ai-engineering-init 1.4.3 → 1.5.0

package/src/skills/banana-image/scripts/banana_image_exec.ts

@@ -0,0 +1,412 @@
+#!/usr/bin/env npx tsx
+/**
+ * Banana Image - Gemini AI 图片生成执行脚本
+ * 直接调用 Gemini API，不依赖 MCP 服务
+ */
+
+import { GoogleGenAI, Part } from "@google/genai";
+import * as fs from "fs";
+import * as path from "path";
+import { ProxyAgent, setGlobalDispatcher } from "undici";
+
+// 配置代理支持
+const proxyUrl = process.env.HTTPS_PROXY || process.env.HTTP_PROXY;
+if (proxyUrl) {
+  const dispatcher = new ProxyAgent(proxyUrl);
+  setGlobalDispatcher(dispatcher);
+}
+
+// ============================================================================
+// 类型定义
+// ============================================================================
+
+interface BananaImageOptions {
+  prompt: string;
+  model?: "flash" | "pro";
+  aspectRatio?: string;
+  resolution?: string;
+  negativePrompt?: string;
+  outputDir?: string;
+  enableGrounding?: boolean;
+  inputImage?: string;
+  systemInstruction?: string;
+  count?: number;
+}
+
+interface GeneratedImage {
+  path: string;
+  width?: number;
+  height?: number;
+  model: string;
+  prompt: string;
+}
+
+interface BananaImageResult {
+  success: boolean;
+  images?: GeneratedImage[];
+  error?: string;
+  metadata?: {
+    model: string;
+    aspectRatio?: string;
+    resolution?: string;
+    duration: number;
+  };
+}
+
+// ============================================================================
+// 常量
+// ============================================================================
+
+const MODEL_MAP = {
+  flash: "gemini-2.5-flash-image",
+  pro: "gemini-3-pro-image-preview",
+} as const;
+
+const VALID_ASPECT_RATIOS = [
+  "1:1",
+  "2:3",
+  "3:2",
+  "3:4",
+  "4:3",
+  "4:5",
+  "5:4",
+  "9:16",
+  "16:9",
+  "21:9",
+];
+
+const VALID_RESOLUTIONS = ["1K", "2K", "4K"];
+
+const DEFAULT_OUTPUT_DIR = path.join(process.cwd(), "images");
+
+// ============================================================================
+// 工具函数
+// ============================================================================
+
+function generateFilename(prefix: string = "banana"): string {
+  const timestamp = new Date().toISOString().replace(/[:.]/g, "-");
+  const random = Math.random().toString(36).substring(2, 8);
+  return `${prefix}_${timestamp}_${random}.png`;
+}
+
+function ensureOutputDir(dir: string): void {
+  if (!fs.existsSync(dir)) {
+    fs.mkdirSync(dir, { recursive: true });
+  }
+}
+
+function loadImageAsBase64(imagePath: string): string {
+  const buffer = fs.readFileSync(imagePath);
+  return buffer.toString("base64");
+}
+
+function getMimeType(imagePath: string): string {
+  const ext = path.extname(imagePath).toLowerCase();
+  const mimeTypes: Record<string, string> = {
+    ".png": "image/png",
+    ".jpg": "image/jpeg",
+    ".jpeg": "image/jpeg",
+    ".gif": "image/gif",
+    ".webp": "image/webp",
+  };
+  return mimeTypes[ext] || "image/png";
+}
+
+// ============================================================================
+// 核心生成逻辑
+// ============================================================================
+
+async function generateImages(
+  options: BananaImageOptions,
+): Promise<BananaImageResult> {
+  const startTime = Date.now();
+
+  // 验证 API Key
+  const apiKey = process.env.GEMINI_API_KEY;
+  if (!apiKey) {
+    return {
+      success: false,
+      error: "GEMINI_API_KEY environment variable is not set",
+    };
+  }
+
+  // 参数处理
+  const modelTier = options.model || "pro";
+  const modelName = MODEL_MAP[modelTier];
+  const outputDir = options.outputDir || DEFAULT_OUTPUT_DIR;
+  const aspectRatio = options.aspectRatio || "1:1";
+  const resolution = options.resolution || (modelTier === "pro" ? "4K" : "1K");
+  const count = Math.min(options.count || 1, 4);
+
+  // 验证参数
+  if (
+    options.aspectRatio &&
+    !VALID_ASPECT_RATIOS.includes(options.aspectRatio)
+  ) {
+    return {
+      success: false,
+      error: `Invalid aspect ratio: ${options.aspectRatio}. Valid: ${VALID_ASPECT_RATIOS.join(", ")}`,
+    };
+  }
+
+  if (
+    options.resolution &&
+    !VALID_RESOLUTIONS.includes(options.resolution.toUpperCase())
+  ) {
+    return {
+      success: false,
+      error: `Invalid resolution: ${options.resolution}. Valid: ${VALID_RESOLUTIONS.join(", ")}`,
+    };
+  }
+
+  ensureOutputDir(outputDir);
+
+  try {
+    // 支持自定义 API 地址（如第三方代理）
