| 1 |
nino.borges |
584 |
"""
|
| 2 |
|
|
HW_IproExtract
|
| 3 |
|
|
|
| 4 |
|
|
This program will take a list of bates that we already have and compare it to a lfp from an ipro proj.
|
| 5 |
|
|
it will export a dat in 200k chunks for those that are in the ipro but you dont already have.
|
| 6 |
|
|
|
| 7 |
|
|
"""
|
| 8 |
|
|
|
| 9 |
|
|
import os
|
| 10 |
|
|
|
| 11 |
|
|
def writeValues(theList,fileCount):
|
| 12 |
|
|
print "200k threshold reached. writing values..."
|
| 13 |
|
|
filePath = r"C:\Users\eborges\Box Sync\Client\Honeywell\HOBX\Microfilm_Irrl_Export_Proj\Loads_B"
|
| 14 |
|
|
fileName = "load_%s"%fileCount
|
| 15 |
|
|
outputFile = open(os.path.join(filePath,fileName),'w')
|
| 16 |
|
|
for i in theList:
|
| 17 |
|
|
outputFile.write(i+"\n")
|
| 18 |
|
|
outputFile.close()
|
| 19 |
|
|
fileCount = fileCount+1
|
| 20 |
|
|
print "Values written to %s. Done."%fileCount
|
| 21 |
|
|
return [], fileCount
|
| 22 |
|
|
|
| 23 |
|
|
|
| 24 |
|
|
|
| 25 |
|
|
if __name__=='__main__':
|
| 26 |
|
|
iproFilePath = r"C:\Users\eborges\Box Sync\Client\Honeywell\HOBX\Microfilm_Irrl_Export_Proj\MicroficheLFPexports\37354040_MICROFICHE_REVIEW.lfp"
|
| 27 |
|
|
concordanceFilePath = r"C:\Users\eborges\Box Sync\Client\Honeywell\HOBX\Microfilm_Irrl_Export_Proj\Concordance\batesList.txt"
|
| 28 |
|
|
kpmgFilePath = r"C:\Users\eborges\Box Sync\Client\Honeywell\HOBX\Microfilm_Irrl_Export_Proj\KPMG\kpmgList.txt"
|
| 29 |
|
|
|
| 30 |
|
|
existingList = []
|
| 31 |
|
|
iproList = []
|
| 32 |
|
|
toBeDoneList = []
|
| 33 |
|
|
docCount = 0
|
| 34 |
|
|
fileCount = 0
|
| 35 |
|
|
|
| 36 |
|
|
|
| 37 |
|
|
print"Now creating existing list from Concordance set..."
|
| 38 |
|
|
contents = open(concordanceFilePath).readlines()
|
| 39 |
|
|
for i in contents:
|
| 40 |
|
|
i = i.replace("\n","")
|
| 41 |
|
|
existingList.append(i)
|
| 42 |
|
|
|
| 43 |
|
|
print "Done."
|
| 44 |
|
|
print "Now adding KPMG set to exisitng list..."
|
| 45 |
|
|
contents = open(kpmgFilePath).readlines()
|
| 46 |
|
|
for i in contents:
|
| 47 |
|
|
i = i.replace("\n","")
|
| 48 |
|
|
existingList.append(i)
|
| 49 |
|
|
|
| 50 |
|
|
print "Done."
|
| 51 |
|
|
print "Now creating Ipro list..."
|
| 52 |
|
|
contents = open(iproFilePath).readlines()
|
| 53 |
|
|
for line in contents:
|
| 54 |
|
|
line = line.split(",")
|
| 55 |
|
|
bates = line[1]
|
| 56 |
|
|
docQl = line[2]
|
| 57 |
|
|
if line[2] == "D":
|
| 58 |
|
|
iproList.append(bates.upper())
|
| 59 |
|
|
print "Done."
|
| 60 |
|
|
print "There are %s existing files and %s files in ipro."%(len(existingList),len(iproList))
|
| 61 |
|
|
print "Reversing the file."
|
| 62 |
|
|
iproList.reverse()
|
| 63 |
|
|
print "Now gathering diff..."
|
| 64 |
|
|
for x in iproList:
|
| 65 |
|
|
if x in existingList:
|
| 66 |
|
|
pass
|
| 67 |
|
|
else:
|
| 68 |
|
|
toBeDoneList.append(x)
|
| 69 |
|
|
docCount = docCount+1
|
| 70 |
|
|
if docCount >200000:
|
| 71 |
|
|
toBeDoneList,fileCount = writeValues(toBeDoneList,fileCount)
|
| 72 |
|
|
docCount = 0 |
