vault-fetch 0.3.0 → 0.3.1

package/README.ja.md

@@ -63,7 +63,9 @@ vault-fetch fetch https://example.com/report.pdf --dest ~/Documents/Obsidian/Cli
 
 サーバーが `Content-Type: application/pdf` を返す場合、vault-fetch は自動的に PDF をダウンロードし、[pdf2md](https://github.com/opendocsg/pdf2md) で Markdown に変換します。追加のフラグは不要です。
 
-- タイトルは変換された Markdown の最初の `#` 見出しから抽出（なければ URL のファイル名を使用）
+- タイトルは優先順位に従って抽出: PDF メタデータ（`dc:title` / `info.Title`）> Markdown の最初の `#` 見出し > URL のファイル名
+- 著者・公開日も PDF メタデータから抽出（存在する場合）
+- 自動抽出が不正確な場合は `--title` で手動指定可能
 - `--selector` および `--raw` オプションは PDF URL と併用不可
 - セッション機能により認証付き PDF のダウンロードにも対応
 
@@ -83,6 +85,7 @@ vault-fetch login https://note.com
 | オプション | 説明 |
 |---|---|
 | `--dest <path>` | 保存先ディレクトリ（必須） |
+| `--title <text>` | タイトル/ファイル名を手動指定 |
 | `--headed` | ブラウザを表示して実行 |
 | `--selector <css>` | CSS セレクタで要素を抽出 |
 | `--timeout <sec>` | タイムアウト秒数（デフォルト: 30） |

package/README.md

@@ -63,7 +63,9 @@ vault-fetch fetch https://example.com/report.pdf --dest ~/Documents/Obsidian/Cli
 
 When the server returns `Content-Type: application/pdf`, vault-fetch automatically downloads the PDF and converts it to Markdown using [pdf2md](https://github.com/opendocsg/pdf2md). No additional flags are needed.
 
-- Title is extracted from the first `#` heading in the converted Markdown, or from the URL filename
+- Title is extracted in priority order: PDF metadata (`dc:title` / `info.Title`) > first `#` heading in converted Markdown > URL filename
+- Author and published date are also extracted from PDF metadata when available
+- Use `--title` to manually override the title if automatic extraction is inaccurate
 - `--selector` and `--raw` options cannot be used with PDF URLs
 - Session support works with authenticated PDF downloads
 
@@ -83,6 +85,7 @@ Subsequent `fetch` commands will automatically use the saved session for that do
 | Option | Description |
 |---|---|
 | `--dest <path>` | Destination directory (required) |
+| `--title <text>` | Override the page title for the output filename |
 | `--headed` | Run with browser visible |
 | `--selector <css>` | Extract elements by CSS selector |
 | `--timeout <sec>` | Timeout in seconds (default: 30) |

package/dist/cli.js

@@ -96,6 +96,7 @@ function resolveConfig(cliOptions, configPath) {
     waitUntil,
     headed: cliOptions.headed ?? false,
     selector: cliOptions.selector ?? null,
+    title: cliOptions.title ?? null,
     noSession: cliOptions.noSession ?? false,
     dryRun: cliOptions.dryRun ?? false,
     blockImages: cliOptions.blockImages ?? true,
@@ -405,7 +406,7 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
 }, []).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").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) => {
+).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").option("--title <text>", "Override the page title for the output filename").action(async (url, options) => {
   try {
     const configPath = existsSync2(CONFIG_PATH) ? CONFIG_PATH : void 0;
     const config = resolveConfig(
@@ -416,6 +417,7 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
         waitUntil: options.waitUntil,
         headed: options.headed,
         selector: options.selector,
+        title: options.title,
         noSession: options.skipSession,
         dryRun: options.dryRun,
         blockImages: options.blockImages,
@@ -442,7 +444,7 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
       if (config.raw) {
         throw new Error("--raw cannot be used with PDF URLs.");
       }
-      const { convertPdfToMarkdown } = await import("./pdf-converter-U3SFA2HY.js");
+      const { convertPdfToMarkdown } = await import("./pdf-converter-FM6DCBO5.js");
       const pdfResult = await convertPdfToMarkdown(
         fetchResult.pdfBuffer,
         fetchResult.finalUrl
@@ -460,6 +462,9 @@ program.command("fetch").description("Fetch a page and save as Markdown").argume
       metadata = result.metadata;
       markdown = convertToMarkdown(result.content);
     }
+    if (config.title !== null) {
+      metadata = { ...metadata, title: config.title };
+    }
     if (config.dryRun) {
       const frontmatter = buildFrontmatter(metadata, config.tags);
       process.stdout.write(`${frontmatter}

