keyword-extraction/kwextraction/extraction.py

# -*- coding: utf-8 -*-

import json
import pkgutil

_file = pkgutil.get_data(__package__, 'equipment_info.json')
_equipment_data = json.loads(_file)


def extractor(text_list):
    for text in text_list:
        result = dict()
        result.update({'str': text})
        for (name, params) in _equipment_data.items():
            if name in text:
                result.update({'equipment': name})
                for param in params:
                    if param in text:
                        result.update({'param': param})

        if 'param' not in result.keys():
            result.update({'param': ''})

        yield result