🏠 back to Observable

load (list of) file(s) from disk


I have read that it is possible to read files from disk from mike’s and trevor’s examples.

However I’d like to know if it’s possible to read a (dozen) file(s) identified by its (their) path(s). In a Jupyter notebook, I would loop over a pd.read_csv(‘path’).

I am vaguely aware that the browser has builtin limitations to access the hard drive. But maybe the notebook can just prompts the user for validation before loading the (list of) file(s) ? Clicking and selecting several files is a bit tedious…

BTW: I am very impressed by the whole product ! :clap:


Hi Oscar,

I don’t believe that it’s possible for web browsers to pre-fill the filename (or names) in a file browsing dialog box. That would make it too easy for malicious sites to try and steal important files from victims browsing by…

That said, if you upload your data file to somewhere private online, you can request it directly by URL, avoiding the need to load the file every time. Currently, if the data isn’t sensitive, we often use GitHub’s private Gists for this purpose.



Thx for taking the time to answer. This is what I feared… That’s understandable but a bit unfortunate. So I tried another approach, as described in this how-to-conveniently-load-local-files notebook. What do you think ? is it perfectible ?

That’s a bit outside the subject by I wonder why the d3.blob and d3.xml are the only ones not working… Any immediate idea ?

Also, the Live Share extension works fine, I don’t know why the a simple ‘npm serve’ would not, a ‘python -m http.server’ neither…