PHP Functions::get_mimetype方法代码示例

 private function get_images($records)
 {
     foreach ($records as $rec) {
         /* not used
            [http://purl.org/dc/terms/format] => jpeg
            [http://purl.org/dc/terms/audience] => experts, general public
            */
         $mr = new \eol_schema\MediaResource();
         $mr->taxonID = $rec["http://rs.tdwg.org/dwc/terms/taxonID"];
         $mr->identifier = $rec["http://purl.org/dc/terms/identifier"];
         $mr->type = "http://purl.org/dc/dcmitype/StillImage";
         $mr->title = (string) @$rec["http://purl.org/dc/terms/title"];
         $mr->description = (string) @$rec["http://purl.org/dc/terms/description"];
         $mr->accessURI = $rec["http://purl.org/dc/terms/identifier"];
         if ($val = Functions::get_mimetype($mr->accessURI)) {
             $mr->format = $val;
         } else {
             continue;
         }
         if ($val = self::get_license(@$rec["http://purl.org/dc/terms/license"])) {
             $mr->UsageTerms = $val;
         } else {
             continue;
         }
         $mr->creator = (string) @$rec["http://purl.org/dc/terms/creator"];
         $mr->publisher = (string) @$rec["http://purl.org/dc/terms/publisher"];
         // $mr->subtype        = (string) $rec["http://rs.tdwg.org/audubon_core/subtype"];
         // $mr->Rating         = (string) $rec["http://ns.adobe.com/xap/1.0/Rating"];
         // $mr->audience       = (string) $rec["http://purl.org/dc/terms/audience"];
         // $mr->language       = (string) $rec["http://purl.org/dc/terms/language"];
         // $mr->CVterm         = (string) $rec["http://iptc.org/std/Iptc4xmpExt/1.0/xmlns/CVterm"];
         // $mr->CreateDate     = (string) $rec["http://ns.adobe.com/xap/1.0/CreateDate"];
         // $mr->modified       = (string) $rec["http://purl.org/dc/terms/modified"];
         // $mr->Owner          = (string) $rec["http://ns.adobe.com/xap/1.0/rights/Owner"];
         // $mr->rights         = (string) $rec["http://purl.org/dc/terms/rights"];
         // $mr->bibliographicCitation = (string) $rec["http://purl.org/dc/terms/bibliographicCitation"];
         // $mr->derivedFrom     = (string) $rec["http://rs.tdwg.org/ac/terms/derivedFrom"];
         // $mr->LocationCreated = (string) $rec["http://iptc.org/std/Iptc4xmpExt/1.0/xmlns/LocationCreated"];
         // $mr->spatial         = (string) $rec["http://purl.org/dc/terms/spatial"];
         // $mr->lat             = (string) $rec["http://www.w3.org/2003/01/geo/wgs84_pos#lat"];
         // $mr->long            = (string) $rec["http://www.w3.org/2003/01/geo/wgs84_pos#long"];
         // $mr->alt             = (string) $rec["http://www.w3.org/2003/01/geo/wgs84_pos#alt"];
         // $mr->contributor    = (string) $rec["http://purl.org/dc/terms/contributor"];
         // if($agentID = (string) $rec["http://eol.org/schema/agent/agentID"])
         // {
         //     $ids = explode(",", $agentID); // not sure yet what separator Worms used, comma or semicolon - or if there are any
         //     $agent_ids = array();
         //     foreach($ids as $id) $agent_ids[] = $id;
         //     $mr->agentID = implode("; ", $agent_ids);
         // }
         // if($referenceID = self::prepare_reference((string) $rec["http://eol.org/schema/reference/referenceID"])) $mr->referenceID = $referenceID;
         // $mr->thumbnailURL   = (string) $rec["http://eol.org/schema/media/thumbnailURL"];
         // if($source = (string) $rec["http://rs.tdwg.org/ac/terms/furtherInformationURL"]) $mr->furtherInformationURL = $source;
         if (!isset($this->object_ids[$mr->identifier])) {
             $this->object_ids[$mr->identifier] = '';
             $this->archive_builder->write_object_to_file($mr);
