@@ -11,32 +11,32 @@ |
||
| 11 | 11 | |
| 12 | 12 | class Utf8 implements OutputInterface |
| 13 | 13 | { |
| 14 | - /** |
|
| 15 | - * {@inheritdoc} |
|
| 16 | - */ |
|
| 17 | - public function output($value) |
|
| 18 | - { |
|
| 19 | - if ($value < 0x80) |
|
| 20 | - { |
|
| 21 | - return chr($value); |
|
| 22 | - } |
|
| 23 | - if ($value < 0x800) |
|
| 24 | - { |
|
| 25 | - return chr(0xC0 | ($value >> 6)) . chr(0x80 | ($value & 0x3F)); |
|
| 26 | - } |
|
| 27 | - if ($value < 0x10000) |
|
| 28 | - { |
|
| 29 | - return chr(0xE0 | ($value >> 12)) |
|
| 30 | - . chr(0x80 | (($value >> 6) & 0x3F)) |
|
| 31 | - . chr(0x80 | ($value & 0x3F)); |
|
| 32 | - } |
|
| 33 | - if ($value < 0x110000) |
|
| 34 | - { |
|
| 35 | - return chr(0xF0 | ($value >> 18)) |
|
| 36 | - . chr(0x80 | (($value >> 12) & 0x3F)) |
|
| 37 | - . chr(0x80 | (($value >> 6) & 0x3F)) |
|
| 38 | - . chr(0x80 | ($value & 0x3F)); |
|
| 39 | - } |
|
| 40 | - throw new InvalidArgumentException('Invalid UTF-8 codepoint 0x' . dechex($value)); |
|
| 41 | - } |
|
| 14 | + /** |
|
| 15 | + * {@inheritdoc} |
|
| 16 | + */ |
|
| 17 | + public function output($value) |
|
| 18 | + { |
|
| 19 | + if ($value < 0x80) |
|
| 20 | + { |
|
| 21 | + return chr($value); |
|
| 22 | + } |
|
| 23 | + if ($value < 0x800) |
|
| 24 | + { |
|
| 25 | + return chr(0xC0 | ($value >> 6)) . chr(0x80 | ($value & 0x3F)); |
|
| 26 | + } |
|
| 27 | + if ($value < 0x10000) |
|
| 28 | + { |
|
| 29 | + return chr(0xE0 | ($value >> 12)) |
|
| 30 | + . chr(0x80 | (($value >> 6) & 0x3F)) |
|
| 31 | + . chr(0x80 | ($value & 0x3F)); |
|
| 32 | + } |
|
| 33 | + if ($value < 0x110000) |
|
| 34 | + { |
|
| 35 | + return chr(0xF0 | ($value >> 18)) |
|
| 36 | + . chr(0x80 | (($value >> 12) & 0x3F)) |
|
| 37 | + . chr(0x80 | (($value >> 6) & 0x3F)) |
|
| 38 | + . chr(0x80 | ($value & 0x3F)); |
|
| 39 | + } |
|
| 40 | + throw new InvalidArgumentException('Invalid UTF-8 codepoint 0x' . dechex($value)); |
|
| 41 | + } |
|
| 42 | 42 | } |
| 43 | 43 | \ No newline at end of file |
@@ -11,189 +11,189 @@ |
||
| 11 | 11 | |
| 12 | 12 | class Serializer |
| 13 | 13 | { |
| 14 | - /** |
|
| 15 | - * @var Escaper |
|
| 16 | - */ |
|
| 17 | - protected $escaper; |
|
| 18 | - |
|
| 19 | - /** |
|
| 20 | - * @var OutputInterface |
|
| 21 | - */ |
|
| 22 | - protected $output; |
|
| 23 | - |
|
| 24 | - /** |
|
| 25 | - * @param OutputInterface $output |
|
| 26 | - * @param Escaper $escaper |
|
| 27 | - */ |
|
| 28 | - public function __construct(OutputInterface $output, Escaper $escaper) |
|
| 29 | - { |
|
| 30 | - $this->escaper = $escaper; |
|
| 31 | - $this->output = $output; |
|
| 32 | - } |
|
| 33 | - |
|
| 34 | - /** |
|
| 35 | - * Serialize given strings into a regular expression |
|
| 36 | - * |
|
| 37 | - * @param array[] $strings |
|
| 38 | - * @return string |
|
| 39 | - */ |
|
| 40 | - public function serializeStrings(array $strings) |
|
| 41 | - { |
|
| 42 | - $info = $this->analyzeStrings($strings); |
|
| 43 | - $alternations = $this->buildAlternations($info); |
|
| 44 | - $expr = implode('|', $alternations); |
|
| 45 | - |
|
| 46 | - if (count($alternations) > 1 || $this->isOneOptionalString($info)) |
|
| 47 | - { |
|
| 48 | - $expr = '(?:' . $expr . ')'; |
|
| 49 | - } |
|
| 50 | - |
|
| 51 | - return $expr . $info['quantifier']; |
|
| 52 | - } |
|
| 53 | - |
|
| 54 | - /** |
|
| 55 | - * Analyze given strings to determine how to serialize them |
|
| 56 | - * |
|
| 57 | - * The returned array may contains any of the following elements: |
|
| 58 | - * |
|
