Spaces:
Runtime error
Runtime error
Leonardo
commited on
Create text_cleaner_tool.py
Browse files- scripts/text_cleaner_tool.py +167 -0
scripts/text_cleaner_tool.py
ADDED
|
@@ -0,0 +1,167 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from smolagents import Tool
|
| 2 |
+
from typing import Dict, Optional, Any
|
| 3 |
+
import logging
|
| 4 |
+
|
| 5 |
+
logger = logging.getLogger(__name__)
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
class TextCleanerTool(Tool):
|
| 9 |
+
name = "clean_text"
|
| 10 |
+
description = (
|
| 11 |
+
"Cleans and normalizes text by removing or replacing unwanted elements"
|
| 12 |
+
)
|
| 13 |
+
inputs = {
|
| 14 |
+
"text": {"type": "string", "description": "The input text to clean"},
|
| 15 |
+
"fix_unicode": {
|
| 16 |
+
"type": "boolean",
|
| 17 |
+
"description": "Fix broken unicode characters and mojibake",
|
| 18 |
+
"default": True,
|
| 19 |
+
},
|
| 20 |
+
"to_ascii": {
|
| 21 |
+
"type": "boolean",
|
| 22 |
+
"description": "Convert non-ASCII characters to their closest ASCII equivalents",
|
| 23 |
+
"default": True,
|
| 24 |
+
},
|
| 25 |
+
"lower": {
|
| 26 |
+
"type": "boolean",
|
| 27 |
+
"description": "Convert text to lowercase",
|
| 28 |
+
"default": True,
|
| 29 |
+
},
|
| 30 |
+
"no_line_breaks": {
|
| 31 |
+
"type": "boolean",
|
| 32 |
+
"description": "Replace line breaks with spaces",
|
| 33 |
+
"default": False,
|
| 34 |
+
},
|
| 35 |
+
"no_urls": {
|
| 36 |
+
"type": "boolean",
|
| 37 |
+
"description": "Replace URLs with a token",
|
| 38 |
+
"default": False,
|
| 39 |
+
},
|
| 40 |
+
"no_emails": {
|
| 41 |
+
"type": "boolean",
|
| 42 |
+
"description": "Replace email addresses with a token",
|
| 43 |
+
"default": False,
|
| 44 |
+
},
|
| 45 |
+
"no_phone_numbers": {
|
| 46 |
+
"type": "boolean",
|
| 47 |
+
"description": "Replace phone numbers with a token",
|
| 48 |
+
"default": False,
|
| 49 |
+
},
|
| 50 |
+
"no_numbers": {
|
| 51 |
+
"type": "boolean",
|
| 52 |
+
"description": "Replace all numbers with a token",
|
| 53 |
+
"default": False,
|
| 54 |
+
},
|
| 55 |
+
"no_digits": {
|
| 56 |
+
"type": "boolean",
|
| 57 |
+
"description": "Replace all digits with 0",
|
| 58 |
+
"default": False,
|
| 59 |
+
},
|
| 60 |
+
"no_currency_symbols": {
|
| 61 |
+
"type": "boolean",
|
| 62 |
+
"description": "Replace currency symbols with a token",
|
| 63 |
+
"default": False,
|
| 64 |
+
},
|
| 65 |
+
"no_punct": {
|
| 66 |
+
"type": "boolean",
|
| 67 |
+
"description": "Remove all punctuation",
|
| 68 |
+
"default": False,
|
| 69 |
+
},
|
| 70 |
+
"no_emoji": {
|
| 71 |
+
"type": "boolean",
|
| 72 |
+
"description": "Remove all emoji characters",
|
| 73 |
+
"default": False,
|
| 74 |
+
},
|
| 75 |
+
"lang": {
|
| 76 |
+
"type": "string",
|
| 77 |
+
"description": "Language code for special handling ('en' or 'de' supported)",
|
| 78 |
+
"default": "en",
|
| 79 |
+
},
|
| 80 |
+
"custom_replacements": {
|
| 81 |
+
"type": "object",
|
| 82 |
+
"description": "Dictionary of custom string replacements to apply",
|
| 83 |
+
"optional": True,
|
| 84 |
+
},
|
| 85 |
+
}
|
| 86 |
+
output_type = "string"
|
| 87 |
+
|
| 88 |
+
def forward(
|
| 89 |
+
self,
|
| 90 |
+
text: str,
|
| 91 |
+
fix_unicode: bool = True,
|
| 92 |
+
to_ascii: bool = True,
|
| 93 |
+
lower: bool = True,
|
| 94 |
+
no_line_breaks: bool = False,
|
| 95 |
+
no_urls: bool = False,
|
| 96 |
+
no_emails: bool = False,
|
| 97 |
+
no_phone_numbers: bool = False,
|
| 98 |
+
no_numbers: bool = False,
|
| 99 |
+
no_digits: bool = False,
|
| 100 |
+
no_currency_symbols: bool = False,
|
| 101 |
+
no_punct: bool = False,
|
| 102 |
+
no_emoji: bool = False,
|
| 103 |
+
lang: str = "en",
|
| 104 |
+
custom_replacements: Optional[Dict[str, str]] = None,
|
| 105 |
+
) -> str:
|
| 106 |
+
"""Clean and normalize text by removing or replacing unwanted elements."""
