start-from-step

autofigure2.py

@@ -1587,6 +1587,68 @@ def crop_and_remove_background(
     return icon_infos
 
 
+def _load_icon_infos_from_existing(output_dir: Path, boxlib_path: str) -> list:
+    """从已有 outputs 目录加载 icon_infos（用于 --start_from 4/5 恢复流程）"""
+    icons_dir = output_dir / "icons"
+    with open(boxlib_path, 'r', encoding='utf-8') as f:
+        boxlib_data = json.load(f)
+    icon_infos = []
+    for box in boxlib_data.get("boxes", []):
+        box_id = box["id"]
+        label = box.get("label", f"<AF>{box_id + 1:02d}")
+        label_clean = label.replace("<", "").replace(">", "")
+        x1, y1, x2, y2 = box["x1"], box["y1"], box["x2"], box["y2"]
+        crop_path = icons_dir / f"icon_{label_clean}.png"
+        nobg_path = icons_dir / f"icon_{label_clean}_nobg.png"
+        nobg_str = str(nobg_path) if nobg_path.exists() else str(crop_path)
+        if not crop_path.exists() and not nobg_path.exists():
+            raise FileNotFoundError(f"start_from>=4 需要图标文件: {crop_path} 或 {nobg_path}")
+        icon_infos.append({
+            "id": box_id, "label": label, "label_clean": label_clean,
+            "x1": x1, "y1": y1, "x2": x2, "y2": y2,
+            "width": x2 - x1, "height": y2 - y1,
+            "crop_path": str(crop_path), "nobg_path": nobg_str,
+        })
+    return icon_infos
+
+
+def _require_files_for_step(output_dir: Path, step: int) -> None:
+    """检查指定步骤所需文件是否存在，不存在则抛出 FileNotFoundError"""
+    figure = output_dir / "figure.png"
+    samed = output_dir / "samed.png"
+    boxlib = output_dir / "boxlib.json"
+    icons_dir = output_dir / "icons"
+    template = output_dir / "template.svg"
+    optimized = output_dir / "optimized_template.svg"
+
+    if step >= 2:
+        if not figure.exists():
+            raise FileNotFoundError(f"从步骤 {step} 开始需要: {figure}")
+    if step >= 3:
+        if not samed.exists():
+            raise FileNotFoundError(f"从步骤 {step} 开始需要: {samed}")
+        if not boxlib.exists():
+            raise FileNotFoundError(f"从步骤 {step} 开始需要: {boxlib}")
+    if step >= 4:
+        if not icons_dir.is_dir():
+            raise FileNotFoundError(f"从步骤 {step} 开始需要目录: {icons_dir}")
+        # 至少需有 boxlib 中对应的裁剪图
+        with open(boxlib, 'r', encoding='utf-8') as f:
+            boxes = json.load(f).get("boxes", [])
+        for box in boxes:
+            label = box.get("label", f"<AF>{box['id']+1:02d}")
+            label_clean = label.replace("<", "").replace(">", "")
+            crop = icons_dir / f"icon_{label_clean}.png"
+            nobg = icons_dir / f"icon_{label_clean}_nobg.png"
+            if not crop.exists() and not nobg.exists():
+                raise FileNotFoundError(f"从步骤 {step} 开始需要: {crop} 或 {nobg}")
+    if step >= 5:
+        if not optimized.exists() and not template.exists():
+            raise FileNotFoundError(
+                f"从步骤 {step} 开始需要: {optimized} 或 {template}"
+            )
+
+
 # ============================================================================
 # 步骤四：多模态调用生成 SVG
 # ============================================================================
@@ -2402,6 +2464,7 @@ def method_to_svg(
     sam_max_masks: int = 32,
     rmbg_model_path: Optional[str] = None,
     stop_after: int = 5,
+    start_from: int = 1,
     placeholder_mode: PlaceholderMode = "label",
     optimize_iterations: int = 2,
     merge_threshold: float = 0.9,
@@ -2424,6 +2487,7 @@ def method_to_svg(
         sam_max_masks: SAM3 API 最大 masks 数（api 模式使用）
         rmbg_model_path: RMBG 模型路径
         stop_after: 执行到指定步骤后停止
+        start_from: 从指定步骤开始（1-5，默认1。>=4 时跳过步骤1-3，使用已有 figure.png/samed.png/boxlib.json/icons）
         placeholder_mode: 占位符模式
             - "none": 无特殊样式
             - "box": 传入 boxlib 坐标
@@ -2463,21 +2527,29 @@ def method_to_svg(
     if sam_backend_value == "fal":
         print(f"SAM3 API max_masks: {sam_max_masks}")
     print(f"执行到步骤: {stop_after}")
+    print(f"从步骤开始: {start_from}")
     print(f"占位符模式: {placeholder_mode}")
     print(f"优化迭代次数: {optimize_iterations}")
     print(f"Box合并阈值: {merge_threshold}")
     print("=" * 60)
 
