config.yaml

# Job specs define the configuration for Bacalhau jobs
jobs:
- id: root-job # A job that accepts a root input and passes it to the output -- the root of all jobs
  type: internal # This job will run on an internal executor. Internal job, doesn't leave Amplify.
  internal_job_id: root-job # Link to internal job ID, must exist in the codebase
- id: metadata-job # This is the key used when looking for a job
  type: bacalhau # This job will run on the Bacalhau executor
  image: ghcr.io/bacalhau-project/amplify/tika:0.0.5
  entrypoint:
  - /usr/local/bin/run
- id: merge-job
  image: ghcr.io/bacalhau-project/amplify/merge:0.0.1
  entrypoint:
  - /usr/local/bin/run
- id: image-resize-job # This job resizes images recursively
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/magick:0.0.1
  entrypoint:
  - /usr/local/bin/run
- id: video-resize-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/ffmpeg:0.0.3
  entrypoint:
  - /usr/local/bin/run
- id: csv-profiling-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/ydata-profiling:0.0.5
  entrypoint:
  - /usr/local/bin/run
- id: csv-frictionless-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/frictionless:0.0.2
  entrypoint:
  - /usr/local/bin/run
- id: convert-table-to-csv-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/frictionless-extract:0.0.6
  entrypoint:
  - /usr/local/bin/run
- id: detection-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/detection:0.0.4
  entrypoint:
  - /usr/local/bin/run
  timeout: 10m
  cpu: 2
  memory: 8Gi
- id: summarization-job
  type: bacalhau
  image: ghcr.io/bacalhau-project/amplify/summarization:0.0.3
  entrypoint:
  - /usr/local/bin/run
  timeout: 10m
  cpu: 4
  memory: 8Gi


# Amplify Work Graph specification
# Each item in the list is a node in the execution graph. A single request 
# (typically a single CID) runs this whole graph. 
graph:
- id: root-node # This is the root of the dag, where the request CID is placed
  job_id: root-job
  inputs:
  - root: true # Identifies that this is a root node
    path: /inputs # Path where inputs will be placed
  outputs:
  - # id: default # Specify custom output id (default: "default")
    path: /outputs # Path where job places outputs
- id: metadata-node 
  job_id: metadata-job 
  inputs:
  - node_id: root-node
    # output_id: custom_id # Connect to custom output id (default: "default")
    path: /inputs
  outputs:
  - path: /outputs
- id: image-resize-node
  job_id: image-resize-job
  inputs:
  - node_id: metadata-node
    predicate: '.*image\/.*'
  - node_id: root-node
    path: /inputs/image/
  outputs:
  - path: /outputs
- id: video-resize-node
  job_id: video-resize-job
  inputs:
  - node_id: metadata-node
    predicate: '.*video\/.*'
  - node_id: root-node
    path: /inputs
  outputs:
  - path: /outputs
- id: convert-table-to-csv-node
  job_id: convert-table-to-csv-job
  inputs:
  - node_id: root-node
    path: /inputs/
  outputs:
  - path: /outputs
- id: csv-validation-node
  job_id: csv-frictionless-job
  inputs:
  - node_id: metadata-node
    predicate: '.*text\/csv.*'
  - node_id: root-node
    path: /inputs/
  outputs:
  - path: /outputs
- id: csv-profiling-node
  job_id: csv-profiling-job
  inputs:
  - node_id: csv-validation-node
    predicate: '.*"valid":true.*'
  - node_id: root-node
    path: /inputs/
  outputs:
  - path: /outputs
- id: detection-node
  job_id: detection-job
  inputs:
  - node_id: metadata-node
    predicate: '.*(video\/|image\/).*'
  - node_id: root-node
    path: /inputs
  outputs:
  - path: /outputs
- id: text-summarization-node
  job_id: summarization-job
  inputs:
  - node_id: metadata-node
    predicate: '.*text\/plain.*'
  - node_id: root-node
    path: /inputs/
  outputs:
  - path: /outputs
- id: merge-node
  job_id: merge-job
  inputs:
  - node_id: root-node
    path: /inputs/src
  - node_id: metadata-node 
    path: /inputs/metadata-node
  - node_id: video-resize-node
    path: /inputs/video-resize-node
  - node_id: detection-node
    path: /inputs/detection-node
  - node_id: image-resize-node
    path: /inputs/image-resize-node
  - node_id: csv-validation-node
    path: /inputs/csv-validation-node
  - node_id: csv-profiling-node
    path: /inputs/csv-profiling-node
  - node_id: convert-table-to-csv-node
    path: /inputs/convert-table-to-csv-node
  - node_id: text-summarization-node
    path: /inputs/text-summarization-node
  outputs:
  - path: /outputs