free-short-video 5.1.2

package/bin/cli.js

@@ -0,0 +1,191 @@
+#!/usr/bin/env node
+'use strict';
+
+/*
+ * free-short-video — npm launcher
+ *
+ * Installs/runs the full Agnes Video Generator (Python/FastAPI) service on the
+ * user's machine with zero manual setup beyond a Python 3.10+ interpreter:
+ *   1. detect python3 >= 3.10
+ *   2. create a venv inside the package
+ *   3. pip install -r requirements.txt  (pulls moviepy + imageio-ffmpeg)
+ *   4. expose imageio-ffmpeg's static ffmpeg binary on PATH (no system ffmpeg needed)
+ *   5. spawn `python server.py` with PORT/HOST + optional AGNES_API_KEY
+ *   6. auto-open the browser, forward Ctrl+C for graceful shutdown
+ */
+
+const { spawn, execSync } = require('child_process');
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+
+const PKG_ROOT = path.resolve(__dirname, '..');
+
+function printHelp() {
+  console.log(`free-short-video — free AI short-video generator
+
+Usage:
+  npx free-short-video [options]
+  fsv [options]
+
+Options:
+  --port <n>     Listen port (default 8765)
+  --host <h>     Bind host (default 127.0.0.1; use 0.0.0.0 for LAN)
+  --no-open      Do not auto-open the browser
+  -h, --help     Show this help
+
+Environment:
+  AGNES_API_KEY  Your Agnes API key (can also be set later in the Web UI)
+
+First run creates a local venv and installs Python dependencies automatically.
+`);
+}
+
+// ── parse args ────────────────────────────────────────────────────────────
+let port = 8765;
+let host = '127.0.0.1';
+let openBrowser = true;
+
+const argv = process.argv.slice(2);
+for (let i = 0; i < argv.length; i++) {
+  const a = argv[i];
+  if (a === '--port') {
+    const v = parseInt(argv[++i], 10);
+    if (!Number.isNaN(v)) port = v;
+  } else if (a === '--host') {
+    host = argv[++i] || host;
+  } else if (a === '--no-open') {
+    openBrowser = false;
+  } else if (a === '-h' || a === '--help') {
+    printHelp();
+    process.exit(0);
+  } else {
+    console.error(`Unknown option: ${a}\n`);
+    printHelp();
+    process.exit(1);
+  }
+}
+
+// ── helpers ───────────────────────────────────────────────────────────────
+function run(cmd, opts) {
+  return execSync(cmd, Object.assign({ stdio: 'inherit', cwd: PKG_ROOT }, opts));
+}
+
+function findPython() {
+  for (const cmd of ['python3', 'python']) {
+    try {
+      const out = execSync(`"${cmd}" --version`, { stdio: ['ignore', 'pipe', 'ignore'] })
+        .toString()
+        .trim();
+      const m = out.match(/Python (\d+)\.(\d+)/);
+      if (m && (parseInt(m[1], 10) > 3 || (parseInt(m[1], 10) === 3 && parseInt(m[2], 10) >= 10))) {
+        return cmd;
+      }
+    } catch (_) {
+      /* not found / wrong version */
+    }
+  }
+  return null;
+}
+
+function venvBin(name) {
+  return process.platform === 'win32'
+    ? path.join(PKG_ROOT, '.venv', 'Scripts', `${name}.exe`)
+    : path.join(PKG_ROOT, '.venv', 'bin', name);
+}
+
+// ── 1. python check ───────────────────────────────────────────────────────
+console.log('================================================');
+console.log('  free-short-video');
+console.log('================================================');
+console.log('');
+
+const py = findPython();
+if (!py) {
+  console.error('❌ Python 3.10+ is required but was not found.');
+  console.error('   macOS:   brew install python3');
+  console.error('   Ubuntu:  sudo apt install python3 python3-venv');
