| 1 |
nino.borges |
674 |
"""
|
| 2 |
|
|
|
| 3 |
|
|
LoadFilesMissing
|
| 4 |
|
|
|
| 5 |
|
|
Created by
|
| 6 |
|
|
Emanuel Borges
|
| 7 |
|
|
2020.03.25
|
| 8 |
|
|
|
| 9 |
|
|
A simple program that will attempt to make load files for an incoming prod where they only gave us text files and images.
|
| 10 |
|
|
This makes Lots of assumptions:
|
| 11 |
|
|
- Images are Tiff or JPG
|
| 12 |
|
|
- Images are single page
|
| 13 |
|
|
- Text is multipage
|
| 14 |
|
|
- There is a text file for all documents and there are not images not represented in the text files
|
| 15 |
|
|
- everything is named for the bates
|
| 16 |
|
|
|
| 17 |
|
|
"""
|
| 18 |
|
|
|
| 19 |
|
|
import os
|
| 20 |
|
|
import BatesRangeFunctions
|
| 21 |
|
|
|
| 22 |
|
|
def LfpWriteLine(openFileObj, bates, filePath, firstPage = True):
|
| 23 |
|
|
if os.path.splitext(filePath)[1].upper() == ".JPG":
|
| 24 |
|
|
docType = "4"
|
| 25 |
|
|
else:
|
| 26 |
|
|
docType = "2"
|
| 27 |
|
|
|
| 28 |
|
|
if firstPage:
|
| 29 |
|
|
firstPageMark = "D"
|
| 30 |
|
|
else:
|
| 31 |
|
|
firstPageMark = ""
|
| 32 |
|
|
#outputImagesFile.write("IM,%s,D,0,@;%s;%s\n"% (prevBates, imageFileMatrix[prevBates],docType))
|
| 33 |
|
|
openFileObj.write("IM,%s,%s,0,@;%s;%s\n"% (bates, firstPageMark, filePath, docType))
|
| 34 |
|
|
|
| 35 |
|
|
if __name__ == '__main__':
|
| 36 |
|
|
textFilesStartDir = r"\\sas40\sas40\33444\Inbound\2\092779\PNC - Response to Subpoena\PNC Subpoena Response (rec'd 9.27.19)\TEXT"
|
| 37 |
|
|
imageFilesStartDir = r"\\sas40\sas40\33444\Inbound\2\092779\PNC - Response to Subpoena\PNC Subpoena Response (rec'd 9.27.19)\IMAGES"
|
| 38 |
|
|
volumeName = "PNC003"
|
| 39 |
|
|
|
| 40 |
|
|
outputDirectory = r"C:\Temp"
|
| 41 |
|
|
|
| 42 |
|
|
approvedImageExtensions = ['.TIFF','.JPG','.TIF']
|
| 43 |
|
|
textFileMatrix = {}
|
| 44 |
|
|
imageFileMatrix = {}
|
| 45 |
|
|
|
| 46 |
|
|
## scan the text files folder and make a matrix of text files
|
| 47 |
|
|
for root, dirs, files in os.walk(textFilesStartDir):
|
| 48 |
|
|
for f in files:
|
| 49 |
|
|
if os.path.splitext(f)[1].upper() == ".TXT":
|
| 50 |
|
|
bates = os.path.splitext(f)[0]
|
| 51 |
|
|
textFileMatrix[bates] = os.path.join(root,f)
|
| 52 |
|
|
|
| 53 |
|
|
|
| 54 |
|
|
|
| 55 |
|
|
## scan the images folder and make a matrix of image files
|
| 56 |
|
|
for root, dirs, files in os.walk(imageFilesStartDir):
|
| 57 |
|
|
for f in files:
|
| 58 |
|
|
if os.path.splitext(f)[1].upper() in approvedImageExtensions:
|
| 59 |
|
|
bates = os.path.splitext(f)[0]
|
| 60 |
|
|
imageFileMatrix[bates] = os.path.join(root,f)
