Data Quality

# -*- coding: utf-8 -*-
import wx
import os
import fileinput
import sys



def clearfile(path):
    file = open(path, 'w')
    file.write(u' ')
    file.close()

def del_regexp(file, string):
    rst = []
    with open(file) as fd:
        t = fd.read()
        for line in t.splitlines():
            if line != string:
                rst.append(line)

    with open(file, 'w') as fd:
        fd.write(u'\n'.join(rst))
        fd.write(u'\n')

def replaceAll(file, searchExp,replaceExp):
    for line in fileinput.input(file, inplace=1):
        if searchExp in line:
            line = line.replace(searchExp,replaceExp)
        sys.stdout.write(line)

class StrangeFileWorker(dict):
    def load_file(self, file_name):
        with open(file_name) as fd:
            self.load_string(fd.read())

    def load_string(self, data):
        cur_key = ''
        for ln in filter(lambda a: a, data.split('\n')):
            if ln.endswith(':'):
                cur_key = ln.strip(':')
                self.setdefault(cur_key, [])
            elif cur_key:
                self[cur_key].append(ln)

    def __repr__(self):
        #return '\n'.join(map(lambda kv: '%s:\n%s\n' % (kv[0], '\n'.join(kv[1])), self.iteritems()))
        return '\n'.join(map(lambda kv: '\n%s:\n%s\n' % (kv[0], '\n'.join(kv[1])), self.iteritems()))

def write_regexp(file, table, regexp):
    sfw = StrangeFileWorker()

    fd = open(file)
    data = fd.read()
    #data = data.decode('cp1251')
    sfw.load_string(data)

    sfw[table].append(regexp)

    data = repr(sfw) # получили представление объекта
    #data = data.encode('cp1251') # вспомнили про кодировку
    fd.write(data) # записали
    fd.close()
#===============================================================================
# def write_regexp(file, table, regexp):
#    result = list()
#    sfw = StrangeFileWorker()
#    store = False
#    with open(file) as fd:
#        for line in fd:
#            line = line.strip(' \n\r:')
#            if result and not line:
#                break
#            if line == table:
#                store = True
#                continue
#            if line == ' ':
#                file.write(regexp)
#===============================================================================
def get_regexp(file, table):
    sfw = StrangeFileWorker()
    list = []
    fd = open(file)
    data = fd.read()
    #data = data.decode('cp1251')
    sfw.load_string(data)
    if table in sfw:
        for reg_exp in sfw[table]:
            list.append(reg_exp)
    else:
        list = []
    return list
    data = repr(sfw) # получили представление объекта
    #data = data.encode('cp1251') # вспомнили про кодировку
    fd.write(data) # записали
    fd.close()
#===============================================================================
# def get_regexp(file, table):
#    sfw = StrangeFileWorker()
#    result = list()
#    store = False
#    fd = open(file) # Открыли файл
#    data = fd.read() #прочитали
#    data = data.decode('utf8') #преобразовали кодировку
#    sfw.load_string(data) #скормили в обьект
#    for line in data:
#        line = line.strip(' \n\r:')
#        if result and not line:
#            break
#        if line == table:
#            store = True
#            continue
#        if store:
#            result = repr(sfw) # получили представление объекта
#            result = result.encode('utf8') # вспомнили про кодировку
#            result.append(line)
#    return result
#===============================================================================

def readfile(path):
    file = open(path, 'r')
    file = file.readlines()
    for i in xrange(len(file) - 1): file[i] = file[i][0:-1]
    return file

def writefile(table, regexp, path):
    file = open(path, 'a')
    file.write((u'\n%s:' % table))
    file.write((u'\n%s' % regexp))
    return file

class dictpick_files(wx.Dialog):
    def __init__(self):
        wildcard = 'All files (*.*)|*.*'
        self.dialog = wx.FileDialog(None, 'Choose a file', os.getcwd(),
            '', wildcard, wx.OPEN)
        if self.dialog.ShowModal() == wx.ID_OK:
            self.Path = self.dialog.GetPath()
            self.dialog.Destroy()
        self.dialog.Destroy()

    def readfile_dlg(self, file, path):
        file = open(path, 'r')
        file = file.readlines()
        for i in xrange(len(file) - 1): file[i] = file[i][0:-1]
        return file

    def writefile_dlg(self, regexp):
        self.file = open(('%s' % (self.Path)), 'a')
        self.file.write((u'\n%s' % regexp))
        return self.file