OpenEditor

An application that allows creation and correction of automated transcriptions of media. The front-end is React based using a Draft.js based editor, the backend runs on AWS as a serverless setup (AWS Lambda, DynamoDB and Fargate).

Note: this application was designed as an internal tool, users can see other users' name and email for collaboration purposes and while the front-end does not allow an user to look at projects that they are not member of, the user's authentication token and the API allows it.

Installation

Prerequisites: a separate AWS account (as some of the permissions are rather permissive), nvm or node 12, yarn, docker and aws cli.

In packages/services/ copy config.sample.yml to config.yml and set inside the AWS profiles to use and an unique suffix (to avoid collisions with other OpenEditor S3 buckets). Similarly in each folder under packages/services/ copy serverless.sample.yml to serverless.yml.

In the root of the repository, be sure to run nvm use or use node 12 and run yarn.

Create the database

In services/database run ./node_modules/.bin/serverless deploy -s prod, the output should look like:

Serverless: Creating deployment bucket 'openeditor-prod-deployment-example'...
Serverless: Packaging service...
Serverless: Uploading CloudFormation file to S3...
Serverless: Uploading artifacts...
Serverless: Validating template...
Serverless: Creating Stack...
Serverless: Checking Stack create progress...
.....
Serverless: Stack create finished...
Service Information
service: openeditor-database
stage: prod
region: us-east-2
stack: openeditor-database-prod
resources: 1

The table created is arn:aws:dynamodb:us-east-2:1234567890:table/openeditor-prod-data.

This is a single table design (see https://www.alexdebrie.com/posts/dynamodb-single-table/), the naming is inspired by https://github.com/amzn/smoke-dynamodb

The data looks like this:

User (Partition key: user id from cognito, Sort key: v0_metadata):

{
  "PK": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee",
  "SK": "v0_metadata",
  "RowType": "user",
  "RowVersion": 0,
  "name": "John",
  "email": "[email protected]",
  "lastLogin": "2021-05-15T11:22:33.337Z",
  "createdAt": "2021-05-15T11:22:33.337Z",
  "updatedAt": "2021-05-15T11:22:33.337Z"
}

Project (Partition key: project id, Sort key: v0_metadata):

{
  "PK": "RVFsoCrgQEe9kxCiYXs5EL",
  "SK": "v0_metadata",
  "RowType": "project",
  "RowVersion": 0,
  "title": "Project 2021-05-15T11:27:07.642Z",
  "createdAt": "2021-05-15T11:27:08.988Z",
  "createdBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee",
  "updatedAt": "2021-05-15T11:27:10.003Z",
  "updatedBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee"
}

And the relation in between is kept by entries like (Partition key: project key, Sort key: v0_user:{user-id}):

{
  "PK": "RVFsoCrgQEe9kxCiYXs5EL",
  "SK": "v0_user:b5c48b36-110c-4c34-b5b2-28315ff2e5ee",
  "RowType": "project-user",
  "RowVersion": 0,
  "createdAt": "2021-05-15T11:27:10.003Z",
  "createdBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee",
  "updatedAt": "2021-05-15T11:27:10.003Z",
  "updatedBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee"
}

Similarly a folder looks like this, where the parent is a project id or a folder id when folders are nested:

{
  "PK": "63LnfUpbPpoPa3WeKEbisF",
  "SK": "v0_metadata",
  "RowType": "folder",
  "RowVersion": 0,
  "parent": "WfkvmVkUQvbbbFHznko6Bq",
  "title": "Test Folder",
  "createdAt": "2021-05-15T11:34:24.539Z",
  "createdBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee",
  "updatedAt": "2021-05-15T11:34:24.539Z",
  "updatedBy": "b5c48b36-110c-4c34-b5b2-28315ff2e5ee"
}

While a transcript has one metadata entry (Partition key: transcript id, Sort key: v0_metadata):

