wp-all-import
/
classes
/
XmlStreamReader
/
prewk
/
xml-string-streamer
/
src
/
XmlStringStreamer
/
Parser
/
UniqueNode.php
wp-all-import
/
classes
/
XmlStreamReader
/
prewk
/
xml-string-streamer
/
src
/
XmlStringStreamer
/
Parser
Last commit date
StringWalker.php
2 years ago
UniqueNode.php
9 years ago
UniqueNode.php
274 lines
| 1 | <?php |
| 2 | /** |
| 3 | * xml-string-streamer UniqueNode parser |
| 4 | * |
| 5 | * @package xml-string-streamer |
| 6 | * @author Oskar Thornblad <oskar.thornblad@gmail.com> |
| 7 | * @author Roman Voloboev <animir@ya.ru> |
| 8 | */ |
| 9 | |
| 10 | namespace Prewk\XmlStringStreamer\Parser; |
| 11 | |
| 12 | use Exception; |
| 13 | use Prewk\XmlStringStreamer\ParserInterface; |
| 14 | use Prewk\XmlStringStreamer\StreamInterface; |
| 15 | |
| 16 | /** |
| 17 | * The unique node parser starts at a given element name and flushes when its corresponding closing tag is found |
| 18 | */ |
| 19 | class UniqueNode implements ParserInterface |
| 20 | { |
| 21 | /** |
| 22 | * Current working XML blob |
| 23 | * @var string |
| 24 | */ |
| 25 | private $workingBlob = ""; |
| 26 | /** |
| 27 | * The flushed node |
| 28 | * @var string |
| 29 | */ |
| 30 | private $flushed = ""; |
| 31 | |
| 32 | /** |
| 33 | * Start position of the given element in the workingBlob |
| 34 | * @var integer |
| 35 | */ |
| 36 | private $startPos = 0; |
| 37 | /** |
| 38 | * Records how far we've searched in the XML blob so far |
| 39 | * @var integer |
| 40 | */ |
| 41 | private $hasSearchedUntilPos = -1; |
| 42 | |
| 43 | const FIND_OPENING_TAG_ACTION = 0; |
| 44 | const FIND_CLOSING_TAG_ACTION = 1; |
| 45 | |
| 46 | /** |
| 47 | * Next action to perform |
| 48 | * @var integer |
| 49 | */ |
| 50 | private $nextAction = 0; |
| 51 | |
| 52 | /** |
| 53 | * Indicates short closing tag |
| 54 | * @var bool |
| 55 | */ |
| 56 | |
| 57 | private $shortClosedTagNow = false; |
| 58 | |
| 59 | /** |
| 60 | * If extractContainer is true, this will grow with the XML captured before and after the specified capture depth |
| 61 | * @var string |
| 62 | */ |
| 63 | protected $containerXml = ""; |
| 64 | |
| 65 | /** |
| 66 | * Whether we're found our first capture target or not |
| 67 | * @var bool |
| 68 | */ |
| 69 | protected $preCapture = true; |
| 70 | |
| 71 | /** |
| 72 | * Parser constructor |
| 73 | * @param array $options An options array |
| 74 | * @throws Exception if the required option uniqueNode isn't set |
| 75 | */ |
| 76 | public function __construct(array $options = array()) |
| 77 | { |
| 78 | $this->options = array_merge(array( |
| 79 | "extractContainer" => false, |
| 80 | ), $options); |
| 81 | |
| 82 | if (!isset($this->options["uniqueNode"])) { |
| 83 | throw new Exception("Required option 'uniqueNode' not set"); |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | /** |
| 88 | * Search the blob for our unique node's opening tag |
| 89 | * @return bool|int Either returns the char position of the opening tag or false |
| 90 | */ |
| 91 | protected function getOpeningTagPos() |
| 92 | { |
| 93 | $startPositionInBlob = false; |
| 94 | if (preg_match("/<" . preg_quote($this->options["uniqueNode"]) . "(>| )/", $this->workingBlob, $matches, PREG_OFFSET_CAPTURE) === 1) { |
| 95 | $startPositionInBlob = $matches[0][1]; |
| 96 | } |
| 97 | |
| 98 | |
| 99 | if ($startPositionInBlob === false) { |
| 100 | $this->hasSearchedUntilPos = strlen($this->workingBlob) - 1; |
| 101 | } |
| 102 | |
| 103 | return $startPositionInBlob; |
