__init__.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445
  1. # coding=utf-8
  2. from __future__ import unicode_literals
  3. from text_unidecode import unidecode
  4. from .. import BaseProvider
  5. from ipaddress import ip_address, ip_network, IPv4Address, IPV4LENGTH, IPV6LENGTH
  6. # from faker.generator import random
  7. # from faker.providers.lorem.la import Provider as Lorem
  8. from faker.utils.decorators import lowercase, slugify, slugify_unicode
  9. localized = True
  10. class _IPv4Constants:
  11. """
  12. IPv4 network constants used to group networks into different categories.
  13. Structure derived from `ipaddress._IPv4Constants`.
  14. Excluded network list is updated to comply with current IANA list of
  15. private and reserved networks.
  16. """
  17. _network_classes = {
  18. 'a': ip_network('0.0.0.0/1'),
  19. 'b': ip_network('128.0.0.0/2'),
  20. 'c': ip_network('192.0.0.0/3')
  21. }
  22. _linklocal_network = ip_network('169.254.0.0/16')
  23. _loopback_network = ip_network('127.0.0.0/8')
  24. _multicast_network = ip_network('224.0.0.0/4')
  25. # Three common private networks from class A, B and CIDR
  26. # to generate private addresses from.
  27. _private_networks = [
  28. ip_network('10.0.0.0/8'),
  29. ip_network('172.16.0.0/12'),
  30. ip_network('192.168.0.0/16')
  31. ]
  32. # List of networks from which IP addresses will never be generated,
  33. # includes other private IANA and reserved networks from
  34. # ttps://www.iana.org/assignments/iana-ipv4-special-registry/iana-ipv4-special-registry.xhtml
  35. _excluded_networks = [
  36. ip_network('0.0.0.0/8'),
  37. ip_network('100.64.0.0/10'),
  38. ip_network('127.0.0.0/8'),
  39. ip_network('169.254.0.0/16'),
  40. ip_network('192.0.0.0/24'),
  41. ip_network('192.0.2.0/24'),
  42. ip_network('192.31.196.0/24'),
  43. ip_network('192.52.193.0/24'),
  44. ip_network('192.88.99.0/24'),
  45. ip_network('192.175.48.0/24'),
  46. ip_network('198.18.0.0/15'),
  47. ip_network('198.51.100.0/24'),
  48. ip_network('203.0.113.0/24'),
  49. ip_network('240.0.0.0/4'),
  50. ip_network('255.255.255.255/32')
  51. ] + [
  52. _linklocal_network,
  53. _loopback_network,
  54. _multicast_network
  55. ]
  56. class Provider(BaseProvider):
  57. safe_email_tlds = ('org', 'com', 'net')
  58. free_email_domains = ('gmail.com', 'yahoo.com', 'hotmail.com')
  59. tlds = (
  60. 'com', 'com', 'com', 'com', 'com', 'com', 'biz', 'info', 'net', 'org'
  61. )
  62. uri_pages = (
  63. 'index', 'home', 'search', 'main', 'post', 'homepage', 'category',
  64. 'register', 'login', 'faq', 'about', 'terms', 'privacy', 'author'
  65. )
  66. uri_paths = (
  67. 'app', 'main', 'wp-content', 'search', 'category', 'tag', 'categories',
  68. 'tags', 'blog', 'posts', 'list', 'explore'
  69. )
  70. uri_extensions = (
  71. '.html', '.html', '.html', '.htm', '.htm', '.php', '.php', '.jsp',
  72. '.asp'
  73. )
  74. user_name_formats = (
  75. '{{last_name}}.{{first_name}}',
  76. '{{first_name}}.{{last_name}}',
  77. '{{first_name}}##',
  78. '?{{last_name}}',
  79. )
  80. email_formats = (
  81. '{{user_name}}@{{domain_name}}',
  82. '{{user_name}}@{{free_email_domain}}',
  83. )
  84. url_formats = (
  85. 'www.{{domain_name}}/',
  86. '{{domain_name}}/',
  87. )
  88. uri_formats = (
  89. '{{url}}',
  90. '{{url}}{{uri_page}}/',
  91. '{{url}}{{uri_page}}{{uri_extension}}',
  92. '{{url}}{{uri_path}}/{{uri_page}}/',
  93. '{{url}}{{uri_path}}/{{uri_page}}{{uri_extension}}',
  94. )
  95. image_placeholder_services = (
  96. 'https://placeholdit.imgix.net/~text'
  97. '?txtsize=55&txt={width}x{height}&w={width}&h={height}',
  98. 'https://www.lorempixel.com/{width}/{height}',
  99. 'https://dummyimage.com/{width}x{height}',
  100. )
  101. replacements = tuple()
  102. def _to_ascii(self, string):
  103. for search, replace in self.replacements:
  104. string = string.replace(search, replace)
  105. string = unidecode(string)
  106. return string
  107. @lowercase
  108. def email(self, domain=None):
  109. if domain:
  110. email = '{0}@{1}'.format(self.user_name(), domain)
  111. else:
  112. pattern = self.random_element(self.email_formats)
  113. email = "".join(self.generator.parse(pattern).split(" "))
  114. return email
  115. @lowercase
  116. def safe_email(self):
  117. return '{}@example.{}'.format(
  118. self.user_name(), self.random_element(self.safe_email_tlds)
  119. )
  120. @lowercase
  121. def free_email(self):
