Getting .part files on some pages

[jp1151419]

Describe the bug

On many creators, both free and paid pages, I end up with some .part files as I download their content. Not all their files necessarily but it's systematic on a per-file basis (ie. deleting the .part files and re-running the script leads to the same result)

To Reproduce

A free creator where some of the paid messages have issues downloading is : midnightstars
Their VIP page is not causing my issues for some reason.

ofscraper --action download --posts ALL -u midnightstars

Expected behavior

Getting all accessible videos downloaded in a ready to play mp4 format

Screenshots/Logs

Log of a single-creator run with failed downloads

Config

Link to my config

System Info

• OS: Windows VM (4c/8t, 16GB RAM)
• Browser : Chrome Version 146.0.7680.80
• ofscraper 3.14.2 installed via pipx

Additional context

Was using "dynamic-mode-default": "rafa" up until my recent tests with a brand new config file and didn't see a difference between the two at this time.

[cjb900]

You are missing the file path for ffmpeg.exe version 7.1.1 or lower in your config.json file. Without it ofscraper can't merge the part files.

[jp1151419]

I used to have it formatted like this and still had issues. Is the path formatting ok?

"binary_options": {
    "ffmpeg": "C:\\ffmpeg\\bin\\ffmpeg.exe"

edit:
tried another run with that included in the config and still getting the .part files
https://pastebin.com/TWTduc4m

That is using ffmpeg.exe from "ffmpeg-7.1.1-full_build"

[cjb900]

You are using ofscraper version 3.14.2. there is an updated version 3.14.3. You can try updating and see if that helps

[soulburn74]

Thank you @jp1151419 for posting this, I was having same issue, had to upgrade my version of ffmpeg to get it to work. (had a new windows build, so had to install of from scratch (good thing I backed up my cert and key for cdrm, didn't want to go through that mess all over again).

[datawhores]

FileNotFoundError: [Errno 2] No such file or directory: 'C:\Users\{home_directory}\.config\ofscraper\key_dumps\pclient_id.bin'

you need to fix these paths as well

I'm guessing it should be client_id.bin and not pclient

[jp1151419]

I fixed that unfortunate typo but it's still not working. The log file is now too big for pastebin so I split it in two.

I see many messages like this:
403, message='Forbidden', url='https://onlyfans.com/api2/v2/users/media/xxxxxxxxx/drm/message/xxxxxxxxxxxxx?type=widevine'

https://pastebin.com/3jziah0n
https://pastebin.com/s0DvWVg6

[datawhores]

Maybe try updating for some reason your client/IP is being blocked from making any requests

[jp1151419]

Just updated to 3.14.4, changed my IP, cleared browser cache and got a fresh cookie, same problem. Anything else I can try?

[datawhores]

What type of browser are you using?
From the logs, it feels like maybe the account got signed out or there is maybe some sort of geoblock

2026-03-18 19:19:44:[sessionmanager.requests_async:631] response text : {"error":{"code":0,"message":"User have no permissions"}}

That is an interesting message to me

[jp1151419]

Here are some fresh logs, split in two again.
https://pastebin.com/vaHxJnUF
https://pastebin.com/8KB9gGMQ

Cookie is extracted from Firefox but I also use Chrome. I just find it easier to find all the necessary values in the Firefox network tab.
It IS a strange response but I have no idea what to make of it since my account remains fully usable through the site.

[datawhores]

You could try changing the ssl_verify to
default or false in the config

Are you subscribed to the models on the account you're using to authenticate and run the script

[jp1151419]

I just tried both default and false, no difference.

Yes,I am subscribed to the account I've been testing this with. I can both visit the page and exchange by DM with the creator so it's not like I'm blocked. It's a free account too but the videos that aren't downloading are categorized as "paid"

[Gamma-Velorum]

This is due to bug #614 in the latest version, where the ffmpeg directory specified in config.json is ignored and the first available ffmpeg binary directory, alphabetically, is chosen instead.