{
  "PK": "EekyyYddRiyw1gBVaeEQ1P",
  "SK": "v0_metadata",
  "RowType": "transcript",
  "RowVersion": 0,
  "parent": "ChcrRMuPQfBbr9PuaHXHXD",
  "status": "transcribed",
  "title": "test.mp4",
  "duration": 1.325,
  "blocks": [
    "WF70k8qJ7N", ...
  ],
  "src": [
    "openeditor-prod-storage-example",
    "media/EekyyYddRiyw1gBVaeEQ1P/input/file-transcoded.m4a"
  ],
  "createdAt": "2021-05-15T18:39:09.482Z",
  "createdBy": "0a023760-1db2-4552-9b47-be6e699fec4e",
  "updatedAt": "2021-05-15T18:39:43.444Z",
  "updatedBy": "0a023760-1db2-4552-9b47-be6e699fec4e"
}

and an entry for each block (paragraph) from the blocks list (Partition key: transcript id, Sort key: v0_block:{block-id}):

{
  "PK": "EekyyYddRiyw1gBVaeEQ1P",
  "SK": "v0_block:WF70k8qJ7N",
  "RowType": "block",
  "RowVersion": 0,
  "status": "transcribed",
  "speaker": "spk_0",
  "text": "It's a test.",
  "start": 740,
  "end": 2065,
  "ends": [
    1020,
    1140,
    2060
  ],
  "key": "WF70k8qJ7N",
  "keys": [
    "zSFxQiwqO",
    "eya0a8ma7G",
    "jiaqwX5rmj"
  ],
  "lengths": [
    4,
    1,
    5
  ],
  "offsets": [
    0,
    5,
    7
  ],
  "starts": [
    740,
    1020,
    1150
  ],
  "createdAt": "2021-05-15T18:39:43.444Z",
  "updatedAt": "2021-05-15T18:39:43.444Z"
}

Where starts, ends, keys, offsets and lengths are data for each token with that specific offset and lenght inside text.

Create the storage bucket

In services/storage run ./node_modules/.bin/serverless deploy -s prod, the output should look like:

Serverless: Using deployment bucket 'openeditor-prod-deployment-example'
Serverless: Packaging service...
Serverless: Uploading CloudFormation file to S3...
Serverless: Uploading artifacts...
Serverless: Validating template...
Serverless: Creating Stack...
Serverless: Checking Stack create progress...
.....
Serverless: Stack create finished...
Service Information
service: openeditor-storage
stage: prod
region: us-east-2
stack: openeditor-storage-prod
resources: 1
api keys:
  None
endpoints:
  None
functions:
  None
layers:
  None
Received Stack Output {
  StorageBucketName: 'openeditor-prod-storage-example',
  StorageBucketArn: 'arn:aws:s3:::openeditor-prod-storage-example',
  ServerlessDeploymentBucketName: 'openeditor-prod-deployment-example'
}
Serverless: Stack Output processed with handler: scripts/output.handler
Serverless: Stack Output saved to file: ../../app/src/openeditor-storage-stack.json

The S3 bucket created is arn:aws:s3:::openeditor-prod-storage-example and a json config for this is created for the front-end: ../../app/src/openeditor-storage-stack.json

Create SQS queues

Create 3 SQS queues named openeditor-prod-transcoded, openeditor-prod-transcribed and openeditor-prod-aligned with default visibility timeout of 15 Minutes.

Create the API

In services/api/config.yml please replace 1234567890 with your account ID, such that various arn: identifiers defined there have the proper ID. Also set TaskSubnet to one of the available subnets in your default VPC.

Setup MediaConvertAPI and MediaConvertQueue in services/api/config.yml with the value from the AWS Console MediaConvert. For the MediaConvertRole create mediaConvertRole in AWS IAM with these policies: AmazonS3FullAccess, AmazonAPIGatewayInvokeFullAccess and AWSElementalMediaConvertFullAccess.

In AWS IAM create ecsTaskExecutionRole role with these policies: AmazonS3FullAccess and AmazonECSTaskExecutionRolePolicy, set execRoleArn in services/api/config.yml with the ARN of the role.

