Filtering in all speech URLS for images and videos

This commit is contained in:
Paul Kirby 2018-10-31 13:17:42 -05:00
parent b758242410
commit b2b16bb129
2 changed files with 312 additions and 218 deletions

View file

@ -17,45 +17,32 @@ class LBRY_Speech
public function __construct() public function __construct()
{ {
add_action('save_post', array($this, 'upload_media'), 10, 2);
if (!is_admin()) {
$this->parser = new LBRY_Speech_Parser(); $this->parser = new LBRY_Speech_Parser();
add_filter('wp_calculate_image_srcset', array($this->parser, 'speech_image_srcset'), 10, 5);
if (is_admin()) {
add_action('save_post', array($this, 'upload_media'), 10, 2);
} else {
// Replace the image srcsets
add_filter('wp_calculate_image_srcset', array($this->parser, 'replace_image_srcset'), 10, 5);
// Core filter for lots of image source calls // Core filter for lots of image source calls
add_filter('wp_get_attachment_image_src', array($this->parser, 'replace_attachment_image_src'), 10, 3); add_filter('wp_get_attachment_image_src', array($this->parser, 'replace_attachment_image_src'), 10, 3);
// Replace any left over urls with speech urls
add_filter('the_content', array($this->parser, 'replace_urls_with_speech')); add_filter('the_content', array($this->parser, 'replace_urls_with_speech'));
} }
} }
/**
* Checks to see if we need to rewrite URLS, does if necessary
*/
public function maybe_rewrite_urls()
{
// See if we have a Spee.ch URL and if we are on the front-end
$speech_url = get_option(LBRY_SETTINGS)[LBRY_SPEECH];
if ($speech_url != '' && !is_admin()) {
ob_start(array($this->parser, 'rewrite'));
}
}
/** /**
* Uploads assets to the speech server * Uploads assets to the speech server
* @param int $post_id The ID of the post to * @param int $post_id The ID of the post to
* @return bool True if successful, false if not or if no Speech URL available * @return bool True if successful, false if not or if no Speech URL available
*/ */
// TODO: set up error reporting
public function upload_media($post_id, $post) public function upload_media($post_id, $post)
{ {
// Only check post_type of Post // Only check post_type of Post
if ('post' !== $post->post_type) { if ('post' !== $post->post_type) {
return; return false;
} }
// error_log('======================== START =====================');
$speech_url = get_option(LBRY_SETTINGS)[LBRY_SPEECH]; $speech_url = get_option(LBRY_SETTINGS)[LBRY_SPEECH];
// Die if we don't have a spee.ch url // Die if we don't have a spee.ch url
@ -67,8 +54,6 @@ class LBRY_Speech
// IDEA: Notify user if post save time will take a while, may be a concern for request timeouts // IDEA: Notify user if post save time will take a while, may be a concern for request timeouts
if ($all_media) { if ($all_media) {
// error_log(print_r($all_media, true));
$requests = array(); $requests = array();
// Build all the Curl Requests // Build all the Curl Requests
@ -102,8 +87,6 @@ class LBRY_Speech
curl_multi_add_handle($mh, $request['request']); curl_multi_add_handle($mh, $request['request']);
} }
// error_log(print_r($requests, true));
// Execute all requests simultaneously // Execute all requests simultaneously
$running = null; $running = null;
do { do {
@ -122,8 +105,6 @@ class LBRY_Speech
$media = $request['media']; $media = $request['media'];
$response_code = curl_getinfo($request['request'], CURLINFO_RESPONSE_CODE); $response_code = curl_getinfo($request['request'], CURLINFO_RESPONSE_CODE);
// error_log(print_r($result, true));
try { try {
// check we got a success code // check we got a success code
if ($response_code != '200') { if ($response_code != '200') {
@ -136,7 +117,6 @@ class LBRY_Speech
// Update image meta // Update image meta
if ($result && $result->success) { if ($result && $result->success) {
// error_log(print_r($result, true));
$meta = wp_get_attachment_metadata($media->id); $meta = wp_get_attachment_metadata($media->id);
if ($media->image_size) { if ($media->image_size) {
$meta['sizes'][$media->image_size]['speech_asset_url'] = $result->data->serveUrl; $meta['sizes'][$media->image_size]['speech_asset_url'] = $result->data->serveUrl;
@ -144,13 +124,12 @@ class LBRY_Speech
$meta['speech_asset_url'] = $result->data->serveUrl; $meta['speech_asset_url'] = $result->data->serveUrl;
} }
wp_update_attachment_metadata($media->id, $meta); wp_update_attachment_metadata($media->id, $meta);
