Révision 0b4524f6
Ajouté par Assos Assos il y a presque 9 ans
| drupal7/sites/all/modules/feeds_xpathparser/FeedsXPathParserBase.inc | ||
|---|---|---|
| 10 | 10 |
*/ |
| 11 | 11 |
abstract class FeedsXPathParserBase extends FeedsParser {
|
| 12 | 12 |
|
| 13 |
/** |
|
| 14 |
* The DOMDocument used for parsing. |
|
| 15 |
* |
|
| 16 |
* @var DOMDocument |
|
| 17 |
*/ |
|
| 18 |
protected $doc; |
|
| 19 |
|
|
| 20 |
/** |
|
| 21 |
* The return value of libxml_disable_entity_loader(). |
|
| 22 |
* |
|
| 23 |
* @var bool |
|
| 24 |
*/ |
|
| 25 |
protected $loader; |
|
| 26 |
|
|
| 27 |
/** |
|
| 28 |
* The elements that should be displayed in raw XML. |
|
| 29 |
* |
|
| 30 |
* @var array |
|
| 31 |
*/ |
|
| 13 | 32 |
protected $rawXML = array(); |
| 14 |
protected $doc = NULL; |
|
| 15 |
protected $xpath = NULL; |
|
| 33 |
|
|
| 34 |
/** |
|
| 35 |
* The DOMXPath objet used for parsing. |
|
| 36 |
* |
|
| 37 |
* @var DOMXPath |
|
| 38 |
*/ |
|
| 39 |
protected $xpath; |
|
| 16 | 40 |
|
| 17 | 41 |
/** |
| 18 | 42 |
* Classes that use FeedsXPathParserBase must implement this. |
| ... | ... | |
| 82 | 106 |
|
| 83 | 107 |
$all_nodes = $this->xpath->namespacedQuery($context_query, NULL, 'context'); |
| 84 | 108 |
|
| 109 |
// The source config could have old values that don't exist in the importer. |
|
| 110 |
$sources = array_intersect_key($source_config['sources'], $mappings); |
|
| 111 |
|
|
| 85 | 112 |
foreach ($all_nodes as $node) {
|
| 86 | 113 |
// Invoke a hook to check whether the domnode should be skipped. |
| 87 | 114 |
if (in_array(TRUE, module_invoke_all('feeds_xpathparser_filter_domnode', $node, $this->doc, $source), TRUE)) {
|
| ... | ... | |
| 89 | 116 |
} |
| 90 | 117 |
|
| 91 | 118 |
$parsed_item = $variables = array(); |
| 92 |
foreach ($source_config['sources'] as $element_key => $query) {
|
|
| 119 |
foreach ($sources as $element_key => $query) {
|
|
| 93 | 120 |
// Variable substitution. |
| 94 | 121 |
$query = strtr($query, $variables); |
| 95 | 122 |
// Parse the item. |
| 96 | 123 |
$result = $this->parseSourceElement($query, $node, $element_key); |
| 97 | 124 |
if (isset($result)) {
|
| 98 |
if (!is_array($result)) {
|
|
| 99 |
$variables['$' . $mappings[$element_key]] = $result; |
|
| 100 |
} |
|
| 101 |
else {
|
|
| 102 |
$variables['$' . $mappings[$element_key]] = ''; |
|
| 103 |
} |
|
| 125 |
$variables['$' . $mappings[$element_key]] = is_array($result) ? reset($result) : $result; |
|
| 104 | 126 |
$parsed_item[$element_key] = $result; |
| 105 | 127 |
} |
| 106 | 128 |
} |
| ... | ... | |
| 532 | 554 |
*/ |
| 533 | 555 |
protected function errorStart() {
|
| 534 | 556 |
libxml_clear_errors(); |
| 557 |
if (function_exists('libxml_disable_entity_loader')) {
|
|
| 558 |
$this->loader = libxml_disable_entity_loader(TRUE); |
|
| 559 |
} |
|
| 560 |
|
|
| 535 | 561 |
return libxml_use_internal_errors(TRUE); |
| 536 | 562 |
} |
| 537 | 563 |
|
| ... | ... | |
| 567 | 593 |
} |
| 568 | 594 |
libxml_clear_errors(); |
| 569 | 595 |
libxml_use_internal_errors($use); |
| 596 |
|
|
| 597 |
if (function_exists('libxml_disable_entity_loader') && isset($this->loader)) {
|
|
| 598 |
libxml_disable_entity_loader($this->loader); |
|
| 599 |
unset($this->loader); |
|
| 600 |
} |
|
| 570 | 601 |
} |
| 571 | 602 |
|
| 572 | 603 |
/** |
| drupal7/sites/all/modules/feeds_xpathparser/FeedsXPathParserDOMXPath.inc | ||
