vault-fetch 0.1.0 → 0.2.0

package/README.md

@@ -5,27 +5,31 @@ Obsidian Clipper では取得できない、JavaScript レンダリングや認
 ## 特徴
 
 - Playwright (Chromium) による JS レンダリング後のページ取得
-- Readability.js による記事本文の抽出（広告・ナビゲーション除去）
+- Readability.js による記事本文の抽出（広告・ナビゲーション除去）、`--raw` モードでフルページ変換も可能
+- リソースブロッキング（画像・フォント・メディア）による高速フェッチ
+- Chrome User-Agent 偽装によるボット対策回避
 - Obsidian Clipper 互換のフロントマター（title, source, author, published, created, description, tags）
 - セッション管理（`storageState`）によるログイン済みページの取得
 - 設定の 3 層解決（CLI オプション > 環境変数 > 設定ファイル）
 
-## セットアップ
+## インストール
 
 ```bash
-npm install
+# グローバルインストール
+npm install -g vault-fetch
+
+# Playwright のブラウザも必要
 npx playwright install chromium
 ```
 
-グローバルインストール:
+## 使い方
+
+`npx` でインストールなしでも実行できます:
 
 ```bash
-npm install -g vault-fetch
-npx playwright install chromium
+npx vault-fetch fetch https://example.com/article --dry-run --dest /tmp
 ```
 
-## 使い方
-
 ### ページの取得・保存
 
 ```bash
@@ -43,6 +47,12 @@ vault-fetch fetch https://example.com/article --dest ~/Documents/Obsidian/Clippi
 
 # タグを追加
 vault-fetch fetch https://example.com/article --dest ~/Documents/Obsidian/Clippings --tag tech --tag ai
+
+# 非記事ページをフルページ変換（Readability をスキップ）
+vault-fetch fetch https://example.com/table-page --dest ~/Documents/Obsidian/Clippings --raw
+
+# 画像を含めてフェッチ（デフォルトではブロック）
+vault-fetch fetch https://example.com/article --dest ~/Documents/Obsidian/Clippings --no-block-images
 ```
 
 ### ログイン（セッション保存）
@@ -68,6 +78,10 @@ vault-fetch login https://note.com
 | `--wait-until <event>` | 待機条件: `load` / `domcontentloaded` / `networkidle`（デフォルト: `networkidle`） |
 | `--skip-session` | 保存済みセッションを使わない |
 | `--dry-run` | 保存せず標準出力に出力 |
+| `--raw` | Readability をスキップし、フルページ HTML を直接変換 |
+| `--no-block-images` | 画像リクエストのブロックを無効化 |
+| `--no-block-fonts` | フォントリクエストのブロックを無効化 |
+| `--no-block-media` | メディアリクエストのブロックを無効化 |
 
 ## 設定
 

package/dist/cli.js

@@ -2,6 +2,7 @@
 
 // src/cli.ts
 import { Command } from "commander";
+import { once } from "events";
 import { existsSync as existsSync2 } from "fs";
 import { homedir as homedir3 } from "os";
 import { join as join3 } from "path";
@@ -20,13 +21,40 @@ function expandTilde(filePath) {
   }
   return filePath;
 }
+var VALID_WAIT_UNTIL = ["load", "domcontentloaded", "networkidle"];
+function validateWaitUntil(value) {
+  if (!VALID_WAIT_UNTIL.includes(value)) {
+    throw new Error(
+      `Invalid waitUntil value: "${value}". Must be one of: ${VALID_WAIT_UNTIL.join(", ")}`
+    );
+  }
+  return value;
+}
 function loadConfigFile(configPath) {
   const content = readFileSync(configPath, "utf-8");
   const parsed = yaml.load(content);
   if (parsed === null || typeof parsed !== "object") {
     throw new Error(`Invalid config file: ${configPath}`);
   }
-  return parsed;
+  const config = parsed;
+  if (config.timeout !== void 0 && typeof config.timeout !== "number") {
+    throw new Error(`Invalid timeout in config file: expected number, got ${typeof config.timeout}`);
+  }
+  if (config.dest !== void 0 && typeof config.dest !== "string") {
+    throw new Error(`Invalid dest in config file: expected string, got ${typeof config.dest}`);
+  }
+  if (config.waitUntil !== void 0) {
+    if (typeof config.waitUntil !== "string") {
+      throw new Error(`Invalid waitUntil in config file: expected string, got ${typeof config.waitUntil}`);
+    }
+    validateWaitUntil(config.waitUntil);
+  }
+  if (config.tags !== void 0) {
+    if (!Array.isArray(config.tags) || !config.tags.every((t) => typeof t === "string")) {
+      throw new Error("Invalid tags in config file: expected array of strings");
+    }
+  }
+  return config;
 }
 function resolveConfig(cliOptions, configPath) {
   let fileConfig = {};
@@ -53,7 +81,8 @@ function resolveConfig(cliOptions, configPath) {
   } else {
     timeout = fileConfig.timeout ?? DEFAULT_TIMEOUT;
   }
-  const waitUntil = cliOptions.waitUntil ?? fileConfig.waitUntil ?? DEFAULT_WAIT_UNTIL;
+  const rawWaitUntil = cliOptions.waitUntil ?? fileConfig.waitUntil ?? DEFAULT_WAIT_UNTIL;
+  const waitUntil = validateWaitUntil(rawWaitUntil);
   const allTags = [
     ...fileConfig.tags ?? [],
     ...cliOptions.tags ?? [],
@@ -68,7 +97,11 @@ function resolveConfig(cliOptions, configPath) {
     headed: cliOptions.headed ?? false,
     selector: cliOptions.selector ?? null,
     noSession: cliOptions.noSession ?? false,
-    dryRun: cliOptions.dryRun ?? false
+    dryRun: cliOptions.dryRun ?? false,
+    blockImages: cliOptions.blockImages ?? true,
+    blockFonts: cliOptions.blockFonts ?? true,
+    blockMedia: cliOptions.blockMedia ?? true,
+    raw: cliOptions.raw ?? false
   };
 }
 
@@ -103,18 +136,38 @@ function ensureSessionDir(sessionsDir) {
 }
 
 // src/fetcher.ts
+var CHROME_USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36";
+function buildBlockedResourceTypes(options) {
+  const blocked = /* @__PURE__ */ new Set();
+  if (options.blockImages) blocked.add("image");
+  if (options.blockFonts) blocked.add("font");
+  if (options.blockMedia) blocked.add("media");
+  return blocked;
+}
 async function fetchPage(url, config, sessionsDir) {
   const browser = await chromium.launch({
     headless: !config.headed
   });
   try {
-    const contextOptions = {};
+    const contextOptions = {
+      userAgent: CHROME_USER_AGENT
+    };
     if (!config.noSession && sessionExists(url, sessionsDir)) {
       const sessionPath = getSessionPath(url, sessionsDir);
       contextOptions.storageState = sessionPath;
     }
     const context = await browser.newContext(contextOptions);
     const page = await context.newPage();
+    const blockedTypes = buildBlockedResourceTypes(config);
+    if (blockedTypes.size > 0) {
+      await page.route("**/*", async (route) => {
+        if (blockedTypes.has(route.request().resourceType())) {
+          await route.abort();
+        } else {
+          await route.continue();
+        }
+      });
+    }
     const timeoutMs = config.timeout * 1e3;
     const response = await page.goto(url, {
       waitUntil: config.waitUntil,
@@ -187,54 +240,48 @@ function extractAuthors(doc, readabilityByline) {
   }
   return [];
 }
+function buildMetadata(doc, article, finalUrl) {
+  const title = article?.title ?? doc.title;
+  const authors = extractAuthors(doc, article?.byline ?? null);
+  const published = extractPublishedDate(doc);
+  const description = getMetaContent(doc, 'meta[property="og:description"]') ?? getMetaContent(doc, 'meta[name="description"]') ?? (article?.excerpt ?? null);
+  const today = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
+  return {
+    title,
+    source: finalUrl,
+    author: authors,
+    published,
+    created: today,
+    description
+  };
+}
+function parseWithReadability(html, url) {
+  const dom = new JSDOM(html, { url });
+  const reader = new Readability(dom.window.document);
+  return reader.parse();
+}
 function extract(html, finalUrl) {
   const metaDom = new JSDOM(html, { url: finalUrl });
   const doc = metaDom.window.document;
-  const readabilityDom = new JSDOM(html, { url: finalUrl });
-  const reader = new Readability(readabilityDom.window.document);
-  const article = reader.parse();
+  const article = parseWithReadability(html, finalUrl);
   if (!article) {
-    throw new Error("Readability failed to extract content from the page");
+    throw new Error(
+      "Readability failed to extract content from the page. Try --raw to convert the full page, or --selector <css> to target specific content."
+    );
   }
   if (!article.content) {
     throw new Error("Readability returned empty content for the page");
   }
-  const title = article.title ?? doc.title;
-  const authors = extractAuthors(doc, article.byline ?? null);
-  const published = extractPublishedDate(doc);
-  const description = getMetaContent(doc, 'meta[property="og:description"]') ?? getMetaContent(doc, 'meta[name="description"]') ?? (article.excerpt ?? null);
-  const today = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
   return {
-    metadata: {
-      title,
-      source: finalUrl,
-      author: authors,
-      published,
-      created: today,
-      description
-    },
+    metadata: buildMetadata(doc, article, finalUrl),
     content: article.content
   };
 }
 function extractMetadata(html, finalUrl) {
   const metaDom = new JSDOM(html, { url: finalUrl });
   const doc = metaDom.window.document;
-  const readabilityDom = new JSDOM(html, { url: finalUrl });
-  const reader = new Readability(readabilityDom.window.document);
-  const article = reader.parse();
-  const title = article?.title ?? doc.title;
-  const authors = extractAuthors(doc, article?.byline ?? null);
-  const published = extractPublishedDate(doc);
-  const description = getMetaContent(doc, 'meta[property="og:description"]') ?? getMetaContent(doc, 'meta[name="description"]') ?? (article?.excerpt ?? null);
-  const today = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
-  return {
-    title,
-    source: finalUrl,
-    author: authors,
-    published,
-    created: today,
-    description
-  };
+  const article = parseWithReadability(html, finalUrl);
+  return buildMetadata(doc, article, finalUrl);
 }
 
