TableConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Configures how tables are rendered in LLMStream, including options for showing captions, styling headers, enabling hover effects, and zebra striping. It allows for custom caption styles and header styles. This configuration is passed to LLMStreamConfiguration.

```swift
import SwiftUI
import LLMStream

let tableConfig = TableConfiguration(
    showCaption: true,
    captionStyle: TableConfiguration.CaptionStyle(
        fontSize: 0.9,
        textColor: Color(white: 0.8)
    ),
    headerStyle: TableConfiguration.HeaderStyle(
        fontWeight: .bold,
        textAlignment: .center,
        borderWidth: 2.0
    ),
    enableHover: true,
    enableZebraStripes: true
)

let config = LLMStreamConfiguration(table: tableConfig)

let markdownTable = """| Feature | Supported |
|---------|-----------|
| Markdown | ✓ |
| LaTeX | ✓ |
| Code Highlighting | ✓ |
"""

LLMStreamView(text: markdownTable, configuration: config, onUrlClicked: { _ in })
```

--------------------------------

### Customize Code Blocks with CodeBlockConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Customizes the appearance and behavior of code blocks, including options to show language labels, copy buttons, and action buttons for code execution. It also allows customization of text sizes, button sizes, opacity, and icons. This configuration is passed to LLMStreamConfiguration.

```swift
import SwiftUI
import LLMStream

let codeConfig = CodeBlockConfiguration(
    showLanguage: true,
    showCopyButton: true,
    showActionButton: true,
    languageTextSize: 13.0,
    copyButtonSize: 16,
    actionButtonSize: 16,
    copyButtonOpacity: 0.5,
    copyButtonHoverOpacity: 1.0,
    copyButtonIcon: Image(systemName: "doc.on.doc"),
    actionButtonOpacity: 0.5,
    actionButtonHoverOpacity: 1.0,
    actionButtonIcon: Image(systemName: "play.circle"),
    actionButtonTooltip: "Execute Code"
)

LLMStreamView(
    text: """
    ```python
    def hello():
        print("Hello, World!")
    ```
    """,
    configuration: LLMStreamConfiguration(codeBlock: codeConfig),
    onUrlClicked: { _ in },
    onCodeAction: { code in
        print("Executing: \(code)")
        // Add your code execution logic here
    }
)
```

--------------------------------

### LLMStream Color Configuration in Swift

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Provides a detailed configuration for colors used within LLMStream, covering text, backgrounds, links, and specific elements like code blocks, tables, and thought states.

```swift
let colorConfig = ColorConfiguration(
    textColor: .white,
    backgroundColor: .clear,
    codeBackgroundColor: Color(red: 0.15, green: 0.15, blue: 0.15),
    codeBorderColor: Color(white: 0.24),
    linkColor: Color(red: 0.29, green: 0.60, blue: 1.0),
    thoughtBackgroundColor: Color.gray.opacity(0.1),
    tableHeaderBackgroundColor: Color(white: 0.24),
    tableBorderColor: Color(white: 0.4),
    tableRowEvenColor: Color(white: 0.2).opacity(0.2),
    tableRowHoverColor: Color(white: 0.27).opacity(0.3),
    theoremBorderColor: Color(red: 0.29, green: 0.60, blue: 1.0),
    proofBorderColor: Color(white: 0.47)
)
```

