Add structured Vision text recognition

CHANGELOG.md

@@ -2,8 +2,10 @@
 
 All notable changes to TextCaptureKit are documented here.
 
-## Unreleased
+## 0.1.0 - 2026-09-12
 
 ### Added
 
 - Bootstrapped the Swift package, DocC catalog, CI/CD workflows, shared AgentGuidelines integration, and repository policy.
+- Added configurable on-device Vision text recognition with structured candidates, confidence, and normalized bounds.
+- Added deterministic plain-text projection and automatic encoded-image orientation handling.

README.md

@@ -11,9 +11,9 @@
 
 # TextCaptureKit
 
-TextCaptureKit is a reusable, UI-agnostic wrapper around Apple's Vision text-recognition APIs. It will convert host-supplied still images into structured recognized text while preserving candidates, confidence, and layout for application-owned review and correction.
+TextCaptureKit is a reusable, UI-agnostic wrapper around Apple's Vision text-recognition APIs. It converts host-supplied still images into structured recognized text while preserving encoded orientation, candidates, confidence, and layout for application-owned review and correction.
 
-The package is currently a bootstrapped foundation. Camera capture, scanner UI, parsing, persistence, and product presentation remain host responsibilities.
+Camera capture, scanner UI, parsing, persistence, and product presentation remain host responsibilities.
 
 ## Documentation
 

Sources/TextCaptureKit/TextCaptureCandidate.swift

@@ -0,0 +1,20 @@
+import Foundation
+
+/// One text candidate reported for a recognized region.
+public struct TextCaptureCandidate: Sendable, Hashable {
+    /// The recognized text.
+    public let text: String
+
+    /// Vision's confidence in the candidate, expressed from zero through one.
+    public let confidence: Float
+
+    /// Creates a recognized-text candidate.
+    ///
+    /// - Parameters:
+    ///   - text: The recognized text.
+    ///   - confidence: Vision's confidence in the candidate.
+    public init(text: String, confidence: Float) {
+        self.text = text
+        self.confidence = confidence
+    }
+}

Sources/TextCaptureKit/TextCaptureImage.swift

@@ -0,0 +1,34 @@
+import Foundation
+
+/// An encoded still image and its display orientation.
+public struct TextCaptureImage: Sendable, Hashable {
+    /// The encoded image bytes.
+    public let data: Data
+
+    /// The orientation required to present the image upright.
+    public let orientation: TextCaptureImageOrientation
+
+    /// Creates a still-image input.
+    ///
+    /// - Parameters:
+    ///   - data: The encoded image bytes.
+    ///   - orientation: The orientation required to present the image upright.
+    public init(data: Data, orientation: TextCaptureImageOrientation = .automatic) {
+        self.data = data
+        self.orientation = orientation
+    }
+}
+
+/// The orientation required to present an encoded still image upright.
+public enum TextCaptureImageOrientation: Sendable, Hashable {
+    /// Reads the encoded image's orientation metadata, falling back to upright.
+    case automatic
+    case down
+    case downMirrored
+    case left
+    case leftMirrored
+    case right
+    case rightMirrored
+    case up
+    case upMirrored
+}