def get_raw_captcha(url, n, outdir=os.curdir, session=None):
for i in range(n):
captcha_name = add_postfix('captcha', str(i))
get_image(url,
captcha_name=captcha_name,
outdir=outdir,
session=session)
def get_image_test():
from minghu6.graphic.captcha.get_image import get_image
session = requests.Session()
imgObj = get_image('http://zyzfw.xidian.edu.cn/site/captcha?v=5832eb2835057', session=session)[0]
imgObj.show()
imgObj = get_image('http://zyzfw.xidian.edu.cn/site/captcha?v=5832eb2835057', session=session)[0]
imgObj.show()
def get_image_test():
from minghu6.graphic.captcha.get_image import get_image
session = requests.Session()
imgObj = get_image(
'http://zyzfw.xidian.edu.cn/site/captcha?v=5832eb2835057',
session=session)[0]
imgObj.show()
imgObj = get_image(
'http://zyzfw.xidian.edu.cn/site/captcha?v=5832eb2835057',
session=session)[0]
imgObj.show()
def main_split(path, num=None, split_method='bisect', outdir=os.path.curdir):
# path, n=None, split_method='bisect', outdir=os.path.curdir
assert split_method in split_method_dict, 'split_method do not exist!'
assert num is not None or split_method != 'bisect', 'bisect method need the param n'
split_method = split_method_dict[split_method]
imgObj, image_path = get_image(path)
box_img = split_method(imgObj, n=num)
# base_path = os.path.join(outdir, os.path.basename(path))
for i, sub_img in enumerate(box_img):
sub_img_path = add_postfix(image_path, '{0:d}'.format(i))
sub_img.save(sub_img_path)
def main_split(path, num=None, split_method='bisect', outdir=os.path.curdir):
# path, n=None, split_method='bisect', outdir=os.path.curdir
assert split_method in split_method_dict, 'split_method do not exist!'
assert num is not None or split_method != 'bisect', 'bisect method need the param n'
split_method = split_method_dict[split_method]
imgObj, image_path = get_image(path)
box_img = split_method(imgObj, n=num)
# base_path = os.path.join(outdir, os.path.basename(path))
for i, sub_img in enumerate(box_img):
sub_img_path = add_postfix(image_path, '{0:d}'.format(i))
sub_img.save(sub_img_path)
def pythonscraping__com_humans_only_test():
from minghu6.internet.simulate_logon import pythonscraping__com_humans_only as logon
from minghu6.graphic.captcha.url_captcha import pythonscraping__com_humans_only as get
from minghu6.graphic.captcha.get_image import get_image
responseSet = get()
captchaUrl, session = responseSet[:2]
params_dict = responseSet[-1]
imgObj, _ = get_image(captchaUrl)
imgObj.show()
captcha = input('input captcha please\n').strip()
params_dict['captcha_response'] = captcha
logon(session=session, **params_dict)
print(params_dict)
def pythonscraping__com_humans_only_test():
from minghu6.internet.simulate_logon import pythonscraping__com_humans_only as logon
from minghu6.graphic.captcha.url_captcha import pythonscraping__com_humans_only as get
from minghu6.graphic.captcha.get_image import get_image
responseSet = get()
captchaUrl, session = responseSet[:2]
params_dict = responseSet[-1]
imgObj, _ = get_image(captchaUrl)
imgObj.show()
captcha = input('input captcha please\n').strip()
params_dict['captcha_response'] = captcha
logon(session=session, **params_dict)
print(params_dict)
def main_preprocessing(path, preprocessing_method, outdir=os.path.curdir, width=None):
imgObj, image_path = get_image(path)
# from list to set
assert preprocessing_method in preprocessing_method_dict.keys(), 'invalid params in -prepro'
other_kwargs = {}
if preprocessing_method == 'remove_frame' and width is not None:
other_kwargs['frame_width'] = int(width)
try:
imgObj = preprocessing_method_dict[preprocessing_method](imgObj, **other_kwargs)
# imgObj.show()
except pp.ImageSizeError as ex:
print(ex)
newpath = add_postfix(image_path, PREPROCESSING_FLAG_DICT[preprocessing_method])
imgObj.save(newpath)
def tesseract(path, limit_config=None, args=None, session: requests = None):
if not has_proper_tesseract():
raise DoNotHaveProperVersion
imgObj, image_path = get_image(path, session=session)
if args is None:
cmd_str = 'tesseract -psm 8 {0} stdout '.format(image_path)
else:
cmd_str = ' '.join(['tesseract', args, image_path, 'stdout'])
print(cmd_str)
if limit_config is not None: # like digits
cmd_str += ' {0}'.format(limit_config)
info_lines, err_lines = exec_cmd(cmd_str, shell=True)
try:
captchaResponse = info_lines[0].strip()
except IndexError:
raise Exception(''.join(err_lines))
else:
return captchaResponse
def main_preprocessing(path,
preprocessing_method,
outdir=os.path.curdir,
width=None):
imgObj, image_path = get_image(path)
# from list to set
assert preprocessing_method in preprocessing_method_dict.keys(
), 'invalid params in -prepro'
other_kwargs = {}
if preprocessing_method == 'remove_frame' and width is not None:
other_kwargs['frame_width'] = int(width)
try:
imgObj = preprocessing_method_dict[preprocessing_method](
imgObj, **other_kwargs)
# imgObj.show()
except pp.ImageSizeError as ex:
print(ex)
newpath = add_postfix(image_path,
PREPROCESSING_FLAG_DICT[preprocessing_method])
imgObj.save(newpath)