getLatestFileName.py

# from DataCleaning.library.functions.getLatestFileName import *
# getLatestFileName(keyWord, fileNames, sep = "_", comparePath = "")
from DataCleaning.library.functions.getFileNames import *

def getLatestFileName(keyWord, fileNames, sep = "_", comparePath = ""):
    tempNames = [f for f in fileNames if keyWord in f]
    numb = 0
    output = 0
    specialKeyWords = ["History order data 2018 April to", "Open OB", "DCG_IBM_In_Transit", "Marathon Price File", "New Buy & Procurement"]
    #                 [USAGE_IBM,                          INV_IBM_OOB,INV_IBM_IT         ]
    invWeeklyWords = ["Weekly Inventory Details - Canada"]
    invKeyWords = ["Inventory_CPPS", "Inventory_PIMS", "Weekly Physical Inventory_PIMS"]
    SAPKeyWords = ["INV_H023", "H041", "INV_DC", "INV_DC_HS", "USAGE_LENOVO", "INV_DC_OpenPO",
                   "Reverse_AFR_QTY_Country", "Reverse_RTV_WIP", "Exchange_Rate", "Moving_Average_Price",
                   "MMPP091", "MMPP099", "MD_PN", "POU_In_Transit",
                   "DCG_SERVICE_PART_ATTRIBUTES", "Sub_PN_Hard_Flag", "SUB_TO_PLANNING", "Sub_PN_TM_STM", "MD_PO"]
    for i in tempNames:
        if keyWord in invKeyWords:
            tempNum = formatInvDate(i, sep)
        elif keyWord in invWeeklyWords:
            tempNum = formatInvWeeklyDate(i)
        elif keyWord == specialKeyWords[0]:
            tempNum = formatUsageDate(i)
        elif keyWord == specialKeyWords[1]:
            tempNum = formatOOBDate(i)
        elif keyWord == specialKeyWords[2]:
            tempNum = formatITDate(i)
        elif keyWord == specialKeyWords[3]:
            tempNum = formatMarathonPriceDate(i)
        elif keyWord == specialKeyWords[4]:
            tempNum = formatCeMemoNewBuyDate(i)
        elif keyWord in SAPKeyWords:
            tempNum = formatSAPDate(i)
        else:
            tempNum = formatITDate(i)

        tempNum = int(tempNum)
        if tempNum > numb:
            numb = tempNum
            output = i
    if comparePath == "":
        return str(output)
    else:
        currentLatest = getLatestFileName(keyWord, getFileNames(comparePath))
        if output == currentLatest:
            return "ERROR"
        else:
            return output


# def getFileNames(wd):
#     fileNames = [i for i in os.listdir(wd)
#                  if i.endswith(".xlsx") or i.endswith(".XLSX")]# and os.path.isfile(os.getcwd() + '\' + i)]
#     return fileNames

def formatITDate(fname):
    tempStr = fname.split(".")[0].replace("_", "").replace(" ", "")
    start = 0
    end = len(tempStr)
    state = 0
    for j in range(len(tempStr)):
        if state == 0:
            if tempStr[j].isnumeric():
                start = j
                state = 1
        elif state == 1:
            if not tempStr[j].isnumeric():
                end = j
                state = 2
        else:
            break
    if state == 0:
        return "-1"
    return tempStr[start:end]

def formatSAPDate(fname, sep = "_"):
    return fname.split(".")[0].split(sep)[-1]

def formatRegularDate(fname, sep = "_"):
    return fname.split(".")[0].split(sep)[-1][:8]

# 07162021 -> 20210716
def formatInvDate(fname, sep = "_"):
    tempString = fname.split(".")[0].split(sep)[-1][:8]
    return tempString[-4:] + tempString[:-4]

def formatInvWeeklyDate(fname):
    tempString = formatITDate(fname)
    return tempString[-4:] + tempString[:-4]

import calendar
month2Num = {name: num for num, name in enumerate(calendar.month_abbr) if num}

def formatUsageDate(fname):
    if "_" not in fname:
        tempStr = fname.split(".")[0].split("April to")[1].replace(" ", "")
        start = 0
        end = len(tempStr)
        state = 0
        for j in range(len(tempStr)):
            if state == 0:
                if not tempStr[j].isnumeric():
                    start = j
                    state = 1
            elif state == 1:
                if tempStr[j].isnumeric():
                    end = j
                    state = 2
            else:
                break
        # mon = datetime.datetime.strptime(tempStr[start:end], "%b")
        year = tempStr[:start]
        mon = ("0" + str(month2Num[tempStr[start:end][:3]]))[-2:]
        day = ("0" + tempStr[end:])[-2:]
        return(year + mon + day)
    else:
        return "-1"

def formatOOBDate(fname):
    tempList = fname.split(".")
    year = tempList[-2][:4]
    mon = tempList[0][-2:]
    day = tempList[1]
    return year + mon + day


def formatMarathonPriceDate(fname):
    tempString = fname.split(".")[-2].replace('-', '').replace(" ", '')
    if ')' in tempString:
        tempString = tempString.split(')')[-1]
        output = scanNums(tempString)
    else:
        output = scanNums(tempString, back = True)
    return output

def formatCeMemoNewBuyDate(fname):
    return fname.split(".")[-2].split(" ")[-1]

def scanNums(string, start = 0, inputEnd = 0, back = False):
    state = 0
    end = len(string)
    step = 1
    first = 0
    last = len(string)
    if back:
        step = -1
        first = len(string) - 1
        last = -1
    for i in range(first, last, step):
        if state == 0:
            if string[i].isnumeric():
                    start = i
                    state = 1
        elif state == 1:
            if not string[i].isnumeric():
                end = i
                state = 2
        else:
            break
    if back:
        tempStart = end + 1
        tempEnd = start + 1
        start = tempStart
        end = tempEnd
    if inputEnd != 0:
        end = start + inputEnd
    return int(string[start:end])

def keepNums(string):
    tempList = []
    for i in range(len(string)):
        if string[i].isnumeric():
            tempList.append(string[i])
    return int(''.join(tempList))