Set download folder

[FRIKIdelTO]

How can I define the directory in which the files will be downloaded when interacting with the pages? by default I know it is "downloaded_files" in the current folder, but I want to keep my project folder free of files being downloaded, in fact I would like them to be saved in f'{gettempdir()}/chrome_downloads'. Is this possible?
Thanks for your amazing work!

[mdmintz]

Duplicate of #1557 (comment)

---

SeleniumBase creates the downloaded_files folder in the working directory where pytest is invoked. Any click-initiated downloads will go there. It's also used for any special files needed. It also holds lock files to prevent issues with multi-threading. The folder resets at the start of every new pytest run so that past test runs don't interfere with new ones. The folder is hard-coded there to prevent issues. And since the folder is reset at the start of new pytest runs, you wouldn't want to use any other existing folders for it.

There are lots of built-in test methods that are specially made for that folder, such as:

self.get_downloads_folder()

self.get_browser_downloads_folder()

self.get_path_of_downloaded_file(file)

self.is_downloaded_file_present(file)

self.delete_downloaded_file_if_present(file)
# Duplicates: self.delete_downloaded_file(file)

self.assert_downloaded_file(file)

self.get_downloaded_files(regex=None, browser=False)

self.get_data_from_downloaded_file(file, timeout=None, browser=False)

self.assert_data_in_downloaded_file(data, file, timeout=None, browser=False)

If you need files in a different folder, use Python os, sys, and shutil libraries for copying the files you need into a different folder after they get downloaded to the downloaded_files folder.

There's also a pytest command-line option to archive existing downloads instead of deleting them:

SeleniumBase/README.md

Line 673 in 570910c

--archive-downloads # (Archive old downloads instead of deleting them.)

(A copy of those downloads will go into a folder in the archived_files/ folder at the end of the pytest run.)

[FRIKIdelTO]

I don't use pytest or multithreading, in fact I just migrated to seleniumbase because I had issues with undetected_chromedriver. I just need to be able to configure where I want to save my downloads

[mdmintz]

If you need files in a different folder, use Python os, sys, and shutil libraries for copying the files you need into a different folder after they get downloaded to the downloaded_files folder.

[Jonathynlee]

@mdmintz I am running this in a lambda and only the tmp folder is writable. There is no way around it. How would I direct the downloads there?

[FRIKIdelTO]

@mdmintz I am running this in a lambda and only the tmp folder is writable. There is no way around it. How would I direct the downloads there?

I did like this:

def set_download_dir(driver, down_path):
    driver.command_executor._commands["send_command"] = ("POST", '/session/$sessionId/chromium/send_command')
    params = {
        'cmd': 'Page.setDownloadBehavior',
        'params': {
            'behavior': 'allow',
            'downloadPath': os.path.normpath(os.path.abspath(down_path))
        }
    }
    driver.execute("send_command", params)

[noahfaro]

I understand that this is a hardcoded folder for a reason, but certain environments (like Lambda as @Jonathynlee mentioned) are read-only except for specific directories, making your response @mdmintz insufficient. This is an actual issue with this repository that prevents us from using it at scale, and would like to hear your thoughts as to why the users cannot be the ones to designate the hardcoded folder that you use. For example, if we could set the DOWNLOADS_FOLDER inside of fixtures.constants.Files, that would already solve the issue at hand.

I am currently trying to implement a workaround where I am quite literally overwriting os commands that point to "downloaded_files" to instead point to "tmp/downloaded_files" so that I can actually deploy this in serverless environments like Lambda, which is certainly (ideally) unnecessary.

[mdmintz]

Nothing really prevents people from importing constants and overriding variables at the start of people's scripts:

SeleniumBase/seleniumbase/fixtures/constants.py

Line 81 in 35dcd82

DOWNLOADS_FOLDER = "downloaded_files"

[gary-conlon]

@mdmintz It is possible to import and set DOWNLOADS_FOLDER. But I am running into the issue when using a dynamic grid. Selenium now has managed downloads which when turned on allow you to talk to the API which will pull the downloads from folders named off session.

https://www.selenium.dev/documentation/grid/configuration/cli_options/#enabling-managed-downloads-by-the-node

Since Seleniumbase has the path hard set here with no option to turn that off, I cannot get back that behaviour that Selenium provided that I need for the grid.

SeleniumBase/seleniumbase/core/browser_launcher.py

Line 2082 in b1b1dee

prefs["download.default_directory"] = downloads_path