73 lines
2.1 KiB
Python
73 lines
2.1 KiB
Python
|
|
from typing import Iterator, Dict, Any, Optional, Tuple
|
|
import re
|
|
import datetime
|
|
|
|
|
|
def parse(filename: str) -> Iterator[Dict[str, Any]]:
|
|
def parse_line(line_str: str) -> Iterator[str]:
|
|
w = None
|
|
s = False
|
|
for ch in line_str:
|
|
if w is None:
|
|
if ch == ';':
|
|
yield ''
|
|
continue
|
|
elif ch in (' ', '\t'):
|
|
continue
|
|
w = ch
|
|
s = ch == '"'
|
|
continue
|
|
elif not s and ch in (';', '\n'):
|
|
yield w.strip()
|
|
w = None
|
|
continue
|
|
elif s and ch == '"':
|
|
s = False
|
|
w += ch
|
|
if w is not None:
|
|
yield w.strip()
|
|
|
|
with open(filename, 'r') as f:
|
|
header: Optional[Tuple[str]] = None
|
|
for line in f:
|
|
if header is None:
|
|
header = tuple([e.strip() for e in line.strip().split(';')])
|
|
continue
|
|
|
|
obj = {}
|
|
for i, part in enumerate(parse_line(line)):
|
|
if part == '':
|
|
part = None
|
|
elif part[0] == '"' and part[-1] == '"':
|
|
part = part[1:-1]
|
|
elif part == 'T':
|
|
part = True
|
|
elif part == 'F':
|
|
part = False
|
|
elif part.isdigit():
|
|
part = int(part)
|
|
elif re.match(r'[0-9]+\.[0-9]+', part):
|
|
part = float(part)
|
|
elif len(part) == 10 and part[4] == '-' and part[7] == '-':
|
|
part = datetime.datetime.strptime(part, '%Y-%m-%d').date()
|
|
else:
|
|
raise RuntimeError(part)
|
|
obj[header[i]] = part
|
|
yield obj
|
|
|
|
|
|
def format_row(*values) -> str:
|
|
row = ''
|
|
for val in values:
|
|
if val is None:
|
|
pass
|
|
elif type(val) == str:
|
|
row += f'"{val}"'
|
|
elif type(val) == bool:
|
|
row += 'T' if val else 'F'
|
|
else:
|
|
row += str(val)
|
|
row += ';'
|
|
return f'{row[:-1]}\n'
|