Parser::getFirstSegmentParser() - Code Metrics - Inspection of "Enhancement for company names, titles and lastname..." - theiconic/name-parser - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#40)

unknown

created 2019-12-16 13:36 UTC

Parser::getFirstSegmentParser() A

↳ Parent: Parser

Complexity

Conditions	1
Paths	1

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
dl	0
loc	17
rs	9.7
c	0
b	0
f	0
cc	1
nc	1
nop	0

<?php

namespace TheIconic\NameParser;

use TheIconic\NameParser\Language\English;
use TheIconic\NameParser\Mapper\NicknameMapper;
use TheIconic\NameParser\Mapper\SalutationMapper;
use TheIconic\NameParser\Mapper\SuffixMapper;
use TheIconic\NameParser\Mapper\InitialMapper;
use TheIconic\NameParser\Mapper\LastnameMapper;
use TheIconic\NameParser\Mapper\FirstnameMapper;
use TheIconic\NameParser\Mapper\MiddlenameMapper;
use TheIconic\NameParser\Mapper\CompanyMapper;
use TheIconic\NameParser\Mapper\ExtensionMapper;
use TheIconic\NameParser\Mapper\MultipartMapper;

class Parser
{
    /**
     * @var string
     */
    protected $whitespace = " \r\n\t";

    /**
     * @var array
     */
    protected $mappers = [];

    /**
     * @var array
     */
    protected $languages = [];

    /**
     * @var array
     */
    protected $nicknameDelimiters = [];

    /**
     * @var int
     */
    protected $maxSalutationIndex = 0;

    /**
     * @var int
     */
    protected $maxCombinedInitials = 2;

    public function __construct(array $languages = [])
    {
        if (empty($languages)) {
            $languages = [new English()];
        }

        $this->languages = $languages;
    }

    /**
     * split full names into the following parts:
     * - prefix / salutation  (Mr., Mrs., etc)
     * - given name / first name
     * - middle initials
     * - surname / last name
     * - suffix (II, Phd, Jr, etc)
     * - extension (Germany: nobility predicate is part of lastname)
     * - title (Germany: academic titles are usually used as name parts between salutation and given name)
     * - company (the string contains typical characteristics for a company name and is returned identically)
     *
     * @param string $name
     * @return Name
     */
    public function parse($name): Name
    {
        $name = $this->normalize($name);

        $segments = explode(',', $name);

        if (1 < count($segments)) {
            return $this->parseSplitName($segments[0], $segments[1], $segments[2] ?? '');
        } else {
            $mapped = $this->getCompany($name);
            if (count($mapped)) {
                return new Name($mapped);
            }
        }

        $parts = explode(' ', $name);

        foreach ($this->getMappers() as $mapper) {
            $parts = $mapper->map($parts);
        }

        return new Name($parts);
    }

    /**
     * handles split-parsing of comma-separated name parts
     *
     * @param string $first - the name part left of the comma
     * @param string $second - the name part right of the comma
     * @param string $third
     * @return Name
     */
    protected function parseSplitName($first, $second, $third): Name
    {
        $parts = array_merge(
            $this->getFirstSegmentParser()->parse($first)->getParts(),
            $this->getSecondSegmentParser()->parse($second)->getParts(),
            $this->getThirdSegmentParser()->parse($third)->getParts()
        );

        return new Name($parts);
    }

    /**
     * @return Parser
     */
    protected function getFirstSegmentParser(): Parser
    {
        $parser = new Parser();

        $parser->setMappers([
            new ExtensionMapper($this->getSamples('Extensions')),
            new MultipartMapper($this->getSamples('Titles'), 'title'),
            new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
            new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
            new SuffixMapper($this->getSamples('Suffixes'), false, 2),
            new LastnameMapper($this->getSamples('LastnamePrefixes'), true),
            new FirstnameMapper(),
            new MiddlenameMapper(),
        ]);

        return $parser;
    }

    /**
     * @return Parser
     */
    protected function getSecondSegmentParser(): Parser
    {
        $parser = new Parser();