         }
     }
 }

 private function create_images($description, $taxon_id, $media_id, $media_url, $source_url)
 {
     if (in_array($media_id, $this->media_ids)) {
         return;
     }
     $this->media_ids[] = $media_id;
     $mr = new \eol_schema\MediaResource();
     if ($val = $this->agent_id) {
         $mr->agentID = implode("; ", $val);
     }
     $mr->taxonID = $taxon_id;
     $mr->identifier = $media_id;
     $mr->type = "http://purl.org/dc/dcmitype/StillImage";
     $mr->language = 'en';
     $mr->format = Functions::get_mimetype($media_url);
     $mr->CVterm = "";
     $mr->rights = "";
     $mr->Owner = "Chaloklum Diving";
     $mr->title = "";
     $mr->UsageTerms = "http://creativecommons.org/licenses/by/3.0/";
     $mr->description = $description;
     $mr->accessURI = $media_url;
     $mr->furtherInformationURL = $source_url;
     $this->archive_builder->write_object_to_file($mr);
 }

 private function get_images($sciname, $description, $taxon_id, $media_id, $media_url, $source_url, $reference_ids, $agent_ids)
 {
     /* this has to be done because there are many maps written in html as jpg but are actually gif. in the site these maps are not showing, meaning typo in html.
        since there is only a handful of jpg maps,  i set all maps to gif */
     $jpeg_maps = array("343dm.jpg", "341dm.jpg", "484ddm.jpg", "535dm2.jpg", "172ms.jpg", "202md.jpg");
     $parts = pathinfo($media_url);
     if (!in_array($parts["basename"], $jpeg_maps)) {
         $media_url = str_ireplace(".jpg", ".gif", $media_url);
     }
     if (in_array($media_id, $this->media_ids)) {
         return;
     }
     $this->media_ids[] = $media_id;
     $mr = new \eol_schema\MediaResource();
     if ($reference_ids) {
         $mr->referenceID = implode("; ", $reference_ids);
     }
     if ($agent_ids) {
         $mr->agentID = implode("; ", $agent_ids);
     }
     $mr->taxonID = (string) $taxon_id;
     $mr->identifier = (string) $media_id;
     $mr->type = "http://purl.org/dc/dcmitype/StillImage";
     $mr->language = 'en';
     $mr->format = Functions::get_mimetype($media_url);
     $mr->furtherInformationURL = $source_url;
     $mr->CVterm = "";
     $mr->Owner = "";
     $mr->rights = "";
     $mr->title = "Distribution of {$sciname} in North America north of Mexico";
     $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc/3.0/";
     $mr->audience = 'Everyone';
     $description = str_ireplace("available on this site", "available on the Singing Insects of North America site", $description);
     $mr->description = $description;
     $mr->subtype = "Map";
     $mr->accessURI = (string) trim($media_url);
     $this->archive_builder->write_object_to_file($mr);
 }

 private function get_images($rec, $reference_ids = null, $agent_ids = null)
 {
     /*
     <specimen_multimedia>
       <media>
         <mediaID>1203947</mediaID>
         <image_link>http://www.boldsystems.org/pics/_w300/BLPDT/10-SRNP-108212_DHJ653087+1309969002.jpg</image_link>
         <photographer>Daniel H. Janzen</photographer>
         <licensing>
           <license>CreativeCommons - Attribution Non-Commercial Share-Alike</license>
           <licenseholder>Daniel H. Janzen</licenseholder>
           <licenseholder_institution>Guanacaste Dry Forest Conservation Fund</licenseholder_institution>
           <year>2010</year>
         </licensing>
       </media>
     </specimen_multimedia>
     */
     if (!@$rec->specimen_multimedia) {
         return;
     }
     $count = 0;
     foreach (@$rec->specimen_multimedia->media as $media) {
         if (trim(@$media->image_link) != "" && !is_numeric(stripos($media->licensing->license, "No Derivatives"))) {
             $SampleID = trim($rec->specimen_identifiers->sampleid);