|---|---|---|
| 36 | 36 |
* |
| 37 | 37 |
* @var arrray |
| 38 | 38 |
*/ |
| 39 |
protected $namepsaces = array();
|
|
| 39 |
protected $namespaces = array();
|
|
| 40 | 40 |
|
| 41 | 41 |
/** |
| 42 | 42 |
* The most recent error from parsing. |
| drupal7/sites/all/modules/feeds_xpathparser/FeedsXPathParserHTML.inc | ||
|---|---|---|
| 10 | 10 |
*/ |
| 11 | 11 |
class FeedsXPathParserHTML extends FeedsXPathParserBase {
|
| 12 | 12 |
|
| 13 |
/** |
|
| 14 |
* Whether this version of PHP has a useable saveHTML() method. |
|
| 15 |
* |
|
| 16 |
* @var bool |
|
| 17 |
*/ |
|
| 13 | 18 |
protected $hasSaveHTML = FALSE; |
| 14 | 19 |
|
| 15 | 20 |
/** |
| 16 |
* Overrides parent::__construct().
|
|
| 21 |
* {@inheritdoc}
|
|
| 17 | 22 |
*/ |
| 18 | 23 |
public function __construct($id) {
|
| 19 | 24 |
parent::__construct($id); |
| ... | ... | |
| 25 | 30 |
} |
| 26 | 31 |
|
| 27 | 32 |
/** |
| 28 |
* Implements FeedsXPathParserBase::setup().
|
|
| 33 |
* {@inheritdoc}
|
|
| 29 | 34 |
*/ |
| 30 | 35 |
protected function setup($source_config, FeedsFetcherResult $fetcher_result) {
|
| 31 |
|
|
| 32 | 36 |
if (!empty($source_config['exp']['tidy']) && extension_loaded('tidy')) {
|
| 33 | 37 |
$config = array( |
| 34 | 38 |
'merge-divs' => FALSE, |
| ... | ... | |
| 42 | 46 |
); |
| 43 | 47 |
// Default tidy encoding is UTF8. |
| 44 | 48 |
$encoding = $source_config['exp']['tidy_encoding']; |
| 45 |
$raw = tidy_repair_string(trim($fetcher_result->getRaw()), $config, $encoding);
|
|
| 49 |
$raw = tidy_repair_string($fetcher_result->getRaw(), $config, $encoding);
|
|
| 46 | 50 |
} |
| 47 | 51 |
else {
|
| 48 | 52 |
$raw = $fetcher_result->getRaw(); |
| 49 | 53 |
} |
| 50 |
$doc = new DOMDocument(); |
|
| 54 |
|
|
| 55 |
$document = new DOMDocument(); |
|
| 56 |
$document->strictErrorChecking = FALSE; |
|
| 57 |
$document->recover = TRUE; |
|
| 58 |
|
|
| 51 | 59 |
// Use our own error handling. |
| 52 | 60 |
$use = $this->errorStart(); |
| 53 |
$success = $doc->loadHTML($raw); |
|
| 54 |
unset($raw); |
|
| 61 |
|
|
| 62 |
if (version_compare(PHP_VERSION, '5.4.0', '>=')) {
|
|
| 63 |
$options = LIBXML_NONET; |
|
| 64 |
$options |= defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0;
|
|
| 65 |
$options |= defined('LIBXML_PARSEHUGE') ? LIBXML_PARSEHUGE : 0;
|
|
| 66 |
|
|
| 67 |
$success = $document->loadHTML($raw, $options); |
|
| 68 |
} |
|
| 69 |
else {
|
|
| 70 |
$success = $document->loadHTML($raw); |
|
| 71 |
} |
|
| 72 |
|
|
| 55 | 73 |
$this->errorStop($use, $source_config['exp']['errors']); |
| 74 |
|
|
| 56 | 75 |
if (!$success) {
|
| 57 | 76 |
throw new Exception(t('There was an error parsing the HTML document.'));
|
| 58 | 77 |
} |
| 59 |
return $doc; |
|
| 78 |
|
|
| 79 |
return $document; |
|
| 60 | 80 |
} |
| 61 | 81 |
|
| 82 |
/** |
|
| 83 |
* {@inheritdoc}
|
|
| 84 |
*/ |
|
| 62 | 85 |
protected function getRaw(DOMNode $node) {
|
| 63 | 86 |
if ($this->hasSaveHTML) {
|
| 64 | 87 |
return $this->doc->saveHTML($node); |
| ... | ... | |
| 66 | 89 |
|
| 67 | 90 |
return $this->doc->saveXML($node, LIBXML_NOEMPTYTAG); |
| 68 | 91 |
} |
| 92 |
|
|
| 69 | 93 |
} |
| drupal7/sites/all/modules/feeds_xpathparser/FeedsXPathParserXML.inc | ||
|---|---|---|
| 11 | 11 |
class FeedsXPathParserXML extends FeedsXPathParserBase {
|
| 12 | 12 |
|
| 13 | 13 |
/** |
| 14 |
* Implements FeedsXPathParserBase::setup().
