Rfc5322Validator::validateDotAtom() - Code Metrics - Inspection of "Chore: prepare for the 2.1.0 release" - stymiee/email-validator - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 82f564...b2a72e )

by John

created 2025-06-10 18:47 UTC

Rfc5322Validator::validateDotAtom() A

↳ Parent: Rfc5322Validator

Complexity

Conditions	4
Paths	4

Size

Total Lines	18
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
eloc	7
c	1
b	0
f	0
dl	0
loc	18
rs	10
cc	4
nc	4
nop	1

<?php

declare(strict_types=1);

namespace EmailValidator\Validator;

use EmailValidator\EmailAddress;

/**
 * Validates email addresses according to RFC 5322 standards
 *
 * This validator implements strict validation rules from RFC 5322 (superseding RFC 822),
 * including proper handling of:
 * - Quoted strings in local part
 * - Comments
 * - Domain literals
 * - Special characters
 * - Length restrictions
 */
class Rfc5322Validator extends AValidator
{
    // Maximum lengths defined by RFC 5322
    private const MAX_LOCAL_PART_LENGTH = 64;
    private const MAX_DOMAIN_LABEL_LENGTH = 63;
    private const MAX_DOMAIN_LENGTH = 255;

    // Character sets for unquoted local part
    private const LOCAL_PART_ALLOWED_CHARS = '!#$%&\'*+-/=?^_`{|}~.';

    /**
     * Validates an email address according to RFC 5322 rules
     *
     * @param EmailAddress $email The email address to validate
     * @return bool True if the email address is valid according to RFC 5322
     */
    public function validate(EmailAddress $email): bool
    {
        $localPart = $email->getLocalPart();
        $domain = $email->getDomain();

        if ($localPart === null || $domain === null) {
            return false;
        }

        return $this->validateLocalPart($localPart) && $this->validateDomain($domain);
    }

    /**
     * Validates the local part of an email address
     *
     * @param string $localPart The local part to validate
     * @return bool True if the local part is valid
     */
    private function validateLocalPart(string $localPart): bool
    {
        // Check length
        if (strlen($localPart) > self::MAX_LOCAL_PART_LENGTH) {
            return false;
        }

        // Empty local part is invalid
        if ($localPart === '') {
            return false;
        }

        // Handle quoted string
        if ($localPart[0] === '"') {
            return $this->validateQuotedString($localPart);
        }

        // Handle dot-atom format
        return $this->validateDotAtom($localPart);
    }

    /**
     * Validates a dot-atom format local part
     *
     * @param string $localPart The unquoted local part to validate
     * @return bool True if the unquoted local part is valid
     */
    private function validateDotAtom(string $localPart): bool
    {
        // Split into atoms
        $atoms = explode('.', $localPart);

        // Check each atom
        foreach ($atoms as $atom) {
            if ($atom === '') {
                return false;
            }

            // Check for valid characters in each atom
            if (!preg_match('/^[a-zA-Z0-9!#$%&\'*+\-\/=?^_`{|}~]+$/', $atom)) {
                return false;
            }
        }

        return true;
    }

    /**
     * Validates a quoted string local part
     *
     * @param string $localPart The quoted string to validate
     * @return bool True if the quoted string is valid
     */
    private function validateQuotedString(string $localPart): bool
    {
        // Must start and end with quotes
        if (!preg_match('/^".*"$/', $localPart)) {
            return false;
        }

        // Remove outer quotes for content validation
        $content = substr($localPart, 1, -1);

        // Empty quoted strings are valid
        if ($content === '') {
            return true;
        }

        $inEscape = false;
        for ($i = 0, $iMax = strlen($content); $i < $iMax; $i++) {
            $char = $content[$i];
            $charCode = ord($char);

            // Non-printable characters are never allowed
            if ($charCode < 32 || $charCode > 126) {
                return false;
            }

            if ($inEscape) {
                // Only quotes and backslashes must be escaped
                // Other characters may be escaped but it's not required
                $inEscape = false;
                continue;
            }

            if ($char === '\\') {
                $inEscape = true;
                continue;
            }

            // Unescaped quotes are not allowed
            if ($char === '"') {
                return false;
            }
        }

        // Can't end with a lone backslash
        return !$inEscape;
    }

    /**
     * Validates the domain part of an email address
     *
     * @param string $domain The domain to validate
     * @return bool True if the domain is valid
     */
    private function validateDomain(string $domain): bool
    {
        // Check for empty domain
        if ($domain === '') {
            return false;
        }

        // Check total length
        if (strlen($domain) > self::MAX_DOMAIN_LENGTH) {
            return false;
        }

        // Handle domain literal
        if ($domain[0] === '[') {
            return $this->validateDomainLiteral($domain);
        }

