ts-sdk 2.0.0

Tetrascience Python SDK

ts-sdk

Tetrascience Python SDK

Version

v2.0.0

Table of Contents

• Install
• Usage
  • Init a new protocol
  • Upload artifact
  • IDS Validation
• Task Script
  • Config.json
    • Overview
    • Allowed IDS
• Changelog
  • v1.4.0
  • v1.3.10
  • v1.3.9
    • CLI Improvements
    • Context Function Improvements
    • Other
  • v1.3.8
  • v1.3.7
  • v1.3.6
  • v1.3.5
  • v1.3.2

Install

pip3 install ts-sdk

Usage

Init a new protocol

ts-sdk init -o <org> -p <protocol-slug> -t <task-script-slug> -f <protocol-folder>
cd <protocol-folder>/task-script
pipenv install --dev
# task-script code modifications...
pipenv run pytest

Upload artifact

export TS_ORG=<your-org-slug>
export TS_API_URL=https://api.tetrascience.com/v1
export TS_AUTH_TOKEN=<token>
ts-sdk put <ids|protocol|task-script> <namespace> <slug> <version> <artifact-folder>

It's also possible to use the configuration JSON file (cfg.json):

{
    "api_url": "https://api.tetrascience.com/v1",
    "auth_token": "your-token",
    "org": "your-org",
    "ignore_ssl": false
}

Usage: ts-sdk put <ids|protocol|task-script> <namespace> <slug> <version> <artifact-folder> -c cfg.json

IDS Validation

When uploading IDS artifact, validation will be performed using ts-ids-validator package. Validation failures for IDS will be printed on the console.

Task Script

Config.json

Overview

All Task Scripts have a config.json file that describes:

• what language a Task Script is written in
• the functions a Task Script has available for a Protocol to use
• the runnerType, an optional field that is only used when letting the artifact builder know a Windows script (identified by using the value “windows-script”) is being built
• the maxCount of container instances that will be used by the task
• and an optional field describing the amount of memory a task will need, memoryInMB

Allowed IDS

Within the functions object of the configuration, the optional field allowedIds describes the IDS types a Task Script can produce.

The allowedIds field is used by the Context API's context.write_file (when writing an IDS) and context.write_ids functions to validate that the ids parameter of either function is allowed before writing an IDS.

allowedIds can be used in the following ways:

1. A single object with three properties: namespace, slug, version

   • When a function only generates one kind of IDS

   • The ids parameter of the context.write_file and context.write_ids functions are optional since there is only one value in the allowedIds field

   • Example:

     "functions": [
       {
         "slug": "generates-ids",
         "function": "main.generate_ids",
         "allowedIds": {
           "namespace": "common",
           "slug": "my-ids",
           "version": "v1.0.0"
         }
       }
     ]
     

2. An array of those objects with the three properties: namespace, slug, version

   • When a function can generate multiple IDS types

   • The ids parameter of the context.write_file and context.write_ids functions is required and will be validated against the allowedIds to confirm the task script is allowed to write that IDS

   • Example:

     "functions": [
       {
         "slug": "generates-ids",
         "function": "main.generate_ids",
         "allowedIds": [
           {
            "namespace": "common",
            "slug": "my-ids",
            "version": "v1.0.0"
           },
           {
            "namespace": "common",
            "slug": "my-ids",
            "version": "v1.1.0"
           },
           {
            "namespace": "common",
            "slug": "my-other-ids",
            "version": "v1.0.0"
           }
         ]
       }
     ]
     

3. null

   • When a function does not generate any IDS

   • Calls to context.write_file (when file_category is “IDS”) and context.write_ids will raise an exception

     • When the file_category parameter is not "IDS", then context.write_file is not writing an IDS and therefore will not raise an exception

   • Example:

     "functions": [
       {
         "slug": "generates-ids",
         "function": "main.generate_ids",
         "allowedIds": null
       }
     ]
     

4. allowedIds not defined

   • Any IDS JSON can be written via context.write_file or context.write_ids
     • no allowed IDSs are specified, so no check is performed
   • Any IDS JSON passed to the context.write_file or context.write_ids still has to pass validation against the IDS specified by the ids parameter

Changelog

v2.0.0

• Secure mode + bug fixes

v1.4.2

• upgrade smart_open (v4.2.0 → v6.3.0)

v1.4.1

• Make allowedIDS non-mandatory for task-script, to maintain backward compatibility

v1.4.0

• Added functionality to support allowedIds feature

v1.3.10

• Internal fixes (/update-status API call retry)

v1.3.9

CLI Improvements

• Make artifact build failure more evident to the user
• ts-sdk put will check for requirements.txt file before uploading

Context Function Improvements

• Add support to write_file to receive a dictionary instead of only strings
• Validate dictionary inputs to write_file against IDS
• Enforce IDS input to write_ids must be a dictionary
• Validate length of the name and value of labels
• Validate labels more thoroughly, in more places
• Add a new DataClass for better label definition

Other

• All code formatted with black
• Adjust abstract dependencies in setup.py to be less strict so that it will work better within task scripts
• Replace json library with simplejsonfor speed and usability

v1.3.8

• Internal fixes (secrets handling)

v1.3.7

• Add --exclude-folders argument to ts-sdk put task-script that excludes common folders that generally do not need to be part of the uploaded task script (e.g. .git, example-input, __tests__)
• Add local check to prevent uploading artifacts using ts-sdk put that would be rejected by the server for being too large
• Improve error messages for adding invalid labels to files

v1.3.6

• Add new s3 meta:
  • DO_NOT_INHERIT_LABELS
  • CONTENT_CREATED_FROM_FILE_ID

v1.3.5

• Fix bug where datalake file_key was incorrectly generated

v1.3.2

• Update context.write_file() to validate file upload path
• Fix logging issues
• Improve namespace validation
• Update print functionality to be more accurate and group arguments to the same call together