czon 0.5.0 → 0.5.2

package/dist/ai/extractMetadataFromMarkdown.js

@@ -1,13 +1,59 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.extractMetadataFromMarkdown = extractMetadataFromMarkdown;
+const metadata_1 = require("../metadata");
 const openai_1 = require("../services/openai");
 /**
- * 从 markdown 内容中提取 metadata
- * @returns Promise<AIMetadata> 提取的元数据，失败时抛出错误
+ * AI Metadata 提取模块
+ *
+ * 优化策略说明：
+ * 1. 从 MetaData 全局状态读取已有 slug，不作为参数传递
+ * 2. 如果已有 slug，条件化 prompt - 完全不提及 slug 相关指令
+ * 3. trade-off: 优先提升 AI 质量（减少无关任务干扰），可能降低 context 缓存命中率
  */
 async function extractMetadataFromMarkdown(filePath, content) {
-    const prompt = buildMetadataPrompt(content);
+    const existingSlug = metadata_1.MetaData.files.find(f => f.path === filePath)?.metadata?.slug;
+    const hasExistingSlug = !!existingSlug;
+    const fields = [
+        'title: 文档的标题（简洁明了，不超过 30 个字）',
+        'tags: 关键词列表（3-8 个关键词，使用中文或英文）',
+        'description: 文档的简短描述，微摘要（用一句话概括本文核心价值，不超过 100 字符），用于 SEO meta description，社交卡片短描述',
+        'summary: 文档中型摘要（用一段话总结文章，包含关键论点和结论，控制在 300 字以内），用于 邮件推送内容，newsletter 介绍',
+        'inferred_date: 文档中隐含的创建日期（如果有的话，格式：YYYY-MM-DD，没有就留空字符串）',
+        'inferred_lang: 文档使用的语言代码（例如：zh-Hans 表示简体中文，en-US 表示美式英语）',
+        'key_points: 文章的关键要点列表（5-10 个要点，简洁明了）',
+        'audience: 目标读者描述（简短描述，50 字以内）',
+        'short_summary: 文档的超短摘要（用 2-3 句话概括文章主要内容，突出核心观点），用于文章列表页摘要，RSS feed 描述',
+        ...(hasExistingSlug
+            ? []
+            : ['slug: URL 友好别名（使用小写字母、数字和连字符，仅包含英文和数字）']),
+    ];
+    const jsonFields = [
+        '{',
+        '  "title": "文档标题",',
+        '  "description": "用一句话概括本文核心价值，不超过 100 字符",',
+        '  "summary": "中型摘要，用一段话总结文章，包含关键论点和结论",',
+        '  "short_summary": "超短摘要，用 2-3 句话概括文章主要内容，突出核心观点",',
+        '  "tags": ["关键词1", "关键词2", "关键词3"],',
+        '  "inferred_date": "2023-01-01",',
+        '  "inferred_lang": "zh-Hans",',
+        '  "key_points": ["要点1", "要点2", "要点3"],',
+        ...(hasExistingSlug ? [] : ['  "slug": "URL 友好别名",']),
+        '  "audience": "目标读者描述"',
+        '}',
+    ];
+    const prompt = `请分析以下文档内容，提取以下信息并返回 JSON 格式：
+
+ 文档内容：
+ """
+ ${content}
+ """
+
+ 请提取：
+${fields.join('\n')}
+
+ 请严格按照以下 JSON 格式返回，不要包含任何其他文本：
+${jsonFields.join('\n')}`;
     const messages = [
         {
             role: 'system',
@@ -22,85 +68,28 @@ async function extractMetadataFromMarkdown(filePath, content) {
         response_format: { type: 'json_object' },
         task_id: `extract-metadata:${filePath}`,
     });
-    const metadata = parseMetadataResponse(response.choices[0].message.content);
-    // 添加 tokens 使用情况
-    metadata.tokens_used = {
-        prompt: response.usage.prompt_tokens,
-        completion: response.usage.completion_tokens,
-        total: response.usage.total_tokens,
+    const metadata = JSON.parse(response.choices[0].message.content);
+    const result = {
+        title: metadata.title?.trim() || '',
+        description: metadata.description?.trim() || '',
+        short_summary: metadata.short_summary?.trim() || '',
+        audience: metadata.audience?.trim() || '',
+        key_points: Array.isArray(metadata.key_points)
+            ? metadata.key_points.map((point) => point.trim()).filter(Boolean)
+            : [],
+        summary: metadata.summary?.trim() || '',
+        slug: metadata.slug?.trim() || existingSlug || '',
+        tags: Array.isArray(metadata.tags)
+            ? metadata.tags.map((tag) => tag.trim()).filter(Boolean)
+            : [],
+        inferred_date: metadata.inferred_date?.trim() || undefined,
+        inferred_lang: metadata.inferred_lang?.trim() || undefined,
+        tokens_used: {
+            prompt: response.usage.prompt_tokens,
+            completion: response.usage.completion_tokens,
+            total: response.usage.total_tokens,
+        },
     };
-    return metadata;
-}
-/**
- * 构建提取 metadata 的 prompt
- */
-function buildMetadataPrompt(content) {
-    // 限制内容长度以避免 token 超限
-    const maxContentLength = Infinity; // 可根据需要调整长度限制
-    const truncatedContent = content.length > maxContentLength
-        ? content.substring(0, maxContentLength) + '... [内容已截断]'
-        : content;
-    return `请分析以下文档内容，提取以下信息并返回 JSON 格式：
-
-文档内容：
-"""
-${truncatedContent}
-"""
-
-请提取：
-1. title: 文档的标题（简洁明了，不超过 30 个字）
-2. slug: URL 友好别名（使用小写字母、数字和连字符，仅包含英文和数字）
-3. tags: 关键词列表（3-8 个关键词，使用中文或英文）
-4. description: 文档的简短描述，微摘要（用一句话概括本文核心价值，不超过 100 字符），用于 SEO meta description，社交卡片短描述
-5. summary: 文档中型摘要（用一段话总结文章，包含关键论点和结论，控制在 300 字以内），用于 邮件推送内容，newsletter 介绍
-6. inferred_date: 文档中隐含的创建日期（如果有的话，格式：YYYY-MM-DD，没有就留空字符串）
-7. inferred_lang: 文档使用的语言代码（例如：zh-Hans 表示简体中文，en-US 表示美式英语）
-8. key_points: 文章的关键要点列表（5-10 个要点，简洁明了）
-9. audience: 目标读者描述（简短描述，50 字以内）
-10. short_summary: 文档的超短摘要（用 2-3 句话概括文章主要内容，突出核心观点），用于文章列表页摘要，RSS feed 描述
-
-
-请严格按照以下 JSON 格式返回，不要包含任何其他文本：
-{
-  "title": "文档标题",
-  "description": "用一句话概括本文核心价值，不超过 100 字符",
-  "summary": "中型摘要，用一段话总结文章，包含关键论点和结论",
-  "short_summary": "超短摘要，用 2-3 句话概括文章主要内容，突出核心观点",
-  "slug": "URL 友好别名",
-  "tags": ["关键词1", "关键词2", "关键词3"],
-  "inferred_date": "2023-01-01",
-  "inferred_lang": "zh-Hans",
-  "key_points": ["要点1", "要点2", "要点3"],
-  "audience": "目标读者描述"
-}`;
-}
-/**
- * 解析 AI 返回的 metadata
- */
-function parseMetadataResponse(responseContent) {
-    try {
-        const metadata = JSON.parse(responseContent);
-        // 验证和清理数据
-        return {
-            title: metadata.title?.trim() || '未命名文档',
-            description: metadata.description?.trim() || '',
-            short_summary: metadata.short_summary?.trim() || '',
-            audience: metadata.audience?.trim() || '',
-            key_points: Array.isArray(metadata.key_points)
-                ? metadata.key_points.map((point) => point.trim()).filter(Boolean)
-                : [],
-            summary: metadata.summary?.trim() || '',
-            slug: metadata.slug?.trim() || '',
-            tags: Array.isArray(metadata.tags)
-                ? metadata.tags.map((tag) => tag.trim()).filter(Boolean)
-                : [],
-            inferred_date: metadata.inferred_date?.trim() || undefined,
-            inferred_lang: metadata.inferred_lang?.trim() || 'zh-Hans',
-        };
-    }
-    catch (error) {
-        console.error('❌ Failed to parse AI response:', error, 'Response:', responseContent);
-        throw error;
-    }
+    return result;
 }
 //# sourceMappingURL=extractMetadataFromMarkdown.js.map

