# ----------------------------------- # Imports des librairies # ----------------------------------- try: from PIL import Image from uuid import uuid4 import json except ImportError: import Image import pytesseract from pytesseract import Output import cv2 import sys import re from pdf2image import convert_from_path from TSLibOCR import * import time import os import numpy as np from TS_OCR import * import MySQLdb from datetime import datetime from collections import Counter """ Retourne le temps passé depuis le lancement du programme """ def getTime(): return "" class Cooperl(TS_OCR): debug = False sqlUser = 'toosmart' sqlHost = '' sqlPasswd = 'i5Hf7MSHKkPN1hy' sqlDb = '' linesPoids = [] linesLots = [] lineMerged = {} dateLivraison = "" typeLivraison = "" enAnomalie = 0 poidsTotal = 0.000 cleCoche = "C[\w\S]*E COUPE EURO [EF]ILIERE LBC" ocrDpi = 170 ocrFolder = '/home/serge/www/toosmart_digicoche/data/tmp/' outputFileName = '' motsASurligner = [] dateTuerie = "" poidsTotalCalcule = 0.000 poidsTotalDetecte = 0 qttCocheCalcule = 0 qttCocheDetecte = 0 nonIntegrable = False idAbattoir = 26 quantieme = 0 def __init__(self, pdf, outputFileName,ocrDirectory,dbName,sqlHost,debug): try: self.debug = debug self.sqlDb = dbName self.sqlHost = sqlHost """ Connexion à la base de données MySQL de Digicoche (Eteko)""" self.conn = MySQLdb.connect(host=self.sqlHost, user=self.sqlUser, passwd=self.sqlPasswd, db=self.sqlDb) self.cursor = self.conn.cursor() self.pdf = pdf self.outputFileName = outputFileName self.ocrFolder = ocrDirectory self.printDebug("Connexion à la base de données OK") self.printDebug("outputFileName = " + str(outputFileName)) except: self.printDebug("Erreur de connexion à la base de données") def printDebug(self,pTxt): self._debugOCR = self._debugOCR + '
' + pTxt if self.debug: print(pTxt) # Analyse du BL de l'abattoir Cooperl def read(self): try: # lecture pagesPdf = convert_from_path(self.pdf, self.ocrDpi) iPage = 1 ocrReturnArray = [] """ On parcours les pages du pdf """ for page in pagesPdf: try: # conversion de l'image en fichier jpeg image_name = "Page_" + str(iPage) + ".jpg" page.save(self.ocrFolder + image_name, "JPEG") # --------------------------------- # Optimisation de l'image img = cv2.imread(self.ocrFolder + image_name) imageAmelioree = img.copy() cv2.imwrite(self.ocrFolder + 'Page_' + str(iPage) + '_result.jpg', imageAmelioree) imageAmelioreeSurlignee = Image.open(self.ocrFolder + 'Page_' + str(iPage) + '_result.jpg') # --------------------------------- # On lance la reconnaissance OCR sur l'image df = pytesseract.image_to_string(np.array(imageAmelioree), lang='fra', config=' --oem 0 --psm 4') ocrReturnArray.append(df) # --------------------------------- iPage = iPage + 1 except Exception as err: self.printDebug(f"Unexpected {err=}, {type(err)=}") self.printDebug("ERROR") try: # Recherche du type de livraison m = re.search(self.cleCoche, ocrReturnArray[0]) if m is not None: self.typeLivraison = m.group() else: # BL Non intégrable car la clef n'est pas correcte # On retourne dans le bash l'information puis on quitte le script # (NE PAS SUPPRIMER CETTE LIGNE !!!) self.printDebug("------") self.printDebug(ocrReturnArray[0]) self.printDebug("------") self.printDebug("Le BL ne comporte pas la bonne clé coche : "+self.cleCoche) print('{"blNonIntegrable":1}') self.nonIntegrable = True return False except Exception as err: self.printDebug(f"Unexpected {err=}, {type(err)=}") self.printDebug("ERROR") # --------------------------------- self.printDebug('1 : ' + ocrReturnArray[0]) # Recherche du numéro du BL try: self.printDebug('1 : '+ocrReturnArray[0]) m = re.findall(r"^([0-9]{6}) [\w]+$", ocrReturnArray[0], re.MULTILINE) if len(m) > 0: for element in m: self._numBl = element else: m = re.findall(r"[A-Z]ON DE LIVRAISON[ ]?N°[\D]+([0-9]{6})", ocrReturnArray[0],re.MULTILINE | re.IGNORECASE) if len(m) > 0: for element in m: self._numBl = element except Exception as err: self._numBl = "" self.printDebug(f"Unexpected1 {err=}, {type(err)=}") self.printDebug("ERROR1") if self._numBl != "": if self.blExist(): # Le BL Existe déjà en base de donneés print('{"blEnDoublon":1}') self.nonIntegrable = True return False # --------------------------------- # Recherche de la date de livraison m = re.findall(r"Date livraison\s+((?:[0-9 ]\s*){1,3}\/\s*(?:[0-9 ]\s*){1,3}\/\s*(?:[0-9 ]\s*){2,3})", ocrReturnArray[0]) if len(m) > 0: self.dateLivraison = str(m[0]).replace(' ','') self.motsASurligner.append(m[0]) #else: # m = re.findall(r"([0-9]{1,2}/[0-9]{1,2}/[0-9]{2})", ocrReturnArray[0]) # if len(m) > 0: # self.dateLivraison = str(m[0]) # self.motsASurligner.append(m[0]) # --------------------------------- # Recherche du poids total try: m = re.findall(r""+self.cleCoche+"\D+(([\d,])*)[| ]*(([\d,])*)", ocrReturnArray[0],re.MULTILINE) if len(m) > 0: for element in m: self.poidsTotalDetecte = element[2] self.qttCocheDetecte = element[0] except Exception as err: self.printDebug(f"Unexpected 1 {err=}, {type(err)=}") self.printDebug("ERROR1") # --------------------------------- # On parcours les pages du BL pour rechercher les données for Sfile in ocrReturnArray: #print(Sfile) # Recherche du nombre de coches, du poids et du lot if self.typeLivraison != "": # On vérifie si le type de livraison du BL est bien autorisé m = re.findall(r"([0-9]{1,3})[. ]{0,2}([0-9]{1,4}) ([0-9]{1,3}[.,][0-9]*)", Sfile) if len(m) > 0: self.linesPoids.extend(m) self.qttCocheCalcule = len(m) for poid in m: # on ajoute le poids détecté au poids total self.poidsTotal = round(self.poidsTotal + round(float(poid[2].replace(",", ".")), 3), 3) self.quantieme = poid[0] # Gestion de l'image pour surlignage des zones tesseract_output = pytesseract.image_to_data(np.array(imageAmelioreeSurlignee),output_type=pytesseract.Output.DICT) try: self.surlignerZones(tesseract_output) except Exception as err: print(f"Unexpected 2 cooperl.py {err=}, {type(err)=}") print("ERROR2") # --------------------------------- except Exception as e: exception_type, exception_object, exception_traceback = sys.exc_info() filename = exception_traceback.tb_frame.f_code.co_filename line_number = exception_traceback.tb_lineno print("Exception type: ", exception_type) print("File name: ", filename) print("Line number: ", line_number) self.printDebug(f"Unexpected {e=}, {type(e)=}") self.printDebug("ERROR") """ Surligne les zones détectées par l'OCR dans le fichier """ def surlignerZones(self, tesseract_output): try: imageASurligner = cv2.imread(self.ocrFolder + 'Page_1_result.jpg') blockASurligner = [] except Exception as err: print(f"Unexpected C {err=}, {type(err)=}") print("ERRORC") try: for i, level_idx in enumerate(tesseract_output['level']): if level_idx == 5: # if tesseract_output['text'][i] in self.motsASurligner: bbox = { 'x': 1 * tesseract_output['left'][i], 'y': 1 * tesseract_output['top'][i], 'width': 1 * tesseract_output['width'][i], 'height': 1 * tesseract_output['height'][i], 'rotation': 0, 'confidence': tesseract_output['conf'][i], 'text': tesseract_output['text'][i] } if int(tesseract_output['conf'][i]) >= 85: color = (42, 219, 151) elif int(tesseract_output['conf'][i]) >= 50: color = (0, 165, 255) else: color = (0, 0, 255) try: overlay = imageASurligner.copy() cv2.rectangle(overlay, (bbox['x'], bbox['y']), (bbox['x'] + bbox['width'], bbox['y'] + bbox['height']), color, cv2.FILLED) alpha = 0.5 imageASurligner = cv2.addWeighted(overlay, alpha, imageASurligner, 1 - alpha, 0) blockASurligner.append(bbox) except Exception as err: print(f"UnexpectedD {err=}, {type(err)=}") print("ERRORD") except Exception as err: print(f"Unexpected D {err=}, {type(err)=}") print("ERRORD") # On sauvegarde le fichier try: imageASurligner = cv2.cvtColor(imageASurligner, cv2.COLOR_BGR2RGB) img = Image.fromarray(imageASurligner, "RGB") img.save(self.ocrFolder+self.outputFileName+"-BL.pdf") #print('Save