CHANGELOG

2.7.1

• [bugfix] No longer allow a block with both a code and a script_ref input, as this can result in confusing behavior
• [improvement] Block.output() and workflow_input() return types adjusted to allow smoother typechecking experience for block definitions

2.7.0

• [feature] Hyperscience Platform V43 has been released!
  • updated blocks
  • starter examples
• [improvement] Generated flow JSON now includes an sdk_version field in the metadata section
• [improvement] MockHsBlockInstance.log() now correctly reflects the difference in logging behaviour between Python versions — see HsBlockInstance.log and Unit Testing Custom Code Blocks for more details
  • In PythonBlock with Python version <= 3.12, logging arguments via HsBlockInstance.log() is not supported
  • In PythonBlock with Python version >= 3.13, logging arguments is supported
• [deprecated] FOLDER_TRIGGER has been deprecated across multiple IDP versions in favor of the Universal Folder Listener
• [bugfix] Fixed BundleExporter hardcoding Python code block identifiers, which prevented correct export of flows using newer PythonBlock versions
• [bugfix] Fixed unstable ordering of roles across IDP versions V38–V42

2.6.0

• [feature] Hyperscience Platform V42.3 has been released!
  • updated blocks
  • starter examples
• [feature] In V42.3+ a new Document Processing with ORCA Subflow has been added for Advanced and Premium license
  • IDP ORCA Flow - a read-only subflow similar to IDP Core Flow but ORCA is used for field identification
  • ORCA QA Flow - Quality assurance for ORCA flow. Meant to be used together with IDP ORCA Flow to spawn QA tasks for human review

2.5.1

• [bugfix] Fix MockHsBlockInstance.log function having an outdated signature

2.5.0

• [feature] Hyperscience Platform V42.2 has been released!
  • updated blocks
  • starter examples

2.4.2

• [bugfix] Allow Code Blocks to use static methods and class methods

2.4.1

• [feature] Added support for using entire modules as code in Code Blocks.
  • Module-backed CCBs enable better code organization, maintainability, and testability
  • For further details see the PythonBlock documentation
• [feature] Comprehensive unit testing support for Custom Code Blocks
  • Added MockHsBlockInstance and MockHsTask classes in flows_sdk.mocks for testing CCBs locally
  • Mock objects simulate blob storage, logging, HTTP requests, and measurements
  • See the new Unit Testing Custom Code Blocks guide for more details and examples
• [improvement] Added comprehensive documentation for several flow utilities
  • Added docstrings with code examples for workflow_input(), FlowDefinitionReferences, and FlowRunReferences

2.4.0

• [feature] Hyperscience Platform V42.1 has been released!
  • updated blocks
  • starter examples
• [improvement] The On-Error flows have been split in two parts:
  • The On-Error flow that is configured on the main flow, which is responsible for handling the error and calling the On-Error subflow
  • The On-Error subflow, which is responsible for generating the error message to be sent to the output blocks
  • Similarly to the Document processing flows, this allows for better separation of concerns and updating the on-error logic without losing any configure output blocks.

2.3.0

• [improvement] Added support for the Python 3.13 PythonBlock

2.1.0

• [feature] Hyperscience Platform V42 has been released!
  • updated blocks
  • starter examples

2.0.0

• [breaking change] Upgrade to Python 3.12 - using blocks from V41 now requires a 3.12 venv
• [feature] Updated blocks for V41.2

1.11.3

• [feature] Backport to V40 IDP Library the support for forcing normalization errors and blank or missing fields to supervision

1.11.2

• [improvement] Added support for empty branches in Routing and Fork blocks:
  • Allow empty branches (branches without any blocks)
  • Allow branches to define an output from an external block, without requiring any blocks to run.**

1.11.1

• [improvement] Updated documentation for chat component for LLM Document Q&A in Custom Supervision

1.11.0

• [feature] Hyperscience Platform V41 has been released!
  • updated blocks
  • starter examples

1.10.5

• [feature] Refer to the business metric measurement occurrence time via time, a datetime object. This datetime must be timezone-aware

1.10.4

• [feature] Added FileFilterBlock to V40 IDP Library

1.10.3

• [feature] Added publish measurements method to HsBlockInstance that allows code in PythonBlock or CodeBlock to tracks business metrics (available since V41)

1.10.2

• [improvement] Added support for the Python 3.12 PythonBlock

1.10.1

• [feature] Added utility methods to HsBlockInstance for making authenticated requests to the Hyperscience instance from inside the code in PythonBlock or CodeBlock (available since R40.2)

1.10.0

• [feature] Hyperscience Platform V40 has been released!
  • updated blocks
  • starter examples
• [bugfix] Removed default Submission Aware On Error flow from V38 examples because it might not exist in a V38 system.

