init datamate

This commit is contained in:
Dallas98
2025-10-21 23:00:48 +08:00
commit 1c97afed7d
692 changed files with 135442 additions and 0 deletions

View File

@@ -0,0 +1,27 @@
"""
Description: 文档表情去除
Create: 2023/12/7 15:43
"""
import time
from typing import Dict, Any
import emoji
from loguru import logger
from datamate.core.base_op import Mapper
class EmojiCleaner(Mapper):
@staticmethod
def _emoji_filter(input_data: str):
res = []
for input_s in input_data.split('\n'):
res.append(emoji.replace_emoji(input_s, replace=''))
return '\n'.join(res)
def execute(self, sample: Dict[str, Any]) -> Dict[str, Any]:
start = time.time()
sample[self.text_key] = self._emoji_filter(sample[self.text_key])
logger.info(f"fileName: {sample[self.filename_key]}, method: EmojiCleaner costs {time.time() - start:6f} s")
return sample