| 59 | - * - (string) quantifier Either '' or '?' |
|
| 60 | - * - (array) chars List of values from single-char strings |
|
| 61 | - * - (array) strings List of multi-char strings |
|
| 62 | - * |
|
| 63 | - * @param array[] $strings |
|
| 64 | - * @return array |
|
| 65 | - */ |
|
| 66 | - protected function analyzeStrings(array $strings) |
|
| 67 | - { |
|
| 68 | - $info = ['quantifier' => '']; |
|
| 69 | - $chars = []; |
|
| 70 | - foreach ($strings as $k => $string) |
|
| 71 | - { |
|
| 72 | - if (empty($string)) |
|
| 73 | - { |
|
| 74 | - $info['quantifier'] = '?'; |
|
| 75 | - unset($strings[$k]); |
|
| 76 | - } |
|
| 77 | - elseif (!isset($string[1])) |
|
| 78 | - { |
|
| 79 | - $chars[$k] = $string[0]; |
|
| 80 | - } |
|
| 81 | - } |
|
| 82 | - |
|
| 83 | - if (count($chars) > 1) |
|
| 84 | - { |
|
| 85 | - $info['chars'] = array_values($chars); |
|
| 86 | - $strings = array_diff_key($strings, $chars); |
|
| 87 | - } |
|
| 88 | - |
|
| 89 | - $info['strings'] = array_values($strings); |
|
| 90 | - |
|
| 91 | - return $info; |
|
| 92 | - } |
|
| 93 | - |
|
| 94 | - /** |
|
| 95 | - * Build the list of alternations based on given info |
|
| 96 | - * |
|
| 97 | - * @param array $info |
|
| 98 | - * @return string[] |
|
| 99 | - */ |
|
| 100 | - protected function buildAlternations(array $info) |
|
| 101 | - { |
|
| 102 | - $alternations = []; |
|
| 103 | - if (!empty($info['chars'])) |
|
| 104 | - { |
|
| 105 | - $alternations[] = $this->serializeCharacterClass($info['chars']); |
|
| 106 | - } |
|
| 107 | - foreach ($info['strings'] as $string) |
|
| 108 | - { |
|
| 109 | - $alternations[] = $this->serializeString($string); |
|
| 110 | - } |
|
| 111 | - |
|
| 112 | - return $alternations; |
|
| 113 | - } |
|
| 114 | - |
|
| 115 | - /** |
|
| 116 | - * Get the list of ranges that cover all given values |
|
| 117 | - * |
|
| 118 | - * @param integer[] $values Ordered list of values |
|
| 119 | - * @return array[] List of ranges in the form [start, end] |
|
| 120 | - */ |
|
| 121 | - protected function getRanges(array $values) |
|
| 122 | - { |
|
| 123 | - $i = 0; |
|
| 124 | - $cnt = count($values); |
|
| 125 | - $start = $values[0]; |
|
| 126 | - $end = $start; |
|
| 127 | - $ranges = []; |
|
| 128 | - while (++$i < $cnt) |
|
| 129 | - { |
|
| 130 | - if ($values[$i] === $end + 1) |
|
| 131 | - { |
|
| 132 | - ++$end; |
|
| 133 | - } |
|
| 134 | - else |
|
| 135 | - { |
|
| 136 | - $ranges[] = [$start, $end]; |
|
| 137 | - $start = $end = $values[$i]; |
|
| 138 | - } |
|
| 139 | - } |
|
| 140 | - $ranges[] = [$start, $end]; |
|
| 141 | - |
|
| 142 | - return $ranges; |
|
| 143 | - } |
|
| 144 | - |
|
| 145 | - /** |
|
| 146 | - * Test whether a string is optional and has more than one character |
|
| 147 | - * |
|
| 148 | - * @param array $info |
|
| 149 | - * @return bool |
|
| 150 | - */ |
|
| 151 | - protected function isOneOptionalString(array $info) |
|
| 152 | - { |
|
| 153 | - // Test whether the first string has a quantifier and more than one element |
|
| 154 | - return (!empty($info['quantifier']) && isset($info['strings'][0][1])); |
|
| 155 | - } |
|
| 156 | - |
|
| 157 | - /** |
|
| 158 | - * Serialize a given list of values into a character class |
|
| 159 | - * |
|
| 160 | - * @param integer[] $values |
|
| 161 | - * @return string |
|
| 162 | - */ |
|
| 163 | - protected function serializeCharacterClass(array $values) |
|
| 164 | - { |
|
| 165 | - $expr = '['; |
|
| 166 | - foreach ($this->getRanges($values) as list($start, $end)) |
|
| 167 | - { |
|
