Expensify / App

Welcome to New Expensify: a complete re-imagination of financial collaboration, centered around chat. Help us build the next generation of Expensify by sharing feedback and contributing to the code.
https://new.expensify.com
MIT License
3.59k stars 2.92k forks source link

[HOLD for payment 2024-07-10] [HOLD for payment 2024-07-02] LOW: [$500] Stop using `reportAction.originalMessage` or `reportAction.message.text` #39797

Closed quinthar closed 4 months ago

quinthar commented 7 months ago

Problem:

Each comment in the server database is stored in reportAction.message in HTML form. However, for some reason lost to time, when we send that same comment out as an Onyx update, we:

  1. Rename the original message in the database to reportAction.originalMessage in the update (which is confusing)
  2. Create a new message object that contains a copy of the html message, as well as a stripped text version.
  3. Put the message object inside a totally unnecessary one-element array

That means every comment is actually sent out three times (four, if you include the update to the report object, but that's out of scope for now). This is clearly wasteful in a number of ways:

  1. It requires postprocessing on the server to do this extraneous operation
  2. It increases network bandwidth by 3x
  3. It increases RAM use 3x
  4. It increases on-device storage 3x
  5. It's confusing AF

Additionally, it's particularly problematic when sending Onyx updates out via UrbanAirship, which is extremely limited in the payload sizes allowed -- with anything over a certain limit just dropped quietly and never delivered, causing "gaps" in our update stream (which require more network calls to "backfill" the Onyx data on app open, which makes things slow). Basically, this was just a mistake introduced for reasons we can't remember, but that we want to undo.

Solution:

To solve this, please:

  1. Phase 1 (External / clientside): Stop using originalMessage, start using array-message
    1. Update every instance of originalMessage.html to use message.html || message[0].html
    2. Update every instance of message.text to dynamically strip the text from message.html || message[0].html
    3. Update every key within originalMessage to use the original message
  2. Phase 2 (Internal / serverside): Once we've confirmed the client no longer uses originalMessage or message.text or any other key within originalMessage:
    1. Stop sending originalMessage
    2. Start sending message without an array
  3. Phase 3 (External / clientside): Remove checking for "array message"

This issue is for phase 1 and phase 3.

Upwork Automation - Do Not Edit
  • Upwork Job URL: https://www.upwork.com/jobs/~01856d1399295ef760
  • Upwork Job ID: 1776729925881405440
  • Last Price Increase: 2024-07-12
  • Automatic offers:
    • cubuspl42 | Reviewer | 0
Issue OwnerCurrent Issue Owner: @puneetlath
melvin-bot[bot] commented 7 months ago

Job added to Upwork: https://www.upwork.com/jobs/~01856d1399295ef760

melvin-bot[bot] commented 7 months ago

Triggered auto assignment to Contributor-plus team member for initial proposal review - @cubuspl42 (External)

maxim-grin commented 7 months ago

Hey I’m going to investigate why the payload is so big in Onyx storage. Possibly track what change exactly caused the problem. Then either revert that piece of change or use one of suggestions by @thienlnam

Contributor details Your Expensify account email: maxim.grin@gmail.com Upwork Profile Link: https://www.upwork.com/freelancers/~01fb7b172c01252917

melvin-bot[bot] commented 7 months ago

πŸ“£ @maxim-grin! πŸ“£ Hey, it seems we don’t have your contributor details yet! You'll only have to do this once, and this is how we'll hire you on Upwork. Please follow these steps:

  1. Make sure you've read and understood the contributing guidelines.
  2. Get the email address used to login to your Expensify account. If you don't already have an Expensify account, create one here. If you have multiple accounts (e.g. one for testing), please use your main account email.
  3. Get the link to your Upwork profile. It's necessary because we only pay via Upwork. You can access it by logging in, and then clicking on your name. It'll look like this. If you don't already have an account, sign up for one here.
  4. Copy the format below and paste it in a comment on this issue. Replace the placeholder text with your actual details. Screen Shot 2022-11-16 at 4 42 54 PM Format:
    Contributor details
    Your Expensify account email: <REPLACE EMAIL HERE>
    Upwork Profile Link: <REPLACE LINK HERE>
maxim-grin commented 7 months ago

Hey I’m going to investigate why the payload is so big in Onyx storage. Possibly track what change exactly caused the problem. Then either revert that piece of change or use one of suggestions by @thienlnam

Contributor details Your Expensify account email: maxim.grin@gmail.com Upwork Profile Link: https://www.upwork.com/freelancers/~01fb7b172c01252917

melvin-bot[bot] commented 7 months ago

βœ… Contributor details stored successfully. Thank you for contributing to Expensify!

ShridharGoel commented 7 months ago

Proposal

Please re-state the problem that we are trying to solve in this issue.

