quasivision 0.2.2

A Rust-based pseudo-visual understanding tool. Analyzes screenshots, UI mockups, and real-world photos — detects UI components, recognizes text via OCR, identifies objects with YOLOE-26n.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
use std::path::Path;

use ab_glyph::{FontArc, PxScale};
use image::{DynamicImage, RgbImage, Rgb};
use imageproc::drawing::{draw_filled_rect_mut, draw_hollow_rect_mut, draw_text_mut};
use imageproc::rect::Rect;
use serde_json;

use crate::element::{Element, OutputResult, CompactElement, AiOutput, TreeOutput, elements_to_text};
use crate::object_detector::DetectionNode;

/// 加载系统字体用于文本渲染
fn load_detection_font() -> Option<FontArc> {
    // 各平台字体目录
    let font_dirs: &[&str] = if cfg!(target_os = "windows") {
        &[r"C:\Windows\Fonts"]
    } else if cfg!(target_os = "macos") {
        &["/System/Library/Fonts", "/Library/Fonts", "~/Library/Fonts"]
    } else {
        &["/usr/share/fonts", "/usr/local/share/fonts", "~/.fonts"]
    };

    // 首选英文字体(跨平台通用)+ 中文字体备选
    #[cfg(target_os = "windows")]
    let preferred = ["arial.ttf", "msyh.ttc", "simsun.ttc", "msyhbd.ttf"];
    #[cfg(target_os = "macos")]
    let preferred = ["Helvetica.ttc", "PingFang.ttc", "STHeiti Light.ttc", "Arial Unicode.ttf"];
    #[cfg(target_os = "linux")]
    let preferred = ["DejaVuSans.ttf", "NotoSansCJK-Regular.ttc", "WenQuanYiMicroHei.ttf", "LiberationSans-Regular.ttf"];

    // 扫描各字体目录
    for dir_str in font_dirs {
        let dir = Path::new(dir_str);
        if !dir.exists() {
            continue;
        }
        // 先试首选字体
        for name in &preferred {
            let path = dir.join(name);
            if path.exists() {
                if let Ok(data) = std::fs::read(&path) {
                    if let Ok(font) = FontArc::try_from_vec(data) {
                        return Some(font);
                    }
                }
            }
        }
        // 再扫目录下任意 .ttf/.ttc
        if let Ok(entries) = std::fs::read_dir(dir) {
            for entry in entries.flatten() {
                let p = entry.path();
                let ext = p.extension().and_then(|e| e.to_str()).unwrap_or("");
                if ext == "ttf" || ext == "ttc" {
                    if let Ok(data) = std::fs::read(&p) {
                        if let Ok(font) = FontArc::try_from_vec(data) {
                            return Some(font);
                        }
                    }
                }
            }
        }
    }

    None
}

/// 递归展平检测树为列表(所有层级)
fn flatten_detections(nodes: &[DetectionNode]) -> Vec<DetectionNode> {
    let mut result = Vec::new();
    for node in nodes {
        result.push(DetectionNode {
            class_name: node.class_name.clone(),
            confidence: node.confidence,
            bbox: node.bbox,
            children: Vec::new(),
        });
        result.extend(flatten_detections(&node.children));
    }
    result
}

/// 写入文件(自动创建父目录)
fn write_output(path: &str, data: &str) -> std::io::Result<()> {
    if let Some(parent) = Path::new(path).parent() {
        std::fs::create_dir_all(parent)?;
    }
    std::fs::write(path, data)
}

/// 将元素列表导出为 JSON 文件
pub fn save_json(elements: &[Element], img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let result = OutputResult {
        comps: elements.to_vec(),
        img_shape,
    };

    let json_str = serde_json::to_string_pretty(&result)?;
    write_output(output_path, &json_str)?;
    println!("[Export] JSON saved to: {}", output_path);
    Ok(())
}

