public function html_input($default_entity_identifier)
{
// The entity can be referenced as URI or ID.
if (is_numeric($default_entity_identifier)) {
$entity = get_post($default_entity_identifier);
} else {
// It is an URI
$entity = wl_get_entity_post_by_uri($default_entity_identifier);
}
if (!is_null($entity)) {
$label = $entity->post_title;
$value = $entity->ID;
} else {
// No ID and no internal uri. Just leave as is.
$label = $default_entity_identifier;
$value = $default_entity_identifier;
}
/*
* Write saved value in page
* The <input> tags host the meta value.
* The visible <input> has the human readable value (i.e. entity name or uri)
* and is accompained by an hidden <input> tag, passed to the server,
* that contains the raw value (i.e. the uri or entity id).
*/
$html = <<<EOF
\t\t\t<div class="wl-input-wrapper wl-autocomplete-wrapper">
\t\t\t\t<input type="text" class="{$this->meta_name} wl-autocomplete" value="{$label}" style="width:88%" />
\t\t\t\t<input type="hidden" class="{$this->meta_name}" name="wl_metaboxes[{$this->meta_name}][]" value="{$value}" />
\t\t\t\t<button class="button wl-remove-input wl-button" type="button" style="width:10%">Remove</button>
\t\t\t\t<div class="wl-input-notice"></div>
\t\t\t</div>
EOF;
return $html;
}
function testFindByURI()
{
$entity_post_id = wl_create_post('', 'test_entity', 'Test Entity', 'draft', 'entity');
$entity_uri = wl_get_entity_uri($entity_post_id);
wl_schema_set_value($entity_post_id, 'sameAs', 'http://example.org/entity/test_entity');
$same_as_array = wl_schema_get_value($entity_post_id, 'sameAs');
$this->assertTrue(is_array($same_as_array));
$this->assertEquals('http://example.org/entity/test_entity', $same_as_array[0]);
wl_schema_set_value($entity_post_id, 'sameAs', array('http://example.org/entity/test_entity', 'http://data.example.org/entity/test_entity'));
$same_as_array = wl_schema_get_value($entity_post_id, 'sameAs');
$this->assertTrue(is_array($same_as_array));
$this->assertEquals('http://example.org/entity/test_entity', $same_as_array[0]);
$this->assertEquals('http://data.example.org/entity/test_entity', $same_as_array[1]);
$post = wl_get_entity_post_by_uri('http://example.org/entity/test_entity');
$this->assertNotNull($post);
$post = wl_get_entity_post_by_uri('http://data.example.org/entity/test_entity');
$this->assertNotNull($post);
$same_as_uri = 'http://example.org/entity/test_entity2';
$entity_post_id = wl_create_post('', 'test_entity_2', 'Test Entity 2', 'draft', 'entity');
$entity_uri = wl_get_entity_uri($entity_post_id);
wl_schema_set_value($entity_post_id, 'sameAs', $same_as_uri);
$same_as_array = wl_schema_get_value($entity_post_id, 'sameAs');
$this->assertTrue(is_array($same_as_array));
$this->assertEquals($same_as_uri, $same_as_array[0]);
$post = wl_get_entity_post_by_uri('http://example.org/entity/test_entity');
$this->assertNotNull($post);
}
function testContentParsing()
{
// Create the sample entity for testing.