--------------------------------

### SwiftUI LLMStream View with Streaming Response

Source: https://context7.com/synth-inc/llmstream/llms.txt

This snippet demonstrates how to use the `LLMStreamView` in SwiftUI to display streaming LLM responses. It configures the appearance, handles URL clicks, and executes code actions. The `simulateStreaming` function mimics a real-time response by appending characters over time.

```swift
import SwiftUI
import LLMStream

struct AIAssistantView: View {
    @State private var streamedResponse = ""
    @State private var isStreaming = false

    let configuration = LLMStreamConfiguration(
        font: FontConfiguration(size: 15.0, lineHeight: 1.5),
        colors: ColorConfiguration(
            textColor: .white,
            backgroundColor: Color(red: 0.1, green: 0.1, blue: 0.12),
            codeBackgroundColor: Color(red: 0.15, green: 0.15, blue: 0.16)
        ),
        thought: ThoughtConfiguration(
            icon: Image(systemName: "brain"),
            thinkingTitle: "Analyzing...",
            thoughtTitle: "Reasoning"
        ),
        codeBlock: CodeBlockConfiguration(
            showActionButton: true,
            actionButtonTooltip: "Run this code"
        )
    )

    var body: some View {
        VStack(spacing: 0) {
            ScrollView {
                LLMStreamView(
                    text: streamedResponse,
                    configuration: configuration,
                    onUrlClicked: { url in
                        if let urlObj = URL(string: url) {
                            NSWorkspace.shared.open(urlObj)
                        }
                    },
                    onCodeAction: { code in
                        executeCode(code)
                    }
                )
                .padding()
            }

            HStack {
                Button("Stream Response") {
                    simulateStreaming()
                }
                .disabled(isStreaming)
            }
            .padding()
        }
    }

    func simulateStreaming() {
        isStreaming = true
        streamedResponse = ""

        let fullResponse = """\
        # Algorithm Analysis

        <think>
        Need to evaluate time complexity...
        Considering space-time tradeoffs...
        </think>

        The optimal approach uses **dynamic programming**:

        ```python
        def solve(n, memo={}):
            if n in memo:
                return memo[n]
            if n <= 1:
                return n
            memo[n] = solve(n-1) + solve(n-2)
            return memo[n]
        ```

        Time complexity: $O(n)$

        Space complexity: $O(n)$
        """

        var index = 0
        Timer.scheduledTimer(withTimeInterval: 0.05, repeats: true) { timer in
            if index < fullResponse.count {
                let currentIndex = fullResponse.index(fullResponse.startIndex, offsetBy: index)
                streamedResponse.append(fullResponse[currentIndex])
                index += 1
            } else {
                timer.invalidate()
                isStreaming = false
            }
        }
    }

    func executeCode(_ code: String) {
        print("Executing code:")
        print(code)
        // Implement your code execution logic
    }
}

```

--------------------------------

### Customize Thought Bubbles with ThoughtConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Customizes the appearance and behavior of LLM thinking states, typically used with `<think>` tags. It allows setting icons, titles, and controls for expanding content. This configuration is passed to LLMStreamConfiguration.

```swift
import SwiftUI
import LLMStream

let thoughtConfig = ThoughtConfiguration(
    icon: Image(systemName: "brain"),
    iconSize: 16,
    thinkingTitle: "Thinking...",
    thoughtTitle: "Thought-process",
    showExpandButton: true
)

let config = LLMStreamConfiguration(thought: thoughtConfig)

let textWithThinking = """# Problem Analysis

<think>
First, I need to understand the problem constraints...
Then calculate the time complexity...
Finally, optimize the algorithm...
</think>

The optimal solution uses dynamic programming with O(n) complexity.
"""

LLMStreamView(text: textWithThinking, configuration: config, onUrlClicked: { _ in })
```

--------------------------------

### Configure UI Layout with LayoutConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Controls spacing, padding, and corner radius for content areas, code blocks, tables, and thought bubbles in LLMStream. It accepts EdgeInsets for padding and CGFloat for spacing and corner radii. This configuration is passed to LLMStreamConfiguration.

```swift
import SwiftUI
import LLMStream

let layoutConfig = LayoutConfiguration(
    contentPadding: EdgeInsets(top: 0, leading: 0, bottom: 0, trailing: 0),
    codePadding: EdgeInsets(top: 12, leading: 16, bottom: 12, trailing: 16),
    thoughtPadding: EdgeInsets(top: 8, leading: 8, bottom: 8, trailing: 8),
    tablePadding: EdgeInsets(top: 8, leading: 8, bottom: 8, trailing: 8),
    citationMargin: EdgeInsets(top: 0, leading: 4, bottom: 0, trailing: 4),
    citationPadding: EdgeInsets(top: 2, leading: 4, bottom: 2, trailing: 4),
    spacing: 8,
    cornerRadius: 8,
    tableCornerRadius: 5,
    theoremCornerRadius: 4,
    citationBorderRadius: 6
)

let config = LLMStreamConfiguration(layout: layoutConfig)
```