|
|
| 14 |
* {@inheritdoc}
|
|
| 15 | 15 |
*/ |
| 16 | 16 |
protected function setup($source_config, FeedsFetcherResult $fetcher_result) {
|
| 17 |
|
|
| 18 | 17 |
if (!empty($source_config['exp']['tidy']) && extension_loaded('tidy')) {
|
| 19 | 18 |
$config = array( |
| 20 | 19 |
'input-xml' => TRUE, |
| ... | ... | |
| 23 | 22 |
); |
| 24 | 23 |
// Default tidy encoding is UTF8. |
| 25 | 24 |
$encoding = $source_config['exp']['tidy_encoding']; |
| 26 |
$raw = tidy_repair_string(trim($fetcher_result->getRaw()), $config, $encoding);
|
|
| 25 |
$raw = tidy_repair_string($fetcher_result->getRaw(), $config, $encoding);
|
|
| 27 | 26 |
} |
| 28 | 27 |
else {
|
| 29 | 28 |
$raw = $fetcher_result->getRaw(); |
| 30 | 29 |
} |
| 31 |
$doc = new DOMDocument(); |
|
| 30 |
|
|
| 31 |
$options = LIBXML_NONET; |
|
| 32 |
$options |= defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0;
|
|
| 33 |
$options |= defined('LIBXML_PARSEHUGE') ? LIBXML_PARSEHUGE : 0;
|
|
| 34 |
|
|
| 35 |
$document = new DOMDocument(); |
|
| 36 |
$document->strictErrorChecking = FALSE; |
|
| 37 |
$document->recover = TRUE; |
|
| 38 |
|
|
| 32 | 39 |
$use = $this->errorStart(); |
| 33 |
$success = $doc->loadXML($raw); |
|
| 34 |
unset($raw); |
|
| 40 |
|
|
| 41 |
$success = $document->loadXML($raw, $options); |
|
| 42 |
|
|
| 35 | 43 |
$this->errorStop($use, $source_config['exp']['errors']); |
| 44 |
|
|
| 36 | 45 |
if (!$success) {
|
| 37 | 46 |
throw new Exception(t('There was an error parsing the XML document.'));
|
| 38 | 47 |
} |
| 39 |
return $doc; |
|
| 48 |
|
|
| 49 |
return $document; |
|
| 40 | 50 |
} |
| 41 | 51 |
|
| 52 |
/** |
|
| 53 |
* {@inheritdoc}
|
|
| 54 |
*/ |
|
| 42 | 55 |
protected function getRaw(DOMNode $node) {
|
| 43 | 56 |
return $this->doc->saveXML($node); |
| 44 | 57 |
} |
| 58 |
|
|
| 45 | 59 |
} |
| drupal7/sites/all/modules/feeds_xpathparser/feeds_xpathparser.info | ||
|---|---|---|
| 14 | 14 |
files[] = tests/feeds_xpathparser_query_parser.test |
| 15 | 15 |
core = 7.x |
| 16 | 16 |
|
| 17 |
; Information added by Drupal.org packaging script on 2014-12-11
|
|
| 18 |
version = "7.x-1.0"
|
|
| 17 |
; Information added by Drupal.org packaging script on 2015-07-03
|
|
| 18 |
version = "7.x-1.1"
|
|
| 19 | 19 |
core = "7.x" |
| 20 | 20 |
project = "feeds_xpathparser" |
| 21 |
datestamp = "1418270885"
|
|
| 21 |
datestamp = "1435895044"
|
|
| 22 | 22 |
|
Formats disponibles : Unified diff
Weekly update of contrib modules