        $parser->setMappers([
            new ExtensionMapper($this->getSamples('Extensions')),
            new MultipartMapper($this->getSamples('Titles'), 'title'),
            new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
            new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
            new SuffixMapper($this->getSamples('Suffixes'), true, 1),
            new NicknameMapper($this->getNicknameDelimiters()),
            new InitialMapper($this->getMaxCombinedInitials(), true),
            new FirstnameMapper(),
            new MiddlenameMapper(true),
        ]);

        return $parser;
    }

    protected function getThirdSegmentParser(): Parser
    {
        $parser = new Parser();

        $parser->setMappers([
            new SuffixMapper($this->getSamples('Suffixes'), true, 0),
        ]);

        return $parser;
    }

    /**
     * get the mappers for this parser
     *
     * @return array
     */
    public function getMappers(): array
    {
        if (empty($this->mappers)) {
            $this->setMappers([
                new ExtensionMapper($this->getSamples('Extensions')),
                new MultipartMapper($this->getSamples('Titles'), 'title'),
                new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
                new NicknameMapper($this->getNicknameDelimiters()),
                new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
                new SuffixMapper($this->getSamples('Suffixes')),
                new InitialMapper($this->getMaxCombinedInitials()),
                new LastnameMapper($this->getSamples('LastnamePrefixes')),
                new FirstnameMapper(),
                new MiddlenameMapper(),
            ]);
        }

        return $this->mappers;
    }

    /**
     * get name as company if parts matches company identifiers
     *
     * @param string $name
     * @return array
     */
    protected function getCompany(string $name): array
    {
        $mapper = new CompanyMapper($this->getSamples('Companies'));

        return $mapper->map([$name]);
    }

    /**
     * set the mappers for this parser
     *
     * @param array $mappers
     * @return Parser
     */
    public function setMappers(array $mappers): Parser
    {
        $this->mappers = $mappers;

        return $this;
    }

    /**
     * normalize the name
     *
     * @param string $name
     * @return string
     */
    protected function normalize(string $name): string
    {
        $whitespace = $this->getWhitespace();

        $name = trim($name);

        return preg_replace('/[' . preg_quote($whitespace) . ']+/', ' ', $name);
    }

    /**
     * get a string of characters that are supposed to be treated as whitespace
     *
     * @return string
     */
    public function getWhitespace(): string
    {
        return $this->whitespace;
    }

    /**
     * set the string of characters that are supposed to be treated as whitespace
     *
     * @param string $whitespace
     * @return Parser
     */
    public function setWhitespace($whitespace): Parser
    {
        $this->whitespace = $whitespace;

        return $this;
    }

    /**
     * @return array
     */
    protected function getSamples(string $sampleName): array
    {
        $samples = [];
        $method = sprintf('get%s', $sampleName);
        foreach ($this->languages as $language) {
            $samples += call_user_func_array([$language, $method], []);
        }

        return $samples;
    }

    /**
     * @return array
     */
    public function getNicknameDelimiters(): array
    {
        return $this->nicknameDelimiters;
    }

    /**
     * @param array $nicknameDelimiters
     * @return Parser
     */
    public function setNicknameDelimiters(array $nicknameDelimiters): Parser
    {
        $this->nicknameDelimiters = $nicknameDelimiters;

        return $this;
    }

    /**
     * @return int
     */
    public function getMaxSalutationIndex(): int
    {
        return $this->maxSalutationIndex;
    }

    /**
     * @param int $maxSalutationIndex
     * @return Parser
     */
    public function setMaxSalutationIndex(int $maxSalutationIndex): Parser
    {
        $this->maxSalutationIndex = $maxSalutationIndex;

        return $this;
    }

    /**
     * @return int
     */
    public function getMaxCombinedInitials(): int
    {
        return $this->maxCombinedInitials;
    }

    /**
     * @param int $maxCombinedInitials
     * @return Parser
     */
    public function setMaxCombinedInitials(int $maxCombinedInitials): Parser
    {
        $this->maxCombinedInitials = $maxCombinedInitials;

