def getReceipt(sim_pred, resultMap, i):
print('input receipt {}'.format(sim_pred))
if sim_pred.find('年') > -1:
return
if priceUtils.checkMnyStr(sim_pred):
return
sim_pred = sim_pred.replace('ｰ', '-').replace('。', '.')
sim_pred = jaconv.z2h(sim_pred, digit=True, ascii=True)
if sim_pred.find('-') == -1:
return
# sim_pred = numberUtils.numberReplacement(sim_pred)
if (sim_pred.find('-') == 0):
sim_pred = str(1) + sim_pred
tmpList = sim_pred.split('-')
tmpHead = tmpList[0]
tmpHead = tmpHead[-1]
tmpHead = numberUtils.numberReplacement(tmpHead)
lstHead = re.findall(r'\d+', tmpHead)
tmpHead = ''.join(lstHead)
if (tmpHead == ''):
tmpHead = '1'
tmpTail = tmpList[-1][:4]
tmpTail = numberUtils.numberReplacement(tmpTail)
lstTail = re.findall(r'\d+', tmpTail)
tmpTail = ''.join(lstTail)
if tmpTail == '':
tmpTail = '1234'
resultMap['6_receiptNO'] = tmpHead + 'ｰ' + tmpTail
# resultMap['6_receiptNO']=tmpHead+'-'+tmpTail
print('output receiptNO {}'.format(resultMap['6_receiptNO']))
def getCategoryAfter(tmpResult,resultMap,i):
tmpResult = jaconv.z2h(tmpResult, digit=True, ascii=True)
# not minus ｰ
if(tmpResult.find('責')>-1 \
or tmpResult.find('No')>-1 \
or tmpResult.find('点')>-1 \
or tmpResult.find('×')>-1 \
# or tmpResult.find('-')>-1 \
or tmpResult.find('ｰ')>-1 \
or tmpResult.find(':')>-1 \
or tmpResult.find('NO')>-1):
return
if not priceUtils.checkMnyStr(tmpResult):
return
print('input_{} category {}'.format(i,tmpResult))
tmpResult=numberUtils.getMny(tmpResult)
if(tmpResult==''):
return
tmpResult=numberUtils.numberReplacement(tmpResult)
lstCatPrice=re.findall(r'\d+', tmpResult)
sCatPrice=''.join(lstCatPrice)
if(sCatPrice==''):
iCatPrice=0
else:
iCatPrice=int(sCatPrice)
if tmpResult.find('-')>-1:
iCatPrice*=-1
if(iCatPrice !=0):
resultMap['suffix_catPrice'].append(iCatPrice)
print('output iCatPrice------------- {}'.format(iCatPrice))
def getYear(sim_pred):
sYear = sim_pred[0:sim_pred.find('年')]
sYear = numberUtils.numberReplacement(sYear)
lstYear = re.findall(r'\d+', sYear)
sYear = ''.join(lstYear)
if (len(sYear) != 4):
sYear = '2017'
if int(sYear) > 2019 or int(sYear) < 2016:
sYear = '2017'
return int(sYear)
def getTax(sim_pred, resultMap, i):
print('input_{} tax {}'.format(i, sim_pred))
resultMap['pos_tax'] = i
sim_pred = numberUtils.getMny(sim_pred)
sim_pred = numberUtils.numberReplacement(sim_pred)
lstTax = re.findall(r'\d+', sim_pred)
sTax = ''.join(lstTax)
if (sTax == ''):
iTax = 0
else:
iTax = int(sTax)
resultMap['a_tax'] = iTax
if (resultMap['a_tax'] > 0):
print('output tax------------- {}'.format(resultMap['a_tax']))
def getTotalPrice(sim_pred, resultMap, i):
print('input_{} totalPrice {}'.format(i, sim_pred))
resultMap['pos_total'] = i
sim_pred = numberUtils.getMny(sim_pred)
sim_pred = numberUtils.numberReplacement(sim_pred)
lstTotal = re.findall(r'\d+', sim_pred)
sTotal = ''.join(lstTotal)
if (sTotal == ''):
iTotal = 0
else:
iTotal = int(sTotal)
resultMap['5_total'] = iTotal
if (resultMap['5_total'] > 0):
print('output total------------- {}'.format(resultMap['5_total']))
def getMin(sim_pred):
if (sim_pred.find(':') == -1):
iMin = 0
else:
sMin = sim_pred[sim_pred.find(':') + 1:]
sMin = numberUtils.numberReplacement(sMin)
lstMin = re.findall(r'\d+', sMin)
sMin = ''.join(lstMin)
if (sMin == ''):
iMin = 0
else:
iMin = int(sMin)
if (iMin > 59):
iMin = 0
return iMin
def getHour(sim_pred):