             $ProcessID = trim($rec->processid);
             $Orientation = trim($media->caption);
             // start checking if image already exists from first/original images resource
             $old_id = $SampleID . "_" . $ProcessID . "_" . $Orientation;
             if (in_array($old_id, $this->old_bolds_image_ids)) {
                 echo "\n [{$old_id}] Found an old ID, will ignore \n";
                 $this->old_bolds_image_ids_count++;
                 continue;
             }
             // end -
             $taxon_id = trim($rec->taxon_id);
             if (@$this->info[$taxon_id]) {
                 if ($this->info[$taxon_id] == $this->max_images_per_taxon) {
                     echo " --- max {$this->max_images_per_taxon} images reached for [{$taxon_id}][{$rec->sciname}] -- ";
                     break;
                 }
                 $this->info[$taxon_id]++;
             } else {
                 $this->info[$taxon_id] = 1;
             }
             $description = "";
             if (@$rec->specimen_identifiers->sampleid) {
                 $description .= "Sample ID = " . $SampleID . "<br>";
             }
             if (@$rec->processid) {
                 $description .= "Process ID = " . $ProcessID . "<br>";
             }
             if (@$media->caption) {
                 $description .= "Caption = " . $Orientation . "<br>";
             }
             $rights = "";
             if (@$media->licensing->year) {
                 $rights = "Copyright " . $media->licensing->year;
             }
             $rightsHolder = "";
             if (@$media->licensing->licenseholder) {
                 $rightsHolder = $media->licensing->licenseholder;
             }
             if (@$media->licensing->licenseholder_institution) {
                 $rightsHolder .= ". " . $media->licensing->licenseholder_institution . ".";
             }
             $agent_ids = self::get_object_agents($media);
             $mediaID = trim($media->mediaID);
             $license = self::get_license($media->licensing->license);
             if (trim($rec->taxon_id) != "" && $mediaID != "" && $license && Functions::get_mimetype($media->image_link) != "") {
                 if (in_array($mediaID, $this->do_ids)) {
                     echo "\n it should not pass here, just in case... \n";
                     continue;
                 } else {
                     $this->do_ids[] = $mediaID;
                 }
                 $mr = new \eol_schema\MediaResource();
                 if ($reference_ids) {
                     $mr->referenceID = implode("; ", $reference_ids);
                 }
                 if ($agent_ids) {
                     $mr->agentID = implode("; ", $agent_ids);
                 }
                 $mr->taxonID = (string) $rec->taxon_id;
                 $mr->identifier = (string) $mediaID;
                 $mr->type = "http://purl.org/dc/dcmitype/StillImage";
                 $mr->language = 'en';
                 $mr->format = (string) Functions::get_mimetype($media->image_link);
                 $mr->furtherInformationURL = (string) $this->sourceURL . $rec->taxon_id;
                 $mr->description = (string) $description;
                 $mr->UsageTerms = (string) $license;
                 $mr->Owner = (string) $rightsHolder;
                 $mr->rights = (string) $rights;
                 $mr->accessURI = (string) $media->image_link;
                 $mr->Rating = 2;
                 $this->archive_builder->write_object_to_file($mr);
             }
         }
     }
 }

 private function get_images($taxon_id, $rec, $reference_ids = null, $agent_ids = null)
 {
     if (!$rec->Filename) {
         return;
     }