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 { 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 { Metadata, 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(\"0.3.0\");\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: ${config.dest}`);\n      }\n\n      const sessionsDir = getSessionDir();\n      const fetchResult = await fetchPage(url, config, sessionsDir);\n\n      let markdown: string;\n      let metadata: Metadata;\n\n      if (fetchResult.kind === \"pdf\") {\n        if (config.selector) {\n          throw new Error(\"--selector cannot be used with PDF URLs.\");\n        }\n        if (config.raw) {\n          throw new Error(\"--raw cannot be used with PDF URLs.\");\n        }\n        const { convertPdfToMarkdown } = await import(\"./pdf-converter.js\");\n        const pdfResult = await convertPdfToMarkdown(\n          fetchResult.pdfBuffer,\n          fetchResult.finalUrl,\n        );\n        markdown = pdfResult.markdown;\n        metadata = pdfResult.metadata;\n      } else if (config.selector) {\n        // --selector mode: skip Readability, extract metadata from full page\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n        markdown = convertToMarkdown(fetchResult.html);\n      } else if (config.raw) {\n        // --raw mode: skip Readability, convert full page HTML directly\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n        markdown = convertToMarkdown(fetchResult.fullHtml);\n      } else {\n        const result = extract(fetchResult.html, fetchResult.finalUrl);\n        metadata = result.metadata;\n        markdown = convertToMarkdown(result.content);\n      }\n\n      if (config.dryRun) {\n        const frontmatter = buildFrontmatter(metadata, config.tags);\n        process.stdout.write(`${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 isPdfContentType(contentType: string): boolean {\n  return contentType.toLowerCase().includes(\"application/pdf\");\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\nconst PDF_MAGIC_BYTES = \"%PDF\";\n\nexport function validatePdfBuffer(pdfBuffer: Buffer, sourceUrl: string): void {\n  if (pdfBuffer.length === 0) {\n    throw new Error(`Empty PDF response received from ${sourceUrl}`);\n  }\n  const header = pdfBuffer.subarray(0, PDF_MAGIC_BYTES.length).toString(\"ascii\");\n  if (!header.startsWith(PDF_MAGIC_BYTES)) {\n    throw new Error(\n      `Response Content-Type is application/pdf but body is not valid PDF data from ${sourceUrl}`,\n    );\n  }\n}\n\nasync function downloadPdf(\n  context: BrowserContext,\n  url: string,\n  timeoutMs: number,\n): Promise<{ pdfBuffer: Buffer; finalUrl: string }> {\n  const apiResponse = await context.request.get(url, { timeout: timeoutMs });\n  const status = apiResponse.status();\n  if (status >= 400) {\n    throw new Error(`HTTP ${status} received when downloading PDF from ${url}`);\n  }\n  const pdfBuffer = Buffer.from(await apiResponse.body());\n  const finalUrl = apiResponse.url();\n  validatePdfBuffer(pdfBuffer, finalUrl);\n  return { pdfBuffer, finalUrl };\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\n    // page.goto throws \"Download is starting\" when the server returns\n    // Content-Disposition: attachment (common for PDF downloads).\n    // Catch this and download the PDF via the context's HTTP client.\n    let response;\n    try {\n      response = await page.goto(url, {\n        waitUntil: config.waitUntil,\n        timeout: timeoutMs,\n      });\n    } catch (error) {\n      if (\n        error instanceof Error &&\n        error.message.includes(\"Download is starting\")\n      ) {\n        const result = await downloadPdf(context, url, timeoutMs);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: result.pdfBuffer, url, finalUrl: result.finalUrl };\n      }\n      throw error;\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 ${response.url()}`);\n    }\n\n    const finalUrl = response.url();\n    const contentType = response.headers()[\"content-type\"] ?? \"\";\n\n    // Inline PDF (Content-Disposition: inline or absent).\n    // Try response.body() first; fall back to context.request if the\n    // browser returned its PDF viewer HTML instead of the actual bytes.\n    if (isPdfContentType(contentType)) {\n      const body = await response.body();\n      try {\n        validatePdfBuffer(body, finalUrl);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: body, url, finalUrl };\n      } catch {\n        // response.body() returned PDF viewer HTML; re-download via API\n        const result = await downloadPdf(context, finalUrl, timeoutMs);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: result.pdfBuffer, url, finalUrl: result.finalUrl };\n      }\n    }\n\n    const fullHtml = await page.content();\n    let html: string;\n\n    if (config.selector) {\n      const element = await page.