package/dist/build/pipeline.js

@@ -102,6 +102,7 @@ async function buildPipeline(options) {
     await (0, processTranslations_1.processTranslations)();
     // 渲染模板
     await (0, template_1.spiderStaticSiteGenerator)();
+    await (0, template_1.downloadCDNResources)();
     // 生成 robots.txt
     await (0, robots_1.generateRobotsTxt)();
     // 生成 sitemap.xml

package/dist/process/processTranslations.js

@@ -36,6 +36,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.processTranslations = processTranslations;
 const promises_1 = require("fs/promises");
 const path_1 = __importStar(require("path"));
+const translateMarkdown_1 = require("../ai/translateMarkdown");
 const languages_1 = require("../languages");
 const metadata_1 = require("../metadata");
 const paths_1 = require("../paths");
@@ -94,9 +95,11 @@ async function processTranslations() {
                         console.info(`ℹ️ Content unchanged for ${file.path}, skipping translation.`);
                     return;
                 }
-                const translatedContent = await translateWithOpenCode(sourcePath, targetPath, lang);
+                const translatedResponse = await (0, translateMarkdown_1.translateMarkdown)(sourcePath, content, lang);
+                const translatedContent = translatedResponse.choices?.[0].message.content?.trim() || '';
                 const translationMeta = ((_a = (file.translations ?? (file.translations = {})))[lang] ?? (_a[lang] = {}));
-                translationMeta.content_length = translatedContent.length;
+                translationMeta.content_length = translatedContent.length; // 记录翻译后内容长度
+                translationMeta.token_used = translatedResponse.usage; // 记录 token 使用情况
                 await (0, writeFile_1.writeFile)(targetPath, translatedContent);
                 // 存储已增强内容的哈希值
                 file.nativeMarkdownHash = hash;