| 104 | } |
| 105 | |
| 106 | /** |
| 107 | * Search short closing tag in $workingBlob before |
| 108 | * |
| 109 | * @param string $workingBlob |
| 110 | * @param int $len |
| 111 | * @return bool|int Either returns the char position of the short closing tag or false |
| 112 | */ |
| 113 | private function checkShortClosingTag($workingBlob, $len) { |
| 114 | $resultEndPositionInBlob = false; |
| 115 | while ($len = strpos($workingBlob, "/>", $len + 1)) { |
| 116 | $subBlob = substr($workingBlob, $this->startPos, $len + strlen("/>") - $this->startPos); |
| 117 | $cntOpen = substr_count($subBlob, "<"); |
| 118 | $cntClose = substr_count($subBlob, "/>"); |
| 119 | if ($cntOpen === $cntClose && $cntOpen === 1) { |
| 120 | $resultEndPositionInBlob = $len + strlen("/>"); |
| 121 | break; // end while. so $endPositionInBlob correct now |
| 122 | } |
| 123 | } |
| 124 | return $resultEndPositionInBlob; |
| 125 | } |
| 126 | |
| 127 | /** |
| 128 | * Search the blob for our unique node's closing tag |
| 129 | * @return bool|int Either returns the char position of the closing tag or false |
| 130 | */ |
| 131 | protected function getClosingTagPos() |
| 132 | { |
| 133 | $endPositionInBlob = strpos($this->workingBlob, "</" . $this->options["uniqueNode"] . ">"); |
| 134 | if ($endPositionInBlob === false) { |
| 135 | |
| 136 | if (isset($this->options["checkShortClosing"]) && $this->options["checkShortClosing"] === true) { |
| 137 | $endPositionInBlob = $this->checkShortClosingTag($this->workingBlob, $this->startPos); |
| 138 | } |
| 139 | |
| 140 | if ($endPositionInBlob === false) { |
| 141 | $this->hasSearchedUntilPos = strlen($this->workingBlob) - 1; |
| 142 | } else { |
| 143 | $this->shortClosedTagNow = true; |
| 144 | } |
| 145 | } else { |
| 146 | if (isset($this->options["checkShortClosing"]) && $this->options["checkShortClosing"] === true) { |
| 147 | $tmpEndPositionInBlob = $this->checkShortClosingTag(substr($this->workingBlob, 0, $endPositionInBlob), $this->startPos); |
| 148 | if ($tmpEndPositionInBlob !== false) { |
| 149 | $this->shortClosedTagNow = true; |
| 150 | $endPositionInBlob = $tmpEndPositionInBlob; |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | return $endPositionInBlob; |
| 156 | } |
| 157 | |
| 158 | /** |
| 159 | * Set the start position in the workingBlob from where we should start reading when the closing tag is found |
| 160 | * @param int $startPositionInBlob Position of starting tag |
| 161 | */ |
| 162 | protected function startSalvaging($startPositionInBlob) |
| 163 | { |
| 164 | $this->startPos = $startPositionInBlob; |
| 165 | } |
| 166 | |
| 167 | /** |
| 168 | * Cut everything from the start position to the end position in the workingBlob (+ tag length) and flush it out for later return in getNodeFrom |
| 169 | * @param int $endPositionInBlob Position of the closing tag |
| 170 | */ |
| 171 | protected function flush($endPositionInBlob) { |
| 172 | $endTagLen = $this->shortClosedTagNow ? 0 : strlen("</" . $this->options["uniqueNode"] . ">"); |
| 173 | $realEndPosition = $endPositionInBlob + $endTagLen; |
| 174 | $this->flushed = substr($this->workingBlob, $this->startPos, $realEndPosition - $this->startPos); |
| 175 | $this->workingBlob = substr($this->workingBlob, $realEndPosition); |
| 176 | $this->hasSearchedUntilPos = 0; |
| 177 | $this->shortClosedTagNow = false; |
| 178 | } |
| 179 | |
| 180 | /** |
| 181 | * Decides whether we're to fetch more chunks from the stream or keep working with what we have. |