     $exclude = array("Animals_Birds_00787.jpg", "Animals_Crustaceans_00008.jpg", "Management_ParksProtectedAreas_OtherProtections_00026.jpg");
     if (in_array($rec->Filename, $exclude)) {
         return;
     }
     $mediaURL = trim($this->eol_defined_image_path . $rec->Filename);
     $description = "";
     if ($rec->navigationalCategoryHierarchy) {
         $description .= "<br>Category hierarchy: " . str_replace("|", " | ", $rec->navigationalCategoryHierarchy);
     }
     if ($rec->description) {
         $description .= "<br>Description: " . $rec->description;
     }
     if ($rec->captureDevice) {
         $description .= "<br>Capture device: " . $rec->captureDevice;
     }
     if ($rec->captureDetails) {
         $description .= "<br>Capture details: " . $rec->captureDetails;
     }
     if ($rec->dateOriginal) {
         if (stripos($rec->dateOriginal, "null") === false) {
             $description .= "<br>Original date: " . $rec->dateOriginal;
         }
     }
     $locality = "";
     if ($rec->Geo_latitude) {
         $locality .= "Latitude: " . $rec->Geo_latitude;
     }
     if ($rec->Geo_longitude) {
         $locality .= "; Longitude: " . $rec->Geo_longitude;
     }
     if ($locality) {
         $description .= "<br><br>Locality: " . $locality;
     }
     if ($life_stage = self::get_lifestage($rec)) {
         $description .= "<br><br>Life stage: " . $life_stage;
     }
     // if($related_media = self::get_related_media($rec)) $description .= "<br><br>Related media: " . $related_media;
     $location_created = "";
     if ($rec->Geo_description) {
         $location_created .= "Locality: " . $rec->Geo_description . ". ";
     }
     if ($rec->Geo_subprovince) {
         $location_created .= "Sub-province: " . $rec->Geo_subprovince . ". ";
     }
     if ($rec->Geo_stateProvince) {
         $location_created .= "State Province: " . $rec->Geo_stateProvince . ". ";
     }
     if ($rec->Geo_country) {
         $location_created .= "Country: " . $rec->Geo_country . ". ";
     }
     if ($rec->Geo_continentOcean) {
         $location_created .= "Continent Ocean: " . $rec->Geo_continentOcean . ". ";
     }
     $mr = new \eol_schema\MediaResource();
     if ($reference_ids) {
         $mr->referenceID = implode("; ", $reference_ids);
     }
     if ($agent_ids) {
         $mr->agentID = implode("; ", $agent_ids);
     }
     $mr->taxonID = (string) $taxon_id;
     $mr->identifier = (string) $rec->Filename;
     $mr->type = "http://purl.org/dc/dcmitype/StillImage";
     $mr->language = 'en';
     $mr->format = Functions::get_mimetype($mediaURL);
     if ($mr->format != "image/jpeg") {
         echo "\n investigate: not jpg: {$mediaURL} [{$mr->format}] \n";
     }
     $mr->title = (string) $rec->title;
     $mr->CreateDate = (string) $rec->dateOriginal;
     $mr->CVterm = "";
     // $mr->rights         = "";
     $info = self::process_credit_line((string) $rec->creditLine);
     $mr->Owner = $info["creditLine"];
     $mr->UsageTerms = $info["license"];
     $mr->audience = 'Everyone';
     $mr->description = (string) $description;
     $mr->LocationCreated = (string) $location_created;
     $mr->accessURI = $mediaURL;
     // for stats
     if (stripos($rec->originalFileName, "(c)") !== false) {
         $this->copyrighted++;
     }
     $resourceid = (string) $rec->resourceid;
     if (!in_array($resourceid, $this->resourceids)) {
         $this->resourceids[] = $resourceid;
     } else {
         return;
     }
     if (!in_array($mr->identifier, $this->object_ids)) {
         $this->object_ids[] = $mr->identifier;
         $this->archive_builder->write_object_to_file($mr);
         // for stats
         $archiveName = (string) $rec->archiveName;
         if ($archiveName) {
//.........这里部分代码省略.........