--------------------------------

### Configure UI Colors with ColorConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Defines the color scheme for all UI elements within LLMStream, including text, backgrounds, code blocks, tables, and interactive elements. It takes various Color objects as parameters to customize different UI parts. This configuration is passed to LLMStreamConfiguration.

```swift
import SwiftUI
import LLMStream

let colorConfig = ColorConfiguration(
    textColor: Color.white,
    backgroundColor: Color.clear,
    codeBackgroundColor: Color(red: 0.15, green: 0.15, blue: 0.15),
    codeBorderColor: Color(white: 0.24),
    linkColor: Color(red: 0.29, green: 0.60, blue: 1.0),
    thoughtBackgroundColor: Color.gray.opacity(0.1),
    tableHeaderBackgroundColor: Color(white: 0.24),
    tableBorderColor: Color(white: 0.4),
    tableRowEvenColor: Color(white: 0.2).opacity(0.2),
    tableRowHoverColor: Color(white: 0.27).opacity(0.3),
    theoremBorderColor: Color(red: 0.29, green: 0.60, blue: 1.0),
    proofBorderColor: Color(white: 0.47),
    citationBackgroundColor: Color(red: 0.15, green: 0.15, blue: 0.15),
    citationHoverBackgroundColor: Color(red: 0.2, green: 0.2, blue: 0.2),
    citationTextColor: Color.white,
    citationHoverTextColor: Color.white,
    citationFontSizeRatio: 0.8
)

let config = LLMStreamConfiguration(colors: colorConfig)
```

--------------------------------

### LLMStream Layout Configuration in Swift

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Configures the layout parameters for LLMStream, including padding for content, code blocks, thoughts, and tables, as well as spacing and corner radius values for various UI elements.

```swift
let layoutConfig = LayoutConfiguration(
    contentPadding: EdgeInsets(top: 0, leading: 0, bottom: 0, trailing: 0),
    codePadding: EdgeInsets(top: 12, leading: 16, bottom: 12, trailing: 16),
    thoughtPadding: EdgeInsets(top: 8, leading: 8, bottom: 8, trailing: 8),
    tablePadding: EdgeInsets(top: 8, leading: 8, bottom: 8, trailing: 8),
    spacing: 8,
    cornerRadius: 8,
    tableCornerRadius: 5,
    theoremCornerRadius: 4
)
```

--------------------------------

### Configure LLMStream Animations in Swift

Source: https://context7.com/synth-inc/llmstream/llms.txt

Defines animation configurations for thought expansion and shimmer effects used in LLMStream. It specifies animation types, durations, and gradient colors for visual feedback during streaming.

```swift
import SwiftUI
import LLMStream

let animationConfig = AnimationConfiguration(
    thoughtExpandAnimation: .spring(response: 0.35, dampingFraction: 0.85),
    shimmerAnimation: .linear(duration: 1.5)
        .delay(0.25)
        .repeatForever(autoreverses: false),
    shimmerGradient: Gradient(colors: [
        .black.opacity(0.3),
        .black,
        .black.opacity(0.3)
    ])
)

let config = LLMStreamConfiguration(animation: animationConfig)
```