/// 保存压缩格式 JSON(短键名、拍平结构、去掉内部字段)
pub fn save_compact_json(elements: &[Element], _img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let compact: Vec<CompactElement> = elements.iter().map(CompactElement::from).collect();
    let json_str = serde_json::to_string(&compact)?;
    let byte_len = json_str.len();
    write_output(output_path, &json_str)?;
    println!("[Export] Compact JSON saved to: {} ({} bytes)", output_path, byte_len);
    Ok(())
}

/// 保存 AI 优化格式(坐标归一化到 0-1000)
pub fn save_ai_json(elements: &[Element], img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let ai = AiOutput::from_elements(elements, img_shape);
    let json_str = serde_json::to_string(&ai)?;
    let byte_len = json_str.len();
    write_output(output_path, &json_str)?;
    println!("[Export] AI-ready JSON saved to: {} ({} bytes)", output_path, byte_len);
    Ok(())
}

/// 保存纯文本表示
pub fn save_text_summary(elements: &[Element], img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let text = elements_to_text(elements, img_shape);
    write_output(output_path, &text)?;
    println!("[Export] Text summary saved to: {} ({} bytes, {} chars)", output_path, text.len(), text.chars().count());
    Ok(())
}

// ── 树形输出 ──

pub fn save_tree_json(elements: &[Element], img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let tree = TreeOutput::from_elements(elements, img_shape);
    let json_str = serde_json::to_string_pretty(&tree)?;
    let byte_len = json_str.len();
    write_output(output_path, &json_str)?;
    println!("[Export] Tree JSON saved to: {} ({} bytes)", output_path, byte_len);
    Ok(())
}

pub fn save_tree_text(elements: &[Element], img_shape: (u32, u32), output_path: &str) -> anyhow::Result<()> {
    let tree = TreeOutput::from_elements(elements, img_shape);
    let text = tree.to_text();
    write_output(output_path, &text)?;
    println!("[Export] Tree text saved to: {} ({} bytes, {} chars)",
        output_path, text.len(), text.chars().count());
    Ok(())
}

// ── 物体检测树形输出 ──

pub fn save_detection_tree_json(
    roots: &[DetectionNode],
    img_shape: (u32, u32),
    output_path: &str,
) -> anyhow::Result<()> {
    let output = serde_json::json!({
        "img_shape": [img_shape.1, img_shape.0],
        "count": count_all(roots),
        "objects": roots,
    });
    let json_str = serde_json::to_string_pretty(&output)?;
    write_output(output_path, &json_str)?;
    println!(
        "[Export] Detection tree JSON saved to: {} ({} bytes, {} objects)",
        output_path, json_str.len(), count_all(roots)
    );
    Ok(())
}

fn count_all(nodes: &[DetectionNode]) -> usize {
    nodes.iter().map(|n| 1 + count_all(&n.children)).sum()
}

pub fn save_detection_tree_text(
    roots: &[DetectionNode],
    img_shape: (u32, u32),
    output_path: &str,
) -> anyhow::Result<()> {
    if roots.is_empty() {
        let text = format!("Objects ({}×{}):\n  (none detected)", img_shape.1, img_shape.0);
        write_output(output_path, &text)?;
        println!("[Export] Detection tree text saved to: {} (no objects)", output_path);
        return Ok(());
    }

    let total = count_all(roots);
    let mut lines = Vec::new();
    lines.push(format!("Objects ({}×{}) — {} found:", img_shape.1, img_shape.0, total));

    fn render_node(node: &DetectionNode, prefix: &str, is_last: bool, lines: &mut Vec<String>) {
        let connector = if is_last { "└─ " } else { "├─ " };
        let x = node.bbox.x_min.round() as i32;
        let y = node.bbox.y_min.round() as i32;
        let w = (node.bbox.x_max - node.bbox.x_min).round() as i32;
        let h = (node.bbox.y_max - node.bbox.y_min).round() as i32;
        let pct = (node.confidence * 100.0).round() as u32;
        lines.push(format!("{}{}[{:>3},{:>3} {:>3}×{:>3}] {} ({}%)", prefix, connector, x, y, w, h, node.class_name, pct));
        let child_prefix = if is_last { "   " } else { "" };
        let new_prefix = format!("{}{}", prefix, child_prefix);
        let count = node.children.len();
        for (i, child) in node.children.iter().enumerate() {
            render_node(child, &new_prefix, i == count - 1, lines);
        }
    }

    let root_count = roots.len();
    for (i, root) in roots.iter().enumerate() {
        render_node(root, "", i == root_count - 1, &mut lines);
    }

    let text = lines.join("\n");
    write_output(output_path, &text)?;
    println!("[Export] Detection tree text saved to: {} ({} bytes, {} chars)", output_path, text.len(), text.chars().count());
    Ok(())
}

