Added support for latest GPT4o model and support for JSON schemas for OpenAI

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.7...v0.4.8

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.6...v0.4.7

[feat] Added gpt-4o mini support
[internal] Adjusted logging directory functionality to support Elucidate

[chore] refreshed internal dependencies
[chore] depreciated old Gemini vision models
[chore] moved away from Anthropic Beta

0.4.5 build exists on PyPI, but no official release exists due to it being an internal build.

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.4...v0.4.6

Added support for claude-3-5-sonnet-20240620

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.3...v0.4.4

Added add support for Message Utility classes.

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.2...v0.4.3

Added support for grabbing credentials from env variables if not given in set_credentials function.
Updated Gemini cost for cost calculation.
Slight bug fix regarding price case for cost calculation.
Bumped google-genai to 0.6.0

What's Changed

• Feature/env support by @alemalvarez in https://github.com/Bikatr7/EasyTL/pull/20

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.1...v0.4.2

Fixed a bug regarding test_credentials() for azure failing due to a bad region. This fix requires passing in your region to test_credentials() for azure.

Better error handling for test_credentials() (exceptions no longer fail silently)

What's Changed

• Add default region in Azure and Note block in the readme by @alemalvarez in https://github.com/Bikatr7/EasyTL/pull/18
• Fix/azure region by @alemalvarez in https://github.com/Bikatr7/EasyTL/pull/19

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.0...v0.4.1

EasyTL v0.4.0 Release Patch Notes

We are excited to announce the official release of EasyTL v0.4.0! This version brings a host of new features, improvements, and bug fixes. Below are the detailed patch notes:

New Features:

• Microsoft Azure Support: Integration with Microsoft Azure services is now available, providing more options for translation needs.
• Anthropic Support: EasyTL now supports Anthropic services, expanding the range of supported LLMs.
• Gemini Flash Models: Added support for Gemini Flash models.
• New Return Formats: Introduced new return formats, including json and raw_json, for more flexible data handling. The raw_json option will return the full response object where applicable and instruct the model to return a valid JSON string.
• Response Schema Support: Implemented response schema support for Gemini & Anthropic services, ensuring consistency and reliability in responses.

Improvements:

• Bug Fixes and Optimizations: Numerous bugs have been addressed, and performance optimizations have been implemented to enhance the overall user experience.
• Enhanced LLM Translations: Improved error handling for improper translation requests.
• Refined Internal Structure: The internal architecture has been refined for clearer and more efficient processing.
• Improved Exception Handling: Enhanced mechanisms for managing exceptions, with all exceptions now prefaced with their origin (e.g., DeepLAuthorizationError and OpenAIAuthorizationError).
• Cost Calculation Improvements: More accurate and efficient cost calculations have been implemented for supported LLMs (Anthropic, Gemini, and OpenAI)
• Documentation: Comprehensive and detailed documentation has been significantly improved, making it easier to understand and use EasyTL. We're also working on external documentation here
• Deprecation Notices: set_api_key and test_api_key methods have been deprecated. Please use set_credentials and test_credentials instead.

Bug Fixes:

• Fixed issues where semaphores were not being followed in certain edge cases, ensuring better concurrency handling.
• Corrected the absence of certain gpt-3.5-turbo models in the list of valid JSON models.
• Internal dependency changes to support new models and functionality.

Other Updates:

• Requirements File: Added a requirements.txt file for easier dependency management, in addition to listing dependencies in pyproject.toml.
• OpenAI Service Refactor: Refactored the OpenAI service to follow the NOT_GIVEN schema, where None values are explicitly marked and missing values are not passed to the API.

We hope you enjoy the new features and improvements in EasyTL v0.4.0! Your feedback is invaluable to us, so please do not hesitate to share your thoughts and suggestions.