$(config.selector);\n      if (!element) {\n        throw new Error(`Selector not found: ${config.selector}`);\n      }\n      html = await element.innerHTML();\n    } else {\n      html = fullHtml;\n    }\n\n    await context.close();\n\n    return { kind: \"html\", html, fullHtml, url, finalUrl };\n  } finally {\n    await browser.close();\n  }\n}\n","import { existsSync, mkdirSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport { homedir } from \"node:os\";\n\nconst CONFIG_DIR = join(homedir(), \".config\", \"vault-fetch\");\nconst SESSIONS_DIR = join(CONFIG_DIR, \"sessions\");\n\nexport function getSessionDir(): string {\n  return SESSIONS_DIR;\n}\n\nfunction extractDomain(url: string): string {\n  const parsed = new URL(url);\n  return parsed.hostname ?? \"\";\n}\n\nexport function getSessionPath(url: string, sessionsDir: string): string {\n  const domain = extractDomain(url);\n  return join(sessionsDir, `${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,iBAAiB,aAA8B;AAC7D,SAAO,YAAY,YAAY,EAAE,SAAS,iBAAiB;AAC7D;AAEO,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,IAAM,kBAAkB;AAEjB,SAAS,kBAAkB,WAAmB,WAAyB;AAC5E,MAAI,UAAU,WAAW,GAAG;AAC1B,UAAM,IAAI,MAAM,oCAAoC,SAAS,EAAE;AAAA,EACjE;AACA,QAAM,SAAS,UAAU,SAAS,GAAG,gBAAgB,MAAM,EAAE,SAAS,OAAO;AAC7E,MAAI,CAAC,OAAO,WAAW,eAAe,GAAG;AACvC,UAAM,IAAI;AAAA,MACR,gFAAgF,SAAS;AAAA,IAC3F;AAAA,EACF;AACF;AAEA,eAAe,YACb,SACA,KACA,WACkD;AAClD,QAAM,cAAc,MAAM,QAAQ,QAAQ,IAAI,KAAK,EAAE,SAAS,UAAU,CAAC;AACzE,QAAM,SAAS,YAAY,OAAO;AAClC,MAAI,UAAU,KAAK;AACjB,UAAM,IAAI,MAAM,QAAQ,MAAM,uCAAuC,GAAG,EAAE;AAAA,EAC5E;AACA,QAAM,YAAY,OAAO,KAAK,MAAM,YAAY,KAAK,CAAC;AACtD,QAAM,WAAW,YAAY,IAAI;AACjC,oBAAkB,WAAW,QAAQ;AACrC,SAAO,EAAE,WAAW,SAAS;AAC/B;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;AAKnC,QAAI;AACJ,QAAI;AACF,iBAAW,MAAM,KAAK,KAAK,KAAK;AAAA,QAC9B,WAAW,OAAO;AAAA,QAClB,SAAS;AAAA,MACX,CAAC;AAAA,IACH,SAAS,OAAO;AACd,UACE,iBAAiB,SACjB,MAAM,QAAQ,SAAS,sBAAsB,GAC7C;AACA,cAAM,SAAS,MAAM,YAAY,SAAS,KAAK,SAAS;AACxD,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,OAAO,WAAW,KAAK,UAAU,OAAO,SAAS;AAAA,MACpF;AACA,YAAM;AAAA,IACR;AAEA,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,cAAc,SAAS,QAAQ,EAAE,cAAc,KAAK;AAK1D,QAAI,iBAAiB,WAAW,GAAG;AACjC,YAAM,OAAO,MAAM,SAAS,KAAK;AACjC,UAAI;AACF,0BAAkB,MAAM,QAAQ;AAChC,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,MAAM,KAAK,SAAS;AAAA,MACvD,QAAQ;AAEN,cAAM,SAAS,MAAM,YAAY,SAAS,UAAU,SAAS;AAC7D,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,OAAO,WAAW,KAAK,UAAU,OAAO,SAAS;AAAA,MACpF;AAAA,IACF;AAEA,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,QAAQ,MAAM,UAAU,KAAK,SAAS;AAAA,EACvD,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF;;;AEnKA,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,YAAY,SAAS,OAAO;AAC9B,UAAI,OAAO,UAAU;AACnB,cAAM,IAAI,MAAM,0CAA0C;AAAA,MAC5D;AACA,UAAI,OAAO,KAAK;AACd,cAAM,IAAI,MAAM,qCAAqC;AAAA,MACvD;AACA,YAAM,EAAE,qBAAqB,IAAI,MAAM,OAAO,6BAAoB;AAClE,YAAM,YAAY,MAAM;AAAA,QACtB,YAAY;AAAA,QACZ,YAAY;AAAA,MACd;AACA,iBAAW,UAAU;AACrB,iBAAW,UAAU;AAAA,IACvB,WAAW,OAAO,UAAU;AAE1B,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AACrE,iBAAW,kBAAkB,YAAY,IAAI;AAAA,IAC/C,WAAW,OAAO,KAAK;AAErB,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AACrE,iBAAW,kBAAkB,YAAY,QAAQ;AAAA,IACnD,OAAO;AACL,YAAM,SAAS,QAAQ,YAAY,MAAM,YAAY,QAAQ;AAC7D,iBAAW,OAAO;AAClB,iBAAW,kBAAkB,OAAO,OAAO;AAAA,IAC7C;AAEA,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"]}
+{"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 { Metadata, 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(\"0.3.0\");\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  .option(\"--title <text>\", \"Override the page title for the output filename\")\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          title: options.title 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: ${config.dest}`);\n      }\n\n      const sessionsDir = getSessionDir();\n      const fetchResult = await fetchPage(url, config, sessionsDir);\n\n      let markdown: string;\n      let metadata: Metadata;\n\n      if (fetchResult.kind === \"pdf\") {\n        if (config.selector) {\n          throw new Error(\"--selector cannot be used with PDF URLs.\");\n        }\n        if (config.raw) {\n          throw new Error(\"--raw cannot be used with PDF URLs.