From 56faf12d2525c63a7138419b2ec6be005b4e8aba Mon Sep 17 00:00:00 2001
From: hp0912 <809211365@qq.com>
Date: Sun, 5 Apr 2026 13:45:39 +0800
Subject: [PATCH] feat: image to image skill
---
skills/image-to-image/SKILL.md | 107 +++++
skills/image-to-image/scripts/bootstrap.py | 127 +++++
.../image-to-image/scripts/image_to_image.py | 442 ++++++++++++++++++
.../image-to-image/scripts/requirements.txt | 1 +
skills/text-to-image/SKILL.md | 11 +-
skills/text-to-image/scripts/bootstrap.py | 25 +
skills/text-to-image/scripts/text_to_image.py | 20 +-
7 files changed, 722 insertions(+), 11 deletions(-)
create mode 100644 skills/image-to-image/SKILL.md
create mode 100644 skills/image-to-image/scripts/bootstrap.py
create mode 100644 skills/image-to-image/scripts/image_to_image.py
create mode 100644 skills/image-to-image/scripts/requirements.txt
diff --git a/skills/image-to-image/SKILL.md b/skills/image-to-image/SKILL.md
new file mode 100644
index 0000000..a4d1245
--- /dev/null
+++ b/skills/image-to-image/SKILL.md
@@ -0,0 +1,107 @@
+---
+name: image-to-image
+description: "图片修改、图生图工具。基于输入的一张或多张图片，结合文本提示词生成新的图片。支持图片混合、风格转换、内容合成等多种创作模式。输入是文字+图片的组合，输出是图片。"
+argument-hint: "需要 prompt（提示词）和 images（图片链接列表），可选 model（模型）、negative_prompt（反向提示词）、ratio（宽高比）、resolution（分辨率）"
+---
+
+# Image To Image Skill
+
+## 描述
+
+这是一个 AI 图生图技能，基于输入的一张或多张图片，结合文本提示词生成新的图片。支持图片混合、风格转换、内容合成等多种创作模式。
+
+支持多个绘图模型：即梦（JiMeng）、豆包（DouBao）、造相（Z-Image）。
+
+从数据库中读取绘图配置（API 密钥、Base URL 等），根据用户选择的模型调用对应的绘图 API，返回生成的图片 URL。
+
+这个仓库里额外提供了一个可执行脚本 `image-to-image/scripts/image_to_image.py`，方便宿主机器人直接调用。
+
+## 触发条件
+
+- 用户想基于图片生成新图片
+- 用户说「把这张图变成……」「把图片修改成……」「风格转换」「图片合成」
+- 用户提到「图生图」「图片编辑」「图片修改」
+- 用户发送了一张或多张图片，并附带修改、合成、风格转换等描述
+
+## 参数说明（JSON Schema）
+
+调用脚本时，需要通过 shell 风格参数传入，参数结构如下：
+
+```json
+{
+ "type": "object",
+ "properties": {
+ "prompt": {
+ "type": "string",
+ "description": "根据用户输入的文本内容，提取出图片混合、风格转换、内容合成等等的提示词，但是不要对提示词进行修改。"
+ },
+ "model": {
+ "type": "string",
+ "description": "画图模型选择（可选）：即梦4.5(jimeng-4.5) / 即梦4.6(jimeng-4.6) / 即梦5.0(jimeng-5.0) / 豆包图生图(doubao-seededit-3.0-i2i) / 造相基础版(Z-Image) / 造相蒸馏版(Z-Image-Turbo) / 造相图片编辑(Qwen-Image-Edit-2511)，默认: 空(none)。",
+ "enum": [
+ "none",
+ "jimeng-4.5",
+ "jimeng-4.6",
+ "jimeng-5.0",
+ "doubao-seededit-3.0-i2i",
+ "Z-Image",
+ "Z-Image-Turbo",
+ "Qwen-Image-Edit-2511"
+ ],
+ "default": "none"
+ },
+ "images": {
+ "type": "array",
+ "items": { "type": "string" },
+ "description": "用于图片编辑、图片混合、风格转换、内容合成等的图片链接列表，至少需要一张图像。"