        return $this;
    }
}


1			<?php
2
3			namespace TheIconic\NameParser;
4
5			use TheIconic\NameParser\Language\English;
6			use TheIconic\NameParser\Mapper\NicknameMapper;
7			use TheIconic\NameParser\Mapper\SalutationMapper;
8			use TheIconic\NameParser\Mapper\SuffixMapper;
9			use TheIconic\NameParser\Mapper\InitialMapper;
10			use TheIconic\NameParser\Mapper\LastnameMapper;
11			use TheIconic\NameParser\Mapper\FirstnameMapper;
12			use TheIconic\NameParser\Mapper\MiddlenameMapper;
13			use TheIconic\NameParser\Mapper\CompanyMapper;
14			use TheIconic\NameParser\Mapper\ExtensionMapper;
15			use TheIconic\NameParser\Mapper\MultipartMapper;
16
17			class Parser
18			{
19			/**
20			* @var string
21			*/
22			protected $whitespace = " \r\n\t";
23
24			/**
25			* @var array
26			*/
27			protected $mappers = [];
28
29			/**
30			* @var array
31			*/
32			protected $languages = [];
33
34			/**
35			* @var array
36			*/
37			protected $nicknameDelimiters = [];
38
39			/**
40			* @var int
41			*/
42			protected $maxSalutationIndex = 0;
43
44			/**
45			* @var int
46			*/
47			protected $maxCombinedInitials = 2;
48
49			public function __construct(array $languages = [])
50			{
51			if (empty($languages)) {
52			$languages = [new English()];
53			}
54
55			$this->languages = $languages;
56			}
57
58			/**
59			* split full names into the following parts:
60			* - prefix / salutation (Mr., Mrs., etc)
61			* - given name / first name
62			* - middle initials
63			* - surname / last name
64			* - suffix (II, Phd, Jr, etc)
65			* - extension (Germany: nobility predicate is part of lastname)
66			* - title (Germany: academic titles are usually used as name parts between salutation and given name)
67			* - company (the string contains typical characteristics for a company name and is returned identically)
68			*
69			* @param string $name
70			* @return Name
71			*/
72			public function parse($name): Name
73			{
74			$name = $this->normalize($name);
75
76			$segments = explode(',', $name);
77
78			if (1 < count($segments)) {
79			return $this->parseSplitName($segments[0], $segments[1], $segments[2] ?? '');
80			} else {
81			$mapped = $this->getCompany($name);
82			if (count($mapped)) {
83			return new Name($mapped);
84			}
85			}
86
87			$parts = explode(' ', $name);
88
89			foreach ($this->getMappers() as $mapper) {
90			$parts = $mapper->map($parts);
91			}
92
93			return new Name($parts);
94			}
95
96			/**
97			* handles split-parsing of comma-separated name parts
98			*
99			* @param string $first - the name part left of the comma
100			* @param string $second - the name part right of the comma
101			* @param string $third
102			* @return Name
103			*/
104			protected function parseSplitName($first, $second, $third): Name
105			{
106			$parts = array_merge(
107			$this->getFirstSegmentParser()->parse($first)->getParts(),
108			$this->getSecondSegmentParser()->parse($second)->getParts(),
109			$this->getThirdSegmentParser()->parse($third)->getParts()
110			);
111
112			return new Name($parts);
113			}
114
115			/**
116			* @return Parser
117			*/
118			protected function getFirstSegmentParser(): Parser
119			{
120			$parser = new Parser();
121
122			$parser->setMappers([
123			new ExtensionMapper($this->getSamples('Extensions')),
124			new MultipartMapper($this->getSamples('Titles'), 'title'),
125			new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
126			new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
127			new SuffixMapper($this->getSamples('Suffixes'), false, 2),
128			new LastnameMapper($this->getSamples('LastnamePrefixes'), true),
129			new FirstnameMapper(),
130			new MiddlenameMapper(),
131			]);
132
133			return $parser;
134			}
135
136			/**
137			* @return Parser
138			*/
139			protected function getSecondSegmentParser(): Parser
140			{
141			$parser = new Parser();
142
143			$parser->setMappers([
144			new ExtensionMapper($this->getSamples('Extensions')),
145			new MultipartMapper($this->getSamples('Titles'), 'title'),
146			new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
147			new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
148			new SuffixMapper($this->getSamples('Suffixes'), true, 1),
149			new NicknameMapper($this->getNicknameDelimiters()),
150			new InitialMapper($this->getMaxCombinedInitials(), true),
151			new FirstnameMapper(),
