function authorHCard(array $mf, $url, $defaultPhoto = null)
{
$hEntries = M\findMicroformatsByType($mf, 'h-entry');
if (count($hEntries) == 0) {
$hEntries[0] = ['type' => ['h-entry'], 'properties' => ['name' => '']];
}
$h = M\getAuthor($hEntries[0], $mf, $url);
if ($h === null) {
return null;
}
return ['name' => M\getPlaintext($h, 'name'), 'photo' => M\getPlaintext($h, 'photo', $defaultPhoto), 'url' => M\getPlaintext($h, 'url', null)];
}
function processHEntry($hEntry, $mf, $url, $resolveRelationships = true, Guzzle\Http\ClientInterface $client = null, $purifier = null)
{
if ($client === null) {
$client = new Guzzle\Http\Client();
}
if ($purifier === null) {
$purifier = function ($value) {
return $value;
};
}
// Use comment-presentation algorithm to clean up.
$cleansed = comments\parse($hEntry);
$referencedPosts = [];
$referencedPostUrls = [];
// Used internally to keep track of what referenced posts have been processed already.
$indexedContent = M\getPlaintext($hEntry, 'content', $cleansed['text']);
$displayContent = $purifier(M\getHtml($hEntry, 'content'));
$cleansed['content'] = $indexedContent;
$cleansed['display_content'] = $displayContent;
// Handle all datetime cases, as per http://indiewebcamp.com/h-entry#How_to_consume_h-entry
try {
$published = new DateTime($cleansed['published']);
$utcPublished = clone $published;
$utcPublished->setTimezone(new DateTimeZone('UTC'));
} catch (Exception $e) {
$published = $utcPublished = false;
}
$inTheFuture = $utcPublished > new DateTime(null, new DateTimeZone('UTC'));
// DateTime() accepts “false” as a constructor param for some reason.
if (!$published and !$cleansed['published'] or $utcPublished > new DateTime(null, new DateTimeZone('UTC'))) {
// If there’s absolutely no datetime, our best guess has to be “now”.
// Additional heuristics could be used in the bizarre case of having a feed where an item without datetime is
// published in between two items with datetimes, allowing us to guess the published datetime is between the two,
// but until that actually happens it’s not worth coding for.
$cleansed['published'] = gmdate('c');
$utcPublished = new DateTime(null, new DateTimeZone('UTC'));
} else {
// “published” is given and parses correctly, into $published.
// Currently it’s not trivial to figure out if a given datetime is floating or not, so assume that the timezone
// given here is correct for the moment. When this can be determined, follow http://indiewebcamp.com/datetime#implying_timezone_from_webmentions
}
// There’s some case causing $utcPublished to still be false and I can’t be bothered to debug it right now, so here’s a fix.
if ($utcPublished === false) {
$utcPublished = new DateTime(null, new DateTimeZone('UTC'));
}
// Store a string representation of published to be indexed+queried upon.
$cleansed['published_utc'] = $utcPublished->format(DateTime::W3C);
if (M\hasProp($hEntry, 'photo')) {
$cleansed['photo'] = $purifier(M\getHtml($hEntry, 'photo'));
}
if (M\hasProp($hEntry, 'logo')) {
$cleansed['logo'] = $purifier(M\getHtml($hEntry, 'logo'));
}
// For every post this post has a relation (in-reply-to, repost-of, like-of etc.), fetch and resolve that URL,
// index it as it’s own post (if it doesn’t already exist) and store only a reference to it here.
$references = ['in-reply-to' => [], 'like-of' => [], 'repost-of' => []];
foreach ($references as $relation => $_) {
$refUrls = [];
// These will be feed pages not permalink pages so cannot check rels, only microformats properties.
if (M\hasProp($hEntry, $relation)) {
foreach ($hEntry['properties'][$relation] as $value) {
if (is_string($value)) {
$refUrls[] = $value;
} elseif (is_array($value) and isset($value['html'])) {
// e-* properties unlikely to be URLs but try all the same.
$refUrls[] = $value['value'];
} elseif (M\isMicroformat($value)) {
if (M\hasProp($value, 'url')) {
$refUrls[] = M\getProp($value, 'url');
} elseif (M\hasProp($value, 'uid')) {
$refUrls[] = M\getProp($value, 'uid');
}
} else {
// If this happens, the microformats parsing spec has changed. Currently do nothing as we don’t know how to interpret this.