| 182 | * @param StreamInterface $stream The stream provider |
| 183 | * @return bool Keep working? |
| 184 | */ |
| 185 | protected function prepareChunk(StreamInterface $stream) |
| 186 | { |
| 187 | if ($this->hasSearchedUntilPos > -1 && $this->hasSearchedUntilPos < (strlen($this->workingBlob) - 1)) { |
| 188 | // More work to do |
| 189 | return true; |
| 190 | } |
| 191 | |
| 192 | $chunk = $stream->getChunk(); |
| 193 | |
| 194 | if ($chunk === false) { |
| 195 | // EOF |
| 196 | if ($this->hasSearchedUntilPos === -1) { |
| 197 | // EOF, but we haven't even started searching, special case that probably means we're dealing with a file of less size than the stream buffer |
| 198 | // Therefore, keep looping |
| 199 | return true; |
| 200 | } |
| 201 | return false; |
| 202 | } else { |
| 203 | // New chunk fetched |
| 204 | $this->workingBlob .= $chunk; |
| 205 | return true; |
| 206 | } |
| 207 | } |
| 208 | |
| 209 | /** |
| 210 | * Tries to retrieve the next node or returns false |
| 211 | * @param StreamInterface $stream The stream to use |
| 212 | * @return string|bool The next xml node or false if one could not be retrieved |
| 213 | */ |
| 214 | public function getNodeFrom(StreamInterface $stream) |
| 215 | { |
| 216 | while ($this->prepareChunk($stream)) { |
| 217 | // What's our next course of action? |
| 218 | if ($this->nextAction === self::FIND_OPENING_TAG_ACTION) { |
| 219 | // Try to find an opening tag |
| 220 | $positionInBlob = $this->getOpeningTagPos(); |
| 221 | |
| 222 | if ($positionInBlob !== false) { |
| 223 | // We found it, start salvaging |
| 224 | if ($this->options["extractContainer"] && $this->preCapture) { |
| 225 | $this->containerXml .= substr($this->workingBlob, 0, $positionInBlob); |
| 226 | $this->preCapture = false; |
| 227 | } |
| 228 | |
| 229 | |
| 230 | $this->startSalvaging($positionInBlob); |
| 231 | |
| 232 | // The next course of action will be to find a closing tag |
| 233 | $this->nextAction = self::FIND_CLOSING_TAG_ACTION; |
| 234 | } |
| 235 | } else if ($this->nextAction === self::FIND_CLOSING_TAG_ACTION) { |
| 236 | // Try to find a closing tag |
| 237 | $positionInBlob = $this->getClosingTagPos(); |
| 238 | if ($positionInBlob !== false) { |
| 239 | // We found it, we now have a full node to flush out |
| 240 | $this->flush($positionInBlob); |
| 241 | |
| 242 | // The next course of action will be to find an opening tag |
| 243 | $this->nextAction = self::FIND_OPENING_TAG_ACTION; |
| 244 | |
| 245 | // Get the flushed node and make way for the next node |
| 246 | $flushed = $this->flushed; |
| 247 | $this->flushed = ""; |
| 248 | |
| 249 | return $flushed; |
| 250 | } |
| 251 | } |
| 252 | } |
| 253 | |
| 254 | if ($this->options["extractContainer"]) { |
| 255 | $this->containerXml .= $this->workingBlob; |
| 256 | } |
| 257 | |
| 258 | return false; |
| 259 | } |
| 260 | |
| 261 | /** |
| 262 | * Get the extracted container XML, if called before the whole stream is parsed, the XML returned can be invalid due to missing closing tags |
| 263 | * @return string XML string |
| 264 | * @throws Exception if the extractContainer option isn't true |
| 265 | */ |
| 266 | public function getExtractedContainer() |
| 267 | { |
| 268 | if (!$this->options["extractContainer"]) { |
| 269 | throw new Exception("This method requires the 'extractContainer' option to be true"); |
| 270 | } |
| 271 | |
| 272 | return $this->containerXml; |
| 273 | } |
| 274 | } |