--------------------------------

### Customize LLMStreamView Appearance with LLMStreamConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Configuration object for customizing the appearance and behavior of LLMStreamView. It includes preset themes (dark, light) and allows for detailed custom options for fonts, colors, layout, thought display, and code blocks.

```swift
import SwiftUI
import LLMStream

// Using preset dark theme
let darkConfig = LLMStreamConfiguration.dark

// Using preset light theme
let lightConfig = LLMStreamConfiguration.light

// Custom configuration
let customConfig = LLMStreamConfiguration(
    font: FontConfiguration(
        size: 16.0,
        lineHeight: 1.6,
        family: "SF Pro, sans-serif",
        codeFontFamily: "SF Mono, monospace"
    ),
    colors: ColorConfiguration(
        textColor: .white,
        backgroundColor: Color(red: 0.1, green: 0.1, blue: 0.12),
        codeBackgroundColor: Color(red: 0.15, green: 0.15, blue: 0.16),
        linkColor: Color(red: 0.4, green: 0.65, blue: 1.0)
    ),
    layout: LayoutConfiguration(
        contentPadding: EdgeInsets(top: 16, leading: 16, bottom: 16, trailing: 16),
        cornerRadius: 12,
        spacing: 12
    ),
    thought: ThoughtConfiguration(
        icon: Image(systemName: "brain"),
        thinkingTitle: "Processing...",
        thoughtTitle: "Analysis"
    ),
    codeBlock: CodeBlockConfiguration(
        showLanguage: true,
        showCopyButton: true,
        showActionButton: true,
        actionButtonTooltip: "Run Code"
    )
)

LLMStreamView(
    text: "Your content here",
    configuration: customConfig,
    onUrlClicked: { _ in },
    onCodeAction: { code in
        // Handle code execution
    }
)
```

--------------------------------

### Required Entitlements for LLMStream Network Access

Source: https://context7.com/synth-inc/llmstream/llms.txt

Specifies the necessary entitlements required in an Xcode project's .entitlements file for LLMStream to access network resources. These include network client and web permissions.

```xml
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>com.apple.security.network.client</key>
    <true/>
    <key>com.apple.security.web</key>
    <true/>
</dict>
</plist>
```

--------------------------------

### Process Markdown LaTeX Documents in Swift

Source: https://context7.com/synth-inc/llmstream/llms.txt

An internal Swift processor designed to clean and prepare LaTeX documents embedded within Markdown for rendering. It extracts content between \begin{document} and \end{document} and removes unnecessary LaTeX commands.

```swift
// Internal usage - automatically handles LaTeX document extraction
// Example input text with LaTeX document:
let inputText = """
\documentclass{article}
\usepackage{amsmath}
\begin{document}
\section{Introduction}
This is a theorem.
\end{document}
"""

// The processor extracts content between \begin{document} and \end{document}
// and removes unnecessary LaTeX commands
let processed = MarkdownLatexTextProcessor.cleanLatexDocuments(in: inputText)
// Result: "# Introduction\nThis is a theorem."
```

--------------------------------

### Swift: Configure Table Appearance and Behavior

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Swift code snippet shows how to configure the appearance and behavior of tables within LLMStream. Options include enabling/disabling captions, styling captions and headers, and enabling hover effects and zebra striping for improved readability.

```swift
let tableConfig = TableConfiguration(
    showCaption: true,
    captionStyle: TableConfiguration.CaptionStyle(
        fontSize: 0.9,
        textColor: Color(white: 0.8)
    ),
    headerStyle: TableConfiguration.HeaderStyle(
        fontWeight: .bold,
        textAlignment: .center,
        borderWidth: 2.0
    ),
    enableHover: true,
    enableZebraStripes: true
)
```

