13831530637

Committed 13 Mar 2025 08:44AM UTC coverage: 44.548% (-5.2%) from 49.787%

Build # 13831530637

Build Type

push

github

Committed by

BenMorel

Commit Message

Merge WKTParser & EWKTParser, add WKTTokenType

Coverage Stats

31 of 32 new or added lines in 3 files covered. (96.88%)

208 existing lines in 11 files now uncovered.

1573 of 3531 relevant lines covered (44.55%)

298.16 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

84.27

/src/IO/WKTParser.php

<?php

declare(strict_types=1);

namespace Brick\Geo\IO;

use Brick\Geo\Exception\GeometryIOException;

/**
 * Well-Known Text parser, with support for EWKT.
 *
 * @internal
 */
final class WKTParser
{
    private const string REGEX_CAPTURE_WORD   = '([a-z]+)';
    private const string REGEX_CAPTURE_NUMBER = '(\-?[0-9]+(?:\.[0-9]+)?(?:e[\+\-]?[0-9]+)?)';
    private const string REGEX_WHITESPACE = '\s+';
    private const string REGEX_CAPTURE_OTHER = '(.+?)';
    private const string REGEX_CAPTURE_SRID = 'SRID\=([0-9]+)\s*;'; // EWKT

    /**
     * The list of tokens.
     *
     * The first element of each token is the token type, the second element is the token value.
     *
     * @var list<array{WKTTokenType, string}>
     */
    private array $tokens = [];

    /**
     * The current token pointer.
     */
    private int $current = 0;

    public function __construct(
        string $wkt,
        private bool $ewkt = false,
    ) {
        $this->scan($wkt);
    }

    private function scan(string $wkt) : void
    {
        $regexPatterns = [
            self::REGEX_CAPTURE_WORD,
            self::REGEX_CAPTURE_NUMBER,
            self::REGEX_WHITESPACE,
            self::REGEX_CAPTURE_OTHER,
        ];

        if ($this->ewkt) {
            array_unshift($regexPatterns, self::REGEX_CAPTURE_SRID);
        }

        $matchKeyToType = $this->ewkt ? [
            1 => WKTTokenType::SRID,
            2 => WKTTokenType::Word,
            3 => WKTTokenType::Number,
            4 => WKTTokenType::Other,
        ] : [
            1 => WKTTokenType::Word,
            2 => WKTTokenType::Number,
            3 => WKTTokenType::Other,
        ];

        $regex = '/' . implode('|', $regexPatterns) . '/i';

        preg_match_all($regex, $wkt, $matches, PREG_SET_ORDER);

        foreach ($matches as $match) {
            foreach ($match as $key => $value) {
                /** @var int $key */

                if ($key === 0) {
                    continue;
                }

                if ($value !== '') {
                    $this->tokens[] = [$matchKeyToType[$key], $value];
                }
            }
        }
    }

    /**
     * @return array{WKTTokenType, string}|null The next token, or null if there are no more tokens.
     */
    private function peekToken(): ?array
    {
        return $this->tokens[$this->current] ?? null;
    }

    /**
     * @return array{WKTTokenType, string}|null The next token, or null if there are no more tokens.
     */
    private function nextToken() : ?array
    {
        $token = $this->peekToken();

        if ($token === null) {
            return null;
        }

        $this->current++;

        return $token;
    }

    /**
     * @throws GeometryIOException
     */
    public function matchOpener() : void
    {
        $token = $this->nextToken();

        if ($token === null) {
            throw new GeometryIOException("Expected '(' but encountered end of stream");
        }
        if ($token[1] !== '(') {
            throw new GeometryIOException("Expected '(' but encountered '" . $token[1] . "'");
        }
    }

    /**
     * @throws GeometryIOException
     */
    public function matchCloser() : void
    {
        $token = $this->nextToken();

        if ($token === null) {
            throw new GeometryIOException("Expected ')' but encountered end of stream");
        }
        if ($token[1] !== ')') {
            throw new GeometryIOException("Expected ')' but encountered '" . $token[1] . "'");
        }
    }