 // src/converter.ts
@@ -253,9 +300,10 @@ import { writeFileSync } from "fs";
 import { join as join2 } from "path";
 import yaml2 from "js-yaml";
 var UNSAFE_CHARS = /[/\\:*?"<>|]/g;
+var CONTROL_CHARS = /[\x00-\x1f\x7f]/g;
 var MAX_FILENAME_LENGTH = 200;
 function sanitizeFilename(title) {
-  const sanitized = title.replace(UNSAFE_CHARS, "").replace(/\s+/g, " ").trim();
+  const sanitized = title.replace(CONTROL_CHARS, "").replace(UNSAFE_CHARS, "").replace(/\s+/g, " ").trim();
   const base = sanitized.slice(0, MAX_FILENAME_LENGTH) || "Untitled";
   return `${base}.md`;
 }
@@ -301,14 +349,14 @@ var CONFIG_PATH = join3(homedir3(), ".config", "vault-fetch", "config.yaml");
 var program = new Command();
 program.name("vault-fetch").description(
   "Fetch JS-rendered web pages and save as Markdown to Obsidian Vault"
-).version("0.1.0");
+).version("0.2.0");
 program.command("fetch").description("Fetch a page and save as Markdown").argument("<url>", "URL to fetch").option("--dest <path>", "Destination directory").option("--headed", "Run browser in headed mode").option("--selector <css>", "CSS selector to extract").option("--timeout <seconds>", "Timeout in seconds", parseInt).option("--tag <name>", "Add tag (repeatable)", (val, acc) => {
   acc.push(val);
   return acc;
 }, []).option(
   "--wait-until <event>",
   "Wait condition: load, domcontentloaded, networkidle"
-).option("--skip-session", "Do not use saved session").option("--dry-run", "Output to stdout instead of saving").action(async (url, options) => {
+).option("--skip-session", "Do not use saved session").option("--dry-run", "Output to stdout instead of saving").option("--no-block-images", "Do not block image requests").option("--no-block-fonts", "Do not block font requests").option("--no-block-media", "Do not block media requests").option("--raw", "Convert full page HTML without Readability extraction").action(async (url, options) => {
   try {
     const configPath = existsSync2(CONFIG_PATH) ? CONFIG_PATH : void 0;
     const config = resolveConfig(
@@ -320,10 +368,17 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
         headed: options.headed,
         selector: options.selector,
         noSession: options.skipSession,
-        dryRun: options.dryRun
+        dryRun: options.dryRun,
+        blockImages: options.blockImages,
+        blockFonts: options.blockFonts,
+        blockMedia: options.blockMedia,
+        raw: options.raw
       },
       configPath
     );
+    if (config.raw && config.selector) {
+      throw new Error("--raw and --selector cannot be used together.");
+    }
     if (!config.dryRun && !existsSync2(config.dest)) {
       throw new Error(`Destination directory does not exist: ${config.dest}`);
     }
@@ -334,6 +389,9 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
     if (config.selector) {
       contentHtml = fetchResult.html;
       metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);
+    } else if (config.raw) {
+      contentHtml = fetchResult.fullHtml;
+      metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);
     } else {
       const result = extract(fetchResult.html, fetchResult.finalUrl);
       metadata = result.metadata;
@@ -372,11 +430,10 @@ program.command("login").description("Login to a site and save session").argumen
     const page = await context.newPage();
     await page.goto(url, { waitUntil: "networkidle", timeout: timeoutSec * 1e3 });
     console.error("Browser opened. Log in manually, then press Enter here to save session.");
-    await new Promise((resolve2) => {
-      process.stdin.once("data", () => {
-        resolve2();
-      });
-    });
+    process.stdin.resume();
+    await once(process.stdin, "data");
+    process.stdin.pause();
+    process.stdin.unref();
     const sessionPath = getSessionPath(url, sessionsDir);
     await context.storageState({ path: sessionPath });
     console.error(`Session saved: ${sessionPath}`);