+ },
+ "negative_prompt": {
+ "type": "string",
+ "description": "用于描述图像中不希望出现的元素或特征的文本，可选。"
+ },
+ "ratio": {
+ "type": "string",
+ "description": "图像的宽高比，可选，默认16:9。",
+ "default": "16:9"
+ },
+ "resolution": {
+ "type": "string",
+ "description": "图像的分辨率，可选，默认2k。",
+ "default": "2k"
+ }
+ },
+ "required": ["prompt", "images"],
+ "additionalProperties": false
+}
+```
+
+对应的命令行参数为：
+
+- `--prompt <提示词>` 必填
+- `--images <图片链接>` 必填，可重复传入多张图片，如 `--images url1 --images url2`
+- `--model <模型名>` 可选
+- `--negative_prompt <反向提示词>` 可选
+- `--ratio <宽高比>` 可选
+- `--resolution <分辨率>` 可选
+
+## 依赖安装
+
+- 脚本首次运行时会自动创建虚拟环境并安装依赖，无需手动执行。
+- 如需手动重新安装，可执行：`python3 image-to-image/scripts/bootstrap.py`
+
+## 执行步骤
+
+1. 当用户发送图片并附带修改、合成、风格转换等描述时触发该技能。
+2. 从用户输入中提取 prompt（提示词），不对提示词做总结或修改。提取 images（图片链接列表）。可选提取 model、negative_prompt、ratio、resolution 参数。
+3. 将参数组装为 shell 风格命令行参数，在仓库根目录下执行本地脚本，例如：`python3 image-to-image/scripts/image_to_image.py --prompt '把这张图变成油画风格' --images 'https://example.com/img1.jpg' --images 'https://example.com/img2.jpg' --model jimeng-5.0`。
+4. 成功时脚本输出
+
+```
+图片URL1
+图片URL2
+```
+
+## 回复要求
+
+- 成功时，脚本输出 `图片URL1图片URL2` 格式，AI 智能体接收到这种格式内容会自动发送图片。
+- 失败时，返回具体的失败信息。
diff --git a/skills/image-to-image/scripts/bootstrap.py b/skills/image-to-image/scripts/bootstrap.py
new file mode 100644
index 0000000..7e6cd70
--- /dev/null
+++ b/skills/image-to-image/scripts/bootstrap.py
@@ -0,0 +1,127 @@
+#!/usr/bin/env python3
+
+from __future__ import annotations
+
+import hashlib
+import subprocess
+import sys
+import traceback
+from pathlib import Path
+
+sys.stderr = sys.stdout
+
+
+def _skill_root_from(script_dir: Path) -> Path:
+ return script_dir.parent
+
+
+def _venv_dir(script_dir: Path) -> Path:
+ return _skill_root_from(script_dir) / ".venv"
+
+
+def _venv_python(venv_dir: Path) -> Path:
+ if sys.platform == "win32":
+ return venv_dir / "Scripts" / "python.exe"
+ return venv_dir / "bin" / "python"
+
+
+def _stamp_file(venv_dir: Path) -> Path:
+ return venv_dir / ".req_hash"
+
+
+def _file_hash(path: Path) -> str:
+ return hashlib.sha256(path.read_bytes()).hexdigest()
+
+
+def _deps_up_to_date(requirements_file: Path, venv_dir: Path) -> bool:
+ stamp = _stamp_file(venv_dir)
+ if not stamp.is_file():
+ return False
+ return stamp.read_text().strip() == _file_hash(requirements_file)
+
+
+def _write_stamp(requirements_file: Path, venv_dir: Path) -> None:
+ _stamp_file(venv_dir).write_text(_file_hash(requirements_file))