// error_log(print_r($meta, true));
} else { // Something unhandled happened here } else { // Something unhandled happened here
throw new \Exception("Unknown Speech Upload issue for asset"); throw new \Exception("Unknown Speech Upload issue for asset");
} }
} catch (\Exception $e) { } catch (\Exception $e) {
$image_size = $media->image_size ? $media->image_size : 'full'; $image_size = $media->image_size ? $media->image_size : 'full';
error_log('Failed to upload asset with ID ' . $media->id . ' for size ' . $size . ' to supplied speech URL.'); error_log('Failed to upload asset with ID ' . $media->id . ' for size ' . $image_size . ' to supplied speech URL.');
error_log($e->getMessage()); error_log($e->getMessage());
} }
} }
@ -162,49 +141,37 @@ class LBRY_Speech
* @param int $post_id The post to search * @param int $post_id The post to search
* @return array An array of Speech Media Objects * @return array An array of Speech Media Objects
*/ */
protected function find_media($post_id) private function find_media($post_id)
{ {
$all_media = array(); $all_media = array();
// Get content and put into a DOMDocument // Get content and put into a DOMDocument
$content = get_post_field('post_content', $post_id); $content = get_post_field('post_content', $post_id);
if (!$content) { if (!$content) {
return $all_media; return $all_media;
} }
// Find all images $images = $this->parser->scrape_images($content);
preg_match_all('/<img [^>]+>/', $content, $images);
// Only MP4 videos for now // Get all the image ID's
preg_match_all('/\[video.*mp4=".*".*\]/', $content, $videos); $image_ids = array();
// Check to make sure we have results
$images = empty($images[0]) ? array() : $images[0];
$videos = empty($videos[0]) ? array() : $videos[0];
// error_log(print_r($images, true));
// error_log(print_r($videos, true));
// TODO: only create media objects if hasn't been uploaded. IE check meta here
// Throw each image into a media object
foreach ($images as $image) { foreach ($images as $image) {
$attachment_id = null; $new_id = $this->parser->get_attachment_id_from_tag('image', $image);
// Looks for wp image class first, if not, pull id from source if ($new_id) {
if (preg_match('/wp-image-([0-9]+)/i', $image, $class_id)) { $image_ids[] = $new_id;
$attachment_id = absint($class_id[1]); }
// error_log('found with wp-image: ' . $attachment_id); }
} elseif (preg_match('/src="((?:https?:)?\/\/[^"]+)"/', $image, $src) && $this->is_local($src[1])) { // Don't forget the featured image
$attachment_id = $this->rigid_attachment_url_to_postid($src[1]); if ($featured_id = get_post_thumbnail_id($post_id)) {
// error_log('found with url: ' . $attachment_id); $image_ids = array_merge($image_ids, array($featured_id));
} }
if ($attachment_id) { // Throw each image into a media object
foreach ($image_ids as $attachment_id) {
// Create main image media object // Create main image media object
$meta = wp_get_attachment_metadata($attachment_id); $meta = wp_get_attachment_metadata($attachment_id);
// error_log(print_r($meta, true));
// If we don't have meta, get out because none of this will work // If we don't have meta, get out because none of this will work
if (!$meta) { if (!$meta) {
break; break;
@ -225,43 +192,34 @@ class LBRY_Speech
} }
} }
} }
}
// Parse video tags based on wordpress shortcode for local embedds $videos = $this->parser->scrape_videos($content);
$video_ids = array();
foreach ($videos as $video) { foreach ($videos as $video) {
$attachment_id = null; $new_id = $this->parser->get_attachment_id_from_tag('mp4', $video);
if (preg_match('/mp4="((?:https?:)?\/\/[^"]+)"/', $video, $src) && $this->is_local($src[1])) { if ($new_id) {
$attachment_id = $this->rigid_attachment_url_to_postid($src[1]); $video_ids[] = $new_id;
}
if ($attachment_id) { }
// Parse video tags based on wordpress shortcode for local embedds
foreach ($video_ids as $attachment_id) {
$meta = wp_get_attachment_metadata($attachment_id); $meta = wp_get_attachment_metadata($attachment_id);
if (!$this->is_published($meta)) { if (!$this->is_published($meta)) {
$all_media[] = new LBRY_Speech_Media($attachment_id); $all_media[] = new LBRY_Speech_Media($attachment_id);
} }
} }
}
}
return $all_media; return $all_media;
} }
/** /**