package/dist/cli.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/cli.ts","../src/config.ts","../src/fetcher.ts","../src/session.ts","../src/extractor.ts","../src/converter.ts","../src/writer.ts"],"sourcesContent":["import { Command } from \"commander\";\nimport { existsSync } from \"node:fs\";\nimport { homedir } from \"node:os\";\nimport { join } from \"node:path\";\nimport { resolveConfig } from \"./config.js\";\nimport { fetchPage } from \"./fetcher.js\";\nimport { extract, extractMetadata } from \"./extractor.js\";\nimport { convertToMarkdown } from \"./converter.js\";\nimport { writeMarkdownFile, buildFrontmatter } from \"./writer.js\";\nimport {\n  getSessionDir,\n  getSessionPath,\n  ensureSessionDir,\n} from \"./session.js\";\nimport type { WaitUntilOption } from \"./types.js\";\n\nconst CONFIG_PATH = join(homedir(), \".config\", \"vault-fetch\", \"config.yaml\");\n\nconst program = new Command();\n\nprogram\n  .name(\"vault-fetch\")\n  .description(\n    \"Fetch JS-rendered web pages and save as Markdown to Obsidian Vault\",\n  )\n  .version(\"0.1.0\");\n\nprogram\n  .command(\"fetch\")\n  .description(\"Fetch a page and save as Markdown\")\n  .argument(\"<url>\", \"URL to fetch\")\n  .option(\"--dest <path>\", \"Destination directory\")\n  .option(\"--headed\", \"Run browser in headed mode\")\n  .option(\"--selector <css>\", \"CSS selector to extract\")\n  .option(\"--timeout <seconds>\", \"Timeout in seconds\", parseInt)\n  .option(\"--tag <name>\", \"Add tag (repeatable)\", (val: string, acc: string[]) => {\n    acc.push(val);\n    return acc;\n  }, [] as string[])\n  .option(\n    \"--wait-until <event>\",\n    \"Wait condition: load, domcontentloaded, networkidle\",\n  )\n  .option(\"--skip-session\", \"Do not use saved session\")\n  .option(\"--dry-run\", \"Output to stdout instead of saving\")\n  .action(async (url: string, options: Record<string, unknown>) => {\n    try {\n      const configPath = existsSync(CONFIG_PATH) ? CONFIG_PATH : undefined;\n      const config = resolveConfig(\n        {\n          dest: options.dest as string | undefined,\n          tags: options.tag as string[] | undefined,\n          timeout: options.timeout as number | undefined,\n          waitUntil: options.waitUntil as WaitUntilOption | undefined,\n          headed: options.headed as boolean | undefined,\n          selector: options.selector as string | undefined,\n          noSession: options.skipSession as boolean | undefined,\n          dryRun: options.dryRun as boolean | undefined,\n        },\n        configPath,\n      );\n\n      // Validate dest directory exists\n      if (!config.dryRun && !existsSync(config.dest)) {\n        throw new Error(`Destination directory does not exist: ${config.dest}`);\n      }\n\n      const sessionsDir = getSessionDir();\n      const fetchResult = await fetchPage(url, config, sessionsDir);\n\n      let contentHtml: string;\n      let metadata;\n\n      if (config.selector) {\n        // --selector mode: skip Readability, extract metadata from full page\n        contentHtml = fetchResult.html;\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n      } else {\n        const result = extract(fetchResult.html, fetchResult.finalUrl);\n        metadata = result.metadata;\n        contentHtml = result.content;\n      }\n\n      const markdown = convertToMarkdown(contentHtml);\n\n      if (config.dryRun) {\n        const frontmatter = buildFrontmatter(metadata, config.tags);\n        process.stdout.write(`${frontmatter}\\n\\n${markdown}\\n`);\n      } else {\n        const filePath = writeMarkdownFile(\n          config.dest,\n          metadata,\n          markdown,\n          config.tags,\n        );\n        console.error(`Saved: ${filePath}`);\n      }\n    } catch (error) {\n      const message = error instanceof Error ? error.message : String(error);\n      console.error(`Error: ${message}`);\n      process.exit(1);\n    }\n  });\n\nprogram\n  .command(\"login\")\n  .description(\"Login to a site and save session\")\n  .argument(\"<url>\", \"URL to login\")\n  .option(\"--timeout <seconds>\", \"Login timeout in seconds\", parseInt)\n  .action(async (url: string, options: Record<string, unknown>) => {\n    const { chromium } = await import(\"playwright\");\n    const sessionsDir = getSessionDir();\n    ensureSessionDir(sessionsDir);\n\n    const timeoutSec = (options.timeout as number | undefined) ?? 300;\n    const browser = await chromium.launch({ headless: false });\n\n    try {\n      const context = await browser.newContext();\n      const page = await context.newPage();\n\n      await page.goto(url, { waitUntil: \"networkidle\", timeout: timeoutSec * 1000 });\n\n      console.error(\"Browser opened. Log in manually, then press Enter here to save session.\");\n\n      await new Promise<void>((resolve) => {\n        process.stdin.once(\"data\", () => {\n          resolve();\n        });\n      });\n\n      const sessionPath = getSessionPath(url, sessionsDir);\n      await context.storageState({ path: sessionPath });\n      console.error(`Session saved: ${sessionPath}`);\n    } catch (error) {\n      const message = error instanceof Error ? error.message : String(error);\n      console.error(`Error: ${message}`);\n      process.exit(1);\n    } finally {\n      await browser.close();\n    }\n  });\n\nprogram.parse();\n","import { readFileSync } from \"node:fs\";\nimport { homedir } from \"node:os\";\nimport { resolve } from \"node:path\";\nimport yaml from \"js-yaml\";\nimport type { ResolvedConfig, WaitUntilOption } from \"./types.js\";\n\nconst DEFAULT_TIMEOUT = 30;\nconst DEFAULT_WAIT_UNTIL: WaitUntilOption = \"networkidle\";\nconst REQUIRED_TAG = \"clippings\";\n\ninterface FileConfig {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n}\n\ninterface CliOptions {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n  headed?: boolean;\n  selector?: string;\n  noSession?: boolean;\n  dryRun?: boolean;\n}\n\nfunction expandTilde(filePath: string): string {\n  if (filePath.startsWith(\"~/\")) {\n    return resolve(homedir(), filePath.slice(2));\n  }\n  return filePath;\n}\n\nfunction loadConfigFile(configPath: string): FileConfig {\n  const content = readFileSync(configPath, \"utf-8\");\n  const parsed = yaml.load(content);\n  if (parsed === null || typeof parsed !== \"object\") {\n    throw new Error(`Invalid config file: ${configPath}`);\n  }\n  return parsed as FileConfig;\n}\n\nexport function resolveConfig(\n  cliOptions: CliOptions,\n  configPath: string | undefined,\n): ResolvedConfig {\n  // Layer 1: Config file\n  let fileConfig: FileConfig = {};\n  if (configPath) {\n    fileConfig = loadConfigFile(configPath);\n  }\n\n  // Layer 2: Environment variables\n  const envDest = process.env.VAULT_FETCH_DEST;\n  const envTimeout = process.env.VAULT_FETCH_TIMEOUT;\n\n  // Resolve each field: CLI > env > file > default\n  const dest = cliOptions.dest ?? envDest ?? fileConfig.dest;\n  if (dest === undefined) {\n    throw new Error(\n      \"dest is required. Set via --dest, VAULT_FETCH_DEST, or config file.\",\n    );\n  }\n\n  let timeout: number;\n  if (cliOptions.timeout !== undefined) {\n    timeout = cliOptions.timeout;\n  } else if (envTimeout !== undefined) {\n    const parsed = Number(envTimeout);\n    if (Number.isNaN(parsed)) {\n      throw new Error(`Invalid VAULT_FETCH_TIMEOUT value: ${envTimeout}`);\n    }\n    timeout = parsed;\n  } else {\n    timeout = fileConfig.timeout ?? DEFAULT_TIMEOUT;\n  }\n\n  const waitUntil =\n    cliOptions.waitUntil ?? fileConfig.waitUntil ?? DEFAULT_WAIT_UNTIL;\n\n  // Merge tags: file tags + CLI tags + always clippings\n  const allTags = [\n    ...(fileConfig.tags ?? []),\n    ...(cliOptions.tags ?? []),\n    REQUIRED_TAG,\n  ];\n  const tags = [...new Set(allTags)];\n\n  return {\n    dest: expandTilde(dest),\n    tags,\n    timeout,\n    waitUntil,\n    headed: cliOptions.headed ?? false,\n    selector: cliOptions.selector ?? null,\n    noSession: cliOptions.noSession ?? false,\n    dryRun: cliOptions.dryRun ?? false,\n  };\n}\n","import { chromium, type BrowserContext } from \"playwright\";\nimport type { FetchResult, ResolvedConfig } from \"./types.js\";\nimport { getSessionPath, sessionExists } from \"./session.js\";\n\nexport async function fetchPage(\n  url: string,\n  config: ResolvedConfig,\n  sessionsDir: string,\n): Promise<FetchResult> {\n  const browser = await chromium.launch({\n    headless: !config.headed,\n  });\n\n  try {\n    const contextOptions: Parameters<typeof browser.newContext>[0] = {};\n\n    // Load session if available and not disabled\n    if (!config.noSession && sessionExists(url, sessionsDir)) {\n      const sessionPath = getSessionPath(url, sessionsDir);\n      contextOptions.storageState = sessionPath;\n    }\n\n    const context: BrowserContext = await browser.newContext(contextOptions);\n    const page = await context.newPage();\n\n    const timeoutMs = config.timeout * 1000;\n    const response = await page.goto(url, {\n      waitUntil: config.waitUntil,\n      timeout: timeoutMs,\n    });\n\n    if (!response) {\n      throw new Error(`No response received from ${url}`);\n    }\n\n    const status = response.status();\n    if (status >= 400) {\n      throw new Error(`HTTP ${status} received from ${response.url()}`);\n    }\n\n    const finalUrl = response.url();\n    const fullHtml = await page.content();\n    let html: string;\n\n    if (config.selector) {\n      const element = await page.