Since there are many edge cases depending on how folks used Roam, I'm posting some Tips on how to preprocess before importing Roam->Tana:
depending on whether you used attributes/tags/pages/nested pages , I can think of the following formats [[page]] , #page , #[[page]] , [[[[page]] [[page1]]]] , #[[[[page]] [[page1]]]] , page:: (are there more?!:zany_face:)
All of these are perfectly handled by Tana. eg: #page and #[[page]] will be converted to an alias #page which points to the node page. in Tana you can remove this alias to turn #page into page by taking the cursor into #page, pressing Ctrl+K, typing "Remove alias for reference".
However, I don't know if this can be done in bulk.
Solution: do this in bulk within Roam using Fabrice Gallet’s Find & Replace extension in Roam Depot. in this extension's settings there's a simple "Allow 'Whole Graph' Find & Replace" toggle, switch it on.
Now, go back to your graph, press Ctrl+P, type "Find &.." you'll see the "Find & Replace in Whole graph option".
choose that, click, you'll see a blue pop-up. now do your thing. turn #[[page]], #page into [[page]]. example below.
this was a simple example. there's regex support too.
afaik, i've only encountered 2 major issues with the roam-import:
URLs (this is in the works i believe)
**a phrase:** should not be parsed as an attribute when importing from roam just like **a phrase**: isn't done. i've used these 2 interchangeably in roam and not as an attribute (for that the double :: notation). They are working on fixing this, but for now, I'm using the same search and replace ":**" so those won't be imported as fields.