Function getOutputPath

scripts/pipeline/pdf-processor.ts:79–81

(outputDir: string, dataSetId: number, fileName: string)

}

function getOutputPath(outputDir: string, dataSetId: number, fileName: string): string {
  return path.join(outputDir, `ds${dataSetId}`, `${path.basename(fileName, ".pdf")}.json`);
}

function isTextQualityPoor(text: string, pageCount: number): boolean {
  const totalChars = text.trim().length;