Merge pull request #3416 from alchemy-fr/PHRAS-3001_es-geopoint-sources_4.1

PHRAS-3001 #comment  merge es geopoint sources 4.1

lib/Alchemy/Phrasea/Core/Provider/SearchEngineServiceProvider.php

@@ -43,6 +43,7 @@ use Silex\ServiceProviderInterface;
 use Symfony\Component\EventDispatcher\EventDispatcherInterface;
 use Symfony\Component\HttpKernel\KernelEvents;
 
+
 class SearchEngineServiceProvider implements ServiceProviderInterface
 {
     public function register(Application $app)
@@ -145,6 +146,7 @@ class SearchEngineServiceProvider implements ServiceProviderInterface
 
         $app['elasticsearch.indexer.databox_fetcher_factory'] = $app->share(function ($app) {
             return new DataboxFetcherFactory(
+                $app['conf'],
                 $app['elasticsearch.record_helper'],
                 $app['elasticsearch.options'],
                 $app,

lib/Alchemy/Phrasea/SearchEngine/Elastic/DataboxFetcherFactory.php

@@ -2,6 +2,7 @@
 
 namespace Alchemy\Phrasea\SearchEngine\Elastic;
 
+use Alchemy\Phrasea\Core\Configuration\PropertyAccess;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\Record\Delegate\FetcherDelegateInterface;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\Record\Fetcher;
 use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\Record\Hydrator\CoreHydrator;
@@ -13,8 +14,14 @@ use Alchemy\Phrasea\SearchEngine\Elastic\Indexer\Record\Hydrator\TitleHydrator;
 use Alchemy\Phrasea\SearchEngine\Elastic\Structure\Structure;
 use Alchemy\Phrasea\SearchEngine\Elastic\Thesaurus\CandidateTerms;
 
+
 class DataboxFetcherFactory
 {
+    /**
+     * @var PropertyAccess       phraseanet configuration
+     */
+    private $conf;
+
     /**
      * @var \ArrayAccess
      */
@@ -39,14 +46,16 @@ class DataboxFetcherFactory
     private $options;
 
     /**
+     * @param PropertyAccess $conf
      * @param RecordHelper $recordHelper
      * @param ElasticsearchOptions $options
      * @param \ArrayAccess $container
      * @param string $structureKey
      * @param string $thesaurusKey
      */
-    public function __construct(RecordHelper $recordHelper, ElasticsearchOptions $options, \ArrayAccess $container, $structureKey, $thesaurusKey)
+    public function __construct(PropertyAccess $conf, RecordHelper $recordHelper, ElasticsearchOptions $options, \ArrayAccess $container, $structureKey, $thesaurusKey)
     {
+        $this->conf         = $conf;
         $this->recordHelper = $recordHelper;
         $this->options      = $options;
         $this->container    = $container;
@@ -70,7 +79,7 @@ class DataboxFetcherFactory
             [
                 new CoreHydrator($databox->get_sbas_id(), $databox->get_viewname(), $this->recordHelper),
                 new TitleHydrator($connection, $this->recordHelper),
-                new MetadataHydrator($connection, $this->getStructure(), $this->recordHelper),
+                new MetadataHydrator($this->conf, $connection, $this->getStructure(), $this->recordHelper),
                 new FlagHydrator($this->getStructure(), $databox),
                 new ThesaurusHydrator($this->getStructure(), $this->getThesaurus(), $candidateTerms),
                 new SubDefinitionHydrator($connection)

lib/Alchemy/Phrasea/SearchEngine/Elastic/Indexer/Record/Hydrator/GpsPosition.php

@@ -15,6 +15,7 @@ use Assert\Assertion;
 
 class GpsPosition
 {
+    const FULL_GEO_NOTATION = 'FullNotation';
     const LONGITUDE_TAG_NAME = 'Longitude';
     const LONGITUDE_REF_TAG_NAME = 'LongitudeRef';
     const LONGITUDE_REF_WEST = 'W';
@@ -29,6 +30,16 @@ class GpsPosition
     private $latitude;
     private $latitude_ref;
 
+    public function __construct()
+    {
+        $this->clear();
+    }
+
+    public function clear()
+    {
+        $this->longitude = $this->longitude_ref = $this->latitude = $this->latitude_ref = null;
+    }
+
     public function set($tag_name, $value)
     {
         switch ($tag_name) {
@@ -64,6 +75,53 @@ class GpsPosition
                 $this->latitude_ref = $normalized;
                 break;
 