$this->createSampleEntity();
$content = '<span class="textannotation highlight organization disambiguated" id="urn:enhancement-16e9b0f6-e792-5b75-ffb7-ec40916d8753" itemid="http://dbpedia.org/resource/Honda" itemscope="itemscope" itemtype="organization">Honda</span> is recalling nearly 900,000 minivans for a defect that could increase fire risk.';
$matches = array();
if (0 < preg_match_all('/ itemid="([^"]+)"/im', $content, $matches, PREG_SET_ORDER)) {
foreach ($matches as $match) {
$item_id = $match[1];
$post = wl_get_entity_post_by_uri($item_id);
$this->assertNotNull($post);
$uri = wl_get_entity_uri($post->ID);
$this->assertNotNull($uri);
$this->assertNotEquals($item_id, $uri);
}
}
$this->assertTrue(0 < count($matches));
}
function testEntityMetadataAreProperlyUpdated()
{
$fake = $this->prepareFakeGlobalPostArrayFromFile('/assets/fake_global_post_array_with_one_entity_linked_as_what.json');
$_POST = $fake;
// Retrieve the entity uri (the first key in wl_entities associative aray)
$original_entity_uri = current(array_keys($fake['wl_entities']));
// Reference the entity to the post content
$content = <<<EOF
<span itemid="{$original_entity_uri}">My entity</span>
EOF;
// Create a post referincing to the created entity
$post_id = wl_create_post($content, 'my-post', 'A post', 'draft');
// Here the entity should have been created
$original_entity = wl_get_entity_post_by_uri($original_entity_uri);
$this->assertNotNull($original_entity);
// Store entity type, images and sameAs (needed later)
$original_type = wl_schema_get_types($original_entity->ID);
$original_thumbnails = $this->getThumbs($original_entity->ID);
$original_sameAs = wl_schema_get_value($original_entity->ID, 'sameAs');
// Query the same entity using the Redlink URI
$entity_uri = wl_get_entity_uri($original_entity->ID);
$e = wl_get_entity_post_by_uri($entity_uri);
$this->assertEquals($original_entity, $e);
// The entity description should be the same we expect
$raw_entity = current(array_values($fake['wl_entities']));
$this->assertEquals($raw_entity['description'], $original_entity->post_content);
// The entity is related as what predicate
$related_entity_ids = wl_core_get_related_entity_ids($post_id, array("predicate" => "what"));
$this->assertCount(1, $related_entity_ids);
// Ensure there are no other relation instances
$relation_instances = wl_tests_get_relation_instances_for($post_id);
$this->assertCount(1, $relation_instances);
/* Now Post is saved again with the same mentioned entity:
* - with different type
* - with different description
* - with one more image
* - with one modified sameAs
* - as WHO instead fo WHAT
*/
$fake = $this->prepareFakeGlobalPostArrayFromFile('/assets/fake_global_post_array_with_one_entity_linked_as_who_and_modified_data.json');
$_POST = $fake;
// The entity url should be the same we expect
$raw_entity = current(array_values($fake['wl_entities']));
$raw_entity_uri = $raw_entity['uri'];
$new_content = <<<EOF
<span itemid="{$raw_entity_uri}">My entity</span>
EOF;
// Update the post content (to force existing entities update)
wp_update_post(array('ID' => $post_id, 'post_content' => $new_content));
// Verify the mentioned entity was already into DB...
$updated_entity = wl_get_entity_post_by_uri($raw_entity_uri);
$this->assertEquals($original_entity->ID, $updated_entity->ID);
$this->assertEquals($original_entity->post_title, $updated_entity->post_title);
// ... but some properties changed!
$this->assertNotEquals($original_entity, $updated_entity);
// Verify entity type has been updated
$updated_type = wl_schema_get_types($updated_entity->ID);
$this->assertNotEquals($original_type, $updated_type);
$this->assertEquals(array('http://schema.org/Organization'), $updated_type);
// Verify entity description has been updated
$this->assertEquals($raw_entity['description'], $updated_entity->post_content);
// Verify entity images have been updated (one was added)
$updated_thumbnails = $this->getThumbs($updated_entity->ID);
$this->assertNotEquals($original_thumbnails, $updated_thumbnails);
$this->assertContains($original_thumbnails[0], $updated_thumbnails);
$this->assertCount(2, $updated_thumbnails);
// There is one more
$this->assertContains('Netherlands_vs_Ivory_Coast', $updated_thumbnails[1]);
// ... about soccer
// Verify entity sameAs have been updated
$updated_sameAs = wl_schema_get_value($updated_entity->ID, 'sameAs');
$this->assertNotEquals($original_sameAs, $updated_sameAs);
$this->assertContains($original_sameAs[1], $updated_sameAs);
$this->assertNotContains($original_sameAs[0], $updated_sameAs);
$this->assertContains('http://sv.dbpedia.org/page/Reason', $updated_sameAs);
// Verify the entity is now related as who predicate
$related_entity_ids = wl_core_get_related_entity_ids($post_id, array("predicate" => "who"));
$this->assertCount(1, $related_entity_ids);
// Ensure there are no other relation instances
$relation_instances = wl_tests_get_relation_instances_for($post_id);
$this->assertCount(1, $relation_instances);
}
/**
* Saves the values of wordlift metaboxes set in the entity editor page
*/
function wl_entity_metabox_save($post_id)
{
if (!isset($_POST['wl_metaboxes'])) {
return;
}
// Loop over the wl_metaboxes array and save metaboxes values
foreach ($_POST['wl_metaboxes'] as $meta_name => $meta_values) {
// First, verify nonce is set for this meta
$nonce_name = 'wordlift_' . $meta_name . '_entity_box_nonce';
$nonce_verify = 'wordlift_' . $meta_name . '_entity_box';
if (!isset($_POST[$nonce_name])) {
return $post_id;
}
// Verify that the nonce is valid.
