| 1 |
<?php |
| 2 |
|
| 3 |
namespace ImportWP\Common\Importer\Preview; |
| 4 |
|
| 5 |
use ImportWP\Common\Importer\File\XMLFile; |
| 6 |
use ImportWP\Common\Importer\PreviewInterface; |
| 7 |
use XMLReader; |
| 8 |
|
| 9 |
class XMLPreview implements PreviewInterface |
| 10 |
{ |
| 11 |
|
| 12 |
/** |
| 13 |
* @var XMLFile $file |
| 14 |
*/ |
| 15 |
private $file; |
| 16 |
|
| 17 |
/** |
| 18 |
* @var \XMLReader $xml |
| 19 |
*/ |
| 20 |
private $xml; |
| 21 |
|
| 22 |
/** |
| 23 |
* @var string |
| 24 |
*/ |
| 25 |
private $record_path; |
| 26 |
|
| 27 |
/** |
| 28 |
* @var array |
| 29 |
*/ |
| 30 |
private $record_path_parts = array(); |
| 31 |
|
| 32 |
public $processed_nodes = array(); |
| 33 |
public $processed_index_nodes = array(); |
| 34 |
public $previous_node = ''; |
| 35 |
public $node_index = 0; |
| 36 |
public $last_element = ''; |
| 37 |
public $last_depth = 0; |
| 38 |
private $prefix = ''; |
| 39 |
private $xpath = false; |
| 40 |
private $class; |
| 41 |
|
| 42 |
/** |
| 43 |
* XMLPreview constructor. |
| 44 |
* |
| 45 |
* @param \ImportWP\Common\Importer\File\XMLFile $file |
| 46 |
* @param string $record_path |
| 47 |
* @param array $args |
| 48 |
*/ |
| 49 |
public function __construct(XMLFile $file, $record_path, $args = array()) |
| 50 |
{ |
| 51 |
$this->file = $file; |
| 52 |
$this->file->processing(true); |
| 53 |
$this->record_path = $record_path; |
| 54 |
|
| 55 |
if (!empty($record_path) && strpos($record_path, '/') !== false) { |
| 56 |
$this->record_path_parts = explode('/', $record_path); |
| 57 |
|
| 58 |
// remove empty first tag |
| 59 |
if (empty($this->record_path_parts[0])) { |
| 60 |
array_shift($this->record_path_parts); |
| 61 |
} |
| 62 |
} |
| 63 |
} |
| 64 |
|
| 65 |
public function data() |
| 66 |
{ |
| 67 |
$this->file->setRecordPath($this->record_path); |
| 68 |
$record = $this->file->getRecord(); |
| 69 |
|
| 70 |
$xml = new \XMLReader(); |
| 71 |
$xml->xml($record); |
| 72 |
|
| 73 |
$result = $this->parseXml2($xml); |
| 74 |
return $result; |
| 75 |
} |
| 76 |
|
| 77 |
private function generateXpath($nodes, $element = false) |
| 78 |
{ |
| 79 |
if (count($nodes) < 1) { |
| 80 |
return ''; |
| 81 |
} |
| 82 |
|
| 83 |
$result = array_slice($nodes, 2); |
| 84 |
if ($element) { |
| 85 |
$result[] = $element; |
| 86 |
} |
| 87 |
|
| 88 |
return !empty($result) ? '/' . implode('/', $result) : ''; |
| 89 |
} |
| 90 |
|
| 91 |
protected function parseXml2(\XMLReader $xml, $path = []) |
| 92 |
{ |
| 93 |
|
| 94 |
$data = []; |
| 95 |
$index = -1; |
| 96 |
|
| 97 |
while ($xml->read()) { |
| 98 |
switch ($xml->nodeType) { |
| 99 |
|
| 100 |
case XMLReader::END_ELEMENT: |
| 101 |
|
| 102 |
return $data; |
| 103 |
|
| 104 |
case XMLReader::ELEMENT: |
| 105 |
|
| 106 |
$element_path = $path; |
| 107 |
$element_path[] = $xml->name; |
| 108 |
|
| 109 |
$row = [ |
| 110 |
'node' => $xml->name, |
| 111 |
'type' => 'element', |
| 112 |
'xpath' => $this->generateXpath($element_path), |
| 113 |
]; |
| 114 |
|
| 115 |
$attributes = []; |
| 116 |
if ($xml->hasAttributes) { |
| 117 |
|
| 118 |
while ($xml->moveToNextAttribute()) { |
| 119 |
$attributes[] = ['name' => $xml->name, 'value' => $xml->value, 'xpath' => $this->generateXpath($element_path, '@' . $xml->name)]; |