+            case self::FULL_GEO_NOTATION:
+                $re = '/(-?\d+(?:\.\d+)?°?)\s*(\d+(?:\.\d+)?\')?\s*(\d+(?:\.\d+)?")?\s*(N|S|E|W)?/um';
+                $normalized = trim(strtoupper($value));
+                $matches = null;
+                preg_match_all($re, $normalized, $matches, PREG_SET_ORDER, 0);
+                if(count($matches) === 2) {     // we need lat and lon
+                    $lat = $lon = null;
+                    foreach ($matches as $imatch => $match) {
+                        if(count($match) != 5) {
+                            continue;
+                        }
+                        $v = 0.0;
+                        for($part=1, $div=1.0; $part<=3; $part++, $div*=60.0) {
+                            $v += floatval($match[$part]) / $div;
+                        }
+                        switch($match[4]) {     // N S E W
+                            case 'N':
+                                $lat = $v;
+                                break;
+                            case 'S':
+                                $lat = -$v;
+                                break;
+                            case 'E':
+                                $lon = $v;
+                                break;
+                            case 'W':
+                                $lon = -$v;
+                                break;
+                            case '':        // no ref -> lat lon (first=lat, second=lon)
+                                if($imatch === 0) {
+                                    $lat = $v;
+                                }
+                                else {
+                                    $lon = $v;
+                                }
+                                break;
+                            default:
+                                throw new \InvalidArgumentException(sprintf('Unsupported reference "%s", should be N|S|E|W.', $match[4]));
+                        }
+                    }
+                    if($lat !== null && $lon != null) {
+                        $this->set(self::LATITUDE_TAG_NAME, $lat);
+                        $this->set(self::LONGITUDE_TAG_NAME, $lon);
+                    }
+                }
+                break;
+
             default:
                 throw new \InvalidArgumentException(sprintf('Unsupported tag name "%s".', $tag_name));
         }
@@ -95,19 +153,11 @@ class GpsPosition
 
     public function getCompositeLongitude()
     {
-        if ($this->longitude === null) {
-            return null;
-        }
-
         return $this->longitude ;
     }
 
     public function getCompositeLatitude()
     {
-        if ($this->latitude === null) {
-            return null;
-        }
-
         return $this->latitude;
     }
 

lib/Alchemy/Phrasea/SearchEngine/Elastic/Indexer/Record/Hydrator/MetadataHydrator.php

@@ -11,6 +11,7 @@
 
 namespace Alchemy\Phrasea\SearchEngine\Elastic\Indexer\Record\Hydrator;
 
+use Alchemy\Phrasea\Core\Configuration\PropertyAccess;
 use Alchemy\Phrasea\SearchEngine\Elastic\Exception\Exception;
 use Alchemy\Phrasea\SearchEngine\Elastic\FieldMapping;
 use Alchemy\Phrasea\SearchEngine\Elastic\Mapping;
@@ -24,28 +25,48 @@ use InvalidArgumentException;
 
 class MetadataHydrator implements HydratorInterface
 {
+    private $conf;
     private $connection;
     private $structure;
     private $helper;
 
-    private $gps_position_buffer = [];
+    private $position_fields_mapping;     // get from conf
 
-    public function __construct(DriverConnection $connection, Structure $structure, RecordHelper $helper)
+    private $caption_gps_position;
+    private $exif_gps_position;
+
+    public function __construct(PropertyAccess $conf, DriverConnection $connection, Structure $structure, RecordHelper $helper)
     {
+        $this->conf = $conf;
         $this->connection = $connection;
         $this->structure = $structure;
         $this->helper = $helper;
+
+        // get the fieldnames of source of lat / lon geo fields (defined in instance conf)
+        $this->position_fields_mapping = [];
+        foreach($conf->get(['geocoding-providers'], []) as $provider) {
+            if($provider['enabled'] && array_key_exists('position-fields', $provider)) {
+                foreach ($provider['position-fields'] as $position_field) {
+                    $this->position_fields_mapping[$position_field['name']] = $position_field['type'];
+                }
+            }
+        }
+
+        $this->caption_gps_position = new GpsPosition();
+        $this->exif_gps_position = new GpsPosition();
     }
 
     public function hydrateRecords(array &$records)
     {
-        $sql = "(SELECT record_id, ms.name AS `key`, m.value AS value, 'caption' AS type, ms.business AS private\n"
+        $sql = "SELECT * FROM ("
+            . "(SELECT record_id, ms.name AS `key`, m.value AS value, 'caption' AS type, ms.business AS private\n"
             . " FROM metadatas AS m INNER JOIN metadatas_structure AS ms ON (ms.id = m.meta_struct_id)\n"
             . " WHERE record_id IN (?))\n"
             . "UNION\n"
             . "(SELECT record_id, t.name AS `key`, t.value AS value, 'exif' AS type, 0 AS private\n"
             . " FROM technical_datas AS t\n"
-            . " WHERE record_id IN (?))\n";
+            . " WHERE record_id IN (?))\n"
+            . ") AS t ORDER BY record_id";
 