if (sim_pred.find(':') == -1):
iHour = 0
else:
sHour = sim_pred[sim_pred.find(':') - 2:sim_pred.find(':')]
sHour = numberUtils.numberReplacement(sHour)
sHour = sHour[-3:]
lstHour = re.findall(r'\d+', sHour)
sHour = ''.join(lstHour)
if (sHour == ''):
iHour = 0
else:
iHour = int(sHour)
if (iHour > 23):
iHour = 0
return iHour
def amendStaff(sim_pred, resultMap, i):
if (not isStaffStr(sim_pred)):
return
print('input staffNo_amend {}'.format(sim_pred))
staffNoOrigin = resultMap['7_staffNO']
sim_pred = sim_pred.replace(',', '.').replace('。', '.').replace('黄', '責')
if (sim_pred.find('.') > -1):
tmpStrs = sim_pred.split('.')[-1]
elif (sim_pred.find('責') > -1):
tmpStrs = sim_pred.split('責')[-1]
else:
return
tmpStrs = numberUtils.numberReplacement(tmpStrs)
lstStaff = re.findall(r'\d+', tmpStrs)
tmpStrs = ''.join(lstStaff)
if (len(tmpStrs) >= 2):
resultMap['7_staffNO'] = tmpStrs
print('output staffNo_amend {}'.format(resultMap['7_staffNO']))
def getMonth(sim_pred):
iPosYear = sim_pred.find('年')
iPosMonth = sim_pred.find('月')
if (iPosMonth == -1):
iMonth = 1
elif (iPosMonth > iPosYear + 3): #maybe week
iMonth = 1
else:
sMonth = sim_pred[iPosYear + 1:iPosMonth]
sMonth = numberUtils.numberReplacement(sMonth)
lstMonth = re.findall(r'\d+', sMonth)
sMonth = ''.join(lstMonth)
if (sMonth == ''):
iMonth = 1
else:
iMonth = int(sMonth)
if (iMonth > 12 or iMonth == 0):
iMonth = 1
return iMonth
def getDay(sim_pred):
iPosYear = sim_pred.find('年')
iPosDay = sim_pred.find('日')
if (iPosDay == -1):
iDay = 1
elif (iPosDay > iPosYear + 6): #maybe week
iDay = 1
else:
sDay = sim_pred[iPosDay - 2:iPosDay]
sDay = numberUtils.numberReplacement(sDay)
lstDate = re.findall(r'\d+', sDay)
sDay = ''.join(lstDate)
if (sDay == ''):
iDay = 1
else:
iDay = int(sDay)
if (iDay > 31 or iDay == 0):
iDay = 1
return iDay
def getNo(sim_pred, resultMap, i):
if (not isStaffStr(sim_pred)):
return
print('input staffNo {}'.format(sim_pred))
# ７ｰ1784 No.0 8
sim_pred = sim_pred.replace(',', '.').replace('。', '.').replace('黄', '責')
sim_pred = jaconv.z2h(sim_pred, digit=True, ascii=True)
if (sim_pred.find('.') > -1):
tmpStrs = sim_pred.split('.')[-1]
elif (sim_pred.find('責') > -1):
tmpStrs = sim_pred.split('責')[-1]
else:
return
tmpStrs = numberUtils.numberReplacement(tmpStrs)
# lstStaff=re.findall(r'\d+', tmpStrs)
# tmpStrs=''.join(lstStaff)
resultMap['7_staffNO'] = tmpStrs
# 08
print('output staffNo {}'.format(resultMap['7_staffNO']))
resultMap['pos_staff'] = i
def getTel(tmpResult, resultMap, i):
print('input tel {}'.format(tmpResult))
if (tmpResult.find('年') > -1):
return
if (resultMap['pos_tel_before'] > 0 and tmpResult.find(' ') > -1):
print('tel_before {}'.format(resultMap['pos_tel_before']))
telBefore = resultMap['tel_before'].split(':')
if (len(telBefore) > 1):
telBeforeNew = telBefore[1]
telAfter = tmpResult.split(':')
if (len(telAfter) > 1):
telAfterNew = telAfter[1]
lsStr1 = list(telAfterNew)
for idx, val in enumerate(telAfterNew):
if (val == ' '):
lsStr1[idx] = telBeforeNew[idx]
tmpResult = jaconv.z2h(tmpResult, digit=True, ascii=True)
if (tmpResult.find('話') > -1):
tmpResult = tmpResult.split('話')[1]
if (tmpResult.find(':') > -1):
tmpResult = tmpResult.split(':')[1]
tmpResult = numberUtils.numberReplacement(tmpResult)
strList = re.findall(r'\d+', tmpResult)
tmpResult = ''.join(strList)
if (len(tmpResult) > 10):