\");\n        }\n        const { convertPdfToMarkdown } = await import(\"./pdf-converter.js\");\n        const pdfResult = await convertPdfToMarkdown(\n          fetchResult.pdfBuffer,\n          fetchResult.finalUrl,\n        );\n        markdown = pdfResult.markdown;\n        metadata = pdfResult.metadata;\n      } else if (config.selector) {\n        // --selector mode: skip Readability, extract metadata from full page\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n        markdown = convertToMarkdown(fetchResult.html);\n      } else if (config.raw) {\n        // --raw mode: skip Readability, convert full page HTML directly\n        metadata = extractMetadata(fetchResult.fullHtml, fetchResult.finalUrl);\n        markdown = convertToMarkdown(fetchResult.fullHtml);\n      } else {\n        const result = extract(fetchResult.html, fetchResult.finalUrl);\n        metadata = result.metadata;\n        markdown = convertToMarkdown(result.content);\n      }\n\n      if (config.title !== null) {\n        metadata = { ...metadata, title: config.title };\n      }\n\n      if (config.dryRun) {\n        const frontmatter = buildFrontmatter(metadata, config.tags);\n        process.stdout.write(`${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: ${message}`);\n      process.exit(1);\n    } finally {\n      await browser.close();\n    }\n  });\n\nprogram.parse();\n","import { readFileSync } from \"node:fs\";\nimport { homedir } from \"node:os\";\nimport { resolve } from \"node:path\";\nimport yaml from \"js-yaml\";\nimport type { ResolvedConfig, WaitUntilOption } from \"./types.js\";\n\nconst DEFAULT_TIMEOUT = 30;\nconst DEFAULT_WAIT_UNTIL: WaitUntilOption = \"networkidle\";\nconst REQUIRED_TAG = \"clippings\";\n\ninterface FileConfig {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n}\n\ninterface CliOptions {\n  dest?: string;\n  tags?: string[];\n  timeout?: number;\n  waitUntil?: WaitUntilOption;\n  headed?: boolean;\n  selector?: string;\n  title?: string;\n  noSession?: boolean;\n  dryRun?: boolean;\n  blockImages?: boolean;\n  blockFonts?: boolean;\n  blockMedia?: boolean;\n  raw?: boolean;\n}\n\nfunction expandTilde(filePath: string): string {\n  if (filePath.startsWith(\"~/\")) {\n    return resolve(homedir(), filePath.slice(2));\n  }\n  return filePath;\n}\n\nconst VALID_WAIT_UNTIL: readonly string[] = [\"load\", \"domcontentloaded\", \"networkidle\"];\n\nfunction validateWaitUntil(value: string): WaitUntilOption {\n  if (!VALID_WAIT_UNTIL.includes(value)) {\n    throw new Error(\n      `Invalid waitUntil value: \"${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    title: cliOptions.title ?? 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 isPdfContentType(contentType: string): boolean {\n  return contentType.toLowerCase().includes(\"application/pdf\");\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\nconst PDF_MAGIC_BYTES = \"%PDF\";\n\nexport function validatePdfBuffer(pdfBuffer: Buffer, sourceUrl: string): void {\n  if (pdfBuffer.length === 0) {\n    throw new Error(`Empty PDF response received from ${sourceUrl}`);\n  }\n  const header = pdfBuffer.subarray(0, PDF_MAGIC_BYTES.length).toString(\"ascii\");\n  if (!header.startsWith(PDF_MAGIC_BYTES)) {\n    throw new Error(\n      `Response Content-Type is application/pdf but body is not valid PDF data from ${sourceUrl}`,\n    );\n  }\n}\n\nasync function downloadPdf(\n  context: BrowserContext,\n  url: string,\n  timeoutMs: number,\n): Promise<{ pdfBuffer: Buffer; finalUrl: string }> {\n  const apiResponse = await context.request.get(url, { timeout: timeoutMs });\n  const status = apiResponse.status();\n  if (status >= 400) {\n    throw new Error(`HTTP ${status} received when downloading PDF from ${url}`);\n  }\n  const pdfBuffer = Buffer.from(await apiResponse.body());\n  const finalUrl = apiResponse.url();\n  validatePdfBuffer(pdfBuffer, finalUrl);\n  return { pdfBuffer, finalUrl };\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\n    // page.goto throws \"Download is starting\" when the server returns\n    // Content-Disposition: attachment (common for PDF downloads).\n    // Catch this and download the PDF via the context's HTTP client.\n    let response;\n    try {\n      response = await page.goto(url, {\n        waitUntil: config.waitUntil,\n        timeout: timeoutMs,\n      });\n    } catch (error) {\n      if (\n        error instanceof Error &&\n        error.message.includes(\"Download is starting\")\n      ) {\n        const result = await downloadPdf(context, url, timeoutMs);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: result.pdfBuffer, url, finalUrl: result.finalUrl };\n      }\n      throw error;\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 ${response.url()}`);\n    }\n\n    const finalUrl = response.url();\n    const contentType = response.headers()[\"content-type\"] ?? \"\";\n\n    // Inline PDF (Content-Disposition: inline or absent).