Bumps [langchain](https://github.com/hwchase17/langchain) from 0.0.242 to 0.0.249.
- [Release notes](https://github.com/hwchase17/langchain/releases)
- [Commits](langchain-ai/langchain@v0.0.242...v0.0.249)

---
updated-dependencies:
- dependency-name: langchain
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <[email protected]>

Bumps [rich](https://github.com/Textualize/rich) from 13.4.2 to 13.5.2.
- [Release notes](https://github.com/Textualize/rich/releases)
- [Changelog](https://github.com/Textualize/rich/blob/master/CHANGELOG.md)
- [Commits](Textualize/rich@v13.4.2...v13.5.2)

---
updated-dependencies:
- dependency-name: rich
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <[email protected]>

The codebase has been refactored to move all interactions with language models to a new class, LLM. This change improves the organization of the code and makes it easier to manage language model interactions. The Coder class has been simplified as a result. The LLM class now handles tasks such as getting the token length of a text, initializing a language model, and determining the appropriate model to use based on token size. This change affects multiple files across the codebase.

…, set up for using different models

The LLM class has been replaced with a more descriptive LanguageModelManager class. This change has been propagated across multiple files including agents.py, coder.py, alignment.py, commit.py, debug.py, learn.py, review.py, sidekick.py, and input.py. The new class provides the same functionality but with a more intuitive name. This change improves code readability and maintainability.

…e of a git repo

Resolves #75

The handling of project files in the SidekickCompleter class has been refactored for efficiency and clarity. The list of project files is now lazily loaded and cached, only when needed. This change improves performance by avoiding unnecessary file system operations, especially in large repositories. Additionally, the logic for determining whether to use the gitignore file for filtering has been encapsulated within the new `project_files` property, making the code more maintainable and easier to understand. 🚀🧹

…user specified model config

… have tests use a test config file.

….249

Bump langchain from 0.0.242 to 0.0.249

This commit introduces a more informative live interaction with the language models across multiple commands. Now, users will be aware of which model they are communicating with and through which provider. This change also ensures that the response from the language model is updated live, providing a more interactive experience. Additionally, support for HuggingFace Hub models has been added, expanding the range of models users can interact with. Let's keep the conversation going! 🗣️💬

In this commit, we've refactored the way input is handled in the sidekick commands. We've replaced the use of `input_prompt` with a more robust `PromptSession` from the `prompt_toolkit` library. This change allows us to bundle history, auto-suggestion, and completion into a single session, making the code cleaner and more maintainable. We've also encapsulated the creation of this session into a new function `generate_prompt_session` for better code organization. 🧹👌

…rience 🚀

This commit introduces several enhancements to AICodeBot. The README.md file has been updated with more detailed instructions on how to configure the language model to use, manage tokens effectively, and handle larger context sizes. The version in the configuration file has been updated to 1.2.

In addition, the language model manager now dynamically switches to a larger model if the context size exceeds the limit of the current model. This feature is supported for both OpenAI and OpenRouter models.

The user interface has also been improved. When a request is sent to the language model, a message is displayed indicating which model is being used. If the model is changed due to the context size, a warning message is displayed.

Finally, the test suite has been updated to cover these new features.

These changes should make AICodeBot more robust and user-friendly, especially when dealing with larger context sizes. 🎉

Bump rich from 13.4.2 to 13.5.2

This commit refactors the way we read the configuration in `lm.py` to ensure that we always return a dictionary, even when the configuration file is not found. This change simplifies the code and makes it more robust.

In `output.py`, we've updated the comment to better reflect the actual code change, which is setting the padding to 0.

In `test_lm.py`, we've expanded the test coverage to include more models. This will help us catch potential issues with these models earlier.

These changes should improve the overall quality and reliability of our code. 🚀👍🏽