+    const apiBase = process.env.GEMINI_API_BASE || process.env.GOOGLE_API_BASE;
+    const aiConfig: { apiKey: string; httpOptions?: { baseUrl: string } } = { apiKey };
+    if (apiBase) {
+      aiConfig.httpOptions = { baseUrl: apiBase };
+    }
+    const ai = new GoogleGenAI(aiConfig);
+
+    // 构建内容
+    const contents: (string | Part)[] = [];
+
+    // 添加系统指令
+    if (options.systemInstruction) {
+      contents.push(options.systemInstruction);
+    }
+
+    // 添加输入图片（编辑模式）
+    if (options.inputImage && fs.existsSync(options.inputImage)) {
+      const imageData = loadImageAsBase64(options.inputImage);
+      const mimeType = getMimeType(options.inputImage);
+      contents.push({
+        inlineData: {
+          data: imageData,
+          mimeType,
+        },
+      });
+    }
+
+    // 构建提示词
+    let fullPrompt = options.prompt;
+    if (options.negativePrompt) {
+      fullPrompt += `\n\nAvoid: ${options.negativePrompt}`;
+    }
+    contents.push(fullPrompt);
+
+    // 构建配置
+    // 注意：只有 Pro 模型支持 imageConfig 和 tools
+    const config: Record<string, unknown> = {};
+
+    if (modelTier === "pro") {
+      // Pro 模型支持宽高比和分辨率配置
+      config.imageConfig = {
+        aspectRatio,
+        imageSize: resolution.toUpperCase(),
+      };
+
+      // Google Search grounding
+      if (options.enableGrounding) {
+        config.tools = [{ googleSearch: {} }];
+      }
+    }
+    // Flash 模型不支持 imageConfig，使用默认配置
+
+    // 调用 API
+    const generatedImages: GeneratedImage[] = [];
+
+    for (let i = 0; i < count; i++) {
+      const response = await ai.models.generateContent({
+        model: modelName,
+        contents: contents.length === 1 ? contents[0] : contents,
+        config,
+      });
+
+      // 提取图片
+      const candidates = response.candidates || [];
+      for (const candidate of candidates) {
+        const parts = candidate.content?.parts || [];
+        for (const part of parts) {
+          if ("inlineData" in part && part.inlineData?.data) {
+            const filename = generateFilename(`banana_${modelTier}`);
+            const filePath = path.join(outputDir, filename);
+
+            // 保存图片
+            const buffer = Buffer.from(part.inlineData.data, "base64");
+            fs.writeFileSync(filePath, buffer);
+
+            generatedImages.push({
+              path: filePath,
+              model: modelName,
+              prompt: options.prompt,
+            });
+          }
+        }
+      }
+    }
+
+    if (generatedImages.length === 0) {
+      return {
+        success: false,
+        error: "No images were generated. The content may have been filtered.",
+      };
+    }
+
+    const duration = Date.now() - startTime;
+
+    return {
+      success: true,
+      images: generatedImages,
+      metadata: {
+        model: modelName,
+        aspectRatio,
+        resolution,
+        duration,
+      },
+    };
+  } catch (error) {
+    const errorMessage = error instanceof Error ? error.message : String(error);
+    return {
+      success: false,
+      error: `Image generation failed: ${errorMessage}`,
+    };
+  }
+}
+
+// ============================================================================
+// CLI 入口
+// ============================================================================
+
+function printUsage(): void {
+  console.log(`
+Banana Image - Gemini AI 图片生成
+
+Usage:
+  npx tsx banana_image_exec.ts [options]
+
+Options:
+  --prompt, -p <text>      图片描述（必需）
+  --model, -m <type>       模型选择: flash | pro (默认: pro)
+  --aspect-ratio, -a <r>   宽高比: 1:1, 16:9, 9:16, 4:3, 3:2, 21:9 等
+  --resolution, -r <res>   分辨率: 1K, 2K, 4K (默认: Pro=4K, Flash=1K)
+  --negative, -n <text>    排除内容描述
+  --output, -o <dir>       输出目录 (默认: ./images)
+  --grounding, -g          启用 Google Search 锚定 (仅 Pro)
+  --input, -i <path>       输入图片路径（编辑模式）
+  --system, -s <text>      系统指令
+  --count, -c <num>        生成数量 (1-4)
+  --help, -h               显示帮助
+
+Examples:
+  # 快速生成
+  npx tsx banana_image_exec.ts -p "可爱的柴犬" -m flash
+
+  # 4K 高质量
+  npx tsx banana_image_exec.ts -p "专业产品照片" -m pro -r 4K -a 4:5
+
+  # 编辑图片
+  npx tsx banana_image_exec.ts -p "将背景改为黄昏" -i /path/to/image.png
+`);
+}
+
+function parseArgs(): BananaImageOptions | null {
+  const args = process.argv.slice(2);
+
+  if (args.length === 0 || args.includes("--help") || args.includes("-h")) {
+    printUsage();
+    return null;