tmpResult = tmpResult[-10:-1]
if (resultMap['3_tel'] == 'none'): #init value
resultMap['3_tel'] = tmpResult
elif (len(resultMap['3_tel']) < len(tmpResult)):
resultMap['3_tel'] = tmpResult
if len(tmpResult) == 9:
tmpResult = str(0) + tmpResult
if (resultMap['3_tel'] == tmpResult and len(tmpResult) > 6):
print('output tel{} {}'.format('-' * 10, tmpResult))
if (tmpResult != 'none'):
resultMap['pos_tel_after'] = i
def parseResult(result, resultMap, im_name):
for i in result: # for trim string, convert string_______step 1
sim_pred = str(result[i][1]).strip()
sim_pred = numberUtils.removeQuote(sim_pred)
sim_pred = jaconv.z2h(sim_pred, digit=True, ascii=True)
result[i][1] = sim_pred
# TODO # get all key position
# ___________________________________________________step 2__prepare_pos
lingUtils.getPosLing(resultMap, result)
pos_ling_after = resultMap['pos_ling_after']
for i in result:
if pos_ling_after > 0 and i <= pos_ling_after:
continue
if resultMap['pos_mny_after'] > 0:
break
sim_pred = str(result[i][1])
if sim_pred.find('￥') > -1:
resultMap['pos_mny_after'] = i
pos_mny = resultMap['pos_mny_after']
for i in result:
if pos_ling_after > 0 and i <= pos_ling_after:
continue
if resultMap['pos_tax_after'] > 0:
break
sim_pred = str(result[i][1])
if (i > 8 and taxUtils.checkIsTaxStr(sim_pred)):
resultMap['pos_tax_after'] = i
pos_tax = resultMap['pos_tax_after']
if pos_tax > 0 and pos_mny > 0 and pos_mny > pos_tax:
pos_mny = 0
if pos_tax > 0 and pos_ling_after > 0 and pos_tax < pos_ling_after:
pos_ling_after = 0
for i in result:
if pos_mny > 0 and i >= pos_mny:
break
if pos_tax > 0 and i >= pos_tax:
break
if resultMap['pos_time_after'] > 0:
break
sim_pred = str(result[i][1])
if (i > 3 and resultMap['pos_time_after'] == 0):
if (sim_pred.find('年') > -1):
resultMap['pos_time_after'] = i
pos_time = resultMap['pos_time_after']
# timeUtils.amendYear(result[pos_time][1], resultMap) #FIXME do nothing
if (pos_time > 0):
timeUtils.amendHour(result[pos_time - 1][1], resultMap)
timeUtils.amendHour(result[pos_time][1], resultMap)
timeUtils.amendHour(result[pos_time + 1][1], resultMap)
if pos_ling_after > 0 and pos_time > 0 and pos_ling_after < pos_time: # ling and then year,match family
resultMap['type_shop'] = 1 # two shop type
for i in result:
if pos_ling_after > 0 and i <= pos_ling_after:
continue
if pos_time > 0 and i <= pos_time:
continue
sim_pred = str(result[i][1])
if (sim_pred.find('No.') > -1):
staffNoUtils.getNo(sim_pred, resultMap, i)
if (resultMap['pos_staff'] > 0):
break
pos_staff = resultMap['pos_staff']
if pos_staff > 0: # receipt
sim_pred = str(result[pos_staff - 1][1])
staffNoUtils.getReceipt(sim_pred, resultMap,
i) #get from current line same with staff
if pos_staff + 1 < len(result):
sim_pred = str(result[pos_staff + 1][1]).strip()
staffNoUtils.getReceipt(sim_pred, resultMap, i)
for i in result:
sim_pred = str(result[i][1])
if (i > pos_time and i < pos_tax - 3):
categoryUtils.getCategoryAfter(sim_pred, resultMap, i)
cardUtils.getCardPos(resultMap, result)
cardUtils.parseCard(resultMap, result)
resultMap['9_category'] = len(resultMap['suffix_catPrice'])
catTotalMny = 0
taxMny = resultMap['a_tax']
subtotal = resultMap['b_subtotal']
catCount = len(resultMap['suffix_catPrice'])
for price in resultMap['suffix_catPrice']:
catTotalMny += price
if (taxMny > 0):