|
| 61 |
|
|
|
| 62 |
|
|
## make the dat and populate it with begno, endno, textPath
|
| 63 |
|
|
## make the LFP and populate with paths.
|
| 64 |
|
|
outputFile = open(os.path.join(outputDirectory,volumeName+".DAT"),'w')
|
| 65 |
|
|
outputFile.write("ProdBegBates|ProdEndBates|TextFilePath\n")
|
| 66 |
|
|
outputImagesFile = open(os.path.join(outputDirectory,volumeName+".LFP"),'w')
|
| 67 |
|
|
batesList = textFileMatrix.keys()
|
| 68 |
|
|
batesList.sort()
|
| 69 |
|
|
imageList = imageFileMatrix.keys()
|
| 70 |
|
|
imageList.sort()
|
| 71 |
|
|
prevBates = ""
|
| 72 |
|
|
|
| 73 |
|
|
for bates in batesList:
|
| 74 |
|
|
if prevBates:
|
| 75 |
|
|
batesEnum = BatesRangeFunctions.EnumerateBates(prevBates,bates)
|
| 76 |
|
|
outputFile.write("%s|%s|%s\n"%(batesEnum[0],batesEnum[-2], textFileMatrix[prevBates]))
|
| 77 |
|
|
firstImage = True
|
| 78 |
|
|
for page in batesEnum[:-1]:
|
| 79 |
|
|
## if os.path.splitext(imageFileMatrix[prevBates])[1].upper() == ".JPG":
|
| 80 |
|
|
## docType = "4"
|
| 81 |
|
|
## else:
|
| 82 |
|
|
## docType = "2"
|
| 83 |
|
|
if firstImage:
|
| 84 |
|
|
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = True)
|
| 85 |
|
|
#outputImagesFile.write("IM,%s,D,0,@;%s;%s\n"% (prevBates, imageFileMatrix[prevBates],docType))
|
| 86 |
|
|
firstImage = False
|
| 87 |
|
|
else:
|
| 88 |
|
|
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = False)
|
| 89 |
|
|
#outputImagesFile.write("IM,%s,,0,@;%s;%s\n"% (prevBates, imageFileMatrix[prevBates],docType))
|
| 90 |
|
|
|
| 91 |
|
|
prevBates = bates
|
| 92 |
|
|
|
| 93 |
|
|
else:
|
| 94 |
|
|
prevBates = bates
|
| 95 |
|
|
|
| 96 |
|
|
#print imageList[-1]
|
| 97 |
|
|
batesEnum = BatesRangeFunctions.EnumerateBates(bates,imageList[-1])
|
| 98 |
|
|
outputFile.write("%s|%s|%s\n"%(batesEnum[0],batesEnum[-1], textFileMatrix[bates]))
|
| 99 |
|
|
outputFile.close()
|
| 100 |
|
|
firstImage = True
|
| 101 |
|
|
for page in batesEnum:
|
| 102 |
|
|
## if os.path.splitext(imageFileMatrix[prevBates])[1].upper() == ".JPG":
|
| 103 |
|
|
## docType = "4"
|
| 104 |
|
|
## else:
|
| 105 |
|
|
## docType = "2"
|
| 106 |
|
|
if firstImage:
|
| 107 |
|
|
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = True)
|
| 108 |
|
|
#outputImagesFile.write("IM,%s,D,0,@;%s;%s\n"% (bates, imageFileMatrix[bates],docType))
|
| 109 |
|
|
firstImage = False
|
| 110 |
|
|
else:
|
| 111 |
|
|
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = False)
|
| 112 |
|
|
#outputImagesFile.write("IM,%s,,0,@;%s;%s\n"% (bates, imageFileMatrix[bates],docType))
|
| 113 |
|
|
outputImagesFile.close()
|
| 114 |
|
|
|
| 115 |
|
|
|
| 116 |
|
|
|
| 117 |
|
|
## run test to see if there are images missing from LFP
|
| 118 |
nino.borges |
675 |
contents = open(os.path.join(outputDirectory,volumeName+".LFP")).readlines()
|
| 119 |
|
|
errLog = open(os.path.join(outputDirectory,volumeName+".ERR"),'w')
|
| 120 |
|
|
testImageList = []
|
| 121 |
|
|
for line in contents:
|
| 122 |
|
|
line = line.replace("\n","")
|
| 123 |
|
|
line = line.split(",")
|
| 124 |
|
|
testImageList.append(line[1])
|
| 125 |
|
|
for bates in imageList:
|
| 126 |
|
|
if bates in testImageList:
|
| 127 |
|
|
pass
|
| 128 |
|
|
else:
|
| 129 |
|
|
errLog.write("%s not accounted for.\n"% bates)
|
| 130 |
|
|
|
| 131 |
|
|
errLog.close()
|
| 132 |
nino.borges |
674 |
|
