Feat/migrate to OpenAI

[SanghunYun95]

Summary by CodeRabbit

릴리스 노트

• 개선 사항

  • AI 모델 제공자를 OpenAI로 전환하였습니다.
  • 채팅 기능의 타임아웃 처리 및 에러 로깅을 개선하였습니다.

• 기술 업데이트

  • 의존성을 최신 OpenAI 라이브러리로 업데이트하였습니다.
  • API 키 관리 설정을 변경하였습니다.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
philo-rag	Ready	Preview, Comment	Mar 25, 2026 0:46am

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 3e0eb802-d7c5-4e96-9271-ceb05d8781fe

📥 Commits

Reviewing files that changed from the base of the PR and between d406f0b and e60cc1a.

📒 Files selected for processing (1)

• backend/app/api/routes/chat.py

🚧 Files skipped from review as they are similar to previous changes (1)

• backend/app/api/routes/chat.py

---

📝 Walkthrough

Walkthrough

Google Gemini API에서 OpenAI API로의 백엔드 마이그레이션을 수행합니다. 구성 설정, 환경 변수, 서비스 로직, 의존성 및 스크립트에서 관련 참조를 업데이트하여 Gemini 클라이언트를 OpenAI 클라이언트로 교체합니다.

Changes

Cohort / File(s)	Summary
API 키 관리 및 설정 backend/app/core/config.py, backend/app/core/env_utils.py	GEMINI_API_KEY를 OPENAI_API_KEY로 변경; Settings 모델에 extra="ignore" 추가; parse_gemini_api_keys() 함수를 parse_openai_api_keys()로 이름 변경 및 환경 변수 참조 업데이트
LLM 서비스 통합 backend/app/services/llm.py	ChatGoogleGenerativeAI → ChatOpenAI로 교체; gemini-2.5-flash-lite → gpt-4o-mini 모델로 변경; get_all_gemini_keys() → get_all_openai_keys()로 함수 이름 변경; 스트리밍 에러 핸들링 강화 및 타임아웃 로깅 개선
채팅 API 및 타임아웃 처리 backend/app/api/routes/chat.py	데이터베이스 검색 흐름에서 asyncio.wait_for 타임아웃 래퍼 제거; 채팅 제목 생성의 타임아웃 로깅 재계산 (t_start 기반 경과 시간 사용)
의존성 업데이트 backend/requirements.txt	langchain-google-genai, google-generativeai 제거; langchain-openai>=0.1.0, openai>=1.0.0 추가
유틸리티 스크립트 backend/scripts/check_models.py, backend/scripts/generate_book_mapping.py	genai 클라이언트를 OpenAI 클라이언트로 교체; API 키 환경 변수 및 함수 참조 업데이트; 에러 핸들링 로직 조정 (리소스 소진 특화 처리 제거)
테스트 설정 backend/tests/unit/test_llm.py	테스트 픽스처에서 GEMINI_API_KEY → OPENAI_API_KEY 참조 변경

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• philo-rag#10: LLM 키 관리 및 설정 코드 경로를 동일하게 수정 (환경 변수 파싱, get_all_* 및 get_llm 함수 변경)
• philo-rag#20: backend/app/api/routes/chat.py의 데이터베이스 검색 타임아웃/세마포 로직을 수정 (이전 PR에서 추가된 세마포와 호출별 타임아웃을 현재 PR에서 제거)
• philo-rag#17: backend/app/api/routes/chat.py에서 타임아웃 처리 및 로컬 임포트 동작 수정 (채팅 제목 생성 및 호출별 타임아웃 관련)

Poem

🐰 OpenAI로 갈아탄 래빗, 반갑네요!
Gemini에서 벗어나 새로운 길을 가요,
타임아웃도 깔끔하게, 에러도 명확하게,
API 키 회전하며 안정적으로 🚀
마이그레이션 완료, 우리 함께 나아가요!

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 61.54% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	Pull request title accurately summarizes the main change: migrating the LLM backend from Gemini to OpenAI across multiple files and dependencies.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings (stacked PR)
• 📝 Generate docstrings (commit on current branch)

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch feat/migrate-to-openai

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@backend/app/api/routes/chat.py`:
- Around line 87-91: The current use of
asyncio.wait_for(asyncio.to_thread(_search_documents, query_vector),
timeout=CHAT_TIMEOUT) can leave the worker thread running after a timeout;
instead, remove the wait_for wrapper and call await
asyncio.to_thread(_search_documents, query_vector) directly while keeping the
semaphore (_db_rpc_semaphore) to limit concurrency, and rely on the DB client’s
internal timeout to abort long DB ops; update error handling around the await to
handle and log DB client timeouts/errors (referencing _search_documents,
_db_rpc_semaphore and CHAT_TIMEOUT) and remove any logic that assumes wait_for
will cancel the background thread.