\n    // Try response.body() first; fall back to context.request if the\n    // browser returned its PDF viewer HTML instead of the actual bytes.\n    if (isPdfContentType(contentType)) {\n      const body = await response.body();\n      try {\n        validatePdfBuffer(body, finalUrl);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: body, url, finalUrl };\n      } catch {\n        // response.body() returned PDF viewer HTML; re-download via API\n        const result = await downloadPdf(context, finalUrl, timeoutMs);\n        await context.close();\n        return { kind: \"pdf\", pdfBuffer: result.pdfBuffer, url, finalUrl: result.finalUrl };\n      }\n    }\n\n    const fullHtml = await page.content();\n    let html: string;\n\n    if (config.selector) {\n      const element = await page.$(config.selector);\n      if (!element) {\n        throw new Error(`Selector not found: ${config.selector}`);\n      }\n      html = await element.innerHTML();\n    } else {\n      html = fullHtml;\n    }\n\n    await context.close();\n\n    return { kind: \"html\", html, fullHtml, url, finalUrl };\n  } finally {\n    await browser.close();\n  }\n}\n","import { existsSync, mkdirSync } from \"node:fs\";\nimport { join } from \"node:path\";\nimport { homedir } from \"node:os\";\n\nconst CONFIG_DIR = join(homedir(), \".config\", \"vault-fetch\");\nconst SESSIONS_DIR = join(CONFIG_DIR, \"sessions\");\n\nexport function getSessionDir(): string {\n  return SESSIONS_DIR;\n}\n\nfunction extractDomain(url: string): string {\n  const parsed = new URL(url);\n  return parsed.hostname ?? \"\";\n}\n\nexport function getSessionPath(url: string, sessionsDir: string): string {\n  const domain = extractDomain(url);\n  return join(sessionsDir, `${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;AAyBrB,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,OAAO,WAAW,SAAS;AAAA,IAC3B,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;;;AC7IA,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,iBAAiB,aAA8B;AAC7D,SAAO,YAAY,YAAY,EAAE,SAAS,iBAAiB;AAC7D;AAEO,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,IAAM,kBAAkB;AAEjB,SAAS,kBAAkB,WAAmB,WAAyB;AAC5E,MAAI,UAAU,WAAW,GAAG;AAC1B,UAAM,IAAI,MAAM,oCAAoC,SAAS,EAAE;AAAA,EACjE;AACA,QAAM,SAAS,UAAU,SAAS,GAAG,gBAAgB,MAAM,EAAE,SAAS,OAAO;AAC7E,MAAI,CAAC,OAAO,WAAW,eAAe,GAAG;AACvC,UAAM,IAAI;AAAA,MACR,gFAAgF,SAAS;AAAA,IAC3F;AAAA,EACF;AACF;AAEA,eAAe,YACb,SACA,KACA,WACkD;AAClD,QAAM,cAAc,MAAM,QAAQ,QAAQ,IAAI,KAAK,EAAE,SAAS,UAAU,CAAC;AACzE,QAAM,SAAS,YAAY,OAAO;AAClC,MAAI,UAAU,KAAK;AACjB,UAAM,IAAI,MAAM,QAAQ,MAAM,uCAAuC,GAAG,EAAE;AAAA,EAC5E;AACA,QAAM,YAAY,OAAO,KAAK,MAAM,YAAY,KAAK,CAAC;AACtD,QAAM,WAAW,YAAY,IAAI;AACjC,oBAAkB,WAAW,QAAQ;AACrC,SAAO,EAAE,WAAW,SAAS;AAC/B;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;AAKnC,QAAI;AACJ,QAAI;AACF,iBAAW,MAAM,KAAK,KAAK,KAAK;AAAA,QAC9B,WAAW,OAAO;AAAA,QAClB,SAAS;AAAA,MACX,CAAC;AAAA,IACH,SAAS,OAAO;AACd,UACE,iBAAiB,SACjB,MAAM,QAAQ,SAAS,sBAAsB,GAC7C;AACA,cAAM,SAAS,MAAM,YAAY,SAAS,KAAK,SAAS;AACxD,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,OAAO,WAAW,KAAK,UAAU,OAAO,SAAS;AAAA,MACpF;AACA,YAAM;AAAA,IACR;AAEA,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,cAAc,SAAS,QAAQ,EAAE,cAAc,KAAK;AAK1D,QAAI,iBAAiB,WAAW,GAAG;AACjC,YAAM,OAAO,MAAM,SAAS,KAAK;AACjC,UAAI;AACF,0BAAkB,MAAM,QAAQ;AAChC,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,MAAM,KAAK,SAAS;AAAA,MACvD,QAAQ;AAEN,cAAM,SAAS,MAAM,YAAY,SAAS,UAAU,SAAS;AAC7D,cAAM,QAAQ,MAAM;AACpB,eAAO,EAAE,MAAM,OAAO,WAAW,OAAO,WAAW,KAAK,UAAU,OAAO,SAAS;AAAA,MACpF;AAAA,IACF;AAEA,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,QAAQ,MAAM,UAAU,KAAK,SAAS;AAAA,EACvD,UAAE;AACA,UAAM,QAAQ,MAAM;AAAA,EACtB;AACF;;;AEnKA,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,kBAAkB,iDAAiD,EAC1E,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,OAAO,QAAQ;AAAA,QACf,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,YAAY,SAAS,OAAO;AAC9B,UAAI,OAAO,UAAU;AACnB,cAAM,IAAI,MAAM,0CAA0C;AAAA,MAC5D;AACA,UAAI,OAAO,KAAK;AACd,cAAM,IAAI,MAAM,qCAAqC;AAAA,MACvD;AACA,YAAM,EAAE,qBAAqB,IAAI,MAAM,OAAO,6BAAoB;AAClE,YAAM,YAAY,MAAM;AAAA,QACtB,YAAY;AAAA,QACZ,YAAY;AAAA,MACd;AACA,iBAAW,UAAU;AACrB,iBAAW,UAAU;AAAA,IACvB,WAAW,OAAO,UAAU;AAE1B,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AACrE,iBAAW,kBAAkB,YAAY,IAAI;AAAA,IAC/C,WAAW,OAAO,KAAK;AAErB,iBAAW,gBAAgB,YAAY,UAAU,YAAY,QAAQ;AACrE,iBAAW,kBAAkB,YAAY,QAAQ;AAAA,IACnD,OAAO;AACL,YAAM,SAAS,QAAQ,YAAY,MAAM,YAAY,QAAQ;AAC7D,iBAAW,OAAO;AAClB,iBAAW,kBAAkB,OAAO,OAAO;AAAA,IAC7C;AAEA,QAAI,OAAO,UAAU,MAAM;AACzB,iBAAW,EAAE,GAAG,UAAU,OAAO,OAAO,MAAM;AAAA,IAChD;AAEA,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/dist/pdf-converter-FM6DCBO5.js