+
+
+def _ensure_venv(venv_dir: Path, venv_python: Path) -> int:
+ if venv_python.is_file():
+ return 0
+
+ sys.stdout.write(f"未检测到技能虚拟环境，正在创建: {venv_dir}\n")
+ command = [
+ sys.executable,
+ "-m",
+ "venv",
+ str(venv_dir),
+ ]
+
+ try:
+ subprocess.run(command, check=True, stdout=sys.stdout, stderr=sys.stdout)
+ except subprocess.CalledProcessError as exc:
+ sys.stdout.write(f"创建虚拟环境失败，退出码: {exc.returncode}\n")
+ return exc.returncode or 1
+
+ return 0
+
+def main() -> int:
+ script_dir = Path(__file__).resolve().parent
+ requirements_file = script_dir / "requirements.txt"
+ venv_dir = _venv_dir(script_dir)
+ venv_python = _venv_python(venv_dir)
+
+ if not requirements_file.is_file():
+ sys.stdout.write(f"未找到依赖文件: {requirements_file}\n")
+ return 1
+
+ ensure_result = _ensure_venv(venv_dir, venv_python)
+ if ensure_result != 0:
+ return ensure_result
+
+ if _deps_up_to_date(requirements_file, venv_dir):
+ sys.stdout.write("依赖已是最新，跳过安装\n")
+ return 0
+
+ command = [
+ str(venv_python),
+ "-m",
+ "pip",
+ "install",
+ "--upgrade",
+ "pip",
+ ]
+
+ try:
+ subprocess.run(command, check=True, stdout=sys.stdout, stderr=sys.stdout)
+ except subprocess.CalledProcessError as exc:
+ sys.stdout.write(f"升级 pip 失败，退出码: {exc.returncode}\n")
+ return exc.returncode or 1
+
+ command = [
+ str(venv_python),
+ "-m",
+ "pip",
+ "install",
+ "-r",
+ str(requirements_file),
+ ]
+
+ try:
+ subprocess.run(command, check=True, stdout=sys.stdout, stderr=sys.stdout)
+ except subprocess.CalledProcessError as exc:
+ sys.stdout.write(f"安装依赖失败，退出码: {exc.returncode}\n")
+ return exc.returncode or 1
+
+ _write_stamp(requirements_file, venv_dir)
+ sys.stdout.write(f"依赖安装完成，当前技能虚拟环境: {venv_dir}\n")
+ return 0
+
+
+if __name__ == "__main__":
+ try:
+ raise SystemExit(main())
+ except SystemExit:
+ raise
+ except Exception:
+ traceback.print_exc(file=sys.stdout)
+ raise SystemExit(1)
diff --git a/skills/image-to-image/scripts/image_to_image.py b/skills/image-to-image/scripts/image_to_image.py
new file mode 100644
index 0000000..c472c37
--- /dev/null
+++ b/skills/image-to-image/scripts/image_to_image.py
@@ -0,0 +1,442 @@
+#!/usr/bin/env python3
+
+from __future__ import annotations
+
+import argparse
+import json
+import os
+import re
+import subprocess
+import sys
+import time
+import traceback
+import urllib.request
+from pathlib import Path
+
+# The skill runner consumes stdout, so route Python error output there as well.
