def setUpClass(self):
""" Setup a trainer with defaults arguments """
self.trainer = TesseractTrainer(dictionary_name='test',
text='text',
font_name='helveticanarrow',
font_path='./font/Helvetica-Narrow.otf',
verbose=False)
self.prefix = '%s.%s.exp%d' % (self.trainer.dictionary_name, self.trainer.font_name, self.trainer.exp_number)
def setUpClass(self):
""" Setup a trainer with defaults arguments """
self.trainer = TesseractTrainer(dictionary_name='test',
text='text',
font_name='helveticanarrow',
font_path='./font/Helvetica-Narrow.otf',
font_size=df.FONT_SIZE,
exp_number=df.EXP_NUMBER,
font_properties=df.FONT_PROPERTIES,
tessdata_path=df.TESSDATA_PATH,
word_list=df.WORD_LIST,
verbose=False)
self.prefix = '%s.%s.exp%d' % (self.trainer.dictionary_name, self.trainer.font_name, self.trainer.exp_number)
class TesseractTrainerTest(unittest.TestCase):
GENERATED_DURING_TRAINING = ['unicharset', 'pffmtable', 'Microfeat', 'inttemp', 'normproto', 'mfunicharset']
@classmethod
def setUpClass(self):
""" Setup a trainer with defaults arguments """
self.trainer = TesseractTrainer(dictionary_name='test',
text='text',
font_name='helveticanarrow',
font_path='./font/Helvetica-Narrow.otf',
font_size=df.FONT_SIZE,
exp_number=df.EXP_NUMBER,
font_properties=df.FONT_PROPERTIES,
tessdata_path=df.TESSDATA_PATH,
word_list=df.WORD_LIST,
verbose=False)
self.prefix = '%s.%s.exp%d' % (self.trainer.dictionary_name, self.trainer.font_name, self.trainer.exp_number)
@classmethod
def tearDownClass(self):
""" Delete all temporary files created during tests """
tempfiles = glob.glob(self.trainer.dictionary_name + '.' + self.trainer.font_name + '*')
for tempfile in tempfiles:
os.remove(tempfile)
os.remove(self.trainer.dictionary_name + '.traineddata')
def assertFileExists(self, f):
try:
open(f)
except IOError:
raise AssertionError('The file %s does not exist.' % (f))
def assertFileDoesNotExist(self, f):
try:
open(f)
except IOError:
pass
else:
raise AssertionError('The file %s does exist.' % (f))
def test1_generate_boxfile(self):
""" Test if the tif and box files are correctly created after executing the self.trainer._generate_boxfile() method. """
self.trainer._generate_boxfile()
self.assertFileExists(self.prefix + '.tif')
self.assertFileExists(self.prefix + '.box')
def test2_train_on_boxfile(self):
""" Test if the tr file is correctly created after executing the self.trainer._train_on_boxfile() method. """
self.trainer._train_on_boxfile()
self.assertFileExists(self.prefix + '.tr')
def test3_compute_character_set(self):
""" Test if the unicharset file is correctly created after executing the self.trainer._compute_character_set() method. """
self.trainer._compute_character_set()
self.assertFileExists('unicharset')
def test4_clustering(self):
""" Test if the mfunicharset, inttemp, Microfeat and pffmtable files are correctly created
after executing the self.trainer._clustering() method.
"""
self.trainer._clustering()
self.assertFileExists('mfunicharset')
self.assertFileExists('inttemp')
self.assertFileExists('Microfeat')
self.assertFileExists('pffmtable')
def test5_normalize(self):
""" Test if the normproto file is correctly created after executing the self.trainer._normalize() method. """
self.trainer._normalize()
self.assertFileExists('normproto')
def test6_rename_files(self):
""" Check if all generated files were renamed to 'self.prefix'.old_name
after executing the self.trainer._rename_files() method.
"""
self.trainer._rename_files()
for filename in self.GENERATED_DURING_TRAINING:
if filename not in "mfunicharset": # mfunicharset does not need to be renamed
self.assertFileExists(self.trainer.dictionary_name + '.' + filename)
self.assertFileDoesNotExist(filename)
def test7_combine_data(self):
""" Test if the traineddata file is correctly created after executing the self.combine_data() method. """
self.trainer._combine_data()
self.assertFileExists(self.trainer.dictionary_name + '.traineddata')
def test8_clean(self):
""" Test if the all generated files were removed after executing the self.trainer.clean() method. """
self.trainer.clean()
for filename in self.GENERATED_DURING_TRAINING:
if filename not in "mfunicharset": # mfunicharset does not need to be renamed
self.assertFileDoesNotExist(self.trainer.dictionary_name + '.' + filename)
else:
self.assertFileDoesNotExist(filename)
parser.add_argument('--experience_number', '-e', type=int, action='store', default=df.EXP_NUMBER,
help="The number of the training experience.")
parser.add_argument('--font-properties', '-f', type=str, action='store', default=df.FONT_PROPERTIES,
help="The path of a file containing font properties for a list of training fonts.")
parser.add_argument('--font-size', '-s', type=int, action='store', default=df.FONT_SIZE,
help="The font size of the training font, in px.")
parser.add_argument('--tessdata-path', '-p', type=str, action='store', default=df.TESSDATA_PATH,
help="The path of the tessdata/ directory on your filesystem.")
parser.add_argument('--word_list', '-w', type=str, action='store', default=df.WORD_LIST,
help="The path of a file containing a list of frequent words.")
parser.add_argument('--verbose', '-v', action='store_true',
help="Use this argument if you want to display the training output.")
args = parser.parse_args()
perform_security_checks(args) # Check validity of args
# Training process
trainer = TesseractTrainer(dictionary_name=args.tesseract_lang,
text=args.training_text,
font_name=args.font_name,
font_path=args.font_path,
font_size=args.font_size,
exp_number=args.experience_number,
font_properties=args.font_properties,
tessdata_path=args.tessdata_path,
word_list=args.word_list,
verbose=args.verbose)
trainer.training() # generate a multipage tif from args.training_text, train on it and generate a traineddata file
trainer.clean() # remove all files generated in the training process (except the traineddata file)
trainer.add_trained_data() # copy the traineddata file to the tessdata/ directory
