Implement podcast index search

conf/podsumer.conf

@@ -33,3 +33,18 @@ media_dir = /opt/media
 playback_interval = 5
 playback_rewind = 5
 
+# Enable/disable SponsorBlock integration auto-skip  timestamps
+# Set to `true` to enable. Default is off.
+sponsorblock_enabled = false
+
+# PodcastIndex API key and secret
+#
+# These are used to search for podcasts and episodes using the PodcastIndex API.
+#
+# You can get your own API key and secret by signing up at https://podcastindex.org/
+#
+# Once you have your API key and secret, you can add them here.
+podcastindex_key = ""
+podcastindex_secret = ""
+
+

conf/test.conf

@@ -14,3 +14,6 @@ ssl = false
 media_dir = /opt/media
 playback_interval = 5
 playback_rewind = 5
+
+podcastindex_key = ""
+podcastindex_secret = ""

src/Brickner/Podsumer/FSState.php

@@ -96,46 +96,70 @@ protected function escapeFilename(string $filename): string
     public function deleteFeed(int $feed_id)
     {
         $feed = $this->getFeed($feed_id);
-        $file_id = $feed['image'];
 
-        $file = $this->getFileById($file_id);
-
-        if ($file['storage_mode'] == 'DISK' && file_exists($file['filename'])) {
-            unlink($file['filename']);
+        # If no feed record is found, or if the feed record lacks a valid
+        # name, run the parent clean-up logic and exit early. Attempting to
+        # continue without a valid feed name can lead to resolving the media
+        # root directory as the feed directory which is unsafe.
+        if (empty($feed) || empty(trim($feed['name'] ?? ''))) {
+            parent::deleteFeed($feed_id);
+            return;
         }
 
-        $items = $this->getFeedItems($feed_id);
-        foreach ($items as $item) {
-            $file_id = $item['image'];
-
-            if (!empty($file_id)) {
+        # Capture all related files before we alter the database so we can
+        # safely remove them from disk afterwards.
+        $files_to_delete = [];
 
-                $file = $this->getFileById($file_id);
+        $image_file_id = $feed['image'] ?? null;
+        if ($image_file_id) {
+            $files_to_delete[] = $this->getFileById($image_file_id);
+        }
 
-                if ($file['storage_mode'] == 'DISK' && file_exists($file['filename'])) {
-                    unlink($file['filename']);
+        # Collect images / audio from each item before DB deletion
+        foreach ($this->getFeedItems($feed_id) as $item) {
+            foreach (['image', 'audio_file'] as $col) {
+                $fid = $item[$col] ?? null;
+                if ($fid) {
+                    $files_to_delete[] = $this->getFileById($fid);
                 }
             }
+        }
 
-            $file_id = $item['audio_file'];
-
-            if (!empty($file_id)) { # The audio for an item may not be downloaded.
-
-                $file = $this->getFileById($file_id);
+        # Delete from the database first (cascades will clean up related rows)
+        parent::deleteFeed($feed_id);
 
-                if ($file['storage_mode'] == 'DISK' && file_exists($file['filename'])) {
-                    unlink($file['filename']);
+        # Remove the on-disk files we captured earlier
+        foreach ($files_to_delete as $file) {
+            if (!empty($file) && ($file['storage_mode'] ?? null) === 'DISK') {
+                $filename = $file['filename'] ?? null;
+                if (!empty($filename) && file_exists($filename)) {
+                    @unlink($filename);
                 }
             }
         }
 
-        # Delete feed dir.
-        $feed_dir = $this->getFeedDir($feed['name']);
-        if (file_exists($feed_dir)) {
-            rmdir($feed_dir);
+        # Finally, try to remove the (now empty) feed directory
+        $feed_name = trim($feed['name']);
+
+        if ($feed_name !== '') {
+            $feed_dir = $this->getFeedDir($feed_name);
+            $media_dir = rtrim($this->getMediaDir(), DIRECTORY_SEPARATOR);
+
+            # Ensure the directory we are about to touch is not the media root
+            if ($feed_dir !== $media_dir && file_exists($feed_dir) && is_dir($feed_dir)) {
+                $dir_contents = @scandir($feed_dir);
+
+                # scandir() returns false on failure. Guard against that so we
+                # do not pass a boolean to array_diff(), which would raise a
+                # TypeError.
+                if (false !== $dir_contents) {
+                    $files_in_dir = array_diff($dir_contents, ['.', '..']);
+                    if (empty($files_in_dir)) {
+                        @rmdir($feed_dir);
+                    }
+                }
+            }
         }
-
-        parent::deleteFeed($feed_id);
     }
 
     public function deleteItemMedia(int $item_id)
@@ -150,8 +174,12 @@ public function deleteItemMedia(int $item_id)
 
