]>
Commit | Line | Data |
---|---|---|
4f5b44bd NL |
1 | <?php |
2 | ||
3 | /* | |
4 | * This file is part of the Symfony package. | |
5 | * | |
6 | * (c) Fabien Potencier <fabien@symfony.com> | |
7 | * | |
8 | * For the full copyright and license information, please view the LICENSE | |
9 | * file that was distributed with this source code. | |
10 | */ | |
11 | ||
12 | namespace Symfony\Component\Routing; | |
13 | ||
14 | /** | |
15 | * RouteCompiler compiles Route instances to CompiledRoute instances. | |
16 | * | |
17 | * @author Fabien Potencier <fabien@symfony.com> | |
18 | * @author Tobias Schultze <http://tobion.de> | |
19 | */ | |
20 | class RouteCompiler implements RouteCompilerInterface | |
21 | { | |
22 | const REGEX_DELIMITER = '#'; | |
23 | ||
24 | /** | |
25 | * This string defines the characters that are automatically considered separators in front of | |
26 | * optional placeholders (with default and no static text following). Such a single separator | |
27 | * can be left out together with the optional placeholder from matching and generating URLs. | |
28 | */ | |
29 | const SEPARATORS = '/,;.:-_~+*=@|'; | |
30 | ||
31 | /** | |
32 | * {@inheritDoc} | |
33 | * | |
34 | * @throws \LogicException If a variable is referenced more than once | |
35 | * @throws \DomainException If a variable name is numeric because PHP raises an error for such | |
36 | * subpatterns in PCRE and thus would break matching, e.g. "(?P<123>.+)". | |
37 | */ | |
38 | public static function compile(Route $route) | |
39 | { | |
40 | $staticPrefix = null; | |
41 | $hostVariables = array(); | |
42 | $pathVariables = array(); | |
43 | $variables = array(); | |
44 | $tokens = array(); | |
45 | $regex = null; | |
46 | $hostRegex = null; | |
47 | $hostTokens = array(); | |
48 | ||
49 | if ('' !== $host = $route->getHost()) { | |
50 | $result = self::compilePattern($route, $host, true); | |
51 | ||
52 | $hostVariables = $result['variables']; | |
53 | $variables = array_merge($variables, $hostVariables); | |
54 | ||
55 | $hostTokens = $result['tokens']; | |
56 | $hostRegex = $result['regex']; | |
57 | } | |
58 | ||
59 | $path = $route->getPath(); | |
60 | ||
61 | $result = self::compilePattern($route, $path, false); | |
62 | ||
63 | $staticPrefix = $result['staticPrefix']; | |
64 | ||
65 | $pathVariables = $result['variables']; | |
66 | $variables = array_merge($variables, $pathVariables); | |
67 | ||
68 | $tokens = $result['tokens']; | |
69 | $regex = $result['regex']; | |
70 | ||
71 | return new CompiledRoute( | |
72 | $staticPrefix, | |
73 | $regex, | |
74 | $tokens, | |
75 | $pathVariables, | |
76 | $hostRegex, | |
77 | $hostTokens, | |
78 | $hostVariables, | |
79 | array_unique($variables) | |
80 | ); | |
81 | } | |
82 | ||
83 | private static function compilePattern(Route $route, $pattern, $isHost) | |
84 | { | |
85 | $tokens = array(); | |
86 | $variables = array(); | |
87 | $matches = array(); | |
88 | $pos = 0; | |
89 | $defaultSeparator = $isHost ? '.' : '/'; | |
90 | ||
91 | // Match all variables enclosed in "{}" and iterate over them. But we only want to match the innermost variable | |
92 | // in case of nested "{}", e.g. {foo{bar}}. This in ensured because \w does not match "{" or "}" itself. | |
93 | preg_match_all('#\{\w+\}#', $pattern, $matches, PREG_OFFSET_CAPTURE | PREG_SET_ORDER); | |
94 | foreach ($matches as $match) { | |
95 | $varName = substr($match[0][0], 1, -1); | |
96 | // get all static text preceding the current variable | |
97 | $precedingText = substr($pattern, $pos, $match[0][1] - $pos); | |
98 | $pos = $match[0][1] + strlen($match[0][0]); | |
99 | $precedingChar = strlen($precedingText) > 0 ? substr($precedingText, -1) : ''; | |
100 | $isSeparator = '' !== $precedingChar && false !== strpos(static::SEPARATORS, $precedingChar); | |
101 | ||
102 | if (is_numeric($varName)) { | |
103 | throw new \DomainException(sprintf('Variable name "%s" cannot be numeric in route pattern "%s". Please use a different name.', $varName, $pattern)); | |
104 | } | |
105 | if (in_array($varName, $variables)) { | |
106 | throw new \LogicException(sprintf('Route pattern "%s" cannot reference variable name "%s" more than once.', $pattern, $varName)); | |
107 | } | |
108 | ||
109 | if ($isSeparator && strlen($precedingText) > 1) { | |
110 | $tokens[] = array('text', substr($precedingText, 0, -1)); | |
111 | } elseif (!$isSeparator && strlen($precedingText) > 0) { | |
112 | $tokens[] = array('text', $precedingText); | |
113 | } | |
114 | ||
115 | $regexp = $route->getRequirement($varName); | |
116 | if (null === $regexp) { | |
117 | $followingPattern = (string) substr($pattern, $pos); | |
118 | // Find the next static character after the variable that functions as a separator. By default, this separator and '/' | |
119 | // are disallowed for the variable. This default requirement makes sure that optional variables can be matched at all | |
120 | // and that the generating-matching-combination of URLs unambiguous, i.e. the params used for generating the URL are | |
121 | // the same that will be matched. Example: new Route('/{page}.{_format}', array('_format' => 'html')) | |