1.9.1

• [bugfix] Updated default inputs for ReprocessingBlock in versions 39 and 39.1, so that additional inputs in custom IDP_CORE flows would get relayed for reprocessing runs

1.9.0

• [feature] Hyperscience Platform V39.1 has been released!
  • updated blocks
  • starter examples
• [improvement] Roles are exposed as an argument to the IDP Core Block
• [improvement] Easier to see where to modify Notification, On Error flows in the Document Processing example flow
• [improvement] Default Notification flow, On Error flow, and IDP Core block in the example flow point to the out-of-the-box flows in the platform

1.8.3

• [bugfix] Reverted the change introduced in Flows SDK version 1.8.2 where the default value for PythonBlock#python_version is set to Python 3.11, which caused breaking behaviour.
• [improvement] Added support for an environment variable HS_PYTHON_CODE_BLOCK_PYTHON_VERSION that allows overriding the default behaviour without having to make code changes to existing flows.
  • To understand how to control Python versions for PythonBlocks, Please refer to the PythonBlock documentation

1.8.2

• [bugfix] Default version of PythonBlock is now Python 3.11, instead of 3.9

1.8.1

• [improvement] Ported quality of life features introduced in the V39 samples to V38 samples
  • Roles are exposed as an argument to the IDP Core Block
  • Easier to see where to modify Notification, On Error flows in the Document Processing example flow for V38
  • Default Notification flow, On Error flow, and IDP Core block in the example V38 flow point to the out-of-the-box flows in the platform for that version
• [improvement] Moved IDPWrapperManifest out of the example Document Processing flow samples for both V38 and V39
• [improvement] Added parameter that controls which Python version a PythonBlock works with

1.8.0

• [feature] Hyperscience Platform V39 has been released!
  • updated blocks
  • starter examples
• [improvement] Roles are exposed as an argument to the IDP Core Block
• [improvement] Easier to see where to modify Notification, On Error flows in the Document Processing example flow
• [improvement] Default Notification flow, On Error flow, and IDP Core block in the example flow point to the out-of-the-box flows in the platform

1.7.2

• Bump asttokens to 2.4.1

1.7.1

• [bugfix] Default CustomSupervision’s notification_workflow to None
• [improvement] Add release.zip to the idp v38 starter flow
• [improvement] Properly identify settings as Classification settings

1.7.0

• [feature] Hyperscience Platform V38 has been released!
  • updated blocks
  • starter examples

1.6.4

• [feature] Add support for on_error_flow construct in error handling.
  • Can be used only in the flow level error handling section
  • Available in Hyperscience Platform V38 and above.

1.6.3

• [feature] Add support for compatibility_spec construct in blocks.
  • Available in Hyperscience Platform V38 and above.

1.6.2

• [bugfix] Remove extra parameter http_config from SubmissionBootstrap block for R32.

1.6.1

• [improvement] Setting a range for pydantic between 1.10.11 inclusive and 2.0 non-inclusive.

1.6.0

• [feature] Hyperscience Platform V37 has been released!
  • updated blocks
  • starter examples

1.5.2

• [improvement] Updated pydantic version to 1.10.8.

1.5.1

• [bugfix] Added missing http_config parameter to SubmissionBootstrapBlock.
• [improvement] Updated documentation for I/O blocks.

1.5.0

• [feature] Hyperscience Platform V36 has been released!
  • updated blocks
  • starter examples

1.4.1

• [feature] Decision dependencies for Custom Supervision is available!
• [feature] Mandatory decisions for Custom Supervision is also available!

1.4.0

• [feature] Hyperscience Platform V35 has been released!
  • updated blocks
  • starter examples
  • Check out some of the new features v35 features, released in recent flows-sdk versions:
    • PythonBlock, how to store and load binary data, use 3rd party libraries and more in 1.1.0
    • How to use structured logging, also visible in the UI in 1.3.0
• [feature] A new system block has been introduced - Foreach!
  • This block enables dynamic parallelization of tasks - check out the docs to learn more

1.3.0

• [improvement] Added a log method to HsBlockInstance

1.2.1

• [improvement] A reference to the code function passed to CodeBlock and PythonBlock is now available on the block object under block.code_fn

1.2.0

• [feature] Hyperscience Platform V34 has been released!
  • [feature] New block - HyperscienceRestApiBlock available for V32, V33, and V34
• [deprecated] Removed the Kofax Folder Listener IOBlock

1.1.0

• [feature] Added a new block - PythonBlock

This requires a Hyperscience platform version 34.0.0+

PythonBlock is a special type of block that can execute custom python code on the Hyperscience platform. Similarly to CodeBlock, the code can utilize packages that are part of the Python Standard Library, but it also allows developers to utilize third-party python packages.

• [feature] Introduces HsTask and HsBlockInstance system arguments for code blocks

This requires a Hyperscience platform version 35.0.0+

1.0.1

• [improvement] new properties for v32 CaseCollationBlock and v33 MachineCollationBlock

1.0.0

• [improvement] reference_name is now an optional property for all blocks.