if (!wp_verify_nonce($_POST[$nonce_name], $nonce_verify)) {
return $post_id;
}
// Delete values before updating
delete_post_meta($post_id, $meta_name);
// Save the property value(s)
if (isset($meta_name) && isset($meta_values) && $meta_values !== '') {
// There can be one or more property values, so we force to array:
if (!is_array($meta_values)) {
$meta_values = array($meta_values);
}
foreach ($meta_values as $meta_value) {
// If the meta expects an entity...
$expecting_uri = wl_get_meta_type($meta_name) === WL_DATA_TYPE_URI;
// ...and the user inputs an entity that is not present in the db...
$absent_from_db = is_null(wl_get_entity_post_by_uri($meta_value));
// ...and that is not a external uri
$name_is_uri = strpos($meta_value, 'http') === 0;
if ($expecting_uri && $absent_from_db && !$name_is_uri) {
// ...we create a new entity!
$new_entity = wl_save_entity('', $meta_value, WL_ENTITY_TYPE_NAME, '');
// Assign type
$constraints = wl_get_meta_constraints($meta_name);
$type = 'http://schema.org/' . $constraints['uri_type'];
wl_set_entity_main_type($new_entity->ID, $type);
// TODO: directly publish the new entity
// Update the value that will be saved as meta
$meta_value = wl_get_entity_uri($new_entity->ID);
}
// TODO: use WL methods
add_post_meta($post_id, $meta_name, $meta_value);
}
}
}
// Push changes on RedLink
wl_linked_data_push_to_redlink($post_id);
}
/**
* Get an array of entities from the *itemid* attributes embedded in the provided content.
*
* @since 3.0.0
*
* @param string $content The content with itemid attributes.
*
* @return array An array of entity posts.
*/
function wl_linked_data_content_get_embedded_entities($content)
{
// Remove quote escapes.
$content = str_replace('\\"', '"', $content);
// Match all itemid attributes.
$pattern = '/<\\w+[^>]*\\sitemid="([^"]+)"[^>]*>/im';
wl_write_log("Getting entities embedded into content [ pattern :: {$pattern} ]");
// Remove the pattern while it is found (match nested annotations).
$matches = array();
// In case of errors, return an empty array.
if (false === preg_match_all($pattern, $content, $matches)) {
wl_write_log("Found no entities embedded in content");
return array();
}
// wl_write_log("wl_update_related_entities [ content :: $content ][ data :: " . var_export($data, true). " ][ matches :: " . var_export($matches, true) . " ]");
// Collect the entities.
$entities = array();
foreach ($matches[1] as $uri) {
$uri_d = html_entity_decode($uri);
$entity = wl_get_entity_post_by_uri($uri_d);
if (null !== $entity) {
array_push($entities, $entity->ID);
}
}
$count = sizeof($entities);
wl_write_log("Found {$count} entities embedded in content");
return $entities;
}
function testEntityAdditionalPropertiesAreSaved()
{
$fake = $this->prepareFakeGlobalPostArrayFromFile('/assets/fake_global_post_array_with_a_new_entity_linked_as_where_with_coordinates.json');
$_POST = $fake;
// Retrieve the entity uri (the first key in wl_entities associative aray)
$entity_uri = current(array_keys($fake['wl_entities']));
// Retrieve the label and compose expected uri
$raw_entity = current(array_values($fake['wl_entities']));
$expected_entity_uri = $this->buildEntityUriForLabel($raw_entity['label']);
// Reference the entity to the post content
$content = <<<EOF
<span itemid="{$entity_uri}">My entity</span>
EOF;
// Create a post referincing to the created entity
$post_id = wl_create_post($content, 'my-post', 'A post', 'draft');
// Here the entity should have been created
$entity = wl_get_entity_post_by_uri($expected_entity_uri);
$this->assertNotNull($entity);
// Verify association to post as where
$related_entity_ids = wl_core_get_related_entity_ids($post_id, array("predicate" => "where"));
$this->assertEquals(array($entity->ID), $related_entity_ids);
// Verify schema type
$this->assertEquals(array('http://schema.org/Place'), wl_schema_get_types($entity->ID));
// Verify coordinates
$this->assertEquals(array(43.21), wl_schema_get_value($entity->ID, 'latitude'));
$this->assertEquals(array(12.34), wl_schema_get_value($entity->ID, 'longitude'));
}
/**
* Replaces the *itemid* attributes URIs with the WordLift URIs.
