Upgrade Tesseract to 5.0

src/CNIRevelator.py

@@ -54,8 +54,8 @@ def main():
     mainw = mainWindow()
 
     try:
-        os.environ['PATH'] = globs.CNIRFolder + '\\Tesseract-OCR4\\'
+        os.environ['PATH'] = globs.CNIRFolder + '\\Tesseract-OCR5\\'
-        os.environ['TESSDATA_PREFIX'] = globs.CNIRFolder + '\\Tesseract-OCR4\\tessdata'
+        os.environ['TESSDATA_PREFIX'] = globs.CNIRFolder + '\\Tesseract-OCR5\\tessdata'
         tesser_version = pytesseract.get_tesseract_version()
     except Exception as e:
         logfile.printerr('ERROR WITH TESSERACT MODULE ' + str(e))

src/critical.py

@@ -44,11 +44,11 @@ def crashCNIR(shutdown=True):
         root.withdraw()
         logfile = logger.logCur
         logfile.printerr("FATAL ERROR : see traceback below.\n{}".format(traceback.format_exc()))
-        showerror(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["CNIRevelator crashed because a fatal error occured. View log for more infos and please open an issue on Github"], parent=root)
+        showerror(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["CNIRevelator crashed because a fatal error occured. View log for more infos and please open an issue on Github"])
-        res = askquestion(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["Would you like to open the log file ?"], parent=root)
+        res = askquestion(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["Would you like to open the log file ?"])
         if res == "yes":
             webbrowser.open_new(globs.CNIRErrLog)
-        res = askquestion(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["Would you like to open an issue on Github to report this bug ?"], parent=root)
+        res = askquestion(lang.all[globs.CNIRlang]["CNIRevelator Fatal Eror"], lang.all[globs.CNIRlang]["Would you like to open an issue on Github to report this bug ?"])
         if res == "yes":
             webbrowser.open_new("https://github.com/neox95/CNIRevelator/issues")
         root.destroy()

src/globs.py

@@ -26,7 +26,7 @@ import os
 
 # CNIRevelator version
 verType             = "final release"
-version             = [3, 1, 1]
+version             = [3, 1, 2]
 verstring_full      = "{}.{}.{} {}".format(version[0], version[1], version[2], verType)
 verstring           = "{}.{}".format(version[0], version[1])
 
@@ -38,7 +38,7 @@ CNIRCryptoKey       = '82Xh!efX3#@P~2eG'
 CNIRNewVersion      = False
 
 CNIRConfig          = CNIRFolder + '\\config\\conf.ig'
-CNIRTesser          = CNIRFolder + '\\Tesseract-OCR4\\'
+CNIRTesser          = CNIRFolder + '\\Tesseract-OCR5\\'
 CNIRErrLog          = CNIRFolder + '\\logs\\error.log'
 CNIRMainLog         = CNIRFolder + '\\logs\\main.log'
 CNIRUrlConfig       = CNIRFolder + '\\config\\urlconf.ig'
@@ -56,7 +56,7 @@ CNIRLangFile        = CNIRFolder + '\\config\\lang.ig'
 CNIRlang            = "fr"
 
 CNIRConfig          = CNIRFolder + '\\config\\conf.ig'
-CNIRTesser          = CNIRFolder + '\\Tesseract-OCR4\\'
+CNIRTesser          = CNIRFolder + '\\Tesseract-OCR5\\'
 CNIRErrLog          = CNIRFolder + '\\logs\\error.log'
 CNIRMainLog         = CNIRFolder + '\\logs\\main.log'
 CNIRUrlConfig       = CNIRFolder + '\\config\\urlconf.ig'

src/ihm.py

@@ -36,7 +36,6 @@ import critical             # critical.py
 import logger               # logger.py
 import globs                # globs.py
 import lang                 # lang.py
-
 import updater              # updater.py
 import critical             # critical.py
 

src/lang.py

@@ -138,7 +138,8 @@ french = \
                                         "Coller :\t\t\t\tCtrl-V \n"
                                         "Forcer une nouvelle détection du document :\tEchap\n",
 