| 168 | - $expr .= $this->escaper->escapeCharacterClass($this->output->output($start)); |
|
| 169 | - if ($end > $start) |
|
| 170 | - { |
|
| 171 | - if ($end > $start + 1) |
|
| 172 | - { |
|
| 173 | - $expr .= '-'; |
|
| 174 | - } |
|
| 175 | - $expr .= $this->escaper->escapeCharacterClass($this->output->output($end)); |
|
| 176 | - } |
|
| 177 | - } |
|
| 178 | - $expr .= ']'; |
|
| 179 | - |
|
| 180 | - return $expr; |
|
| 181 | - } |
|
| 182 | - |
|
| 183 | - /** |
|
| 184 | - * Serialize a given string into a regular expression |
|
| 185 | - * |
|
| 186 | - * @param array $string |
|
| 187 | - * @return string |
|
| 188 | - */ |
|
| 189 | - protected function serializeString(array $string) |
|
| 190 | - { |
|
| 191 | - $expr = ''; |
|
| 192 | - foreach ($string as $element) |
|
| 193 | - { |
|
| 194 | - $expr .= (is_array($element)) ? $this->serializeStrings($element) : $this->escaper->escapeLiteral($this->output->output($element)); |
|
| 195 | - } |
|
| 196 | - |
|
| 197 | - return $expr; |
|
| 198 | - } |
|
| 14 | + /** |
|
| 15 | + * @var Escaper |
|
| 16 | + */ |
|
| 17 | + protected $escaper; |
|
| 18 | + |
|
| 19 | + /** |
|
| 20 | + * @var OutputInterface |
|
| 21 | + */ |
|
| 22 | + protected $output; |
|
| 23 | + |
|
| 24 | + /** |
|
| 25 | + * @param OutputInterface $output |
|
| 26 | + * @param Escaper $escaper |
|
| 27 | + */ |
|
| 28 | + public function __construct(OutputInterface $output, Escaper $escaper) |
|
| 29 | + { |
|
| 30 | + $this->escaper = $escaper; |
|
| 31 | + $this->output = $output; |
|
| 32 | + } |
|
| 33 | + |
|
| 34 | + /** |
|
| 35 | + * Serialize given strings into a regular expression |
|
| 36 | + * |
|
| 37 | + * @param array[] $strings |
|
| 38 | + * @return string |
|
| 39 | + */ |
|
| 40 | + public function serializeStrings(array $strings) |
|
| 41 | + { |
|
| 42 | + $info = $this->analyzeStrings($strings); |
|
| 43 | + $alternations = $this->buildAlternations($info); |
|
| 44 | + $expr = implode('|', $alternations); |
|
| 45 | + |
|
| 46 | + if (count($alternations) > 1 || $this->isOneOptionalString($info)) |
|
| 47 | + { |
|
| 48 | + $expr = '(?:' . $expr . ')'; |
|
| 49 | + } |
|
| 50 | + |
|
| 51 | + return $expr . $info['quantifier']; |
|
| 52 | + } |
|
| 53 | + |
|
| 54 | + /** |
|
| 55 | + * Analyze given strings to determine how to serialize them |
|
| 56 | + * |
|
| 57 | + * The returned array may contains any of the following elements: |
|
| 58 | + * |
|
| 59 | + * - (string) quantifier Either '' or '?' |
|
| 60 | + * - (array) chars List of values from single-char strings |
|
| 61 | + * - (array) strings List of multi-char strings |
|
| 62 | + * |
|
| 63 | + * @param array[] $strings |
|
| 64 | + * @return array |
|
| 65 | + */ |
|
| 66 | + protected function analyzeStrings(array $strings) |
|
| 67 | + { |
|
| 68 | + $info = ['quantifier' => '']; |
|
| 69 | + $chars = []; |
|
| 70 | + foreach ($strings as $k => $string) |
|
| 71 | + { |
|
| 72 | + if (empty($string)) |
|
| 73 | + { |
|
| 74 | + $info['quantifier'] = '?'; |
|
| 75 | + unset($strings[$k]); |
|
| 76 | + } |
|
| 77 | + elseif (!isset($string[1])) |
|
| 78 | + { |
|
| 79 | + $chars[$k] = $string[0]; |
|
| 80 | + } |
|
| 81 | + } |
|
| 82 | + |
|
| 83 | + if (count($chars) > 1) |
|
| 84 | + { |
|
| 85 | + $info['chars'] = array_values($chars); |
|
| 86 | + $strings = array_diff_key($strings, $chars); |
|
| 87 | + } |
|
| 88 | + |
|
| 89 | + $info['strings'] = array_values($strings); |
|
| 90 | + |
|
| 91 | + return $info; |
|
| 92 | + } |
|
| 93 | + |
|
| 94 | + /** |