| 120 |
} |
| 121 |
$xml->moveToElement(); |
| 122 |
} |
| 123 |
|
| 124 |
$row['attr'] = $attributes; |
| 125 |
|
| 126 |
if (!$xml->isEmptyElement) { |
| 127 |
$row['value'] = $this->parseXml2($xml, $element_path); |
| 128 |
} |
| 129 |
|
| 130 |
$data[] = $row; |
| 131 |
break; |
| 132 |
case XMLReader::TEXT: |
| 133 |
case XMLReader::CDATA: |
| 134 |
$row = [ |
| 135 |
'node' => 'text', |
| 136 |
'xpath' => $this->generateXpath($path), |
| 137 |
'value' => "" . $xml->value, |
| 138 |
'type' => 'text' |
| 139 |
]; |
| 140 |
$data[] = $row; |
| 141 |
break; |
| 142 |
} |
| 143 |
} |
| 144 |
|
| 145 |
return $data; |
| 146 |
} |
| 147 |
|
| 148 |
/** |
| 149 |
* Generate a nested list visualising the xml document nodes |
| 150 |
* and attributes based on the XML data being previewed. |
| 151 |
* |
| 152 |
* @return string |
| 153 |
*/ |
| 154 |
public function output() |
| 155 |
{ |
| 156 |
$this->file->setRecordPath($this->record_path); |
| 157 |
$record = $this->file->getRecord(); |
| 158 |
|
| 159 |
return $this->generatePreview($record); |
| 160 |
} |
| 161 |
|
| 162 |
/** |
| 163 |
* Load xml string into XML Previewer |
| 164 |
* |
| 165 |
* @param string $record XML Record as string |
| 166 |
* |
| 167 |
* @return string |
| 168 |
*/ |
| 169 |
private function generatePreview($record) |
| 170 |
{ |
| 171 |
$this->xml = new \XMLReader(); |
| 172 |
$this->xml->xml($record); |
| 173 |
|
| 174 |
// loop through document |
| 175 |
$output = $this->iterate_element(); |
| 176 |
return '<ul>' . $output . '</ul>'; |
| 177 |
} |
| 178 |
|
| 179 |
/** |
| 180 |
* Loop through XML document element by element. |
| 181 |
* |
| 182 |
* @return string |
| 183 |
*/ |
| 184 |
public function iterate_element() |
| 185 |
{ |
| 186 |
$output = ''; |
| 187 |
|
| 188 |
while ($this->xml->read()) { |
| 189 |
|
| 190 |
switch ($this->xml->nodeType) { |
| 191 |
case XMLReader::ELEMENT: |
| 192 |
|
| 193 |
$element_name = $this->xml->name; |
| 194 |
|
| 195 |
// remove elements from if depth has decreased |
| 196 |
if ($this->last_depth > $this->xml->depth) { |
| 197 |
|
| 198 |
for ($i = ($this->xml->depth + 1); $i < count($this->processed_index_nodes); $i++) { |
| 199 |
$this->processed_index_nodes[$i] = array(); |
| 200 |
} |
| 201 |
} |
| 202 |
|
| 203 |
// increase index of element if found |
| 204 |
$found = false; |
| 205 |
if (!empty($this->processed_index_nodes[$this->xml->depth])) { |
| 206 |
|
| 207 |
foreach ($this->processed_index_nodes[$this->xml->depth] as &$node) { |
| 208 |
|
| 209 |
if (isset($node['n']) && $node['n'] == $element_name) { |
| 210 |
$node['i'] = ($node['i'] + 1); |
| 211 |
$found = true; |
| 212 |
} |
| 213 |
} |
| 214 |
} |
| 215 |
|
| 216 |
// otherwise reset back to 1 if not found |
| 217 |
if (!$found) { |
| 218 |
$this->processed_index_nodes[$this->xml->depth][] = array('n' => $element_name, 'i' => 1); |
| 219 |
} |
| 220 |
|
| 221 |
$output .= $this->open_element($element_name); |
| 222 |
|
| 223 |
$isEmpty = $this->xml->isEmptyElement; |
| 224 |
if ($isEmpty) { |
| 225 |
$output .= $this->close_element($element_name); |
| 226 |
} |
| 227 |
|
| 228 |
|
| 229 |
break; |
| 230 |
case XMLReader::END_ELEMENT: |
| 231 |
$output .= $this->close_element($this->xml->name); |
