Ural

A helper library full of URL-related heuristics.

Installation

You can install ural with pip with the following command:

pip install ural

Usage

Generic functions

• ensure_protocol
• get_domain_name
• get_hostname
• get_normalized_hostname
• force_protocol
• is_shortened_url
• is_url
• normalize_url
• strip_protocol
• urls_from_html
• urls_from_text

LRU-related functions (What on earth is a LRU?)

• lru.url_to_lru
• lru.lru_to_url
• lru.lru_stems
• lru.normalized_lru_stems

LRU-related classes

• LRUTrie

  • set
  • set_lru
  • match
  • match_lru

• NormalizedLRUTrie

Platform-specific functions

• facebook
  • is_facebook_url
  • convert_facebook_url_to_mobile
  • extract_user_from_url
• google
  • is_amp_url
  • extract_url_from_google_link
• youtube
  • is_youtube_url
  • is_youtube_video_id
  • parse_youtube_url
  • extract_video_id_from_youtube_url
  • normalize_youtube_url

ensure_protocol

Function checking if the url has a protocol, and adding the given one if there is none.

from ural import ensure_protocol

ensure_protocol('www2.lemonde.fr', protocol='https')
>>> 'https://www2.lemonde.fr'

Arguments

• url string: URL to format.
• protocol string: protocol to use if there is none in url. Is 'http' by default.

get_domain_name

Function returning an url's domain name. This function is of course tld-aware and will return None if no valid domain name can be found.

from ural import get_domain_name

get_domain_name('https://facebook.com/path')
>>> 'facebook.com'

get_hostname

Function returning the given url's full hostname. It can work on scheme-less urls.

from ural import get_hostname

get_hostname('http://www.facebook.com/path')
>>> 'www.facebook.com'

get_normalized_hostname

Function returning the given url's normalized hostname, i.e. without usually irrelevant subdomains etc. Works a lot like normalize_url.

from ural import get_normalized_hostname

get_normalized_hostname('http://www.facebook.com/path')
>>> 'facebook.com'

get_normalized_hostname('http://fr-FR.facebook.com/path', strip_lang_subdomains=True)
>>> 'facebook.com'

Arguments

• url str: Target url.
• normalized_amp ?bool: Whether to attempt to normalize Google AMP subdomains.
• strip_lang_subdomains ?bool: Whether to drop language-specific subdomains.

force_protocol

Function force-replacing the protocol of the given url.

from ural import force_protocol

force_protocol('https://www2.lemonde.fr', protocol='ftp')
>>> 'ftp://www2.lemonde.fr'

Arguments

• url string: URL to format.
• protocol string: protocol wanted in the output url. Is 'http' by default.

is_shortened_url

Function returning whether the given url is probably a shortened url. It works by matching the given url domain against most prominent shortener domains. So the result could be a false negative.

from ural import is_shortened_url

is_shortened_url('http://lemonde.fr')
>>> False

is_shortened_url('http://bit.ly/1sNZMwL')
>>> True

is_url

Function returning whether the given string is a valid url.

from ural import is_url

is_url('https://www2.lemonde.fr')
>>> True

is_url('lemonde.fr/economie/article.php', require_protocol=False)
>>> True

is_url('lemonde.falsetld/whatever.html', tld_aware=True)
>>> False

Arguments

• string string: string to test.
• require_protocol bool [True]: whether the argument has to have a protocol to be considered a url.
• tld_aware bool [False]: whether to check if the url's tld actually exists or not.
• allow_spaces_in_path bool [False]: whether the allow spaces in URL paths.
• only_http_https bool [True]: whether to only allow the http and https protocols.

normalize_url

Function normalizing the given url by stripping it of usually non-discriminant parts such as irrelevant query items or sub-domains etc.