* Checks to see if a url is local to this installation * Checks meta to see if a spee.ch url exists
* @param string $url * @param array $meta An array of meta which would possibly contain a speech_asset_url
* @return boolean * @return boolean Whether or not its published to speech
*/ */
private function is_local($url) public function is_published($meta)
{
if (strpos($url, home_url()) !== false) {
return true;
}
}
/**
* Checks array to see if a spee.ch url exists
*/
private function is_published($meta)
{ {
if (key_exists('speech_asset_url', $meta) && $meta['speech_asset_url'] !== '') { if (key_exists('speech_asset_url', $meta) && $meta['speech_asset_url'] !== '') {
return true; return true;
@ -270,34 +228,6 @@ class LBRY_Speech
return false; return false;
} }
/**
* Checks for image crop sizes and filters out query params
* Courtesy of this post: http://bordoni.me/get-attachment-id-by-image-url/
* @param string $url The url of the attachment you want an ID for
* @return int The found post_id
*/
private function rigid_attachment_url_to_postid($url)
{
$scrubbed_url = strtok($url, '?'); // Clean up query params first
$post_id = attachment_url_to_postid($scrubbed_url);
if (! $post_id) {
$dir = wp_upload_dir();
$path = $scrubbed_url;
if (0 === strpos($path, $dir['baseurl'] . '/')) {
$path = substr($path, strlen($dir['baseurl'] . '/'));
}
if (preg_match('/^(.*)(\-\d*x\d*)(\.\w{1,})/i', $path, $matches)) {
$url = $dir['baseurl'] . '/' . $matches[1] . $matches[3];
$post_id = attachment_url_to_postid($url);
}
}
return (int) $post_id;
}
/** /**
* Builds a cURL request to the Speech URL * Builds a cURL request to the Speech URL
* @param string $method The method to call on the Speech API * @param string $method The method to call on the Speech API

View file

@ -7,10 +7,6 @@
class LBRY_Speech_Parser class LBRY_Speech_Parser
{ {
public function __construct()
{
}
/** /**
* [speech_image_srcset description] * [speech_image_srcset description]
* @param [type] $sources [description] * @param [type] $sources [description]
@ -19,15 +15,13 @@ class LBRY_Speech_Parser
* @param [type] $image_meta [description] * @param [type] $image_meta [description]
* @param [type] $attachment_id [description] * @param [type] $attachment_id [description]
*/ */
public function speech_image_srcset($sources, $size_array, $image_src, $image_meta, $attachment_id) public function replace_image_srcset($sources, $size_array, $image_src, $image_meta, $attachment_id)
{ {
$time_start = microtime(true); $time_start = microtime(true);
$new_sources = $sources; $new_sources = $sources;
$sizes = $image_meta['sizes']; $sizes = $image_meta['sizes'];
// error_log(print_r($image_meta, true));
foreach ($sources as $width => $source) { foreach ($sources as $width => $source) {
$speech_url = $this->find_speech_url_by_width($sizes, $width); $speech_url = $this->find_speech_url_by_width($sizes, $width);
@ -43,90 +37,196 @@ class LBRY_Speech_Parser
return $new_sources; return $new_sources;
} }
/**
* [replace_urls_with_speech description]
* @param [type] $content [description]
* @return [type] [description]
*/
public function replace_urls_with_speech($content) public function replace_urls_with_speech($content)
{ {
// // Find all images $new_content = $content;
// preg_match_all('/<img [^>]+>/', $content, $images);
//
// // Check to make sure we have results
// $images = empty($images[0]) ? array() : $images[0];
//
// foreach ($images as $image) {
// $attachment_id = null;
// // Looks for wp image class first, if not, pull id from source
// if (preg_match('/wp-image-([0-9]+)/i', $image, $class_id)) {
// $attachment_id = absint($class_id[1]);
// } elseif (preg_match('/src="((?:https?:)?\/\/[^"]+)"/', $image, $src) && $this->is_local($src[1])) {
// $attachment_id = $this->rigid_attachment_url_to_postid($src[1]);
// }
//
// // // Look for size class, if not continue
// // if (!preg_match('/ size-([\w-_]+) /i', $image, $size_class)) {
// // continue;
// // }
//
// if ($attachment_id) {
// // Create main image media object
// $meta = wp_get_attachment_metadata($attachment_id);
//
// // If we don't have meta, get out because none of this will work
// if (!$meta) {
// break;
// }
//
// if (!$this->is_published($meta)) {
// $all_media[] = new LBRY_Speech_Media($attachment_id);
// }
//
// // COMBAK: find a way to make this more efficient?
