| 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" |
| 36 |
> |
textFilesStartDir = r"\\sas40\sas40\33444\Inbound\2\092779\PNC - Response to Subpoena\PNC Subpoena Response (rec'd 11.15.19)\VOL0001\TEXT" |
| 37 |
> |
imageFilesStartDir = r"\\sas40\sas40\33444\Inbound\2\092779\PNC - Response to Subpoena\PNC Subpoena Response (rec'd 11.15.19)\VOL0001\IMAGES" |
| 38 |
> |
volumeName = "PNC007" |
| 39 |
|
|
| 40 |
|
outputDirectory = r"C:\Temp" |
| 41 |
+ |
errLog = open(os.path.join(outputDirectory,volumeName+"_XDD"+".ERR"),'w') |
| 42 |
|
|
| 43 |
|
approvedImageExtensions = ['.TIFF','.JPG','.TIF'] |
| 44 |
|
textFileMatrix = {} |
| 62 |
|
|
| 63 |
|
## make the dat and populate it with begno, endno, textPath |
| 64 |
|
## make the LFP and populate with paths. |
| 65 |
< |
outputFile = open(os.path.join(outputDirectory,volumeName+".DAT"),'w') |
| 65 |
> |
outputFile = open(os.path.join(outputDirectory,volumeName+"_XDD"+".DAT"),'w') |
| 66 |
|
outputFile.write("ProdBegBates|ProdEndBates|TextFilePath\n") |
| 67 |
< |
outputImagesFile = open(os.path.join(outputDirectory,volumeName+".LFP"),'w') |
| 67 |
> |
outputImagesFile = open(os.path.join(outputDirectory,volumeName+"_XDD"+".LFP"),'w') |
| 68 |
|
batesList = textFileMatrix.keys() |
| 69 |
|
batesList.sort() |
| 70 |
|
imageList = imageFileMatrix.keys() |
| 77 |
|
outputFile.write("%s|%s|%s\n"%(batesEnum[0],batesEnum[-2], textFileMatrix[prevBates])) |
| 78 |
|
firstImage = True |
| 79 |
|
for page in batesEnum[:-1]: |
| 80 |
< |
## if os.path.splitext(imageFileMatrix[prevBates])[1].upper() == ".JPG": |
| 81 |
< |
## docType = "4" |
| 82 |
< |
## else: |
| 83 |
< |
## docType = "2" |
| 84 |
< |
if firstImage: |
| 85 |
< |
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = True) |
| 86 |
< |
#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)) |
| 80 |
> |
try: |
| 81 |
> |
imagePath = imageFileMatrix[page] |
| 82 |
> |
except: |
| 83 |
> |
imagePath = "ERROR FILE NOT FOUND.TIFF" |
| 84 |
> |
errLog.write("%s Image FILE NOT FOUND.\n"% page) |
| 85 |
> |
LfpWriteLine(outputImagesFile, page, imagePath, firstImage) |
| 86 |
> |
firstImage = False |
| 87 |
|
|
| 88 |
|
prevBates = bates |
| 89 |
|
|
| 96 |
|
outputFile.close() |
| 97 |
|
firstImage = True |
| 98 |
|
for page in batesEnum: |
| 99 |
< |
## if os.path.splitext(imageFileMatrix[prevBates])[1].upper() == ".JPG": |
| 100 |
< |
## docType = "4" |
| 101 |
< |
## else: |
| 102 |
< |
## docType = "2" |
| 103 |
< |
if firstImage: |
| 104 |
< |
LfpWriteLine(outputImagesFile, page, imageFileMatrix[page], firstPage = True) |
| 105 |
< |
#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)) |
| 99 |
> |
try: |
| 100 |
> |
imagePath = imageFileMatrix[page] |
| 101 |
> |
except: |
| 102 |
> |
imagePath = "ERROR FILE NOT FOUND.TIFF" |
| 103 |
> |
errLog.write("%s Image FILE NOT FOUND.\n"% page) |
| 104 |
> |
LfpWriteLine(outputImagesFile, page, imagePath, firstImage) |
| 105 |
> |
firstImage = False |
| 106 |
|
outputImagesFile.close() |
| 107 |
|
|
| 108 |
|
|
| 109 |
|
|
| 110 |
|
## run test to see if there are images missing from LFP |
| 111 |
< |
contents = open(os.path.join(outputDirectory,volumeName+".LFP")).readlines() |
| 112 |
< |
errLog = open(os.path.join(outputDirectory,volumeName+".ERR"),'w') |
| 111 |
> |
contents = open(os.path.join(outputDirectory,volumeName+"_XDD"+".LFP")).readlines() |
| 112 |
> |
|
| 113 |
|
testImageList = [] |
| 114 |
|
for line in contents: |
| 115 |
|
line = line.replace("\n","") |