feat(AI): 统一多模态模型架构,整合文本和视觉推理路径 - 将文本生成和VL(图→文)功能合并到单一的Qwen3.5-4B多模态MNN模型 - 移除独立的Qwen3-VL-4B模型依赖,MLX VL改为使用.llm的多模态模型 - 更新ModelKind枚举,新增userFacing集合用于面向用户展示 - MNN后端现在同时支持文本和视觉任务,模拟器回退到MLX refactor(models): 模型管理和界面调整以适应新的多模态架构 - 更新模型管理界面,只显示统一的Qwen3.5-4B(MNN)模型给用户 - 修改就绪状态检查逻辑,使用ModelKind.userFacing替代allCases - 更新模型文件清单,从Qwen3.5-2B升级到Qwen3.5-4B-4bit - 调整模型管理页面UI,突出MNN+SME2端侧加速功能 feat(camera): 添加拍照识别引擎切换功能 - 实现双路径拍照识别:Apple Vision OCR + 文本模型 和 Qwen3-VL直接识别 - 添加预处理逻辑,优化Qwen3-VL对窄长区域图片的识别效果 - 在模型管理页面添加拍照识别引擎选择组件 - 提供用户界面选项,在两种识别方式间切换 style(ui): 优化输入框样式和颜色主题一致性 - 为指标快速表单添加浅色主题偏好 - 统一所有文本输入框的颜色样式(theme) - 创建EntryInputField组件,替换原有的单行输入+按钮模式 - 实现聊天框风格的条目输入,支持多行自适应和圆形发送按钮 fix(build): 修正Xcode项目配置中的重复框架搜索路径 - 清理project.pbxproj中重复的FRAMEWORK_SEARCH_PATHS配置 - 重新排列Swift桥接头文件配置确保正确引用 - 修复因路径配置重复导致的编译警告问题 test: 增加区域图片预处理和模型清单测试覆盖 - 添加RegionImageCropper.prepareForQwenVL的单元测试 - 验证宽而矮图片的放大和填充逻辑 - 更新ModelManifestTests中的字节数预期值以匹配新模型 - 修正OCRService中VNRecognizedTextObservation类型的处理 ```
153 lines
6.9 KiB
Swift
153 lines
6.9 KiB
Swift
import XCTest
|
|
import CoreGraphics
|
|
import AVFoundation
|
|
@testable import 康康
|
|
|
|
/// 异常项快拍的局部裁剪几何。
|
|
/// 回归用例:屏上「宽而矮」的小框,必须裁出「宽 > 高」的照片 rect。
|
|
/// 旧实现用 `metadataOutputRectConverted`(传感器横向坐标)套到竖屏照片 → x/y 轴对调,
|
|
/// 把宽框裁成竖窄条(2026-05-31 真机 bug)。本组用例钉住正确的纯几何映射。
|
|
final class RegionImageCropperTests: XCTestCase {
|
|
|
|
/// 竖屏照片 + 竖屏屏幕 + 宽框 → 裁出的 rect 必须是横向(宽 > 高)。
|
|
func testWideBoxYieldsLandscapeCropRect() {
|
|
let photo = CGSize(width: 3024, height: 4032) // 竖屏照片(像素)
|
|
let view = CGSize(width: 393, height: 852) // 竖屏屏幕(点)
|
|
let box = RegionFraming.box(in: view) // 宽 84% / 矮 160
|
|
|
|
let rect = RegionImageCropper.cropRect(photoPixelSize: photo, box: box, in: view)
|
|
|
|
XCTAssertGreaterThan(rect.width, rect.height,
|
|
"宽框应裁出宽 rect;若 width<height 说明轴对调 bug 复现")
|
|
XCTAssertGreaterThan(rect.width, 0)
|
|
XCTAssertGreaterThan(rect.height, 0)
|
|
}
|
|
|
|
/// 裁出的 rect 必须完整落在照片像素范围内(越界夹紧)。
|
|
func testCropRectStaysInsidePhotoBounds() {
|
|
let photo = CGSize(width: 3024, height: 4032)
|
|
let view = CGSize(width: 393, height: 852)
|
|
let box = RegionFraming.box(in: view)
|
|
|
|
let rect = RegionImageCropper.cropRect(photoPixelSize: photo, box: box, in: view)
|
|
|
|
XCTAssertGreaterThanOrEqual(rect.minX, 0)
|
|
XCTAssertGreaterThanOrEqual(rect.minY, 0)
|
|
XCTAssertLessThanOrEqual(rect.maxX, photo.width)
|
|
XCTAssertLessThanOrEqual(rect.maxY, photo.height)
|
|
}
|
|
|
|
/// 宽框宽高比应与照片裁剪 rect 宽高比一致(aspect-fill 等比映射,不拉伸)。
|
|
func testCropRectPreservesBoxAspectRatio() {
|
|
let photo = CGSize(width: 3024, height: 4032)
|
|
let view = CGSize(width: 393, height: 852)
|
|
let box = RegionFraming.box(in: view)
|
|
|
|
let rect = RegionImageCropper.cropRect(photoPixelSize: photo, box: box, in: view)
|
|
|
|
let boxAspect = box.width / box.height
|
|
let rectAspect = rect.width / rect.height
|
|
XCTAssertEqual(rectAspect, boxAspect, accuracy: 0.05,
|
|
"等比映射下,裁剪 rect 宽高比应与屏上小框一致")
|
|
}
|
|
|
|
/// 退化输入:零尺寸不应崩溃,返回 .zero。
|
|
func testZeroInputsReturnZero() {
|
|
XCTAssertEqual(
|
|
RegionImageCropper.cropRect(photoPixelSize: .zero,
|
|
box: CGRect(x: 0, y: 0, width: 10, height: 10),
|
|
in: CGSize(width: 100, height: 200)),
|
|
.zero)
|
|
}
|
|
|
|
// MARK: - aspect-FIT(静态图框选)
|
|
|
|
/// 静态图以 aspect-fit 显示;宽框应裁出宽 rect 且落在照片范围内。
|
|
func testAspectFitWideBoxYieldsLandscapeInsideBounds() {
|
|
let photo = CGSize(width: 3024, height: 4032)