        // Validate regular domain
        return $this->validateDomainName($domain);
    }

    /**
     * Validates a domain literal (IP address in brackets)
     *
     * @param string $domain The domain literal to validate
     * @return bool True if the domain literal is valid
     */
    private function validateDomainLiteral(string $domain): bool
    {
        // Must be enclosed in brackets
        if (!preg_match('/^\[(.*)]$/', $domain, $matches)) {
            return false;
        }

        $content = $matches[1];

        // Handle IPv6
        if (stripos($content, 'IPv6:') === 0) {
            $ipv6 = substr($content, 5);
            // Remove any whitespace
            $ipv6 = trim($ipv6);

            // Handle compressed notation
            if (strpos($ipv6, '::') !== false) {
                // Only one :: allowed
                if (substr_count($ipv6, '::') > 1) {
                    return false;
                }

                // Split on ::
                $parts = explode('::', $ipv6);
                if (count($parts) !== 2) {
                    return false;
                }

                // Count segments on each side
                $leftSegments = $parts[0] ? explode(':', $parts[0]) : [];
                $rightSegments = $parts[1] ? explode(':', $parts[1]) : [];

                // Calculate missing segments
                $totalSegments = count($leftSegments) + count($rightSegments);
                if ($totalSegments >= 8) {
                    return false;
                }

                // Fill in missing segments
                $middleSegments = array_fill(0, 8 - $totalSegments, '0');

                // Combine all segments
                $segments = array_merge($leftSegments, $middleSegments, $rightSegments);
            } else {
                $segments = explode(':', $ipv6);
                if (count($segments) !== 8) {
                    return false;
                }
            }

            // Validate each segment
            foreach ($segments as $segment) {
                if (!preg_match('/^[0-9A-Fa-f]{1,4}$/', $segment)) {
                    return false;
                }
            }

            // Convert to standard format for final validation
            $ipv6 = implode(':', array_map(function ($segment) {
                return str_pad($segment, 4, '0', STR_PAD_LEFT);
            }, $segments));

            // Final validation using filter_var
            if (!filter_var($ipv6, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6)) {
                return false;
            }

            return true;
        }

        // Handle IPv4
        $ipv4 = trim($content);

        // Split into octets
        $octets = explode('.', $ipv4);
        if (count($octets) !== 4) {
            return false;
        }

        // Validate each octet
        foreach ($octets as $octet) {
            // Remove leading zeros
            $octet = ltrim($octet, '0');
            if ($octet === '') {
                $octet = '0';
            }

            // Check numeric value
            if (!is_numeric($octet) || intval($octet) < 0 || intval($octet) > 255) {
                return false;
            }
        }

        // Convert to standard format for final validation
        $ipv4 = implode('.', array_map(function ($octet) {
            return ltrim($octet, '0') ?: '0';
        }, $octets));

        return filter_var($ipv4, FILTER_VALIDATE_IP, FILTER_FLAG_IPV4) !== false;
    }

    /**
     * Validates a domain name
     *
     * @param string $domain The domain name to validate
     * @return bool True if the domain name is valid
     */
    private function validateDomainName(string $domain): bool
    {
        // Split into labels
        $labels = explode('.', $domain);

        // Must have at least two labels
        if (count($labels) < 2) {
            return false;
        }

        // Validate each label
        foreach ($labels as $label) {
            if (!$this->validateDomainLabel($label)) {
                return false;
            }
        }

        return true;
    }

    /**
     * Validates a single domain label
     *
     * @param string $label The domain label to validate
     * @return bool True if the domain label is valid
     */
    private function validateDomainLabel(string $label): bool
    {
        // Check length
        if (strlen($label) > self::MAX_DOMAIN_LABEL_LENGTH || $label === '') {
            return false;
        }

        // Must start and end with alphanumeric
        if (!ctype_alnum($label[0]) || !ctype_alnum(substr($label, -1))) {
            return false;
        }

        // Check for valid characters (alphanumeric and hyphen)
        if (!preg_match('/^[a-zA-Z0-9][a-zA-Z0-9-]*[a-zA-Z0-9]$/', $label)) {
            return false;
        }

