Python: file format

Abstract:Transformation among *.txt, *xls file formats

This script contains two class known as os_file and file_format. Both combined can finish:

1. Read txt file into dictionary

2. Convert *.txt/*.csv file into *xls file

Here is the script:

import os

import xlwt

import xlrd # work xls file

import csv

import shutil

import sh

######################

#class

class os_file:

def __init__(self, file):

self.file=file

def file_name(self):

items=self.file.split('/')

file_name=items[-1]

return file_name

def name_head(self):

file_name=self.file_name()

head, tail=file_name.split('.')

return(head)

def name_tail(self):

file_name=self.file_name()

head, tail=file_name.split('.')

return(tail)

def change_tail(self, tail):

head=self.name_head()

new_name=head+'.'+tail

return new_name

def file_size(self):

statinfo = os.stat(self.file)

file_size=statinfo.st_size

return(file_size)

def format_dir(self, dir):

if not dir[-1] == '/':

dir=dir+'/'

if not os.path.isdir(dir):

os.mkdir(dir, 0755)

return dir

def cp_file(self, dir):

shutil.copy(self.file, dir)

out_file=self.format_dir(dir)+self.file_name()

return out_file

def soft_cp(self, dir):

out_file=self.format_dir(dir)+self.file_name()

if os.path.isfile(out_file):

return 'no'

else:

shutil.copy(self.file, dir)

return out_file

######################################

#class

class file_format:

def __init__(self, infile):

self.infile=infile

self.OF=os_file(infile)

self.infile_tail=self.OF.name_tail()

if self.infile_tail == 'txt':

self.sep="\t"

elif self.infile_tail == 'csv':

self.sep=","

#self.outfile=outfile

def out_file(self, outdir="", file_tail=""):

if not os.path.isdir(outdir):

outdir=os.path.dirname(self.infile)

outfile=self.OF.format_dir(outdir)+self.OF.name_head()

if not file_tail=="":

outfile=outfile+'.'+file_tail

return outfile

def read_file(self):

arr=[]

in_obj=open(self.infile,'r')

for line in in_obj:

arr.append(line)

print line

in_obj.close()

return arr

def read_lines(self, num=1):

arr=[]

in_obj=open(self.infile,'r')

for i in range(num):

line=in_obj.readline()

if line=="":

break

else:

line=line.strip("\n")

arr.append(line)

in_obj.close()

return arr

def txt_to_nested_dict(self):

dict={}

in_obj=open(self.infile,'r')

first_line=in_obj.readline()

first_line=first_line.strip("\n")

header=first_line.split("\t")#colnames

header.pop(0)#remove the first item

#

for line in in_obj:

line=line.strip("\n")

items=line.split("\t")

rowname=items.pop(0)

dict2={}

for index, value in enumerate(items):

colname=header[index]

dict2[colname]=value

dict[rowname]=dict2

return dict

def gothrough_dict(self, dict):

for key in dict:

print key, dict[key], dict[key]['PC_DU145b']

def txt_to_xls(self, outdir=""):

outfile=self.out_file(outdir, 'xls')

#read infile

file_obj = open(self.infile, 'r')

row_list = []

for row in file_obj:

row_list.append(row.split())

column_list = zip(*row_list)

#set workbook and add sheet

workbook = xlwt.Workbook()

worksheet = workbook.add_sheet('Sheet1')

i = 0

for column in column_list:

for item in range(len(column)):

worksheet.write(item, i, column[item])

workbook.save(outfile)

i+=1

return(outfile)

#############

#main program

ff=file_format('/home/yuan/data_2/test/test.txt')

#read the first 1 lines

print ff.read_lines(1)

#read txt into dict

dict=ff.txt_to_nested_dict()

#read dict

ff.gothrough_dict(dict)

#convert txt file to xls file

ff.txt_to_xls()

Writing date: 20150515