  122. return self.user_name() + '@' + self.free_email_domain()
  123. @lowercase
  124. def company_email(self):
  125. return self.user_name() + '@' + self.domain_name()
  126. @lowercase
  127. def free_email_domain(self):
  128. return self.random_element(self.free_email_domains)
  129. @lowercase
  130. def ascii_email(self):
  131. pattern = self.random_element(self.email_formats)
  132. return self._to_ascii(
  133. "".join(self.generator.parse(pattern).split(" "))
  134. )
  135. @lowercase
  136. def ascii_safe_email(self):
  137. return self._to_ascii(
  138. self.user_name() +
  139. '@example.' +
  140. self.random_element(self.safe_email_tlds)
  141. )
  142. @lowercase
  143. def ascii_free_email(self):
  144. return self._to_ascii(
  145. self.user_name() + '@' + self.free_email_domain()
  146. )
  147. @lowercase
  148. def ascii_company_email(self):
  149. return self._to_ascii(
  150. self.user_name() + '@' + self.domain_name()
  151. )
  152. @slugify_unicode
  153. def user_name(self):
  154. pattern = self.random_element(self.user_name_formats)
  155. username = self._to_ascii(
  156. self.bothify(self.generator.parse(pattern)).lower()
  157. )
  158. return username
  159. @lowercase
  160. def domain_name(self, levels=1):
  161. """
  162. Produce an Internet domain name with the specified number of
  163. subdomain levels.
  164. >>> domain_name()
  165. nichols-phillips.com
  166. >>> domain_name(2)
  167. williamson-hopkins.jackson.com
  168. """
  169. if levels < 1:
  170. raise ValueError("levels must be greater than or equal to 1")
  171. if levels == 1:
  172. return self.domain_word() + '.' + self.tld()
  173. else:
  174. return self.domain_word() + '.' + self.domain_name(levels - 1)
  175. @lowercase
  176. @slugify_unicode
  177. def domain_word(self,):
  178. company = self.generator.format('company')
  179. company_elements = company.split(' ')
  180. company = self._to_ascii(company_elements.pop(0))
  181. return company
  182. def tld(self):
  183. return self.random_element(self.tlds)
  184. def url(self, schemes=None):
  185. """
  186. :param schemes: a list of strings to use as schemes, one will chosen randomly.
  187. If None, it will generate http and https urls.
  188. Passing an empty list will result in schemeless url generation like "://domain.com".
  189. :returns: a random url string.
  190. """
  191. if schemes is None:
  192. schemes = ['http', 'https']
  193. pattern = '{}://{}'.format(
  194. self.random_element(schemes) if schemes else "",
  195. self.random_element(self.url_formats)
  196. )
  197. return self.generator.parse(pattern)
  198. def _random_ipv4_address_from_subnet(self, subnet, network=False):
  199. """
  200. Produces a random IPv4 address or network with a valid CIDR
  201. from within a given subnet.
  202. :param subnet: IPv4Network to choose from within
  203. :param network: Return a network address, and not an IP address
  204. """
  205. address = str(
  206. subnet[self.generator.random.randint(
  207. 0, subnet.num_addresses - 1
  208. )]
  209. )
  210. if network:
  211. address += '/' + str(self.generator.random.randint(
  212. subnet.prefixlen,
  213. subnet.max_prefixlen
  214. ))
  215. address = str(ip_network(address, strict=False))
  216. return address
  217. def _exclude_ipv4_networks(self, networks, networks_to_exclude):
  218. """
  219. Exclude the list of networks from another list of networks
  220. and return a flat list of new networks.
  221. :param networks: List of IPv4 networks to exclude from
  222. :param networks_to_exclude: List of IPv4 networks to exclude
  223. :returns: Flat list of IPv4 networks
  224. """
  225. for network_to_exclude in networks_to_exclude:
  226. def _exclude_ipv4_network(network):
  227. """
  228. Exclude a single network from another single network
  229. and return a list of networks. Network to exclude
  230. comes from the outer scope.
  231. :param network: Network to exclude from
  232. :returns: Flat list of IPv4 networks after exclusion.
  233. If exclude fails because networks do not
  234. overlap, a single element list with the
  235. orignal network is returned. If it overlaps,
  236. even partially, the network is excluded.
  237. """
  238. try:
  239. return list(network.address_exclude(network_to_exclude))
  240. except ValueError:
  241. # If networks overlap partially, `address_exclude`
  242. # will fail, but the network still must not be used
  243. # in generation.
