the-final-cl-test-2/wonderful _cleanup.py

import os
import re

crime_separator = '*******\nCrime'
not_sus_regex = r"[Nn][o\']t[ a-z]*sus"
not_sus = re.compile(not_sus_regex, re.IGNORECASE)
people = dict()
companies = set()
wonderful_blacklist = {'alice', 'hatter', 'rabbit', 'march mare'}
# in_to_cm('6\'0"') == 182.88


class Person:
    def __init__(self, name):
        n = name.strip().split()
        if len(n) > 2:
            print(name)
        self.first = n[0]
        self.last = n[1]
        self.memberships = set()
        self.gender = None
        self.age = None
        self.not_sus = None
        self.addr = None
        self.interview = None
        self.height = None
        self.weight = None
        self.plate = None
        self.make = None
        self.color = None

    def add_member(self, comp):
        self.memberships.add(comp)

    def __repr__(self):
        out = list()
        for i in sorted(filter(lambda x: x[0] != '_', dir(self))):
            if not callable(eval('self.' + i)):
                out.append(': '.join([i, str(eval('self.' + i))]))
        return '\n'.join(out)


def in_to_cm(inch):
    return round(int(inch.split('\'')[0]) * 30.48 + int(inch.split('\'')[1][:-1]) * 2.54, 2)


def lb_to_kg(lb):
    return round(int(lb.split()[0]) * 0.4536, 2)


def check_wonders(x):
    x = x.lower()
    for alice in wonderful_blacklist:
        if alice in x:
            return False
    return True


def format_output(field):
    if not field:
        return ''
    elif type(field) in {list, tuple}:
        return ', '.join(list(map(format_output, field)))
    elif type(field) in {set}:
        return ', '.join(list(map(format_output, sorted(field))))
    elif type(field) is float:
        return str(field).replace('.', ',')
    else:
        return str(field).replace('\n', '  ')


with open(r'mystery\crimescene') as f:
    scene = filter(lambda x: check_wonders(x), f.read().split(crime_separator))

with open(r'mystery\crimescene2', 'w') as f:
    t = list(sorted(scene, key=lambda x: (not 'clue' in x.lower(), x)))[:10]
    t = crime_separator + crime_separator.join(t)
    print(t.replace('\n\n\n', '\n'), file=f)

addresses = dict()
for streetname in os.listdir(r'mystery\streets'):
    with open(r'mystery\streets' + '\\' + streetname) as f:
        street = f.readlines()
        ints = set()
        for nr in range(len(street)):
            if 'SEE INTER' in street[nr]:
                ints.add((streetname, nr + 1, street[nr].strip().split('#')[1]))
    for inter in ints:
        intr = inter[2]
        with open(r'mystery\interviews\interview-' + intr) as inf:
            inview = inf.read().strip().lower().replace('\n', ' ')
            ns = bool(not_sus.findall(inview))
            # if ns: print(inview)
            streetname = inter[0].replace('_', ' ')
            addresses[(streetname, str(inter[1]))] = {'ns': ns, 'interv': inview}

for comp in os.listdir(r'mystery\memberships'):
    with open(r'mystery\memberships' + '\\' + comp) as f:
        for name in f:
            name = name.strip()
            if name not in people:
                people[name] = Person(name)
            people[name].add_member(comp)

with open(r'mystery\people') as f:
    for line in f.readlines()[8:]:
        name, g, age, addr = line.strip().split('\t')
        name = name.strip()
        if name not in people:
            people[name] = Person(name)
        people[name].age = age.strip()
        people[name].gender = g.strip()
        people[name].addr = tuple(addr.split(', line '))
        if people[name].addr in addresses:
            people[name].interview = addresses[people[name].addr]['interv']
            people[name].not_sus = addresses[people[name].addr]['ns']

with open(r'mystery\vehicles') as f:
    for car in f.read().split('\n\n')[1:]:
        t = list(map(lambda x: x.replace('se Plate', 'te:').split(': ')[1].strip(), car.strip().split('\n')))
        plate, make, color, name, hei, wei = t
        if name not in people:
            people[name] = Person(name)
        people[name].height = in_to_cm(hei)
        people[name].weight = lb_to_kg(wei)
        people[name].plate = plate
        people[name].make = make
        people[name].color = color

f = open('people.csv', 'w')
print('Full name;First;Last;G;Height;Age;Weight;Address;Sus;Make;Color;Plate;Membership;Interview', file=f)
for name in sorted(people):
    output_row = [name]
    pp = people[name]
    output_row.append(pp.first)
    output_row.append(pp.last)
    output_row.append(pp.gender)
    output_row.append(pp.height)
    output_row.append(pp.age)
    output_row.append(pp.weight)
    output_row.append(pp.addr)
    output_row.append(pp.not_sus)
    output_row.append(pp.make)
    output_row.append(pp.color)
    output_row.append(pp.plate)
    output_row.append(pp.memberships)
    output_row.append(pp.interview)
    print(*list(map(format_output, output_row)), sep=';', file=f)
f.close()