-"CHANGELOG"                         :   "Version 3.1.1 \nMise-à-jour mineure avec les progressions suivantes :\n- Correction d'un bug sévère du système de mise à jour\n\n" + \
+"CHANGELOG"                         :   "Version 3.1.2 \nMise-à-jour mineure avec les progressions suivantes :\n- Montée de version de Tesseract OCR : 5.0\n\n" + \
+"Version 3.1.1 \nMise-à-jour mineure avec les progressions suivantes :\n- Correction d'un bug sévère du système de mise à jour\n\n" + \
 "Version 3.1.0 \nMise-à-jour majeure avec les progressions suivantes :\n- Modifications cosmétiques de l'interface utilisateur\n- Stabilisation des changements effectués sur la version mineure 3.0 : interface utilisateur, OCR, VISA A et B, logging\n- Rationalisation du système de langues\n- Ajout des canaux de mise-à-jour\n\n" + \
 "Version 3.0.8 finale\nCorrectif : bug du système de mise-à-jour'\n\n" + \
 "Version 3.0.6 \nMise-à-jour mineure avec les corrections suivantes :\n- Changement de l'apparence du launcher de l'application\n- Améliorations de l'interface, notamment de la stabilité\n- Ajout de la signature numérique de l'exécutable\n\n" + \
@@ -189,6 +190,8 @@ french = \
 "Pièce d'identité/de voyage"        :   "Pièce d'identité/de voyage",
 "Pièce d'identité FR"               :   "Pièce d'identité FR",
 "Permis de conduire"                :   "Permis de conduire",
+"The file you provided is not "
+"valid : {}"                        :   "Le fichier transmis n'est pas valide : {}",
 
 "LANDCODE2"                         :   {
                                             'AW': 'Aruba',
@@ -827,7 +830,8 @@ english = \
                                         "Paste:\t\t\t\tCtrl-V\n"
                                         "Force a new document detection:\tEchap\n",
 
-"CHANGELOG"                         :   "Version 3.1.1 \nMinor update with the following progressions: \n- Fixed a severe bug in the update system" + \
+"CHANGELOG"                         :   "Version 3.1.2 \nMinor update with the following progressions:\n- Upgrade Tesseract OCR to 5.0\n\n" + \
+"Version 3.1.1 \nMinor update with the following progressions: \n- Fixed a severe bug in the update system\n\n" + \
 "Version 3.1.0 \nMajor update with the following progressions: \n- Cosmetic modifications of the user interface \n- Stabilization of the changes made on the minor version 3.0 : user interface, OCR, VISA A and B, logging\n- Rationalization of the language system\n- Added update channels\n\n" + \
 "Version 3.0.8 final\nCorrection: bug in the update system'\n\n" + \
 "Version 3.0.6 \nMinor update with the following fixes:\n- Change in the appearance of the application launcher\n- Improvements to the interface, including stability\n- Added digital signature of the executable\n" + \
@@ -879,6 +883,8 @@ english = \
 "Pièce d'identité/de voyage"        :   "Identity/travel document",
 "Pièce d'identité FR"               :   "French Identity card",
 "Permis de conduire"                :   "Driver License",
+"The file you provided is not "
+"valid : {}"                        :   "The file you provided is not valid : {}",
 
 "LANDCODE2"                         :   {
                                             "AW": "Aruba",

src/main.py

@@ -451,7 +451,7 @@ class mainWindow(Tk):
                 os.environ['PATH'] = globs.CNIRTesser
                 os.environ['TESSDATA_PREFIX'] =  globs.CNIRTesser + '\\tessdata'
                 
-                text = pytesseract.image_to_string(crop_img, lang='ocrb', boxes=False, config='--psm 6 --oem 0 -c tessedit_char_whitelist=ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890<')
+                text = pytesseract.image_to_string(crop_img, lang='ocrb', config='--psm 6 --oem 1 -c tessedit_char_whitelist=ABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890<')
                 
                 # manual validation
                 # the regex

src/updater.py

@@ -171,18 +171,18 @@ def tessInstall(PATH, credentials):
      
     
     # Verifying that Tesseract is installed
-    if not os.path.exists(PATH + '\\Tesseract-OCR4\\'):
+    if not os.path.exists(PATH + '\\Tesseract-OCR5\\'):
         finalver, finalurl, finalchecksum = getLatestVersion(credentials)
         
         if finalurl == None:
             logfile.printerr('Unable to get the Tesseract url')
             return False
         
-        tesseracturl = finalurl.replace("CNIRevelator.zip", "tesseract_4.zip")
+        tesseracturl = finalurl.replace("CNIRevelator.zip", "tesseract_5.zip")
         
-        # WE ASSUME THAT THE MAIN FILE IS CNIRevelator.zip AND THAT THE TESSERACT PACKAGE IS tesseract_4.zip
+        # WE ASSUME THAT THE MAIN FILE IS CNIRevelator.zip AND THAT THE TESSERACT PACKAGE IS tesseract_5.zip
         logfile.printdbg('Preparing download of Tesseract OCR 4...')
-        getTesseract = downloader.newdownload(credentials, tesseracturl, PATH + '\\downloads\\TsrtPackage.zip', "Tesseract 4 OCR Module").download()
+        getTesseract = downloader.newdownload(credentials, tesseracturl, PATH + '\\downloads\\TsrtPackage.zip', "Tesseract 5 OCR Module").download()
         
         try:
             # CHECKSUM