into : '+self.ocrFolder + 'BL.pdf') except Exception as err: print("erreur") print(f"UnexpectedE {err=}, {type(err)=}") return True def mergeData(self): try: if self.nonIntegrable: return False for linePoids in self.linesPoids: self.lineMerged[linePoids[1]] = {"poids" : linePoids[2], "quantieme" : linePoids[0],"lot":linePoids[1]} try: # On stocke la date de livraison dans le format YYYYMMDD self.dateLivraison = datetime.strptime(self.dateLivraison, "%d/%m/%y").strftime("%Y%m%d") except: self.dateLivraison = "" self.qttCocheCalcule = len(self.lineMerged) try: for key in self.lineMerged: # on ajoute le poids détecté au poids total self.poidsTotalCalcule = round(self.poidsTotalCalcule + round(float(self.lineMerged[key]["poids"].replace(",", ".")), 3), 3) except Exception as err: print(f"Unexpected {err=}, {type(err)=}") print("ERROR A1") # On créé l'entête du BL dans eteko # Date de création de la ligne currentDate = datetime.now().strftime('%Y%m%d%H%M') sql = "INSERT INTO tier_att(id_base, id_type, id_tier,attdatcre, id_work, actif, newobj, att1, att2, att3,att6,att7,att8,att9,attf100,att17,att90) VALUES (1, 10, NULL,'"+str(currentDate)+"', 16, 1, 0, "+str(self.idAbattoir)+", '" + self.typeLivraison + "', '" + str( self.poidsTotalCalcule) + "','" + str(self.poidsTotalDetecte) + "','" + str( self.poidsTotalCalcule) + "',ROUND('" + str(self.qttCocheDetecte) + "'),'" + str( self.qttCocheCalcule) + "', '" + self.dateLivraison + "','"+ self._numBl +"','"+ self._debugOCR.replace("'", "\\'") +"');" try: self.cursor.execute(sql) except Exception as err: print(f"Unexpected {err=}, {type(err)=}") print("ERROR A2 "+ sql) # On récupère l'id du BL créé dans Eteko blId = self.cursor.lastrowid # On retourne dans le bash l'id du BL créé # (NE PAS SUPPRIMER CETTE LIGNE !!!) print('{"blId":' + str(blId) + "}") # ------------------------------------------ if self.quantieme == '': self.quantieme = 0 for key in self.lineMerged: numeroTuerie = str(self.lineMerged[key]["lot"])[-4:] sql = ("INSERT INTO prod_att(id_base, id_type, id_prod, id_work, actif, newobj, fid_prod, type, name, revision, `desc`, att1, att2, att3, att4, attf100, attf101,att10,attf104) VALUES (1, 11, NULL, 11, 1, 0, null, null, null, null, null, '" + str( blId) + "','', '" + str(numeroTuerie) + "', '" + str(self.lineMerged[key]["poids"]) + "', '" + str(self.quantiemeVersDate(int(self.quantieme))) + "', '" + str( self.enAnomalie) + "','" + str(self.quantieme) + "','" + self.dateLivraison + "');") try: self.cursor.execute(sql) except Exception as err: print(f"Unexpected {err=}, {type(err)=}") print("ERROR A3 " + sql) print(sql) self.conn.close() except Exception as e: exception_type, exception_object, exception_traceback = sys.exc_info() filename = exception_traceback.tb_frame.f_code.co_filename line_number = exception_traceback.tb_lineno print("Exception type: ", exception_type) print("File name: ", filename) print("Line number: ", line_number) self.printDebug(f"Unexpected {e=}, {type(e)=}") self.printDebug("ERROR") self.printDebug(sql) def quantiemeVersDate(self,n): try: a = datetime.now().year """Donne la date d=[j,m,a] qui est le nième jour de l'année a""" if ((a%4==0 and a%100!=0) or a%400==0): # bissextile? jm = (0,31,60,91,121,152,182,213,244,274,305,335,366) else: jm = (0,31,59,90,120,151,181,212,243,273,304,334,365) for m in range(1,13): if jm[m]>=n: #return str(n)+str(jm[m-1])+ str(m)+ str(a) return str(a) + str("{:02d}".format(m)) + str("{:02d}".format(n-jm[m-1])) except Exception as e: exception_type, exception_object, exception_traceback = sys.exc_info() filename = exception_traceback.tb_frame.f_code.co_filename line_number = exception_traceback.tb_lineno print("Exception type: ", exception_type) print("File name: ", filename) print("Line number: ", line_number) print("n: ", n) self.printDebug(f"Unexpected {e=}, {type(e)=}") self.printDebug("ERROR")