Figure & Table Extraction

[huypham37]

Overview

Extract and save figures/tables from PDFs for easy reference and citation.

Why

Scientists often need to reference specific figures. "Figure 3 shows..." should link to the actual figure.

Plan

1. Create Figure Model

// Model/Figure.swift
struct Figure: Codable, Identifiable {
    let id: String
    let paperId: String
    let pageNumber: Int
    let bounds: CGRect
    let imageData: Data  // PNG snapshot
    var caption: String?
    var label: String?  // "Figure 3", "Table 2"
    var linkedNoteIds: [String]
    let extractedAt: Date
}

2. Extraction Methods

• Manual: User draws rectangle on PDF → extract region
• Semi-auto: Detect image regions on page
• Caption parsing: Find "Figure X:" patterns

3. FigureService

• Extract region from PDF page as image
• OCR caption (Vision framework)
• Store images in ~/.pdfscribe/figures/

4. UI Components

• Figure extraction mode (toolbar toggle)
• Draw rectangle → confirm → add caption
• Figure gallery per paper
• Thumbnail previews

5. Integration

• Link figures in notes: [[figure:id]]
• Insert figure in note (embedded image)
• Copy figure to clipboard

Acceptance Criteria

• Extract region from PDF as image
• Save with caption/label
• View all figures for a paper
• Link figures in notes
• Copy figure to clipboard