This is a very useful utility when attempting to match similar urls written slightly differently when shared on social media etc.

from ural import normalize_url

normalize_url('https://www2.lemonde.fr/index.php?utm_source=google')
>>> 'lemonde.fr'

Arguments

• url string: URL to normalize.
• normalize_amp bool [True]: whether to attempt to normalize Google AMP urls.
• sort_query bool [True]: whether to sort query items.
• strip_authentication bool [True]: whether to strip authentication.
• strip_fragment bool|str ['except-routing']: whether to strip the url's fragment. If set to except-routing, will only strip the fragment if the fragment is not deemed to be js routing (i.e. if it contains a /).
• strip_index bool [True]: whether to strip trailing index.
• strip_lang_subdomains bool [False]: whether to strip language subdomains (ex: 'fr-FR.lemonde.fr' to only 'lemonde.fr' because 'fr-FR' isn't a relevant subdomain, it indicates the language and the country).
• strip_trailing_slash bool [False]: whether to strip trailing slash.

strip_protocol

Function removing the protocol from the url.

from ural import strip_protocol

strip_protocol('https://www2.lemonde.fr/index.php')
>>> 'www2.lemonde.fr/index.php'

Arguments

• url string: URL to format.

urls_from_html

Function returning an iterator over the urls present in the links of given HTML text.

from ural import urls_from_html

html = """<p>Hey! Check this site: <a href="https://medialab.sciencespo.fr/">médialab</a></p>"""

for url in urls_from_html(html):
    print(url)
>>> 'https://medialab.sciencespo.fr/'

Arguments

• string string: html string.

urls_from_text

Function returning an iterator over the urls present in the string argument. Extracts only the urls with a protocol.

from ural import urls_from_text

text = "Hey! Check this site: https://medialab.sciencespo.fr/, it looks really cool. They're developing many tools on https://github.com/"

for url in urls_from_text(text):
    print(url)
>>> 'https://medialab.sciencespo.fr/'
>>> 'https://github.com/'

Arguments

• string string: source string.

lru.url_to_lru

Function converting the given url to a serialized lru.

from ural.lru import url_to_lru

url_to_lru('http://www.lemonde.fr:8000/article/1234/index.html?field=value#2')
>>> 's:http|t:8000|h:fr|h:lemonde|h:www|p:article|p:1234|p:index.html|q:field=value|f:2|'

Arguments

• url string: url to convert.
• tld_aware ?bool: whether to acknowledge TLDs when converting.

lru.lru_to_url

Function converting the given serialized lru or lru stems to a proper url.

from ural.lru import lru_to_url

lru_to_url('s:http|t:8000|h:fr|h:lemonde|h:www|p:article|p:1234|p:index.html|')
>>> 'http://www.lemonde.fr:8000/article/1234/index.html'

lru_to_url(['s:http', 'h:fr', 'h:lemonde', 'h:www', 'p:article', 'p:1234', 'p:index.html'])
>>> 'http://www.lemonde.fr:8000/article/1234/index.html'

lru.lru_stems

Function returning url parts in hierarchical order.

from ural.lru import lru_stems

lru_stems('http://www.lemonde.fr:8000/article/1234/index.html?field=value#2')
>>> ['s:http', 't:8000', 'h:fr', 'h:lemonde', 'h:www', 'p:article', 'p:1234', 'p:index.html', 'q:field=value', 'f:2']

Arguments

• url string: URL to parse.
• tld_aware ?bool: whether to acknowledge TLDs when stemming.

lru.normalized_lru_stems

Function normalizing url and returning its parts in hierarchical order.

from ural.lru import normalized_lru_stems

normalized_lru_stems('http://www.lemonde.fr:8000/article/1234/index.html?field=value#2')
>>> ['t:8000', 'h:fr', 'h:lemonde', 'h:www', 'p:article', 'p:1234', 'q:field=value']

Arguments

This function accepts the same arguments as normalize_url.

LRUTrie

Class implementing a prefix tree (Trie) storing URLs hierarchically by storing them as LRUs along with some arbitrary metadata. It is very useful when needing to match URLs by longest common prefix.

