import logging
from typing import List, Optional, Dict, Type
from pycognaize.common.enums import (
IqDocumentKeysEnum,
ID,
IqFieldKeyEnum,
IqDataTypesEnum
)
from pycognaize.document.field import Field
from pycognaize.document.html_info import HTML
from pycognaize.document.tag import ExtractionTag
from pycognaize.document.page import Page
[docs]
class DateField(Field):
"""Base class for all pycognaize date fields"""
tag_class: Type[ExtractionTag] = ExtractionTag
def __init__(self,
name: str,
value: str = '',
tags: Optional[List[ExtractionTag]] = None,
field_id: Optional[str] = None,
group_key: str = None,
confidence: Optional[float] = -1.0,
group_name: str = None,
mapping: Optional[List[Dict[str, str]]] = None
):
super().__init__(name=name, value=value, tags=tags,
group_key=group_key, confidence=confidence,
group_name=group_name, mapping=mapping)
self._field_id = field_id
self._value = '; '.join([i.raw_value
for i in self.tags]) if self.tags else value
@property
def field_id(self):
return self._field_id
@property
def value(self):
return self._value
[docs]
@classmethod
def construct_from_raw(
cls, raw: dict, pages: Dict[int, Page],
html: Optional[HTML] = None,
labels=None) -> 'DateField':
"""Create DateField object from dictionary"""
tag_dicts: List[dict] = raw[IqDocumentKeysEnum.tags.value]
tags = []
for i in tag_dicts:
try:
tags.append(cls.tag_class.construct_from_raw(
raw=i, page=pages[i['page']]))
except Exception as e:
logging.debug(f"Failed creating tag for field {raw[ID]}: {e}")
return cls(name=raw[IqDocumentKeysEnum.name.value],
value=raw[IqFieldKeyEnum.value.value],
tags=tags,
field_id=str(raw[ID]),
group_key=raw.get(IqFieldKeyEnum.group_key.value, ''),
group_name=raw.get(IqFieldKeyEnum.group.value, ''),
mapping=raw.get(IqFieldKeyEnum.mapping.value, []),
)
[docs]
def to_dict(self) -> dict:
"""Converts DateField object to dictionary"""
field_dict = super().to_dict()
field_dict[IqFieldKeyEnum.value.value] = self.value
field_dict[ID] = self._field_id
field_dict[IqFieldKeyEnum.data_type.value] = IqDataTypesEnum.date.value
return field_dict
def __repr__(self):
return (f"<{self.__class__.__name__}: {self.name}:"
f" {'|'.join([i.raw_value for i in self.tags])}>")
def __str__(self):
return f"{'|'.join([i.raw_value for i in self.tags])}"