🚑 little hotfix with windows

little hotfix with windows

app/config.py

@@ -2,6 +2,7 @@
 import os
 import typing
 from typing import Literal
+import platform
 
 import boto3
 from dotenv import load_dotenv
@@ -43,3 +44,5 @@ def get_secret():
 KEYWORD_MODEL_S3_PATH = os.getenv("KEYWORD_MODEL_S3_PATH")
 CONTENT_MODEL_S3_PATH = os.getenv("CONTENT_MODEL_S3_PATH")
 STOPWORD_FILE_PATH = "app/static/stopwords.txt"
+OS = platform.system()
+MECAB_DIC_PATH = "C:\mecab/mecab-ko-dic"

app/runnable.py

@@ -23,10 +23,16 @@
     Problem,
 )
 from app.utils.utils import get_stopwords
+from app.config import OS, MECAB_DIC_PATH
 
 log = logging.getLogger("__main__")
 
 
+if OS == "Windows":
+    import win32file
+    win32file._setmaxstdio(2048)
+
+
 class KeywordPredictRunnable(bentoml.Runnable):
     SUPPORTS_CPU_MULTI_THREADING = True
     threshold = 0.5
@@ -37,7 +43,7 @@ def __init__(self, problem_dict: Optional[dict] = None):
         self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
         log.info(f"keyword predict model is running on {self.device}")
         self.model = get_keyword_grading_model().to(self.device)
-        self.tokenizer = Mecab()
+        self.tokenizer = Mecab(MECAB_DIC_PATH) if OS == "Windows" else Mecab()
         self.stopwords = get_stopwords()
 
     def create_problem(self, input_data: KeywordGradingRequest) -> None:

app/utils/utils.py

@@ -5,4 +5,4 @@
 
 
 def get_stopwords() -> Set:
-    return set(Path(STOPWORD_FILE_PATH).read_text().split("\n"))
+    return set(Path(STOPWORD_FILE_PATH).read_text(encoding="UTF-8").split("\n"))