// // Create a media object for each image size
// // Get images sizes for this attachment, as not all image sizes implemented
// $image_sizes = wp_get_attachment_metadata($attachment_id)['sizes'];
//
// foreach ($image_sizes as $size => $meta) {
// if (!$this->is_published($meta)) {
// $all_media[] = new LBRY_Speech_Media($attachment_id, array('image_size' => $size));
// }
// }
// }
// }
// error_log(print_r($content, true));
return $content; $assets = array();
// Get Images
$images = $this->scrape_images($content);
error_log(print_r($images, true));
foreach ($images as $image) {
if ($src = $this->get_src_from_image_tag($image)) {
$assets[] = $src;
}
} }
public function replace_attachment_image_src($image, $attachment_id, $size) // Get Videos
{ $videos = $this->scrape_videos($content);
// TODO: Need to come back to this. error_log(print_r($videos, true));
// if (!$image) { foreach ($videos as $video) {
// return; if ($src = $this->get_src_from_video_tag($video)) {
// } $assets[] = $src;
// }
// $new_image = $image; }
// $sizes = $image_meta['sizes'];
//
// // If we have a given size, then use that immediately
// if (is_string($size)) {
//
// if ($size == 'full') {
//
// }
// $new_image[0] = $sizes[$size]['speech_asset_url'];
// return $image;
// }
//
// // Otherwise, we can find it by the url provided
// $new_image[0] = $this->find_speech_url_by_file_url($sizes, $image[0]);
return $image; // Replace with Speech Urls
error_log(print_r($assets, true));
foreach ($assets as $asset) {
$id = $this->rigid_attachment_url_to_postid($asset);
$meta = wp_get_attachment_metadata($id);
if ($meta && LBRY()->speech->is_published($meta) && $speech_url = $this->find_speech_url_by_file_url($meta, $asset)) {
$new_content = str_replace($asset, $speech_url, $new_content);
}
}
return $new_content;
} }
/** /**
* [find_speech_url description] * [replace_attachment_image_src description]
* @param [type] $sizes [description] * @param [type] $image [description]
* @param [type] $width [description] * @param [type] $attachment_id [description]
* @param [type] $size [description]
* @return [type] [description] * @return [type] [description]
*/ */
public function replace_attachment_image_src($image, $attachment_id, $size)
{
if (!$image) {
return $image;
}
// Die if we don't have a speech_asset_url
$image_meta = wp_get_attachment_metadata($attachment_id);
if (!LBRY()->speech->is_published($image_meta)) {
return $image;
}
$new_image = $image;
$sizes = $image_meta['sizes'];
// If we have a given size, then use that immediately
if (is_string($size)) {
switch ($size) {
case 'full':
$new_image[0] = $image_meta['speech_asset_url'];
break;
case 'post-thumbnail':
if (LBRY()->speech->is_published($sizes['thumbnail'])) {
$new_image[0] = $sizes['thumbnail']['speech_asset_url'];
}
break;
default:
if (key_exists($size, $sizes) && LBRY()->speech->is_published($sizes[$size])) {
$new_image[0] = $sizes[$size]['speech_asset_url'];
}
break;
}
return $new_image;
}
// Otherwise, we can find it by the url provided
$speech_url = $this->find_speech_url_by_file_url($image_meta, $image[0]);
if ($speech_url) {
$new_image[0] = $speech_url;
}
return $new_image;
}
/**
* Scrapes all image tags from content
* @param string $content The content to scrape
* @return array Array of image tags, empty if none found
*/
public function scrape_images($content)
{
// Find all images
preg_match_all('/<img [^>]+>/', $content, $images);
// Check to make sure we have results
$images = empty($images[0]) ? array() : $images[0];
return array_unique($images);
}
/**
* Scrapes all video shortcodes from content
* @param string $content The content to scrape
* @return array Array of video tags, empty if none found
*/
public function scrape_videos($content)
{
// Only MP4 videos for now
preg_match_all('/\[video.*mp4=".*".*\]/', $content, $videos);
$videos = empty($videos[0]) ? array() : $videos[0];
return array_unique($videos);
}
/**
* Retrives an attachment ID via an html tag
* @param string $type Can be 'image' or 'mp4'