// ── 元素可视化 ──

pub fn draw_elements(img: &DynamicImage, elements: &[Element]) -> RgbImage {
    let mut rgb = img.to_rgb8();

    let color_map: Vec<(&str, Rgb<u8>)> = vec![
        ("Text", Rgb([0, 0, 255])),
        ("Compo", Rgb([255, 0, 255])),
        ("Block", Rgb([0, 255, 0])),
        ("Image", Rgb([255, 0, 0])),
        ("Button", Rgb([0, 200, 0])),
        ("Icon", Rgb([255, 165, 0])),
        ("Text Content", Rgb([255, 0, 255])),
        ("Noise", Rgb([6, 6, 255])),
    ];

    fn put_safe(rgb: &mut RgbImage, x: i32, y: i32, color: Rgb<u8>) {
        if x >= 0 && (x as u32) < rgb.width() && y >= 0 && (y as u32) < rgb.height() {
            rgb.put_pixel(x as u32, y as u32, color);
        }
    }

    fn draw_rect(rgb: &mut RgbImage, c1: i32, r1: i32, c2: i32, r2: i32, color: Rgb<u8>) {
        for x in c1..=c2 {
            put_safe(rgb, x, r1, color);
            put_safe(rgb, x, r1 + 1, color);
            put_safe(rgb, x, r2, color);
            put_safe(rgb, x, r2 - 1, color);
        }
        for y in r1..=r2 {
            put_safe(rgb, c1, y, color);
            put_safe(rgb, c1 + 1, y, color);
            put_safe(rgb, c2, y, color);
            put_safe(rgb, c2 - 1, y, color);
        }
    }

    for ele in elements {
        let color = color_map.iter()
            .find(|(name, _)| *name == ele.class)
            .map(|(_, c)| *c)
            .unwrap_or(Rgb([0, 255, 0]));
        let (c1, r1, c2, r2) = ele.put_bbox();
        draw_rect(&mut rgb, c1, r1, c2, r2, color);
    }

    rgb
}

pub fn save_visualization(img: &DynamicImage, elements: &[Element], output_path: &str) -> anyhow::Result<()> {
    let vis = draw_elements(img, elements);
    if let Some(parent) = Path::new(output_path).parent() {
        std::fs::create_dir_all(parent)?;
    }
    vis.save(output_path)?;
    println!("[Export] Visualization saved to: {}", output_path);
    Ok(())
}

// ── 物体检测可视化(所有层级,系统字体渲染标签) ──

/// 在图片上绘制物体检测结果(所有层级)
///
/// - 所有检测框用 2px 彩色边框绘制
/// - 框内右上角用系统字体渲染"类别 置信度%"标签(空间够且不遮挡则画)
pub fn draw_object_detections(img: &DynamicImage, roots: &[DetectionNode]) -> RgbImage {
    let mut rgb = img.to_rgb8();

    // 展平所有节点
    let all_nodes = flatten_detections(roots);
    if all_nodes.is_empty() {
        return rgb;
    }