+  }
+
+  const options: BananaImageOptions = {
+    prompt: "",
+  };
+
+  for (let i = 0; i < args.length; i++) {
+    const arg = args[i];
+    const next = args[i + 1];
+
+    switch (arg) {
+      case "--prompt":
+      case "-p":
+        options.prompt = next || "";
+        i++;
+        break;
+      case "--model":
+      case "-m":
+        if (next === "flash" || next === "pro") {
+          options.model = next;
+        }
+        i++;
+        break;
+      case "--aspect-ratio":
+      case "-a":
+        options.aspectRatio = next;
+        i++;
+        break;
+      case "--resolution":
+      case "-r":
+        options.resolution = next;
+        i++;
+        break;
+      case "--negative":
+      case "-n":
+        options.negativePrompt = next;
+        i++;
+        break;
+      case "--output":
+      case "-o":
+        options.outputDir = next;
+        i++;
+        break;
+      case "--grounding":
+      case "-g":
+        options.enableGrounding = true;
+        break;
+      case "--input":
+      case "-i":
+        options.inputImage = next;
+        i++;
+        break;
+      case "--system":
+      case "-s":
+        options.systemInstruction = next;
+        i++;
+        break;
+      case "--count":
+      case "-c":
+        options.count = parseInt(next, 10) || 1;
+        i++;
+        break;
+    }
+  }
+
+  if (!options.prompt) {
+    console.error("Error: --prompt is required");
+    printUsage();
+    return null;
+  }
+
+  return options;
+}
+
+async function main(): Promise<void> {
+  const options = parseArgs();
+  if (!options) {
+    process.exit(1);
+  }
+
+  const result = await generateImages(options);
+  console.log(JSON.stringify(result, null, 2));
+
+  if (!result.success) {
+    process.exit(1);
+  }
+}
+
+main().catch((error) => {
+  console.error(JSON.stringify({ success: false, error: String(error) }));
+  process.exit(1);
+});

package/src/skills/banana-image/scripts/batch_prep.py

@@ -0,0 +1,82 @@
+#!/usr/bin/env python3
+"""
+批量图片生成准备脚本
+将文本文件（每行一个 prompt）或 CSV 转换为 JSON 格式的 prompt 列表
+"""
+
+import sys
+import json
+import csv
+import argparse
+from pathlib import Path
+
+
+def parse_text_file(filepath: Path) -> list[dict]:
+    """解析纯文本文件，每行一个 prompt"""
+    prompts = []
+    with open(filepath, "r", encoding="utf-8") as f:
+        for i, line in enumerate(f, 1):
+            line = line.strip()
+            if line and not line.startswith("#"):
+                prompts.append({"id": i, "prompt": line})
+    return prompts
+
+
+def parse_csv_file(filepath: Path, prompt_column: str = "prompt") -> list[dict]:
+    """解析 CSV 文件，提取指定列作为 prompt"""
+    prompts = []
+    with open(filepath, "r", encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for i, row in enumerate(reader, 1):
+            prompt = row.get(prompt_column) or row.get(list(row.keys())[0])
+            if prompt and prompt.strip():
+                entry = {"id": i, "prompt": prompt.strip()}
+                # 保留其他列作为参数
+                for key, value in row.items():
+                    if key != prompt_column and value:
+                        entry[key] = value
+                prompts.append(entry)
+    return prompts
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="将文本或 CSV 文件转换为批量生成的 JSON 格式"
+    )
+    parser.add_argument("file", help="输入文件路径")
+    parser.add_argument(
+        "--column", "-c", default="prompt", help="CSV 中 prompt 所在列名 (默认: prompt)"
+    )
+    parser.add_argument("--output", "-o", help="输出文件路径 (默认: 标准输出)")
+
+    args = parser.parse_args()
+    filepath = Path(args.file)
+
+    if not filepath.exists():
+        print(json.dumps({"error": f"文件不存在: {filepath}"}))
+        sys.exit(1)
+
+    try:
+        if filepath.suffix.lower() == ".csv":
+            prompts = parse_csv_file(filepath, args.column)
+        else:
+            prompts = parse_text_file(filepath)
+
+        result = {"source": str(filepath), "count": len(prompts), "prompts": prompts}
+
+        output = json.dumps(result, ensure_ascii=False, indent=2)
+
+        if args.output:
+            with open(args.output, "w", encoding="utf-8") as f:
+                f.write(output)
+            print(f"已保存到: {args.output}")
+        else:
+            print(output)
+
+    except Exception as e:
+        print(json.dumps({"error": str(e)}))
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()