+  console.error('   Windows: https://www.python.org/downloads/');
+  process.exit(1);
+}
+console.log(`✓ Using ${py}`);
+
+const venvPython = venvBin('python');
+const venvPip = venvBin('pip');
+
+// ── 2. venv + deps ─────────────────────────────────────────────────────────
+if (!fs.existsSync(venvPython)) {
+  console.log('[1/3] Creating virtual environment...');
+  run(`"${py}" -m venv "${path.join(PKG_ROOT, '.venv')}"`);
+}
+
+console.log('[2/3] Installing Python dependencies (first run only)...');
+run(`"${venvPip}" install -q -r "${path.join(PKG_ROOT, 'requirements.txt')}"`, {
+  env: process.env,
+});
+
+// ── 3. ffmpeg via imageio-ffmpeg (no system ffmpeg required) ──────────────
+let ffmpegDir = null;
+try {
+  const exe = execSync(`"${venvPython}" -c "import imageio_ffmpeg;print(imageio_ffmpeg.get_ffmpeg_exe())"`, {
+    stdio: ['ignore', 'pipe', 'ignore'],
+  })
+    .toString()
+    .trim();
+  if (exe) ffmpegDir = path.dirname(exe);
+} catch (_) {
+  /* imageio-ffmpeg missing is non-fatal; system ffmpeg may still work */
+}
+if (ffmpegDir) {
+  console.log('✓ Using bundled ffmpeg (imageio-ffmpeg)');
+} else {
+  console.log('⚠ No bundled ffmpeg found; relying on system ffmpeg in PATH');
+}
+
+// ── 4. launch server ──────────────────────────────────────────────────────
+const env = Object.assign({}, process.env, {
+  HOST: host,
+  PORT: String(port),
+});
+if (ffmpegDir) {
+  env.PATH = ffmpegDir + path.delimiter + (env.PATH || '');
+}
+if (process.env.AGNES_API_KEY) {
+  env.AGNES_API_KEY = process.env.AGNES_API_KEY;
+}
+
+console.log('[3/3] Starting service...');
+console.log('');
+console.log(`  Web UI will be at http://${host === '0.0.0.0' ? '127.0.0.1' : host}:${port}`);
+console.log('  Press Ctrl+C to stop.');
+console.log('');
+
+const child = spawn(venvPython, ['server.py'], {
+  cwd: PKG_ROOT,
+  env,
+  stdio: 'inherit',
+});
+
+child.on('exit', (code) => {
+  process.exit(code === null ? 0 : code);
+});
+
+const shutdown = (sig) => {
+  if (child.exitCode === null) child.kill(sig);
+};
+process.on('SIGINT', () => shutdown('SIGINT'));
+process.on('SIGTERM', () => shutdown('SIGTERM'));
+
+// ── auto-open browser ─────────────────────────────────────────────────────
+if (openBrowser) {
+  const url = `http://${host === '0.0.0.0' ? '127.0.0.1' : host}:${port}`;
+  setTimeout(() => {
+    try {
+      if (process.platform === 'darwin') execSync(`open "${url}"`);
+      else if (process.platform === 'win32') execSync(`start "" "${url}"`);
+      else execSync(`xdg-open "${url}"`);
+    } catch (_) {
+      /* browser open is best-effort */
+    }
+  }, 1500);
+}

package/core/__init__.py

@@ -0,0 +1,35 @@
+"""core — Agnes Video Generator v2.0 核心模块
+
+导出所有子包的核心类和工具函数。
+"""
+
+from core.api import AgnesImageAPI, AgnesVideoAPI, AgnesChatAPI
+from core.audio import EdgeTTSEngine, SilentTTSEngine, SubtitleGenerator
+from core.compositor import VideoConcatenator, VideoProcessor
+from core.pipelines import (
+    BasePipeline,
+    PipelineShutdown,
+    SimpleVideoPipeline,
+    CreativeVideoPipeline,
+    ManuscriptVideoPipeline,
+)
+
+__all__ = [
+    # API 层
+    "AgnesImageAPI",
+    "AgnesVideoAPI",
+    "AgnesChatAPI",
+    # 音频层
+    "EdgeTTSEngine",
+    "SilentTTSEngine",
+    "SubtitleGenerator",
+    # 拼接层
+    "VideoConcatenator",
+    "VideoProcessor",
+    # 流水线层
+    "BasePipeline",
+    "PipelineShutdown",
+    "SimpleVideoPipeline",