@@ -0,0 +1,69 @@
+#!/usr/bin/env node
+
+// src/pdf-converter.ts
+import pdf2md from "@opendocsg/pdf2md";
+var PDF_DATE_PATTERN = /^D:(\d{4})(\d{2})(\d{2})/;
+function parsePdfDate(dateStr) {
+  const match = dateStr.match(PDF_DATE_PATTERN);
+  if (!match) return null;
+  const [, year, month, day] = match;
+  return `${year}-${month}-${day}`;
+}
+var AUTHOR_SEPARATOR = /\s*(?:;|\s+and\s+|&)\s*/;
+function formatAuthor(raw) {
+  const sanitized = raw.trim().replace(/[[\]]/g, "");
+  return `[[${sanitized}]]`;
+}
+function parseAuthors(raw) {
+  return raw.split(AUTHOR_SEPARATOR).map((s) => s.trim()).filter(Boolean).map(formatAuthor);
+}
+async function convertPdfToMarkdown(pdfBuffer, sourceUrl) {
+  let pdfMeta = null;
+  const markdown = await pdf2md(pdfBuffer, {
+    metadataParsed: (metadata) => {
+      pdfMeta = metadata;
+    }
+  });
+  if (!markdown.trim()) {
+    throw new Error("pdf2md returned empty content from the PDF");
+  }
+  const xmpTitle = pdfMeta?.metadata?.get("dc:title");
+  const infoTitle = pdfMeta?.info.Title;
+  const title = (xmpTitle && xmpTitle.trim() ? xmpTitle.trim() : null) ?? (infoTitle && infoTitle.trim() ? infoTitle.trim() : null) ?? extractTitleFromMarkdown(markdown) ?? extractTitleFromUrl(sourceUrl);
+  const rawAuthor = pdfMeta?.info.Author;
+  const author = rawAuthor && rawAuthor.trim() ? parseAuthors(rawAuthor) : [];
+  const rawDate = pdfMeta?.info.CreationDate;
+  const published = rawDate ? parsePdfDate(rawDate) : null;
+  const today = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
+  return {
+    markdown,
+    metadata: {
+      title,
+      source: sourceUrl,
+      author,
+      published,
+      created: today,
+      description: null
+    }
+  };
+}
+function extractTitleFromMarkdown(markdown) {
+  const match = markdown.match(/^#\s+(.+?)(?:\s+#+)?$/m);
+  if (!match) return null;
+  const trimmed = match[1].trim();
+  return trimmed || null;
+}
+function extractTitleFromUrl(url) {
+  const pathname = new URL(url).pathname;
+  const lastSegment = pathname.split("/").filter(Boolean).pop();
+  if (!lastSegment) {
+    throw new Error(`Cannot extract filename from URL: ${url}`);
+  }
+  return decodeURIComponent(lastSegment.replace(/\.pdf$/i, ""));
+}
+export {
+  convertPdfToMarkdown,
+  extractTitleFromUrl,
+  parsePdfDate
+};
+//# sourceMappingURL=pdf-converter-FM6DCBO5.js.map