Run ./node_modules/.bin/serverless deploy -s prod, the output should look like:

Serverless: Cleaning dependency symlinks
Serverless: Creating dependency symlinks
Serverless: Using deployment bucket 'openeditor-prod-deployment-example'
Serverless: Bundling with Webpack...
Serverless: Packaging service...
Serverless: Uploading CloudFormation file to S3...
Serverless: Uploading artifacts...
Serverless: Uploading service getTranscript.zip file to S3 (91.63 KB)...
........
Serverless: Uploading service create.zip file to S3 (88.68 KB)...
Serverless: Uploading service reparagraphTranscript.zip file to S3 (91.63 KB)...
Serverless: Validating template...
Serverless: Creating Stack...
Serverless: Checking Stack create progress...
........
Serverless: Stack create finished...
Service Information
service: openeditor
stage: prod
region: us-east-2
stack: openeditor-prod
resources: 100
api keys:
  prod-import: SECRET_API_KEY
endpoints:
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/transcript/{PK}
  PUT - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/transcript/{PK}
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/transcript/{PK}
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/reparagraph/{PK}
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/data
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/import
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/export
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/export/{PK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/data/{PK}/{SK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/data
  PUT - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/data/{PK}/{SK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/user/{sub}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/users
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/users/{sub}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/users/{sub}/projects
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/projects/{PK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/user/{sub}/projects
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/users/{sub}/projects
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/users/{sub}/projects/{PK}
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/user/{sub}/projects
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/transcribe
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/transcode
  POST - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/align
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/tree/{PK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/children/{PK}
  GET - https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod/breadcrumbs/{PK}
functions:
  getTranscript: openeditor-prod-getTranscript
  updateTranscript: openeditor-prod-updateTranscript
  duplicateTranscript: openeditor-prod-duplicateTranscript
  reparagraphTranscript: openeditor-prod-reparagraphTranscript
  create: openeditor-prod-create
  import: openeditor-prod-import
  exportList: openeditor-prod-exportList
  exportTranscript: openeditor-prod-exportTranscript
  get: openeditor-prod-get
  list: openeditor-prod-list
  update: openeditor-prod-update
  getUser: openeditor-prod-getUser
  postAuth: openeditor-prod-postAuth
  users: openeditor-prod-users
  user: openeditor-prod-user
  projects: openeditor-prod-projects
  projectUsers: openeditor-prod-projectUsers
  getUserProjects: openeditor-prod-getUserProjects
  joinProject: openeditor-prod-joinProject
  leaveProject: openeditor-prod-leaveProject
  addUser2Project: openeditor-prod-addUser2Project
  transcribe: openeditor-prod-transcribe
  transcode: openeditor-prod-transcode
  align: openeditor-prod-align
  tree: openeditor-prod-tree
  children: openeditor-prod-children
  breadcrumbs: openeditor-prod-breadcrumbs
  transcribed: openeditor-prod-transcribed
  transcoded: openeditor-prod-transcoded
  aligned: openeditor-prod-aligned
layers:
  None
Received Stack Output {
  ApiGatewayRestApiId: 'APIGW_ID',
  Region: 'us-east-2',
  ServiceEndpoint: 'https://APIGW_ID.execute-api.us-east-2.amazonaws.com/prod',
  ApiGatewayRestApiRootResourceId: 'APIGW_ROOT_ID',
  ServerlessDeploymentBucketName: 'openeditor-prod-deployment-example'
}
Serverless: Stack Output processed with handler: scripts/output.handler
Serverless: Stack Output saved to file: ../../app/src/openeditor-stack.json

This also created in the front-end ../../app/src/openeditor-stack.json.

Setup Authentication

In services/auth run ./node_modules/.bin/serverless deploy -s prod, the output should look like:

Serverless: Using deployment bucket 'openeditor-prod-deployment-example'
Serverless: Packaging service...
Serverless: Uploading CloudFormation file to S3...
Serverless: Uploading artifacts...
Serverless: Validating template...
Serverless: Creating Stack...
Serverless: Checking Stack create progress...
.................
Serverless: Stack create finished...
Service Information
service: openeditor-auth
stage: prod
region: us-east-2
stack: openeditor-auth-prod
resources: 5
api keys:
  None
endpoints:
  None
functions:
  None
layers:
  None
Received Stack Output {
  UserPoolClientId: 'example',
  UserPoolId: 'us-east-2_example',
  IdentityPoolId: 'us-east-2:example',
  UserPoolArn: 'arn:aws:cognito-idp:us-east-2:1234567890:userpool/us-east-2_example',
  ServerlessDeploymentBucketName: 'openeditor-prod-deployment-example'
}
Serverless: Stack Output processed with handler: scripts/output.handler
Serverless: Stack Output saved to file: ../../app/src/openeditor-auth-stack.json

This also created in the front-end ../../app/src/openeditor-auth-stack.json.