$(config.selector);\n      if (!element) {\n        throw new Error(`Selector not found: ${config.selector}`);\n      }\n      html = await element.innerHTML();\n    } else {\n      html = fullHtml;\n    }\n\n    await context.close();\n\n    return { html, fullHtml, url, finalUrl };\n  } finally {\n    await browser.close();\n  }\n}\n","import { existsSync, mkdirSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport { homedir } from \"node:os\";\n\nconst CONFIG_DIR = join(homedir(), \".config\", \"vault-fetch\");\nconst SESSIONS_DIR = join(CONFIG_DIR, \"sessions\");\n\nexport function getSessionDir(): string {\n  return SESSIONS_DIR;\n}\n\nfunction extractDomain(url: string): string {\n  const parsed = new URL(url);\n  return parsed.hostname ?? \"\";\n}\n\nexport function getSessionPath(url: string, sessionsDir: string): string {\n  const domain = extractDomain(url);\n  return join(sessionsDir, `${domain}.json`);\n}\n\nexport function sessionExists(url: string, sessionsDir: string): boolean {\n  const sessionPath = getSessionPath(url, sessionsDir);\n  return existsSync(sessionPath);\n}\n\nexport function ensureSessionDir(sessionsDir: string): void {\n  if (!existsSync(sessionsDir)) {\n    mkdirSync(sessionsDir, { recursive: true });\n  }\n}\n","import { Readability } from \"@mozilla/readability\";\nimport { JSDOM } from \"jsdom\";\nimport type { Metadata } from \"./types.js\";\n\nfunction getMetaContent(doc: Document, selector: string): string | null {\n  const el = doc.querySelector(selector);\n  return el?.getAttribute(\"content\") ?? null;\n}\n\nfunction formatAuthor(raw: string): string {\n  return `[[${raw.trim()}]]`;\n}\n\nfunction extractPublishedDate(doc: Document): string | null {\n  const published =\n    getMetaContent(doc, 'meta[property=\"article:published_time\"]') ??\n    getMetaContent(doc, 'meta[name=\"datePublished\"]');\n\n  if (!published) {\n    const jsonLd = doc.querySelector('script[type=\"application/ld+json\"]');\n    if (jsonLd?.textContent) {\n      try {\n        const data = JSON.parse(jsonLd.textContent) as Record<string, unknown>;\n        if (typeof data.datePublished === \"string\") {\n          return data.datePublished.split(\"T\")[0];\n        }\n      } catch {\n        // JSON-LD parse failed\n      }\n    }\n    return null;\n  }\n\n  return published.split(\"T\")[0];\n}\n\nfunction extractAuthors(\n  doc: Document,\n  readabilityByline: string | null,\n): string[] {\n  const articleAuthors = doc.querySelectorAll('meta[property=\"article:author\"]');\n  if (articleAuthors.length > 0) {\n    return Array.from(articleAuthors)\n      .map((el) => el.getAttribute(\"content\"))\n      .filter((v): v is string => v !== null)\n      .map(formatAuthor);\n  }\n\n  const ogAuthor = getMetaContent(doc, 'meta[property=\"og:author\"]');\n  if (ogAuthor) {\n    return [formatAuthor(ogAuthor)];\n  }\n\n  if (readabilityByline) {\n    return [formatAuthor(readabilityByline)];\n  }\n\n  return [];\n}\n\nexport interface ExtractResult {\n  metadata: Metadata;\n  content: string;\n}\n\nexport function extract(html: string, finalUrl: string): ExtractResult {\n  // One JSDOM for metadata DOM queries\n  const metaDom = new JSDOM(html, { url: finalUrl });\n  const doc = metaDom.window.document;\n\n  // One JSDOM for Readability (which mutates the DOM)\n  const readabilityDom = new JSDOM(html, { url: finalUrl });\n  const reader = new Readability(readabilityDom.window.document);\n  const article = reader.parse();\n\n  if (!article) {\n    throw new Error(\"Readability failed to extract content from the page\");\n  }\n\n  if (!article.content) {\n    throw new Error(\"Readability returned empty content for the page\");\n  }\n\n  const title = article.title ?? doc.title;\n  const authors = extractAuthors(doc, article.byline ?? null);\n  const published = extractPublishedDate(doc);\n\n  const description =\n    getMetaContent(doc, 'meta[property=\"og:description\"]') ??\n    getMetaContent(doc, 'meta[name=\"description\"]') ??\n    (article.excerpt ?? null);\n\n  const today = new Date().toISOString().split(\"T\")[0];\n\n  return {\n    metadata: {\n      title,\n      source: finalUrl,\n      author: authors,\n      published,\n      created: today,\n      description,\n    },\n    content: article.content,\n  };\n}\n\nexport function extractMetadata(html: string, finalUrl: string): Metadata {\n  const metaDom = new JSDOM(html, { url: finalUrl });\n  const doc = metaDom.window.document;\n\n  const readabilityDom = new JSDOM(html, { url: finalUrl });\n  const reader = new Readability(readabilityDom.window.document);\n  const article = reader.parse();\n\n  const title = article?.title ?? doc.title;\n  const authors = extractAuthors(doc, article?.byline ?? null);\n  const published = extractPublishedDate(doc);\n\n  const description =\n    getMetaContent(doc, 'meta[property=\"og:description\"]') ??\n    getMetaContent(doc, 'meta[name=\"description\"]') ??\n    (article?.excerpt ?? null);\n\n  const today = new Date().toISOString().split(\"T\")[0];\n\n  return {\n    title,\n    source: finalUrl,\n    author: authors,\n    published,\n    created: today,\n    description,\n  };\n}\n","import TurndownService from \"turndown\";\n\nexport function convertToMarkdown(html: string): string {\n  const turndown = new TurndownService({\n    headingStyle: \"atx\",\n    codeBlockStyle: \"fenced\",\n    bulletListMarker: \"-\",\n  });\n\n  return turndown.turndown(html);\n}\n","import { writeFileSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport yaml from \"js-yaml\";\nimport type { Metadata } from \"./types.js\";\n\nconst UNSAFE_CHARS = /[/\\\\:*?\"<>|]/g;\nconst MAX_FILENAME_LENGTH = 200;\n\nexport function sanitizeFilename(title: string): string {\n  const sanitized = title\n    .replace(UNSAFE_CHARS, \"\")\n    .replace(/\\s+/g, \" \")\n    .trim();\n  const base = sanitized.slice(0, MAX_FILENAME_LENGTH) || \"Untitled\";\n  return `${base}.md`;\n}\n\nexport function buildFrontmatter(metadata: Metadata, tags: string[]): string {\n  const data: Record<string, unknown> = {\n    title: metadata.title,\n    source: metadata.source,\n  };\n\n  if (metadata.author.length > 0) {\n    data.author = metadata.author;\n  }\n\n  if (metadata.published) {\n    data.published = metadata.published;\n  }\n\n  data.created = metadata.created;\n\n  if (metadata.description) {\n    data.description = metadata.description;\n  }\n\n  data.tags = tags;\n\n  const yamlStr = yaml.dump(data, {\n    quotingType: '\"',\n    forceQuotes: false,\n    lineWidth: -1,\n    sortKeys: false,\n  });\n\n  return `---\\n${yamlStr}---`;\n}\n\nexport function writeMarkdownFile(\n  dest: string,\n  metadata: Metadata,\n  markdownContent: string,\n  tags: string[],\n): string {\n  const filename = sanitizeFilename(metadata.title);\n  const filePath = join(dest, filename);\n  const frontmatter = buildFrontmatter(metadata, tags);\n  const fullContent = `${frontmatter}\\n\\n${markdownContent}\\n`;\n\n  writeFileSync(filePath, fullContent, \"utf-8\");\n\n  return filePath;\n}\n"],"mappings":";;;AAAA,SAAS,eAAe;AACxB,SAAS,cAAAA,mBAAkB;AAC3B,SAAS,WAAAC,gBAAe;AACxB,SAAS,QAAAC,aAAY;;;ACHrB,SAAS,oBAAoB;AAC7B,SAAS,eAAe;AACxB,SAAS,eAAe;AACxB,OAAO,UAAU;AAGjB,IAAM,kBAAkB;AACxB,IAAM,qBAAsC;AAC5C,IAAM,eAAe;AAoBrB,SAAS,YAAY,UAA0B;AAC7C,MAAI,SAAS,WAAW,IAAI,GAAG;AAC7B,WAAO,QAAQ,QAAQ,GAAG,SAAS,MAAM,CAAC,CAAC;AAAA,EAC7C;AACA,SAAO;AACT;AAEA,SAAS,eAAe,YAAgC;AACtD,QAAM,UAAU,aAAa,YAAY,OAAO;AAChD,QAAM,SAAS,KAAK,KAAK,OAAO;AAChC,MAAI,WAAW,QAAQ,OAAO,WAAW,UAAU;AACjD,UAAM,IAAI,MAAM,wBAAwB,UAAU,EAAE;AAAA,EACtD;AACA,SAAO;AACT;AAEO,SAAS,cACd,YACA,YACgB;AAEhB,MAAI,aAAyB,CAAC;AAC9B,MAAI,YAAY;AACd,iBAAa,eAAe,UAAU;AAAA,EACxC;AAGA,QAAM,UAAU,QAAQ,IAAI;AAC5B,QAAM,aAAa,QAAQ,IAAI;AAG/B,QAAM,OAAO,WAAW,QAAQ,WAAW,WAAW;AACtD,MAAI,SAAS,QAAW;AACtB,UAAM,IAAI;AAAA,MACR;AAAA,IACF;AAAA,EACF;AAEA,MAAI;AACJ,MAAI,WAAW,YAAY,QAAW;AACpC,cAAU,WAAW;AAAA,EACvB,WAAW,eAAe,QAAW;AACnC,UAAM,SAAS,OAAO,UAAU;AAChC,QAAI,OAAO,MAAM,MAAM,GAAG;AACxB,YAAM,IAAI,MAAM,sCAAsC,UAAU,EAAE;AAAA,IACpE;AACA,cAAU;AAAA,EACZ,OAAO;AACL,cAAU,WAAW,WAAW;AAAA,EAClC;AAEA,QAAM,YACJ,WAAW,aAAa,WAAW,aAAa;AAGlD,QAAM,UAAU;AAAA,IACd,GAAI,WAAW,QAAQ,CAAC;AAAA,IACxB,GAAI,WAAW,QAAQ,CAAC;AAAA,IACxB;AAAA,EACF;AACA,QAAM,OAAO,CAAC,GAAG,IAAI,IAAI,OAAO,CAAC;AAEjC,SAAO;AAAA,IACL,MAAM,YAAY,IAAI;AAAA,IACtB;AAAA,IACA;AAAA,IACA;AAAA,IACA,QAAQ,WAAW,UAAU;AAAA,IAC7B,UAAU,WAAW,YAAY;AAAA,IACjC,WAAW,WAAW,aAAa;AAAA,IACnC,QAAQ,WAAW,UAAU;AAAA,EAC/B;AACF;;;ACpGA,SAAS,gBAAqC;;;ACA9C,SAAS,YAAY,iBAAiB;AACtC,SAAS,YAAY;AACrB,SAAS,WAAAC,gBAAe;AAExB,IAAM,aAAa,KAAKA,SAAQ,GAAG,WAAW,aAAa;AAC3D,IAAM,eAAe,KAAK,YAAY,UAAU;AAEzC,SAAS,gBAAwB;AACtC,SAAO;AACT;AAEA,SAAS,cAAc,KAAqB;AAC1C,QAAM,SAAS,IAAI,IAAI,GAAG;AAC1B,SAAO,OAAO,YAAY;AAC5B;AAEO,SAAS,eAAe,KAAa,aAA6B;AACvE,QAAM,SAAS,cAAc,GAAG;AAChC,SAAO,KAAK,aAAa,GAAG,MAAM,OAAO;AAC3C;AAEO,SAAS,cAAc,KAAa,aAA8B;AACvE,QAAM,cAAc,eAAe,KAAK,WAAW;AACnD,SAAO,WAAW,WAAW;AAC/B;AAEO,SAAS,iBAAiB,aAA2B;AAC1D,MAAI,CAAC,WAAW,WAAW,GAAG;AAC5B,cAAU,aAAa,EAAE,WAAW,KAAK,CAAC;AAAA,EAC5C;AACF;;;AD1BA,eAAsB,UACpB,KACA,QACA,aACsB;AACtB,QAAM,UAAU,MAAM,SAAS,OAAO;AAAA,IACpC,UAAU,CAAC,OAAO;AAAA,EACpB,CAAC;AAED,MAAI;AACF,UAAM,iBAA