    // 加载字体
    let font = load_detection_font();
    if font.is_none() {
        eprintln!("  [WARN] No Chinese font found — detection labels will use fallback");
    }

    const COLORS: &[Rgb<u8>] = &[
        Rgb([255, 0, 0]),     Rgb([0, 180, 0]),
        Rgb([0, 100, 255]),   Rgb([255, 165, 0]),
        Rgb([180, 0, 255]),   Rgb([0, 200, 200]),
        Rgb([255, 0, 180]),   Rgb([200, 200, 0]),
    ];

    struct BoxInfo { x1: i32, y1: i32, x2: i32, y2: i32, label: String, color: Rgb<u8> }

    let boxes: Vec<BoxInfo> = all_nodes.iter().enumerate().map(|(i, node)| {
        let x1 = node.bbox.x_min.round() as i32;
        let y1 = node.bbox.y_min.round() as i32;
        let x2 = node.bbox.x_max.round() as i32;
        let y2 = node.bbox.y_max.round() as i32;
        let pct = (node.confidence * 100.0).round() as u32;
        BoxInfo {
            x1, y1, x2, y2,
            label: format!("{} {}%", node.class_name, pct),
            color: COLORS[i % COLORS.len()],
        }
    }).collect();

    // 第一遍:画所有框(2px 粗)
    for info in &boxes {
        let w = (info.x2 - info.x1).max(0) as u32;
        let h = (info.y2 - info.y1).max(0) as u32;
        if w < 2 || h < 2 { continue; }
        let r = Rect::at(info.x1, info.y1).of_size(w, h);
        draw_hollow_rect_mut(&mut rgb, r, info.color);
        let r2 = Rect::at(info.x1 + 1, info.y1 + 1).of_size(w.saturating_sub(2), h.saturating_sub(2));
        draw_hollow_rect_mut(&mut rgb, r2, info.color);
    }

    // 第二遍:画标签(框内右上角)
    if let Some(ref f) = font {
        let scale = PxScale::from(13.0);
        let black = Rgb([0, 0, 0]);

        for info in &boxes {
            let tw = imageproc::drawing::text_size(scale, f, &info.label).0 as i32;
            let th = 16;
            let pad = 3;

            // 右上角位置
            let lx1 = info.x2 - tw - pad;
            let ly1 = info.y1 + 2;
            let lx2 = info.x2 - 2;
            let ly2 = ly1 + th;

            if lx1 <= info.x1 || ly2 >= info.y2 { continue; }

            // 检查是否遮挡其他框(>30% 标签面积重叠则跳过)
            let mut blocked = false;
            for other in &boxes {
                if other.x1 == info.x1 && other.y1 == info.y1 { continue; }
                let ox = lx1.max(other.x1);
                let oy = ly1.max(other.y1);
                let ox2 = lx2.min(other.x2);
                let oy2 = ly2.min(other.y2);
                if ox < ox2 && oy < oy2 {
                    let inter = (ox2 - ox) as i64 * (oy2 - oy) as i64;
                    let label_area = (lx2 - lx1) as i64 * (ly2 - ly1) as i64;
                    if label_area > 0 && (inter * 100 / label_area) > 30 {
                        blocked = true;
                        break;
                    }
                }
            }
            if blocked { continue; }

            // 白色背景
            let bg_r = Rect::at(lx1, ly1).of_size((lx2 - lx1) as u32, (ly2 - ly1) as u32);
            draw_filled_rect_mut(&mut rgb, bg_r, Rgb([255, 255, 255]));
            // 标签边框
            draw_hollow_rect_mut(&mut rgb, bg_r, info.color);
            // 文字
            draw_text_mut(&mut rgb, black, lx1 + 2, ly1, scale, f, &info.label);
        }
    } else {
        // 无字体 → 画小色块代替标签
        for info in &boxes {
            let s = 8;
            let ix1 = info.x2 - s - 2;
            let iy1 = info.y1 + 2;
            let ir = Rect::at(ix1, iy1).of_size(s as u32, s as u32);
            draw_filled_rect_mut(&mut rgb, ir, info.color);
        }
    }

    rgb
}

pub fn save_object_detection_visualization(
    img: &DynamicImage,
    roots: &[DetectionNode],
    output_path: &str,
) -> anyhow::Result<()> {
    let vis = draw_object_detections(img, roots);
    if let Some(parent) = Path::new(output_path).parent() {
        std::fs::create_dir_all(parent)?;
    }
    vis.save(output_path)?;
    println!("[Export] Object detection visualization saved to: {}", output_path);
    Ok(())
}