make the search to follow the the nested objects #67

Te-cho · Mar 20, 2017 · 381734f · 381734f
1 parent 0d50334
commit 381734f
Show file tree

Hide file tree

Showing 6 changed files with 47 additions and 70 deletions.
diff --git a/circle.yml b/circle.yml
@@ -8,4 +8,4 @@ deployment:
     branch: master
     commands:
       - rsync -avz /home/ubuntu/youtubeSearch/ [email protected]:/youtubeSearch
-      - ssh -i ~/.ssh/id_rsa [email protected] 'make -C /youtubeSearch/ upProduction; chmod -R 777 /youtubeSearch/y2bsearch/storage/; chmod -R 777 /youtubeSearch/y2bsearch/public/counter.txt; mv /youtubeSearch/y2bsearch/.env.prod /youtubeSearch/y2bsearch/.env'
+      - ssh -i ~/.ssh/id_rsa [email protected] 'make -C /youtubeSearch/ upProduction; chmod -R 777 /youtubeSearch/y2bsearch/storage/; chmod -R 777 /youtubeSearch/y2bsearch/public/counter.txt;'
diff --git a/y2bsearch/app/Http/Controllers/Controller.php b/y2bsearch/app/Http/Controllers/Controller.php
@@ -30,21 +30,21 @@ public function show(Request $request)
 //        (new GoogleAnalyticsService)->printResultss();
         $search_keywords = strtolower($this->searchKeywords);
         if (empty($search_keywords)) {
-            return $this->mainPage($search_keywords);
+            return $this->mainPage();
         } else {
             return $this->searchPage($search_keywords);
         }
 
     }
 
-    public function mainPage($search_keywords)
+    public function mainPage()
     {
         $searchProcessor = new SearchProcessor();
         $client = EsService::generateESConnection();
         $params = $searchProcessor->generateTopSearchQuery();
         $response = $client->search($params);
         $subtitlesService = new SubtitleAnalyzer();
-        $response = $subtitlesService->analyzeAndProcess($response, $search_keywords);
+        $response = $subtitlesService->analyzeAndProcess($response);
         $data['videos'] = $response['hits']['hits'];
         $data['mainPage'] = true;
 
@@ -58,7 +58,7 @@ public function searchPage($search_keywords)
         $params = $searchProcessor->generateSearchQuery($search_keywords);
         $response = $client->search($params);
         $subtitlesService = new SubtitleAnalyzer();
-        $response = $subtitlesService->analyzeAndProcess($response, $search_keywords);
+        $response = $subtitlesService->analyzeAndProcess($response);
         $data['videos'] = $response['hits']['hits'];
         $data['searchKeywords'] = $this->searchKeywords;
 

diff --git a/y2bsearch/app/Src/SearchService/SearchProcessor.php b/y2bsearch/app/Src/SearchService/SearchProcessor.php
@@ -48,30 +48,45 @@ public function generateSearchQuery($searchKeywords)
             'body' => [
                 'size' => 9,
                 'query' => [
-                    'query_string' => [
-                        'default_field' => "subtitles",
-                        'query' => "",
+                    'bool' => [
+                        'must' => [
+                            "nested" => [
+                                'path' => "subtitles",
+                                'inner_hits' => [
+                                    'highlight' => [
+                                        'pre_tags' => ['<b>'],
+                                        'post_tags' => ['</b>'],
+                                        "order" => "score",
+                                        'fields' => [
+                                            "subtitles.sentence" => [
+                                                "fragment_size" => 300,
+                                                "number_of_fragments" => 100,
+                                            ],
+                                        ],
+                                    ],
+                                ],
+                                'query' => [
+                                    'bool' => [
+                                        'must' => [],
+                                    ],
+                                ],
+                            ],
+                        ],
                     ],
                 ],
                 'sort' => [
                     '_score' => ['order' => 'desc'],
                 ],
-                'highlight' => [
-                    'pre_tags' => ['<b>'],
-                    'post_tags' => ['</b>'],
-                    "order" => "score",
-                    'fields' => [
-                        "subtitles" => [
-                            "fragment_size" => 300,
-                            "number_of_fragments" => 100,
-                        ],
-                    ],
-                ],
             ],
         ];
-        $searchKeywords = '*' . str_replace(' ', ' AND ', $searchKeywords);
-        $searchKeywords .= '*';
-        $params['body']['query']['query_string']['query'] = $searchKeywords;
+        $searchKeywords = explode(' ', $searchKeywords);
+        foreach ($searchKeywords as $keyword) {
+            $params['body']['query']['bool']['must']['nested']['query']['bool']['must'][] = [
+                'term' => [
+                    'subtitles.sentence' => $keyword,
+                ],
+            ];
+        }
 