In services/api/config.yml uncomment lines 69-73 and set the proper ARN for the User Pool:

- Action:
    - cognito-idp:ListUsers
    - cognito-idp:AdminGetUser
  Resource: arn:aws:cognito-idp:us-east-2:1234567890:userpool/us-east-2_example
  Effect: Allow

and run ./node_modules/.bin/serverless deploy -s prod again to update the API.

In AWS Cognito console set the Post Authentication trigger on the user pool to openeditor-prod-postAuth

Run the front-end

In app run yarn start and open in browser http://localhost:3000/, you should see the sign in, choose to create an account.

On first login an user entry and a project will be created in the database. At this moment you can create projects and folders, uploads will work but stay in uploaded limbo.

Setup S3 event notifications

On the 3 SQS queues created, edit the access policy and add (and modify queue ARN):

{
  "Sid": "s3-to-sqs",
  "Effect": "Allow",
  "Principal": {
    "AWS": "*"
  },
  "Action": "SQS:SendMessage",
  "Resource": "arn:aws:sqs:us-east-2:1234567890:openeditor-prod-transcoded",
  "Condition": {
    "ArnLike": {
      "aws:SourceArn": "arn:aws:s3:*:*:openeditor-prod-storage-example"
    }
  }
}

Setup on the already created S3 bucket, three event notifications:

Name: transcoded, suffix: -transcoded.m4a, on all object created events; with destination SQS queue openeditor-prod-transcoded
Name: transcribed, suffix: -transcription.json, on all object created events; with destination SQS queue openeditor-prod-transcribed
Name: aligned, suffix: -align.json, on all object created events; with destination SQS queue openeditor-prod-aligned

Setup the aligner

In AWS ECR create a repository hyperaudio/gentle.

Clone somewhere https://github.com/hyperaudio/gentle and run git submodule init && git submodule update

Build and push to the new ECR repository (follow the push commands provided example in AWS ECR), it should look like:

Retrieve an authentication token and authenticate your Docker client to your registry. Use the AWS CLI: aws ecr get-login-password --region us-east-2 | docker login --username AWS --password-stdin 1234567890.dkr.ecr.us-east-2.amazonaws.com
Build your Docker image using the following command: docker build -t hyperaudio/gentle . (if the build fails try increasing your docker memory)
After the build completes, tag your image so you can push the image to this repository: docker tag hyperaudio/gentle:latest 1234567890.dkr.ecr.us-east-2.amazonaws.com/hyperaudio/gentle:latest
Run the following command to push this image to your newly created AWS repository: docker push 1234567890.dkr.ecr.us-east-2.amazonaws.com/hyperaudio/gentle:latest

In AWS IAM, modify role ecsTaskExecutionRole trust relationships as per https://docs.aws.amazon.com/AmazonECS/latest/developerguide/task_execution_IAM_role.html

In AWS ECS create a Fargate cluster named OpenEditor, and a Fargate Task named gentle witn ecsTaskExecutionRole role and 8GB memory and 4 vCPUs, with container named gentle and the image from the above repository 1234567890.dkr.ecr.us-east-2.amazonaws.com/hyperaudio/gentle:latest

Build and deploy the front-end

In packages/app run yarn build and then deploy the build folder to any static hosting of choice.

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
packages		packages
.editorconfig		.editorconfig
.gitignore		.gitignore
.nvmrc		.nvmrc
LICENSE		LICENSE
README.md		README.md
package.json		package.json
yarn.lock		yarn.lock

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

OpenEditor

Installation

Create the database

Create the storage bucket

Create SQS queues

Create the API

Setup Authentication

Run the front-end

Setup S3 event notifications

Setup the aligner

Build and deploy the front-end

About

Languages

License

wwciweb/openeditor

Folders and files

Latest commit

History

Repository files navigation

OpenEditor

Installation

Create the database

Create the storage bucket

Create SQS queues

Create the API

Setup Authentication

Run the front-end

Setup S3 event notifications

Setup the aligner

Build and deploy the front-end

About

Topics

Resources

License

Stars

Watchers

Forks

Languages