}
}
}
if ($resolveRelationships) {
foreach ($refUrls as $refUrl) {
try {
$resp = $client->get($refUrl)->send();
$refResolvedUrl = $resp->getEffectiveUrl();
$refMf = Mf2\parse($resp->getBody(1), $refResolvedUrl);
$refHEntries = M\findMicroformatsByType($refMf, 'h-entry');
$relatedUrl = $refResolvedUrl;
if (count($refHEntries) > 0) {
$refHEntry = $refHEntries[0];
$refSearchUrl = M\hasProp($refHEntry, 'url') ? M\getProp($refHEntry, 'url') : $refResolvedUrl;
if (!in_array($refSearchUrl, $referencedPostUrls)) {
list($refCleansed, $_) = processHEntry($refHEntry, $refMf, $refResolvedUrl, false, $client, $purifier);
$referencedPosts[] = $refCleansed;
$referencedPostUrls[] = $refSearchUrl;
$relatedUrl = $refSearchUrl;
}
}
$references[$relation][] = $relatedUrl;
} catch (Guzzle\Common\Exception\GuzzleException $e) {
$references[$relation][] = $refUrl;
}
}
} else {
// If we’re not resolving relationships, the most accurate data we have is the data given already.
$references[$relation] = $refUrls;
}
// Now we have the best possible list of URLs, attach it to $cleansed.
$cleansed[$relation] = array_unique($references[$relation]);
}
if (!M\hasProp($hEntry, 'author') or !M\isMicroformat($hEntry['properties']['author'][0])) {
// No authorship data given, we need to find the author!
// TODO: proper /authorship implementation.
// TODO: wrap proper /authorship implementation in layer which does purification, simplification, fallback.
$potentialAuthor = M\getAuthor($hEntry, $mf, $url);
if (M\isMicroformat($potentialAuthor)) {
$cleansed['author'] = flattenHCard($potentialAuthor, $url);
} elseif (!empty($mf['rels']['author'])) {
// TODO: look in elasticsearch index for a person with the first rel-author URL then fall back to fetching.
// Fetch the first author URL and look for a representative h-card there.
$relAuthorMf = Mf2\fetch($mf['rels']['author'][0]);
$relAuthorHCards = M\findMicroformatsByType($relAuthorMf, 'h-card');
foreach ($relAuthorHCards as $raHCard) {
$relMes = @($relAuthorMf['rels']['me'] ?: []);
if ((M\getProp($raHCard, 'url') === M\getProp($raHCard, 'url')) === $mf['rels']['author'][0]) {
$cleansed['author'] = flattenHCard($raHCard, $mf['rels']['author'][0]);
} elseif (M\hasProp($raHCard, 'url') and count(array_intersect($raHCard['properties']['url'], $relMes)) > 0) {
$cleansed['author'] = flattenHCard($raHCard, $mf['rels']['author'][0]);
}
}
}
// If after all that there’s still no authorship data, fake some.
if ($cleansed['author']['name'] === false) {
$cleansed['author'] = flattenHCard(['properties' => []], $url);
try {
$response = $client->head("{$cleansed['author']['url']}/favicon.ico")->send();
if (strpos($response->getHeader('content-type'), 'image') !== false) {
// This appears to be a valid image!
$cleansed['author']['photo'] = $response->getEffectiveUrl();
}
} catch (Guzzle\Common\Exception\GuzzleException $e) {
// No photo fallback could be found.
}
}
}
// TODO: this will be M\getLocation when it’s ported to the other library.
if (($location = getLocation($hEntry)) !== null) {
$cleansed['location'] = $location;
// TODO: do additional reverse lookups of address details if none are provided.
if (!empty($location['latitude']) and !empty($location['longitude'])) {
// If this is a valid point, add a point with mashed names for elasticsearch to index.
$cleansed['location_point'] = ['lat' => $location['latitude'], 'lon' => $location['longitude']];
}
}
// TODO: figure out what other properties need storing/indexing, and whether anything else needs mashing for
// elasticsearch to index more easily.
return [$cleansed, $referencedPosts];
}