[twitter] Rework twitter_expand_entities

- Uses Twitter-provided entity indices to avoid blanket string replacement and unwarranted tag search
This commit is contained in:
Hypolite Petovan 2020-03-30 02:01:41 -04:00
parent 96925a3a75
commit 1df61e8dc2

View file

@ -1202,38 +1202,65 @@ function twitter_fetchuser(App $a, $uid, $screen_name = "", $user_id = "")
return $contact_id; return $contact_id;
} }
function twitter_expand_entities(App $a, $body, $item, $picture) /**
* Replaces Twitter entities with Friendica-friendly links.
*
* The Twitter API gives indices for each entity, which allows for fine-grained replacement.
*
* First, we need to collect everything that needs to be replaced, what we will replace it with, and the start index.
* Then we sort the indices decreasingly, and we replace from the end of the body to the start in order for the next
* index to be correct even after the last replacement.
*
* @param string $body
* @param stdClass $status
* @param string $picture
* @return array
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
function twitter_expand_entities($body, stdClass $status, $picture)
{ {
$plain = $body; $plain = $body;
$tags_arr = []; $tags = [];
foreach ($item->entities->hashtags AS $hashtag) { $replacementList = [];
$url = '#[url=' . DI::baseUrl()->get() . '/search?tag=' . $hashtag->text . ']' . $hashtag->text . '[/url]';
$tags_arr['#' . $hashtag->text] = $url; foreach ($status->entities->hashtags AS $hashtag) {
$body = str_replace('#' . $hashtag->text, $url, $body); $replace = '#[url=' . DI::baseUrl()->get() . '/search?tag=' . $hashtag->text . ']' . $hashtag->text . '[/url]';
$tags['#' . $hashtag->text] = $replace;
$replacementList[$hashtag->indices[0]] = [
'replace' => $replace,
'length' => $hashtag->indices[1] - $hashtag->indices[0],
];
} }
foreach ($item->entities->user_mentions AS $mention) { foreach ($status->entities->user_mentions AS $mention) {
$url = '@[url=https://twitter.com/' . rawurlencode($mention->screen_name) . ']' . $mention->screen_name . '[/url]'; $replace = '@[url=https://twitter.com/' . rawurlencode($mention->screen_name) . ']' . $mention->screen_name . '[/url]';
$tags_arr['@' . $mention->screen_name] = $url; $tags['@' . $mention->screen_name] = $replace;
$body = str_replace('@' . $mention->screen_name, $url, $body);
$replacementList[$mention->indices[0]] = [
'replace' => $replace,
'length' => $mention->indices[1] - $mention->indices[0],
];
} }
if (isset($item->entities->urls)) { // This URL if set will be used to add an attachment at the bottom of the post
$type = ''; $attachmentUrl = '';
$footerurl = '';
$footerlink = '';
$footer = '';
foreach ($item->entities->urls as $url) { foreach ($status->entities->urls ?? [] as $url) {
$plain = str_replace($url->url, '', $plain); $plain = str_replace($url->url, '', $plain);
if ($url->url && $url->expanded_url && $url->display_url) { if ($url->url && $url->expanded_url && $url->display_url) {
// Quote tweet, we just remove the quoted tweet URL from the body, the share block will be added later. // Quote tweet, we just remove the quoted tweet URL from the body, the share block will be added later.
if (!empty($item->quoted_status) && isset($item->quoted_status_id_str) if (!empty($status->quoted_status) && isset($status->quoted_status_id_str)
&& substr($url->expanded_url, -strlen($item->quoted_status_id_str)) == $item->quoted_status_id_str ) { && substr($url->expanded_url, -strlen($status->quoted_status_id_str)) == $status->quoted_status_id_str
$body = str_replace($url->url, '', $body); ) {
$replacementList[$url->indices[0]] = [
'replace' => '',
'length' => $url->indices[1] - $url->indices[0],
];
continue; continue;
} }
@ -1252,20 +1279,13 @@ function twitter_expand_entities(App $a, $body, $item, $picture)
$expanded_url = $url->url; $expanded_url = $url->url;
} }
if ($type == '') {
$type = $oembed_data->type;
}
if ($oembed_data->type == 'video') { if ($oembed_data->type == 'video') {
$type = $oembed_data->type; $attachmentUrl = $expanded_url;
$footerurl = $expanded_url; $replace = '';
$footerlink = '[url=' . $expanded_url . ']' . $url->display_url . '[/url]';
$body = str_replace($url->url, $footerlink, $body);
} elseif (($oembed_data->type == 'photo') && isset($oembed_data->url)) { } elseif (($oembed_data->type == 'photo') && isset($oembed_data->url)) {