*
* @param string $content The post content.
*
* @return string The updated post content.
*/
function wl_replace_item_id_with_uri($content)
{
// wl_write_log( "wl_replace_item_id_with_uri" );
// Strip slashes, see https://core.trac.wordpress.org/ticket/21767
$content = stripslashes($content);
// If any match are found.
$matches = array();
if (0 < preg_match_all('/ itemid="([^"]+)"/i', $content, $matches, PREG_SET_ORDER)) {
foreach ($matches as $match) {
// Get the item ID.
$item_id = $match[1];
// Get the post bound to that item ID (looking both in the 'official' URI and in the 'same-as' .
$post = wl_get_entity_post_by_uri($item_id);
// If no entity is found, continue to the next one.
if (null === $post) {
continue;
}
// Get the URI for that post.
$uri = wl_get_entity_uri($post->ID);
// wl_write_log( "wl_replace_item_id_with_uri [ item id :: $item_id ][ uri :: $uri ]" );
// If the item ID and the URI differ, replace the item ID with the URI saved in WordPress.
if ($item_id !== $uri) {
$uri_e = esc_html($uri);
$content = str_replace(" itemid=\"{$item_id}\"", " itemid=\"{$uri_e}\"", $content);
}
}
}
// Reapply slashes.
$content = addslashes($content);
return $content;
}
/**
* Fills up the microdata_template with entity's values.
*
* @param string $entity_id An entity ID.
* @param string $entity_type Entity type stracture.
* @param integer $recursion_level Recursion depth level in microdata compiling. Recursion depth limit is defined by WL_MAX_NUM_RECURSIONS_WHEN_PRINTING_MICRODATA constant.
*
* @return string The content with embedded microdata.
*/
function wl_content_embed_compile_microdata_template($entity_id, $entity_type, $recursion_level = 0)
{
wl_write_log("[ entity id :: {$entity_id} ][ entity type :: " . var_export($entity_type, true) . " ][ recursion level :: {$recursion_level} ]");
$regex = '/{{(.*?)}}/';
$matches = array();
if (null === $entity_type) {
return '';
}
$template = $entity_type['microdata_template'];
// Return empty string if template fields have not been found.
if (false === preg_match_all($regex, $template, $matches, PREG_SET_ORDER)) {
return '';
}
foreach ($matches as $match) {
$placeholder = $match[0];
$field_name = $match[1];
// Get property value.
$meta_collection = wl_schema_get_value($entity_id, $field_name);
// If no value is given, just remove the placeholder from the template
if (null == $meta_collection) {
$template = str_replace($placeholder, '', $template);
continue;
}
// What kind of value is it?
// TODO: Performance issue here: meta type retrieving should be centralized
$expected_type = wl_get_meta_type($field_name);
foreach ($meta_collection as $field_value) {
if (WL_DATA_TYPE_URI == $expected_type) {
// If is a numeric value we assume it is an ID referencing for an internal entity.
if (is_numeric($field_value)) {
// Found id, get uri.
