0.9.2

What's Changed

• Added ability for sources and segments to have multiple names in chatterlang.
• Removed signature segments.
• Added Anthropic as a chat provider.
• Updated the install options so that not all the providers have to be installed. Include the provider
  you want as a parameter (e.g. pip install talkpipe[anthropic]). Current options are anthropic, openai,
  and ollama. Use "all" to install all three.

• deps(deps): bump numba from 0.61.2 to 0.62.0 by @dependabot[bot] in #28
• deps(deps): bump numba from 0.62.0 to 0.62.1 by @dependabot[bot] in #29

Full Changelog: v0.9.1...v0.9.2

0.9.1

Added an import in the talkpipe init.py so lancedb is available to chatterlang.

Full Changelog: v0.9.1...v0.9.1

Version 0.9.0

New and Updated Segments and Sources

• Added set, which simply assigns some constant to a key.
• Added isTrue and isFalse segments that can function as both a filter or an evaluation of a field.
• Added llmBinaryAnswer segment for coercing true/false answers from an LLM.
• Added searchLanceDB and addToLanceDB for creating and searching LanceDB vector databases. This
  involved creating a LanceDBDocumentStore class that operates in ways compatible with simplevectordb and
  whoosh.
• Renamed extractor to readFile.

Improvements

• Configured CI/CD system for GitHub to run tests, security checks, and container builds.
• Fixed identified security vulnerabilities.
• Renamed append_as to set_as and appendAs to setAs throughout the codebase. "Set" is more accurate than "append" for the associated behavior.
• Changed asFunction to as_function.
• Added a check and throw an exception in simplevectordb if the user has clustered and then tries to use cosine for search.
• Added a plugin system so it is easier for external whl files to add commands to chatterlang.
• Refactored the documentation system:
  • Pulls from the registry in real time. This ensures that plugin commands are
    included in the documentation system. It also reduces potential problems from bad parsing of source code.
  • Pulls "Annotated" typing from parameter names to create the Parameters section of the documentation.
    Makes for cleaner, more consistently up-to-date documentation. The use of Annotated is optional.
• Updated isIn and isNotIn to function like isTrue so that they need not always be filters.
• Created an AbstractFieldSegment class and changed the field_segment decorator to use it. This makes it easier
  to create segments with consistent "field_segment" behavior but require additional initialization.
• Renamed ExtractFile to ReadFile and refactored it to be descended from AbstractFieldSegment.
• Renamed extract to readFile in chatterlang for consistency.
• Deprecated simplevectordb classes and segments. These will be removed in 1.0.0. The lancedb vector database
  provides equivalent functionality and doesn't require all the data to be in memory at once.
• Added an AbstractFieldSegment class and refactored field_segment to use it. This is so classes can be
  implemented with field_segment functionality that also need some one-time initialization.

Refactored the extract segment.

Full Changelog: v0.9.0a4...v0.9.0a5

Minor installation fix

• Minor change to fix the version of numba to address an install issue when installing with --pre because of an issue with a numba release candidate.

Full Changelog: v0.9.0a3...v0.9.0.a4

Updated Documentation

Full Changelog: v0.9.0a2...v0.9.0a3

Same as a1, but with potential fix for docker image creation

v0.9.0a2

Potential fix for docker container build for a release.

Cumulative changes

Includes new CI-CD workflows, refactored docker image, plugin system, refactored documentation browswer, some API breaking changes from 0.8.