| 232 |
break; |
| 233 |
case XMLReader::TEXT: |
| 234 |
case XMLReader::CDATA: |
| 235 |
|
| 236 |
$output .= ($this->xml->value); |
| 237 |
break; |
| 238 |
} |
| 239 |
} |
| 240 |
|
| 241 |
return $output; |
| 242 |
} |
| 243 |
|
| 244 |
/** |
| 245 |
* Create string representation of current element and its attributes. |
| 246 |
* |
| 247 |
* @param $element_name |
| 248 |
* |
| 249 |
* @return string |
| 250 |
*/ |
| 251 |
public function open_element($element_name) |
| 252 |
{ |
| 253 |
$output = ''; |
| 254 |
|
| 255 |
$this->processed_nodes[] = $this->xml->name; |
| 256 |
$this->last_element = $this->xml->name; |
| 257 |
|
| 258 |
if ($this->xml->depth > $this->last_depth) { |
| 259 |
$output .= "<ul>"; |
| 260 |
} |
| 261 |
|
| 262 |
$this->last_depth = $this->xml->depth; |
| 263 |
|
| 264 |
$attrs = ''; |
| 265 |
|
| 266 |
// generate elements xpath |
| 267 |
$temp = $this->get_xpath(); |
| 268 |
|
| 269 |
// output elements class |
| 270 |
$this->class = "class=\"xml-node xml-draggable\""; |
| 271 |
|
| 272 |
if (count($this->processed_nodes) > 1) { |
| 273 |
$attrs .= $this->get_attributes(); |
| 274 |
$output .= "<li><span {$this->class}data-xpath=\"" . $temp . "\"><{$element_name}</span>{$attrs}>"; |
| 275 |
} else { |
| 276 |
$output .= "<li><span><{$element_name}</span>{$attrs}>"; |
| 277 |
} |
| 278 |
|
| 279 |
|
| 280 |
return $output; |
| 281 |
} |
| 282 |
|
| 283 |
/** |
| 284 |
* Get xPath string for current attribute. |
| 285 |
* |
| 286 |
* @param string $attr |
| 287 |
* |
| 288 |
* @return string |
| 289 |
*/ |
| 290 |
public function get_xpath($attr = '') |
| 291 |
{ |
| 292 |
|
| 293 |
if ($this->xpath && count($this->processed_index_nodes) <= 1) { |
| 294 |
return str_replace('//', '/', $attr); |
| 295 |
} |
| 296 |
|
| 297 |
$temp = $this->gen_xpath($this->processed_nodes); |
| 298 |
|
| 299 |
|
| 300 |
if ($this->xpath !== false && !empty($this->prefix) && strpos($temp, $this->prefix) === 0) { |
| 301 |
|
| 302 |
$temp = str_replace($this->prefix, "", $temp); |
| 303 |
} |
| 304 |
|
| 305 |
$temp .= $attr; |
| 306 |
|
| 307 |
return str_replace('//', '/', $temp); |
| 308 |
} |
| 309 |
|
| 310 |
/** |
| 311 |
* Generate xPath string for current nodes. |
| 312 |
* |
| 313 |
* @param array $nodes |
| 314 |
* |
| 315 |
* @return string |
| 316 |
*/ |
| 317 |
public function gen_xpath($nodes) |
| 318 |
{ |
| 319 |
// TODO: if base node is set don't return the base node part |
| 320 |
$temp = array(); |
| 321 |
|
| 322 |
foreach ($nodes as $depth => $node) { |
| 323 |
|
| 324 |
if ($this->xpath && $depth <= 0) { |
| 325 |
continue; |
| 326 |
} |
| 327 |
|
| 328 |
foreach ($this->processed_index_nodes[$depth] as $p_node) { |
| 329 |
if ($p_node['n'] == $node) { |
| 330 |
$temp[] = $node . '[' . $p_node['i'] . ']'; |
| 331 |
} |
| 332 |
} |
| 333 |
} |
| 334 |
|
| 335 |
$temp = $this->strip_record_path($temp); |
| 336 |
|
| 337 |
return '/' . implode('/', $temp); |
| 338 |
} |
| 339 |
|
| 340 |
/** |
| 341 |
* Create string representation of element attributes |
| 342 |
* |
| 343 |
* @return string |
| 344 |
*/ |
| 345 |
public function get_attributes() |
| 346 |
{ |
| 347 |
|
| 348 |
$output = ''; |
| 349 |
|
| 350 |
// get element attributes |
