Working Writing

Tmp/box/1.txt

-Tmp
\ No newline at end of file
+187,90,460,90,460,139,187,139 MIGROS
+188,163,444,163,444,179,188,179 SOCIETE COOPERATIVE MIGROS GENDVE
+260,182,375,182,375,199,260,199 N CORNAVIN GARE
+253,203,391,203,391,219,253,219 TEL. 058 573 43 20
+246,225,391,225,391,239,246,239 WWW. NIGROSGENEVE. CH
+101,255,237,255,237,269,101,269 DÉSIGNATIOA C
+118,254,535,254,535,270,118,270 SIGNATION ARTICLE QUANT. PRIX RABAIS TOTAL #
+101,287,230,287,230,303,101,303 \IRAP POULET PANE
+296,284,498,284,498,301,296,301 1 5 55
+102,280,507,280,507,329,102,329 PON BAR ORIGINAL 1009 1 16S 2 ‘3
+102,306,510,306,510,342,102,342 DOCE-COLA 501 IEEE (50 1150
+104,353,529,353,529,396,104,396 TOTAL CHE ES £5
+105,392,521,392,521,425,105,425 TWIN) QR AE
+106,427,556,427,556,445,106,445 TOTAL EN EUR SE
+107,446,450,446,450,477,107,477 OC! OO UN
+210,462,256,462,256,518,210,518 NMIN
+211,487,453,487,453,517,211,517 75.04 17:52
+213,498,454,498,454,537,213,537 2 OY E00568176/308156/0000
+425,548,457,548,457,564,425,564 9.65
+379,576,535,576,535,620,379,620 CHE-105. 842. 900 “
+380,612,417,612,417,628,380,628 TOTAL
+199,673,486,673,486,700,199,700 H-CUNULUS VOTRE FIDÉLITE RÉCONPENSÉE
+256,721,368,721,368,736,256,736 MERCI DE VOTRE
+235,734,455,734,455,754,235,754 ET AU PLAISIR DE VOUS REVOIR
+303,766,372,766,372,786,303,786 $65
+225,855,459,855,459,907,225,907 ANOOOKANZZZS¥Z4SST0000
+154,877,493,877,493,922,154,922 JLIALE UTILIS. ENP TICK. HEURE
+154,891,519,891,519,943,154,943 PIED 0537999 282 0537 B. M. 2024 17:51:59

Tmp/box/2.txt

-Tmp
\ No newline at end of file
+216,104,426,104,426,138,216,138 TANIGAMI
+217,145,263,145,263,156,217,156 RUE RO
+235,147,427,147,427,175,235,175 : I SEAU 14. 1201 GENEVE
+132,167,396,167,396,187,132,187 AE EL: 441 22 741 03 33
+128,176,517,176,517,207,128,207 CHE-115.261.972 “OWT
+445,187,486,187,486,199,445,199 NN 60
+211,210,216,210,216,220,211,220 L
+218,211,443,211,443,225,218,225 @ MERCREDI, 05 JUIN 2024 & 17:28:21
+223,241,436,241,436,258,223,258 VOTRE TOTAL DE POINTS EST DE : 5936
+133,280,207,280,207,293,133,293 QT ARTICLE
+165,308,379,308,379,320,165,320 POKEMON - TERRARIUM - SÉRIE 13
+166,326,432,326,432,342,166,342 ONE PIECE - TCG CARD - DRESSROSA (30)
+167,346,442,346,442,364,167,364 ONE PIECE - TCG CARD - CAKE ISLAND (18)
+168,367,448,367,448,386,168,386 ONE PIECE - TCG CARD - FISHERMAN'S ISLAN
+140,388,426,388,426,408,140,408 1 ONE PIECE - TCG CARD - SKYPEA (36)
+141,434,273,434,273,448,141,448 MONTANT TOTAL CHF
+143,460,229,460,229,473,143,473 MONTANT T/A
+143,504,178,504,178,516,143,516 TWINT
+145,521,543,521,543,556,145,556 REMBOURSEMENT EN BON CADEAU, SOUS UN DÉLAI DE 7 JOURS
+313,549,382,549,382,562,313,562 OUVRABLES.
+309,555,531,555,531,580,309,580 CTION SONT NI REPRIS NI ECHANGES:
+312,576,516,576,516,598,312,598 RUNE COMMANDE, NE SONT PAS
+390,593,495,593,495,610,390,610 LATION DU CLIENT.
+161,573,307,573,307,591,161,591 LES ARTICLES EN PROM
+179,590,310,590,310,608,179,608 LES ACOMPTES POU
+204,601,389,601,389,624,204,624 REMBOURSÉS EN CAS D'ANNUL
+299,638,409,638,409,657,299,657 EMPLOYÉ : JOHANN
+225,658,486,658,486,690,225,690 NOUS VOUS REMERCIONS DE VOTRE VISITE
+264,696,449,696,449,722,264,722 ‘VANTAGES CARTE DE FIDELITÉ