|
| 95 | + * Build the list of alternations based on given info |
|
| 96 | + * |
|
| 97 | + * @param array $info |
|
| 98 | + * @return string[] |
|
| 99 | + */ |
|
| 100 | + protected function buildAlternations(array $info) |
|
| 101 | + { |
|
| 102 | + $alternations = []; |
|
| 103 | + if (!empty($info['chars'])) |
|
| 104 | + { |
|
| 105 | + $alternations[] = $this->serializeCharacterClass($info['chars']); |
|
| 106 | + } |
|
| 107 | + foreach ($info['strings'] as $string) |
|
| 108 | + { |
|
| 109 | + $alternations[] = $this->serializeString($string); |
|
| 110 | + } |
|
| 111 | + |
|
| 112 | + return $alternations; |
|
| 113 | + } |
|
| 114 | + |
|
| 115 | + /** |
|
| 116 | + * Get the list of ranges that cover all given values |
|
| 117 | + * |
|
| 118 | + * @param integer[] $values Ordered list of values |
|
| 119 | + * @return array[] List of ranges in the form [start, end] |
|
| 120 | + */ |
|
| 121 | + protected function getRanges(array $values) |
|
| 122 | + { |
|
| 123 | + $i = 0; |
|
| 124 | + $cnt = count($values); |
|
| 125 | + $start = $values[0]; |
|
| 126 | + $end = $start; |
|
| 127 | + $ranges = []; |
|
| 128 | + while (++$i < $cnt) |
|
| 129 | + { |
|
| 130 | + if ($values[$i] === $end + 1) |
|
| 131 | + { |
|
| 132 | + ++$end; |
|
| 133 | + } |
|
| 134 | + else |
|
| 135 | + { |
|
| 136 | + $ranges[] = [$start, $end]; |
|
| 137 | + $start = $end = $values[$i]; |
|
| 138 | + } |
|
| 139 | + } |
|
| 140 | + $ranges[] = [$start, $end]; |
|
| 141 | + |
|
| 142 | + return $ranges; |
|
| 143 | + } |
|
| 144 | + |
|
| 145 | + /** |
|
| 146 | + * Test whether a string is optional and has more than one character |
|
| 147 | + * |
|
| 148 | + * @param array $info |
|
| 149 | + * @return bool |
|
| 150 | + */ |
|
| 151 | + protected function isOneOptionalString(array $info) |
|
| 152 | + { |
|
| 153 | + // Test whether the first string has a quantifier and more than one element |
|
| 154 | + return (!empty($info['quantifier']) && isset($info['strings'][0][1])); |
|
| 155 | + } |
|
| 156 | + |
|
| 157 | + /** |
|
| 158 | + * Serialize a given list of values into a character class |
|
| 159 | + * |
|
| 160 | + * @param integer[] $values |
|
| 161 | + * @return string |
|
| 162 | + */ |
|
| 163 | + protected function serializeCharacterClass(array $values) |
|
| 164 | + { |
|
| 165 | + $expr = '['; |
|
| 166 | + foreach ($this->getRanges($values) as list($start, $end)) |
|
| 167 | + { |
|
| 168 | + $expr .= $this->escaper->escapeCharacterClass($this->output->output($start)); |
|
| 169 | + if ($end > $start) |
|
| 170 | + { |
|
| 171 | + if ($end > $start + 1) |
|
| 172 | + { |
|
| 173 | + $expr .= '-'; |
|
| 174 | + } |
|
| 175 | + $expr .= $this->escaper->escapeCharacterClass($this->output->output($end)); |
|
| 176 | + } |
|
| 177 | + } |
|
| 178 | + $expr .= ']'; |
|
| 179 | + |
|
| 180 | + return $expr; |
|
| 181 | + } |
|
| 182 | + |
|
| 183 | + /** |
|
| 184 | + * Serialize a given string into a regular expression |
|
| 185 | + * |
|
| 186 | + * @param array $string |
|
| 187 | + * @return string |
|
| 188 | + */ |
|
| 189 | + protected function serializeString(array $string) |
|
| 190 | + { |
|
| 191 | + $expr = ''; |
|
| 192 | + foreach ($string as $element) |
|
| 193 | + { |
|
| 194 | + $expr .= (is_array($element)) ? $this->serializeStrings($element) : $this->escaper->escapeLiteral($this->output->output($element)); |
|
| 195 | + } |
|
| 196 | + |
|
| 197 | + return $expr; |
|
| 198 | + } |
|
| 199 | 199 | } |
| 200 | 200 | \ No newline at end of file |