+sys.stderr = sys.stdout
+
+
+def _skill_root() -> Path:
+ script_dir = Path(__file__).resolve().parent
+ return script_dir.parent
+
+
+def _skill_venv_python() -> Path:
+ venv_dir = _skill_root() / ".venv"
+ if sys.platform == "win32":
+ return venv_dir / "Scripts" / "python.exe"
+ return venv_dir / "bin" / "python"
+
+
+def _run_bootstrap() -> None:
+ bootstrap = Path(__file__).resolve().parent / "bootstrap.py"
+ result = subprocess.run([sys.executable, str(bootstrap)])
+ if result.returncode != 0:
+ raise SystemExit(result.returncode)
+
+
+def _ensure_skill_venv_python() -> None:
+ venv_python = _skill_venv_python()
+ if not venv_python.is_file():
+ _run_bootstrap()
+ venv_python = _skill_venv_python()
+ if not venv_python.is_file():
+ sys.stdout.write("bootstrap 后仍未找到虚拟环境\n")
+ raise SystemExit(1)
+
+ venv_dir = _skill_root() / ".venv"
+ if Path(sys.prefix) == venv_dir.resolve():
+ return
+
+ os.execv(str(venv_python), [str(venv_python), str(Path(__file__).resolve()), *sys.argv[1:]])
+
+
+_ensure_skill_venv_python()
+
+try:
+ import pymysql # type: ignore # noqa: E402
+except ModuleNotFoundError:
+ _run_bootstrap()
+ os.execv(sys.executable, [sys.executable, str(Path(__file__).resolve()), *sys.argv[1:]])
+
+
+# ---------------------------------------------------------------------------
+# Database helpers
+# ---------------------------------------------------------------------------
+
+def _mysql_connect():
+ host = os.environ.get("MYSQL_HOST", "127.0.0.1")
+ port = int(os.environ.get("MYSQL_PORT", "3306"))
+ user = os.environ.get("MYSQL_USER", "root")
+ password = os.environ.get("MYSQL_PASSWORD", "")
+ database = os.environ.get("ROBOT_CODE", "")
+ if not database:
+ raise RuntimeError("环境变量 ROBOT_CODE 未配置")
+
+ return pymysql.connect(
+ host=host, port=port, user=user, password=password,
+ database=database, charset="utf8mb4",
+ connect_timeout=10, read_timeout=30,
+ )
+
+
+def _query_one(conn, sql: str, params: tuple = ()) -> dict | None:
+ cur = conn.cursor()
+ cur.execute(sql, params)
+ columns = [desc[0] for desc in cur.description] if cur.description else []
+ row = cur.fetchone()
+ cur.close()
+ if row is None:
+ return None
+ return dict(zip(columns, row))
+
+
+# ---------------------------------------------------------------------------
+# Settings resolution (mirrors the Go service logic)
+# ---------------------------------------------------------------------------
+
+def load_drawing_settings(conn, from_wx_id: str) -> tuple[bool, dict]:
+ """Return (enabled, image_ai_settings_dict)."""
+ gs = _query_one(conn, "SELECT image_ai_enabled, image_ai_settings FROM global_settings LIMIT 1")
+ enabled = False
+ settings_json: dict = {}
+
+ if gs:
+ if gs.get("image_ai_enabled"):
+ enabled = bool(gs["image_ai_enabled"])
+ raw = gs.get("image_ai_settings")
+ if raw:
+ if isinstance(raw, (bytes, bytearray)):
+ raw = raw.decode("utf-8")
+ if isinstance(raw, str) and raw.strip():
+ settings_json = json.loads(raw)
+
+ if from_wx_id.endswith("@chatroom"):
+ override = _query_one(
+ conn,
+ "SELECT image_ai_enabled, image_ai_settings FROM chat_room_settings WHERE chat_room_id = %s LIMIT 1",
+ (from_wx_id,),
+ )
+ else:
+ override = _query_one(
+ conn,
+ "SELECT image_ai_enabled, image_ai_settings FROM friend_settings WHERE wechat_id = %s LIMIT 1",
+ (from_wx_id,),
+ )
+
+ if override:
+ if override.get("image_ai_enabled") is not None:
+ enabled = bool(override["image_ai_enabled"])
+ raw = override.get("image_ai_settings")
+ if raw:
+ if isinstance(raw, (bytes, bytearray)):
+ raw = raw.decode("utf-8")
+ if isinstance(raw, str) and raw.strip():
+ settings_json = json.loads(raw)
+
+ return enabled, settings_json
+
+
+# ---------------------------------------------------------------------------
+# API callers
+# ---------------------------------------------------------------------------
+
+def _http_post_json(url: str, body: dict, headers: dict, timeout: int = 300) -> dict:
+ data = json.dumps(body).encode("utf-8")
+ req = urllib.request.Request(url, data=data, headers=headers, method="POST")
+ with urllib.request.urlopen(req, timeout=timeout) as resp:
+ return json.loads(resp.read().decode("utf-8"))
+
+
+def _http_get_json(url: str, headers: dict, timeout: int = 30) -> dict:
+ req = urllib.request.Request(url, headers=headers, method="GET")
+ with urllib.request.urlopen(req, timeout=timeout) as resp:
+ return json.loads(resp.read().decode("utf-8"))
+
+
+def call_jimeng(config: dict, prompt: str, model: str, images: list[str],
+ negative_prompt: str, ratio: str, resolution: str) -> list[str]:
+ """Call JiMeng (即梦) image compositions API (图生图)."""