$body = str_replace($url->url, '[url=' . $expanded_url . '][img]' . $oembed_data->url . '[/img][/url]', $body); $replace = '[url=' . $expanded_url . '][img]' . $oembed_data->url . '[/img][/url]';
} elseif ($oembed_data->type != 'link') { } elseif ($oembed_data->type != 'link') {
$body = str_replace($url->url, '[url=' . $expanded_url . ']' . $url->display_url . '[/url]', $body); $replace = '[url=' . $expanded_url . ']' . $url->display_url . '[/url]';
} else { } else {
$img_str = Network::fetchUrl($final_url, true, 4); $img_str = Network::fetchUrl($final_url, true, 4);
@ -1282,83 +1302,41 @@ function twitter_expand_entities(App $a, $body, $item, $picture)
unlink($tempfile); unlink($tempfile);
if (substr($mime, 0, 6) == 'image/') { if (substr($mime, 0, 6) == 'image/') {
$type = 'photo'; $replace = '[img]' . $final_url . '[/img]';
$body = str_replace($url->url, '[img]' . $final_url . '[/img]', $body);
} else { } else {
$type = $oembed_data->type; $attachmentUrl = $expanded_url;
$footerurl = $expanded_url; $replace = '';
$footerlink = '[url=' . $expanded_url . ']' . $url->display_url . '[/url]'; }
}
$body = str_replace($url->url, $footerlink, $body); $replacementList[$url->indices[0]] = [
} 'replace' => $replace,
'length' => $url->indices[1] - $url->indices[0],
];
} }
} }
krsort($replacementList);
foreach ($replacementList as $startIndex => $parameters) {
$body = Strings::substringReplace($body, $parameters['replace'], $startIndex, $parameters['length']);
} }
// Footer will be taken care of with a share block in the case of a quote // Footer will be taken care of with a share block in the case of a quote
if (empty($item->quoted_status)) { if (empty($status->quoted_status)) {
if ($footerurl != '') { $footer = '';
$footer = add_page_info($footerurl, false, $picture); if ($attachmentUrl) {
} $footer = add_page_info($attachmentUrl, false, $picture);
if (($footerlink != '') && (trim($footer) != '')) {
$removedlink = trim(str_replace($footerlink, '', $body));
if (($removedlink == '') || strstr($body, $removedlink)) {
$body = $removedlink;
} }
if (trim($footer)) {
$body .= $footer; $body .= $footer;
} } elseif ($picture) {
if ($footer == '' && $picture != '') {
$body .= "\n\n[img]" . $picture . "[/img]\n"; $body .= "\n\n[img]" . $picture . "[/img]\n";
} elseif ($footer == '' && $picture == '') { } else {
$body = add_page_info_to_body($body); $body = add_page_info_to_body($body);
} }
} }
}
// it seems as if the entities aren't always covering all mentions. So the rest will be checked here
$tags = BBCode::getTags($body);
if (count($tags)) {
foreach ($tags as $tag) {
if (strstr(trim($tag), ' ')) {
continue;
}
if (strpos($tag, '#') === 0) {
if (strpos($tag, '[url=')) {
continue;
}
// don't link tags that are already embedded in links
if (preg_match('/\[(.*?)' . preg_quote($tag, '/') . '(.*?)\]/', $body)) {
continue;
}
if (preg_match('/\[(.*?)\]\((.*?)' . preg_quote($tag, '/') . '(.*?)\)/', $body)) {
continue;
}
$basetag = str_replace('_', ' ', substr($tag, 1));
$url = '#[url=' . DI::baseUrl()->get() . '/search?tag=' . $basetag . ']' . $basetag . '[/url]';
$body = str_replace($tag, $url, $body);
$tags_arr['#' . $basetag] = $url;
} elseif (strpos($tag, '@') === 0) {
if (strpos($tag, '[url=')) {
continue;
}
$basetag = substr($tag, 1);
$url = '@[url=https://twitter.com/' . rawurlencode($basetag) . ']' . $basetag . '[/url]';
$body = str_replace($tag, $url, $body);
$tags_arr['@' . $basetag] = $url;
}
}
}
$tags = implode($tags_arr, ',');
return ['body' => $body, 'tags' => $tags, 'plain' => $plain]; return ['body' => $body, 'tags' => $tags, 'plain' => $plain];
} }
@ -1554,9 +1532,9 @@ function twitter_createpost(App $a, $uid, $post, array $self, $create_user, $onl
// Search for media links // Search for media links
$picture = twitter_media_entities($post, $postarray); $picture = twitter_media_entities($post, $postarray);
$converted = twitter_expand_entities($a, $postarray['body'], $post, $picture); $converted = twitter_expand_entities($postarray['body'], $post, $picture);
$postarray['body'] = $converted["body"]; $postarray['body'] = $converted['body'];
$postarray['tag'] = $converted["tags"]; $postarray['tag'] = implode($converted['tags'], ',');
$postarray['created'] = DateTimeFormat::utc($post->created_at); $postarray['created'] = DateTimeFormat::utc($post->created_at);
$postarray['edited'] = DateTimeFormat::utc($post->created_at); $postarray['edited'] = DateTimeFormat::utc($post->created_at);