         $file = $this->getFileById($file_id);
 
-        if ($file['storage_mode'] == 'DISK' && file_exists($file['filename'])) {
-            unlink($file['filename']);
+        if (!empty($file) && ($file['storage_mode'] ?? null) === 'DISK') {
+            $filename = $file['filename'] ?? null;
+
+            if (!empty($filename) && file_exists($filename)) {
+                @unlink($filename);
+            }
         }
 
         parent::deleteItemMedia($item_id);

src/Brickner/Podsumer/PodcastIndex.php

@@ -0,0 +1,49 @@
+<?php declare(strict_types = 1);
+
+namespace Brickner\Podsumer;
+
+class PodcastIndex
+{
+    static public function search(string $query, int $max, string $key, string $secret, ?int $time = null): array
+    {
+        if (empty($key) || empty($secret) || empty($query)) {
+            return [];
+        }
+
+        $time = $time ?? time();
+        $hash = sha1($key . $secret . $time);
+
+        $headers = [
+            'User-Agent: Podsumer',
+            'X-Auth-Date: ' . $time,
+            'X-Auth-Key: ' . $key,
+            'Authorization: ' . $hash,
+        ];
+
+        $url = 'https://api.podcastindex.org/api/1.0/search/byterm?q=' . urlencode($query) . '&max=' . $max;
+
+        $curl = curl_init();
+        curl_setopt($curl, \CURLOPT_URL, $url);
+        curl_setopt($curl, \CURLOPT_RETURNTRANSFER, true);
+        curl_setopt($curl, \CURLOPT_HTTPHEADER, $headers);
+        curl_setopt($curl, \CURLOPT_FOLLOWLOCATION, true);
+        curl_setopt($curl, \CURLOPT_CONNECTTIMEOUT, 30);
+        curl_setopt($curl, \CURLOPT_TIMEOUT, 60);
+
+        $result = curl_exec($curl);
+        // Always close the handle to avoid leaking resources.
+        curl_close($curl);
+
+        if (false === $result) {
+            return [];
+        }
+
+        $data = json_decode($result, true);
+        if (!is_array($data) || !array_key_exists('feeds', $data)) {
+            return [];
+        }
+
+        return $data['feeds'];
+    }
+}
+

src/Brickner/Podsumer/SponsorBlock.php

@@ -0,0 +1,54 @@
+<?php declare(strict_types = 1);
+
+namespace Brickner\Podsumer;
+
+/**
+ * Very small wrapper around the public SponsorBlock API
+ *   https://wiki.sponsor.ajay.app/
+ *
+ * We currently only use it to obtain *skip segments* for a YouTube video so we
+ * can automatically skip sponsor reads during podcast playback.
+ */
+class SponsorBlock
+{
+    /**
+     * @param string   $videoId      YouTube video ID.
+     * @param string[] $categories   Array of category slugs we care about.
+     *                               Defaults to the standard sponsor + selfpromo + interaction set.
+     *
+     * @return array<int,array{segment:array{0:float,1:float},category:string}> | []
+     */
+    public static function getSegments(string $videoId, array $categories = ['sponsor', 'selfpromo', 'interaction', 'intro', 'outro']): array
+    {
+        $categories = ['sponsor', 'selfpromo', 'interaction', 'intro', 'outro'];
+        if (empty($videoId)) {
+            return [];
+        }
+
+        $url = 'https://sponsor.ajay.app/api/skipSegments?videoID=' . urlencode($videoId) . '&categories=' . rawurlencode(json_encode($categories));
+
+        $curl = curl_init();
+        curl_setopt($curl, \CURLOPT_URL, $url);
+        curl_setopt($curl, \CURLOPT_RETURNTRANSFER, true);
+        curl_setopt($curl, \CURLOPT_HTTPHEADER, [
+            'User-Agent: Podsumer',
+            'Accept: application/json',
+        ]);
+        curl_setopt($curl, \CURLOPT_CONNECTTIMEOUT, 10);
+        curl_setopt($curl, \CURLOPT_TIMEOUT, 20);
+
+        $response = curl_exec($curl);
+        curl_close($curl);
+
+        if (false === $response) {
+            return [];
+        }
+
+        $decoded = json_decode($response, true);
+        if (!is_array($decoded)) {
+            return [];
+        }
+
+        return $decoded;
+    }
+}