2D,CAAC;AAGlE,QAAI,CAAC,OAAO,aAAa,cAAc,KAAK,WAAW,GAAG;AACxD,YAAM,cAAc,eAAe,KAAK,WAAW;AACnD,qBAAe,eAAe;AAAA,IAChC;AAEA,UAAM,UAA0B,MAAM,QAAQ,WAAW,cAAc;AACvE,UAAM,OAAO,MAAM,QAAQ,QAAQ;AAEnC,UAAM,YAAY,OAAO,UAAU;AACnC,UAAM,WAAW,MAAM,KAAK,KAAK,KAAK;AAAA,MACpC,WAAW,OAAO;AAAA,MAClB,SAAS;AAAA,IACX,CAAC;AAED,QAAI,CAAC,UAAU;AACb,YAAM,IAAI,MAAM,6BAA6B,GAAG,EAAE;AAAA,IACpD;AAEA,UAAM,SAAS,SAAS,OAAO;AAC/B,QAAI,UAAU,KAAK;AACjB,YAAM,IAAI,MAAM,QAAQ,MAAM,kBAAkB,SAAS,IAAI,CAAC,EAAE;AAAA,IAClE;AAEA,UAAM,WAAW,SAAS,IAAI;AAC9B,UAAM,WAAW,MAAM,KAAK,QAAQ;AACpC,QAAI;AAEJ,QAAI,OAAO,UAAU;AACnB,YAAM,UAAU,MAAM,KAAK,EAAE,OAAO,QAAQ;AAC5C,UAAI,CAAC,SAAS;AACZ,cAAM,IAAI,MAAM,uBAAuB,OAAO,QAAQ,EAAE;AAAA,MAC1D;AACA,aAAO,MAAM,QAAQ,UAAU;AAAA,IACjC,OAAO;AACL,aAAO;AAAA,IACT;AAEA,UAAM,QAAQ,MAAM;AAEpB,WAAO,EAAE,MAAM,UAAU,KAAK,SAAS;AAAA,EACzC,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF;;;AE5DA,SAAS,mBAAmB;AAC5B,SAAS,aAAa;AAGtB,SAAS,eAAe,KAAe,UAAiC;AACtE,QAAM,KAAK,IAAI,cAAc,QAAQ;AACrC,SAAO,IAAI,aAAa,SAAS,KAAK;AACxC;AAEA,SAAS,aAAa,KAAqB;AACzC,SAAO,KAAK,IAAI,KAAK,CAAC;AACxB;AAEA,SAAS,qBAAqB,KAA8B;AAC1D,QAAM,YACJ,eAAe,KAAK,yCAAyC,KAC7D,eAAe,KAAK,4BAA4B;AAElD,MAAI,CAAC,WAAW;AACd,UAAM,SAAS,IAAI,cAAc,oCAAoC;AACrE,QAAI,QAAQ,aAAa;AACvB,UAAI;AACF,cAAM,OAAO,KAAK,MAAM,OAAO,WAAW;AAC1C,YAAI,OAAO,KAAK,kBAAkB,UAAU;AAC1C,iBAAO,KAAK,cAAc,MAAM,GAAG,EAAE,CAAC;AAAA,QACxC;AAAA,MACF,QAAQ;AAAA,MAER;AAAA,IACF;AACA,WAAO;AAAA,EACT;AAEA,SAAO,UAAU,MAAM,GAAG,EAAE,CAAC;AAC/B;AAEA,SAAS,eACP,KACA,mBACU;AACV,QAAM,iBAAiB,IAAI,iBAAiB,iCAAiC;AAC7E,MAAI,eAAe,SAAS,GAAG;AAC7B,WAAO,MAAM,KAAK,cAAc,EAC7B,IAAI,CAAC,OAAO,GAAG,aAAa,SAAS,CAAC,EACtC,OAAO,CAAC,MAAmB,MAAM,IAAI,EACrC,IAAI,YAAY;AAAA,EACrB;AAEA,QAAM,WAAW,eAAe,KAAK,4BAA4B;AACjE,MAAI,UAAU;AACZ,WAAO,CAAC,aAAa,QAAQ,CAAC;AAAA,EAChC;AAEA,MAAI,mBAAmB;AACrB,WAAO,CAAC,aAAa,iBAAiB,CAAC;AAAA,EACzC;AAEA,SAAO,CAAC;AACV;AAOO,SAAS,QAAQ,MAAc,UAAiC;AAErE,QAAM,UAAU,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACjD,QAAM,MAAM,QAAQ,OAAO;AAG3B,QAAM,iBAAiB,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACxD,QAAM,SAAS,IAAI,YAAY,eAAe,OAAO,QAAQ;AAC7D,QAAM,UAAU,OAAO,MAAM;AAE7B,MAAI,CAAC,SAAS;AACZ,UAAM,IAAI,MAAM,qDAAqD;AAAA,EACvE;AAEA,MAAI,CAAC,QAAQ,SAAS;AACpB,UAAM,IAAI,MAAM,iDAAiD;AAAA,EACnE;AAEA,QAAM,QAAQ,QAAQ,SAAS,IAAI;AACnC,QAAM,UAAU,eAAe,KAAK,QAAQ,UAAU,IAAI;AAC1D,QAAM,YAAY,qBAAqB,GAAG;AAE1C,QAAM,cACJ,eAAe,KAAK,iCAAiC,KACrD,eAAe,KAAK,0BAA0B,MAC7C,QAAQ,WAAW;AAEtB,QAAM,SAAQ,oBAAI,KAAK,GAAE,YAAY,EAAE,MAAM,GAAG,EAAE,CAAC;AAEnD,SAAO;AAAA,IACL,UAAU;AAAA,MACR;AAAA,MACA,QAAQ;AAAA,MACR,QAAQ;AAAA,MACR;AAAA,MACA,SAAS;AAAA,MACT;AAAA,IACF;AAAA,IACA,SAAS,QAAQ;AAAA,EACnB;AACF;AAEO,SAAS,gBAAgB,MAAc,UAA4B;AACxE,QAAM,UAAU,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACjD,QAAM,MAAM,QAAQ,OAAO;AAE3B,QAAM,iBAAiB,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACxD,QAAM,SAAS,IAAI,YAAY,eAAe,OAAO,QAAQ;AAC7D,QAAM,UAAU,OAAO,MAAM;AAE7B,QAAM,QAAQ,SAAS,SAAS,IAAI;AACpC,QAAM,UAAU,eAAe,KAAK,SAAS,UAAU,IAAI;AAC3D,QAAM,YAAY,qBAAqB,GAAG;AAE1C,QAAM,cACJ,eAAe,KAAK,iCAAiC,KACrD,eAAe,KAAK,0BAA0B,MAC7C,SAAS,WAAW;AAEvB,QAAM,SAAQ,oBAAI,KAAK,GAAE,YAAY,EAAE,MAAM,GAAG,EAAE,CAAC;AAEnD,SAAO;AAAA,IACL;AAAA,IACA,QAAQ;AAAA,IACR,QAAQ;AAAA,IACR;AAAA,IACA,SAAS;AAAA,IACT;AAAA,EACF;AACF;;;ACtIA,OAAO,qBAAqB;AAErB,SAAS,kBAAkB,MAAsB;AACtD,QAAM,WAAW,IAAI,gBAAgB;AAAA,IACnC,cAAc;AAAA,IACd,gBAAgB;AAAA,IAChB,kBAAkB;AAAA,EACpB,CAAC;AAED,SAAO,SAAS,SAAS,IAAI;AAC/B;;;ACVA,SAAS,qBAAqB;AAC9B,SAAS,QAAAC,aAAY;AACrB,OAAOC,WAAU;AAGjB,IAAM,eAAe;AACrB,IAAM,sBAAsB;AAErB,SAAS,iBAAiB,OAAuB;AACtD,QAAM,YAAY,MACf,QAAQ,cAAc,EAAE,EACxB,QAAQ,QAAQ,GAAG,EACnB,KAAK;AACR,QAAM,OAAO,UAAU,MAAM,GAAG,mBAAmB,KAAK;AACxD,SAAO,GAAG,IAAI;AAChB;AAEO,SAAS,iBAAiB,UAAoB,MAAwB;AAC3E,QAAM,OAAgC;AAAA,IACpC,OAAO,SAAS;AAAA,IAChB,QAAQ,SAAS;AAAA,EACnB;AAEA,MAAI,SAAS,OAAO,SAAS,GAAG;AAC9B,SAAK,SAAS,SAAS;AAAA,EACzB;AAEA,MAAI,SAAS,WAAW;AACtB,SAAK,YAAY,SAAS;AAAA,EAC5B;AAEA,OAAK,UAAU,SAAS;AAExB,MAAI,SAAS,aAAa;AACxB,SAAK,cAAc,SAAS;AAAA,EAC9B;AAEA,OAAK,OAAO;AAEZ,QAAM,UAAUA,MAAK,KAAK,MAAM;AAAA,IAC9B,aAAa;AAAA,IACb,aAAa;AAAA,IACb,WAAW;AAAA,IACX,UAAU;AAAA,EACZ,CAAC;AAED,SAAO;AAAA,EAAQ,OAAO;AACxB;AAEO,SAAS,kBACd,MACA,UACA,iBACA,MACQ;AACR,QAAM,WAAW,iBAAiB,SAAS,KAAK;AAChD,QAAM,WAAWD,MAAK,MAAM,QAAQ;AACpC,QAAM,cAAc,iBAAiB,UAAU,IAAI;AACnD,QAAM,cAAc,GAAG,WAAW;AAAA;AAAA,EAAO,eAAe;AAAA;AAExD,gBAAc,UAAU,aAAa,OAAO;AAE5C,SAAO;AACT;;;AN/CA,IAAM,cAAcE,MAAKC,SAAQ,GAAG,WAAW,eAAe,aAAa;AAE3E,IAAM,UAAU,IAAI,QAAQ;AAE5B,QACG,KAAK,aAAa,EAClB;AAAA,EACC;AACF,EACC,QAAQ,OAAO;AAElB,QACG,QAAQ,OAAO,EACf,YAAY,mCAAmC,EAC/C,SAAS,SAAS,cAAc,EAChC,OAAO,iBAAiB,uBAAuB,EAC/C,OAAO,YAAY,4BAA4B,EAC/C,OAAO,oBAAoB,yBAAyB,EACpD,OAAO,uBAAuB,sBAAsB,QAAQ,EAC5D,OAAO,gBAAgB,wBAAwB,CAAC,KAAa,QAAkB;AAC9E,MAAI,KAAK,GAAG;AACZ,SAAO;AACT,GAAG,CAAC,CAAa,EAChB;AAAA,EACC;AAAA,EACA;AACF,EACC,OAAO,kBAAkB,0BAA0B,EACnD,OAAO,aAAa,oCAAoC,EACxD,OAAO,OAAO,KAAa,YAAqC;AAC/D,MAAI;AACF,UAAM,aAAaC,YAAW,WAAW,IAAI,cAAc;AAC3D,UAAM,SAAS;AAAA,MACb;AAAA,QACE,MAAM,QAAQ;AAAA,QACd,MAAM,QAAQ;AAAA,QACd,SAAS,QAAQ;AAAA,QACjB,WAAW,QAAQ;AAAA,QACnB,QAAQ,QAAQ;AAAA,QAChB,UAAU,QAAQ;AAAA,QAClB,WAAW,QAAQ;AAAA,QACnB,QAAQ,QAAQ;AAAA,MAClB;AAAA,MACA;AAAA,IACF;AAGA,QAAI,CAAC,OAAO,UAAU,CAACA,YAAW,OAAO,IAAI,GAAG;AAC9C,YAAM,IAAI,MAAM,yCAAyC,OAAO,IAAI,EAAE;AAAA,IACxE;AAEA,UAAM,cAAc,cAAc;AAClC,UAAM,cAAc,MAAM,UAAU,KAAK,QAAQ,WAAW;AAE5D,QAAI;AACJ,QAAI;AAEJ,QAAI,OAAO,UAAU;AAEnB,oBAAc,YAAY;AAC1B,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AAAA,IACvE,OAAO;AACL,YAAM,SAAS,QAAQ,YAAY,MAAM,YAAY,QAAQ;AAC7D,iBAAW,OAAO;AAClB,oBAAc,OAAO;AAAA,IACvB;AAEA,UAAM,WAAW,kBAAkB,WAAW;AAE9C,QAAI,OAAO,QAAQ;AACjB,YAAM,cAAc,iBAAiB,UAAU,OAAO,IAAI;AAC1D,cAAQ,OAAO,MAAM,GAAG,WAAW;AAAA;AAAA,EAAO,QAAQ;AAAA,CAAI;AAAA,IACxD,OAAO;AACL,YAAM,WAAW;AAAA,QACf,OAAO;AAAA,QACP;AAAA,QACA;AAAA,QACA,OAAO;AAAA,MACT;AACA,cAAQ,MAAM,UAAU,QAAQ,EAAE;AAAA,IACpC;AAAA,EACF,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAQ,MAAM,UAAU,OAAO,EAAE;AACjC,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF,CAAC;AAEH,QACG,QAAQ,OAAO,EACf,YAAY,kCAAkC,EAC9C,SAAS,SAAS,cAAc,EAChC,OAAO,uBAAuB,4BAA4B,QAAQ,EAClE,OAAO,OAAO,KAAa,YAAqC;AAC/D,QAAM,EAAE,UAAAC,UAAS,IAAI,MAAM,OAAO,YAAY;AAC9C,QAAM,cAAc,cAAc;AAClC,mBAAiB,WAAW;AAE5B,QAAM,aAAc,QAAQ,WAAkC;AAC9D,QAAM,UAAU,MAAMA,UAAS,OAAO,EAAE,UAAU,MAAM,CAAC;AAEzD,MAAI;AACF,UAAM,UAAU,MAAM,QAAQ,WAAW;AACzC,UAAM,OAAO,MAAM,QAAQ,QAAQ;AAEnC,UAAM,KAAK,KAAK,KAAK,EAAE,WAAW,eAAe,SAAS,aAAa,IAAK,CAAC;AAE7E,YAAQ,MAAM,yEAAyE;AAEvF,UAAM,IAAI,QAAc,CAACC,aAAY;AACnC,cAAQ,MAAM,KAAK,QAAQ,MAAM;AAC/B,QAAAA,SAAQ;AAAA,MACV,CAAC;AAAA,IACH,CAAC;AAED,UAAM,cAAc,eAAe,KAAK,WAAW;AACnD,UAAM,QAAQ,aAAa,EAAE,MAAM,YAAY,CAAC;AAChD,YAAQ,MAAM,kBAAkB,WAAW,EAAE;AAAA,EAC/C,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAQ,MAAM,UAAU,OAAO,EAAE;AACjC,YAAQ,KAAK,CAAC;AAAA,EAChB,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF,CAAC;AAEH,QAAQ,MAAM;","names":["existsSync","homedir","join","homedir","join","yaml","join","homedir","existsSync","chromium","resolve"]}
+{"version":3,"sources":["../src/cli.ts","../src/config.ts","../src/fetcher.ts","../src/session.ts","../src/extractor.ts","../src/converter.ts","../src/writer.ts"],"sourcesContent":["import { Command } from \"commander\";\nimport { once } from \"node:events\";\nimport { existsSync } from \"node:fs\";\nimport { homedir } from \"node:os\";\nimport { join } from \"node:path\";\nimport { resolveConfig } from \"./config.js\";\nimport { fetchPage } from \"./fetcher.js\";\nimport { extract, extractMetadata } from \"./extractor.js\";\nimport { convertToMarkdown } from \"./converter.js\";\nimport { writeMarkdownFile, buildFrontmatter } from \"./writer.js\";\nimport {\n  getSessionDir,\n  getSessionPath,\n  ensureSessionDir,\n} from \"./session.js\";\nimport type { WaitUntilOption } from \"./types.js\";\n\nconst CONFIG_PATH = join(homedir(), \".config\", \"vault-fetch\", \"config.yaml\");\n\nconst program = new Command();\n\nprogram\n  .name(\"vault-fetch\")\n  .description(\n    \"Fetch JS-rendered web pages and save as Markdown to Obsidian Vault\",\n  )\n  .version(\"0.2.0\");\n\nprogram\n  .command(\"fetch\")\n  .description(\"Fetch a page and save as Markdown\")\n  .argument(\"<url>\", \"URL to fetch\")\n  .option(\"--dest <path>\", \"Destination directory\")\n  .option(\"--headed\", \"Run browser in headed mode\")\n  .option(\"--selector <css>\", \"CSS selector to extract\")\n  .option(\"--timeout <seconds>\", \"Timeout in seconds\", parseInt)\n  .option(\"--tag <name>\", \"Add tag (repeatable)\", (val: string, acc: string[]) => {\n    acc.push(val);\n    return acc;\n  }, [] as string[])\n  .option(\n    \"--wait-until <event>\",\n    \"Wait condition: load, domcontentloaded, networkidle\",\n  )\n  .option(\"--skip-session\", \"Do not use saved session\")\n  .option(\"--dry-run\", \"Output to stdout instead of saving\")\n  .option(\"--no-block-images\", \"Do not block image requests\")\n  .option(\"--no-block-fonts\", \"Do not block font requests\")\n  .option(\"--no-block-media\", \"Do not block media requests\")\n  .option(\"--raw\", \"Convert full page HTML without Readability extraction\")\n  .action(async (url: string, options: Record<string, unknown>) => {\n    try {\n      const configPath = existsSync(CONFIG_PATH) ? CONFIG_PATH : undefined;\n      const config = resolveConfig(\n        {\n          dest: options.dest as string | undefined,\n          tags: options.tag as string[] | undefined,\n          timeout: options.timeout as number | undefined,\n          waitUntil: options.waitUntil as WaitUntilOption | undefined,\n          headed: options.headed as boolean | undefined,\n          selector: options.selector as string | undefined,\n          noSession: options.skipSession as boolean | undefined,\n          dryRun: options.dryRun as boolean | undefined,\n          blockImages: options.blockImages as boolean | undefined,\n          blockFonts: options.blockFonts as boolean | undefined,\n          blockMedia: options.blockMedia as boolean | undefined,\n          raw: options.raw as boolean | undefined,\n        },\n        configPath,\n      );\n\n      if (config.raw && config.selector) {\n        throw new Error(\"--raw and --selector cannot be used together.