|
|
let view = CGSize(width: 393, height: 852)
|
|
let fitted = AVMakeRect(aspectRatio: photo, insideRect: CGRect(origin: .zero, size: view))
|
|
// fitted 内的一个宽框
|
|
let box = CGRect(x: fitted.minX + 30, y: fitted.midY - 50, width: fitted.width - 60, height: 100)
|
|
|
|
let rect = RegionImageCropper.cropRectAspectFit(photoPixelSize: photo, box: box, imageFrame: fitted)
|
|
|
|
XCTAssertGreaterThan(rect.width, rect.height)
|
|
XCTAssertGreaterThanOrEqual(rect.minX, 0)
|
|
XCTAssertGreaterThanOrEqual(rect.minY, 0)
|
|
XCTAssertLessThanOrEqual(rect.maxX, photo.width)
|
|
XCTAssertLessThanOrEqual(rect.maxY, photo.height)
|
|
}
|
|
|
|
/// 框正好等于整个显示区 → 裁出整张照片(像素)。
|
|
func testAspectFitFullFrameBoxYieldsFullPhoto() {
|
|
let photo = CGSize(width: 2000, height: 1000) // 横向照片
|
|
let view = CGSize(width: 400, height: 800)
|
|
let fitted = AVMakeRect(aspectRatio: photo, insideRect: CGRect(origin: .zero, size: view))
|
|
|
|
let rect = RegionImageCropper.cropRectAspectFit(photoPixelSize: photo, box: fitted, imageFrame: fitted)
|
|
|
|
XCTAssertEqual(rect.minX, 0, accuracy: 1)
|
|
XCTAssertEqual(rect.minY, 0, accuracy: 1)
|
|
XCTAssertEqual(rect.width, photo.width, accuracy: 2)
|
|
XCTAssertEqual(rect.height, photo.height, accuracy: 2)
|
|
}
|
|
|
|
/// 等比映射:选框宽高比应与裁出 rect 宽高比一致。
|
|
func testAspectFitPreservesBoxAspectRatio() {
|
|
let photo = CGSize(width: 3024, height: 4032)
|
|
let view = CGSize(width: 393, height: 852)
|
|
let fitted = AVMakeRect(aspectRatio: photo, insideRect: CGRect(origin: .zero, size: view))
|
|
let box = CGRect(x: fitted.minX + 20, y: fitted.minY + 40, width: 180, height: 120)
|
|
|
|
let rect = RegionImageCropper.cropRectAspectFit(photoPixelSize: photo, box: box, imageFrame: fitted)
|
|
|
|
XCTAssertEqual(rect.width / rect.height, box.width / box.height, accuracy: 0.05)
|
|
}
|
|
|
|
func testAspectFitZeroInputsReturnZero() {
|
|
XCTAssertEqual(
|
|
RegionImageCropper.cropRectAspectFit(photoPixelSize: .zero,
|
|
box: CGRect(x: 0, y: 0, width: 10, height: 10),
|
|
imageFrame: CGRect(x: 0, y: 0, width: 100, height: 100)),
|
|
.zero)
|
|
}
|
|
|
|
/// Qwen3-VL 对宽而矮的局部小图更容易在 processor 缩放后丢字。
|
|
/// 进入 VL 前应把短边放大并加白边,但不拉伸内容。
|
|
func testPrepareForQwenVLUpscalesWideShortCropAndAddsPadding() {
|
|
let image = solidImage(size: CGSize(width: 320, height: 80))
|
|
|
|
let prepared = RegionImageCropper.prepareForQwenVL(image)
|
|
|
|
XCTAssertGreaterThanOrEqual(prepared.size.height, 448)
|
|
XCTAssertGreaterThan(prepared.size.width, 320)
|
|
XCTAssertEqual(prepared.size.width / prepared.size.height,
|
|
4.0,
|
|
accuracy: 0.8,
|
|
"预处理应大致保留宽条内容比例,只允许白边造成轻微变化")
|
|
}
|
|
|
|
func testPrepareForQwenVLDoesNotEnlargePastLongEdgeLimit() {
|
|
let image = solidImage(size: CGSize(width: 5000, height: 900))
|
|
|
|
let prepared = RegionImageCropper.prepareForQwenVL(image)
|
|
|
|
XCTAssertLessThanOrEqual(max(prepared.size.width, prepared.size.height), 2400 + 128)
|
|
}
|
|
|
|
private func solidImage(size: CGSize) -> UIImage {
|
|
let format = UIGraphicsImageRendererFormat.default()
|
|
format.scale = 1
|
|
return UIGraphicsImageRenderer(size: size, format: format).image { ctx in
|
|
UIColor.white.setFill()
|
|
ctx.fill(CGRect(origin: .zero, size: size))
|
|
UIColor.black.setFill()
|
|
ctx.fill(CGRect(x: size.width * 0.1,
|
|
y: size.height * 0.35,
|
|
width: size.width * 0.8,
|
|
height: size.height * 0.3))
|
|
}
|
|
}
|
|
}
|