Note that this class directly inherits from the phylactery library's TrieDict so you can also use any of its methods.

from ural.lru import LRUTrie

trie = LRUTrie()

# To respect tlds
trie = LRUTrie(tld_aware=True)

set

Method storing a URL in a LRUTrie along with its metadata.

from ural.lru import LRUTrie

trie = LRUTrie()
trie.set('http://www.lemonde.fr', {'type': 'general press'})

trie.match('http://www.lemonde.fr')
>>> {'type': 'general press'}

Arguments

• url string: url to store in the LRUTrie.
• metadata any: metadata of the url.

set_lru

Method storing a URL already represented as a LRU or LRU stems along with its metadata.

from ural.lru import LRUTrie

trie = LRUTrie()

# Using stems
trie.set_lru(['s:http', 'h:fr', 'h:lemonde', 'h:www'], {'type': 'general press'})

# Using serialized lru
trie.set_lru('s:http|h:fr|h:lemonde|h:www|', {'type': 'general_press'})

Arguments

• lru string|list: lru to store in the Trie.
• metadata any: metadata to attach to the lru.

match

Method returning the metadata attached to the longest prefix match of your query URL. Will return None if no common prefix can be found.

from ural.lru import LRUTrie

trie = LRUTrie()
trie.set('http://www.lemonde.fr', {'media': 'lemonde'})

trie.match('http://www.lemonde.fr')
>>> {'media': 'lemonde'}
trie.match('http://www.lemonde.fr/politique')
>>> {'media': 'lemonde'}

trie.match('http://www.lefigaro.fr')
>>> None

Arguments

• url string: url to match in the LRUTrie.

match_lru

Method returning the metadata attached to the longest prefix match of your query LRU. Will return None if no common prefix can be found.

from ural.lru import LRUTrie

trie = LRUTrie()
trie.set(['s:http', 'h:fr', 'h:lemonde', 'h:www'], {'media': 'lemonde'})

trie.match(['s:http', 'h:fr', 'h:lemonde', 'h:www'])
>>> {'media': 'lemonde'}
trie.match('s:http|h:fr|h:lemonde|h:www|p:politique|')
>>> {'media': 'lemonde'}

trie.match(['s:http', 'h:fr', 'h:lefigaro', 'h:www'])
>>> None

Arguments

• lru string|list: lru to match in the LRUTrie.

NormalizedLRUTrie

The NormalizedLRUTrie is nearly identical to the standard LRUTrie except that it normalized urls given to it before attempting any operation. It is a good choice if you want to avoid prefix queries issues related to http vs https or www shenanigans, for instance.

To tweak its normalization, you can give to NormalizedLRUTrie the same options you would give to normalize_url:

from ural.lru import NormalizedLRUTrie

trie = NormalizedLRUTrie(strip_trailing_slash=True)

Note that there are still some differences between the LRUTrie and the NormalizedLRUTrie:

1. The NormalizedLRUTrie cannot be TLD aware.
2. The NormalizedLRUTrie does not have the #.set_lru and #.match_lru methods.

Facebook

is_facebook_url

Function returning whether given url is from Facebook or not.

from ural.facebook import is_facebook_url

is_facebook_url('http://www.facebook.com/post/974583586343')
>>> True

is_facebook_url('https://fb.me/846748464')
>>> True

is_facebook_url('https://www.lemonde.fr')
>>> False

convert_facebook_url_to_mobile

Function returning the mobile version of the given Facebook url. Will raise an exception if a non-Facebook url is given.

from ural.facebook import convert_facebook_url_to_mobile

convert_facebook_url_to_mobile('http://www.facebook.com/post/974583586343')
>>> 'http://m.facebook.com/post/974583586343'

extract_user_from_url

Function extracting user information from a facebook user url.

from ural.facebook import extract_user_from_url

extract_user_from_url('https://www.facebook.com/people/Sophia-Aman/102016783928989')
>>> FacebookUser(id='102016783928989', handle=None, url='https://www.facebook.com/profile.php?id=102016783928989)

extract_user_from_url('/annelaure.rivolu?rc=p&__tn__=R')
>>> FacebookUser(id=None, handle='annelaure.rivolu', url='https://www.facebook.com/annelaure.rivolu)

Google

is_amp_url

Returns whether the given url is probably a Google AMP url.

from ural.google import is_amp_url

is_amp_url('http://www.europe1.fr/sante/les-onze-vaccins.amp')
>>> True

is_amp_url('https://www.lemonde.fr')
>>> False

extract_url_from_google_link

Extracts the url from the given Google search link. This is useful to "resolve" the links scraped from Google's search results.

from ural.google import extract_url_from_google_link

extract_url_from_google_link('https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=2ahUKEwicu4K-rZzmAhWOEBQKHRNWA08QFjAAegQIARAB&url=https%3A%2F%2Fwww.facebook.com%2Fieff.ogbeide&usg=AOvVaw0vrBVCiIHUr5pncjeLpPUp')

