def _get_structure(self):
"""
This method return the structure we are goinng to work with.
"""
structure_file = ""
req = ""
if path.isfile(self.structure):
structure_file = self.structure
elif path.isfile(self.base + "dir_structure_production.json"):
structure_file = self.base + "dir_structure_production.json"
else:
if "dev" not in PyFunceble.VERSION:
req = requests.get(PyFunceble.LINKS["dir_structure"])
else:
req = requests.get(PyFunceble.LINKS["dir_structure"].replace(
"master", "dev"))
if structure_file.endswith("_production.json"):
structure = Dict().from_json(File(structure_file).read())
return self._update_structure_from_config(structure)
elif structure_file.endswith(".json"):
return Dict().from_json(File(structure_file).read())
return self._update_structure_from_config(Dict().from_json(req.text))
def __wordpress_dot_com(self):
"""
Handle the wordpress.com SPECIAL case.
:return:
:code:`(new status, new source)` or :code:`None` if there is any
change to apply.
:rtype: tuple|None
"""
# We initiate a variable which whill have to be into the HTML
# in order to be considered as inactive.
does_not_exist = "doesn’t exist"
# We get the content of the page.
wordpress_com_content = requests.get(
"http://%s:80" % PyFunceble.INTERN["to_test"], headers=self.headers
)
if does_not_exist in wordpress_com_content.text:
# The marker is into the page content.
# We return the new status and source.
return self.__special_down()
# We return None, there is no changes.
return None
def _special_wordpress_com(self):
"""
Handle the wordpress.com special case.
"""
# We initiate the domain to match.
wordpress_com = ".wordpress.com"
# We initiate a variable which whill have to be into the HTML
# in order to be considered as inactive.
does_not_exist = "doesn’t exist"
if self.tested.endswith(wordpress_com):
# The currently tested element ends with wordpress.com.
# We get the content of the page.
wordpress_com_content = requests.get("http://%s:80" % self.tested,
headers=self.headers)
if does_not_exist in wordpress_com_content.text:
# The marker is into the page content.
# We update the source.
self.source = "SPECIAL"
# We update the status.
self.domain_status = PyFunceble.STATUS["official"]["down"]
# We update the output file.
self.output = (self.output_parent_dir +
PyFunceble.OUTPUTS["splited"]["directory"] +
self.domain_status)
def _special_blogspot(self):
"""
Handle the blogspot SPECIAL case.
"""
# We initate a variable whith a regex which will match all blogpost domain.
regex_blogspot = ".blogspot."
# We iniate a list of elements in the HTML which will tell us more about
# the status of the domain.
regex_blogger = ["create-blog.g?", "87065", "doesn’t exist"]
if PyFunceble.INTERN["to_test_type"] == "domain":
# The element we are testing is a domain.
# We construct the url to get.
url_to_get = "http://%s" % self.tested
elif PyFunceble.INTERN["to_test_type"] == "url":
# The element we are testing is a URL.
# We construct the url to get.
url_to_get = self.tested
else:
raise Exception("Unknow test type.")
if Regex(self.tested, regex_blogspot, return_data=False,
escape=True).match():
# The element we are testing is a blogspot subdomain.
# We get the HTML of the home page.
blogger_content_request = requests.get(url_to_get,
headers=self.headers)
for regx in regex_blogger:
# We loop through the list of regex to match.
if (regx in blogger_content_request.text or Regex(
blogger_content_request.text,
regx,
return_data=False,
escape=False,
).match()):
# The content match the currently read regex.
# We update the source.
self.source = "SPECIAL"
# We update the domain status.
self.domain_status = PyFunceble.STATUS["official"]["down"]
# We update the output file.
self.output = (self.output_parent_dir +
PyFunceble.OUTPUTS["splited"]["directory"] +
self.domain_status)
# And we break the loop as we matched something.
break
def __blogspot(self):
"""
Handle the blogspot SPECIAL case.
:return:
:code:`(new status, new source)` or :code:`None` if there is any
change to apply.
:rtype: tuple|None
"""
# We iniate a list of elements in the HTML which will tell us more about
# the status of the domain.
regex_blogger = ["create-blog.g?", "87065", "doesn’t exist"]
if PyFunceble.INTERN["to_test_type"] == "domain":
# The element we are testing is a domain.
# We construct the url to get.
url_to_get = "http://%s" % PyFunceble.INTERN["to_test"]
elif PyFunceble.INTERN["to_test_type"] == "url":
# The element we are testing is a URL.
# We construct the url to get.
url_to_get = PyFunceble.INTERN["to_test"]
else:
raise NotImplementedError(
"to_test_type not implemented: `{}`".format(
PyFunceble.INTERN["to_test_type"]
)
)
# We get the HTML of the home page.
blogger_content_request = requests.get(url_to_get, headers=self.headers)
for regx in regex_blogger:
# We loop through the list of regex to match.
if (
regx in blogger_content_request.text
or Regex(
blogger_content_request.text,
regx,
return_data=False,
escape=False,
).match()
):
# * The currently read regex is present into the docuement.
# or
# * Something in the document match the currently read regex.
# We update the status and source.
return self.__special_down()
# We return None, there is no changes.
return None
def text(self):
"""
This method download the given link and return its requests.text.
"""
req = requests.get(self.link)
if req.status_code == 200:
if self.return_data:
return req.text
File(self.destination).write(req.text, overwrite=True)
return True
raise Exception("Unable to download %s." % repr(self.link))
def special_wordpress_com(self):
"""
Handle the wordpress.com special case.
"""
wordpress_com = ".wordpress.com"
does_not_exist = "doesn’t exist"
if self.tested.endswith(wordpress_com):
wordpress_com_content = requests.get("http://%s:80" % self.tested)
if does_not_exist in wordpress_com_content.text:
self.source = "SPECIAL"
self.domain_status = PyFunceble.STATUS["official"]["down"]
self.output = self.output_parent_dir + PyFunceble.OUTPUTS["splited"][
"directory"
] + self.domain_status
def text(self):
"""
Download the given link and return or save its :code:`requests.text`
at the given destination.
:rtype: mixed
:raises:
:code:`Exception`
If the status code is not :code:`200`.
"""
try:
# We request the link.
req = requests.get(self.link, verify=self.verification)
if req.status_code == 200:
# The request http status code is equal to 200.
if self.return_data:
# We have to return the data.
# We return the link content.
return req.text
# We save the link content to the parsed destination.
File(self.destination).write(req.text, overwrite=True)
# We return True.
return True
# The request http status code is not equal to 200.
# We raise an exception saying that we were unable to download.
raise Exception("Unable to download %s." % repr(self.link))
except requests.exceptions.ConnectionError:
print(Fore.RED + "No Internet connection available." +
Style.RESET_ALL)
exit(1)
def special_blogspot(self):
"""
Handle the blogspot SPECIAL case.
"""
regex_blogspot = ".blogspot."
regex_blogger = ["create-blog.g?", "87065", "doesn’t exist"]
if Regex(self.tested, regex_blogspot, return_data=False, escape=True).match():
blogger_content_request = requests.get("http://%s:80" % self.tested)
for regx in regex_blogger:
if regx in blogger_content_request.text or Regex(
blogger_content_request.text, regx, return_data=False, escape=False
).match():
self.source = "SPECIAL"
self.domain_status = PyFunceble.STATUS["official"]["down"]
self.output = self.output_parent_dir + PyFunceble.OUTPUTS[
"splited"
][
"directory"
] + self.domain_status
break