Tmp/box/3.txt

-Tmp
\ No newline at end of file
+257,7,476,7,476,50,257,50 MANOR*
+227,77,303,77,303,89,227,89 WIT. TRANOR CH
+227,92,347,92,347,106,227,106 RUE SAINT-LAURENT 7
+228,109,308,109,308,123,228,123 1002 LAUSANNE
+227,126,308,126,308,139,227,139 021 321 36 99
+458,163,464,163,464,170,458,170 I
+457,171,469,171,469,182,457,182 FA
+223,213,478,213,478,229,223,229 YOUS ÉTES SERVI LE: 28.05.2024 14:46
+220,230,382,230,382,243,220,243 ‘A. ALBUQUERQUE DE OLIVEIRA
+219,255,476,255,476,274,219,274 FIL A TRICOTER 2.55 A
+232,276,292,276,292,287,232,287 40556080512
+221,290,475,290,475,306,221,306 FAT DE TRAVAUX MANUELS 1.95 4
+234,309,294,309,294,319,234,319 SB 581658
+223,323,474,323,474,338,223,338 FIL DE TRAVAUX MANUELS 1,95 &
+235,341,295,341,295,351,235,351 405081551885
+224,355,474,355,474,369,224,369 FIL A TRICOTER 7.95 8
+237,372,295,372,295,381,237,381 DOOSETSA00
+225,387,473,387,473,399,225,399 FIL A TRICOTER 7.95 8
+237,403,296,403,296,412,237,412 ‘OSSIGSAI6
+225,418,472,418,472,430,225,430 CROCHET BOUCLE D'OREILLES 2.50 8
+237,434,297,434,297,442,237,442 OSR771092836
+225,449,472,449,472,460,225,460 FIL D'OREILLE 2.95
+237,464,297,464,297,472,237,472 DOSRCRIOSEER2
+213,504,459,504,459,531,213,531 TOTAL ARTICLE: 7 27.80
+214,528,459,528,459,559,214,559 TWINT 21.80
+222,566,458,566,458,578,222,578 TOTAL MOYENS DE PAIEMENT 27.80
+225,603,465,603,465,624,225,624 SOLDE DE POINTS AVANT ACHAT ON
+225,618,364,618,364,629,225,629 NOUVEAU SOLDE DE POINTS:
+227,639,467,639,467,653,227,653 TWINT. XIXIOOG GORA] 183496 31503669
+287,665,413,665,413,677,287,677 HERC! POUR VOTRE ACHAT
+230,688,312,688,312,696,230,696 -16.261.650 TUR
+230,699,401,699,401,708,230,708 ADE MONTANE TAU 8
+219,709,400,709,400,719,219,719 JA 7780 WW
+232,720,400,720,400,730,232,730 ANOR A SAS) IS
+291,785,405,785,405,800,291,800 RETOUR AVEC EABAL LAGE
+275,799,420,799,420,814,275,814 ORIGINE ET PREWE D-ACHAT
+299,813,396,813,396,827,299,827 DNS TES 90 JOURS.
+251,824,417,824,417,841,251,841 XCLUS DES PRETATION DOS ART
+235,837,433,837,433,857,235,857 MH SERVI OT TH CAR TAS FEATTRUTS AC
+254,851,462,851,462,870,254,870 ME VOUS DRE GTOLOF OK
+259,864,463,864,463,885,259,885 = OH RABATS SUR VOTRE PROCHAIN AN
+323,877,454,877,454,891,323,891 EN TOW, A
+278,890,417,890,417,908,278,908 SEANNEZ OT ISTRIVER-VOUS;