--------------------------------

### LLMStream Code Block Configuration in Swift

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Customizes the appearance and behavior of code blocks within LLMStream. Options include showing language labels, copy buttons, action buttons, and adjusting their sizes and opacities.

```swift
let codeConfig = CodeBlockConfiguration(
    showLanguage: true,           // Show language label
    showCopyButton: true,         // Show copy button
    showActionButton: true,       // Show action button
    languageTextSize: 13.0,       // Language label size
    copyButtonSize: 16,           // Copy button size
    actionButtonSize: 16,         // Action button size
    copyButtonOpacity: 0.5,       // Copy button normal opacity
    copyButtonHoverOpacity: 1.0,  // Copy button hover opacity
    actionButtonOpacity: 0.5,     // Action button normal opacity
    actionButtonHoverOpacity: 1.0,// Action button hover opacity
    actionButtonIcon: "play.circle", // SF Symbol name for action button
    actionButtonTooltip: "Execute"   // Action button tooltip
)
```

--------------------------------

### Configure Typography Settings with FontConfiguration (Swift)

Source: https://context7.com/synth-inc/llmstream/llms.txt

Configures typography settings for text, code, tables, and mathematical expressions within the LLMStreamView. Allows customization of font size, line height, and specific font families for different content types.

```swift
import LLMStream

let fontConfig = FontConfiguration(
    size: 15.0,
    lineHeight: 1.5,
    family: "-apple-system, BlinkMacSystemFont, sans-serif",
    codeFontFamily: "SF Mono, Monaco, Menlo, monospace",
    tableFontFamily: "system-ui, serif",
    mathFontFamily: "STIX Two Math, Latin Modern Math, serif"
)

let config = LLMStreamConfiguration(font: fontConfig)
```

--------------------------------

### Swift: Customize Thought Process Display

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Swift code defines a `ThoughtConfiguration` for customizing the display of thought processes in LLMStream. It allows setting an optional icon, icon size, loading state title, normal state title, and whether to show an expand button.

```swift
let thoughtConfig = ThoughtConfiguration(
    icon: Image(systemName: "brain"),  // Optional icon
    iconSize: 16,
    thinkingTitle: "Thinking…",      // Loading state title
    thoughtTitle: "Thought-process",    // Normal state title
    showExpandButton: true
)
```

--------------------------------

### Swift: Add Custom Action Button to Code Blocks

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Swift code demonstrates how to add a custom action button to code blocks within an LLMStreamView. It allows triggering a callback function when the button is pressed, passing the code content to the handler. Configuration options include the button's icon and tooltip.

```swift
LLMStreamView(
    text: "Your content here",
    configuration: LLMStreamConfiguration(
        codeBlock: CodeBlockConfiguration(
            showActionButton: true,
            actionButtonIcon: "play.circle",
            actionButtonTooltip: "Run this code"
        )
    ),
    onCodeAction: { code in
        // Handle the code execution here
        print("Code to execute:", code)
    }
)
```

--------------------------------

### Swift: Configure Animations for LLMStream

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

This Swift code snippet illustrates how to configure animations within LLMStream, including animations for expanding thoughts and shimmering effects. It allows specifying animation types, durations, delays, and gradients for visual feedback.

```swift
let animationConfig = AnimationConfiguration(
    thoughtExpandAnimation: .spring(response: 0.35, dampingFraction: 0.85),
    shimmerAnimation: .linear(duration: 1.5).delay(0.25).repeatForever(autoreverses: false),
    shimmerGradient: Gradient(colors: [
        .black.opacity(0.3),
        .black,
        .black.opacity(0.3)
    ])
)
```

--------------------------------

### LLMStream Font Configuration in Swift

Source: https://github.com/synth-inc/llmstream/blob/main/README.md

Defines custom font settings for LLMStream, allowing adjustments to base font size, line height, and specific font families for regular text, code blocks, tables, and mathematical expressions.

```swift
let fontConfig = FontConfiguration(
    size: 14.0,                    // Base font size
    lineHeight: 1.4,               // Line height multiplier
    family: "-apple-system...",    // Main font family
    codeFontFamily: "SF Mono...",  // Code blocks font
    tableFontFamily: "system-ui...",// Table font
    mathFontFamily: "STIX Two Math..." // Math expressions font
)
```
Tokens:
Raw