\");\n      }\n\n      // Validate dest directory exists\n      if (!config.dryRun && !existsSync(config.dest)) {\n        throw new Error(`Destination directory does not exist: ${config.dest}`);\n      }\n\n      const sessionsDir = getSessionDir();\n      const fetchResult = await fetchPage(url, config, sessionsDir);\n\n      let contentHtml: string;\n      let metadata;\n\n      if (config.selector) {\n        // --selector mode: skip Readability, extract metadata from full page\n        contentHtml = fetchResult.html;\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n      } else if (config.raw) {\n        // --raw mode: skip Readability, convert full page HTML directly\n        contentHtml = fetchResult.fullHtml;\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n      } else {\n        const result = extract(fetchResult.html, fetchResult.finalUrl);\n        metadata = result.metadata;\n        contentHtml = result.content;\n      }\n\n      const markdown = convertToMarkdown(contentHtml);\n\n      if (config.dryRun) {\n        const frontmatter = buildFrontmatter(metadata, config.tags);\n        process.stdout.write(`${frontmatter}\\n\\n${markdown}\\n`);\n      } else {\n        const filePath = writeMarkdownFile(\n          config.dest,\n          metadata,\n          markdown,\n          config.tags,\n        );\n        console.error(`Saved: ${filePath}`);\n      }\n    } catch (error) {\n      const message = error instanceof Error ? error.message : String(error);\n      console.error(`Error: ${message}`);\n      process.exit(1);\n    }\n  });\n\nprogram\n  .command(\"login\")\n  .description(\"Login to a site and save session\")\n  .argument(\"<url>\", \"URL to login\")\n  .option(\"--timeout <seconds>\", \"Login timeout in seconds\", parseInt)\n  .action(async (url: string, options: Record<string, unknown>) => {\n    const { chromium } = await import(\"playwright\");\n    const sessionsDir = getSessionDir();\n    ensureSessionDir(sessionsDir);\n\n    const timeoutSec = (options.timeout as number | undefined) ?? 300;\n    const browser = await chromium.launch({ headless: false });\n\n    try {\n      const context = await browser.newContext();\n      const page = await context.newPage();\n\n      await page.goto(url, { waitUntil: \"networkidle\", timeout: timeoutSec * 1000 });\n\n      console.error(\"Browser opened. Log in manually, then press Enter here to save session.\");\n\n      process.stdin.resume();\n      await once(process.stdin, \"data\");\n      process.stdin.pause();\n      process.stdin.unref();\n\n      const sessionPath = getSessionPath(url, sessionsDir);\n      await context.storageState({ path: sessionPath });\n      console.error(`Session saved: ${sessionPath}`);\n    } catch (error) {\n      const message = error instanceof Error ? error.message : String(error);\n      console.error(`Error: ${message}`);\n      process.exit(1);\n    } finally {\n      await browser.close();\n    }\n  });\n\nprogram.parse();\n","import { readFileSync } from \"node:fs\";\nimport { homedir } from \"node:os\";\nimport { resolve } from \"node:path\";\nimport yaml from \"js-yaml\";\nimport type { ResolvedConfig, WaitUntilOption } from \"./types.js\";\n\nconst DEFAULT_TIMEOUT = 30;\nconst DEFAULT_WAIT_UNTIL: WaitUntilOption = \"networkidle\";\nconst REQUIRED_TAG = \"clippings\";\n\ninterface FileConfig {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n}\n\ninterface CliOptions {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n  headed?: boolean;\n  selector?: string;\n  noSession?: boolean;\n  dryRun?: boolean;\n  blockImages?: boolean;\n  blockFonts?: boolean;\n  blockMedia?: boolean;\n  raw?: boolean;\n}\n\nfunction expandTilde(filePath: string): string {\n  if (filePath.startsWith(\"~/\")) {\n    return resolve(homedir(), filePath.slice(2));\n  }\n  return filePath;\n}\n\nconst VALID_WAIT_UNTIL: readonly string[] = [\"load\", \"domcontentloaded\", \"networkidle\"];\n\nfunction validateWaitUntil(value: string): WaitUntilOption {\n  if (!VALID_WAIT_UNTIL.includes(value)) {\n    throw new Error(\n      `Invalid waitUntil value: \"${value}\". Must be one of: ${VALID_WAIT_UNTIL.join(\", \")}`,\n    );\n  }\n  return value as WaitUntilOption;\n}\n\nfunction loadConfigFile(configPath: string): FileConfig {\n  const content = readFileSync(configPath, \"utf-8\");\n  const parsed = yaml.load(content);\n  if (parsed === null || typeof parsed !== \"object\") {\n    throw new Error(`Invalid config file: ${configPath}`);\n  }\n  const config = parsed as Record<string, unknown>;\n\n  if (config.timeout !== undefined && typeof config.timeout !== \"number\") {\n    throw new Error(`Invalid timeout in config file: expected number, got ${typeof config.timeout}`);\n  }\n  if (config.dest !== undefined && typeof config.dest !== \"string\") {\n    throw new Error(`Invalid dest in config file: expected string, got ${typeof config.dest}`);\n  }\n  if (config.waitUntil !== undefined) {\n    if (typeof config.waitUntil !== \"string\") {\n      throw new Error(`Invalid waitUntil in config file: expected string, got ${typeof config.waitUntil}`);\n    }\n    validateWaitUntil(config.waitUntil);\n  }\n  if (config.tags !== undefined) {\n    if (!Array.isArray(config.tags) || !config.tags.every((t: unknown) => typeof t === \"string\")) {\n      throw new Error(\"Invalid tags in config file: expected array of strings\");\n    }\n  }\n\n  return config as FileConfig;\n}\n\nexport function resolveConfig(\n  cliOptions: CliOptions,\n  configPath: string | undefined,\n): ResolvedConfig {\n  // Layer 1: Config file\n  let fileConfig: FileConfig = {};\n  if (configPath) {\n    fileConfig = loadConfigFile(configPath);\n  }\n\n  // Layer 2: Environment variables\n  const envDest = process.env.VAULT_FETCH_DEST;\n  const envTimeout = process.env.VAULT_FETCH_TIMEOUT;\n\n  // Resolve each field: CLI > env > file > default\n  const dest = cliOptions.dest ?? envDest ?? fileConfig.dest;\n  if (dest === undefined) {\n    throw new Error(\n      \"dest is required. Set via --dest, VAULT_FETCH_DEST, or config file.\",\n    );\n  }\n\n  let timeout: number;\n  if (cliOptions.timeout !== undefined) {\n    timeout = cliOptions.timeout;\n  } else if (envTimeout !== undefined) {\n    const parsed = Number(envTimeout);\n    if (Number.isNaN(parsed)) {\n      throw new Error(`Invalid VAULT_FETCH_TIMEOUT value: ${envTimeout}`);\n    }\n    timeout = parsed;\n  } else {\n    timeout = fileConfig.timeout ?? DEFAULT_TIMEOUT;\n  }\n\n  const rawWaitUntil = cliOptions.waitUntil ?? fileConfig.waitUntil ?? DEFAULT_WAIT_UNTIL;\n  const waitUntil = validateWaitUntil(rawWaitUntil);\n\n  // Merge tags: file tags + CLI tags + always clippings\n  const allTags = [\n    ...(fileConfig.tags ?? []),\n    ...(cliOptions.tags ?? []),\n    REQUIRED_TAG,\n  ];\n  const tags = [...new Set(allTags)];\n\n  return {\n    dest: expandTilde(dest),\n    tags,\n    timeout,\n    waitUntil,\n    headed: cliOptions.headed ?? false,\n    selector: cliOptions.selector ?? null,\n    noSession: cliOptions.noSession ?? false,\n    dryRun: cliOptions.dryRun ?? false,\n    blockImages: cliOptions.blockImages ?? true,\n    blockFonts: cliOptions.blockFonts ?? true,\n    blockMedia: cliOptions.blockMedia ?? true,\n    raw: cliOptions.raw ?? false,\n  };\n}\n","import { chromium, type BrowserContext } from \"playwright\";\nimport type { FetchResult, ResolvedConfig } from \"./types.js\";\nimport { getSessionPath, sessionExists } from \"./session.js\";\n\nexport const CHROME_USER_AGENT =\n  \"Mozilla/5.0 (Windows NT 10.0; Win64; x64) \" +\n  \"AppleWebKit/537.36 (KHTML, like Gecko) \" +\n  \"Chrome/134.0.0.0 Safari/537.36\";\n\ninterface BlockingOptions {\n  blockImages: boolean;\n  blockFonts: boolean;\n  blockMedia: boolean;\n}\n\nexport function buildBlockedResourceTypes(options: BlockingOptions): Set<string> {\n  const blocked = new Set<string>();\n  if (options.blockImages) blocked.add(\"image\");\n  if (options.blockFonts) blocked.add(\"font\");\n  if (options.blockMedia) blocked.add(\"media\");\n  return blocked;\n}\n\nexport async function fetchPage(\n  url: string,\n  config: ResolvedConfig,\n  sessionsDir: string,\n): Promise<FetchResult> {\n  const browser = await chromium.launch({\n    headless: !config.headed,\n  });\n\n  try {\n    const contextOptions: Parameters<typeof browser.newContext>[0] = {\n      userAgent: CHROME_USER_AGENT,\n    };\n\n    // Load session if available and not disabled\n    if (!config.noSession && sessionExists(url, sessionsDir)) {\n      const sessionPath = getSessionPath(url, sessionsDir);\n      contextOptions.storageState = sessionPath;\n    }\n\n    const context: BrowserContext = await browser.newContext(contextOptions);\n    const page = await context.newPage();\n\n    // Block specified resource types for faster loading\n    const blockedTypes = buildBlockedResourceTypes(config);\n    if (blockedTypes.size > 0) {\n      await page.route(\"**/*\", async (route) => {\n        if (blockedTypes.has(route.request().resourceType())) {\n          await route.abort();\n        } else {\n          await route.continue();\n        }\n      });\n    }\n\n    const timeoutMs = config.timeout * 1000;\n    const response = await page.goto(url, {\n      waitUntil: config.waitUntil,\n      timeout: timeoutMs,\n    });\n\n    if (!response) {\n      throw new Error(`No response received from ${url}`);\n    }\n\n    const status = response.status();\n    if (status >= 400) {\n      throw new Error(`HTTP ${status} received from ${response.url()}`);\n    }\n\n    const finalUrl = response.url();\n    const fullHtml = await page.content();\n    let html: string;\n\n    if (config.selector) {\n      const element = await page.