package/dist/process/template.js

@@ -33,13 +33,14 @@ var __importStar = (this && this.__importStar) || (function () {
     };
 })();
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.spiderStaticSiteGenerator = void 0;
+exports.downloadCDNResources = exports.spiderStaticSiteGenerator = void 0;
 const fs = __importStar(require("fs/promises"));
 const path = __importStar(require("path"));
+const sitemap_1 = require("../build/sitemap");
 const metadata_1 = require("../metadata");
 const paths_1 = require("../paths");
-const sitemap_1 = require("../build/sitemap");
 const ssg_1 = require("../ssg");
+const resourceMap_1 = require("../ssg/resourceMap");
 const convertMarkdownToHtml_1 = require("../utils/convertMarkdownToHtml");
 const frontmatter_1 = require("../utils/frontmatter");
 const writeFile_1 = require("../utils/writeFile");
@@ -151,4 +152,19 @@ const spiderStaticSiteGenerator = async () => {
     }
 };
 exports.spiderStaticSiteGenerator = spiderStaticSiteGenerator;
+const downloadCDNResources = async () => {
+    for (const resource of resourceMap_1.EXTERNAL_RESOURCES) {
+        const targetFilePath = path.join(paths_1.CZON_DIST_DIR, 'assets', resource.name);
+        console.info(`⬇️ Downloading resource: ${resource.url} -> ${targetFilePath}`);
+        const response = await fetch(resource.url);
+        if (!response.ok) {
+            console.error(`❌ Failed to download resource: ${resource.url}, status: ${response.status}`);
+            throw new Error(`Failed to download resource: ${resource.url}`);
+        }
+        const buffer = await response.arrayBuffer();
+        await (0, writeFile_1.writeFile)(targetFilePath, Buffer.from(buffer));
+        console.info(`✅ Resource downloaded: ${targetFilePath}`);
+    }
+};
+exports.downloadCDNResources = downloadCDNResources;
 //# sourceMappingURL=template.js.map

package/dist/ssg/ContentPage.js

@@ -30,7 +30,7 @@ const ContentPage = props => {
             react_1.default.createElement("meta", { name: "viewport", content: "width=device-width, initial-scale=1.0" }),
             react_1.default.createElement("title", null, title),
             react_1.default.createElement("meta", { name: "description", content: `tags: ${tags.join(', ')}` }),
-            react_1.default.createElement("script", { src: "https://cdn.jsdelivr.net/npm/@tailwindcss/browser@4", defer: true }),
+            react_1.default.createElement("script", { src: (0, node_path_1.relative)(props.ctx.path, '/assets/tailwindcss.js'), defer: true }),
             react_1.default.createElement("style", null, style_1.style),
             react_1.default.createElement("script", { dangerouslySetInnerHTML: {
                     __html: `

package/dist/ssg/IndexPage.js

@@ -4,6 +4,7 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 };
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.IndexPage = void 0;
+const node_path_1 = require("node:path");
 const react_1 = __importDefault(require("react"));
 const sortBy_1 = require("../utils/sortBy");
 const ContentMeta_1 = require("./components/ContentMeta");
@@ -59,7 +60,7 @@ const IndexPage = props => {
                     react_1.default.createElement("footer", null,
                         react_1.default.createElement(LanguageSwitcher_1.LanguageSwitcher, { ctx: props.ctx, lang: props.lang }),
                         react_1.default.createElement(CZONFooter_1.CZONFooter, null))), footer: null }),
-            react_1.default.createElement("script", { src: "https://cdn.jsdelivr.net/npm/@tailwindcss/browser@4", defer: true }))));
+            react_1.default.createElement("script", { src: (0, node_path_1.relative)(props.ctx.path, '/assets/tailwindcss.js'), defer: true }))));
     // TODO: 渲染多语言首页列表
     // return (
     //   <div>

package/dist/ssg/resourceMap.js

@@ -0,0 +1,10 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.EXTERNAL_RESOURCES = void 0;
+exports.EXTERNAL_RESOURCES = [
+    {
+        name: 'tailwindcss.js',
+        url: 'https://cdn.jsdelivr.net/npm/@tailwindcss/browser@4',
+    },
+];
+//# sourceMappingURL=resourceMap.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "czon",
-  "version": "0.5.0",
+  "version": "0.5.2",
   "description": "CZone - AI enhanced Markdown content engine",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",