+ base_url = config.get("base_url", "").rstrip("/")
+ session_ids = config.get("sessionid", [])
+ if not base_url or not session_ids:
+ raise RuntimeError("即梦绘图配置缺少 base_url 或 sessionid")
+
+ if not model or model == "none":
+ model = "jimeng-5.0"
+
+ if not ratio:
+ ratio = "16:9"
+ if not resolution:
+ resolution = "2k"
+
+ # 如果分辨率大于4k，重置为2k
+ m = re.search(r"(\d+)", resolution)
+ if m and int(m.group(1)) > 4:
+ resolution = "2k"
+
+ token = ",".join(session_ids)
+ body = {
+ "model": model,
+ "prompt": prompt,
+ "images": images,
+ "ratio": ratio,
+ "resolution": resolution,
+ "response_format": "url",
+ "sample_strength": 0.5,
+ }
+ if negative_prompt:
+ body["negative_prompt"] = negative_prompt
+
+ # 图生图使用 /v1/images/compositions 端点
+ resp = _http_post_json(
+ f"{base_url}/v1/images/compositions",
+ body,
+ {"Content-Type": "application/json", "Authorization": f"Bearer {token}"},
+ timeout=300,
+ )
+ urls = [item["url"] for item in resp.get("data", []) if item.get("url")]
+ return urls
+
+
+def call_doubao(config: dict, prompt: str, model: str, image: str) -> list[str]:
+ """Call DouBao (豆包) image-to-image API."""
+ api_key = config.get("api_key", "")
+ if not api_key:
+ raise RuntimeError("豆包绘图配置缺少 api_key")
+
+ if not model or model == "none":
+ model = "doubao-seededit-3.0-i2i"
+
+ model_map = {
+ "doubao-seededit-3.0-i2i": "doubao-seededit-3-0-i2i-250628",
+ }
+ actual_model = model_map.get(model, model)
+
+ body = {
+ "model": actual_model,
+ "prompt": prompt,
+ "response_format": "url",
+ "size": config.get("size", "2K"),
+ "sequential_image_generation": config.get("sequential_image_generation", "auto"),
+ "watermark": config.get("watermark", False),
+ }
+ if image:
+ body["image"] = image
+
+ resp = _http_post_json(
+ "https://ark.cn-beijing.volces.com/api/v3/images/generations",
+ body,
+ {"Content-Type": "application/json", "Authorization": f"Bearer {api_key}"},
+ timeout=300,
+ )
+ urls = []
+ for item in resp.get("data", []):
+ url = item.get("url")
+ if url:
+ urls.append(url)
+ return urls
+
+
+def call_zimage(config: dict, prompt: str, model: str, images: list[str]) -> list[str]:
+ """Call Z-Image (造相) image generation API (async task-based)."""
