Developer Tools

The main developer tool for pipelogic is called ppl. ppl is a command-line tool for creating pipelogic application and deploying them on streaming platform. These applications are built as containerized building blocks, ensuring consistency and portability throughout the development and deployment processes.

PPL tool only requires basic bash knowledge. It is a powerful tool that enables users to develop and manage solutions with few command lines.

Contents

Installation

export PPL_TOKEN="..."
curl "https://app.pipelogic.ai/api/v1/cli/$(uname -s | tr '[:upper:]' '[:lower:]')_$(arch)/$PPL_TOKEN" -o /home/$USER/bin/ppl
chmod +x /home/$USER/bin/ppl

Check PPL Version

Check Tool version

ppl version

Prompt

ppl v2.1.3

Upgrade PPL

ppl upgrade

Login

ppl login
E-Mail: orangecat@catperson.com
Password:

List Workspaces

List available workspaces

ppl workspace list

"*" Demonstrates the current workspace.

Prompt

┌───┬──────────────────────────────────────┬─────────────────────┬──────────────────────────────────────┬──────────────────────────────────────┐
│   │ ID                                   │ NAME                │ CREATED AT                           │ UPDATED AT                           │
├───┼──────────────────────────────────────┼─────────────────────┼──────────────────────────────────────┼──────────────────────────────────────┤
│ * │ s4da6e27-49f4-99ed-9b31-0242k91l1019 │ User workspace      │ 2024-10-12 06:11:30.696971 +0000 UTC │ 2024-10-12 16:11:30.696971 +0000 UTC │
│   │ 113ab75d-6ekk-11ee-a0b8-0132kc143006 │ Company workspace   │ 2024-10-19 20:08:52.935997 +0000 UTC │ 2024-10-19 10:08:52.935997 +0000 UTC │
└───┴──────────────────────────────────────┴─────────────────────┴──────────────────────────────────────┴──────────────────────────────────────┘

Switch Workspaces

ppl workspace switch

Prompt

Please choose the index of the workspace you want to use.
 1. User workspace
 2. Company workspace
 Select workspace [1-2]:

Initialize Component

Component initialization will create required folder structure and files.

ppl code init -t

You will be prompted to create a new component and initialize it with boilerplate project

No component provided, use --component flag to indicate an existing component.
Do you want to create and initialize a new component? (Y/n): y
Initialized successfully a new git repository.
Please select an identifier for the component. 
Identifier: Testing
Display Name [Testing]: 
No ReadMe provided. An editor will be opened to create the ReadMe. Please provide a description of the component in Markdown syntax. Press enter to confirm.
Successfully created component!

┌───┬───────────────────┬────────┬──────────┬────────────────────────────────┐
│ # │ NAME              │ SIZE   │ LANGUAGE │ TAGS                           │
├───┼───────────────────┼────────┼──────────┼────────────────────────────────┤
│ 1 │ Crop Image Python │ 1.2 kB │ py       │ one-to-one, image-manipulation │
│ 2 │ Dot Product       │ 3.6 kB │ py       │ many-to-many                   │
│ 3 │ Crop Image        │ 6.7 kB │ cpp      │ one-to-one, image-manipulation │
│ 4 │ Double Value      │ 585 B  │ py       │ default, one-to-one            │
│ 5 │ Dot Product       │ 5.5 kB │ cpp      │ many-to-many                   │
│ 6 │ Double Value      │ 1.8 kB │ cpp      │ default, one-to-one            │
└───┴───────────────────┴────────┴──────────┴────────────────────────────────┘
This operation will delete current configuration, tests or sources you might have right now.

Select a Component from list to finish your initialization

Select template [1-6]: 2
Component codebase has been initialized succesfully.