+    "CreativeVideoPipeline",
+    "ManuscriptVideoPipeline",
+]

package/core/api/__init__.py

@@ -0,0 +1,7 @@
+"""core.api — Agnes AI API 调用层"""
+
+from core.api.agnes_image import AgnesImageAPI, ImageOutput
+from core.api.agnes_video import AgnesVideoAPI, VideoOutput
+from core.api.agnes_chat import AgnesChatAPI
+
+__all__ = ["AgnesImageAPI", "ImageOutput", "AgnesVideoAPI", "VideoOutput", "AgnesChatAPI"]

package/core/api/agnes_chat.py

@@ -0,0 +1,294 @@
+"""core.api.agnes_chat — Agnes Chat API 封装（从 core/screenwriter.py 提取）
+
+P5: 健壮 JSON 解析（strip_code_fence + 正则提取 + 降级重试）
+P11: chat/chat_multimodal 统一重试（5xx/超时/连接错 3 次指数退避，4xx 不重试）
+"""
+
+import base64
+import json
+import logging
+import mimetypes
+import os
+import re
+import time
+from typing import List
+
+import requests
+
+from core.api.error_collector import collect_error, collect_error_from_exception
+from core.api.rate_limiter import get_rate_limiter
+
+logger = logging.getLogger(__name__)
+
+BASE_URL = "https://apihub.agnes-ai.com/v1"
+
+# 重试配置
+_MAX_RETRIES = 3
+_RETRY_BASE_DELAY = 15  # 秒，指数退避基数
+
+# 正则：匹配首个 {…} 块（支持嵌套大括号）
+_JSON_BLOCK_RE = re.compile(r"\{[\s\S]*\}")
+
+
+def strip_code_fence(text: str) -> str:
+    """去除 LLM 响应中的代码围栏标记。
+
+    处理常见变体：```json ... ```、``` ... ```、前后有多余空白/换行。
+
+    Args:
+        text: LLM 原始响应文本。
+
+    Returns:
+        去除围栏后的文本。
+    """
+    text = text.strip()
+    # 去除首行 ``` 或 ```json 等
+    if text.startswith("```"):
+        first_newline = text.index("\n") if "\n" in text else len(text)
+        text = text[first_newline + 1:]
+    # 去除尾部 ```
+    if text.endswith("```"):
+        text = text[:-3]
+    return text.strip()
+
+
+class AgnesChatAPI:
+    """Agnes LLM Chat API 封装（text + multimodal）。"""
+
+    def __init__(self, api_key: str, model: str = "agnes-2.0-flash"):
+        self.api_key = api_key
+        self.model = model
+        self.headers = {
+            "Authorization": f"Bearer {api_key}",
+            "Content-Type": "application/json",
+        }
+
+    def _image_to_b64_uri(self, path: str) -> str:
+        with open(path, "rb") as f:
+            b64 = base64.b64encode(f.read()).decode("utf-8")
+        mime = mimetypes.guess_type(path)[0] or "image/png"
+        return f"data:{mime};base64,{b64}"
+
+    @staticmethod
+    def _should_retry(resp: requests.Response) -> bool:
+        """判断 HTTP 响应是否应重试（5xx 和 429 重试，4xx 不重试）。"""
+        return resp.status_code >= 500 or resp.status_code == 429
+
+    @staticmethod
+    def _extract_prompt_from_payload(payload: dict) -> str:
+        """从 Chat payload 中提取 user prompt 文本（用于错误收集）。"""
+        messages = payload.get("messages", [])
+        for msg in reversed(messages):
+            content = msg.get("content", "")
+            if isinstance(content, list):
+                # 多模态：提取 text 部分
+                texts = [item.get("text", "") for item in content if isinstance(item, dict) and item.get("type") == "text"]
+                if texts:
+                    return texts[0]
+            elif isinstance(content, str) and content.strip():
+                return content
+        return ""
+
+    def _request_with_retry(self, payload: dict, timeout: int = 120) -> dict:
+        """带重试的 API 请求。
+
+        对 5xx/429/超时/连接错误进行最多 3 次指数退避重试。
+        4xx 错误（非 429）直接抛出不重试。
+        每次请求前通过全局限速器控制调用频率。
+
+        Args:
+            payload: 请求 JSON body。
+            timeout: 请求超时秒数。
+