| 351 |
if ($this->xml->hasAttributes) { |
| 352 |
while ($this->xml->moveToNextAttribute()) { |
| 353 |
$output .= " <span {$this->class} data-xpath=\"" . $this->get_xpath('/@' . $this->xml->name) . "\">{$this->xml->name}=\"{$this->xml->value}\"</span>"; |
| 354 |
} |
| 355 |
|
| 356 |
$this->xml->moveToElement(); |
| 357 |
} |
| 358 |
|
| 359 |
return $output; |
| 360 |
} |
| 361 |
|
| 362 |
/** |
| 363 |
* Create string representation of current closing element, |
| 364 |
* and remove it from the currently opened list. |
| 365 |
* |
| 366 |
* @param $element_name |
| 367 |
* |
| 368 |
* @return string |
| 369 |
*/ |
| 370 |
public function close_element($element_name) |
| 371 |
{ |
| 372 |
$output = ''; |
| 373 |
$prefix = ''; |
| 374 |
|
| 375 |
if ($this->last_element != $this->xml->name) { |
| 376 |
if ($this->xml->depth < $this->last_depth) { |
| 377 |
$output .= "</ul>"; |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
array_pop($this->processed_nodes); |
| 382 |
|
| 383 |
if (!$this->xml->isEmptyElement) { |
| 384 |
$output .= "{$prefix}</{$element_name}></li>"; |
| 385 |
} else { |
| 386 |
$output .= "</{$element_name}></li>"; |
| 387 |
} |
| 388 |
|
| 389 |
return $output; |
| 390 |
} |
| 391 |
|
| 392 |
/** |
| 393 |
* Generate array of elements xpath from file |
| 394 |
* |
| 395 |
* @param boolean $show_attrs |
| 396 |
* |
| 397 |
* @return array |
| 398 |
*/ |
| 399 |
public function generate_xpath($show_attrs = false) |
| 400 |
{ |
| 401 |
|
| 402 |
// retrive parsed list of nodes/attributes |
| 403 |
$nodes = $this->get_nodes($show_attrs); |
| 404 |
$list = array(); |
| 405 |
|
| 406 |
// list all possible xpath |
| 407 |
foreach ($nodes as $depth_nodes) { |
| 408 |
|
| 409 |
// display nodes |
| 410 |
foreach ($depth_nodes as $node) { |
| 411 |
$list[] = $node['xpath']; |
| 412 |
|
| 413 |
// skip generating attributes |
| 414 |
if (!$show_attrs) { |
| 415 |
continue; |
| 416 |
} |
| 417 |
|
| 418 |
// display attributes |
| 419 |
foreach ($node['attrs'] as $attr) { |
| 420 |
$list[] = $attr['xpath']; |
| 421 |
} |
| 422 |
} |
| 423 |
} |
| 424 |
|
| 425 |
return $list; |
| 426 |
} |
| 427 |
|
| 428 |
/** |
| 429 |
* Loop through file and retreive array of nodes / attributes |
| 430 |
* |
| 431 |
* @param boolean $show_attrs |
| 432 |
* |
| 433 |
* @return array |
| 434 |
*/ |
| 435 |
public function get_nodes($show_attrs = false) |
| 436 |
{ |
| 437 |
|
| 438 |
$output = array(); |
| 439 |
|
| 440 |
$previous_depth = -1; |
| 441 |
$previous_element = false; |
| 442 |
$element_index = 0; |
| 443 |
$nodes = array(); |
| 444 |
$xpath_nodes = array(); |
| 445 |
|
| 446 |
while ($this->xml->read()) { |
| 447 |
|
| 448 |
// if($this->xml->nodeType == XMLReader::TEXT || $this->xml->nodeType == XMLReader::CDATA){ |
| 449 |
// var_dump($this->xml->value); |
| 450 |
// var_dump($previous_element); |
| 451 |
// echo "===<br />"; |
| 452 |
// } |
| 453 |
|
| 454 |
if ($this->xml->nodeType == XMLReader::ELEMENT) { |
| 455 |
|
| 456 |
$d = $this->xml->depth; |
| 457 |
$e = $this->xml->name; |
| 458 |
$x = '/'; |
| 459 |
$p = false; |
| 460 |
$a = array(); |
| 461 |
$v = $this->xml->value; |
| 462 |
|
| 463 |
// var_dump($this->xml->hasValue); |
| 464 |
|
| 465 |