Tmp/box/4.txt

-Tmp
\ No newline at end of file
+305,98,416,98,416,124,305,124 PRONTO
+150,356,221,356,221,368,150,368 ‘SOUS-TOTAL
+208,377,373,377,373,389,208,389 HT TVA, TE.
+147,389,389,389,389,405,147,405 2,60 % 10,04 0,26 10,30
+143,419,161,419,161,431,143,431 COA
+144,433,289,433,289,446,144,446 CHE 060 TVA
+314,449,382,449,382,460,314,460 GARE GMBH
+237,449,266,449,266,461,237,461 HO MI
+245,488,450,488,450,517,245,517 CHF 10,30
+383,527,467,527,467,541,383,541 CHF 10,30
+293,544,311,544,311,552,293,552 HE

Tmp/fileConverter.py

-from PIL import Image
 import pytesseract
 from pytesseract import Output
 import cv2
 import numpy as np
+import os
+from pathlib import Path
 
-# Transforming image to string and printing it!
-#img = cv2.imread("X00016469670.jpg")
-img = cv2.imread("img/3.jpg")
 
 # get grayscale image
 def get_grayscale(image):
@@ -39,35 +37,58 @@ def canny(image):
     return cv2.Canny(image, 100, 200)
 
 
-gray = get_grayscale(img)
-thresh = thresholding(gray)
-opend = opening(gray)
-end = closing(opend)
+def write_box(filename) -> str:
+    img = cv2.imread(filename)
+    d = pytesseract.image_to_data(img, output_type=Output.DICT)
+    n_boxes = len(d['level'])
+    res = ""
+    z = ""
+    (x, y, w, h) = (0, 0, 0, 0)
+    for i in range(n_boxes):
+        if d['conf'][i] == -1:
+            if d['line_num'][i] > 0:
+                if not z == "":
+                    res += z + "\n"
+                    z = ""
+                (x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
+                z += str(x) + "," + str(y) + "," + str(x + w) + "," + str(y) + "," + str(x + w) + "," + str(y + h) \
+                     + "," + str(x) + "," + str(y + h)
+            cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
+        elif d['conf'][i] != -1:
+            if not str(d['text'][i]).isspace():
+                z += " " + str(d["text"][i]).upper()
+            else:
+                z = ""
+    return res
+
+
+#gray = get_grayscale(img)
+#thresh = thresholding(gray)
+#opend = opening(gray)
+#end = closing(opend)
 #faire closing
-canny = canny(gray)
+#canny = canny(gray)
+
+# Nom du dossier à ouvrir
+dossier = 'img'
+
+# Obtenir le chemin absolu du dossier
+chemin_dossier = os.path.join(os.path.dirname(__file__), dossier)
+
+# Vérifier si le dossier existe
+if os.path.exists(chemin_dossier):
+    # Lire les noms de tous les fichiers dans le dossier
+    fichiers = os.listdir(chemin_dossier)
+
+    # Afficher les noms des fichiers
+    for fichier in fichiers:
+        if os.path.isfile(os.path.join(chemin_dossier, fichier)):
+            fileName = os.path.splitext(fichier)[0]
+            filePath = dossier + "/" + fichier
+            with open(Path("entities") / f"{fileName}.txt", "w+", encoding="utf8") as file_entities, \
+                    open(Path("box") / f"{fileName}.txt", "w+", encoding="utf8") as file_box:
+                file_entities.write('{\n\t"company": "",\n\t"date": "",\n\t"address": "",\n\t"total": ""\n}')
+                file_box.write(write_box(filePath))
+        else:
+            print(f"Le dossier '{dossier}' n'existe pas.")
 
-#expected_keys = ['level', 'page_num', 'block_num', 'par_num', 'line_num', 'word_num',    'left', 'top', 'width', 'height', 'conf', 'text']
-
-d = pytesseract.image_to_data(img, output_type=Output.DICT)
-n_boxes = len(d['level'])
-
-z = ""
-for i in range(n_boxes):
-    (x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i], d['height'][i])
-    if d['word_num'][i] != 0:
-        z += d['text'][i] + "(" + str(d["conf"][i]) + ")  "
-        cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
-    else:
-        if z != "":
-            print(z)
-        z = ""
-    #cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
-cv2.imshow('img', img)
-cv2.waitKey(0)
-
-
-#img2 = Image.open('X00016469670.jpg').convert('L')
-#img2.show()
-
-#print(pytesseract.image_to_data(img))
-#print(pytesseract.image_to_string(img2))