What's Changed

• Azure snippet and start of core functionality for AzureService by @alemalvarez in https://github.com/Bikatr7/EasyTL/pull/7
• Add gemini flash and response schema support by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/10
• Add proper json mode and response schema to anthropic by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/11
• Make easytl utilize anthropic by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/12
• Merge in mostly completed azure-dev to be completed. by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/13
• Complete azure service by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/14
• Reorganized src by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/15
• added not_given to openai (thought) this would be longer by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/16
• Refactor readme also fixed a bug with text length validation by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/17

New Contributors

• @alemalvarez made their first contribution in https://github.com/Bikatr7/EasyTL/pull/7

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.3.3...v0.4.0

Here are the release notes for the first beta build of EasyTL v0.4.0. Detailed patch notes will be provided soon.

EasyTL v0.4.0 Beta Build Highlights:

• Microsoft Azure Support: Integration with Microsoft Azure is now available.
• Anthropic Support: Added support for Anthropic services.
• Bug Fixes and Optimizations: Numerous bugs have been addressed, and performance optimizations have been implemented.
• Enhanced LLM Translations: Improved error handling for improper translation requests.
• Refined Internal Structure: The internal architecture has been made clearer and more efficient.
• Improved Exception Handling: Enhanced mechanisms for managing exceptions.
• Comprehensive Documentation: Significantly better and more detailed documentation.
• Cost Calculation Improvements: More accurate and efficient cost calculations.
• Extended Model Support: Increased support for various LLM models.
• New Return Formats: Added support for new return formats, including json and raw_json. Additionally, JSON response schema support has been implemented where applicable.

Stay tuned for detailed patch notes.

What's Changed

• Azure snippet and start of core functionality for AzureService by @alemalvarez in https://github.com/Bikatr7/EasyTL/pull/7
• Merge in mostly completed azure-dev to be completed. by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/13
• Complete azure service by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/14
• Reorganized src by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/15

New Contributors

• @alemalvarez made their first contribution in https://github.com/Bikatr7/EasyTL/pull/7

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.0-alpha-3...v0.4.0-beta

Third Alpha build for Kudasai, proper patch notes pending, but v0.4.0 intends to add Microsoft Azure Translate and Anthropic API support to EasyTL, along with a multitude of bug fixes, improvements

Since last alpha:

• Added RAW_JSON mode.
• Stability improvements
• EasyTL know checks if incoming text is too much to translate for LLM models.
• More cost calculation bug fixes

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.4.0-alpha-2...v0.4.0-alpha-3

Second Alpha build for Kudasai, proper patch notes pending, but v0.4.0 intends to add Microsoft Azure Translate and Anthropic API support to EasyTL, along with a multitude of bug fixes and improvements.

Since last alpha:

• Fixed Gemini cost calculation

v0.4.0-alpha build

Alpha build for Kudasai, proper patch notes pending, but v0.4.0 intends to add Microsoft Azure Translate and Anthropic API support to EasyTL, along with a multitude of bug fixes and improvements.

What's Changed

• Add Gemini flash and response schema support by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/10
• Add proper JSON mode and response schema to anthropic by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/11
• Make easytl utilize anthropic by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/12

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.3.3...v0.4.0-alpha

EasyTL v0.3.3 Patch Notes

New Features

• GPT-4o Support: Added support for the GPT-4o branch of OpenAI models.

Improvements and Bug Fixes

• JSON Mode Fix: Fixed a bug where some OpenAI models were incorrectly marked as not supporting JSON mode.
• Platform Identification: Added platform identification information to DeepL.
• OpenAI Enhancements: Implemented slight bug fixes and internal cleanup for OpenAI models.

Internal Changes

• Anthropic Prep: Prepared internal systems for Anthropic support.
• Service Addition: Added Anthropic service and necessary constants.

What's Changed

• [internal] Added anthropic service and some constants it needs with it. by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/9

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.3.1...v0.3.3

EasyTL v0.3.1 Patch Notes

New Features

• Automatic Retry: OpenAI and DeepL now feature automatic retry mechanisms for requests. These are applied automatically, unless overridden by user decorators.

Improvements and Bug Fixes

• Decorator Bug Fix: Resolved an issue where user decorators were not correctly applying to DeepL and Google Translate.
• Security Policy Update: Provided instructions for encrypted reporting.

