A script written to correct a file management software incorrectly flipping images on a file server. This would scan certain directories and correct the pdfs.

PeDeEff.py

from PyPDF2 import PdfFileWriter, PdfFileReader
import os, datetime, time

cwd = os.getcwd()
root_dir = cwd
main_dir = cwd + '\\00\\00'
log_path = cwd + '\\PeDeEfLog'
log_file_path = log_path + '\\log.txt'

class Parser():
    def __init__(self, pdfnum):
        self.folders = os.listdir(main_dir)
        self.log_file = ''
        self.current_folder = ''
        self.current_temp_folder = ''
        self.current_folder_files = ''
        self.current_file_log = ''
        self.current_file_log_contents = ''
        self.number_of_pdfs = pdfnum

    def log(self, message):
        ts = time.time()
        time_stamp = datetime.datetime.fromtimestamp(ts).strftime('%Y-%m-%d %H:%M:%S')
        self.log_file.write(time_stamp + ' : ' + message + '\n')

    def log_pdf(self, pdf):
        self.current_file_log.write(pdf+'\n')

    def log_check(self):
        if os.path.isdir(log_path):
            print "Log Path Exists"
        else:
            os.mkdir(log_path)
            print "Created Log Path"

        if os.path.isfile(log_file_path):
            print "Log File Exists"
        else:
            try:
                f = open(log_file_path, "w")
            except IOError:
                pass
            f.close()
            print "Created Log File"

        self.log_file = open(log_file_path, "a")

    def file_check(self):
        self.current_file_log = self.current_temp_folder + '\\files.txt'
        if os.path.isfile(self.current_file_log):
            self.log(str(self.current_file_log)+ "Temp file exists")
            x = open(self.current_file_log, "r")
            self.current_file_log_contents = x.read().split('\n')
            x.close()

        else:
            try:
                f = open(self.current_file_log, "w")
            except IOError:
                pass

            f.close()
            self.log("File created")

        self.current_file_log = open(self.current_file_log, "a")

    def temp_check(self):
        if 'temp' in self.current_folder_files:
            self.current_temp_folder = self.current_folder +'\\temp'
            self.file_check()
            self.log('Directory Exists')

        else:
            os.mkdir(self.current_folder +'\\temp')
            self.current_temp_folder = self.current_folder +'\\temp'
            self.file_check()
            self.log(str(self.current_temp_folder)+'Temp directory Created')

    def alter_pdf(self, pdf):
        pdf_path = self.current_folder + '\\' + pdf
        destination = self.current_temp_folder + '\\' + pdf
        output = PdfFileWriter()

        with open(pdf_path, "rb") as f:
            input = PdfFileReader(pdf_path, 'rb')

        pagecount = int(input.getNumPages())
        p = input.getPage(0)
        (w, h) = p.mediaBox.upperRight


        for page in range(pagecount):
            # print page
            if page is 0 or page is pagecount - self.number_of_pdfs:
                # print "no change"
                output.addPage(input.getPage(page))
            else:
                # print "change"
                p = input.getPage(page)
                p.scaleTo(w, h)

                output.addPage(p)

        os.rename(pdf_path, destination)
        with open(pdf_path, 'wb') as f:
           output.write(f)

    def run(self):
        self.log_check()
        self.log("Script Started.")
        self.log("Log Check Complete.")
        for folder in self.folders:
            self.current_folder = main_dir+'\\'+folder
            self.log('Processing '+str(self.current_folder))
            self.current_folder_files = os.listdir(self.current_folder)

            self.temp_check()

            for pdf in self.current_folder_files:
                if pdf.endswith('.pdf'):
                    if pdf in self.current_file_log_contents:
                        self.log("The file " + pdf + " is in file log")
                    else:
                        self.log(pdf +" "+ "File is not in file log")
                        self.alter_pdf(pdf)
                        self.log_pdf(pdf)

                else:
                    self.log(pdf +"File is not a pdf, skipping...")

        self.log("Script ending, closing log files...")
        self.log_file.close()
        self.current_file_log.close()


os.chdir(root_dir +'\\Accounting\\Credit Cards\\Credit Card Charges Forms')
cwd = os.getcwd()
main_dir = cwd + '\\00\\00'
log_file_path = log_path + '\\log.txt'
if os.path.isdir(main_dir):
    x = Parser(1)
    x.run()
    x = None

os.chdir(root_dir +'\\Accounting\\Accounts Payable\\Purchase Requisitions')
cwd = os.getcwd()
main_dir = cwd + '\\00\\00'
log_file_path = log_path + '\\log.txt'
if os.path.isdir(main_dir):
    x = Parser(1)
    x.run()
    x = None

os.chdir(root_dir +'\\Accounting\\Credit Cards\\Travel Expense Forms')
cwd = os.getcwd()
main_dir = cwd + '\\00\\00'
log_file_path = log_path + '\\log.txt'
if os.path.isdir(main_dir):
    x = Parser(2)
    x.run()
    x = None

os.chdir(root_dir +'\\Accounting\\Credit Cards\\Employee Travel Statements')
cwd = os.getcwd()
main_dir = cwd + '\\00\\00'
log_file_path = log_path + '\\log.txt'
if os.path.isdir(main_dir):
    x = Parser(2)
    x.run()
    x = None