         $ids = array_keys($records);
         $statement = $this->connection->executeQuery(
@@ -54,7 +75,17 @@ class MetadataHydrator implements HydratorInterface
             array(Connection::PARAM_INT_ARRAY, Connection::PARAM_INT_ARRAY)
         );
 
+        $record_id = -1;
         while ($metadata = $statement->fetch()) {
+
+            if($metadata['record_id'] !== $record_id) {
+                // record has changed, don't mix with previous one
+                $this->caption_gps_position->clear();
+                $this->exif_gps_position->clear();
+
+                $record_id = $metadata['record_id'];
+            }
+
             // Store metadata value
             $key = $metadata['key'];
             $value = trim($metadata['value']);
@@ -64,10 +95,10 @@ class MetadataHydrator implements HydratorInterface
                 continue;
             }
 
-            $id = $metadata['record_id'];
+            if (isset($records[$record_id])) {
-            if (isset($records[$id])) {
+                $record =& $records[$record_id];
-                $record =& $records[$id];
+            }
-            } else {
+            else {
                 throw new Exception('Received metadata from unexpected record');
             }
 
@@ -89,11 +120,31 @@ class MetadataHydrator implements HydratorInterface
                         $record[$field] = array();
                     }
                     $record[$field][] = $value;
+
+                    if(array_key_exists($key, $this->position_fields_mapping)) {
+                        // this field is mapped as a position part (lat, lon, latlon), push it
+                        switch($this->position_fields_mapping[$key]) {
+                            case 'lat':
+                                $this->handleGpsPosition($this->caption_gps_position, $record, GpsPosition::LATITUDE_TAG_NAME, $value);
+                                break;
+                            case 'lng':
+                            case 'lon':
+                                $this->handleGpsPosition($this->caption_gps_position, $record, GpsPosition::LONGITUDE_TAG_NAME, $value);
+                                break;
+                            case 'latlng':
+                            case 'latlon':
+                                $this->handleGpsPosition($this->caption_gps_position, $record, GpsPosition::FULL_GEO_NOTATION, $value);
+                                break;
+                        }
+                    }
+
                     break;
 
                 case 'exif':
-                    if (GpsPosition::isSupportedTagName($key)) {
+                    // exif gps is a first-chance if caption is not yet set
-                        $this->handleGpsPosition($records, $id, $key, $value);
+                    // anyway if caption is set later, it will override the exif values
+                    if (GpsPosition::isSupportedTagName($key) && !$this->caption_gps_position->isCompleteComposite()) {
+                        $this->handleGpsPosition($this->exif_gps_position, $record, $key, $value);
                         break;
                     }
                     $tag = $this->structure->getMetadataTagByName($key);
@@ -109,38 +160,25 @@ class MetadataHydrator implements HydratorInterface
                     break;
             }
         }
-
-        $this->clearGpsPositionBuffer();
     }
 
-    private function handleGpsPosition(&$records, $id, $tag_name, $value)
+    private function handleGpsPosition(GpsPosition &$position, &$record, $tag_name, $value)
     {
-        // Get position object
-        if (!isset($this->gps_position_buffer[$id])) {
-            $this->gps_position_buffer[$id] = new GpsPosition();
-        }
-        $position = $this->gps_position_buffer[$id];
         // Push this tag into object
         $position->set($tag_name, $value);
+
         // Try to output complete position
         if ($position->isCompleteComposite()) {
             $lon = $position->getCompositeLongitude();
             $lat = $position->getCompositeLatitude();
 
-            $records[$id]['metadata_tags']['Longitude'] = $lon;
+            $record['metadata_tags']['Longitude'] = $lon;
-            $records[$id]['metadata_tags']['Latitude'] = $lat;
+            $record['metadata_tags']['Latitude'] = $lat;
 
-            $records[$id]["location"] = [
+            $record["location"] = [
                 "lat" => $lat,
                 "lon" => $lon
             ];
-
-            unset($this->gps_position_buffer[$id]);
         }
     }
-
-    private function clearGpsPositionBuffer()
-    {
-        $this->gps_position_buffer = [];
-    }
 }