brlala / async_request_facebook_attachment_id

asynchronous code for calling facebook graph api(speed up API calls)

Upload Facebook Attachment(async)

The current process sends an attachment to user only when user request the file. This causes two problem:

1. There's a lag time from when user request the file to the time that they receive it (due to the file size)
2. The same file is being sent multiple times when being requested.
3. In addition this also solves the problem of transferring files from a bucket to another bucket

Proposed Approach:

1. Leverage the use of attachment_id by Facebook
2. Create an additional collection in mongo using URL as a primary key.

Problem Encountered:

1. There are a few hundread links in our database for each client(each with file size 300KB-15MB), to upload each of them synchronously will take a long time. (IO bound)

Objective:

1. Convert all links to attachment_id provided by Facebook
2. Using asynchronous access
3. Migrate from dev bucket to production bucket

Tools used:

1. aiohttp
2. aiofiles
3. Python

Process Breakdown:

1. Get all URLs
2. Filter valid and working URLs
3. Download all URLs (async)
4. Upload to new cloud bucket (blocking)
5. Upload to Facebook for attachment ID (async)
6. Modify database entries(removing existing attachment_id and update URL)

p/s The program's semaphore is set at 8, can be increased or lowered depending on the network stability