package/dist/pdf-converter-FM6DCBO5.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/pdf-converter.ts"],"sourcesContent":["import pdf2md from \"@opendocsg/pdf2md\";\nimport type { Metadata } from \"./types.js\";\n\nexport interface PdfConvertResult {\n  markdown: string;\n  metadata: Metadata;\n}\n\ninterface PdfMetadataInfo {\n  Title?: string;\n  Author?: string;\n  Creator?: string;\n  CreationDate?: string;\n}\n\ninterface PdfRawMetadata {\n  info: PdfMetadataInfo;\n  metadata: { get: (name: string) => string | null } | null;\n}\n\nconst PDF_DATE_PATTERN = /^D:(\\d{4})(\\d{2})(\\d{2})/;\n\nexport function parsePdfDate(dateStr: string): string | null {\n  const match = dateStr.match(PDF_DATE_PATTERN);\n  if (!match) return null;\n  const [, year, month, day] = match;\n  return `${year}-${month}-${day}`;\n}\n\nconst AUTHOR_SEPARATOR = /\\s*(?:;|\\s+and\\s+|&)\\s*/;\n\nfunction formatAuthor(raw: string): string {\n  const sanitized = raw.trim().replace(/[[\\]]/g, \"\");\n  return `[[${sanitized}]]`;\n}\n\nfunction parseAuthors(raw: string): string[] {\n  return raw\n    .split(AUTHOR_SEPARATOR)\n    .map((s) => s.trim())\n    .filter(Boolean)\n    .map(formatAuthor);\n}\n\nexport async function convertPdfToMarkdown(\n  pdfBuffer: Buffer,\n  sourceUrl: string,\n): Promise<PdfConvertResult> {\n  // TypeScript narrows this to `null` without the assertion because\n  // it cannot see that metadataParsed is called synchronously inside pdf2md.\n  let pdfMeta = null as PdfRawMetadata | null;\n\n  const markdown = await pdf2md(pdfBuffer, {\n    metadataParsed: (metadata) => {\n      pdfMeta = metadata as PdfRawMetadata;\n    },\n  });\n\n  if (!markdown.trim()) {\n    throw new Error(\"pdf2md returned empty content from the PDF\");\n  }\n\n  // Title priority: XMP dc:title > info.Title > Markdown H1 > URL segment\n  const xmpTitle = pdfMeta?.metadata?.get(\"dc:title\");\n  const infoTitle = pdfMeta?.info.Title;\n  const title =\n    (xmpTitle && xmpTitle.trim() ? xmpTitle.trim() : null) ??\n    (infoTitle && infoTitle.trim() ? infoTitle.trim() : null) ??\n    extractTitleFromMarkdown(markdown) ??\n    extractTitleFromUrl(sourceUrl);\n\n  // Author\n  const rawAuthor = pdfMeta?.info.Author;\n  const author = rawAuthor && rawAuthor.trim() ? parseAuthors(rawAuthor) : [];\n\n  // Published\n  const rawDate = pdfMeta?.info.CreationDate;\n  const published = rawDate ? parsePdfDate(rawDate) : null;\n\n  const today = new Date().toISOString().split(\"T\")[0];\n\n  return {\n    markdown,\n    metadata: {\n      title,\n      source: sourceUrl,\n      author,\n      published,\n      created: today,\n      description: null,\n    },\n  };\n}\n\nfunction extractTitleFromMarkdown(markdown: string): string | null {\n  const match = markdown.match(/^#\\s+(.+?)(?:\\s+#+)?$/m);\n  if (!match) return null;\n  const trimmed = match[1].trim();\n  return trimmed || null;\n}\n\nexport function extractTitleFromUrl(url: string): string {\n  const pathname = new URL(url).pathname;\n  const lastSegment = pathname.split(\"/\").filter(Boolean).pop();\n  if (!lastSegment) {\n    throw new Error(`Cannot extract filename from URL: ${url}`);\n  }\n  return decodeURIComponent(lastSegment.replace(/\\.pdf$/i, \"\"));\n}\n"],"mappings":";;;AAAA,OAAO,YAAY;AAoBnB,IAAM,mBAAmB;AAElB,SAAS,aAAa,SAAgC;AAC3D,QAAM,QAAQ,QAAQ,MAAM,gBAAgB;AAC5C,MAAI,CAAC,MAAO,QAAO;AACnB,QAAM,CAAC,EAAE,MAAM,OAAO,GAAG,IAAI;AAC7B,SAAO,GAAG,IAAI,IAAI,KAAK,IAAI,GAAG;AAChC;AAEA,IAAM,mBAAmB;AAEzB,SAAS,aAAa,KAAqB;AACzC,QAAM,YAAY,IAAI,KAAK,EAAE,QAAQ,UAAU,EAAE;AACjD,SAAO,KAAK,SAAS;AACvB;AAEA,SAAS,aAAa,KAAuB;AAC3C,SAAO,IACJ,MAAM,gBAAgB,EACtB,IAAI,CAAC,MAAM,EAAE,KAAK,CAAC,EACnB,OAAO,OAAO,EACd,IAAI,YAAY;AACrB;AAEA,eAAsB,qBACpB,WACA,WAC2B;AAG3B,MAAI,UAAU;AAEd,QAAM,WAAW,MAAM,OAAO,WAAW;AAAA,IACvC,gBAAgB,CAAC,aAAa;AAC5B,gBAAU;AAAA,IACZ;AAAA,EACF,CAAC;AAED,MAAI,CAAC,SAAS,KAAK,GAAG;AACpB,UAAM,IAAI,MAAM,4CAA4C;AAAA,EAC9D;AAGA,QAAM,WAAW,SAAS,UAAU,IAAI,UAAU;AAClD,QAAM,YAAY,SAAS,KAAK;AAChC,QAAM,SACH,YAAY,SAAS,KAAK,IAAI,SAAS,KAAK,IAAI,UAChD,aAAa,UAAU,KAAK,IAAI,UAAU,KAAK,IAAI,SACpD,yBAAyB,QAAQ,KACjC,oBAAoB,SAAS;AAG/B,QAAM,YAAY,SAAS,KAAK;AAChC,QAAM,SAAS,aAAa,UAAU,KAAK,IAAI,aAAa,SAAS,IAAI,CAAC;AAG1E,QAAM,UAAU,SAAS,KAAK;AAC9B,QAAM,YAAY,UAAU,aAAa,OAAO,IAAI;AAEpD,QAAM,SAAQ,oBAAI,KAAK,GAAE,YAAY,EAAE,MAAM,GAAG,EAAE,CAAC;AAEnD,SAAO;AAAA,IACL;AAAA,IACA,UAAU;AAAA,MACR;AAAA,MACA,QAAQ;AAAA,MACR;AAAA,MACA;AAAA,MACA,SAAS;AAAA,MACT,aAAa;AAAA,IACf;AAAA,EACF;AACF;AAEA,SAAS,yBAAyB,UAAiC;AACjE,QAAM,QAAQ,SAAS,MAAM,wBAAwB;AACrD,MAAI,CAAC,MAAO,QAAO;AACnB,QAAM,UAAU,MAAM,CAAC,EAAE,KAAK;AAC9B,SAAO,WAAW;AACpB;AAEO,SAAS,oBAAoB,KAAqB;AACvD,QAAM,WAAW,IAAI,IAAI,GAAG,EAAE;AAC9B,QAAM,cAAc,SAAS,MAAM,GAAG,EAAE,OAAO,OAAO,EAAE,IAAI;AAC5D,MAAI,CAAC,aAAa;AAChB,UAAM,IAAI,MAAM,qCAAqC,GAAG,EAAE;AAAA,EAC5D;AACA,SAAO,mBAAmB,YAAY,QAAQ,WAAW,EAAE,CAAC;AAC9D;","names":[]}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vault-fetch",
-  "version": "0.3.0",
+  "version": "0.3.1",
   "description": "Fetch JS-rendered web pages with Playwright and save as Markdown to Obsidian Vault",
   "type": "module",
   "license": "MIT",

