-
-
Notifications
You must be signed in to change notification settings - Fork 526
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: change
event allows getting a list of the changes made
#1585
base: feat/blocknote-transactions
Are you sure you want to change the base?
Conversation
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.
4ce1f42
to
4a3d364
Compare
24382b7
to
96b242b
Compare
@blocknote/ariakit
@blocknote/code-block
@blocknote/core
@blocknote/mantine
@blocknote/react
@blocknote/server-util
@blocknote/shadcn
@blocknote/xl-docx-exporter
@blocknote/xl-multi-column
@blocknote/xl-odt-exporter
@blocknote/xl-pdf-exporter
commit: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is so awesome, and implementation actually cleaner / clearer than expected! really cool
} else if (transaction.getMeta("uiEvent") === "drop") { | ||
source = { type: "drop" }; | ||
} else if (transaction.getMeta("history$")) { | ||
source = { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
does this work for collaboration as well (yjs history)?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This works only for the history plugin. But, I think y-prosemirror tells us if an undo happened (unsure if locally or remote or both), so maybe I can do it for that too
const changes: BlocksChanged<BSchema, ISchema, SSchema> = []; | ||
// TODO when we upgrade to Tiptap v3, we can get the appendedTransactions which would give us things like the actual inserted Block IDs. | ||
// since they are appended to the transaction via the unique-id plugin | ||
const combinedTransaction = combineTransactionSteps(transaction.before, [ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't completely understand this. How do we get the block ids now then, if we don't yet support the uniqueid transactions?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Tiptap V2 gives the root tr that was applied which does not include any appended transactions (since it is only the root tr).
Tiptap V3 gives both root tr & any appended transaction in that callback which means you can see the appended transactions & re-combine them to get the "actual" changed ranges.
Technically, the editor state is up to date by the time we are reading this, but we can't know what transactions have been appended so our transaction ranges can be incorrect.
The reason that we are getting ids out of this right now is that nodeToBlock will generate an ID if it does not find one. But, those IDs are not the same ID that actually would have been persisted into the document.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What do we actually need to need the UniqueId transactions for? Since this is more for tracking user changes, so giving a new block an ID doesn't seem like smth we need to listen for. So is the issue just that if a new block is created, when we try to get it with nodeToBlock
, it will generate a random ID as the UniqueId transaction is not detected here?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The reason the unique-id appended transaction is important is that if you do getChanges
, any inserted block technically hasn't been assigned an ID yet, it just generates one at random (in nodeToBlock
as you described). So, you can't actually trust that ID to do lookups anywhere (in the document or otherwise). So, in the future, we can actually append that transaction to get the correct ID and not have that issue - making the ID reported by getChanges
accurate to what is actually in the document.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The reason that we are getting ids out of this right now is that nodeToBlock will generate an ID if it does not find one. But, those IDs are not the same ID that actually would have been persisted into the document.
is this a blocker for the PR? it makes the "insert" changes not very useful then, right?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It depends on what they are using this for. If this is just to inspect the content to do things like remove image uploads, then it won't matter.
I'm uncertain of what you'd need the insert changes for, much less whether it is a hard requirement to have the ID be correct. I get that correctness is important, in general. But, I hope that Tiptap V3 is around the corner and alleviates us of this.
Thinking out loud: One way to get around this would be to get the previous editor state & apply the transaction again (which would run the plugins again). But, the ID would still be incorrect since it was randomly generated. So, not helpful either.
Another way would be to try to find the node afterward, but technically any plugin could append a transaction that makes the position no longer valid. So not helpful in the general case, but probably fine just for blocknote.
Or, on inserting a node, we just already give it an ID
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm, those workarounds sound painful. I figured a quick and easy fix be to update insertBlocks
to add the id
before issueing the transaction. However, then we'd still have scenarios where user operations like enter
doesn't show the right id
It depends on what they are using this for. If this is just to inspect the content to do things like remove image uploads, then it won't matter.
Thinking of two scenarios atm that will be difficult now:
- Table of contents: detect inserted headings and turn them into a ToC. While this would now work, you can't make clicking the ToC heading work because you don't know to which block in the document you need to scroll
- post-processing: maybe you'd want to use this API to do some kind of post-processing / validation and update the block to make sure it always satisfies "some" requirements - (ofc; doing this after a change is still rather hacky; we should probably know more about these use-cases and expose a better API for this)
let originalDispatch: typeof editor.dispatch; | ||
|
||
beforeEach(() => { | ||
transaction = undefined as unknown as Transaction; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
not easier to test the onchange event directly instead of patching like this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Likely easier, but the test is not for the change event, but to see if a transaction can give us the changes. I'd want it isolated
const changes: BlocksChanged<BSchema, ISchema, SSchema> = []; | ||
// TODO when we upgrade to Tiptap v3, we can get the appendedTransactions which would give us things like the actual inserted Block IDs. | ||
// since they are appended to the transaction via the unique-id plugin | ||
const combinedTransaction = combineTransactionSteps(transaction.before, [ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What do we actually need to need the UniqueId transactions for? Since this is more for tracking user changes, so giving a new block an ID doesn't seem like smth we need to listen for. So is the issue just that if a new block is created, when we try to get it with nodeToBlock
, it will generate a random ID as the UniqueId transaction is not detected here?
removedBlockIds.forEach((blockId) => { | ||
changes.push({ | ||
type: "delete", | ||
block: prevAffectedBlocks.find((block) => block.id === blockId)!, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Shouldn't prevBlock
be set instead of block
? Since the block previously existed and no longer does
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I thought of doing that too. But felt that if someone just wanted, "list of all changed blocks", it'd be just getChanges().map(c=>c.block)
, whereas branching on the type is more cumbersome.
I think an argument can be made for either way (i.e. the type is 'deleted' so obviously block
is the deleted block)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yep fair, I don't have a strong opinion one way or another, just wanted to make sure it was intentional
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
small last notes added!
}, | ||
"type": "delete", | ||
}, | ||
{ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Q: do we want to show parent as an "update" to the consumer? can we clearly explain this?
@@ -0,0 +1,190 @@ | |||
[ | |||
{ |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is the case we discussed. The we want it like this (can we clearly explain this), or prefer only to share the update children?
Let's at least make sure it's clearly documented :)
/** | ||
* Whether the change is from this client or another client. | ||
*/ | ||
isChangeOrigin: boolean; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
if this is false
, shouldn't we just trigger type: "local"
instead?
This implements a
getChanges
API which will derive the changes made within a transaction as a list of: inserts, updates and deletions of blocknote blocksTo test this you can update an example like so:
Things left:
Resolves #1541 #482 #1566