+        Returns:
+            解析后的响应 JSON dict。
+
+        Raises:
+            requests.HTTPError: 4xx 客户端错误或重试耗尽。
+        """
+        last_exc = None
+        for attempt in range(_MAX_RETRIES):
+            try:
+                get_rate_limiter().acquire()
+                resp = requests.post(
+                    f"{BASE_URL}/chat/completions",
+                    headers=self.headers,
+                    json=payload,
+                    timeout=timeout,
+                )
+                if self._should_retry(resp) and attempt < _MAX_RETRIES - 1:
+                    delay = _RETRY_BASE_DELAY * (attempt + 1)
+                    logger.warning(
+                        f"[AgnesChat] Server error {resp.status_code}, "
+                        f"retry {attempt + 1}/{_MAX_RETRIES} in {delay}s..."
+                    )
+                    collect_error(
+                        "chat", "chat",
+                        prompt=self._extract_prompt_from_payload(payload),
+                        error_type=f"HTTP{resp.status_code}",
+                        error_message=f"HTTP {resp.status_code}: server error",
+                        status_code=resp.status_code,
+                        response_body=resp.text,
+                        retry_count=attempt + 1,
+                    )
+                    time.sleep(delay)
+                    continue
+                resp.raise_for_status()
+                return resp.json()
+            except (requests.ConnectionError, requests.Timeout) as e:
+                last_exc = e
+                # 每次失败都记录（包括中间重试）
+                collect_error_from_exception(
+                    "chat", "chat",
+                    exc=e, prompt=self._extract_prompt_from_payload(payload),
+                    retry_count=attempt + 1,
+                )
+                if attempt < _MAX_RETRIES - 1:
+                    delay = _RETRY_BASE_DELAY * (attempt + 1)
+                    logger.warning(
+                        f"[AgnesChat] {type(e).__name__}, "
+                        f"retry {attempt + 1}/{_MAX_RETRIES} in {delay}s..."
+                    )
+                    time.sleep(delay)
+                    continue
+                raise
+        # 重试耗尽
+        if last_exc:
+            collect_error_from_exception(
+                "chat", "chat",
+                exc=last_exc, prompt=self._extract_prompt_from_payload(payload),
+                retry_count=_MAX_RETRIES,
+            )
+            raise last_exc
+        # 不可重试的 HTTP 错误（4xx 非 429）
+        try:
+            resp.raise_for_status()  # type: ignore[possibly-undefined]
+        except requests.HTTPError as e:
+            collect_error(
+                "chat", "chat",
+                prompt=self._extract_prompt_from_payload(payload),
+                error_type=type(e).__name__,
+                error_message=str(e),
+                status_code=resp.status_code,  # type: ignore[possibly-undefined]
+                response_body=resp.text[:5000],  # type: ignore[possibly-undefined]
+                retry_count=_MAX_RETRIES,
+            )
+            raise
+        return resp.json()  # type: ignore[possibly-undefined]
+
+    def chat(self, system_prompt: str, user_prompt: str, max_tokens: int = 4096) -> str:
+        """纯文本 Chat 调用（含重试）。"""
+        logger.info(f"[AgnesChat] Calling chat ({self.model}), prompt: {len(user_prompt)} chars...")
