Upload processors.py with huggingface_hub
Browse files- processors.py +46 -0
processors.py
CHANGED
|
@@ -75,6 +75,16 @@ class TakeFirstNonEmptyLine(BaseFieldOperator):
|
|
| 75 |
return splitted[0].strip()
|
| 76 |
|
| 77 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
class LowerCaseTillPunc(BaseFieldOperator):
|
| 79 |
def process(self, instance):
|
| 80 |
non_empty_line = instance.lower()
|
|
@@ -85,6 +95,11 @@ class LowerCaseTillPunc(BaseFieldOperator):
|
|
| 85 |
return non_empty_line
|
| 86 |
|
| 87 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
class FirstCharacter(BaseFieldOperator):
|
| 89 |
def process(self, instance):
|
| 90 |
match = re.search(r"\s*(\w)", instance)
|
|
@@ -93,6 +108,37 @@ class FirstCharacter(BaseFieldOperator):
|
|
| 93 |
return ""
|
| 94 |
|
| 95 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
class StringOrNotString(BaseFieldOperator):
|
| 97 |
string: str
|
| 98 |
|
|
|
|
| 75 |
return splitted[0].strip()
|
| 76 |
|
| 77 |
|
| 78 |
+
class ConvertToBoolean(BaseFieldOperator):
|
| 79 |
+
def process(self, instance):
|
| 80 |
+
clean_instance = str(instance).strip().lower()
|
| 81 |
+
if any(w in clean_instance for w in ["no", "not", "wrong", "false"]):
|
| 82 |
+
return "FALSE"
|
| 83 |
+
if any(w in clean_instance for w in ["yes", "right", "correct", "true"]):
|
| 84 |
+
return "TRUE"
|
| 85 |
+
return "OTHER"
|
| 86 |
+
|
| 87 |
+
|
| 88 |
class LowerCaseTillPunc(BaseFieldOperator):
|
| 89 |
def process(self, instance):
|
| 90 |
non_empty_line = instance.lower()
|
|
|
|
| 95 |
return non_empty_line
|
| 96 |
|
| 97 |
|
| 98 |
+
class LowerCase(BaseFieldOperator):
|
| 99 |
+
def process(self, instance):
|
| 100 |
+
return instance.lower()
|
| 101 |
+
|
| 102 |
+
|
| 103 |
class FirstCharacter(BaseFieldOperator):
|
| 104 |
def process(self, instance):
|
| 105 |
match = re.search(r"\s*(\w)", instance)
|
|
|
|
| 108 |
return ""
|
| 109 |
|
| 110 |
|
| 111 |
+
class TakeFirstWord(BaseFieldOperator):
|
| 112 |
+
def process(self, instance):
|
| 113 |
+
match = re.search(r"[\w]+", instance)
|
| 114 |
+
if match:
|
| 115 |
+
return instance[match.start() : match.end()]
|
| 116 |
+
return ""
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
class YesNoToInt(BaseFieldOperator):
|
| 120 |
+
def process(self, instance):
|
| 121 |
+
if instance == "yes":
|
| 122 |
+
return "1"
|
| 123 |
+
return "0"
|
| 124 |
+
|
| 125 |
+
|
| 126 |
+
class ToYesOrNone(BaseFieldOperator):
|
| 127 |
+
def process(self, instance):
|
| 128 |
+
if instance == "yes":
|
| 129 |
+
return "yes"
|
| 130 |
+
return "none"
|
| 131 |
+
|
| 132 |
+
|
| 133 |
+
class StanceToProCon(BaseFieldOperator):
|
| 134 |
+
def process(self, instance):
|
| 135 |
+
if instance == "positive":
|
| 136 |
+
return "PRO"
|
| 137 |
+
if instance in ["negative", "suggestion"]:
|
| 138 |
+
return "CON"
|
| 139 |
+
return "none"
|
| 140 |
+
|
| 141 |
+
|
| 142 |
class StringOrNotString(BaseFieldOperator):
|
| 143 |
string: str
|
| 144 |
|