         return $params;
     }

diff --git a/y2bsearch/app/Src/SubtitleAnalyzer/SubtitleAnalyzer.php b/y2bsearch/app/Src/SubtitleAnalyzer/SubtitleAnalyzer.php
@@ -7,52 +7,10 @@
 
 class SubtitleAnalyzer
 {
-    /**
-     * @param array $videos
-     * @param $searchTerm
-     *
-     * @return array
-     */
-    public function getTiming(array $videos, $searchTerm)
-    {
-        $terms = explode(' ', $searchTerm);
-        $termsCount = count($terms);
-        foreach ($videos as $videoKey => $value) {
-            $index = 0;
-            $subtitles = $value['_source']['_subtitles'];
-            foreach ($subtitles['sentence'] as $key => $subtitle) {
-                $matchedTerms = 0;
-                foreach ($terms as $term) {
-                    if (str_contains($subtitle, $term)) {
-                        $matchedTerms++;
-                    } else {
-                        break;
-                    }
-                }
-                if ($matchedTerms === $termsCount) {
-                    $index = $key;
-                    break;
-                }
-            }
-            $start = strtotime($subtitles['start'][$index]) - strtotime('TODAY');
-//            $end = strtotime($subtitles['end'][$index]) - strtotime('TODAY');
-            $sentence = $subtitles['sentence'][$index];
-            if($start === 0){
-                unset($videos[$videoKey]);
-                continue;
-            }
-            $videos[$videoKey]['_source']['start'] = $start;
-            $videos[$videoKey]['_source']['sentence'] = $sentence;
-            $videos[$videoKey]['_source']['video_url'] = $videos[$videoKey]['_source']['video_url'] . '&t=' . $start;
-        }
-
-        return $videos;
-    }
-
-    public function analyzeAndProcess($response, $searchKeywords)
+    public function analyzeAndProcess($response)
     {
         $responseProcessed = $response['hits']['hits'];
-        $responseProcessed = $this->getTiming($responseProcessed, $searchKeywords);
+//        $responseProcessed = $this->getTiming($responseProcessed);
         $responseProcessed = $this->cleanHighlights($responseProcessed);
         $response['hits']['hits'] = $responseProcessed;
 
@@ -70,9 +28,10 @@ private function cleanHighlights($videos)
     {
         $sanitizer = new Sanitizer();
         foreach ($videos as $index => $video) {
-            $highlight = $video['highlight']['subtitles'][0];
+            $highlight = $video['inner_hits']['subtitles']['hits']['hits'][0]['highlight']['subtitles.sentence'];
             $highlight = $sanitizer->cleanFromXMLTags($highlight);
-            $videos[$index]['highlight']['subtitles'][0] = $highlight;
+            $videos[$index]['inner_hits']['subtitles']['hits']['hits'][0]['highlight']['subtitles.sentence'] =
+                $highlight;
         }
 
         return $videos;

diff --git a/y2bsearch/resources/views/components/videoCard.blade.php b/y2bsearch/resources/views/components/videoCard.blade.php
@@ -4,7 +4,7 @@
             <div class="content title">
                 {{ $video['video_title'] }}
             </div>
-            <a target="_blank" data-embedhref="https://www.youtube.com/embed/{{ $video['video_hash_id'] }}?start={{ $video['start'] }}&cc_load_policy=1&autoplay=1"
+            <a target="_blank" data-embedhref="https://www.youtube.com/embed/{{ $video['video_hash_id'] }}?start={{ strtotime($start ) - strtotime('TODAY')}}&cc_load_policy=1&autoplay=1"
                data-id="{{ $video['video_hash_id'] }}" class="image .valencia .lark">
                 <img src="{{$video['image_medium']}}">
             </a>

diff --git a/y2bsearch/resources/views/subviews/videos.blade.php b/y2bsearch/resources/views/subviews/videos.blade.php
@@ -1,5 +1,8 @@
 <div id="videos" class="ui link cards three column stackable grid container">
     @foreach ($videos as $video)
-        @include('components.videoCard', ['video' => $video['_source'], 'highlighted_search' => $video['highlight']['subtitles'][0]])
+        @include('components.videoCard', ['video' => $video['_source'],
+        'highlighted_search' => $video['inner_hits']['subtitles']['hits']['hits'][0]['highlight']['subtitles.sentence'][0],
+        'start' => $video['inner_hits']['subtitles']['hits']['hits'][0]['_source']['start']
+        ])
     @endforeach
 </div>