Feature/service separation

apps/pre-processing-service/app/api/endpoints/product.py

@@ -18,7 +18,7 @@ async def search(request: RequestSadaguSearch):
     """
     상품 검색 엔드포인트
     """
-    return search_products(request)
+    return await search_products(request)
 
 @router.post("/match", response_model=ResponseSadaguMatch)
 async def match(request: RequestSadaguMatch):

apps/pre-processing-service/app/core/config.py

@@ -1,9 +1,65 @@
 # pydantic_settings에서 SettingsConfigDict를 추가로 import 합니다.
 from pydantic_settings import BaseSettings, SettingsConfigDict
 import os
+import platform
+import subprocess
 from typing import Optional
 
 
+def detect_mecab_dicdir() -> Optional[str]:
+    """MeCab 사전 경로 자동 감지"""
+
+    # 1. mecab-config 명령어로 사전 경로 확인 (가장 정확한 방법)
+    try:
+        result = subprocess.run(['mecab-config', '--dicdir'],
+                                capture_output=True, text=True, timeout=5)
+        if result.returncode == 0:
+            dicdir = result.stdout.strip()
+            if os.path.exists(dicdir):
+                print(f"mecab-config에서 사전 경로 발견: {dicdir}")
+                return dicdir
+    except (subprocess.CalledProcessError, FileNotFoundError, subprocess.TimeoutExpired):
+        pass
+
+    # 2. 플랫폼별 일반적인 경로들 확인
+    system = platform.system().lower()
+
+    if system == "darwin":  # macOS
+        candidate_paths = [
+            "/opt/homebrew/lib/mecab/dic/mecab-ko-dic",  # Apple Silicon
+            "/usr/local/lib/mecab/dic/mecab-ko-dic",  # Intel Mac
+            "/opt/homebrew/lib/mecab/dic/mecab-ipadic",  # 기본 사전
+            "/usr/local/lib/mecab/dic/mecab-ipadic"
+        ]
+    elif system == "linux":
+        candidate_paths = [
+            "/usr/lib/x86_64-linux-gnu/mecab/dic/mecab-ko-dic",
+            "/usr/lib/mecab/dic/mecab-ko-dic",
+            "/usr/local/lib/mecab/dic/mecab-ko-dic",
+            "/usr/share/mecab/dic/mecab-ko-dic",
+            "/usr/lib/mecab/dic/mecab-ipadic",
+            "/usr/local/lib/mecab/dic/mecab-ipadic"
+        ]
+    elif system == "windows":
+        candidate_paths = [
+            "C:/Program Files/MeCab/dic/mecab-ko-dic",
+            "C:/mecab/dic/mecab-ko-dic",
+            "C:/Program Files/MeCab/dic/mecab-ipadic"
+        ]
+    else:
+        candidate_paths = []
+
+    # 경로 존재 여부 확인
+    for path in candidate_paths:
+        if os.path.exists(path):
+            # dicrc 파일 존재 확인 (실제 사전인지 검증)
+            dicrc_path = os.path.join(path, "dicrc")
+            if os.path.exists(dicrc_path):
+                print(f"플랫폼 기본 경로에서 사전 발견: {path}")
+                return path
+
+    return None
+
 # 공통 설정을 위한 BaseSettings
 class BaseSettingsConfig(BaseSettings):
 
@@ -13,7 +69,19 @@ class BaseSettingsConfig(BaseSettings):
     db_user: str
     db_pass: str
     db_name: str
-    env_name: str = "dev"
+    env_name: str = ".dev"
+
+    # MeCab 사전 경로 (자동 감지)
+    mecab_path: Optional[str] = None
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        # mecab_path가 설정되지 않았으면 자동 감지
+        if not self.mecab_path:
+            self.mecab_path = detect_mecab_dicdir()
+            if not self.mecab_path:
+                print("MeCab 사전 경로를 찾을 수 없어 기본 설정으로 실행합니다.")
 