    /**
     * @throws GeometryIOException
     */
    public function getNextWord() : string
    {
        $token = $this->nextToken();

        if ($token === null) {
            throw new GeometryIOException("Expected word but encountered end of stream");
        }
        if ($token[0] !== WKTTokenType::Word) {
            throw new GeometryIOException("Expected word but encountered '" . $token[1] . "'");
        }

        return $token[1];
    }

    /**
     * @return string|null The next word, or NULL if the next token is not a word, or there are no more tokens.
     */
    public function getOptionalNextWord() : ?string
    {
        $token = $this->tokens[$this->current] ?? null;

        if ($token === null) {
            return null;
        }

        if ($token[0] !== WKTTokenType::Word) {
            return null;
        }

        $this->current++;

        return $token[1];
    }

    public function matchOptionalOpener(): bool
    {
        $token = $this->peekToken();

        $isOpener = ($token !== null && $token[1] === '(');

        if ($isOpener) {
            $this->current++;
        }

        return $isOpener;
    }

    /**
     * Returns whether the next token is an opener or a word.
     *
     * @return bool True if the next token is an opener, false if it is a word.
     *
     * @throws GeometryIOException If the next token is not an opener or a word, or if there is no next token.
     */
    public function isNextOpenerOrWord() : bool
    {
        $token = $this->tokens[$this->current] ?? null;

        if ($token === null) {
            throw new GeometryIOException("Expected '(' or word but encountered end of stream");
        }

        if ($token[1] === '(') {
            return true;
        }

        if ($token[0] === WKTTokenType::Word) {
            return false;
        }

        throw new GeometryIOException("Expected '(' or word but encountered '" . $token[1] . "'");
    }

    /**
     * @throws GeometryIOException
     */
    public function getNextNumber() : float
    {
        $token = $this->nextToken();

        if ($token === null) {
            throw new GeometryIOException("Expected number but encountered end of stream");
        }

        if ($token[0] !== WKTTokenType::Number) {
            throw new GeometryIOException("Expected number but encountered '" . $token[1] . "'");
        }

        return (float) $token[1];
    }

    /**
     * @throws GeometryIOException
     */
    public function getNextCloserOrComma() : string
    {
        $token = $this->nextToken();

        if ($token === null) {
            throw new GeometryIOException("Expected ')' or ',' but encountered end of stream");
        }
        if ($token[1] !== ')' && $token[1] !== ',') {
            throw new GeometryIOException("Expected ')' or ',' but encountered '" . $token[1] . "'");
        }

        return $token[1];
    }

    public function getOptionalSRID() : int
    {
        $token = $this->tokens[$this->current] ?? null;

        if ($token === null) {
            return 0;
        }

        if ($token[0] !== WKTTokenType::SRID) {
            return 0;
        }

        $this->current++;

        return (int) $token[1];
    }

    public function isEndOfStream() : bool
    {
        return $this->nextToken() === null;
    }
}