We don't need to use text from reportAction, it would be better to use originalMessage.html.

What is the root cause of that problem?

New change.

text from message in reportAction is being used at multiple places.

What changes do you think we should make in order to solve the problem?

We have text defined here:

https://github.com/Expensify/App/blob/f1502291a14d4e15e04249b241109f146bdb512f/src/types/onyx/ReportAction.ts#L18-L19

Example usage:

https://github.com/Expensify/App/blob/f1502291a14d4e15e04249b241109f146bdb512f/src/components/ReportActionItem/ReportPreview.tsx#L173

We can change this to:

const parser = new ExpensiMark();
const actionMessage = parser.htmlToText(action.message?.[0]?.html ?? '');

Add the above in a common method getActionMessagePostParsing in ReportUtils. Similarly, check all usages and update them to use the method mentioned above.

Finally, we can remove text.

There's one instance of originalMessage?.html in ReportUtils.ts which also needs to be updated to use the common method.

if (ReportActionsUtils.isWhisperAction(reportAction)) {
    // Allow flagging welcome message whispers as they can be set by any room creator
    if (report?.description && !isCurrentUserAction && isOriginalMessageHaveHtml && `getActionMessagePostParsing(reportAction)` === report.description) {
        return true;
    }

    // Disallow flagging the rest of whisper as they are sent by us
    return false;
}
nkdengineer commented 7 months ago

Proposal

Please re-state the problem that we are trying to solve in this issue.

We want to remove the usage of text of reportAction in the payload sent to client side. And also changes how we're using message.html

What is the root cause of that problem?

This is new feature request.

What changes do you think we should make in order to solve the problem?

  1. Define 2 common methods here, to get the report action html (with forward compatibility: message as single item), and to get the report action text dynamically by parsing the html:
    
    function getReportActionHtml(reportAction: OnyxEntry<ReportAction>) {
    return reportAction?.message?.html || reportAction?.message?.[0]?.html;
    }

function getReportActionText(reportAction: OnyxEntry) { const html = getReportActionHtml(reportAction); const parser = new ExpensiMark(); return html ? parser.htmlToText(html) : ''; }

2. > Update every instance of originalMessage.html to use message.html || message[0].html

In [here](https://github.com/Expensify/App/blob/4e13dada426c067e2945fbd2bd06ddd73c0b8409/src/libs/ReportUtils.ts#L4618), replace `reportAction?.originalMessage?.html` by `getReportActionHtml(reportAction)`

3. > Update every instance of message.text to dynamically strip the text from message.html || message[0].html

There're some places that are using `message?.[0]?.text`, examples: [here](https://github.com/Expensify/App/blob/4e13dada426c067e2945fbd2bd06ddd73c0b8409/src/libs/ReportUtils.ts#L2877), [here](https://github.com/Expensify/App/blob/4e13dada426c067e2945fbd2bd06ddd73c0b8409/src/libs/actions/Report.ts#L853), [here](https://github.com/Expensify/App/blob/4e13dada426c067e2945fbd2bd06ddd73c0b8409/src/libs/actions/Report.ts#L1523), ... (There're more)
We can replace them by `getReportActionText(reportAction)`.

4. > Phase 3 (External / clientside): Remove checking for "array message"

After phase 2 is done, we can just change the implementation of `getReportActionHtml` to

function getReportActionHtml(reportAction: OnyxEntry) { return reportAction?.message?.html; }


5. There're other places that use `originalMessage`, but not its `html` field, like [here](https://github.com/Expensify/App/blob/4e13dada426c067e2945fbd2bd06ddd73c0b8409/src/libs/ReportActionsUtils.ts#L777)

We might need to replace them with `getReportActionMessage(reportAction)`

Where `getReportActionMessage` will be the following:

function getReportActionMessage(reportAction: OnyxEntry) { return reportAction?.message || reportAction?.message?.[0]; }


To be compatible with both current format of `message` (as array) and new message format later (single item). The `reportAction?.message?.[0]` part can be removed after Phase 2 is done.

### What alternative solutions did you explore? (Optional)
NA

<!---
ATTN: Contributor+

You are the first line of defense in making sure every proposal has a clear and easily understood problem with a "root cause". Do not approve any proposals that lack a satisfying explanation to the first two prompts. It is CRITICALLY important that we understand the root cause at a minimum even if the solution doesn't directly address it. When we avoid this step we can end up solving the wrong problems entirely or just writing hacks and workarounds.

Instructions for how to review a proposal:

1. Address each contributor proposal one at a time and address each part of the question one at a time e.g. if a solution looks acceptable, but the stated problem is not clear then you should provide feedback and make suggestions to improve each prompt before moving on to the next. Avoid responding to all sections of a proposal at once. Move from one question to the next each time asking the contributor to "Please update your original proposal and tag me again when it's ready for review".

