abydos.distance._synoname._synoname_word_approximation()

Complexity

Conditions

73

Size

Total Lines	202
Code Lines	140

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	103
CRAP Score	73.0047

Importance

Changes

0

# -*- coding: utf-8 -*-
        for n, val in enumerate(
            (
                'exact',
                'omission',
                'substitution',
                'transposition',
                'punctuation',
                'initials',
                'extension',
                'inclusion',
                'no_first',
                'word_approx',
                'confusions',
                'char_approx',
            )
        )
    }
    match_name = (
        '',
        'exact',
        'omission',
        'substitution',
        'transposition',
        'punctuation',
        'initials',
        'extension',
        'inclusion',
        'no_first',
        'word_approx',
        'confusions',
        'char_approx',
        'no_match',
    )
    match_type_dict = {val: n for n, val in enumerate(match_name)}

    def _synoname_strip_punct(self, word):

This method could be written as a function/class method.

        """Return a word with punctuation stripped out.

        Parameters
        ----------
        word : str
            A word to strip punctuation from

        Returns
        -------
        str
            The word stripped of punctuation

        Examples
        --------
        >>> pe = Synoname()
        >>> pe._synoname_strip_punct('AB;CD EF-GH$IJ')
        'ABCD EFGHIJ'

        """
        stripped = ''
        for char in word:
            if char not in set(',-./:;"&\'()!{|}?$%*+<=>[\\]^_`~'):
                stripped += char
        return stripped.strip()

    def _synoname_word_approximation(

Too many arguments (6/5)

This function exceeds the maximum number of variables (32/15).

Too many return statements (10/6)

        self, src_ln, tar_ln, src_fn='', tar_fn='', features=None

Wrong hanging indentation before block (add 4 spaces).

    ):
        """Return the Synoname word approximation score for two names.

        Parameters
        ----------
        src_ln : str
            Last name of the source
        tar_ln : str
            Last name of the target
        src_fn : str
            First name of the source (optional)
        tar_fn : str
            First name of the target (optional)
        features : dict
            A dict containing special features calculated using
            :py:class:`fingerprint.SynonameToolcode` (optional)

        Returns
        -------
        float
            The word approximation score

        Examples
        --------
        >>> pe = Synoname()
        >>> pe._synoname_word_approximation('Smith Waterman', 'Waterman',
        ... 'Tom Joe Bob', 'Tom Joe')
        0.6

        """
        if features is None:
            features = {}
        if 'src_specials' not in features:
            features['src_specials'] = []
        if 'tar_specials' not in features:
            features['tar_specials'] = []

        src_len_specials = len(features['src_specials'])
        tar_len_specials = len(features['tar_specials'])

        # 1
        if ('gen_conflict' in features and features['gen_conflict']) or (
            'roman_conflict' in features and features['roman_conflict']

Wrong hanging indentation before block (add 4 spaces).

        ):
            return 0

        # 3 & 7
        full_tar1 = ' '.join((tar_ln, tar_fn)).replace('-', ' ').strip()
        for s_pos, s_type in features['tar_specials']:
            if s_type == 'a':
                full_tar1 = full_tar1[
                    : -(1 + len(self.stc.synoname_special_table[s_pos][1]))
                ]
            elif s_type == 'b':
                loc = (
                    full_tar1.find(
                        ' ' + self.stc.synoname_special_table[s_pos][1] + ' '
                    )
                    + 1
                )
                full_tar1 = (
                    full_tar1[:loc]
                    + full_tar1[
                        loc + len(self.stc.synoname_special_table[s_pos][1]) :
                    ]
                )
            elif s_type == 'c':
                full_tar1 = full_tar1[
                    1 + len(self.stc.synoname_special_table[s_pos][1]) :
                ]

        full_src1 = ' '.join((src_ln, src_fn)).replace('-', ' ').strip()
        for s_pos, s_type in features['src_specials']:
            if s_type == 'a':
                full_src1 = full_src1[
                    : -(1 + len(self.stc.synoname_special_table[s_pos][1]))
                ]
            elif s_type == 'b':
                loc = (
                    full_src1.find(
                        ' ' + self.stc.synoname_special_table[s_pos][1] + ' '
                    )
                    + 1
                )
                full_src1 = (
                    full_src1[:loc]
                    + full_src1[
                        loc + len(self.stc.synoname_special_table[s_pos][1]) :
                    ]
                )
            elif s_type == 'c':
                full_src1 = full_src1[
                    1 + len(self.stc.synoname_special_table[s_pos][1]) :
                ]

        full_tar2 = full_tar1
        for s_pos, s_type in features['tar_specials']:
            if s_type == 'd':
                full_tar2 = full_tar2[
                    len(self.stc.synoname_special_table[s_pos][1]) :
                ]
            elif (
                s_type == 'X'

Wrong hanging indentation before block (add 4 spaces).

                and self.stc.synoname_special_table[s_pos][1] in full_tar2

Wrong hanging indentation before block (add 4 spaces).

            ):
                loc = full_tar2.find(
                    ' ' + self.stc.synoname_special_table[s_pos][1]
                )
                full_tar2 = (
                    full_tar2[:loc]
                    + full_tar2[
                        loc + len(self.stc.synoname_special_table[s_pos][1]) :
                    ]
                )

        full_src2 = full_src1
        for s_pos, s_type in features['src_specials']:
            if s_type == 'd':
                full_src2 = full_src2[
                    len(self.stc.synoname_special_table[s_pos][1]) :
                ]
            elif (
                s_type == 'X'

Wrong hanging indentation before block (add 4 spaces).

                and self.stc.synoname_special_table[s_pos][1] in full_src2

Wrong hanging indentation before block (add 4 spaces).

            ):
                loc = full_src2.find(
                    ' ' + self.stc.synoname_special_table[s_pos][1]
                )
                full_src2 = (
                    full_src2[:loc]
                    + full_src2[
                        loc + len(self.stc.synoname_special_table[s_pos][1]) :
                    ]
                )

        full_tar1 = self._synoname_strip_punct(full_tar1)
        tar1_words = full_tar1.split()
        tar1_num_words = len(tar1_words)

        full_src1 = self._synoname_strip_punct(full_src1)