 private function create_image_objects($records)
 {
     foreach ($records as $rec) {
         if (!$rec->url_foto) {
             continue;
         }
         $rec->especie = self::clean_name($rec->especie);
         $taxon_id = strtolower(str_replace(" ", "_", $rec->especie));
         $mr = new \eol_schema\MediaResource();
         $mr->taxonID = $taxon_id;
         $mr->identifier = md5($rec->url_foto);
         $mr->type = 'http://purl.org/dc/dcmitype/StillImage';
         $mr->language = 'es';
         $mr->format = Functions::get_mimetype($rec->url_foto);
         $mr->furtherInformationURL = "http://www.acguanacaste.ac.cr/biodiversidad/voucher.php?voucher=" . $rec->voucher;
         $mr->accessURI = $rec->url_foto;
         $mr->thumbnailURL = $rec->url_thumb;
         $mr->Owner = 'Guanacaste Conservation Area';
         $mr->rights = '';
         $mr->title = '';
         $mr->UsageTerms = 'http://creativecommons.org/licenses/by-nc-sa/3.0/';
         $mr->description = self::get_description($rec);
         $mr->LocationCreated = self::get_location($rec);
         if ($agent_ids = self::get_agent_ids($rec->autor)) {
             $mr->agentID = implode("; ", $agent_ids);
         }
         if (!isset($this->object_ids[$mr->identifier])) {
             $this->object_ids[$mr->identifier] = '';
             $this->archive_builder->write_object_to_file($mr);
         }
     }
 }

 private function prepare_images($taxon, $images)
 {
     $reference_ids = array();
     $ref_ids = array();
     $agent_ids = array();
     foreach ($images as $rec) {
         echo "\n - " . $taxon . " - " . $rec['url'];
         $media_url = $rec["image"];
         echo "\n media url: " . $media_url . "\n\n";
         $path_parts = pathinfo($rec["image"]);
         $identifier = (string) $rec["taxon_id"] . "_" . str_replace(" ", "_", $path_parts["basename"]);
         if (in_array($identifier, $this->do_ids)) {
             continue;
         } else {
             $this->do_ids[] = $identifier;
         }
         $mr = new \eol_schema\MediaResource();
         if ($reference_ids) {
             $mr->referenceID = implode("; ", $reference_ids);
         }
         if ($agent_ids) {
             $mr->agentID = implode("; ", $agent_ids);
         }
         $mr->taxonID = (string) $rec["taxon_id"];
         $mr->identifier = $identifier;
         $mr->type = "http://purl.org/dc/dcmitype/StillImage";
         $mr->language = 'en';
         $mr->format = (string) Functions::get_mimetype($media_url);
         $mr->furtherInformationURL = (string) $rec['url'];
         $mr->accessURI = (string) $media_url;
         $mr->Owner = "";
         $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
         $mr->description = (string) $rec["caption"];
         $this->archive_builder->write_object_to_file($mr);
         $this->create_instances_from_taxon_object($taxon, $rec, $reference_ids);
     }
 }

 private function prepare_images($rec)
 {
     if ($mediaURLs = self::get_image_urls($rec)) {
         print "\n images: " . count($mediaURLs) . "\n";
         foreach ($mediaURLs as $mediaURL) {
             /* not used for now
                $desc = "";
                if($val = $rec["Locality"]) $desc .= "Locality: " . $val . "<br>";
                if($val = $rec["Size"])     $desc .= "Size: " . $val . "<br>";
                if($val = $rec["Book"])     $desc .= "Book: " . $val . "<br>";
                if($val = $rec["Synonym"])  $desc .= "Synonym: " . $val . "<br>";
                */
             $mr = new \eol_schema\MediaResource();
             $mr->taxonID = $rec["taxon_id"];
             $mr->identifier = md5($mediaURL);
             $mr->type = "http://purl.org/dc/dcmitype/StillImage";
             $mr->format = Functions::get_mimetype($mediaURL);
             $mr->Owner = "Femorale";
             $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc/3.0/";
             $mr->accessURI = $mediaURL;
             $mr->furtherInformationURL = str_replace(" ", "%20", $rec["Expr1"]);
             if (!isset($this->object_ids[$mr->identifier])) {
                 $this->archive_builder->write_object_to_file($mr);