>>> 'https://www.facebook.com/ieff.ogbeide'

Youtube

is_youtube_url

Returns whether the given url is from Youtube.

from ural.youtube import is_youtube_url

is_youtube_url('https://lemonde.fr')
>>> False

is_youtube_url('https://www.youtube.com/watch?v=otRTOE9i51o')
>>> True

is_youtube_url('https://youtu.be/otRTOE9i51o)
>>> True

is_youtube_video_id

Returns whether the given string is a formally valid Youtube id. Note that it won't validate the fact that this id actually refers to an existing video or not. You will need to call Youtube servers for that.

from ural.youtube import is_youtube_video_id

is_youtube_video_id('otRTOE9i51o')
>>> True

is_youtube_video_id('bDYTYET')
>>> False

parse_youtube_url

Returns parsed information about the given youtube url: either about the linked video, user or channel. If the url is an invalid Youtube url or if not a Youtube url, the function returns None.

from ural.youtube import (
  parse_youtube_url,

  # You can also import the named tuples if you need them
  YoutubeVideo,
  YoutubeUser,
  YoutubeChannel
)

parse_youtube_url('https://www.youtube.com/watch?v=otRTOE9i51o')
>>> YoutubeVideo(id='otRTOE9i51o')

parse_youtube_url('https://lemonde.fr')
>>> None

parse_youtube_url('http://www.youtube.com/channel/UCWvUxN9LAjJ-sTc5JJ3gEyA/videos')
>>> YoutubeChannel(id='UCWvUxN9LAjJ-sTc5JJ3gEyA', name=None)

parse_youtube_url('http://www.youtube.com/user/ojimfrance')
>>> YoutubeUser(id=None, name='ojimfrance')

parse_youtube_url('https://www.youtube.com/taranisnews')
>>> YoutubeChannel(id=None, name='taranisnews')

Arguments

• url str: Youtube url to parse.
• fix_common_mistakes bool [True]: Whether to fix common mistakes that can be found in Youtube urls as you can find them when crawling the web.

extract_video_id_from_youtube_url

Return a video id from the given Youtube url or None if we could not find one.

from ural.youtube import extract_video_id_from_youtube_url

extract_video_id_from_youtube_url('https://www.youtube.com/watch?v=otRTOE9i51o')
>>> 'otRTOE9i51o'

extract_video_id_from_youtube_url('https://lemonde.fr')
>>> None

extract_video_id_from_youtube_url('http://youtu.be/afa-5HQHiAs')
>>> 'afa-5HQHiAs'

normalize_youtube_url

Returns a normalized version of the given Youtube url. It will normalize video, user and channel urls so you can easily match them.

from ural.youtube import normalize_youtube_url

normalize_youtube_url('https://www.youtube.com/watch?v=otRTOE9i51o')
>>> 'https://www.youtube.com/watch?v=otRTOE9i51o'

normalize_youtube_url('http://youtu.be/afa-5HQHiAs')
>>> 'https://www.youtube.com/watch?v=afa-5HQHiAs'

Miscellaneous

About LRUs

TL;DR: a LRU is a hierarchical reordering of a URL so that one can perform meaningful prefix queries on URLs.

If you observe many URLs, you will quickly notice that they are not written in sound hierarchical order. In this URL, for instance:

http://business.lemonde.fr/articles/money.html?id=34#content

Some parts, such as the subdomain, are written in an "incorrect order". And this is fine, really, this is how URLs always worked.

But if what you really want is to match URLs, you will need to reorder them so that their order closely reflects the hierarchy of their targeted content. And this is exactly what LRUs are (that and also a bad pun on URL, since a LRU is basically a "reversed" URL).

Now look how the beforementioned URL could be splitted into LRU stems:

[
  's:http',
  'h:fr',
  'h:lemonde',
  'h:business',
  'p:articles',
  'p:money.html',
  'q:id=34',
  'f:content'
]

And typically, this list of stems will be serialized thusly:

s:http|h:fr|h:lemonde|h:business|p:articles|p:money.html|q:id=34|f:content|

The trailing slash is added so that serialized LRUs can be prefix-free.