        // Check for consecutive hyphens
        if (strpos($label, '--') !== false) {
            return false;
        }

        return true;
    }
}


1			<?php
2
3			declare(strict_types=1);
4
5			namespace EmailValidator\Validator;
6
7			use EmailValidator\EmailAddress;
8
9			/**
10			* Validates email addresses according to RFC 5322 standards
11			*
12			* This validator implements strict validation rules from RFC 5322 (superseding RFC 822),
13			* including proper handling of:
14			* - Quoted strings in local part
15			* - Comments
16			* - Domain literals
17			* - Special characters
18			* - Length restrictions
19			*/
20			class Rfc5322Validator extends AValidator
21			{
22			// Maximum lengths defined by RFC 5322
23			private const MAX_LOCAL_PART_LENGTH = 64;
24			private const MAX_DOMAIN_LABEL_LENGTH = 63;
25			private const MAX_DOMAIN_LENGTH = 255;
26
27			// Character sets for unquoted local part
28			private const LOCAL_PART_ALLOWED_CHARS = '!#$%&\'*+-/=?^_`{\|}~.';
29
30			/**
31			* Validates an email address according to RFC 5322 rules
32			*
33			* @param EmailAddress $email The email address to validate
34			* @return bool True if the email address is valid according to RFC 5322
35			*/
36			public function validate(EmailAddress $email): bool
37			{
38			$localPart = $email->getLocalPart();
39			$domain = $email->getDomain();
40
41			if ($localPart === null \|\| $domain === null) {
42			return false;
43			}
44
45			return $this->validateLocalPart($localPart) && $this->validateDomain($domain);
46			}
47
48			/**
49			* Validates the local part of an email address
50			*
51			* @param string $localPart The local part to validate
52			* @return bool True if the local part is valid
53			*/
54			private function validateLocalPart(string $localPart): bool
55			{
56			// Check length
57			if (strlen($localPart) > self::MAX_LOCAL_PART_LENGTH) {
58			return false;
59			}
60
61			// Empty local part is invalid
62			if ($localPart === '') {
63			return false;
64			}
65
66			// Handle quoted string
67			if ($localPart[0] === '"') {
68			return $this->validateQuotedString($localPart);
69			}
70
71			// Handle dot-atom format
72			return $this->validateDotAtom($localPart);
73			}
74
75			/**
76			* Validates a dot-atom format local part
77			*
78			* @param string $localPart The unquoted local part to validate
79			* @return bool True if the unquoted local part is valid
80			*/
81			private function validateDotAtom(string $localPart): bool
82			{
83			// Split into atoms
84			$atoms = explode('.', $localPart);
85
86			// Check each atom
87			foreach ($atoms as $atom) {
88			if ($atom === '') {
89			return false;
90			}
91
92			// Check for valid characters in each atom
93			if (!preg_match('/^[a-zA-Z0-9!#$%&\'*+\-\/=?^_`{\|}~]+$/', $atom)) {
94			return false;
95			}
96			}
97
98			return true;
99			}
100
101			/**
102			* Validates a quoted string local part
103			*
104			* @param string $localPart The quoted string to validate
105			* @return bool True if the quoted string is valid
106			*/
107			private function validateQuotedString(string $localPart): bool
108			{
109			// Must start and end with quotes
110			if (!preg_match('/^".*"$/', $localPart)) {
111			return false;
112			}
113
114			// Remove outer quotes for content validation
115			$content = substr($localPart, 1, -1);
116
117			// Empty quoted strings are valid
118			if ($content === '') {
119			return true;
120			}
121
122			$inEscape = false;
123			for ($i = 0, $iMax = strlen($content); $i < $iMax; $i++) {
124			$char = $content[$i];
125			$charCode = ord($char);
126
127			// Non-printable characters are never allowed
128			if ($charCode < 32 \|\| $charCode > 126) {
129			return false;
130			}
131
132			if ($inEscape) {
133			// Only quotes and backslashes must be escaped
134			// Other characters may be escaped but it's not required
135			$inEscape = false;
136			continue;
137			}
138
139			if ($char === '\\') {
140			$inEscape = true;
141			continue;
142			}
143
144			// Unescaped quotes are not allowed
145			if ($char === '"') {
146			return false;
147			}
148			}
149
150			// Can't end with a lone backslash
151			return !$inEscape;
152			}
153
154			/**
155			* Validates the domain part of an email address
156			*
157			* @param string $domain The domain to validate
158			* @return bool True if the domain is valid
159			*/
160			private function validateDomain(string $domain): bool
161			{
162			// Check for empty domain
163			if ($domain === '') {
164			return false;
165			}
166
167			// Check total length
168			if (strlen($domain) > self::MAX_DOMAIN_LENGTH) {
169			return false;
170			}
171
172			// Handle domain literal
173			if ($domain[0] === '[') {
