"""Defines NumericParser for parsing a number string into a float values"""
import re
from typing import Union
from math import copysign
[docs]
class NumericParser:
"""
Parser for any kind of text input into a corresponding format.
Numeric data is assumed to be valid and clean.
Use NumericColumnFormatter for validating your data.
"""
numeric = (r"(^0+$|^[\\[\\(+-]*(([1-9][0-9]{0,2}([.,][0-9]{3})*)|[0-9]"
r"|[1-9][0-9]+)([,.][0-9]*)?[\]\)]*$)")
delimiters = r"[.,]"
REGEX_NO_ALPHANUM_CHARS = re.compile(r'[^a-zA-Z0-9)\[\](-.,]')
brackets = r"[\[\]\(\)]"
# match a pattern with whitespace followed by either a single comma
# from the left or whitespace surrounded by numbers
single_white_space_regex = "(?<=\\d)(,{0,1}) {1,2}(?=\\d)"
def __init__(self, raw: str):
"""Takes string as an input and create ValueParser object
:param raw: input raw string representation of the number
"""
raw = re.sub(self.single_white_space_regex, '', raw)
raw = raw.replace('$', '').strip()
self.raw = raw
self.format = None
self.parsed = None
self.value_type = None
self._separators = None
self._separated_digits = None
self.sign = None
self.removed_sign = False
def __get_separated_digits_and_separators(self):
"""Assigns the list of separators and separated digits to self.
Ex. self.raw=12,345.6
[',','.']->self.separators
['12', '345', '6']->self._separated_digits
Applicable only if type is numeric
"""
# TODO: check if strip_value is used properly
val = self.REGEX_NO_ALPHANUM_CHARS.sub('', self.raw)
val = re.sub(self.brackets, '', val)
val = self.strip_value(val)
# strip from multiple zeros. ex: 000 -> 0
val = val[:-1].lstrip('0') + val[-1]
self._separators = re.findall(self.delimiters, val)
self._separated_digits = re.split(self.delimiters, val)
[docs]
def infer_sign(self):
"""
Infer the sign of the number either with brackets
or a trailing minus sign"""
self.sign = 1
if self.raw.startswith('(') and self.raw.endswith(')'):
self.sign = -1
self.remove_sign_parentheses()
elif self.raw.lstrip().startswith('-'):
self.sign = -1
self.remove_dash_sign()
def remove_dash_sign(self):
self.raw = self.raw.lstrip().lstrip('-')
self.removed_sign = True
def remove_sign_parentheses(self):
self.raw = self.raw.lstrip('(')
self.raw = self.raw.rstrip(')')
self.removed_sign = True
[docs]
def parse_regular_float(self):
"""Parse Like a regular float number"""
# noinspection PyBroadException
try:
# try parsing like a regular float number
# if it doesn't work use more complicated logic
delimiter = any(re.findall(self.delimiters, self.raw))
decimal_part = len(re.split(self.delimiters, self.raw)[-1]) if\
delimiter else 0
if decimal_part < 3:
self.parsed = float(self.raw.strip()) * self.sign
elif decimal_part > 2:
return None
return self.parsed
except Exception:
return None
[docs]
def parse_regular_float_with_semicolon(self):
"""Parse Like a regular float number with semicolon"""
# noinspection PyBroadException
try:
if self.raw.strip().startswith('0'):
# try parsing like a regular float number after replacing
# commas if it doesn't work use more complicated logic
if self.removed_sign:
self.parsed = float(self.raw.strip().replace(',', '.')) \
* self.sign
else:
self.parsed = float(self.raw.strip().replace(',', '.'))
return self.parsed
except Exception:
return None
[docs]
def parse_numeric(self) -> Union[int, float]:
"""
Parse the raw data, and return parsed numeric value
and get the numeric, separator and value type of the raw data
:return: Parsed float or int value
"""
self.raw = self.raw.replace('–', '-')
self.infer_sign()
parse_float = self.parse_regular_float()
if parse_float is not None:
return parse_float
else:
parse_float_semicolon = self.parse_regular_float_with_semicolon()
if parse_float_semicolon is not None:
return parse_float_semicolon
if '%' in self.raw:
return float('nan')
try:
self.__get_separated_digits_and_separators()
self.parsed = self.parse_raw_numeric()
self.parsed = self.parsed * self.sign
except Exception:
return float('nan')
return self.parsed
[docs]
def parse_raw_numeric(self) -> Union[int, float]:
"""Used for parsing raw text without getting any context
from the column
"""
if len(self._separators) == 0:
return int(self._separated_digits[0])
elif len(self._separated_digits[-1]) < 3:
int_part = int(''.join(self._separated_digits[:-1]))
dec_part = float('0.' + self._separated_digits[-1])
dec_part = copysign(dec_part, int_part)
return int_part + dec_part
elif len(self._separators) > 1 and self._separators[0] != \
self._separators[-1] and len(self._separated_digits[-1]) < 3:
int_part = int(''.join(self._separated_digits[:-1]))
dec_part = float('0.' + self._separated_digits[-1])
dec_part = copysign(dec_part, int_part)
return int_part + dec_part
else:
return int(''.join(self._separated_digits))
[docs]
@staticmethod
def strip_value(text: str):
"""Strip non-numeric characters from a string representing a number
:param str text: Input string representing a number
:return: String with stripped non-numeric characters
:rtype str:
"""
currencies = u'$¢£¤¥֏؋৲৳৻૱௹฿៛₠₡₢₣₤₥₦₧₨₩₪₫€₭₮₯₰₱₲₳₴₵₶₷₸₹₺₻₼₽₾꠸﷼﹩＄￠￡￥￦'
text = text.strip(
'\n\r\t ' + currencies + "'+*/%^_:;'\"\\|").rstrip('—-.,')
text = text.replace(' ', '')
return text
[docs]
def is_numeric(self) -> bool:
"""Determines if the text is a representation of a numeric value
:return: True if the raw text corresponds to numeric representation
"""
if re.match(self.numeric,
self.strip_value(self.raw)) and '%' not in self.raw:
return True
return False