def ocr(self):
def is_rect(column):
column = self.model.headerData(c, QtCore.Qt.Horizontal)
if column[:4] == 'rect' or column[:4] == 'crop':
try:
temp = int(column[4:])
return True
except:
return False
return False
columns = list(range(self.model.columnCount()))[::-1]
for c in columns:
column = self.model.headerData(c, QtCore.Qt.Horizontal)
if not is_rect(column):
continue
column_text = column + '_text'
if not self.model.headerData(c + 1,
QtCore.Qt.Horizontal) == column_text:
self.model.insertColumn(c + 1)
self.model.setHeaderData(c + 1, QtCore.Qt.Horizontal,
column_text)
children = self.model.root().children()
child_count = len(children)
column_count = self.model.columnCount()
loop_count = child_count * column_count
self.progress_bar.setMaximum(loop_count)
self.progress_bar.show()
for r, item in enumerate(children):
for c in range(column_count):
print(r, c, c + column_count * r, loop_count)
self.progress_bar.setValue(c + column_count * r)
QtWidgets.QApplication.processEvents()
column = self.model.headerData(c, QtCore.Qt.Horizontal)
if not is_rect(column):
continue
rect = item.data(column)
if rect is None:
continue
item.data(column + '_text', Tesseract().OCR(rect).strip())
self.progress_bar.hide()
def __init__(self, proxy=None):
self.PAGETRY, rc = startup.CFG.value('pagetry', Avito.PAGETRY).toInt()
self.SLEEP, rc = startup.CFG.value('sleep', Avito.SLEEP).toInt()
self.g = Grab()
self.g.setup(hammer_mode=True, hammer_timeouts=((2, 5), (5, 10)))
if __name__ == '__main__':
self.g.setup(log_dir='dump')
if proxy:
self.g.load_proxylist(proxy,
'text_file',
'http',
auto_init=True,
auto_change=False)
self.tess = Tesseract()
self.tess.set_pagemode(TessPageSegMode.PSM_SINGLE_LINE)
def ocr(self):
for item in self.model.root().children():
i = 0
while i < 1000:
rect = item.data('Rect' + str(i))
if rect is None:
break
item.data('Text' + str(i), Tesseract().OCR(rect).strip())
i = i + 1
columns = list(range(self.model.columnCount()))[::-1]
for c in columns:
column = self.model.headerData(c, QtCore.Qt.Horizontal)
if not 'Rect' in column:
continue
replace = column.replace('Rect', 'Text')
if self.model.headerData(c + 1, QtCore.Qt.Horizontal) == replace:
continue
self.model.insertColumn(c + 1)
self.model.setHeaderData(c + 1, QtCore.Qt.Horizontal, replace)
def __init__(self, url):
""""""
HTMLParser.__init__(self)
self.link = None
self.located = False
while not self.link:
p = CaptchaParser(URLOpen().open(url).read())
if p.captcha:
handle = URLOpen().open(p.captcha)
if handle.info()["Content-Type"] == "image/gif":
self.tess = Tesseract(handle.read())
captcha = self.get_captcha()
if captcha:
handle = URLOpen().open(
url,
urllib.urlencode([(CAPTCHACODE, p.captchacode),
(MEGAVAR, p.megavar),
("captcha", captcha)]))
self.reset()
self.feed(handle.read())
self.close()
logger.info("Captcha %s: %s" % (p.captcha, captcha))
def build_candidates(self, characters4_pixels_list, uncertain_pixels):
"""Build word candidates from characters and uncertains groups."""
for plindex, characters4_pixels in enumerate(characters4_pixels_list):
logging.debug("Generating words (%d) %d/%d" % (2**len(uncertain_pixels), plindex+1, len(characters4_pixels_list)))
for length in range(len(uncertain_pixels)+1):
for groups in combinations_no_repetition(uncertain_pixels, length):
characters4_pixels_test = [x.copy() for x in characters4_pixels]
for pixels in groups:
pair = get_pair_inclussion(characters4_pixels_test, center_of_mass(pixels)[0], pred=lambda x: center_of_mass(x)[0])
if not pair:
continue
char1, char2 = pair
char1.update(pixels)
char2.update(pixels)
images = [self.rotate_character(pixels, cindex) for cindex, pixels in enumerate(characters4_pixels_test)]
clean_image = smooth(join_images_horizontal(images), 0)
ocr = Tesseract(self.data, lambda x: clean_image)
text = ocr.get_captcha().strip()
filtered_text = filter_word(text)
if filtered_text:
yield filtered_text
""""""
if tag == "form":
if ((len(attrs) == 3) and (attrs[2][1] == "formDownload")):
self.form_action = attrs[0][1]
if __name__ == "__main__":
urllib2.install_opener(
urllib2.build_opener(urllib2.HTTPCookieProcessor(
cookielib.CookieJar())))
urllib2.urlopen(
urllib2.Request(
"http://www.gigasize.com/get.php/3196987695/p3x03sp.avi"))
tes = Tesseract(
urllib2.urlopen(
urllib2.Request("http://www.gigasize.com/randomImage.php")).read(),
True)
captcha = tes.get_captcha(3)
data = urllib.urlencode({
"txtNumber": captcha,
"btnLogin.x": "124",
"btnLogin.y": "12",
"btnLogin": "******"
})
handle = urllib2.urlopen(
urllib2.Request("http://www.gigasize.com/formdownload.php"), data)
f = FormParser(handle.read())
handle.close()
if f.form_action:
timer = 60
from pathlib import Path
from subprocess import PIPE, CalledProcessError, Popen, check_output
from tempfile import TemporaryDirectory
from cv2 import cv2
from pdf2image import convert_from_path
from tesseract import Tesseract, PageSegMode
__all__ = [
"pdf_to_text",
"ocr_to_text",
"get_page_count",
]
TESS = Tesseract()
def pdf_to_text(pdf_path: str, target_dir: str):
"""
Convert pdf at `pdf_path` to a txt file in `target_dir` using XpdfReader's pdftotext.
"""
file_name = Path(pdf_path).stem
command = [
"pdftotext",
"-layout",
pdf_path,
str(Path(target_dir) / f"{file_name}.txt"),
]
proc = Popen(command, stdout=PIPE, stderr=PIPE)
proc.wait()
def test_tesseract_exists(self):
self.assertEqual(Tesseract().run().returncode, 0)