174			return $this->validateDomainLiteral($domain);
175			}
176
177			// Validate regular domain
178			return $this->validateDomainName($domain);
179			}
180
181			/**
182			* Validates a domain literal (IP address in brackets)
183			*
184			* @param string $domain The domain literal to validate
185			* @return bool True if the domain literal is valid
186			*/
187			private function validateDomainLiteral(string $domain): bool
188			{
189			// Must be enclosed in brackets
190			if (!preg_match('/^\[(.*)]$/', $domain, $matches)) {
191			return false;
192			}
193
194			$content = $matches[1];
195
196			// Handle IPv6
197			if (stripos($content, 'IPv6:') === 0) {
198			$ipv6 = substr($content, 5);
199			// Remove any whitespace
200			$ipv6 = trim($ipv6);
201
202			// Handle compressed notation
203			if (strpos($ipv6, '::') !== false) {
204			// Only one :: allowed
205			if (substr_count($ipv6, '::') > 1) {
206			return false;
207			}
208
209			// Split on ::
210			$parts = explode('::', $ipv6);
211			if (count($parts) !== 2) {
212			return false;
213			}
214
215			// Count segments on each side
216			$leftSegments = $parts[0] ? explode(':', $parts[0]) : [];
217			$rightSegments = $parts[1] ? explode(':', $parts[1]) : [];
218
219			// Calculate missing segments
220			$totalSegments = count($leftSegments) + count($rightSegments);
221			if ($totalSegments >= 8) {
222			return false;
223			}
224
225			// Fill in missing segments
226			$middleSegments = array_fill(0, 8 - $totalSegments, '0');
227
228			// Combine all segments
229			$segments = array_merge($leftSegments, $middleSegments, $rightSegments);
230			} else {
231			$segments = explode(':', $ipv6);
232			if (count($segments) !== 8) {
233			return false;
234			}
235			}
236
237			// Validate each segment
238			foreach ($segments as $segment) {
239			if (!preg_match('/^[0-9A-Fa-f]{1,4}$/', $segment)) {
240			return false;
241			}
242			}
243
244			// Convert to standard format for final validation
245			$ipv6 = implode(':', array_map(function ($segment) {
246			return str_pad($segment, 4, '0', STR_PAD_LEFT);
247			}, $segments));
248
249			// Final validation using filter_var
250			if (!filter_var($ipv6, FILTER_VALIDATE_IP, FILTER_FLAG_IPV6)) {
251			return false;
252			}
253
254			return true;
255			}
256
257			// Handle IPv4
258			$ipv4 = trim($content);
259
260			// Split into octets
261			$octets = explode('.', $ipv4);
262			if (count($octets) !== 4) {
263			return false;
264			}
265
266			// Validate each octet
267			foreach ($octets as $octet) {
268			// Remove leading zeros
269			$octet = ltrim($octet, '0');
270			if ($octet === '') {
271			$octet = '0';
272			}
273
274			// Check numeric value
275			if (!is_numeric($octet) \|\| intval($octet) < 0 \|\| intval($octet) > 255) {
276			return false;
277			}
278			}
279
280			// Convert to standard format for final validation
281			$ipv4 = implode('.', array_map(function ($octet) {
282			return ltrim($octet, '0') ?: '0';
283			}, $octets));
284
285			return filter_var($ipv4, FILTER_VALIDATE_IP, FILTER_FLAG_IPV4) !== false;
286			}
287
288			/**
289			* Validates a domain name
290			*
291			* @param string $domain The domain name to validate
292			* @return bool True if the domain name is valid
293			*/
294			private function validateDomainName(string $domain): bool
295			{
296			// Split into labels
297			$labels = explode('.', $domain);
298
299			// Must have at least two labels
300			if (count($labels) < 2) {
301			return false;
302			}
303
304			// Validate each label
305			foreach ($labels as $label) {
306			if (!$this->validateDomainLabel($label)) {
307			return false;
308			}
309			}
310
311			return true;
312			}
313
314			/**
315			* Validates a single domain label
316			*
317			* @param string $label The domain label to validate
318			* @return bool True if the domain label is valid
319			*/
320			private function validateDomainLabel(string $label): bool
321			{
322			// Check length
323			if (strlen($label) > self::MAX_DOMAIN_LABEL_LENGTH \|\| $label === '') {
324			return false;
325			}
326
327			// Must start and end with alphanumeric
328			if (!ctype_alnum($label[0]) \|\| !ctype_alnum(substr($label, -1))) {
329			return false;
330			}
331
332			// Check for valid characters (alphanumeric and hyphen)
333			if (!preg_match('/^[a-zA-Z0-9][a-zA-Z0-9-]*[a-zA-Z0-9]$/', $label)) {
334			return false;
335			}
336
337			// Check for consecutive hyphens
338			if (strpos($label, '--') !== false) {
339			return false;
340			}
341
342			return true;
343			}
344			}
345

stymiee / email-validator

Push — master ( 82f564...b2a72e )

Rfc5322Validator::validateDotAtom() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like