// use to set element index = /posts/post[0], /posts/post[1] |
| 466 |
if ($previous_element == $e) { |
| 467 |
$element_index++; |
| 468 |
} else { |
| 469 |
$element_index = 0; |
| 470 |
} |
| 471 |
|
| 472 |
// manage depth |
| 473 |
if ($d < $previous_depth) { |
| 474 |
|
| 475 |
$c = ($previous_depth - $d); |
| 476 |
for ($x = 0; $x <= $c; $x++) { |
| 477 |
|
| 478 |
// parent node |
| 479 |
array_pop($nodes); |
| 480 |
} |
| 481 |
} elseif ($d > $previous_depth) { |
| 482 |
|
| 483 |
// child node |
| 484 |
} else { |
| 485 |
|
| 486 |
// partner node, same level |
| 487 |
array_pop($nodes); |
| 488 |
} |
| 489 |
|
| 490 |
$nodes[] = $this->xml->name; |
| 491 |
|
| 492 |
// generate xpath |
| 493 |
$xpath = $x = '/' . implode('/', $nodes); |
| 494 |
if (!in_array($xpath, $xpath_nodes)) { |
| 495 |
$xpath_nodes[] = $xpath; |
| 496 |
} |
| 497 |
|
| 498 |
// get element attributes |
| 499 |
if ($show_attrs && $this->xml->hasAttributes) { |
| 500 |
while ($this->xml->moveToNextAttribute()) { |
| 501 |
$a[] = array('attr' => $this->xml->name, 'xpath' => $x . '/@' . $this->xml->name); |
| 502 |
} |
| 503 |
} |
| 504 |
|
| 505 |
// if nodes array is not empty get parent from it |
| 506 |
if (count($nodes) > 1) { |
| 507 |
$p = $nodes[count($nodes) - 2]; |
| 508 |
} |
| 509 |
|
| 510 |
// create array if no array exists for depth |
| 511 |
if (!isset($output[$d])) { |
| 512 |
$output[$d] = array(); |
| 513 |
} |
| 514 |
|
| 515 |
// check to see if element doesnt exist in depth array |
| 516 |
if (!$this->in_element_array($e, $output[$d], $x, $element_index)) { |
| 517 |
|
| 518 |
$output[$d][] = array( |
| 519 |
'node' => $e, |
| 520 |
'parent' => $p, |
| 521 |
'index' => $element_index, |
| 522 |
'xpath' => $x, |
| 523 |
'attrs' => $a, |
| 524 |
'depth' => $d, |
| 525 |
'value' => $v |
| 526 |
|
| 527 |
); |
| 528 |
} |
| 529 |
|
| 530 |
// set previous depth to current depth |
| 531 |
$previous_depth = $d; |
| 532 |
$previous_element = $e; |
| 533 |
} |
| 534 |
} |
| 535 |
|
| 536 |
return $output; |
| 537 |
} |
| 538 |
|
| 539 |
/** |
| 540 |
* Check to see if node is in array |
| 541 |
* |
| 542 |
* @param string $element node name |
| 543 |
* @param array $array array to compare against |
| 544 |
* @param string $xpath xpath string for node |
| 545 |
* |
| 546 |
* @return boolean |
| 547 |
*/ |
| 548 |
public function in_element_array($element, $array = array(), $xpath = '/', $index = 0) |
| 549 |
{ |
| 550 |
|
| 551 |
foreach ($array as $node) { |
| 552 |
|
| 553 |
if ($node['node'] == $element && $node['xpath'] == $xpath && $node['index'] == $index) { |
| 554 |
return true; |
| 555 |
} |
| 556 |
} |
| 557 |
|
| 558 |
return false; |
| 559 |
} |
| 560 |
|
| 561 |
/** |
| 562 |
* Remove the xml wrapping element <record> and next node from xpath node list |
| 563 |
* |
| 564 |
* @param array $node_list |
| 565 |
* |
| 566 |
* @return array |
| 567 |
*/ |
| 568 |
private function strip_record_path($node_list) |
| 569 |
{ |
| 570 |
|
| 571 |
$output = array(); |
| 572 |
if (count($node_list) > 2) { |
| 573 |
for ($i = 2; $i < count($node_list); $i++) { |
| 574 |
$output[] = $node_list[$i]; |
| 575 |
} |
| 576 |
} |
| 577 |
|
| 578 |
return $output; |
| 579 |
} |
| 580 |
} |
| 581 |
|