+ base_url = config.get("base_url", "").rstrip("/")
+ api_key = config.get("api_key", "")
+ if not base_url or not api_key:
+ raise RuntimeError("造相绘图配置缺少 base_url 或 api_key")
+
+ if not model or model == "none":
+ model = "Qwen-Image-Edit-2511"
+
+ model_map = {
+ "Z-Image": "Tongyi-MAI/Z-Image",
+ "Z-Image-Turbo": "Tongyi-MAI/Z-Image-Turbo",
+ "Qwen-Image-Edit-2511": "Qwen/Qwen-Image-Edit-2511",
+ }
+ actual_model = model_map.get(model)
+ if actual_model is None:
+ raise RuntimeError(f"不支持的造相模型: {model}")
+
+ body = {
+ "model": actual_model,
+ "prompt": prompt,
+ "image_url": images,
+ }
+ headers = {
+ "Content-Type": "application/json",
+ "Authorization": f"Bearer {api_key}",
+ "X-ModelScope-Async-Mode": "true",
+ }
+
+ # Step 1: create task
+ resp = _http_post_json(f"{base_url}/v1/images/generations", body, headers, timeout=30)
+ task_id = resp.get("task_id", "")
+ if not task_id:
+ raise RuntimeError("造相接口未返回 task_id")
+
+ # Step 2: poll for result
+ poll_headers = {
+ "Content-Type": "application/json",
+ "Authorization": f"Bearer {api_key}",
+ "X-ModelScope-Task-Type": "image_generation",
+ }
+ deadline = time.time() + 15 * 60 # 15 minutes
+ while time.time() < deadline:
+ task_resp = _http_get_json(f"{base_url}/v1/tasks/{task_id}", poll_headers, timeout=30)
+ status = task_resp.get("task_status", "")
+ if status == "SUCCEED":
+ images_result = task_resp.get("output_images", [])
+ if images_result:
+ return images_result
+ raise RuntimeError("造相任务成功但未返回图片")
+ if status == "FAILED":
+ raise RuntimeError("造相绘图任务失败")
+ time.sleep(5)
+
+ raise RuntimeError("造相绘图任务超时")
+
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+
+JIMENG_MODELS = {"jimeng-4.5", "jimeng-4.6", "jimeng-5.0"}
+DOUBAO_MODELS = {"doubao-seededit-3.0-i2i"}
+ZIMAGE_MODELS = {"Z-Image", "Z-Image-Turbo", "Qwen-Image-Edit-2511"}
+
+
+def _parse_cli_params(argv: list[str]) -> dict:
+ parser = argparse.ArgumentParser(add_help=False)
+ parser.add_argument("--prompt", default="")
+ parser.add_argument("--images", action="append", default=[])
+ parser.add_argument("--model", default="")
+ parser.add_argument("--negative_prompt", default="")
+ parser.add_argument("--ratio", default="")
+ parser.add_argument("--resolution", default="")
+
+ namespace, unknown = parser.parse_known_args(argv)
+ if unknown:
+ raise ValueError(f"存在不支持的参数: {' '.join(unknown)}")
+
+ return {
+ "prompt": namespace.prompt,
+ "images": [img for img in namespace.images if img.strip()],
+ "model": namespace.model,
+ "negative_prompt": namespace.negative_prompt,
+ "ratio": namespace.ratio,
+ "resolution": namespace.resolution,
+ }
+
+
+def main() -> int:
+ if len(sys.argv) < 2:
+ sys.stdout.write("缺少输入参数\n")
+ return 1
+
+ try:
+ params = _parse_cli_params(sys.argv[1:])
+ except ValueError as exc:
+ sys.stdout.write(f"参数格式错误: {exc}\n")
+ return 1
+
+ prompt = params.get("prompt", "").strip()
+ if not prompt:
+ sys.stdout.write("缺少提示词\n")
+ return 1
+
+ images = params.get("images", [])
+ if not images:
+ sys.stdout.write("图片链接列表为空\n")
+ return 1
+
+ model = params.get("model", "").strip()
+ negative_prompt = params.get("negative_prompt", "").strip()
