feat(llm): enhance error logging and add comprehensive unit tests

[marevol]

Summary

• Improve LLM client error handling by including response body in error log messages for better debugging of API failures
• Add extensive unit tests using MockWebServer to verify chat and streaming functionality for both OpenAI and Gemini clients
• Add mockwebserver test dependency to pom.xml

Changes Made

• GeminiLlmClient.java: Added response body extraction and logging for error responses in both chat() and streamChat() methods
• OpenAiLlmClient.java: Added response body extraction and logging for error responses in both chat() and streamChat() methods
• pom.xml: Added mockwebserver test dependency for HTTP mocking in unit tests
• GeminiLlmClientTest.java: Added comprehensive tests for:
  • chat() method: success cases, error responses (401, 429, 500, 503), empty candidates, null finish reason, partial usage metadata
  • streamChat() method: success cases, multiple chunks, error responses, finish reasons, malformed JSON handling
• OpenAiLlmClientTest.java: Added comprehensive tests for:
  • chat() method: success cases, error responses (401, 429, 500, 503), empty choices, null finish reason, partial usage
  • streamChat() method: success cases, multiple chunks, error responses, SSE format handling, malformed JSON handling

Testing

• All new tests use MockWebServer to simulate API responses
• Tests cover both happy path and error scenarios
• Existing tests remain passing

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR enhances error handling and testing infrastructure for LLM client implementations by adding error response body logging and comprehensive unit tests.

Changes:

• Enhanced error logging in OpenAiLlmClient and GeminiLlmClient to include response bodies for better debugging
• Added extensive unit tests for both clients covering success cases, error responses, and edge cases
• Added mockwebserver test dependency to support HTTP mocking in tests

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/main/java/org/codelibs/fess/llm/openai/OpenAiLlmClient.java	Added response body extraction and logging in error handling for chat() and streamChat() methods
src/main/java/org/codelibs/fess/llm/gemini/GeminiLlmClient.java	Added response body extraction and logging in error handling for chat() and streamChat() methods
src/test/java/org/codelibs/fess/llm/openai/OpenAiLlmClientTest.java	Added comprehensive test suite using MockWebServer for chat and streaming functionality
src/test/java/org/codelibs/fess/llm/gemini/GeminiLlmClientTest.java	Added comprehensive test suite using MockWebServer for chat and streaming functionality
pom.xml	Added mockwebserver test dependency

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The variable name errorBody should follow Java naming conventions with camelCase starting with lowercase. While it already does, consider a more descriptive name like responseBody since this represents the full response body, not just error content.

Suggested change

	String errorBody = "";
	if (response.body() != null) {
	try {
	errorBody = response.body().string();
	} catch (final IOException e) {
	// ignore
	}
	}
	logger.warn("OpenAI API error. url={}, statusCode={}, message={}, body={}", url, response.code(), response.message(),
	errorBody);
	String responseBody = "";
	if (response.body() != null) {
	try {
	responseBody = response.body().string();
	} catch (final IOException e) {
	// ignore
	}
	}
	logger.warn("OpenAI API error. url={}, statusCode={}, message={}, body={}", url, response.code(), response.message(),
	responseBody);

[Copilot]

The variable name errorBody should be more descriptive. Consider renaming to responseBody since it represents the full response body, not just error content.

Suggested change

	String errorBody = "";
	if (response.body() != null) {
	try {
	errorBody = response.body().string();
	} catch (final IOException e) {
	// ignore
	}
	}
	logger.warn("Gemini API error. url={}, statusCode={}, message={}, body={}", url, response.code(), response.message(),
	errorBody);
	String responseBody = "";
	if (response.body() != null) {
	try {
	responseBody = response.body().string();
	} catch (final IOException e) {
	// ignore
	}
	}
	logger.warn("Gemini API error. url={}, statusCode={}, message={}, body={}", url, response.code(), response.message(),
	responseBody);

[Copilot]

The trailing slash removal logic is duplicated in both test files. Consider extracting this to a shared utility method to reduce duplication and improve maintainability.

Suggested change

	private void setupClientForMockServer() {
	final String baseUrl = mockServer.url("").toString();
	// Remove trailing slash
	final String apiUrl = baseUrl.endsWith("/") ? baseUrl.substring(0, baseUrl.length() - 1) : baseUrl;
	private String removeTrailingSlash(final String url) {
	if (url == null) {
	return null;
	}
	return url.endsWith("/") ? url.substring(0, url.length() - 1) : url;
	}
	private void setupClientForMockServer() {
	final String baseUrl = mockServer.url("").toString();
	// Remove trailing slash
	final String apiUrl = removeTrailingSlash(baseUrl);

[Copilot]

The trailing slash removal logic is duplicated from OpenAiLlmClientTest. Consider extracting this to a shared utility method to reduce duplication and improve maintainability.

Suggested change

	private void setupClientForMockServer() {
	final String baseUrl = mockServer.url("").toString();
	// Remove trailing slash
	final String apiUrl = baseUrl.endsWith("/") ? baseUrl.substring(0, baseUrl.length() - 1) : baseUrl;
	private String normalizeBaseUrl(final String url) {
	if (url == null || url.isEmpty()) {
	return url;
	}
	return url.endsWith("/") ? url.substring(0, url.length() - 1) : url;
	}
	private void setupClientForMockServer() {
	final String baseUrl = mockServer.url("").toString();
	final String apiUrl = normalizeBaseUrl(baseUrl);