matcher

Matcher

Check token patterns for spaCy's rule-based Matcher against a text and return the matches in the text, as well as the individual tokens and whether they're part of a match. For a usage example, see the Rule-based Matcher Explorer demo.

Installation

pip install -r requirements.txt
python app.py

API

GET /models

Get a list of available models and their human-readable name, keyed by model name.

Example response

{
    "en_core_web_sm": "English - en_core_web_sm (v2.0.0)"
}

POST /match

Match a pattern and return the matches and tokens.

Example request

{
    "text": "A match is a tool for starting a fire. Typically, modern matches are made of small wooden sticks or stiff paper. ",
    "model":"en_core_web_sm",
    "pattern": [
        {
            "POS": "ADJ",
            "OP": "?"
        },
        {
            "LEMMA": "match",
            "POS": "NOUN"
        },
        {
            "LEMMA": "be"
        }
    ]
}

Name	Type	Description
text	string	The text to match on.
model	string	The statistical model to use for tokenization.
pattern	list	The token pattern to match. Each object in the list describes one token and is keyed by token attributes.

Example response

{
    "matches": [
        {
            "start": 2,
            "end": 10,
            "label": "MATCH"
        },
        {
            "start": 50,
            "end": 68,
            "label": "MATCH"
        }
    ],
    "tokens": [
        {
            "start": 0,
            "end": 1,
            "label": "TOKEN"
        },
        {
            "start": 2,
            "end": 7,
            "label": "MATCH"
        },
        {
            "start": 8,
            "end": 10,
            "label": "MATCH"
        },
        {
            "start": 11,
            "end": 12,
            "label": "TOKEN"
        },
        {
            "start": 13,
            "end": 17,
            "label": "TOKEN"
        },
        {
            "start": 18,
            "end": 21,
            "label": "TOKEN"
        },
        {
            "start": 22,
            "end": 30,
            "label": "TOKEN"
        },
        {
            "start": 31,
            "end": 32,
            "label": "TOKEN"
        },
        {
            "start": 33,
            "end": 37,
            "label": "TOKEN"
        },
        {
            "start": 37,
            "end": 38,
            "label": "TOKEN"
        },
        {
            "start": 39,
            "end": 48,
            "label": "TOKEN"
        },
        {
            "start": 48,
            "end": 49,
            "label": "TOKEN"
        },
        {
            "start": 50,
            "end": 56,
            "label": "MATCH"
        },
        {
            "start": 57,
            "end": 64,
            "label": "MATCH"
        },
        {
            "start": 65,
            "end": 68,
            "label": "MATCH"
        },
        {
            "start": 69,
            "end": 73,
            "label": "TOKEN"
        },
        {
            "start": 74,
            "end": 76,
            "label": "TOKEN"
        },
        {
            "start": 77,
            "end": 82,
            "label": "TOKEN"
        },
        {
            "start": 83,
            "end": 89,
            "label": "TOKEN"
        },
        {
            "start": 90,
            "end": 96,
            "label": "TOKEN"
        },
        {
            "start": 97,
            "end": 99,
            "label": "TOKEN"
        },
        {
            "start": 100,
            "end": 105,
            "label": "TOKEN"
        },
        {
            "start": 106,
            "end": 111,
            "label": "TOKEN"
        },
        {
            "start": 111,
            "end": 112,
            "label": "TOKEN"
        }
    ]
}

Name	Type	Description
matches	list	The matches in the text.
tokens	list	The individual tokens in the text and whether they're part of a match.
start	number	Character offset the match or token starts on.
end	number	Character offset the match or token ends after.
label	string	"MATCH" for matched span, "TOKEN" for token span.

Usage Example (JavaScript)

function getMatches(text, model, pattern) {
    const options = {
        method: 'POST',
        headers: { 'Accept': 'application/json', 'Content-Type': 'application/json' },
        credentials: 'same-origin',
        body: JSON.stringify({ text, model, pattern })
    };
    fetch('/match', options)
        .then(res => res.json())
        .then(({ tokens, matches }) => {
            console.log(tokens, matches);
        });
}