Internal Changes

• Testing Adjustments: Made changes to internal testing.

What's Changed

• Standardize decorator parameter usage by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/8

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.3.0...v0.3.1

EasyTL v0.3.0 Patch Notes

Depreciation

• Authentication Update: Replaced set_api_key with set_credentials to accommodate Google Translate
• Testing Methodology: Shifted from test_api_key_validity to test_credentials to align with new authentication methods.

New Features

• Google Translate Support: Integrated Google Translate for additional translation capabilities.
• Credential Functions: Added set_credentials and test_credentials for streamlined user authentication.

Improvements and Bug Fixes

• Documentation Updates: Minor corrections and updates in documentation; removed unnecessary import.
• Security and Workflow: Added a new security policy, updated the README, and made workflow enhancements.
• Deprecation Warnings: Informed users about the removal of deprecated functions in upcoming versions.

What's Changed

• Add GoogleTLService and some other requirements excluding easytl.py needs by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/5
• Implement-googletlservice by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/6

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.2.1...v0.3.0

Fix bug regarding rate limit delays not properly applying for OpenAI or Gemini

EasyTL v0.2.0 Patch Notes

Upgrades

• Python Version: Updated to Python 3.10
• Gemini Python Library: Bumped the Gemini python library to version 0.5.1, improving integration and support for newer features such a json output and system message
• Cost and Model Adjustments: Updated the cost calculations and expanded the whitelist of models for more precise and efficient resource management.

New Features

• Raw Response Type Option: Added a raw response type option to all services, allowing users to obtain the original translation object directly from the API.
• JSON Response Type: Introduced a json type return option for Gemini and OpenAI services, enabling models to return a JSON-parseable string, facilitating better integration with applications requiring JSON formatted responses.
• Rate Limit Management: Added a rate_limit parameter to control the delay between translations for iterable inputs, improving handling under rate limit constraints.

Improvements and Bug Fixes

• Documentation and Minor Bugs: Made various documentation updates and minor bug fixes, including a fix for handling null system messages when using previous settings and correcting default translation instructions for LLMs.
• System Message Handling in Gemini: Addressed a bug related to the Gemini system message when not overriding previous settings, ensuring consistent behavior across sessions.
• Codebase Optimizations: Implemented several optimizations across the codebase to improve performance and reliability.

What's Changed

• v0.2.0 beta build | Add-json-and-raw-output-response-options-to-openai-and-gemini by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/4

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.1.2...v0.2.0

Some Internal testing changes

Beta v0.2.0

• Proper Patch Notes will be introduced upon full build, but in summary, raw and json output modes were added.

What's Changed

• v0.2.0 beta build | Add-json-and-raw-output-response-options-to-openai-and-gemini by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/4

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.1.2...v0.2.0-beta

EasyTL v0.1.2 Patch Notes

Upgrades

• Google GenerativeAI 0.5.0 Integration: Moved to google-generativeai 0.5 to utilize its system prompt feature, enhancing prompt handling capabilities.

New Features

• OpenAI Exception Handling: Enhanced exception management with more OpenAI exceptions made importable, improving error resolution processes.
• Gemini System Parameter: Introduced system parameter support for the pro-1.5 Gemini model, enabling more refined control over translation configurations.
• Asynchronous Request Management: Added semaphore mechanisms for managing asynchronous requests and included the ability to customize semaphore settings, optimizing resource usage during high concurrency.

Improvements and Bug Fixes

• DeepL Parameter Fix: Corrected the default parameter for target language in DeepL translation methods, ensuring accuracy in language targeting.
• Parameter Validation Enhancements: Made several minor improvements in parameter validation to ensure robustness and reliability in API interactions.

What's Changed

• Implement system message for gemini-pro 1.5 by @Bikatr7 in https://github.com/Bikatr7/EasyTL/pull/3
• Thanks to @trulybeloved for the idea to add semaphore and planned logging functionality.

Full Changelog: https://github.com/Bikatr7/EasyTL/compare/v0.1.1...v0.1.2