package/dist/pdf-converter-U3SFA2HY.js

@@ -1,42 +0,0 @@
-#!/usr/bin/env node
-
-// src/pdf-converter.ts
-import pdf2md from "@opendocsg/pdf2md";
-async function convertPdfToMarkdown(pdfBuffer, sourceUrl) {
-  const markdown = await pdf2md(pdfBuffer);
-  if (!markdown.trim()) {
-    throw new Error("pdf2md returned empty content from the PDF");
-  }
-  const title = extractTitleFromMarkdown(markdown) ?? extractTitleFromUrl(sourceUrl);
-  const today = (/* @__PURE__ */ new Date()).toISOString().split("T")[0];
-  return {
-    markdown,
-    metadata: {
-      title,
-      source: sourceUrl,
-      author: [],
-      published: null,
-      created: today,
-      description: null
-    }
-  };
-}
-function extractTitleFromMarkdown(markdown) {
-  const match = markdown.match(/^#\s+(.+?)(?:\s+#+)?$/m);
-  if (!match) return null;
-  const trimmed = match[1].trim();
-  return trimmed || null;
-}
-function extractTitleFromUrl(url) {
-  const pathname = new URL(url).pathname;
-  const lastSegment = pathname.split("/").filter(Boolean).pop();
-  if (!lastSegment) {
-    throw new Error(`Cannot extract filename from URL: ${url}`);
-  }
-  return decodeURIComponent(lastSegment.replace(/\.pdf$/i, ""));
-}
-export {
-  convertPdfToMarkdown,
-  extractTitleFromUrl
-};
-//# sourceMappingURL=pdf-converter-U3SFA2HY.js.map

package/dist/pdf-converter-U3SFA2HY.js.map

@@ -1 +0,0 @@
-{"version":3,"sources":["../src/pdf-converter.ts"],"sourcesContent":["import pdf2md from \"@opendocsg/pdf2md\";\nimport type { Metadata } from \"./types.js\";\n\nexport interface PdfConvertResult {\n  markdown: string;\n  metadata: Metadata;\n}\n\nexport async function convertPdfToMarkdown(\n  pdfBuffer: Buffer,\n  sourceUrl: string,\n): Promise<PdfConvertResult> {\n  const markdown = await pdf2md(pdfBuffer);\n  if (!markdown.trim()) {\n    throw new Error(\"pdf2md returned empty content from the PDF\");\n  }\n  const title = extractTitleFromMarkdown(markdown) ?? extractTitleFromUrl(sourceUrl);\n  const today = new Date().toISOString().split(\"T\")[0];\n\n  return {\n    markdown,\n    metadata: {\n      title,\n      source: sourceUrl,\n      author: [],\n      published: null,\n      created: today,\n      description: null,\n    },\n  };\n}\n\nfunction extractTitleFromMarkdown(markdown: string): string | null {\n  const match = markdown.match(/^#\\s+(.+?)(?:\\s+#+)?$/m);\n  if (!match) return null;\n  const trimmed = match[1].trim();\n  return trimmed || null;\n}\n\nexport function extractTitleFromUrl(url: string): string {\n  const pathname = new URL(url).pathname;\n  const lastSegment = pathname.split(\"/\").filter(Boolean).pop();\n  if (!lastSegment) {\n    throw new Error(`Cannot extract filename from URL: ${url}`);\n  }\n  return decodeURIComponent(lastSegment.replace(/\\.pdf$/i, \"\"));\n}\n"],"mappings":";;;AAAA,OAAO,YAAY;AAQnB,eAAsB,qBACpB,WACA,WAC2B;AAC3B,QAAM,WAAW,MAAM,OAAO,SAAS;AACvC,MAAI,CAAC,SAAS,KAAK,GAAG;AACpB,UAAM,IAAI,MAAM,4CAA4C;AAAA,EAC9D;AACA,QAAM,QAAQ,yBAAyB,QAAQ,KAAK,oBAAoB,SAAS;AACjF,QAAM,SAAQ,oBAAI,KAAK,GAAE,YAAY,EAAE,MAAM,GAAG,EAAE,CAAC;AAEnD,SAAO;AAAA,IACL;AAAA,IACA,UAAU;AAAA,MACR;AAAA,MACA,QAAQ;AAAA,MACR,QAAQ,CAAC;AAAA,MACT,WAAW;AAAA,MACX,SAAS;AAAA,MACT,aAAa;AAAA,IACf;AAAA,EACF;AACF;AAEA,SAAS,yBAAyB,UAAiC;AACjE,QAAM,QAAQ,SAAS,MAAM,wBAAwB;AACrD,MAAI,CAAC,MAAO,QAAO;AACnB,QAAM,UAAU,MAAM,CAAC,EAAE,KAAK;AAC9B,SAAO,WAAW;AACpB;AAEO,SAAS,oBAAoB,KAAqB;AACvD,QAAM,WAAW,IAAI,IAAI,GAAG,EAAE;AAC9B,QAAM,cAAc,SAAS,MAAM,GAAG,EAAE,OAAO,OAAO,EAAE,IAAI;AAC5D,MAAI,CAAC,aAAa;AAChB,UAAM,IAAI,MAAM,qCAAqC,GAAG,EAAE;AAAA,EAC5D;AACA,SAAO,mBAAmB,YAAY,QAAQ,WAAW,EAAE,CAAC;AAC9D;","names":[]}