  244. if network.overlaps(network_to_exclude):
  245. return []
  246. else:
  247. return [network]
  248. networks = list(map(_exclude_ipv4_network, networks))
  249. # flatten list of lists
  250. networks = [
  251. item for nested in networks for item in nested
  252. ]
  253. return networks
  254. def ipv4_network_class(self):
  255. """
  256. Returns a IPv4 network class 'a', 'b' or 'c'.
  257. :returns: IPv4 network class
  258. """
  259. return self.random_element('abc')
  260. def ipv4(self, network=False, address_class=None, private=None):
  261. """
  262. Produce a random IPv4 address or network with a valid CIDR.
  263. :param network: Network address
  264. :param address_class: IPv4 address class (a, b, or c)
  265. :param private: Public or private
  266. :returns: IPv4
  267. """
  268. if private is True:
  269. return self.ipv4_private(address_class=address_class,
  270. network=network)
  271. elif private is False:
  272. return self.ipv4_public(address_class=address_class,
  273. network=network)
  274. # if neither private nor public is required explicitly,
  275. # generate from whole requested address space
  276. if address_class:
  277. all_networks = [_IPv4Constants._network_classes[address_class]]
  278. else:
  279. # if no address class is choosen, use whole IPv4 pool
  280. all_networks = [ip_network('0.0.0.0/0')]
  281. # exclude special networks
  282. all_networks = self._exclude_ipv4_networks(
  283. all_networks,
  284. _IPv4Constants._excluded_networks
  285. )
  286. # choose random network from the list
  287. random_network = self.generator.random.choice(all_networks)
  288. return self._random_ipv4_address_from_subnet(random_network, network)
  289. def ipv4_private(self, network=False, address_class=None):
  290. """
  291. Returns a private IPv4.
  292. :param network: Network address
  293. :param address_class: IPv4 address class (a, b, or c)
  294. :returns: Private IPv4
  295. """
  296. # compute private networks from given class
  297. supernet = _IPv4Constants._network_classes[
  298. address_class or self.ipv4_network_class()
  299. ]
  300. private_networks = [
  301. subnet for subnet in _IPv4Constants._private_networks
  302. if subnet.overlaps(supernet)
  303. ]
  304. # exclude special networks
  305. private_networks = self._exclude_ipv4_networks(
  306. private_networks,
  307. _IPv4Constants._excluded_networks
  308. )
  309. # choose random private network from the list
  310. private_network = self.generator.random.choice(private_networks)
  311. return self._random_ipv4_address_from_subnet(private_network, network)
  312. def ipv4_public(self, network=False, address_class=None):
  313. """
  314. Returns a public IPv4 excluding private blocks.
  315. :param network: Network address
  316. :param address_class: IPv4 address class (a, b, or c)
  317. :returns: Public IPv4
  318. """
  319. # compute public networks
  320. public_networks = [_IPv4Constants._network_classes[
  321. address_class or self.ipv4_network_class()
  322. ]]
  323. # exclude private and excluded special networks
  324. public_networks = self._exclude_ipv4_networks(
  325. public_networks,
  326. _IPv4Constants._private_networks +
  327. _IPv4Constants._excluded_networks
  328. )
  329. # choose random public network from the list
  330. public_network = self.generator.random.choice(public_networks)
  331. return self._random_ipv4_address_from_subnet(public_network, network)
  332. def ipv6(self, network=False):
  333. """Produce a random IPv6 address or network with a valid CIDR"""
  334. address = str(ip_address(self.generator.random.randint(
  335. 2 ** IPV4LENGTH, (2 ** IPV6LENGTH) - 1)))
  336. if network:
  337. address += '/' + str(self.generator.random.randint(0, IPV6LENGTH))
  338. address = str(ip_network(address, strict=False))
  339. return address
  340. def mac_address(self):
  341. mac = [self.generator.random.randint(0x00, 0xff) for _ in range(0, 6)]
  342. return ":".join(map(lambda x: "%02x" % x, mac))
  343. def uri_page(self):
  344. return self.random_element(self.uri_pages)
  345. def uri_path(self, deep=None):
  346. deep = deep if deep else self.generator.random.randint(1, 3)
  347. return "/".join(
  348. self.random_elements(self.uri_paths, length=deep)
  349. )
  350. def uri_extension(self):
  351. return self.random_element(self.uri_extensions)
  352. def uri(self):
  353. pattern = self.random_element(self.uri_formats)
  354. return self.generator.parse(pattern)
  355. @slugify
  356. def slug(self, value=None):
  357. """Django algorithm"""
  358. if value is None:
  359. value = self.generator.text(20)
  360. return value
  361. def image_url(self, width=None, height=None):
  362. """
  363. Returns URL to placeholder image
  364. Example: http://placehold.it/640x480
  365. """
  366. width_ = width or self.random_int(max=1024)
  367. height_ = height or self.random_int(max=1024)
  368. placeholder_url = self.random_element(self.image_placeholder_services)
  369. return placeholder_url.format(width=width_, height=height_)