1	<?php
2
3	declare(strict_types=1);
4
5	namespace Brick\Geo\IO;
6
7	use Brick\Geo\Exception\GeometryIOException;
8
9	/**
10	* Well-Known Text parser, with support for EWKT.
11	*
12	* @internal
13	*/
14	final class WKTParser
15	{
16	private const string REGEX_CAPTURE_WORD = '([a-z]+)';
17	private const string REGEX_CAPTURE_NUMBER = '(\-?[0-9]+(?:\.[0-9]+)?(?:e[\+\-]?[0-9]+)?)';
18	private const string REGEX_WHITESPACE = '\s+';
19	private const string REGEX_CAPTURE_OTHER = '(.+?)';
20	private const string REGEX_CAPTURE_SRID = 'SRID\=([0-9]+)\s*;'; // EWKT
21
22	/**
23	* The list of tokens.
24	*
25	* The first element of each token is the token type, the second element is the token value.
26	*
27	* @var list<array{WKTTokenType, string}>
28	*/
29	private array $tokens = [];
30
31	/**
32	* The current token pointer.
33	*/
34	private int $current = 0;
35
36	public function __construct(
37	string $wkt,
38	private bool $ewkt = false,
39	) {
40	$this->scan($wkt);	4,350✔
41	}
42
43	private function scan(string $wkt) : void
44	{
45	$regexPatterns = [	4,350✔
46	self::REGEX_CAPTURE_WORD,	4,350✔
47	self::REGEX_CAPTURE_NUMBER,	4,350✔
48	self::REGEX_WHITESPACE,	4,350✔
49	self::REGEX_CAPTURE_OTHER,	4,350✔
50	];	4,350✔
51
52	if ($this->ewkt) {	4,350✔
53	array_unshift($regexPatterns, self::REGEX_CAPTURE_SRID);	1,392✔
54	}
55
56	$matchKeyToType = $this->ewkt ? [	4,350✔
57	1 => WKTTokenType::SRID,	4,350✔
58	2 => WKTTokenType::Word,	4,350✔
59	3 => WKTTokenType::Number,	4,350✔
60	4 => WKTTokenType::Other,	4,350✔
61	] : [	2,958✔
62	1 => WKTTokenType::Word,	2,958✔
63	2 => WKTTokenType::Number,	2,958✔
64	3 => WKTTokenType::Other,	2,958✔
65	];	2,958✔
66
67	$regex = '/' . implode('\|', $regexPatterns) . '/i';	4,350✔
68
69	preg_match_all($regex, $wkt, $matches, PREG_SET_ORDER);	4,350✔
70
71	foreach ($matches as $match) {	4,350✔
72	foreach ($match as $key => $value) {	4,350✔
73	/** @var int $key */
74
75	if ($key === 0) {	4,350✔
76	continue;	4,350✔
77	}
78
79	if ($value !== '') {	4,350✔
80	$this->tokens[] = [$matchKeyToType[$key], $value];	4,350✔
81	}
82	}
83	}
84	}
85
86	/**
87	* @return array{WKTTokenType, string}\|null The next token, or null if there are no more tokens.
88	*/
89	private function peekToken(): ?array
90	{
91	return $this->tokens[$this->current] ?? null;	4,350✔
92	}
93
94	/**
95	* @return array{WKTTokenType, string}\|null The next token, or null if there are no more tokens.
96	*/
97	private function nextToken() : ?array
98	{
99	$token = $this->peekToken();	4,350✔
100
101	if ($token === null) {	4,350✔
102	return null;	4,336✔
103	}
104
105	$this->current++;	4,350✔
106
107	return $token;	4,350✔
108	}
109
110	/**
111	* @throws GeometryIOException
112	*/
113	public function matchOpener() : void
114	{
115	$token = $this->nextToken();	2,834✔
116
117	if ($token === null) {	2,834✔
UNCOV 118	throw new GeometryIOException("Expected '(' but encountered end of stream");	×
119	}
120	if ($token[1] !== '(') {	2,834✔
UNCOV 121	throw new GeometryIOException("Expected '(' but encountered '" . $token[1] . "'");	×
122	}
123	}
124
125	/**
126	* @throws GeometryIOException
127	*/
128	public function matchCloser() : void
129	{
130	$token = $this->nextToken();	484✔
131
132	if ($token === null) {	484✔
UNCOV 133	throw new GeometryIOException("Expected ')' but encountered end of stream");	×
134	}
135	if ($token[1] !== ')') {	484✔
UNCOV 136	throw new GeometryIOException("Expected ')' but encountered '" . $token[1] . "'");	×
137	}
138	}
139
140	/**
