chemdataextractor ---- Chemistry Articles NLP Library

 1
 2
 3
 4
 5
 6
 7
 8
 9
10

class Pressure(QuantityModel):
    """ A boiling point property"""
    specifier = StringType(parse_expression=(I('Boiling') + I('Point')).add_action(join), required=True)
    compound = ModelType(Compound)

class BoilingPoint(TemperatureModel):
    """ A boiling point property"""
    specifier = StringType(parse_expression=(I('Boiling') + I('Point')).add_action(join), required=True)
    compound = ModelType(Compound, binding=True)
    pressure = ModelType(Pressure, contextual=True)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45

def construct_unit_element(dimensions):
    """
    Construct an element for detecting units for the dimensions given.
    Any magnitude modifiers (e.g. kilo) will be automatically handled.

    :param Dimension dimensions: The dimensions that the element produced will look for.
    :returns: An Element to look for units of given dimensions. If None or Dimensionless are passed in, returns None.
    :rtype: BaseParserElement or None
    """
    if not dimensions or not dimensions.units_dict:
        return None

    #------------------------- 单个 单位的处理 -----------------------
    # Handle all the magnitudes
    # 数量关系（kilo）等规则
    units_regex = '^(('
    for element in magnitudes_dict.keys():
        units_regex += '(' + element.pattern + ')|'
        units_regex = units_regex[:-1]
        units_regex += ')?'
        units_regex += '('
        # Case where we have a token that's just brackets

        # 多余 =括号= 和 =-= 处理
    units_regex += r'((\(|\[))|((\)|\]))|\-|'

    # Handle all the units
    # 单位符号的加入
    for element in dimensions.units_dict:
        units_regex += '(' + element.pattern + ')|'
        units_regex += r'(\/)'
        # Case when we have powers, or one or more units

    # 符号的 幂运算 规则 <--- 完成的
    # ---> kilo meter, /, +3.2, meter, m
    units_regex2 = units_regex + r'|([\+\-–−]?\d+(\.\d+)?)'
    units_regex2 += '))+$'

    # 不包括 数字 规则  <---- 完成的， kilo meter, meter, /
    units_regex += '))+'

    # ~(不包含数字） + (包含数字）*~ ,即 (前后挨在一起的情况), eg: m / s , m, m 2
    units_regex += (units_regex2[1:-2] + '*')
    units_regex += '$'
    return (R(pattern=units_regex) + ZeroOrMore(R(pattern=units_regex) | R(pattern=units_regex2))).add_action(merge)

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21

@memoize
def value_element_plain():
    """
    Returns an element similar to value_element but without any units.

    :returns: An Element to look for values.
    :rtype: BaseParserElement
    """
    fraction = R('^[\+\-–−]?\d+/\d+$') | (R('^[\+\-–−]?\d+$') + R('^/$') + R('^\d+$')).add_action(merge)
    pure_number = R('^[\+\-–−]?\d+(([\.・,\d])+)?$')
    number = fraction | pure_number
    joined_range = R('^[\+\-–−]?\d+(([\.・,\d])+)?[\-–−~∼˜]\d+(([\.・,\d])+)?$')('raw_value').add_action(merge)
    spaced_range = (number + R('^[\-–−~∼˜]$') + number)('raw_value').add_action(merge)
    to_range = (number + I('to') + number)('raw_value').add_action(join)
    plusminus_range = (number + R('±') + number)('raw_value').add_action(join)
    bracket_range = R(pure_number.pattern[:-1] + '\(\d+\)' + '$')('raw_value')
    between_range = (I('between').hide() + number + I('and') + number).add_action(join)
    value_range = (Optional(R('^[\-–−]$')) + (plusminus_range | joined_range | spaced_range | to_range | between_range | bracket_range))('raw_value').add_action(merge)
    value_single = (Optional(R('^[~∼˜\<\>]$')) + Optional(R('^[\-–−]$')) + number)('raw_value').add_action(merge)
    value = Optional(lbrct).hide() + (value_range | value_single)('raw_value') + Optional(rbrct).hide()
    return value

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19

def value_element(units=(OneOrMore(T('NN')) | OneOrMore(T('NNP')) | OneOrMore(T('NNPS')) | OneOrMore(T('NNS')))('raw_units').add_action(merge)):
    """
    Returns an Element for values with given units. By default, uses tags to guess that a unit exists.

    :param BaseParserElement units: (Optional) A parser element for the units that are to be looked for. Default option looks for nouns.
    :returns: An Element to look for values and units.
    :rtype: BaseParserElement
    """
    number = R('^[\+\-–−]?\d+(([\.・,\d])+)?$')
    joined_range = R('^[\+\-–−]?\d+(([\.・,\d])+)?[\-–−~∼˜]\d+(([\.・,\d])+)?$')('raw_value').add_action(merge)
    spaced_range = (number + Optional(units).hide() + (R('^[\-–−~∼˜]$') + number | number))('raw_value').add_action(merge)
    to_range = (number + Optional(units).hide() + I('to') + number)('raw_value').add_action(join)
    plusminus_range = (number + R('±') + number)('raw_value').add_action(join)
    bracket_range = R(number.pattern[:-1] + '\(\d+\)' + '$')('raw_value')
    between_range = (I('between').hide() + number + I('and') + number).add_action(join)
    value_range = (Optional(R('^[\-–−]$')) + (bracket_range | plusminus_range | joined_range | spaced_range | to_range | between_range))('raw_value').add_action(merge)
    value_single = (Optional(R('^[~∼˜\<\>]$')) + Optional(R('^[\-–−]$')) + number)('raw_value').add_action(merge)
    value = Optional(lbrct).hide() + (value_range | value_single)('raw_value') + Optional(rbrct).hide()
    return value + units

1
2
3
4
5
6

# self.models = [MolModel] -->
records: [{'MolModel': {'raw_value': '0.2', 'raw_units': 'mmol', 'value': [0.2], 'units': 'MilliMol^(1.0)', 'specifier': 'substance amount', 'compound': {'Compound': {'names': ['PbX2']}}, 'fake_field': 'of'}}]

# self.models = [Compound, MolModel] -->
self.models = [Compound, MolModel] 结果不一致
records: [{'Compound': {'names': ['PbX2']}}, {'Compound': {'names': ['Br']}}, {'Compound': {'names': ['CH3NH3X']}}, {'Compound': {'labels': ['I']}}, {'MolModel': {'raw_value': '0.2', 'raw_units': 'mmol', 'value': [0.2], 'units': 'MilliMol^(1.0)', 'specifier': 'substance amount', 'compound': {'Compound': {'names': ['PbX2']}}, 'fake_field': 'of'}}]

1
2
3
4
5
6
7
8

if model is QuantityModel and model is Dimensionless:
    ...
elif ...:
    ...

combined_entity_list = entities_1 | entities_2 | ....

root_phrase = OneOrMore(combined_entities + Optional(SkipTo(combined_entities)))('root_phrase')