+ ratio = params.get("ratio", "").strip()
+ resolution = params.get("resolution", "").strip()
+
+ from_wx_id = os.environ.get("ROBOT_FROM_WX_ID", "").strip()
+ if not from_wx_id:
+ sys.stdout.write("环境变量 ROBOT_FROM_WX_ID 未配置\n")
+ return 1
+
+ # Connect to DB and load settings
+ try:
+ conn = _mysql_connect()
+ except Exception as exc:
+ sys.stdout.write(f"数据库连接失败: {exc}\n")
+ return 1
+
+ try:
+ enabled, settings_json = load_drawing_settings(conn, from_wx_id)
+ except Exception as exc:
+ sys.stdout.write(f"加载绘图配置失败: {exc}\n")
+ return 1
+ finally:
+ try:
+ conn.close()
+ except Exception:
+ pass
+
+ if not enabled:
+ sys.stdout.write("AI 绘图未开启\n")
+ return 0
+
+ # Default model
+ if not model or model == "none":
+ model = "jimeng-5.0"
+
+ # Route to correct API
+ try:
+ image_urls: list[str] = []
+
+ if model in JIMENG_MODELS:
+ jimeng_config = settings_json.get("JiMeng", {})
+ if not jimeng_config.get("enabled", False):
+ sys.stdout.write("即梦绘图未开启\n")
+ return 0
+ image_urls = call_jimeng(jimeng_config, prompt, model, images, negative_prompt, ratio, resolution)
+
+ elif model in DOUBAO_MODELS:
+ doubao_config = settings_json.get("DouBao", {})
+ if not doubao_config.get("enabled", False):
+ sys.stdout.write("豆包绘图未开启\n")
+ return 0
+ # 豆包图生图只支持单张图片
+ image_urls = call_doubao(doubao_config, prompt, model, images[0])
+
+ elif model in ZIMAGE_MODELS:
+ zimage_config = settings_json.get("Z-Image", {})
+ if not zimage_config.get("enabled", False):
+ sys.stdout.write("造相绘图未开启\n")
+ return 0
+ image_urls = call_zimage(zimage_config, prompt, model, images)
+
+ else:
+ sys.stdout.write("不支持的 AI 图像模型\n")
+ return 1
+
+ except Exception as exc:
+ sys.stdout.write(f"调用绘图接口失败: {exc}\n")
+ return 1
+
+ if not image_urls:
+ sys.stdout.write("未生成任何图像\n")
+ return 1
+
+ for url in image_urls:
+ if url:
+ sys.stdout.write(f"{url}")
+
+ sys.stdout.write("\n")
+ return 0
+
+
+if __name__ == "__main__":
+ try:
+ raise SystemExit(main())
+ except SystemExit:
+ raise
+ except Exception:
+ traceback.print_exc(file=sys.stdout)
+ raise SystemExit(1)
diff --git a/skills/image-to-image/scripts/requirements.txt b/skills/image-to-image/scripts/requirements.txt
new file mode 100644
index 0000000..e99b1f8
--- /dev/null
+++ b/skills/image-to-image/scripts/requirements.txt
@@ -0,0 +1 @@
+pymysql>=1.1,<2
diff --git a/skills/text-to-image/SKILL.md b/skills/text-to-image/SKILL.md
index b84fc84..6f4de23 100644
--- a/skills/text-to-image/SKILL.md
+++ b/skills/text-to-image/SKILL.md
@@ -81,16 +81,15 @@ argument-hint: "需要 prompt 参数（画图提示词），可选 model（模
## 依赖安装
-- 在执行 `text-to-image/scripts/text_to_image.py` 之前，必须先安装依赖。
-- 执行安装脚本：`python3 text-to-image/scripts/bootstrap.py`
+- 脚本首次运行时会自动创建虚拟环境并安装依赖，无需手动执行。
+- 如需手动重新安装，可执行：`python3 text-to-image/scripts/bootstrap.py`
## 执行步骤
-1. 当用户输入绘图相关内容时触发该技能。
+1. 当用户想通过文本描述生成图像时触发该技能。
2. 从用户输入中提取 prompt（画图提示词），不对提示词做总结或修改。可选提取 model、negative_prompt、ratio、resolution 参数。
-3. 在执行脚本前，先安装依赖：`python3 text-to-image/scripts/bootstrap.py`。
-4. 将参数组装为 shell 风格命令行参数，在仓库根目录下执行本地脚本，例如：`python3 text-to-image/scripts/text_to_image.py --prompt '一只小白兔' --model jimeng-5.0`。
