PDF 转 JSON

The full object structure of the PDF as cpdf reads it: the trailer, catalog, page tree, every dictionary, and each page's content stream parsed into a list of drawing and text operators. It's a complete dump of the file, not a summary.

Usually not. The text here is scattered across content stream operators in the order it was drawn, with fonts and positions but no paragraphs. For readable page text in a structured format use Prepare PDF for AI, and for plain text use PDF to Text.

Not with BentoPDF. The output follows cpdf's own JSON format, which the cpdf command line tool can read back into a PDF, but the browser tool only goes in one direction.

Content streams are expanded into operator arrays, binary stream data such as images is encoded as text, and every object is written out with its keys, so a JSON file several times the size of the source is normal. Large or image-heavy PDFs produce very large files.

Select as many as you like; each is converted in a background worker and the results are packaged into pdfs-to-json.zip with one .json per PDF, named after the source file. A single PDF also arrives as a ZIP.

You're asked for the password before conversion starts and the decrypted copy is serialized. Without the password the object structure can't be read.

Inspecting a PDF's internals when debugging a generator, checking what a page's content stream contains, comparing two versions structurally, or feeding the object tree into a script. For document properties alone, View Metadata is quicker.

No. cpdf runs as WebAssembly in a web worker in your browser, and the JSON is written and zipped on your device before download.