152			new MiddlenameMapper(true),
153			]);
154
155			return $parser;
156			}
157
158			protected function getThirdSegmentParser(): Parser
159			{
160			$parser = new Parser();
161
162			$parser->setMappers([
163			new SuffixMapper($this->getSamples('Suffixes'), true, 0),
164			]);
165
166			return $parser;
167			}
168
169			/**
170			* get the mappers for this parser
171			*
172			* @return array
173			*/
174			public function getMappers(): array
175			{
176			if (empty($this->mappers)) {
177			$this->setMappers([
178			new ExtensionMapper($this->getSamples('Extensions')),
179			new MultipartMapper($this->getSamples('Titles'), 'title'),
180			new MultipartMapper($this->getSamples('LastnamePrefixes'), 'prefix'),
181			new NicknameMapper($this->getNicknameDelimiters()),
182			new SalutationMapper($this->getSamples('Salutations'), $this->getMaxSalutationIndex()),
183			new SuffixMapper($this->getSamples('Suffixes')),
184			new InitialMapper($this->getMaxCombinedInitials()),
185			new LastnameMapper($this->getSamples('LastnamePrefixes')),
186			new FirstnameMapper(),
187			new MiddlenameMapper(),
188			]);
189			}
190
191			return $this->mappers;
192			}
193
194			/**
195			* get name as company if parts matches company identifiers
196			*
197			* @param string $name
198			* @return array
199			*/
200			protected function getCompany(string $name): array
201			{
202			$mapper = new CompanyMapper($this->getSamples('Companies'));
203
204			return $mapper->map([$name]);
205			}
206
207			/**
208			* set the mappers for this parser
209			*
210			* @param array $mappers
211			* @return Parser
212			*/
213			public function setMappers(array $mappers): Parser
214			{
215			$this->mappers = $mappers;
216
217			return $this;
218			}
219
220			/**
221			* normalize the name
222			*
223			* @param string $name
224			* @return string
225			*/
226			protected function normalize(string $name): string
227			{
228			$whitespace = $this->getWhitespace();
229
230			$name = trim($name);
231
232			return preg_replace('/[' . preg_quote($whitespace) . ']+/', ' ', $name);
233			}
234
235			/**
236			* get a string of characters that are supposed to be treated as whitespace
237			*
238			* @return string
239			*/
240			public function getWhitespace(): string
241			{
242			return $this->whitespace;
243			}
244
245			/**
246			* set the string of characters that are supposed to be treated as whitespace
247			*
248			* @param string $whitespace
249			* @return Parser
250			*/
251			public function setWhitespace($whitespace): Parser
252			{
253			$this->whitespace = $whitespace;
254
255			return $this;
256			}
257
258			/**
259			* @return array
260			*/
261			protected function getSamples(string $sampleName): array
262			{
263			$samples = [];
264			$method = sprintf('get%s', $sampleName);
265			foreach ($this->languages as $language) {
266			$samples += call_user_func_array([$language, $method], []);
267			}
268
269			return $samples;
270			}
271
272			/**
273			* @return array
274			*/
275			public function getNicknameDelimiters(): array
276			{
277			return $this->nicknameDelimiters;
278			}
279
280			/**
281			* @param array $nicknameDelimiters
282			* @return Parser
283			*/
284			public function setNicknameDelimiters(array $nicknameDelimiters): Parser
285			{
286			$this->nicknameDelimiters = $nicknameDelimiters;
287
288			return $this;
289			}
290
291			/**
292			* @return int
293			*/
294			public function getMaxSalutationIndex(): int
295			{
296			return $this->maxSalutationIndex;
297			}
298
299			/**
300			* @param int $maxSalutationIndex
301			* @return Parser
302			*/
303			public function setMaxSalutationIndex(int $maxSalutationIndex): Parser
304			{
305			$this->maxSalutationIndex = $maxSalutationIndex;
306
307			return $this;
308			}
309
310			/**
311			* @return int
312			*/
313			public function getMaxCombinedInitials(): int
314			{
315			return $this->maxCombinedInitials;
316			}
317
318			/**
319			* @param int $maxCombinedInitials
320			* @return Parser
321			*/
322			public function setMaxCombinedInitials(int $maxCombinedInitials): Parser
323			{
324			$this->maxCombinedInitials = $maxCombinedInitials;
325
326			return $this;
327			}
328			}
329

theiconic / name-parser

Pull Request — master (#40)

Parser::getFirstSegmentParser() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like