-    # 步骤一：生成图片
     figure_path = output_dir / "figure.png"
-    generate_figure_from_method(
-        method_text=method_text,
-        output_path=str(figure_path),
-        api_key=api_key,
-        model=image_gen_model,
-        base_url=base_url,
-        provider=provider,
-    )
+    samed_path = str(output_dir / "samed.png")
+    boxlib_path = str(output_dir / "boxlib.json")
+
+    # 预先检查所需资源
+    _require_files_for_step(output_dir, start_from)
+
+    if start_from <= 1:
+        # 步骤一：生成图片
+        generate_figure_from_method(
+            method_text=method_text,
+            output_path=str(figure_path),
+            api_key=api_key,
+            model=image_gen_model,
+            base_url=base_url,
+            provider=provider,
+        )
 
     if stop_after == 1:
         print("\n" + "=" * 60)
@@ -2493,95 +2565,107 @@ def method_to_svg(
             "final_svg_path": None,
         }
 
-    # 步骤二：SAM3 分割（包含Box合并）
-    samed_path, boxlib_path, valid_boxes = segment_with_sam3(
-        image_path=str(figure_path),
-        output_dir=str(output_dir),
-        text_prompts=sam_prompts,
-        min_score=min_score,
-        merge_threshold=merge_threshold,
-        sam_backend=sam_backend_value,
-        sam_api_key=sam_api_key,
-        sam_max_masks=sam_max_masks,
-    )
-
-    if len(valid_boxes) == 0:
-        print("\n警告: 没有检测到有效的图标，流程终止")
-        return {
-            "figure_path": str(figure_path),
-            "samed_path": samed_path,
-            "boxlib_path": boxlib_path,
-            "icon_infos": [],
-            "template_svg_path": None,
-            "final_svg_path": None,
-        }
-
-    print(f"\n检测到 {len(valid_boxes)} 个图标")
-
-    if stop_after == 2:
-        print("\n" + "=" * 60)
-        print("已在步骤 2 后停止")
-        print("=" * 60)
-        return {
-            "figure_path": str(figure_path),
-            "samed_path": samed_path,
-            "boxlib_path": boxlib_path,
-            "icon_infos": [],
-            "template_svg_path": None,
-            "optimized_template_path": None,
-            "final_svg_path": None,
-        }
-
-    # 步骤三：裁切 + 去背景
-    icon_infos = crop_and_remove_background(
-        image_path=str(figure_path),
-        boxlib_path=boxlib_path,
-        output_dir=str(output_dir),
-        rmbg_model_path=rmbg_model_path,
-    )
-
-    if stop_after == 3:
-        print("\n" + "=" * 60)
-        print("已在步骤 3 后停止")
-        print("=" * 60)
-        return {
-            "figure_path": str(figure_path),
-            "samed_path": samed_path,
-            "boxlib_path": boxlib_path,
-            "icon_infos": icon_infos,
-            "template_svg_path": None,
-            "optimized_template_path": None,
-            "final_svg_path": None,
-        }
-
-    # 步骤四：生成 SVG 模板
-    template_svg_path = output_dir / "template.svg"
-    generate_svg_template(
-        figure_path=str(figure_path),
-        samed_path=samed_path,
-        boxlib_path=boxlib_path,
-        output_path=str(template_svg_path),
-        api_key=api_key,
-        model=svg_gen_model,
-        base_url=base_url,
-        provider=provider,
-        placeholder_mode=placeholder_mode,
-    )
+    if start_from <= 2:
+        # 步骤二：SAM3 分割（包含Box合并）
+        samed_path, boxlib_path, valid_boxes = segment_with_sam3(
+            image_path=str(figure_path),
+            output_dir=str(output_dir),
+            text_prompts=sam_prompts,
+            min_score=min_score,
+            merge_threshold=merge_threshold,
+            sam_backend=sam_backend_value,
+            sam_api_key=sam_api_key,
+            sam_max_masks=sam_max_masks,
+        )
+        if len(valid_boxes) == 0:
+            print("\n警告: 没有检测到有效的图标，流程终止")
+            return {
+                "figure_path": str(figure_path),
+                "samed_path": samed_path,
+                "boxlib_path": boxlib_path,
+                "icon_infos": [],
+                "template_svg_path": None,
+                "final_svg_path": None,
+            }
+        print(f"\n检测到 {len(valid_boxes)} 个图标")
+        if stop_after == 2:
+            print("\n" + "=" * 60)
+            print("已在步骤 2 后停止")
+            print("=" * 60)
+            return {
+                "figure_path": str(figure_path),
+                "samed_path": samed_path,
+                "boxlib_path": boxlib_path,
+                "icon_infos": [],
+                "template_svg_path": None,
+                "optimized_template_path": None,
+                "final_svg_path": None,
+            }
+
+    if start_from <= 3:
+        # 步骤三：裁切 + 去背景
+        icon_infos = crop_and_remove_background(
+            image_path=str(figure_path),
+            boxlib_path=boxlib_path,
+            output_dir=str(output_dir),
+            rmbg_model_path=rmbg_model_path,
+        )
+        if stop_after == 3:
+            print("\n" + "=" * 60)
+            print("已在步骤 3 后停止")
+            print("=" * 60)
+            return {
+                "figure_path": str(figure_path),
+                "samed_path": samed_path,
+                "boxlib_path": boxlib_path,
+                "icon_infos": icon_infos,
+                "template_svg_path": None,
+                "optimized_template_path": None,
+                "final_svg_path": None,
+            }
+
+    if start_from >= 4:
+        # 从步骤 4 恢复：使用已有 figure.png, samed.png, boxlib.json, icons/
+        samed_path = str(output_dir / "samed.png")
+        boxlib_path = str(output_dir / "boxlib.json")
+        icon_infos = _load_icon_infos_from_existing(output_dir, boxlib_path)
+        print(f"\n从步骤 {start_from} 恢复，已加载 {len(icon_infos)} 个图标")
+
+    if start_from <= 4:
+        # 步骤四：生成 SVG 模板
+        template_svg_path = output_dir / "template.svg"
+        generate_svg_template(
+            figure_path=str(figure_path),
+            samed_path=samed_path,
+            boxlib_path=boxlib_path,
+            output_path=str(template_svg_path),
+            api_key=api_key,
+            model=svg_gen_model,
+            base_url=base_url,
+            provider=provider,
+            placeholder_mode=placeholder_mode,
+        )
 
