From 3c60274dc7579e4dee26f2de51764ce0d818747d Mon Sep 17 00:00:00 2001
From: "Peter Droogmans (attiks)" <peter@attiks.com>
Date: Wed, 24 Jul 2024 14:20:39 +0200
Subject: [PATCH] Calculate embedding

---
 .../src/Services/OchaAiTagTagger.php          | 30 +++++++++----------
 1 file changed, 15 insertions(+), 15 deletions(-)

diff --git a/modules/ocha_ai_tag/src/Services/OchaAiTagTagger.php b/modules/ocha_ai_tag/src/Services/OchaAiTagTagger.php
index 4f2824c..9e5c480 100644
--- a/modules/ocha_ai_tag/src/Services/OchaAiTagTagger.php
+++ b/modules/ocha_ai_tag/src/Services/OchaAiTagTagger.php
@@ -299,6 +299,10 @@ protected function getSimilarTerms(array $embeddings, bool $average_embeddings =
 
     // Sort by similarity score descending.
     foreach ($types as $type) {
+      if (!isset($results[$type])) {
+        continue;
+      }
+
       foreach ($results[$type] as $vocabulary => $terms) {
         if (!empty($terms)) {
           arsort($results[$type][$vocabulary]);
@@ -431,23 +435,19 @@ public function embedDocument(int $id): bool {
    */
   public function getSimilarDocuments($id, string $text = '') {
     $query_embedding = [];
-    $doc = $this->getVectorStorePlugin()->getDocument('vector_jobs', $id);
 
-    if (empty($doc)) {
-      // Doc doesn't exist in vector store.
-      $data = $this->getSourcePlugin()->getDocument('jobs', $id);
-      if (empty($data)) {
-        $query_embedding = $this->getEmbeddings($text);
-      }
-      else {
-        $job = reset($data['jobs']);
-        $job = $this->processDocument($job);
-        $query_embedding = $job['embedding'] ?? [];
-      }
+    // Will fail for unpublished documents.
+    $doc = NULL;
+    try {
+      $doc = $this->getVectorStorePlugin()->getDocument('vector_jobs', $id);
     }
-    else {
-      $doc = $doc['_source'];
-      $query_embedding = $doc['embedding'] ?? [];
+    catch (\Throwable $e) {
+      $doc = NULL;
+    }
+
+    if (empty($doc)) {
+      $query_embedding = $this->getEmbeddings($text);
+      $query_embedding = reset($query_embedding);
     }
 
     if (!is_array($query_embedding) || empty($query_embedding)) {