Open Source

scrapper/4 kenney_data_to_local.py

@@ -0,0 +1,68 @@
+import os
+import json
+from urllib.parse import urlparse, unquote
+
+# === 配置路径 ===
+input_json = "kenney_data.json"
+output_json = "kenney_data_local.json"
+
+zip_root = "kenney_assets"
+img_root = "kenney_assets_images"
+
+# === 工具函数 ===
+
+
+def sanitize_filename(name):
+    return "".join(c for c in name if c.isalnum() or c in "._- ()").strip()
+
+
+def build_zip_path(entry):
+    title = entry["title"]
+    version = entry["changelog"][0]["version"] if entry["changelog"] else "1.0"
+    category = entry["properties"].get("Category", ["Uncategorized"])[0]
+    series = entry["properties"].get("Series", [None])[0]
+
+    folder = os.path.join(zip_root, sanitize_filename(category))
+    if series:
+        folder = os.path.join(folder, sanitize_filename(series))
+    filename = f"{sanitize_filename(title)} V{version}.zip"
+    return os.path.join(folder, filename)
+
+
+def build_image_paths(entry):
+    title = entry["title"]
+    category = entry["properties"].get("Category", ["Uncategorized"])[0]
+    series = entry["properties"].get("Series", [None])[0]
+    images = entry.get("images", [])
+
+    folder = os.path.join(img_root, sanitize_filename(category))
+    if series:
+        folder = os.path.join(folder, sanitize_filename(series))
+    folder = os.path.join(folder, sanitize_filename(title))
+
+    local_paths = []
+    for img_url in images:
+        parsed = urlparse(img_url)
+        filename = unquote(os.path.basename(parsed.path))
+        local_paths.append(os.path.join(folder, filename))
+
+    return local_paths
+
+
+# === 主处理 ===
+with open(input_json, "r", encoding="utf-8") as f:
+    data = json.load(f)
+
+for entry in data:
+    if "download" in entry and entry["download"].endswith(".zip"):
+        zip_path = build_zip_path(entry)
+        if os.path.exists(zip_path):
+            entry["download"] = zip_path  # 替换为本地路径
+    if "images" in entry and isinstance(entry["images"], list):
+        entry["images"] = build_image_paths(entry)
+
+# === 保存修改后的 JSON ===
+with open(output_json, "w", encoding="utf-8") as f:
+    json.dump(data, f, ensure_ascii=False, indent=2)
+
+print("✅ 已更新 JSON：本地路径写入完毕！")