$field_value = wl_get_entity_uri($field_value);
}
// Just if the linked entity does exist I can go further with template compiling
$nested_entity = wl_get_entity_post_by_uri($field_value);
if (!is_null($nested_entity)) {
$content = '<span itemid="' . esc_attr($field_value) . '">' . $nested_entity->post_title . '</span>';
$compiled_template = wl_content_embed_item_microdata($content, $field_value, $field_name, ++$recursion_level);
$template = str_replace($placeholder, $compiled_template, $template);
} else {
$template = str_replace($placeholder, '', $template);
}
continue;
}
// Standard condition: field containing a raw value
$value = '<span itemprop="' . esc_attr($field_name) . '" content="' . esc_attr($field_value) . '"></span>';
$template = str_replace($placeholder, $value, $template);
}
}
return $template;
}
/**
* Fills up the microdata_template with entity's values.
*
* @param string $entity_id An entity ID.
* @param string $entity_type Entity type structure.
* @param integer $recursion_level Recursion depth level in microdata compiling. Recursion depth limit is defined by WL_MAX_NUM_RECURSIONS_WHEN_PRINTING_MICRODATA constant.
*
* @return string The content with embedded microdata.
*/
function wl_content_embed_compile_microdata_template($entity_id, $entity_type, $recursion_level = 0)
{
global $wl_logger;
if (WP_DEBUG) {
$wl_logger->trace("Embedding microdata [ entity id :: {$entity_id} ][ entity type :: " . var_export($entity_type, true) . " ][ recursion level :: {$recursion_level} ]");
}
$regex = '/{{(.*?)}}/';
$matches = array();
if (null === $entity_type) {
return '';
}
$template = $entity_type['microdata_template'];
// Return empty string if template fields have not been found.
if (false === preg_match_all($regex, $template, $matches, PREG_SET_ORDER)) {
return '';
}
foreach ($matches as $match) {
$placeholder = $match[0];
$field_name = $match[1];
// Get property value.
$meta_collection = wl_schema_get_value($entity_id, $field_name);
// If no value is given, just remove the placeholder from the template
if (null == $meta_collection) {
$template = str_replace($placeholder, '', $template);
continue;
}
// What kind of value is it?
// TODO: Performance issue here: meta type retrieving should be centralized
$expected_type = wl_get_meta_type($field_name);
if (WP_DEBUG) {
$wl_logger->trace("Embedding microdata [ placeholder :: {$placeholder} ][ field name :: {$field_name} ][ meta collection :: " . (is_array($meta_collection) ? var_export($meta_collection, true) : $meta_collection) . " ][ expected type :: {$expected_type} ]");
}
foreach ($meta_collection as $field_value) {
// Quick and dirty patch for #163:
// - only apply to URIs, i.e. to properties pointing to another post ( $field_value should be a post ID ),
// - check that $field_value is actually a number,
// - check that the referenced post is published.
// OR
// - if the value is empty then we don't display it.
if (Wordlift_Schema_Service::DATA_TYPE_URI === $expected_type && is_numeric($field_value) && 'publish' !== ($post_status = get_post_status($field_value)) || empty($field_value)) {
if (WP_DEBUG) {
$wl_logger->trace("Microdata refers to a non-published post [ field value :: {$field_value} ][ post status :: {$post_status} ]");
}
// Remove the placeholder.
$template = str_replace($placeholder, '', $template);
continue;
}
if (Wordlift_Schema_Service::DATA_TYPE_URI == $expected_type) {
// If is a numeric value we assume it is an ID referencing for an internal entity.
if (is_numeric($field_value)) {
// Found id, get uri.
$field_value = wl_get_entity_uri($field_value);
}
// Just if the linked entity does exist I can go further with template compiling
$nested_entity = wl_get_entity_post_by_uri($field_value);
if (!is_null($nested_entity)) {
$content = '<span itemid="' . esc_attr($field_value) . '">' . $nested_entity->post_title . '</span>';
$compiled_template = wl_content_embed_item_microdata($content, $field_value, $field_name, ++$recursion_level);
$template = str_replace($placeholder, $compiled_template, $template);
} else {
$template = str_replace($placeholder, '', $template);
}
continue;
}
// Standard condition: field containing a raw value
// For non visible test, schema.org dictates to use the *meta* tag.
// see http://schema.org/docs/gs.html#advanced_missing
$value = '<meta itemprop="' . esc_attr($field_name) . '" content="' . esc_attr($field_value) . '" />';
$template = str_replace($placeholder, $value, $template);
}
}
return $template;
}