In `@backend/requirements.txt`:
- Line 6: The langchain-openai version constraint currently prevents installing
the latest 1.x releases; update the requirement for the package named
"langchain-openai" by removing the upper bound or changing the spec to allow 1.x
(e.g., replace "langchain-openai>=0.1.0,<1.0.0" with "langchain-openai>=0.1.0"
or a range that includes 1.x) so the latest 1.1.10 (2026-02-17) can be
installed.

In `@backend/scripts/check_models.py`:
- Line 4: The code imports OpenAI directly with "from openai import OpenAI" but
the openai package is not listed explicitly in requirements.txt; add an explicit
dependency entry for "openai" (pinned to a compatible version with your
langchain-openai dependency) to requirements.txt so direct imports are
satisfied, then run your dependency/install workflow (pip install / lockfile
update) to ensure CI and local installs include the package.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 8bec5acd-744c-4c2a-a02e-5abb5dcadebc

📥 Commits

Reviewing files that changed from the base of the PR and between c220735 and ff949ea.

📒 Files selected for processing (9)

• backend/app/api/routes/chat.py
• backend/app/core/config.py
• backend/app/core/env_utils.py
• backend/app/services/database.py
• backend/app/services/llm.py
• backend/requirements.txt
• backend/scripts/check_models.py
• backend/scripts/generate_book_mapping.py
• backend/tests/unit/test_llm.py

[coderabbitai]

🧹 Nitpick comments (2)

backend/app/api/routes/chat.py (2)

92-93: logging.exception 호출에서 중복된 예외 객체를 제거하세요.

logger.exception은 자동으로 예외 정보와 트레이스백을 포함하므로 {e}를 메시지에 포함할 필요가 없습니다.

♻️ 제안된 수정

     except Exception as e:
-        logger.exception(f"Database search failed: {e}")
+        logger.exception("Database search failed")

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@backend/app/api/routes/chat.py` around lines 92 - 93, In the except block
that currently reads "except Exception as e: logger.exception(f\"Database search
failed: {e}\")" (in backend/app/api/routes/chat.py), remove the redundant "{e}"
from the log message and call logger.exception with a plain message like
"Database search failed" so the exception and traceback are captured only once
by logger.exception; do not pass exc_info or the exception object separately.

---

207-209: 타임아웃 로그 형식의 일관성을 고려하세요.

다른 타임아웃 로그는 실제 경과 시간(time.perf_counter() - t0)을 기록하지만, 여기서는 상수 값만 출력합니다. 일관성을 위해 동일한 패턴을 적용하는 것이 좋습니다.

♻️ 제안된 수정

+    t_start = time.perf_counter()
     try:
         title = await asyncio.wait_for(generate_chat_title_async(query), timeout=CHAT_TIMEOUT)
         # ...
     except asyncio.TimeoutError:
-        logger.warning(f"Timeout generating chat title after {CHAT_TIMEOUT}s")
+        logger.warning(f"Timeout generating chat title after {time.perf_counter() - t_start:.2f}s")
         return {"title": DEFAULT_CHAT_TITLE}

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@backend/app/api/routes/chat.py` around lines 207 - 209, The timeout handler
currently logs the constant CHAT_TIMEOUT; change the logger.warning in the
except asyncio.TimeoutError block to report the actual elapsed time (e.g., use
time.perf_counter() - t0) for consistency with other timeout logs; update the
reference in that block (where DEFAULT_CHAT_TITLE is returned) to compute
elapsed from the existing t0 variable and include that elapsed value in the
warning message instead of CHAT_TIMEOUT.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@backend/app/api/routes/chat.py`:
- Around line 92-93: In the except block that currently reads "except Exception
as e: logger.exception(f\"Database search failed: {e}\")" (in
backend/app/api/routes/chat.py), remove the redundant "{e}" from the log message
and call logger.exception with a plain message like "Database search failed" so
the exception and traceback are captured only once by logger.exception; do not
pass exc_info or the exception object separately.
- Around line 207-209: The timeout handler currently logs the constant
CHAT_TIMEOUT; change the logger.warning in the except asyncio.TimeoutError block
to report the actual elapsed time (e.g., use time.perf_counter() - t0) for
consistency with other timeout logs; update the reference in that block (where
DEFAULT_CHAT_TITLE is returned) to compute elapsed from the existing t0 variable
and include that elapsed value in the warning message instead of CHAT_TIMEOUT.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 00a268b6-1540-4691-92e9-6d15e65d632d

📥 Commits

Reviewing files that changed from the base of the PR and between ff949ea and d406f0b.

📒 Files selected for processing (2)

• backend/app/api/routes/chat.py
• backend/requirements.txt