$(config.selector);\n      if (!element) {\n        throw new Error(`Selector not found: ${config.selector}`);\n      }\n      html = await element.innerHTML();\n    } else {\n      html = fullHtml;\n    }\n\n    await context.close();\n\n    return { html, fullHtml, url, finalUrl };\n  } finally {\n    await browser.close();\n  }\n}\n","import { existsSync, mkdirSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport { homedir } from \"node:os\";\n\nconst CONFIG_DIR = join(homedir(), \".config\", \"vault-fetch\");\nconst SESSIONS_DIR = join(CONFIG_DIR, \"sessions\");\n\nexport function getSessionDir(): string {\n  return SESSIONS_DIR;\n}\n\nfunction extractDomain(url: string): string {\n  const parsed = new URL(url);\n  return parsed.hostname ?? \"\";\n}\n\nexport function getSessionPath(url: string, sessionsDir: string): string {\n  const domain = extractDomain(url);\n  return join(sessionsDir, `${domain}.json`);\n}\n\nexport function sessionExists(url: string, sessionsDir: string): boolean {\n  const sessionPath = getSessionPath(url, sessionsDir);\n  return existsSync(sessionPath);\n}\n\nexport function ensureSessionDir(sessionsDir: string): void {\n  if (!existsSync(sessionsDir)) {\n    mkdirSync(sessionsDir, { recursive: true });\n  }\n}\n","import { Readability } from \"@mozilla/readability\";\nimport { JSDOM } from \"jsdom\";\nimport type { Metadata } from \"./types.js\";\n\nfunction getMetaContent(doc: Document, selector: string): string | null {\n  const el = doc.querySelector(selector);\n  return el?.getAttribute(\"content\") ?? null;\n}\n\nfunction formatAuthor(raw: string): string {\n  return `[[${raw.trim()}]]`;\n}\n\nfunction extractPublishedDate(doc: Document): string | null {\n  const published =\n    getMetaContent(doc, 'meta[property=\"article:published_time\"]') ??\n    getMetaContent(doc, 'meta[name=\"datePublished\"]');\n\n  if (!published) {\n    const jsonLd = doc.querySelector('script[type=\"application/ld+json\"]');\n    if (jsonLd?.textContent) {\n      try {\n        const data = JSON.parse(jsonLd.textContent) as Record<string, unknown>;\n        if (typeof data.datePublished === \"string\") {\n          return data.datePublished.split(\"T\")[0];\n        }\n      } catch {\n        // JSON-LD parse failed\n      }\n    }\n    return null;\n  }\n\n  return published.split(\"T\")[0];\n}\n\nfunction extractAuthors(\n  doc: Document,\n  readabilityByline: string | null,\n): string[] {\n  const articleAuthors = doc.querySelectorAll('meta[property=\"article:author\"]');\n  if (articleAuthors.length > 0) {\n    return Array.from(articleAuthors)\n      .map((el) => el.getAttribute(\"content\"))\n      .filter((v): v is string => v !== null)\n      .map(formatAuthor);\n  }\n\n  const ogAuthor = getMetaContent(doc, 'meta[property=\"og:author\"]');\n  if (ogAuthor) {\n    return [formatAuthor(ogAuthor)];\n  }\n\n  if (readabilityByline) {\n    return [formatAuthor(readabilityByline)];\n  }\n\n  return [];\n}\n\nexport interface ExtractResult {\n  metadata: Metadata;\n  content: string;\n}\n\ninterface ReadabilityArticle {\n  title: string;\n  byline: string | null;\n  excerpt: string;\n  content: string;\n}\n\nfunction buildMetadata(\n  doc: Document,\n  article: ReadabilityArticle | null,\n  finalUrl: string,\n): Metadata {\n  const title = article?.title ?? doc.title;\n  const authors = extractAuthors(doc, article?.byline ?? null);\n  const published = extractPublishedDate(doc);\n\n  const description =\n    getMetaContent(doc, 'meta[property=\"og:description\"]') ??\n    getMetaContent(doc, 'meta[name=\"description\"]') ??\n    (article?.excerpt ?? null);\n\n  const today = new Date().toISOString().split(\"T\")[0];\n\n  return {\n    title,\n    source: finalUrl,\n    author: authors,\n    published,\n    created: today,\n    description,\n  };\n}\n\nfunction parseWithReadability(html: string, url: string): ReadabilityArticle | null {\n  const dom = new JSDOM(html, { url });\n  const reader = new Readability(dom.window.document);\n  return reader.parse() as ReadabilityArticle | null;\n}\n\nexport function extract(html: string, finalUrl: string): ExtractResult {\n  const metaDom = new JSDOM(html, { url: finalUrl });\n  const doc = metaDom.window.document;\n\n  const article = parseWithReadability(html, finalUrl);\n\n  if (!article) {\n    throw new Error(\n      \"Readability failed to extract content from the page. \" +\n        \"Try --raw to convert the full page, or --selector <css> to target specific content.\",\n    );\n  }\n\n  if (!article.content) {\n    throw new Error(\"Readability returned empty content for the page\");\n  }\n\n  return {\n    metadata: buildMetadata(doc, article, finalUrl),\n    content: article.content,\n  };\n}\n\nexport function extractMetadata(html: string, finalUrl: string): Metadata {\n  const metaDom = new JSDOM(html, { url: finalUrl });\n  const doc = metaDom.window.document;\n\n  const article = parseWithReadability(html, finalUrl);\n\n  return buildMetadata(doc, article, finalUrl);\n}\n","import TurndownService from \"turndown\";\n\nexport function convertToMarkdown(html: string): string {\n  const turndown = new TurndownService({\n    headingStyle: \"atx\",\n    codeBlockStyle: \"fenced\",\n    bulletListMarker: \"-\",\n  });\n\n  return turndown.turndown(html);\n}\n","import { writeFileSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport yaml from \"js-yaml\";\nimport type { Metadata } from \"./types.js\";\n\nconst UNSAFE_CHARS = /[/\\\\:*?\"<>|]/g;\nconst CONTROL_CHARS = /[\\x00-\\x1f\\x7f]/g;\nconst MAX_FILENAME_LENGTH = 200;\n\nexport function sanitizeFilename(title: string): string {\n  const sanitized = title\n    .replace(CONTROL_CHARS, \"\")\n    .replace(UNSAFE_CHARS, \"\")\n    .replace(/\\s+/g, \" \")\n    .trim();\n  const base = sanitized.slice(0, MAX_FILENAME_LENGTH) || \"Untitled\";\n  return `${base}.md`;\n}\n\nexport function buildFrontmatter(metadata: Metadata, tags: string[]): string {\n  const data: Record<string, unknown> = {\n    title: metadata.title,\n    source: metadata.source,\n  };\n\n  if (metadata.author.length > 0) {\n    data.author = metadata.author;\n  }\n\n  if (metadata.published) {\n    data.published = metadata.published;\n  }\n\n  data.created = metadata.created;\n\n  if (metadata.description) {\n    data.description = metadata.description;\n  }\n\n  data.tags = tags;\n\n  const yamlStr = yaml.dump(data, {\n    quotingType: '\"',\n    forceQuotes: false,\n    lineWidth: -1,\n    sortKeys: false,\n  });\n\n  return `---\\n${yamlStr}---`;\n}\n\nexport function writeMarkdownFile(\n  dest: string,\n  metadata: Metadata,\n  markdownContent: string,\n  tags: string[],\n): string {\n  const filename = sanitizeFilename(metadata.title);\n  const filePath = join(dest, filename);\n  const frontmatter = buildFrontmatter(metadata, tags);\n  const fullContent = `${frontmatter}\\n\\n${markdownContent}\\n`;\n\n  writeFileSync(filePath, fullContent, \"utf-8\");\n\n  return