| 1 |
nino.borges |
560 |
"""
|
| 2 |
|
|
eDiiToDAT.py
|
| 3 |
|
|
|
| 4 |
|
|
Created by Emanuel Borges
|
| 5 |
|
|
08.22.2014
|
| 6 |
|
|
|
| 7 |
|
|
This program will take and parse an eDii file and convert it to a dat file, so that it can be loaded into Relativity
|
| 8 |
|
|
or Concordance.
|
| 9 |
|
|
|
| 10 |
|
|
"""
|
| 11 |
|
|
|
| 12 |
|
|
import os
|
| 13 |
|
|
|
| 14 |
|
|
class ParseDiiFile(object):
|
| 15 |
|
|
"""This is the main class for the program that parses and converts 1 or multiple files"""
|
| 16 |
|
|
def __init__(self):
|
| 17 |
|
|
pass
|
| 18 |
|
|
|
| 19 |
|
|
class HeadderInfo(object):
|
| 20 |
|
|
def __init__(self):
|
| 21 |
|
|
self.headderRow = ['BegNo']
|
| 22 |
|
|
|
| 23 |
|
|
def ParseForHeadderInfo(self,diiPath):
|
| 24 |
|
|
contents = open(diiPath).readlines()
|
| 25 |
|
|
for line in contents:
|
| 26 |
|
|
if line[:2] == '@C':
|
| 27 |
|
|
line = line.replace("\n","")
|
| 28 |
|
|
fieldName = line.split(" ")[1]
|
| 29 |
|
|
if fieldName in self.headderRow:
|
| 30 |
|
|
pass
|
| 31 |
|
|
else:
|
| 32 |
|
|
self.headderRow.append(fieldName)
|
| 33 |
|
|
def WriteHeadderRow(self,outputFilePath):
|
| 34 |
|
|
count = 1
|
| 35 |
|
|
outputFile = open(outputFilePath,'w')
|
| 36 |
|
|
for i in self.headderRow:
|
| 37 |
|
|
outputFile.write(i)
|
| 38 |
|
|
if count < len(self.headderRow):
|
| 39 |
|
|
outputFile.write("|")
|
| 40 |
|
|
else:
|
| 41 |
|
|
outputFile.write("\n")
|
| 42 |
|
|
count = count + 1
|
| 43 |
|
|
outputFile.close()
|
| 44 |
|
|
|
| 45 |
|
|
class SingleRecord(object):
|
| 46 |
|
|
def __init__(self,begno):
|
| 47 |
|
|
self.BegNo = begno
|
| 48 |
|
|
def ReturnInformation(self,info):
|
| 49 |
|
|
data = getattr(self,info)
|
| 50 |
|
|
return data
|
| 51 |
|
|
def SaveInformation(self,field,info):
|
| 52 |
|
|
data = setattr(self,field,info)
|
| 53 |
|
|
|
| 54 |
|
|
|
| 55 |
|
|
if __name__ == '__main__':
|
| 56 |
|
|
hi = HeadderInfo()
|
| 57 |
|
|
for root, dirs, files in os.walk(r"W:\Manny\Client\DII FILES"):
|
| 58 |
|
|
for f in files:
|
| 59 |
|
|
if os.path.splitext(f)[1].upper() == ".DII":
|
| 60 |
|
|
hi.ParseForHeadderInfo(os.path.join(root,f))
|
| 61 |
|
|
hi.WriteHeadderRow(r"W:\Manny\Client\output.dat")
|
| 62 |
|
|
count = 0
|
| 63 |
|
|
matrix = {}
|
| 64 |
|
|
for root, dirs, files in os.walk(r"W:\Manny\Client\DII FILES"):
|
| 65 |
|
|
for f in files:
|
| 66 |
|
|
if os.path.splitext(f)[1].upper() == ".DII":
|
| 67 |
|
|
contents = open(os.path.join(root,f))
|
| 68 |
|
|
for line in contents:
|
| 69 |
|
|
line = line.replace("\n","")
|
| 70 |
|
|
if line[:2] == '@T':
|
| 71 |
|
|
begno = line.split("@T")[1]
|
| 72 |
|
|
## I'd like to just use the bates as teh key but what if it exists...
|
| 73 |
|
|
count = count+1
|
| 74 |
|
|
matrix[count] = SingleRecord(begno)
|
| 75 |
|
|
|
| 76 |
|
|
elif line[:2] == '@C':
|
| 77 |
|
|
field = line.split("@C")[1]
|
| 78 |
|
|
field = field.split(" ")[1]
|
| 79 |
|
|
field = field.strip()
|
| 80 |
|
|
#print field
|
| 81 |
|
|
matrix[count].SaveInformation(field,line.split(field)[1])
|
| 82 |
|
|
recordList = matrix.keys()
|
| 83 |
|
|
recordList.sort()
|
| 84 |
|
|
outputFile = open(r"W:\Manny\Client\output.dat",'a')
|
| 85 |
|
|
|
| 86 |
|
|
for i in recordList:
|
| 87 |
|
|
cnt = 1
|
| 88 |
|
|
for field in hi.headderRow:
|
| 89 |
|
|
try:
|
| 90 |
|
|
outputFile.write(matrix[i].ReturnInformation(field).strip())
|
| 91 |
|
|
if cnt < len(hi.headderRow):
|
| 92 |
|
|
outputFile.write("|")
|
| 93 |
|
|
else:
|
| 94 |
|
|
outputFile.write("\n")
|
| 95 |
|
|
cnt = cnt + 1
|
| 96 |
|
|
except:
|
| 97 |
|
|
if cnt < len(hi.headderRow):
|
| 98 |
|
|
outputFile.write("|")
|
| 99 |
|
|
else:
|
| 100 |
|
|
outputFile.write("\n")
|
| 101 |
|
|
cnt = cnt +1
|
| 102 |
|
|
#outputFile.writelines(matrix[i].__dict__)
|
| 103 |
|
|
#outputFile.write("\n")
|
| 104 |
|
|
outputFile.close()
|
| 105 |
|
|
|