2. Limit excessive conversation and moderate issues to keep them on track. If someone is doing any of the following things please kindly and humbly course-correct them:

- Posting PRs.
- Posting large multi-line diffs (this is basically a PR).
- Skipping any of the required questions.
- Not using the proposal template at all.
- Suggesting that an existing issue is related to the current issue before a problem or root cause has been established.
- Excessively wordy explanations.

3. Choose the first proposal that has a reasonable answer to all the required questions.
-->
cubuspl42 commented 7 months ago

@ShridharGoel

Similarly, check all usages and update them

Are you saying to literally repeat this code snippet in all places? Couldn't we refactor the code to avoid this repetition?

As the scope of the issue is well-defined and rather narrow, I believe this part of the plan can be a part of the proposal.

@nkdengineer

Aren't you quoting browser-specific local notifications code? πŸ€” As I understand it, the ultimate goal is to optimize the payload size of remote push notifications.

nkdengineer commented 7 months ago

Aren't you quoting browser-specific local notifications code? πŸ€” As I understand it, the ultimate goal is to optimize the payload size of remote push notifications.

@cubuspl42 In the OP it's mentioned

@thienlnam thinks we added text for notifications

So I'm quoting the place that we're showing notification based on text, so it's clear where it's used.

It's only 1 part of the proposal though, what do you think about the rest of my proposal?

Are you saying to literally repeat this code snippet in all places? Couldn't we refactor the code to avoid this repetition?

@cubuspl42 My alternative solution achieves this

ShridharGoel commented 7 months ago

Proposal

Updated. @cubuspl42 We can add a method in ReportUtils, updated the proposal.

cubuspl42 commented 7 months ago

@quinthar

Would you summarize the Expensify notifications architecture in a few sentences so that we can be more confident with proposal preparation and review?

As I understand that most (if not all?) notification processing on mobile Native is server-side (push-based), and we're talking about optimizing the pushed payload size. But I'm not 100% confident how that relates to the Onyx schema.

Also, there were some doubts about the scope...

I'm assuming here that we're only omitting the text from reportActions in urban airship native updates, if we're going to remove text in Pusher update too, then we need to populate the text in the Pusher report actions updates by htmlToText.

Would you clarify this too? My intuition was that we wanted to nuke all usage of the text property client-side, also removing it from the typing. But it would be good to be sure about it.

quinthar commented 7 months ago

Great question. I've updated the original post with a more detailed description of the problem and solutions. Thanks!

ShridharGoel commented 7 months ago

Proposal

Updated

quinthar commented 7 months ago

Updated the proposal again (sorry!!!) in light of message being in this weird one-element array. Thanks for your understanding and help in cleaning up this mess!

nkdengineer commented 7 months ago

Proposal updated after latest requirements update

cubuspl42 commented 7 months ago

@nkdengineer I think your proposal shows more preparation. How is your capacity, though? Would you be able to start working on this within 24-48 hours?

nkdengineer commented 7 months ago

Would you be able to start working on this within 24-48 hours?

@cubuspl42 Yes, I definitely can

cubuspl42 commented 7 months ago

I approve the proposal by @nkdengineer

C+ reviewed πŸŽ€ πŸ‘€ πŸŽ€

melvin-bot[bot] commented 7 months ago

Triggered auto assignment to @thienlnam, see https://stackoverflow.com/c/expensify/questions/7972 for more details.

melvin-bot[bot] commented 7 months ago

πŸ“£ @cubuspl42 πŸŽ‰ An offer has been automatically sent to your Upwork account for the Reviewer role πŸŽ‰ Thanks for contributing to the Expensify app!

Offer link Upwork job

melvin-bot[bot] commented 7 months ago

πŸ“£ @nkdengineer You have been assigned to this job! Please apply to the Upwork job and leave a comment on the Github issue letting us know when we can expect a PR to be ready for review πŸ§‘β€πŸ’» Once you apply to this job, your Upwork ID will be stored and you will be automatically hired for future jobs! Keep in mind: Code of Conduct | Contributing πŸ“–

nkdengineer commented 7 months ago

@thienlnam Hi! I see this is a HIGH priority issue so it should start at $500 (from the announcement here)

While bugs/feature requests will now start at $250, any that are deemed Critical or High to our roadmap (noted via the label High Priority) will automatically be updated to $500. This is to ensure we get the necessary attention and urgency applied to these issues.

Perhaps the automation did not work well in this case.