122 | // If {page} would also match the separating dot, {_format} would never match as {page} will eagerly consume everything. | |
123 | // Also even if {_format} was not optional the requirement prevents that {page} matches something that was originally | |
124 | // part of {_format} when generating the URL, e.g. _format = 'mobile.html'. | |
125 | $nextSeparator = self::findNextSeparator($followingPattern); | |
126 | $regexp = sprintf( | |
127 | '[^%s%s]+', | |
128 | preg_quote($defaultSeparator, self::REGEX_DELIMITER), | |
129 | $defaultSeparator !== $nextSeparator && '' !== $nextSeparator ? preg_quote($nextSeparator, self::REGEX_DELIMITER) : '' | |
130 | ); | |
131 | if (('' !== $nextSeparator && !preg_match('#^\{\w+\}#', $followingPattern)) || '' === $followingPattern) { | |
132 | // When we have a separator, which is disallowed for the variable, we can optimize the regex with a possessive | |
133 | // quantifier. This prevents useless backtracking of PCRE and improves performance by 20% for matching those patterns. | |
134 | // Given the above example, there is no point in backtracking into {page} (that forbids the dot) when a dot must follow | |
135 | // after it. This optimization cannot be applied when the next char is no real separator or when the next variable is | |
136 | // directly adjacent, e.g. '/{x}{y}'. | |
137 | $regexp .= '+'; | |
138 | } | |
139 | } | |
140 | ||
141 | $tokens[] = array('variable', $isSeparator ? $precedingChar : '', $regexp, $varName); | |
142 | $variables[] = $varName; | |
143 | } | |
144 | ||
145 | if ($pos < strlen($pattern)) { | |
146 | $tokens[] = array('text', substr($pattern, $pos)); | |
147 | } | |
148 | ||
149 | // find the first optional token | |
150 | $firstOptional = PHP_INT_MAX; | |
151 | if (!$isHost) { | |
152 | for ($i = count($tokens) - 1; $i >= 0; $i--) { | |
153 | $token = $tokens[$i]; | |
154 | if ('variable' === $token[0] && $route->hasDefault($token[3])) { | |
155 | $firstOptional = $i; | |
156 | } else { | |
157 | break; | |
158 | } | |
159 | } | |
160 | } | |
161 | ||
162 | // compute the matching regexp | |
163 | $regexp = ''; | |
164 | for ($i = 0, $nbToken = count($tokens); $i < $nbToken; $i++) { | |
165 | $regexp .= self::computeRegexp($tokens, $i, $firstOptional); | |
166 | } | |
167 | ||
168 | return array( | |
169 | 'staticPrefix' => 'text' === $tokens[0][0] ? $tokens[0][1] : '', | |
170 | 'regex' => self::REGEX_DELIMITER.'^'.$regexp.'$'.self::REGEX_DELIMITER.'s', | |
171 | 'tokens' => array_reverse($tokens), | |
172 | 'variables' => $variables, | |
173 | ); | |
174 | } | |
175 | ||
176 | /** | |
177 | * Returns the next static character in the Route pattern that will serve as a separator. | |
178 | * | |
179 | * @param string $pattern The route pattern | |
180 | * | |
181 | * @return string The next static character that functions as separator (or empty string when none available) | |
182 | */ | |
183 | private static function findNextSeparator($pattern) | |
184 | { | |
185 | if ('' == $pattern) { | |
186 | // return empty string if pattern is empty or false (false which can be returned by substr) | |
187 | return ''; | |
188 | } | |
189 | // first remove all placeholders from the pattern so we can find the next real static character | |
190 | $pattern = preg_replace('#\{\w+\}#', '', $pattern); | |
191 | ||
192 | return isset($pattern[0]) && false !== strpos(static::SEPARATORS, $pattern[0]) ? $pattern[0] : ''; | |
193 | } | |
194 | ||
195 | /** | |
196 | * Computes the regexp used to match a specific token. It can be static text or a subpattern. | |
197 | * | |
198 | * @param array $tokens The route tokens | |
199 | * @param integer $index The index of the current token | |
200 | * @param integer $firstOptional The index of the first optional token | |
201 | * | |
202 | * @return string The regexp pattern for a single token | |
203 | */ | |
204 | private static function computeRegexp(array $tokens, $index, $firstOptional) | |
205 | { | |
206 | $token = $tokens[$index]; | |
207 | if ('text' === $token[0]) { | |
208 | // Text tokens | |
209 | return preg_quote($token[1], self::REGEX_DELIMITER); | |
210 | } else { | |
211 | // Variable tokens | |
212 | if (0 === $index && 0 === $firstOptional) { | |
213 | // When the only token is an optional variable token, the separator is required | |
214 | return sprintf('%s(?P<%s>%s)?', preg_quote($token[1], self::REGEX_DELIMITER), $token[3], $token[2]); | |
215 | } else { | |
216 | $regexp = sprintf('%s(?P<%s>%s)', preg_quote($token[1], self::REGEX_DELIMITER), $token[3], $token[2]); | |
217 | if ($index >= $firstOptional) { | |
218 | // Enclose each optional token in a subpattern to make it optional. | |
219 | // "?:" means it is non-capturing, i.e. the portion of the subject string that | |
220 | // matched the optional subpattern is not passed back. | |
221 | $regexp = "(?:$regexp"; | |
222 | $nbTokens = count($tokens); | |
223 | if ($nbTokens - 1 == $index) { | |
224 | // Close the optional subpatterns | |
225 | $regexp .= str_repeat(")?", $nbTokens - $firstOptional - (0 === $firstOptional ? 1 : 0)); | |
226 | } | |
227 | } | |
228 | ||
229 | return $regexp; | |
230 | } | |
231 | } | |
232 | } | |
233 | } |