-    # 步骤 4.6：LLM 优化 SVG 模板（可配置迭代次数，0 表示跳过）
-    optimized_template_path = output_dir / "optimized_template.svg"
-    optimize_svg_with_llm(
-        figure_path=str(figure_path),
-        samed_path=samed_path,
-        final_svg_path=str(template_svg_path),
-        output_path=str(optimized_template_path),
-        api_key=api_key,
-        model=svg_gen_model,
-        base_url=base_url,
-        provider=provider,
-        max_iterations=optimize_iterations,
-        skip_base64_validation=True,
-    )
+        # 步骤 4.6：LLM 优化 SVG 模板（可配置迭代次数，0 表示跳过）
+        optimized_template_path = output_dir / "optimized_template.svg"
+        optimize_svg_with_llm(
+            figure_path=str(figure_path),
+            samed_path=samed_path,
+            final_svg_path=str(template_svg_path),
+            output_path=str(optimized_template_path),
+            api_key=api_key,
+            model=svg_gen_model,
+            base_url=base_url,
+            provider=provider,
+            max_iterations=optimize_iterations,
+            skip_base64_validation=True,
+        )
+    else:
+        # start_from >= 5：使用已有的 template / optimized_template
+        template_svg_path = output_dir / "template.svg"
+        optimized_template_path = output_dir / "optimized_template.svg"
+        if not optimized_template_path.exists():
+            optimized_template_path = template_svg_path
 
     if stop_after == 4:
         print("\n" + "=" * 60)
@@ -2728,6 +2812,13 @@ def method_to_svg(
         default=5,
         help="执行到指定步骤后停止（1-5，默认: 5 完整流程）"
     )
+    parser.add_argument(
+        "--start_from",
+        type=int,
+        choices=[1, 2, 3, 4, 5],
+        default=1,
+        help="从指定步骤开始（4 或 5 时跳过 1-3，使用 outputs 中已有的 figure.png/samed.png/boxlib.json/icons）"
+    )
 
     # 占位符模式参数
     parser.add_argument(
@@ -2787,6 +2878,7 @@ def method_to_svg(
         sam_max_masks=args.sam_max_masks,
         rmbg_model_path=args.rmbg_model_path,
         stop_after=args.stop_after,
+        start_from=args.start_from,
         placeholder_mode=args.placeholder_mode,
         optimize_iterations=args.optimize_iterations,
         merge_threshold=args.merge_threshold,

web/app.js

@@ -40,7 +40,6 @@
     }
 
     if (uploadZone && referenceFile) {
-      uploadZone.addEventListener("click", () => referenceFile.click());
       uploadZone.addEventListener("dragover", (event) => {
         event.preventDefault();
         uploadZone.classList.add("dragging");

web/index.html

@@ -67,7 +67,7 @@
           </div>
 
           <div class="field-group">
-            <label class="field-label" for="referenceFile">Reference Image</label>
+            <div class="field-label">Reference Image</div>
             <div class="upload-zone" id="uploadZone">
               <input id="referenceFile" class="file-input" type="file" accept="image/*" />
               <div class="upload-text">Drop image here or click to upload</div>