Oven logo

Oven

linkify-it-py

CI pypi Anaconda-Server Badge Documentation Status codecov Maintainability

This is Python port of linkify-it.

Links recognition library with FULL unicode support. Focused on high quality link patterns detection in plain text.

Demo

Javascript Demo

Why it's awesome:

  • Full unicode support, with astral characters!
  • International domains support.
  • Allows rules extension & custom normalizers.

Install

pip install linkify-it-py

or

conda install -c conda-forge linkify-it-py

Usage examples

Example 1. Simple use

from linkify_it import LinkifyIt


linkify = LinkifyIt()

print(linkify.test("Site github.com!"))
# => True

print(linkify.match("Site github.com!"))
# => [linkify_it.main.Match({
#         'schema': '',
#         'index': 5,
#         'last_index': 15,
#         'raw': 'github.com',
#         'text': 'github.com',
#         'url': 'http://github.com'
#     }]

Example 2. With options

from linkify_it import LinkifyIt
from linkify_it.tlds import TLDS


# Reload full tlds list & add unofficial `.onion` domain.
linkify = (
    LinkifyIt()
    .tlds(TLDS)               # Reload with full tlds list
    .tlds("onion", True)      # Add unofficial `.onion` domain
    .add("git:", "http:")     # Add `git:` protocol as "alias"
    .add("ftp:", None)        # Disable `ftp:` protocol
    .set({"fuzzy_ip": True})  # Enable IPs in fuzzy links (without schema)
)
print(linkify.test("Site tamanegi.onion!"))
# => True

print(linkify.match("Site tamanegi.onion!"))
# => [linkify_it.main.Match({
#         'schema': '',
#         'index': 5,
#         'last_index': 19,
#         'raw': 'tamanegi.onion',
#         'text': 'tamanegi.onion',
#         'url': 'http://tamanegi.onion'
#     }]

Example 3. Add twitter mentions handler

from linkify_it import LinkifyIt


linkify = LinkifyIt()

def validate(obj, text, pos):
    tail = text[pos:]

    if not obj.re.get("twitter"):
        obj.re["twitter"] = re.compile(
            "^([a-zA-Z0-9_]){1,15}(?!_)(?=$|" + obj.re["src_ZPCc"] + ")"
        )
    if obj.re["twitter"].search(tail):
        if pos > 2 and tail[pos - 2] == "@":
            return False
        return len(obj.re["twitter"].search(tail).group())
    return 0

def normalize(obj, match):
    match.url = "https://twitter.com/" + re.sub(r"^@", "", match.url)

linkify.add("@", {"validate": validate, "normalize": normalize})

API

API documentation

LinkifyIt(schemas, options)

Creates new linkifier instance with optional additional schemas.

By default understands:

  • http(s)://... , ftp://..., mailto:... & //... links
  • "fuzzy" links and emails (google.com, [email protected]).

schemas is an dict, where each key/value describes protocol/rule:

  • key - link prefix (usually, protocol name with : at the end, skype: for example). linkify-it-py makes sure that prefix is not preceded with alphanumeric char.
  • value - rule to check tail after link prefix
    • str
      • just alias to existing rule
    • dict
      • validate - either a re.Pattern (start with ^, and don't include the link prefix itself), or a validator function which, given arguments self, text and pos, returns the length of a match in text starting at index pos. pos is the index right after the link prefix. self can be used to access the linkify object to cache data.
      • normalize - optional function to normalize text & url of matched result (for example, for twitter mentions).

options:

  • fuzzy_link - recognize URL-s without http(s):// head. Default True.
  • fuzzy_ip - allow IPs in fuzzy links above. Can conflict with some texts like version numbers. Default False.
  • fuzzy_email - recognize emails without mailto: prefix. Default True.
  • --- - set True to terminate link with --- (if it's considered as long dash).

.test(text)

Searches linkifiable pattern and returns True on success or False on fail.

.pretest(text)

Quick check if link MAY BE can exist. Can be used to optimize more expensive .test() calls. Return False if link can not be found, True - if .test() call needed to know exactly.

.test_schema_at(text, name, position)

Similar to .test() but checks only specific protocol tail exactly at given position. Returns length of found pattern (0 on fail).

.match(text)

Returns list of found link matches or null if nothing found.

Each match has:

  • schema - link schema, can be empty for fuzzy links, or // for protocol-neutral links.
  • index - offset of matched text
  • last_index - index of next char after mathch end
  • raw - matched text
  • text - normalized text
  • url - link, generated from matched text

.matchAtStart(text)

Checks if a match exists at the start of the string. Returns Match (see docs for match(text)) or null if no URL is at the start. Doesn't work with fuzzy links.

.tlds(list_tlds, keep_old=False)

Load (or merge) new tlds list. Those are needed for fuzzy links (without schema) to avoid false positives. By default:

  • 2-letter root zones are ok.
  • biz|com|edu|gov|net|org|pro|web|xxx|aero|asia|coop|info|museum|name|shop|рф are ok.
  • encoded (xn--...) root zones are ok.

If that's not enough, you can reload defaults with more detailed zones list.

.add(key, value)

Add a new schema to the schemas object. As described in the constructor definition, key is a link prefix (skype:, for example), and value is a str to alias to another schema, or an dict with validate and optionally normalize definitions. To disable an existing rule, use .add(key, None).

.set(options)

Override default options. Missed properties will not be changed.

License

MIT