|
| 107 |
+
# Input validation
|
| 108 |
+
if not text:
|
| 109 |
+
return ""
|
| 110 |
+
|
| 111 |
+
if not isinstance(text, str):
|
| 112 |
+
try:
|
| 113 |
+
text = str(text)
|
| 114 |
+
except Exception as e:
|
| 115 |
+
logger.error(f"Failed to convert input to string: {e}")
|
| 116 |
+
return f"Error: Could not process input of type {type(text)}"
|
| 117 |
+
|
| 118 |
+
# Import cleantext safely
|
| 119 |
+
try:
|
| 120 |
+
from cleantext import clean
|
| 121 |
+
except ImportError:
|
| 122 |
+
logger.error(
|
| 123 |
+
"cleantext package not installed. Install with: pip install clean-text"
|
| 124 |
+
)
|
| 125 |
+
return "Error: Required dependency 'clean-text' is not installed."
|
| 126 |
+
|
| 127 |
+
# Special parameter handling
|
| 128 |
+
replace_params = {
|
| 129 |
+
"replace_with_url": "<URL>",
|
| 130 |
+
"replace_with_email": "<EMAIL>",
|
| 131 |
+
"replace_with_phone_number": "<PHONE>",
|
| 132 |
+
"replace_with_number": "<NUMBER>",
|
| 133 |
+
"replace_with_digit": "0",
|
| 134 |
+
"replace_with_currency_symbol": "<CUR>",
|
| 135 |
+
"replace_with_punct": "",
|
| 136 |
+
}
|
| 137 |
+
|
| 138 |
+
try:
|
| 139 |
+
# Apply cleantext with parameters
|
| 140 |
+
cleaned_text = clean(
|
| 141 |
+
text,
|
| 142 |
+
fix_unicode=fix_unicode,
|
| 143 |
+
to_ascii=to_ascii,
|
| 144 |
+
lower=lower,
|
| 145 |
+
no_line_breaks=no_line_breaks,
|
| 146 |
+
no_urls=no_urls,
|
| 147 |
+
no_emails=no_emails,
|
| 148 |
+
no_phone_numbers=no_phone_numbers,
|
| 149 |
+
no_numbers=no_numbers,
|
| 150 |
+
no_digits=no_digits,
|
| 151 |
+
no_currency_symbols=no_currency_symbols,
|
| 152 |
+
no_punct=no_punct,
|
| 153 |
+
no_emoji=no_emoji,
|
| 154 |
+
lang=lang,
|
| 155 |
+
**replace_params,
|
| 156 |
+
)
|
| 157 |
+
|
| 158 |
+
# Apply any custom replacements
|
| 159 |
+
if custom_replacements:
|
| 160 |
+
for old, new in custom_replacements.items():
|
| 161 |
+
cleaned_text = cleaned_text.replace(old, new)
|
| 162 |
+
|
| 163 |
+
return cleaned_text
|
| 164 |
+
|
| 165 |
+
except Exception as e:
|
| 166 |
+
logger.error(f"Error cleaning text: {e}")
|
| 167 |
+
return f"Error during text cleaning: {str(e)}"
|