sakuya9t / social-x Goto Github PK

backend endpoint should be able to return value in realtime mode.

When running the script, should download npm, create pipenv, install things needed for the app. Can be done in shell or ansible.

We want the similarity always between 0 and 1

test case:
account1 = {'platform': 'twitter', 'account': 'ClassicVines1'}
account2 = {'platform': 'pinterest', 'account': 'mariesbazaar'}

which returns 1.55

When data is missing from the database, should be able to retrieve simplified data and put into DB.

1. retrieved data should not have post content.
2. should have separate table to save the real-time mode value. So we would be able to redo the calculation in batch later.

The URL linked to the Instagram image expires after a few weeks. We cannot reuse the image URL in the Instagram account. Instead, we can download the image, convert it to base64 and saves it.

• Instagram parser returns image base64 instead of URL.
• Calculator should be able to detect image base64 and should be able to compare it with image URL.

Train two models:

• Real-time mode model
• Batch mode model

Then apply both models to the backend, so I can calculate the similarity according to the data format.

• Attach both models to the backend.

Rearrange the UI so that it would fit screens with width<600

• narrow UI
• should auto switch UI when the window width changes
• narrow UI should function same as the normal UI

Test to make sure the project works when

• two accounts are from Instagram and Instagram.
• two accounts are from Instagram and Twitter.
• two accounts are from Instagram and Pinterest.
• two accounts are from Instagram and Flickr.
• two accounts are from Twitter and Twitter.
• two accounts are from Twitter and Pinterest.
• two accounts are from Twitter and Flickr.
• two accounts are from Pinterest and Pinterest.
• two accounts are from Pinterest and Flickr.
• two accounts are from Flickr and Flickr.

Goal: Should have integrated test for each case and guarantee that all tests can pass.

There should be a script when executes, both frontend and backend are stopped.

• Write the script
• Test on the dev
• Test on the demo

In the batch, if an account is parsed in realtime mode, reparse it in batch mode.

Can do it just calling the "retrieve" function with username, it will auto-detect this. Just need a trigger.

Similarity module based on image labels

Though not necessarily get labels of all images, we need to make sure we have such function.

• Auto extract image labels given an image using MRISA
• Calculate the image label similarity
• Put the image similarity into feature vectors.

This is a part of post content similarity. Should not affect the profile-only part.

Probably not doing this in the final project itself.

There should be a restart button, resetting everything on the page so that the user can start over. (can be just a simple refresh)

• such a restart button exists

• When click on the restart button, everything goes back to the beginning

The chromedriver is not uploaded to the repository. Should have a script to set it up.

• Auto check and install the latest version of chrome.
• Download the chrome driver which has same version as the browser (otherwise will pop an error)

When frontend get error from backend in querying (e.g. invalid account), it should have an error page indicating this.

• Handle when 'error' is True
• Handle when response code is 500.
• Auto-scroll down when clicking "calculate"

As said in the report, we should have the following endpoints:

• /info | GET | Get raw information of a given account.

• /query | POST | Query details of similarity between two accounts

1. trigger to acquire from db
2. function to insert calculated value into db
3. handle timeout stuff

We are generating samples using the sample file and the pairing files. We want to get rid of these files.

• Database for the pairing file.
• When selecting database mode, should get pairing, insta, twitter data from the database.
• If the database is empty, should auto initialize it.
• If there are not enough samples, should return error.

As mentioned, we should refactor the calculator module into a single module, peeling off all related calculation functionalities.

• Calculation module should exist in a standalone project other than the backend service.
• Should connect between service and calculator through JSON packets.
• Should have a manual about the endpoints.

Save a config template. When running, the template should be renamed and copied to the target folder.

*Notice: have to be tested in other directories other than the project folder.

• 4 config files on the backend side
• front end config, should auto-detect current IP and attach it to the config file.

Currently, we only use the default way to run the application. When a user uses their own account to run the app, we should be able to get protected accounts with these accounts.

• Parse accounts using log in mode
• A calculate button in the login page
• Should return the same result as not logging in

The MRISA repo is here https://github.com/vivithemage/mrisa

We need to deploy MRISA as one of our dependencies.
It should be deployed to port 7999 and keep running with the server.

Following things should be completed in automation.

• Auto download and unzip the source code. Install dependencies.
• Copy it to somewhere in the disk.
• Specify the path that MRISA is deployed in the config file.
• Auto testing the MRISA to make sure it's working after deployment.

Probably not doing this in the final project because of the Instagram parsing efficency.

When a user gives feedback to a data item, we should label it according to user feedback and place it into the labelled database while deleting old items.

• 1. make sure this logic works

• 2. user can use this function from frontend.

We need bash scripts to start the frontend and the backend at one time.

The backend should run in a pipenv

• starting script
• tested in dev env
• tested in demo env.

Explain each parameter. What it is used for, why, and where to get the resources.

• front end config.
• algomodule
• config-deploy
• config

Should have a readme file briefly talk through what this project is