Fetch Component

The component is needed to be initalized and .pipecomponent needs to be present.

  ppl code fetch

Displays all available versions of the component in the server

┌───┬──────────────────────────────────────┬────────────────┬──────┬──────────────────────────────────────┐
│ # │ ID                                   │ AUTHOR         │ TAGS │ CREATED AT                           │
├───┼──────────────────────────────────────┼────────────────┼──────┼──────────────────────────────────────┤
│ 1 │ 54a136d3-bf91-401f-11ec-f3142436774d │ Orange Cat     │      │ 2024-03-21 17:43:57.852828 +0000 UTC │
│ 2 │ 11a336d3-9991-111f-11ec-f3111436774d │ Orange Cat     │      │ 2024-03-21 18:43:57.852828 +0000 UTC │
└───┴──────────────────────────────────────┴────────────────┴──────┴──────────────────────────────────────┘
This operation will delete current configuration, tests or sources you might have right now.
Select release [1-2]:

Compile Component

ppl code compile

Prompt

Build log: Successfully tagged registry.pipelogic.ai/pipelogic/dot_product:1714344292

Code has been compiled successfully.

Test Components

This works only if the component has tests defined in its folder.

Code Test

ppl code test

Prompt

Test log: Test 0: Success
Test log: Test 1: Success
Code test has run successfully.

Release Component

  ppl code release

Prompt

There are uncommitted changes in the repository.
Do you wish to continue? (y/N): y
No version notes provided. An editor will be opened to create them. Please provide a description of the component version using Markdown syntax. Press enter to confirm.
Code has been released succesfully.

Generate Type Definitions

Each component with named types needs to have type definition file. This file is used for converting incoming type into programming variables.

Generate Type definitions automatically based on config.yml

ppl code gentypes

Detects types present and saves them as pipetypes

The currently used types are:
1. Image.RGBA
2. BoundingBox
3. Point
4. DetectedClass
5. Image
6. Image.BGR
7. Image.BGRA
8. Image.GRAY
9. Image.RGB
10. Rectangle
Do you want to use some other types? (y/N):

Upload files

There are two file categories currently accepted on the Pipelogic platform: video, model. These categories exists to enable their specific use case.

To upload files to pipelogic storage:

Uploading Files

ppl file upload_video video.mp4

Prompt

Name: example file
No ReadMe provided. An editor will be opened to create the ReadMe. Please provide a description of the file in Markdown syntax. Press enter to confirm.
File has been uploaded successfully.

Upload Models

Pipelogic depends on triton server for inference. In order to upload a model to the platform to be readily used in a component, it's necessary to configure it to onnx format. Refer export models and then make sure it fits into expected model file structure described in features.

Upload Model

  ppl file upload_model example-model/

Prompt

  Name: example-model
  No ReadMe provided. An editor will be opened to create the ReadMe. Please provide a description of the file in Markdown syntax. Press enter to confirm.
  File has been uploaded successfully.

Uploaded models should follow Triton Inference Server file structure.

Rename Files

Rename Files

ppl file update file_id --name new_name

Add Readme

Appends a readme to file, when the file is downloaded the appended readme file will be downloaded as well.

Append Readme

ppl file update file_id --readme readme.md

Set Default Parameters for Model

Selected configuration file will added to model. When a model is selected inside a component, it will fill up all the configuration fields if a component is added to it.

Add Configuration File

ppl file update file_id --config configuration_file.yml

It also makes sure the model and component has the matching configuration fields, if they don`t the model will be greyed out and cannot be selected.

Inspect Existing Types

Inspect Type Definition

ppl type get type_name

Inspect BoundingBox Type

ppl type get BoundingBox

Prompt

BoundingBox := {class:DetectedClass,rectangle:Rectangle<Double>}

Create New Types

Create a type definition and save it as a text file rboundingbox

RBoundingBox := {class:DetectedClass,rectangle:Rectangle<Double>, angle:Double}

Add it to Local Fork

ppl type add rboundingbox

Push it to Pipelogic Server

ppl type push rboundingbox

Was this page helpful?