* @param string $tag The Tag / shortcode to scrub for an ID
* @return int|bool An ID if one is found, false otherwise
*/
public function get_attachment_id_from_tag($type, $tag)
{
$attachment_id = false;
switch ($type) {
case 'image':
// Looks for wp image class first, if not, pull id from source
if (preg_match('/wp-image-([0-9]+)/i', $tag, $class_id)) {
$attachment_id = absint($class_id[1]);
} elseif ($src = $this->get_src_from_image_tag($tag)) {
$attachment_id = $this->rigid_attachment_url_to_postid($src);
}
break;
case 'mp4':
if ($src = $this->get_src_from_video_tag($tag)) {
$attachment_id = $this->rigid_attachment_url_to_postid($src);
}
break;
}
return $attachment_id;
}
/**
* Pulls src from image tag
* @param string $image HTML Image tag
* @return string The source if found and is a local source, otherwise false
*/
private function get_src_from_image_tag($image)
{
if (preg_match('/src="((?:https?:)?\/\/[^"]+)"/', $image, $src) && $src[1]) {
if ($this->is_local($src[1])) {
return $src[1];
}
}
return false;
}
/**
* Pulls src from video tag
* @param string $video The video shortcode
* @return string The source if found and is a local source, otherwise false
*/
private function get_src_from_video_tag($video)
{
if (preg_match('/mp4="((?:https?:)?\/\/[^"]+)"/', $video, $src) && $src[1]) {
if ($this->is_local($src[1])) {
return $src[1];
}
}
return false;
}
/**
* Retrieves a speech_asset_url based sizes meta provided
* @param array $sizes Image sizes meta array
* @param string $width The width to look for
* @return string An asset url if found, false if not
*/
private function find_speech_url_by_width($sizes, $width) private function find_speech_url_by_width($sizes, $width)
{ {
foreach ($sizes as $key => $size) { foreach ($sizes as $key => $size) {
@ -138,14 +238,78 @@ class LBRY_Speech_Parser
return false; return false;
} }
private function find_speech_url_by_file_name($sizes, $url) /**
* Retrieves a speech_asset_url based on a passed url
* @param array $meta The attachment meta data for the asset
* @param string $url The URL of the asset being exchanged
* @return string The URL of the Speech Asset
*/
private function find_speech_url_by_file_url($meta, $url)
{ {
for // See if this looks like video meta
if (!key_exists('file', $meta) && key_exists('mime_type', $meta) && $meta['mime_type'] == 'video/mp4') {
return $meta['speech_asset_url'];
} }
private function microtime_float() $pathinfo = pathinfo($url);
$basename = $pathinfo['basename'];
// Check main file or if $meta is just a url (video) first
if (key_exists('file', $meta) && $basename == wp_basename($meta['file'])) {
return $meta['speech_asset_url'];
}
// Check to see if we have a meta option here
if (key_exists('sizes', $meta)) {
// Then check sizes
foreach ($meta['sizes'] as $size => $meta) {
if ($basename == $meta['file'] && key_exists('speech_asset_url', $meta)) {
return $meta['speech_asset_url'];
}
}
}
// Couldn't make a match
return false;
}
/**
* Checks to see if a url is local to this installation
* @param string $url
* @return boolean
*/
private function is_local($url)
{ {
list($usec, $sec) = explode(" ", microtime()); if (strpos($url, home_url()) !== false) {
return ((float)$usec + (float)$sec); return true;
}
}
/**
* Checks for image crop sizes and filters out query params
* Courtesy of this post: http://bordoni.me/get-attachment-id-by-image-url/
* @param string $url The url of the attachment you want an ID for
* @return int The found post_id
*/
private function rigid_attachment_url_to_postid($url)
{
$scrubbed_url = strtok($url, '?'); // Clean up query params first
$post_id = attachment_url_to_postid($scrubbed_url);
if (! $post_id) {
$dir = wp_upload_dir();
$path = $scrubbed_url;
if (0 === strpos($path, $dir['baseurl'] . '/')) {
$path = substr($path, strlen($dir['baseurl'] . '/'));
}
if (preg_match('/^(.*)(\-\d*x\d*)(\.\w{1,})/i', $path, $matches)) {
$url = $dir['baseurl'] . '/' . $matches[1] . $matches[3];
$post_id = attachment_url_to_postid($url);
}
}
return (int) $post_id;
} }
} }