[coati] fix inference profanity check (#3299)

applications/Chat/inference/requirements.txt

@@ -10,3 +10,4 @@ uvicorn
 git+https://github.com/huggingface/transformers
 accelerate
 bitsandbytes
+jieba

applications/Chat/inference/utils.py

@@ -2,6 +2,7 @@ import re
 from threading import Lock
 from typing import Any, Callable, Generator, List, Optional
 import json
+import jieba
 
 import torch
 import torch.distributed as dist
@@ -130,10 +131,7 @@ class ChatPromptProcessor:
         self.tokenizer = tokenizer
         self.context = context
         self.max_len = max_len
-        if len(censored_words) > 0:
-            self.censored_pat = re.compile(f'({"|".join(map(re.escape, censored_words))})', flags=re.I)
-        else:
-            self.censored_pat = None
+        self.censored_words = set([word.lower() for word in censored_words])
         # These will be initialized after the first call of preprocess_prompt()
         self.context_len: Optional[int] = None
         self.dialogue_placeholder_len: Optional[int] = None
@@ -179,9 +177,10 @@ class ChatPromptProcessor:
         return output.strip()
 
     def has_censored_words(self, text: str) -> bool:
-        if self.censored_pat is None:
+        if len(self.censored_words) == 0:
             return False
-        return self.censored_pat.search(text) is not None
+        intersection = set(jieba.cut(text.lower())) & self.censored_words
+        return len(intersection) > 0
 
 class LockedIterator: