FL Bulk PDF Loader

PDFs in ComfyUI? It's a stranger sight than you'd expect, but Fill-Nodes has a whole corner devoted to document handling, and FL_BulkPDFLoader is the front door for the batch case. Point it at a directory and it loads every PDF in there at once, handing back a list of PDF objects that the pack's other PDF nodes know how to open, read, and convert. If your workflow involves pulling text or images out of a stack of documents - think building a caption or prompt corpus from a pile of reference PDFs, or bulk-converting pages to images - this is where you start.

How it works

You give it a directory_path and it walks the folder, reading each PDF into an internal PDF data object that carries the file's path, filename, page count, and raw binary content. It validates that each one exists and can actually be read, collecting errors for any that fail so a single corrupt file doesn't kill the whole batch. The output is a list of those objects - one per document - which you then feed into a downstream PDF node like FL_PDFTextExtractor (to pull text) or the pack's image-extraction and page-to-image converters.

The key thing to understand: that PDF type is a Fill-Nodes construct. It's not an image, not a string - it's a container the pack's own PDF nodes pass around. So this node only makes sense as the first link in a chain of Fill-Nodes PDF nodes.

The inputs and outputs

• directory_path - the folder to load PDFs from. That's the only input.

Output is pdfs - a list of PDF objects. Wire it into FL_PDFTextExtractor, an image extractor, or a page-to-images converter depending on what you're after.

Installing it

Comes with the Fill-Nodes pack. ComfyUI Manager: search ComfyUI_Fill-Nodes, install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI_Fill-Nodes

then restart. The PDF nodes rely on libraries like PyPDF2 and PyMuPDF (fitz), which the pack installs on first load - if the console spends a while pulling those in, that's normal. And the loud "Machine Delusions" banner at startup is just the pack's splash.

Where people get tripped up

• It needs a real directory path, which is awkward on hosted setups. On your own machine, easy - point it at a folder. On cloud/serverless ComfyUI you don't have a persistent local filesystem to browse, so getting the PDFs onto the runner is the actual problem to solve first. This node assumes the files are already sitting in a readable directory.
• The PDF output only feeds Fill-Nodes PDF nodes. Don't try to wire it into a generic node expecting an image or text - it's a custom type. The text extractor and page converters are its only valid destinations.
• A missing dependency shows up as the node not loading at all. If FL_BulkPDFLoader (and the rest of the PDF nodes) never appear in the menu, PyPDF2 or PyMuPDF probably failed to install - the classic ComfyUI dependency clash. Check the startup log.
• It loads binaries into memory, so a folder of huge PDFs is a memory hit. For a handful of documents it's nothing; for hundreds of large scanned files, mind your RAM.