                 $this->object_ids[$mr->identifier] = '';
             }
         }
     }
 }

 private function process_records($records)
 {
     /*
     [257] => Array
             (
                 [img] => /caribbean/resources/img/images/species/3908_3406.jpg
                 [url] => /caribbean/en/thefishes/species/3908
                 [sciname] => Xyrichtys novacula
                 [vernacular] => Pearly razorfish
             )
     */
     foreach ($records as $contributor => $contributor_images) {
         foreach ($contributor_images as $rec) {
             if ($rec["sciname"]) {
                 $taxon = new \eol_schema\Taxon();
                 $taxon->taxonID = str_replace(" ", "_", $rec["sciname"]);
                 $taxon->scientificName = $rec["sciname"];
                 if ($val = $rec["url"]) {
                     $taxon->furtherInformationURL = $this->domain . $val;
                 }
                 if (!isset($this->taxa[$taxon->taxonID])) {
                     $this->taxa[$taxon->taxonID] = '';
                     $this->archive_builder->write_object_to_file($taxon);
                 }
             }
             if ($rec["img"]) {
                 $temp = pathinfo($rec["img"]);
                 $identifier = $temp["filename"];
                 $agent_ids = self::process_agent($contributor);
                 $mr = new \eol_schema\MediaResource();
                 if ($agent_ids) {
                     $mr->agentID = implode("; ", $agent_ids);
                 }
                 $mr->taxonID = $taxon->taxonID;
                 $mr->identifier = $identifier;
                 $mr->type = 'http://purl.org/dc/dcmitype/StillImage';
                 $mr->format = Functions::get_mimetype($rec["img"]);
                 if (preg_match("/_(.*?)\\./ims", $rec["img"], $arr)) {
                     $mr->furtherInformationURL = $this->image_summary_page . $arr[1];
                 }
                 $mr->UsageTerms = 'http://creativecommons.org/licenses/by-nc-sa/3.0/';
                 $mr->accessURI = $this->domain . $rec["img"];
                 $mr->Owner = $contributor;
                 $mr->description = '';
                 $mr->publisher = '';
                 $mr->CVterm = '';
                 $mr->title = '';
                 $mr->spatial = '';
                 //this is 'location' in attribution in EOL's data_object page
                 if (!isset($this->object_ids[$mr->identifier])) {
                     $this->object_ids[$mr->identifier] = '';
                     $this->archive_builder->write_object_to_file($mr);
                 }
             }
         }
         //inner loop
     }
     //outer loop
 }

 function get_taxon_details($id)
 {
     $languages = array("en" => self::ENGLISH_DIR, "nl" => self::DUTCH_DIR);
     $description = "";
     $title = "";
     $identifier = "";
     $mediaURL = "";
     $mimeType = "";
     $rightsHolder = "";
     $source = "";
     $texts = array();
     $objects = array();
     $text_rightsHolder = "";
     foreach ($languages as $language => $dir) {
         debug("\n {$language} - {$dir} \n");
         if ($language == "nl") {
             if (substr($description, strlen($description) - 2, 2) == ". ") {
                 $description = substr($description, 0, strlen($description) - 2) . "; ";
             }
             if (substr($title, strlen($title) - 2, 2) == ". ") {
                 $title = substr($title, 0, strlen($title) - 2) . "; ";
             }
         }
         $xml_file = self::ECOMARE_SOURCE_DIR . "/{$dir}/" . $id . ".xml";
         debug("\n Accessing: {$xml_file} \n");
         if ($xml = Functions::get_hashed_response($xml_file, array('download_wait_time' => 1000000, 'timeout' => 600, 'download_attempts' => 2, 'delay_in_minutes' => 2))) {
             debug(" - OK ");
             foreach ($xml->content_item as $item) {
                 if ($item->type == 9) {
                     $source_url = str_ireplace("L=0", "L=2", $xml->subject_setup->source_url);
                     if (trim($item->file_name) != "") {
                         $description .= $item->image_description != '' ? "" . $item->image_description . ". " : '';
