Function extractPDFPages

source code/utils/pdf.ts:181–302

(
  filePath: string,
  options?: { firstPage?: number; lastPage?: number },
)

 * @param options Optional page range (1-indexed, inclusive)
 */
export async function extractPDFPages(
  filePath: string,
  options?: { firstPage?: number; lastPage?: number },
): Promise<PDFResult<PDFExtractPagesResult>> {
  try {
    const fs = getFsImplementation()
    const stats = await fs.stat(filePath)
    const originalSize = stats.size

    if (originalSize === 0) {
      return {
        success: false,
        error: { reason: 'empty', message: `PDF file is empty: ${filePath}` },
      }
    }

    if (originalSize > PDF_MAX_EXTRACT_SIZE) {
      return {
        success: false,
        error: {
          reason: 'too_large',
          message: `PDF file exceeds maximum allowed size for text extraction (${formatFileSize(PDF_MAX_EXTRACT_SIZE)}).`,
        },
      }
    }

    const available = await isPdftoppmAvailable()
    if (!available) {
      return {
        success: false,
        error: {
          reason: 'unavailable',
          message:
            'pdftoppm is not installed. Install poppler-utils (e.g. `brew install poppler` or `apt-get install poppler-utils`) to enable PDF page rendering.',
        },
      }
    }

    const uuid = randomUUID()
    const outputDir = join(getToolResultsDir(), `pdf-${uuid}`)
    await mkdir(outputDir, { recursive: true })

    // pdftoppm produces files like <prefix>-01.jpg, <prefix>-02.jpg, etc.
    const prefix = join(outputDir, 'page')
    const args = ['-jpeg', '-r', '100']
    if (options?.firstPage) {
      args.push('-f', String(options.firstPage))
    }
    if (options?.lastPage && options.lastPage !== Infinity) {
      args.push('-l', String(options.lastPage))
    }
    args.push(filePath, prefix)
    const { code, stderr } = await execFileNoThrow('pdftoppm', args, {
      timeout: 120_000,
      useCwd: false,
    })

    if (code !== 0) {