     @property
     def db_url(self) -> str:
@@ -25,11 +93,11 @@ def db_url(self) -> str:
 
 # 환경별 설정 클래스
 class DevSettings(BaseSettingsConfig):
-    model_config = SettingsConfigDict(env_file=['.env', 'dev.env'])
+    model_config = SettingsConfigDict(env_file=['.env', '.dev.env'])
 
 
 class PrdSettings(BaseSettingsConfig):
-    model_config = SettingsConfigDict(env_file=['.env', 'prd.env'])
+    model_config = SettingsConfigDict(env_file=['.env', '.prd.env'])
 
 def get_settings() -> BaseSettingsConfig:
     """환경 변수에 따라 적절한 설정 객체를 반환하는 함수"""

apps/pre-processing-service/app/model/schemas.py

@@ -1,95 +1,94 @@
 from datetime import datetime
-from typing import Optional, List, Dict
+from typing import Optional, List, Dict, Union
 from pydantic import BaseModel, Field, HttpUrl
 
-#기본 요청
+# 기본 요청
 class RequestBase(BaseModel):
     job_id: int
     schedule_id: int
-    sschdule_his_id: Optional[int] = None
+    schedule_his_id: Optional[int] = None
 
-#기본 응답
+# 기본 응답
 class ResponseBase(BaseModel):
     job_id: int
     schedule_id: int
-    sschdule_his_id : Optional[int] = None
+    schedule_his_id: Optional[int] = None
     status: str
 
-
-#네이버 키워드 추출
+# 네이버 키워드 추출
 class RequestNaverSearch(RequestBase):
     tag: str
     category: Optional[str] = None
-    start_date : Optional[str] = None
-    end_date : Optional[str] = None
+    start_date: Optional[str] = None
+    end_date: Optional[str] = None
 
 class ResponseNaverSearch(ResponseBase):
     category: Optional[str] = None
     keyword: str
-    total_keyword: dict[int, str]
-
-# #키워드 사다구몰 검증
-# class RequestSadaguValidate(RequestBase):
-#     tag: str
-#     category: str
-#
-# class ResponsetSadaguValidate(ResponseBase):
-#     keyword: str
+    total_keyword: Dict[int, str]
 
 # 2단계: 검색
 class RequestSadaguSearch(RequestBase):
     keyword: str
 
 class ResponseSadaguSearch(ResponseBase):
     keyword: str
-    search_results: list[dict]
+    search_results: List[Dict]
 
 # 3단계: 매칭
 class RequestSadaguMatch(RequestBase):
     keyword: str
-    search_results: list[dict]
+    search_results: List[Dict]
 
 class ResponseSadaguMatch(ResponseBase):
     keyword: str
-    matched_products: list[dict]
+    matched_products: List[Dict]
 
 # 4단계: 유사도
 class RequestSadaguSimilarity(RequestBase):
     keyword: str
-    matched_products: list[dict]
+    matched_products: List[Dict]
+    search_results: Optional[List[Dict]] = None  # 3단계에서 매칭 실패시 폴백용
 
 class ResponseSadaguSimilarity(ResponseBase):
     keyword: str
-    selected_product: dict | None = None
-    reason: str | None = None
-
-#사다구몰 크롤링
-class RequestSadaguCrawl(RequestBase):
+    selected_product: Optional[Dict] = None
+    reason: Optional[str] = None
+
+# 사다구몰 크롤링
+class RequestSadaguCrawl(BaseModel):
+    job_id: int = Field(..., description="작업 ID")
+    schedule_id: int = Field(..., description="스케줄 ID")
+    schedule_his_id: int = Field(..., description="스케줄 히스토리 ID")
     tag: str = Field(..., description="크롤링 태그 (예: 'detail')")
     product_url: HttpUrl = Field(..., description="크롤링할 상품의 URL")
     use_selenium: bool = Field(default=True, description="Selenium 사용 여부")
     include_images: bool = Field(default=False, description="이미지 정보 포함 여부")
 
-class ResponseSadaguCrawl(ResponseBase):
+class ResponseSadaguCrawl(BaseModel):
+    job_id: int
+    schedule_id: int
+    schedule_his_id: int
     tag: str
     product_url: str
     use_selenium: bool
     include_images: bool
-    product_detail: Optional[dict] = None
+    product_detail: Optional[Dict] = None
+    status: str
     crawled_at: Optional[str] = None
 
-#블로그 생성
+# 블로그 생성
 class RequestBlogCreate(RequestBase):
     tag: str
     category: str
 
 class ResponseBlogCreate(ResponseBase):
     pass
 
-#블로그 배포
+# 블로그 배포
 class RequestBlogPublish(RequestBase):
     tag: str
     category: str
 
 class ResponseBlogPublish(ResponseBase):
-    pass
+    pass