141	* @throws GeometryIOException
142	*/
143	public function getNextWord() : string
144	{
145	$token = $this->nextToken();	4,350✔
146
147	if ($token === null) {	4,350✔
UNCOV 148	throw new GeometryIOException("Expected word but encountered end of stream");	×
149	}
150	if ($token[0] !== WKTTokenType::Word) {	4,350✔
UNCOV 151	throw new GeometryIOException("Expected word but encountered '" . $token[1] . "'");	×
152	}
153
154	return $token[1];	4,350✔
155	}
156
157	/**
158	* @return string\|null The next word, or NULL if the next token is not a word, or there are no more tokens.
159	*/
160	public function getOptionalNextWord() : ?string
161	{
162	$token = $this->tokens[$this->current] ?? null;	4,350✔
163
164	if ($token === null) {	4,350✔
UNCOV 165	return null;	×
166	}
167
168	if ($token[0] !== WKTTokenType::Word) {	4,350✔
169	return null;	2,834✔
170	}
171
172	$this->current++;	3,186✔
173
174	return $token[1];	3,186✔
175	}
176
177	public function matchOptionalOpener(): bool
178	{
179	$token = $this->peekToken();	2,528✔
180
181	$isOpener = ($token !== null && $token[1] === '(');	2,528✔
182
183	if ($isOpener) {	2,528✔
184	$this->current++;	24✔
185	}
186
187	return $isOpener;	2,528✔
188	}
189
190	/**
191	* Returns whether the next token is an opener or a word.
192	*
193	* @return bool True if the next token is an opener, false if it is a word.
194	*
195	* @throws GeometryIOException If the next token is not an opener or a word, or if there is no next token.
196	*/
197	public function isNextOpenerOrWord() : bool
198	{
199	$token = $this->tokens[$this->current] ?? null;	402✔
200
201	if ($token === null) {	402✔
UNCOV 202	throw new GeometryIOException("Expected '(' or word but encountered end of stream");	×
203	}
204
205	if ($token[1] === '(') {	402✔
206	return true;	394✔
207	}
208
209	if ($token[0] === WKTTokenType::Word) {	346✔
210	return false;	346✔
211	}
212
UNCOV 213	throw new GeometryIOException("Expected '(' or word but encountered '" . $token[1] . "'");	×
214	}
215
216	/**
217	* @throws GeometryIOException
218	*/
219	public function getNextNumber() : float
220	{
221	$token = $this->nextToken();	2,812✔
222
223	if ($token === null) {	2,812✔
224	throw new GeometryIOException("Expected number but encountered end of stream");	×
225	}
226
227	if ($token[0] !== WKTTokenType::Number) {	2,812✔
UNCOV 228	throw new GeometryIOException("Expected number but encountered '" . $token[1] . "'");	×
229	}
230
231	return (float) $token[1];	2,812✔
232	}
233
234	/**
235	* @throws GeometryIOException
236	*/
237	public function getNextCloserOrComma() : string
238	{
239	$token = $this->nextToken();	2,560✔
240
241	if ($token === null) {	2,560✔
UNCOV 242	throw new GeometryIOException("Expected ')' or ',' but encountered end of stream");	×
243	}
244	if ($token[1] !== ')' && $token[1] !== ',') {	2,560✔
UNCOV 245	throw new GeometryIOException("Expected ')' or ',' but encountered '" . $token[1] . "'");	×
246	}
247
248	return $token[1];	2,560✔
249	}
250
251	public function getOptionalSRID() : int
252	{
253	$token = $this->tokens[$this->current] ?? null;	1,392✔
254
255	if ($token === null) {	1,392✔
NEW 256	return 0;	×
257	}
258
259	if ($token[0] !== WKTTokenType::SRID) {	1,392✔
260	return 0;	696✔
261	}
262
263	$this->current++;	696✔
264
265	return (int) $token[1];	696✔
266	}
267
268	public function isEndOfStream() : bool
269	{
270	return $this->nextToken() === null;	4,336✔
271	}
272	}

brick / geo / 13831530637

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous