Skip to main content

spaCy pipeline component for adding emoji metadata to Doc, Token and Span objects

Project description

spacymoji: emoji for spaCy

spaCy extension and pipeline component for adding emoji meta data to Doc objects. Detects emoji consisting of one or more unicode characters, and can optionally merge multi-char emoji (combined pictures, emoji with skin tone modifiers) into one token. Human-readable emoji descriptions are added as a custom attribute, and an optional lookup table can be provided for your own descriptions. The extension sets the custom Doc, Token and Span attributes ._.is_emoji, ._.emoji_desc, ._.has_emoji and ._.emoji. You can read more about custom pipeline components and extension attributes here.

Emoji are matched using spaCy's PhraseMatcher, and looked up in the data table provided by the emoji package.

tests Current Release Version pypi Version

⏳ Installation

spacymoji requires spacy v3.0.0 or higher. For spaCy v2.x, install spacymoji==2.0.0.

pip install spacymoji

☝️ Usage

Import the component and add it anywhere in your pipeline using the string name of the "emoji" component factory:

import spacy

nlp = spacy.load("en_core_web_sm")
nlp.add_pipe("emoji", first=True)
doc = nlp("This is a test 😻 👍🏿")
assert doc._.has_emoji is True
assert doc[2:5]._.has_emoji is True
assert doc[0]._.is_emoji is False
assert doc[4]._.is_emoji is True
assert doc[5]._.emoji_desc == "thumbs up dark skin tone"
assert len(doc._.emoji) == 2
assert doc._.emoji[1] == ("👍🏿", 5, "thumbs up dark skin tone")

spacymoji only cares about the token text, so you can use it on a blank Language instance (it should work for all available languages!), or in a pipeline with a loaded pipeline. If your pipeline includes a tagger, parser and entity recognizer, make sure to add the emoji component as first=True, so the spans are merged right after tokenization, and before the document is parsed. If your text contains a lot of emoji, this might even give you a nice boost in parser accuracy.

Available attributes

The extension sets attributes on the Doc, Span and Token. You can change the attribute names (and other parameters of the Emoji component) by passing them via the config parameter in the nlp.add_pipe(...) method. For more details on custom components and attributes, see the processing pipelines documentation.

Attribute Type Description
Token._.is_emoji bool Whether the token is an emoji.
Token._.emoji_desc str A human-readable description of the emoji.
Doc._.has_emoji bool Whether the document contains emoji.
Doc._.emoji List[Tuple[str, int, str]] (emoji, index, description) tuples of the document's emoji.
Span._.has_emoji bool  Whether the span contains emoji.
Span._.emoji List[Tuple[str, int, str]] (emoji, index, description) tuples of the span's emoji.


You can configure the emoji factory by setting any of the following parameters in the config dictionary:

Setting Type Description
attrs Tuple[str, str, str, str] Attributes to set on the ._ property. Defaults to ('has_emoji', 'is_emoji', 'emoji_desc', 'emoji').
pattern_id str ID of match pattern, defaults to 'EMOJI'. Can be changed to avoid ID conflicts.
merge_spans bool Merge spans containing multi-character emoji, defaults to True. Will only merge combined emoji resulting in one icon, not sequences.
lookup Dict[str, str] Optional lookup table that maps emoji strings to custom descriptions, e.g. translations or other annotations.
emoji_config = {"attrs": ("has_e", "is_e", "e_desc", "e"), lookup={"👨‍🎤": "David Bowie"})
nlp.add_pipe(emoji, first=True, config=emoji_config)
doc = nlp("We can be 👨‍🎤 heroes")
assert doc[3]._.is_e
assert doc[3]._.e_desc == "David Bowie"

If you're training a pipeline, you can define the component config in your config.cfg:

pipeline = ["emoji", "ner"]
# ...

factory = "emoji"
merge_spans = false

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

spacymoji-3.1.0.tar.gz (9.0 kB view hashes)

Uploaded source

Built Distribution

spacymoji-3.1.0-py2.py3-none-any.whl (8.5 kB view hashes)

Uploaded py2 py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page