If that's right, could you help to adjust the bounty here, thanks πŸ™

cubuspl42 commented 7 months ago

Hm, it's likely because the issue is missing the "High priority" label. But I'm not sure if the "HIGH:" prefix and the "High priority" label represent the same concept πŸ€”

nkdengineer commented 7 months ago

We already have a draft PR here to fix things in phase 1, @thienlnam please let me know when phase 2 is completed on the server side, I will then update phase 3 accordingly. Cc @cubuspl42

quinthar commented 7 months ago

Great work! What's your ETA for submitting the PR for review?

nkdengineer commented 7 months ago

I will open PR tomorrow

cubuspl42 commented 7 months ago

@nkdengineer Just so we're on the same page, phase 1 must be deployed to production first. Also, some time has to pass, so Native users have the time to update. So we're not blocked by phase 2 in any way.

nkdengineer commented 7 months ago

@cubuspl42 The PR is here.

quinthar commented 7 months ago

Great! @cubuspl42 what is your ETA for review?

cubuspl42 commented 7 months ago

I'm on it πŸ™‚

quinthar commented 7 months ago

Great, what is your ETA for review?

cubuspl42 commented 7 months ago

It was yesterday. I raised concerns about whether we actually covered the whole scope of the Phase 1. I even pinged you πŸ™‚ We can continue the dialogue on the PR.

melvin-bot[bot] commented 7 months ago

Triggered auto assignment to Contributor Plus for review of internal employee PR - @parasharrajat (Internal)

mallenexpensify commented 7 months ago

@parasharrajat reassigning, please take over as C+. If you don't have bandwidth, unassign yourself. Thanks. PR is

parasharrajat commented 7 months ago

Sorry, I don't have the bandwidth to work on this. I was OOO since Thursday. let me request someone to push this forward.

parasharrajat commented 7 months ago

@c3024 will review this.

mallenexpensify commented 7 months ago

Thanks @parasharrajat , πŸ‘‹ @c3024 @c3024 commented on the PR earlier today

melvin-bot[bot] commented 6 months ago

This issue has not been updated in over 15 days. @thienlnam, @c3024, @nkdengineer eroding to Monthly issue.

P.S. Is everyone reading this sure this is really a near-term priority? Be brave: if you disagree, go ahead and close it out. If someone disagrees, they'll reopen it, and if they don't: one less thing to do!

mallenexpensify commented 6 months ago

@c3024 looks like the PR is awaiting an action from you

c3024 commented 6 months ago

@thienlnam @quinthar

The issue talks about only text and html

Phase 1 (External / clientside): Stop using originalMessage, start using array-message Update every instance of originalMessage.html to use message.html || message[0].html Update every instance of message.text to dynamically strip the text from message.html || message[0].html

but originalMessage has many more fields like reason, resolution, IOUTransactionID, events, oldName, newName, participantAccountIDs, etc. which are not present in message. So this phase 1 should be to update every instance of originalMessage (not just fields of html or text) with array message

Also in phase 2

Phase 2 (Internal / serverside): Once we've confirmed the client no longer uses originalMessage or message.text: Stop sending originalMessage Start sending message without an array

message should be sent including all the above mentioned other fields to completely deprecate originalMessage.

Can you please update phase 1 and 2 of the issue description to include these?

melvin-bot[bot] commented 6 months ago

Triggered auto assignment to @puneetlath (Bug), see https://stackoverflow.com/c/expensify/questions/14418 for more details. Please add this bug to a GH project, as outlined in the SO.

mallenexpensify commented 6 months ago

Added Bug to assign a BZ. @quinthar @thienlnam , can you address @c3024 's request above when you have a min https://github.com/Expensify/App/issues/39797#issuecomment-2130753109

thienlnam commented 6 months ago

I've updated the description

roryabraham commented 6 months ago

I did some exploratory work on this topic in https://github.com/Expensify/App/pull/41872 - maybe useful for you @nkdengineer @c3024 @thienlnam.

The main idea is that we make ReportAction take a generic param for the type, and that determines the originalMessage it will have. This helps ensure that we correctly and safely narrow the ReportAction type to ensure the originalMessage is as we expect before trying to access any of its properties

roryabraham commented 6 months ago

self-assigning as I've been pulled into the PR for review

melvin-bot[bot] commented 6 months ago

@puneetlath, @thienlnam, @roryabraham, @c3024, @nkdengineer Whoops! This issue is 2 days overdue. Let's get this updated quick!

melvin-bot[bot] commented 5 months ago

@puneetlath, @thienlnam, @roryabraham, @c3024, @nkdengineer Huh... This is 4 days overdue. Who can take care of this?

melvin-bot[bot] commented 5 months ago

@puneetlath, @thienlnam, @roryabraham, @c3024, @nkdengineer Now this issue is 8 days overdue. Are you sure this should be a Daily? Feel free to change it!

puneetlath commented 5 months ago

@nkdengineer @c3024 how is this going? Are we almost done?