if (taxMny / 0.07 + 50 < catTotalMny and catCount > 1):
catTotalMny -= resultMap['suffix_catPrice'][-1]
resultMap['9_category'] = len(resultMap['suffix_catPrice']) - 1
# if(resultMap['5_total']==0):
# if(subtotal>0)
else:
if (catCount > 1):
lastMny = resultMap['suffix_catPrice'][-1]
if (catTotalMny - lastMny == lastMny):
catTotalMny = lastMny
resultMap['9_category'] = catCount - 1
if (resultMap['5_total'] < 100):
if (subtotal > 0):
resultMap['5_total'] = subtotal
else:
resultMap['5_total'] = catTotalMny
for i in result: # tel parser at last
sim_pred = result[i][1]
if (resultMap['pos_tel_after'] == 0):
telUtils.getTel(sim_pred, resultMap, i)
if (resultMap['pos_tel_after'] != 0):
break
if pos_time > 0 and i > pos_time:
break
if pos_ling_after > 0 and i > pos_ling_after:
break
if pos_tax > 0 and i > pos_tax:
break
if pos_staff > 0 and i > pos_staff:
break
for i in result: # shop name parser at last
sim_pred = result[i][1]
if (resultMap['1_shopName'] == 'none'):
shopNameUtils.getShopName(sim_pred, resultMap)
if (resultMap['1_shopName'] != 'none'):
break
if pos_time > 0 and i > pos_time:
break
if pos_ling_after > 0 and i > pos_ling_after:
break
if pos_tax > 0 and i > pos_tax:
break
if pos_staff > 0 and i > pos_staff:
break
# city
for i in result: # city parser at last
sim_pred = result[i][1]
if (resultMap['2_city'] == 'none'):
cityUtils.getCity(sim_pred, resultMap)
if (resultMap['2_city'] != 'none'):
break
if pos_ling_after > 0 and i > pos_ling_after:
break
if pos_time > 0 and i > pos_time:
break
if pos_tax > 0 and i > pos_tax:
break
if pos_staff > 0 and i > pos_staff:
break
if (resultMap['1_shopName'] == 'ファミリマート'):
if (resultMap['pos_time'] + 1 == resultMap['pos_staff']):
for i in range(pos_time + 1, pos_time + 4):
sim_pred = str(result[i][1]).strip()
if (sim_pred.find('-') == 1 and len(sim_pred) <= 8):
sim_pred = numberUtils.numberReplacement(sim_pred)
if (sim_pred[0].isdigit()
and sim_pred[0] != resultMap['6_receiptNO'][0]):
resultMap['6_receiptNO'] = sim_pred[0] + resultMap[
'6_receiptNO'][1:]
if (len(sim_pred) <= 8):
staffNoUtils.amendStaff(sim_pred, resultMap, i)
# set default at last--------------vvvvv
if pos_tax>0 and \
(result[pos_tax][1].find('づ')>-1 \
or result[pos_tax][1].find('う')>-1):
resultMap['1_shopName'] = 'サンクス'
if resultMap['1_shopName'] == 'none':
resultMap['1_shopName'] = 'サンクス'
if len(resultMap['2_city'].strip()) == 0:
resultMap['2_city'] = '東京都'
if resultMap['3_tel'] == 'none':
resultMap['test_tel'] = 'unknown_tel'
if len(resultMap['3_tel'].strip()) == 0:
resultMap['3_tel'] = 'none'
resultMap['test_tel'] = 'unknown_tel'
if resultMap['6_receiptNO'] == 'none':
resultMap['test_receiptNO'] = 'unknown_receipt'
if resultMap['7_staffNO'] == 'none':
resultMap['test_staffNO'] = 'unknown_staff'
if len(resultMap['7_staffNO']) == 0:
resultMap['7_staffNO'] = 'none'
if resultMap['8_pointcard'] == 'none':
resultMap['test_Card'] = 'unknown_card'
# --------------------^^^^^^^
for key in sorted(resultMap):
print('{} = {}'.format(key, resultMap[key]))
base_name = im_name.split('/')[-1]
with open('./' + 'result_submit.tsv', 'a+') as f:
f.write(
base_name.split('.')[0] + '.jpg-----------' +
str(datetime.datetime.now().time()) + '\r\n')
resultStr = ''
i = 0
for key in sorted(resultMap):
if (i > 9):
break
resultStr += '{}\r\n'.format(resultMap.get(key))
i += 1
f.write(resultStr)