+        data = self._request_with_retry(
+            {
+                "model": self.model,
+                "messages": [
+                    {"role": "system", "content": system_prompt},
+                    {"role": "user", "content": user_prompt},
+                ],
+                "temperature": 0.7,
+                "max_tokens": max_tokens,
+            },
+            timeout=120,
+        )
+        return data["choices"][0]["message"]["content"]
+
+    def chat_json(self, system_prompt: str, user_prompt: str, max_tokens: int = 4096) -> dict:
+        """Chat 调用并解析 JSON 响应（健壮版）。
+
+        处理流程：
+        1. 调用 chat 获取文本
+        2. strip_code_fence 去除围栏
+        3. 尝试直接 json.loads
+        4. 失败则用正则提取首个 {…} 块
+        5. 仍失败则重试一次 chat 调用
+        6. 最终失败抛出 ValueError
+
+        Args:
+            system_prompt: System 提示词。
+            user_prompt: User 提示词。
+            max_tokens: 最大生成 token 数。
+
+        Returns:
+            解析后的 JSON dict。
+
+        Raises:
+            ValueError: JSON 解析最终失败。
+        """
+        for retry in range(2):
+            content = self.chat(system_prompt, user_prompt, max_tokens=max_tokens)
+            # Step 1: 去围栏
+            cleaned = strip_code_fence(content)
+            # Step 2: 直接解析
+            try:
+                return json.loads(cleaned)
+            except (json.JSONDecodeError, ValueError):
+                pass
+            # Step 3: 正则提取首个 {…} 块
+            match = _JSON_BLOCK_RE.search(cleaned)
+            if match:
+                try:
+                    return json.loads(match.group())
+                except (json.JSONDecodeError, ValueError):
+                    pass
+            # Step 4: 首轮失败则重试一次 chat 调用
+            if retry == 0:
+                logger.warning("[AgnesChat] JSON parse failed, retrying chat call...")
+                continue
+            # 最终失败
+            preview = content[:200]
+            error_msg = (
+                f"[AgnesChat] Failed to parse JSON after 2 attempts. "
+                f"Response preview: {preview}..."
+            )
+            collect_error(
+                "chat", "chat_json",
+                prompt=user_prompt, system_prompt=system_prompt,
+                error_type="JSONParseError",
+                error_message=error_msg,
+                response_body=content[:5000],
+                retry_count=2,
+            )
+            raise ValueError(error_msg)
+        # 不应到达此处，但保险起见
+        raise ValueError("[AgnesChat] Unexpected flow in chat_json")
+
+    def chat_multimodal(
+        self,
+        system_prompt: str,
+        text_prompt: str,
+        image_paths: List[str],
+        max_tokens: int = 4096,
+    ) -> str:
+        """多模态 Chat 调用（文本 + 图片，含重试）。"""
+        messages = [{"role": "system", "content": system_prompt}]
+
+        user_content = [{"type": "text", "text": text_prompt}]
+        for img_path in image_paths:
+            if img_path.startswith(("http://", "https://")):
+                user_content.append({
+                    "type": "image_url",
+                    "image_url": {"url": img_path},
+                })
+            elif os.path.exists(img_path):
+                b64_uri = self._image_to_b64_uri(img_path)
+                user_content.append({
+                    "type": "image_url",
+                    "image_url": {"url": b64_uri},
+                })
+        messages.append({"role": "user", "content": user_content})
+
+        logger.info(
+            f"[AgnesChat] Calling multimodal ({self.model}), "
+            f"{len(image_paths)} image(s), prompt: {len(text_prompt)} chars..."
+        )
+        data = self._request_with_retry(
+            {
+                "model": self.model,
+                "messages": messages,
+                "temperature": 0.7,
+                "max_tokens": max_tokens,
+            },
+            timeout=300,
+        )
+        return data["choices"][0]["message"]["content"]