fix: Allow injecting ObjectMapper in FunctionTool, default to ObjectMapper (re. #473) #487
Conversation
Summary of ChangesHello @vorburger, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly improves the Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
7d99d03 to
f628dc9
Compare
There was a problem hiding this comment.
Code Review
This pull request successfully refactors FunctionTool to allow injecting a custom ObjectMapper, which is a great step towards better flexibility and testability.
I've left a couple of comments on areas that could be improved for consistency and correctness:
- The
ObjectMapperused for schema generation inFunctionCallingUtilsis still separate from the one injected intoFunctionTool, which could lead to inconsistencies. - There's a potential bug in how
Listparameters are handled in thecallLivemethod, which is different from the more robust implementation in thecallmethod.
Addressing these points will make the FunctionTool more robust, especially when dealing with custom types. Overall, this is a valuable change.
| this.funcDeclaration = | ||
| FunctionCallingUtils.buildFunctionDeclaration( | ||
| this.func, ImmutableList.of("toolContext", "inputStream")); | ||
| this.objectMapper = objectMapper; |
There was a problem hiding this comment.
While the runtime argument conversion now uses the injected objectMapper, the function declaration schema is built using FunctionCallingUtils, which has its own static, non-configurable ObjectMapper. This can lead to inconsistencies if the injected objectMapper has custom serializers, as the generated schema will not reflect them. This could cause mismatches between what the LLM expects and what the tool can handle at runtime. To ensure consistency, FunctionCallingUtils should also be updated to use the same ObjectMapper instance.
| } | ||
|
|
||
| private static List<Object> createList(List<Object> values, Class<?> type) { | ||
| private List<Object> createList(List<Object> values, Class<?> type) { |
There was a problem hiding this comment.
The callLive method, a caller of this function, has a potential bug. It obtains a java.lang.reflect.Type and casts it directly to Class before passing it here. This is unsafe and will cause a ClassCastException if the type is a ParameterizedType (e.g., for a List<Map<String, String>>). The call method contains the correct, more robust logic for handling this, which should be replicated in callLive to prevent runtime errors.
Related to #473.
Intentionally just for
FunctionTool, for now; but this should eventually really be done everywhere like this (IMHO).Alternative (and superior) to the initial #486 - but put on hold, until at least that basic version is merged (thus this is Draft).