-5. 成功是脚本输出
+3. 将参数组装为 shell 风格命令行参数，在仓库根目录下执行本地脚本，例如：`python3 text-to-image/scripts/text_to_image.py --prompt '一只可爱的猫咪在花园里玩耍' --model jimeng-5.0`。
+4. 成功是脚本输出
```
图片URL1
diff --git a/skills/text-to-image/scripts/bootstrap.py b/skills/text-to-image/scripts/bootstrap.py
index 357c37d..f090e58 100644
--- a/skills/text-to-image/scripts/bootstrap.py
+++ b/skills/text-to-image/scripts/bootstrap.py
@@ -2,6 +2,7 @@
from __future__ import annotations
+import hashlib
import subprocess
import sys
import traceback
@@ -24,6 +25,25 @@ def _venv_python(venv_dir: Path) -> Path:
return venv_dir / "bin" / "python"
+def _stamp_file(venv_dir: Path) -> Path:
+ return venv_dir / ".req_hash"
+
+
+def _file_hash(path: Path) -> str:
+ return hashlib.sha256(path.read_bytes()).hexdigest()
+
+
+def _deps_up_to_date(requirements_file: Path, venv_dir: Path) -> bool:
+ stamp = _stamp_file(venv_dir)
+ if not stamp.is_file():
+ return False
+ return stamp.read_text().strip() == _file_hash(requirements_file)
+
+
+def _write_stamp(requirements_file: Path, venv_dir: Path) -> None:
+ _stamp_file(venv_dir).write_text(_file_hash(requirements_file))
+
+
def _ensure_venv(venv_dir: Path, venv_python: Path) -> int:
if venv_python.is_file():
return 0
@@ -58,6 +78,10 @@ def main() -> int:
if ensure_result != 0:
return ensure_result
+ if _deps_up_to_date(requirements_file, venv_dir):
+ sys.stdout.write("依赖已是最新，跳过安装\n")
+ return 0
+
command = [
str(venv_python),
"-m",
@@ -88,6 +112,7 @@ def main() -> int:
sys.stdout.write(f"安装依赖失败，退出码: {exc.returncode}\n")
return exc.returncode or 1
+ _write_stamp(requirements_file, venv_dir)
sys.stdout.write(f"依赖安装完成，当前技能虚拟环境: {venv_dir}\n")
return 0
diff --git a/skills/text-to-image/scripts/text_to_image.py b/skills/text-to-image/scripts/text_to_image.py
index ea5f53f..b298ab0 100644
--- a/skills/text-to-image/scripts/text_to_image.py
+++ b/skills/text-to-image/scripts/text_to_image.py
@@ -6,6 +6,7 @@ import argparse
import json
import os
import re
+import subprocess
import sys
import time
import traceback
@@ -28,10 +29,21 @@ def _skill_venv_python() -> Path:
return venv_dir / "bin" / "python"
+def _run_bootstrap() -> None:
+ bootstrap = Path(__file__).resolve().parent / "bootstrap.py"
+ result = subprocess.run([sys.executable, str(bootstrap)])
+ if result.returncode != 0:
+ raise SystemExit(result.returncode)
+
+
def _ensure_skill_venv_python() -> None:
venv_python = _skill_venv_python()
if not venv_python.is_file():
- return
+ _run_bootstrap()
+ venv_python = _skill_venv_python()
+ if not venv_python.is_file():
+ sys.stdout.write("bootstrap 后仍未找到虚拟环境\n")
+ raise SystemExit(1)
venv_dir = _skill_root() / ".venv"
if Path(sys.prefix) == venv_dir.resolve():
@@ -45,10 +57,8 @@ _ensure_skill_venv_python()
try:
import pymysql # type: ignore # noqa: E402
except ModuleNotFoundError:
- sys.stdout.write(
- "缺少依赖 pymysql，请先执行 python3 text-to-image/scripts/bootstrap.py 安装当前 skill 的依赖\n"
- )
- raise SystemExit(1)
+ _run_bootstrap()
+ os.execv(sys.executable, [sys.executable, str(Path(__file__).resolve()), *sys.argv[1:]])
# ---------------------------------------------------------------------------