more clever check for only new toots

themes/hugo-mastodon-comments/static/comments/getcomments.php

@@ -5,18 +5,22 @@ $instance = $config['mastodon-instance'];
 $uid = $config['user-id'];
 $searchurl = $config['search-url'];
 $search = isset($_GET['search']) ? $_GET['search'] : '';
+$debug_on = true;
 /* cache files */
 $dbt = "cache-toots.json";
 
 /* Exit if search empty */
 if (empty($search)) {
-  out("No proper search given");
+  debug("No proper search given");
   die();
 }
 
 /* MISC FUNCTIONS */
-function out($data) {
+function debug($data) {
+  global $debug_on;
+  if ($debug_on === true) {
     error_log("[getcomments.php] " . print_r($data, TRUE));
+  }
 }
 
 /* CACHE FUNCTIONS */
@@ -28,52 +32,28 @@ function write_db($db, $data) {
 }
 /* access data from file */
 function read_db($db, &$data) {
+  // if DB does not exist, create it with empty array
   if (! file_exists($db)) {
     touch($db);
+    write_db($db, array());
   }
   $file = file_get_contents($db, true);
   $data = json_decode($file, true);
 }
 
 /* TOOT FUNCTIONS */
-function collectToots($instance, $uid, $min_id) {
+function collectToots($instance, $uid, $min_id, $searchurl) {
   $raw = file_get_contents("$instance/api/v1/accounts/$uid/statuses?exclude_reblogs=true&exclude_replies=true&limit=50&min_id=$min_id");
-  $json = json_decode($raw, true);
+  $json_complete = json_decode($raw, true);
+  $json = array();
+  foreach ($json_complete as $toot) {
+    $json[] = array('id' => $toot['id'], 'date' => $toot['created_at'] ,'url' => analyzeToot($instance, $toot['id'], $searchurl));
+  }
   return($json);
 }
-
+/* Find out if a toot contains the searched URL */
-
+function analyzeToot($instance, $id, $searchurl) {
-/* Collect all the toots */
+  debug("Searching for $searchurl in $id");
-$toots = array();
-/* get id of latest cached toot, and set as $min_id */
-read_db($dbt, $toots);
-if (!empty($toots['0']['id'])) {
-  $min_id_cached = $toots['0']['id'];
-  $min_id = $min_id_cached;
-} else {
-  /* if cached toots do not exist, start from oldest toot */
-  $min_id = "0";
-  $min_id_cached = "0";
-}
-
-
-/* test whether there are new toots available */
-$min_id_new = "1";
-while ($min_id_new !== $min_id) {
-  $min_id_new = $min_id;
-
-  $toots = array_merge(collectToots($instance, $uid, $min_id), $toots);
-  $min_id = $toots['0']['id'];
-}
-
-/* if newer toot has been found, find new URLs */
-// TODO: only look up newly found toots
-if ($min_id !== $min_id_cached) {
-  out("New toots found");
-  $ids = array_column($toots, 'id');
-
-  /* Find out if a toot contains the searched URL */
-  function analyzeToot($instance, $id, $searchurl) {
   $raw = file_get_contents("$instance/api/v1/statuses/$id");
   $json = json_decode($raw, true);
 
@@ -84,23 +64,45 @@ if ($min_id !== $min_id_cached) {
   } else {
     return("");
   }
-  }
+}
 
-  $toots = array();
-  foreach ($ids as $id) {
-    $toots[] = array('id' => $id, 'url' => analyzeToot($instance, $id, $searchurl));
-  }
 
-  write_db($dbt, $toots);
+/* Collect all the toots */
+/* get id of latest cached toot, and set as $min_id */
+read_db($dbt, $toots);
+$toots_cached = $toots;
+if (!empty($toots['0']['id'])) {
+  $min_id_cached = $toots['0']['id'];
+  $min_id = $min_id_cached;
 } else {
-  out("No new toots found");
+  /* if cached toots do not exist, start from oldest toot */
+  $min_id = "0";
+  $min_id_cached = "0";
+}
+
+/* test whether there are new toots available */
+// Search for toots older than the cached latest toot ID ($min_id)
+$uptodate = false;
+while ($uptodate === false) {
+  $toots = array_merge(collectToots($instance, $uid, $min_id, $searchurl), $toots);
+  $min_id_new = $toots['0']['id']; // the latest ID of the recent search
+  if ($min_id_new === $min_id) {
+    // min_id is the latest, let's write the new DB and end this loop
+    $uptodate = true;
+    debug("Rewrite cache DB.");
+    write_db($dbt, $toots);
+  } else {
+    // next round looks for toots newer than the newly found ID
+    debug("Newer toots than in cache found. Starting another search for new toots");
+    $min_id = $min_id_new;
+  }
 }
 
 /* check if URL from $search exists in $toots */
 // if multiple exist, take the oldest one (highest array position)
 $id = array_keys(array_column($toots, 'url'), strtolower($search));
 if (empty($id)) {
-  out("Blog URL \"$search\" has not been found");
+  debug("Blog URL \"$search\" has not been found");
   die();
 }
 $id = $toots[end($id)]['id'];
@@ -143,7 +145,7 @@ function filterStats($stats) {
 }
 
 function tootStats($instance, $id, &$result) {
-  out("Checking ID $id");
+  debug("Checking ID $id");
   $raw = file_get_contents("$instance/api/v1/statuses/$id");
   $json = json_decode($raw, true);
   $newStats = filterStats($json);