Roam JSON export -> Obsidian Markdown, preserving block references

Share & showcase
#1

Hi there,

I just converted my Roam DB over to Obsidian. I love being back in control over my notes now :smile:!

Here is the little python script I used to convert Roam’s JSON into Markdown, preserving block references: github.com/renerocksai/rj2obs

Admittedly, it doesn’t have the most fancy command line interface yet, but that’s because it was originally just intended to be used by me.

Might make it more user-friendly, though.

Feedback is welcome!

1 Like
My Migration from Roam to Obsidian
#2

Just used it, it worked like a charm, thanks a lot. Quick question (if you used namespaces in Roam): how did you deal with them? I found that on export they create empty folders but the actual content of the blocks is still contained in the original note where the namespace was created. Thanks!

#3

Hi there, great it worked!

I used namespaces in exactly 1 occasion, which is why I took extra care that they are dealt with. I had noticed that Roam’s Markdown export also created folders, so I decided to mimic that behavior:

[[ [[A]] / [[B]] ]] - creates folder A and puts note B in there if such a note exists

In my case, one note B existed :smile:.

So it can happen that you end up with empty folders - but you can ignore or delete them.
Cheers!

#4

Hi, I am very keen to use this script to import my notes to Obsidian, but I am not sure how to execute this. Is there a simple explainer which you can share? I am stuck at executing the requirements file. Hoping for a reply. Thanks!

#5

Hi,

So you have Python installed? Are you on Linux, Windows, Mac?

#6

Hey, the conversion just worked out for me, thanks for your response. Also, do I need to run the Markdown importer even after I run your script?

I am still to figure out now the parallels between Roam and Obsidian. For example, in Roam if I have made a page with the name of a person, the page shows all the notes about that person taken over time. When I converted that to Obsidian, the note with the name of the person is blank. References show in Linked mentions, but I would like to see it in the same page similar to Roam. Any idea about that?

#7

Hi,

Thanks for the script but for some reason it doesn’t seem to work. I keep getting the following error:

    Pass 1: scan all pages
  0%|                                                                                                                                 | 0/1907 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "r2o.py", line 122, in <module>
    created = page.get('create-time', page['edit-time'])
KeyError: 'edit-time'

Do you know what I could be doing wrong?

#8

I ran into this also. what I found were the timestamps in the roam json file for a particular page had gone bad. they are large integers which represent the number of seconds since january 1, 1970. in my case I found they should be 10-digit integers but were 13 for some reason. so the year was 56,816 instead of, say, 2019. I used perl on the command line to fix them throughout the whole file. I’ll post the one-liner I ran at the shell in a sec.

#9

here is the command I ran (the last word is the name of the Roam json file)

perl -p -i -e 's/:(1[56]\d{8})\d{3},/:$1,/g' roam.json
#10

I’m running into an error as well.
I’ve analyzed my JSON and I don’t have the variable of this py
image
Thanks beforehand

#11

the uid xFq6qser5 might be a block ref?

#12

I had one very long file that had around 2228 lines (it was from the Day One data I had imported into roam, on one day, the script put out two bullet points for every journal entry). the .md file ended up stopping right around line 2000. not a big deal, since I noticed it and just imported the last few hundred lines by hand, but it was interesting. my next-longest file was only 700-something lines, and that all made it over.