filePath;\n}\n"],"mappings":";;;AAAA,SAAS,eAAe;AACxB,SAAS,YAAY;AACrB,SAAS,cAAAA,mBAAkB;AAC3B,SAAS,WAAAC,gBAAe;AACxB,SAAS,QAAAC,aAAY;;;ACJrB,SAAS,oBAAoB;AAC7B,SAAS,eAAe;AACxB,SAAS,eAAe;AACxB,OAAO,UAAU;AAGjB,IAAM,kBAAkB;AACxB,IAAM,qBAAsC;AAC5C,IAAM,eAAe;AAwBrB,SAAS,YAAY,UAA0B;AAC7C,MAAI,SAAS,WAAW,IAAI,GAAG;AAC7B,WAAO,QAAQ,QAAQ,GAAG,SAAS,MAAM,CAAC,CAAC;AAAA,EAC7C;AACA,SAAO;AACT;AAEA,IAAM,mBAAsC,CAAC,QAAQ,oBAAoB,aAAa;AAEtF,SAAS,kBAAkB,OAAgC;AACzD,MAAI,CAAC,iBAAiB,SAAS,KAAK,GAAG;AACrC,UAAM,IAAI;AAAA,MACR,6BAA6B,KAAK,sBAAsB,iBAAiB,KAAK,IAAI,CAAC;AAAA,IACrF;AAAA,EACF;AACA,SAAO;AACT;AAEA,SAAS,eAAe,YAAgC;AACtD,QAAM,UAAU,aAAa,YAAY,OAAO;AAChD,QAAM,SAAS,KAAK,KAAK,OAAO;AAChC,MAAI,WAAW,QAAQ,OAAO,WAAW,UAAU;AACjD,UAAM,IAAI,MAAM,wBAAwB,UAAU,EAAE;AAAA,EACtD;AACA,QAAM,SAAS;AAEf,MAAI,OAAO,YAAY,UAAa,OAAO,OAAO,YAAY,UAAU;AACtE,UAAM,IAAI,MAAM,wDAAwD,OAAO,OAAO,OAAO,EAAE;AAAA,EACjG;AACA,MAAI,OAAO,SAAS,UAAa,OAAO,OAAO,SAAS,UAAU;AAChE,UAAM,IAAI,MAAM,qDAAqD,OAAO,OAAO,IAAI,EAAE;AAAA,EAC3F;AACA,MAAI,OAAO,cAAc,QAAW;AAClC,QAAI,OAAO,OAAO,cAAc,UAAU;AACxC,YAAM,IAAI,MAAM,0DAA0D,OAAO,OAAO,SAAS,EAAE;AAAA,IACrG;AACA,sBAAkB,OAAO,SAAS;AAAA,EACpC;AACA,MAAI,OAAO,SAAS,QAAW;AAC7B,QAAI,CAAC,MAAM,QAAQ,OAAO,IAAI,KAAK,CAAC,OAAO,KAAK,MAAM,CAAC,MAAe,OAAO,MAAM,QAAQ,GAAG;AAC5F,YAAM,IAAI,MAAM,wDAAwD;AAAA,IAC1E;AAAA,EACF;AAEA,SAAO;AACT;AAEO,SAAS,cACd,YACA,YACgB;AAEhB,MAAI,aAAyB,CAAC;AAC9B,MAAI,YAAY;AACd,iBAAa,eAAe,UAAU;AAAA,EACxC;AAGA,QAAM,UAAU,QAAQ,IAAI;AAC5B,QAAM,aAAa,QAAQ,IAAI;AAG/B,QAAM,OAAO,WAAW,QAAQ,WAAW,WAAW;AACtD,MAAI,SAAS,QAAW;AACtB,UAAM,IAAI;AAAA,MACR;AAAA,IACF;AAAA,EACF;AAEA,MAAI;AACJ,MAAI,WAAW,YAAY,QAAW;AACpC,cAAU,WAAW;AAAA,EACvB,WAAW,eAAe,QAAW;AACnC,UAAM,SAAS,OAAO,UAAU;AAChC,QAAI,OAAO,MAAM,MAAM,GAAG;AACxB,YAAM,IAAI,MAAM,sCAAsC,UAAU,EAAE;AAAA,IACpE;AACA,cAAU;AAAA,EACZ,OAAO;AACL,cAAU,WAAW,WAAW;AAAA,EAClC;AAEA,QAAM,eAAe,WAAW,aAAa,WAAW,aAAa;AACrE,QAAM,YAAY,kBAAkB,YAAY;AAGhD,QAAM,UAAU;AAAA,IACd,GAAI,WAAW,QAAQ,CAAC;AAAA,IACxB,GAAI,WAAW,QAAQ,CAAC;AAAA,IACxB;AAAA,EACF;AACA,QAAM,OAAO,CAAC,GAAG,IAAI,IAAI,OAAO,CAAC;AAEjC,SAAO;AAAA,IACL,MAAM,YAAY,IAAI;AAAA,IACtB;AAAA,IACA;AAAA,IACA;AAAA,IACA,QAAQ,WAAW,UAAU;AAAA,IAC7B,UAAU,WAAW,YAAY;AAAA,IACjC,WAAW,WAAW,aAAa;AAAA,IACnC,QAAQ,WAAW,UAAU;AAAA,IAC7B,aAAa,WAAW,eAAe;AAAA,IACvC,YAAY,WAAW,cAAc;AAAA,IACrC,YAAY,WAAW,cAAc;AAAA,IACrC,KAAK,WAAW,OAAO;AAAA,EACzB;AACF;;;AC3IA,SAAS,gBAAqC;;;ACA9C,SAAS,YAAY,iBAAiB;AACtC,SAAS,YAAY;AACrB,SAAS,WAAAC,gBAAe;AAExB,IAAM,aAAa,KAAKA,SAAQ,GAAG,WAAW,aAAa;AAC3D,IAAM,eAAe,KAAK,YAAY,UAAU;AAEzC,SAAS,gBAAwB;AACtC,SAAO;AACT;AAEA,SAAS,cAAc,KAAqB;AAC1C,QAAM,SAAS,IAAI,IAAI,GAAG;AAC1B,SAAO,OAAO,YAAY;AAC5B;AAEO,SAAS,eAAe,KAAa,aAA6B;AACvE,QAAM,SAAS,cAAc,GAAG;AAChC,SAAO,KAAK,aAAa,GAAG,MAAM,OAAO;AAC3C;AAEO,SAAS,cAAc,KAAa,aAA8B;AACvE,QAAM,cAAc,eAAe,KAAK,WAAW;AACnD,SAAO,WAAW,WAAW;AAC/B;AAEO,SAAS,iBAAiB,aAA2B;AAC1D,MAAI,CAAC,WAAW,WAAW,GAAG;AAC5B,cAAU,aAAa,EAAE,WAAW,KAAK,CAAC;AAAA,EAC5C;AACF;;;AD1BO,IAAM,oBACX;AAUK,SAAS,0BAA0B,SAAuC;AAC/E,QAAM,UAAU,oBAAI,IAAY;AAChC,MAAI,QAAQ,YAAa,SAAQ,IAAI,OAAO;AAC5C,MAAI,QAAQ,WAAY,SAAQ,IAAI,MAAM;AAC1C,MAAI,QAAQ,WAAY,SAAQ,IAAI,OAAO;AAC3C,SAAO;AACT;AAEA,eAAsB,UACpB,KACA,QACA,aACsB;AACtB,QAAM,UAAU,MAAM,SAAS,OAAO;AAAA,IACpC,UAAU,CAAC,OAAO;AAAA,EACpB,CAAC;AAED,MAAI;AACF,UAAM,iBAA2D;AAAA,MAC/D,WAAW;AAAA,IACb;AAGA,QAAI,CAAC,OAAO,aAAa,cAAc,KAAK,WAAW,GAAG;AACxD,YAAM,cAAc,eAAe,KAAK,WAAW;AACnD,qBAAe,eAAe;AAAA,IAChC;AAEA,UAAM,UAA0B,MAAM,QAAQ,WAAW,cAAc;AACvE,UAAM,OAAO,MAAM,QAAQ,QAAQ;AAGnC,UAAM,eAAe,0BAA0B,MAAM;AACrD,QAAI,aAAa,OAAO,GAAG;AACzB,YAAM,KAAK,MAAM,QAAQ,OAAO,UAAU;AACxC,YAAI,aAAa,IAAI,MAAM,QAAQ,EAAE,aAAa,CAAC,GAAG;AACpD,gBAAM,MAAM,MAAM;AAAA,QACpB,OAAO;AACL,gBAAM,MAAM,SAAS;AAAA,QACvB;AAAA,MACF,CAAC;AAAA,IACH;AAEA,UAAM,YAAY,OAAO,UAAU;AACnC,UAAM,WAAW,MAAM,KAAK,KAAK,KAAK;AAAA,MACpC,WAAW,OAAO;AAAA,MAClB,SAAS;AAAA,IACX,CAAC;AAED,QAAI,CAAC,UAAU;AACb,YAAM,IAAI,MAAM,6BAA6B,GAAG,EAAE;AAAA,IACpD;AAEA,UAAM,SAAS,SAAS,OAAO;AAC/B,QAAI,UAAU,KAAK;AACjB,YAAM,IAAI,MAAM,QAAQ,MAAM,kBAAkB,SAAS,IAAI,CAAC,EAAE;AAAA,IAClE;AAEA,UAAM,WAAW,SAAS,IAAI;AAC9B,UAAM,WAAW,MAAM,KAAK,QAAQ;AACpC,QAAI;AAEJ,QAAI,OAAO,UAAU;AACnB,YAAM,UAAU,MAAM,KAAK,EAAE,OAAO,QAAQ;AAC5C,UAAI,CAAC,SAAS;AACZ,cAAM,IAAI,MAAM,uBAAuB,OAAO,QAAQ,EAAE;AAAA,MAC1D;AACA,aAAO,MAAM,QAAQ,UAAU;AAAA,IACjC,OAAO;AACL,aAAO;AAAA,IACT;AAEA,UAAM,QAAQ,MAAM;AAEpB,WAAO,EAAE,MAAM,UAAU,KAAK,SAAS;AAAA,EACzC,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF;;;AE7FA,SAAS,mBAAmB;AAC5B,SAAS,aAAa;AAGtB,SAAS,eAAe,KAAe,UAAiC;AACtE,QAAM,KAAK,IAAI,cAAc,QAAQ;AACrC,SAAO,IAAI,aAAa,SAAS,KAAK;AACxC;AAEA,SAAS,aAAa,KAAqB;AACzC,SAAO,KAAK,IAAI,KAAK,CAAC;AACxB;AAEA,SAAS,qBAAqB,KAA8B;AAC1D,QAAM,YACJ,eAAe,KAAK,yCAAyC,KAC7D,eAAe,KAAK,4BAA4B;AAElD,MAAI,CAAC,WAAW;AACd,UAAM,SAAS,IAAI,cAAc,oCAAoC;AACrE,QAAI,QAAQ,aAAa;AACvB,UAAI;AACF,cAAM,OAAO,KAAK,MAAM,OAAO,WAAW;AAC1C,YAAI,OAAO,KAAK,kBAAkB,UAAU;AAC1C,iBAAO,KAAK,cAAc,MAAM,GAAG,EAAE,CAAC;AAAA,QACxC;AAAA,MACF,QAAQ;AAAA,MAER;AAAA,IACF;AACA,WAAO;AAAA,EACT;AAEA,SAAO,UAAU,MAAM,GAAG,EAAE,CAAC;AAC/B;AAEA,SAAS,eACP,KACA,mBACU;AACV,QAAM,iBAAiB,IAAI,iBAAiB,iCAAiC;AAC7E,MAAI,eAAe,SAAS,GAAG;AAC7B,WAAO,MAAM,KAAK,cAAc,EAC7B,IAAI,CAAC,OAAO,GAAG,aAAa,SAAS,CAAC,EACtC,OAAO,CAAC,MAAmB,MAAM,IAAI,EACrC,IAAI,YAAY;AAAA,EACrB;AAEA,QAAM,WAAW,eAAe,KAAK,4BAA4B;AACjE,MAAI,UAAU;AACZ,WAAO,CAAC,aAAa,QAAQ,CAAC;AAAA,EAChC;AAEA,MAAI,mBAAmB;AACrB,WAAO,CAAC,aAAa,iBAAiB,CAAC;AAAA,EACzC;AAEA,SAAO,CAAC;AACV;AAcA,SAAS,cACP,KACA,SACA,UACU;AACV,QAAM,QAAQ,SAAS,SAAS,IAAI;AACpC,QAAM,UAAU,eAAe,KAAK,SAAS,UAAU,IAAI;AAC3D,QAAM,YAAY,qBAAqB,GAAG;AAE1C,QAAM,cACJ,eAAe,KAAK,iCAAiC,KACrD,eAAe,KAAK,0BAA0B,MAC7C,SAAS,WAAW;AAEvB,QAAM,SAAQ,oBAAI,KAAK,GAAE,YAAY,EAAE,MAAM,GAAG,EAAE,CAAC;AAEnD,SAAO;AAAA,IACL;AAAA,IACA,QAAQ;AAAA,IACR,QAAQ;AAAA,IACR;AAAA,IACA,SAAS;AAAA,IACT;AAAA,EACF;AACF;AAEA,SAAS,qBAAqB,MAAc,KAAwC;AAClF,QAAM,MAAM,IAAI,MAAM,MAAM,EAAE,IAAI,CAAC;AACnC,QAAM,SAAS,IAAI,YAAY,IAAI,OAAO,QAAQ;AAClD,SAAO,OAAO,MAAM;AACtB;AAEO,SAAS,QAAQ,MAAc,UAAiC;AACrE,QAAM,UAAU,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACjD,QAAM,MAAM,QAAQ,OAAO;AAE3B,QAAM,UAAU,qBAAqB,MAAM,QAAQ;AAEnD,MAAI,CAAC,SAAS;AACZ,UAAM,IAAI;AAAA,MACR;AAAA,IAEF;AAAA,EACF;AAEA,MAAI,CAAC,QAAQ,SAAS;AACpB,UAAM,IAAI,MAAM,iDAAiD;AAAA,EACnE;AAEA,SAAO;AAAA,IACL,UAAU,cAAc,KAAK,SAAS,QAAQ;AAAA,IAC9C,SAAS,QAAQ;AAAA,EACnB;AACF;AAEO,SAAS,gBAAgB,MAAc,UAA4B;AACxE,QAAM,UAAU,IAAI,MAAM,MAAM,EAAE,KAAK,SAAS,CAAC;AACjD,QAAM,MAAM,QAAQ,OAAO;AAE3B,QAAM,UAAU,qBAAqB,MAAM,QAAQ;AAEnD,SAAO,cAAc,KAAK,SAAS,QAAQ;AAC7C;;;ACtIA,OAAO,qBAAqB;AAErB,SAAS,kBAAkB,MAAsB;AACtD,QAAM,WAAW,IAAI,gBAAgB;AAAA,IACnC,cAAc;AAAA,IACd,gBAAgB;AAAA,IAChB,kBAAkB;AAAA,EACpB,CAAC;AAED,SAAO,SAAS,SAAS,IAAI;AAC/B;;;ACVA,SAAS,qBAAqB;AAC9B,SAAS,QAAAC,aAAY;AACrB,OAAOC,WAAU;AAGjB,IAAM,eAAe;AACrB,IAAM,gBAAgB;AACtB,IAAM,sBAAsB;AAErB,SAAS,iBAAiB,OAAuB;AACtD,QAAM,YAAY,MACf,QAAQ,eAAe,EAAE,EACzB,QAAQ,cAAc,EAAE,EACxB,QAAQ,QAAQ,GAAG,EACnB,KAAK;AACR,QAAM,OAAO,UAAU,MAAM,GAAG,mBAAmB,KAAK;AACxD,SAAO,GAAG,IAAI;AAChB;AAEO,SAAS,iBAAiB,UAAoB,MAAwB;AAC3E,QAAM,OAAgC;AAAA,IACpC,OAAO,SAAS;AAAA,IAChB,QAAQ,SAAS;AAAA,EACnB;AAEA,MAAI,SAAS,OAAO,SAAS,GAAG;AAC9B,SAAK,SAAS,SAAS;AAAA,EACzB;AAEA,MAAI,SAAS,WAAW;AACtB,SAAK,YAAY,SAAS;AAAA,EAC5B;AAEA,OAAK,UAAU,SAAS;AAExB,MAAI,SAAS,aAAa;AACxB,SAAK,cAAc,SAAS;AAAA,EAC9B;AAEA,OAAK,OAAO;AAEZ,QAAM,UAAUA,MAAK,KAAK,MAAM;AAAA,IAC9B,aAAa;AAAA,IACb,aAAa;AAAA,IACb,WAAW;AAAA,IACX,UAAU;AAAA,EACZ,CAAC;AAED,SAAO;AAAA,EAAQ,OAAO;AACxB;AAEO,SAAS,kBACd,MACA,UACA,iBACA,MACQ;AACR,QAAM,WAAW,iBAAiB,SAAS,KAAK;AAChD,QAAM,WAAWD,MAAK,MAAM,QAAQ;AACpC,QAAM,cAAc,iBAAiB,UAAU,IAAI;AACnD,QAAM,cAAc,GAAG,WAAW;AAAA;AAAA,EAAO,eAAe;AAAA;AAExD,gBAAc,UAAU,aAAa,OAAO;AAE5C,SAAO;AACT;;;ANhDA,IAAM,cAAcE,MAAKC,SAAQ,GAAG,WAAW,eAAe,aAAa;AAE3E,IAAM,UAAU,IAAI,QAAQ;AAE5B,QACG,KAAK,aAAa,EAClB;AAAA,EACC;AACF,EACC,QAAQ,OAAO;AAElB,QACG,QAAQ,OAAO,EACf,YAAY,mCAAmC,EAC/C,SAAS,SAAS,cAAc,EAChC,OAAO,iBAAiB,uBAAuB,EAC/C,OAAO,YAAY,4BAA4B,EAC/C,OAAO,oBAAoB,yBAAyB,EACpD,OAAO,uBAAuB,sBAAsB,QAAQ,EAC5D,OAAO,gBAAgB,wBAAwB,CAAC,KAAa,QAAkB;AAC9E,MAAI,KAAK,GAAG;AACZ,SAAO;AACT,GAAG,CAAC,CAAa,EAChB;AAAA,EACC;AAAA,EACA;AACF,EACC,OAAO,kBAAkB,0BAA0B,EACnD,OAAO,aAAa,oCAAoC,EACxD,OAAO,qBAAqB,6BAA6B,EACzD,OAAO,oBAAoB,4BAA4B,EACvD,OAAO,oBAAoB,6BAA6B,EACxD,OAAO,SAAS,uDAAuD,EACvE,OAAO,OAAO,KAAa,YAAqC;AAC/D,MAAI;AACF,UAAM,aAAaC,YAAW,WAAW,IAAI,cAAc;AAC3D,UAAM,SAAS;AAAA,MACb;AAAA,QACE,MAAM,QAAQ;AAAA,QACd,MAAM,QAAQ;AAAA,QACd,SAAS,QAAQ;AAAA,QACjB,WAAW,QAAQ;AAAA,QACnB,QAAQ,QAAQ;AAAA,QAChB,UAAU,QAAQ;AAAA,QAClB,WAAW,QAAQ;AAAA,QACnB,QAAQ,QAAQ;AAAA,QAChB,aAAa,QAAQ;AAAA,QACrB,YAAY,QAAQ;AAAA,QACpB,YAAY,QAAQ;AAAA,QACpB,KAAK,QAAQ;AAAA,MACf;AAAA,MACA;AAAA,IACF;AAEA,QAAI,OAAO,OAAO,OAAO,UAAU;AACjC,YAAM,IAAI,MAAM,+CAA+C;AAAA,IACjE;AAGA,QAAI,CAAC,OAAO,UAAU,CAACA,YAAW,OAAO,IAAI,GAAG;AAC9C,YAAM,IAAI,MAAM,yCAAyC,OAAO,IAAI,EAAE;AAAA,IACxE;AAEA,UAAM,cAAc,cAAc;AAClC,UAAM,cAAc,MAAM,UAAU,KAAK,QAAQ,WAAW;AAE5D,QAAI;AACJ,QAAI;AAEJ,QAAI,OAAO,UAAU;AAEnB,oBAAc,YAAY;AAC1B,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AAAA,IACvE,WAAW,OAAO,KAAK;AAErB,oBAAc,YAAY;AAC1B,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AAAA,IACvE,OAAO;AACL,YAAM,SAAS,QAAQ,YAAY,MAAM,YAAY,QAAQ;AAC7D,iBAAW,OAAO;AAClB,oBAAc,OAAO;AAAA,IACvB;AAEA,UAAM,WAAW,kBAAkB,WAAW;AAE9C,QAAI,OAAO,QAAQ;AACjB,YAAM,cAAc,iBAAiB,UAAU,OAAO,IAAI;AAC1D,cAAQ,OAAO,MAAM,GAAG,WAAW;AAAA;AAAA,EAAO,QAAQ;AAAA,CAAI;AAAA,IACxD,OAAO;AACL,YAAM,WAAW;AAAA,QACf,OAAO;AAAA,QACP;AAAA,QACA;AAAA,QACA,OAAO;AAAA,MACT;AACA,cAAQ,MAAM,UAAU,QAAQ,EAAE;AAAA,IACpC;AAAA,EACF,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAQ,MAAM,UAAU,OAAO,EAAE;AACjC,YAAQ,KAAK,CAAC;AAAA,EAChB;AACF,CAAC;AAEH,QACG,QAAQ,OAAO,EACf,YAAY,kCAAkC,EAC9C,SAAS,SAAS,cAAc,EAChC,OAAO,uBAAuB,4BAA4B,QAAQ,EAClE,OAAO,OAAO,KAAa,YAAqC;AAC/D,QAAM,EAAE,UAAAC,UAAS,IAAI,MAAM,OAAO,YAAY;AAC9C,QAAM,cAAc,cAAc;AAClC,mBAAiB,WAAW;AAE5B,QAAM,aAAc,QAAQ,WAAkC;AAC9D,QAAM,UAAU,MAAMA,UAAS,OAAO,EAAE,UAAU,MAAM,CAAC;AAEzD,MAAI;AACF,UAAM,UAAU,MAAM,QAAQ,WAAW;AACzC,UAAM,OAAO,MAAM,QAAQ,QAAQ;AAEnC,UAAM,KAAK,KAAK,KAAK,EAAE,WAAW,eAAe,SAAS,aAAa,IAAK,CAAC;AAE7E,YAAQ,MAAM,yEAAyE;AAEvF,YAAQ,MAAM,OAAO;AACrB,UAAM,KAAK,QAAQ,OAAO,MAAM;AAChC,YAAQ,MAAM,MAAM;AACpB,YAAQ,MAAM,MAAM;AAEpB,UAAM,cAAc,eAAe,KAAK,WAAW;AACnD,UAAM,QAAQ,aAAa,EAAE,MAAM,YAAY,CAAC;AAChD,YAAQ,MAAM,kBAAkB,WAAW,EAAE;AAAA,EAC/C,SAAS,OAAO;AACd,UAAM,UAAU,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACrE,YAAQ,MAAM,UAAU,OAAO,EAAE;AACjC,YAAQ,KAAK,CAAC;AAAA,EAChB,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF,CAAC;AAEH,QAAQ,MAAM;","names":["existsSync","homedir","join","homedir","join","yaml","join","homedir","existsSync","chromium"]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vault-fetch",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Fetch JS-rendered web pages with Playwright and save as Markdown to Obsidian Vault",
   "type": "module",
   "license": "MIT",