                         $title .= $item->header != '' ? "" . $item->header . ". " : '';
                         $identifier = "img_" . $item->uid;
                         $mediaURL = self::ECOMARE_DOMAIN . $item->file_path . $item->file_name;
                         $mimeType = Functions::get_mimetype($item->file_name);
                         $rightsHolder = $item->image_copyright;
                         $source = $source_url;
                     }
                     if (!in_array($item->bodytext, $texts)) {
                         $objects[] = array("identifier" => $item->uid, "mediaURL" => '', "mimeType" => 'text/html', "dataType" => 'http://purl.org/dc/dcmitype/Text', "rightsHolder" => '', "title" => '', "description" => strip_tags($item->bodytext), "source" => $source_url, "license" => "http://creativecommons.org/licenses/by-nc/3.0/", "subject" => "http://rs.tdwg.org/ontology/voc/SPMInfoItems#TaxonBiology", "language" => $language);
                         $texts[] = $item->bodytext;
                     }
                 } elseif ($item->type == 10) {
                     $text_rightsHolder = strip_tags($item->bodytext);
                 }
             }
         }
     }
     // language loop
     $objects[] = array("identifier" => $identifier, "mediaURL" => $mediaURL, "mimeType" => $mimeType, "dataType" => 'http://purl.org/dc/dcmitype/StillImage', "rightsHolder" => $rightsHolder, "title" => $title, "description" => strip_tags($description), "source" => $source, "license" => "http://creativecommons.org/licenses/by-nc/3.0/", "subject" => "", "language" => "en");
     $i = 0;
     foreach ($objects as $object) {
         if ($object["mimeType"] == 'text/html') {
             $objects[$i]["rightsHolder"] = $text_rightsHolder;
         }
         $i++;
     }
     return $objects;
 }

 private function get_images($description, $taxon_id, $media_id, $media_url, $reference_ids, $agent_ids)
 {
     $description = utf8_encode(self::remove_quotes($description));
     if (in_array($media_id, $this->media_ids)) {
         return;
     }
     $this->media_ids[] = $media_id;
     $bibliographicCitation = false;
     if ($reference_ids = @$this->image_references[$media_id]) {
         $reference_ids = array_unique($reference_ids);
         $bibliographicCitation = self::get_citation($reference_ids);
     }
     $mr = new \eol_schema\MediaResource();
     if ($reference_ids) {
         $mr->referenceID = implode("; ", $reference_ids);
     }
     if ($agent_ids) {
         $mr->agentID = implode("; ", $agent_ids);
     }
     $mr->taxonID = (string) $taxon_id;
     $mr->identifier = (string) $media_id;
     $mr->type = "http://purl.org/dc/dcmitype/StillImage";
     $mr->language = 'en';
     $mr->format = Functions::get_mimetype($media_url);
     $mr->CVterm = "";
     $mr->Owner = "";
     $mr->rights = "";
     $mr->title = "";
     $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc/3.0/";
     $mr->audience = 'Everyone';
     $mr->description = (string) $description;
     $mr->accessURI = $media_url;
     $mr->furtherInformationURL = $this->taxon_url . $taxon_id;
     if ($bibliographicCitation) {
         $mr->bibliographicCitation = $bibliographicCitation;
     }
     $this->archive_builder->write_object_to_file($mr);
 }

 private function get_images($rec, $agent_ids, $reference_ids)
 {
     $texts = $rec['texts'];
     if (@$texts['map_caption'] && @$texts['map_image']) {
         $description = (string) utf8_encode($texts['map_caption']);
         $mediaURL = (string) utf8_encode($texts['map_image']);
         $dc_identifier = "map_" . $rec['taxonID'];
         if (trim($description) && trim($mediaURL)) {
             $mr = new \eol_schema\MediaResource();
             if ($reference_ids) {
                 $mr->referenceID = implode("; ", $reference_ids);
             }
             if ($agent_ids) {
                 $mr->agentID = implode("; ", $agent_ids);
             }
             $mr->taxonID = (string) $rec['taxonID'];
             $mr->identifier = (string) $dc_identifier;
             $mr->type = "http://purl.org/dc/dcmitype/StillImage";
             $mr->language = 'en';
             $mr->format = Functions::get_mimetype($mediaURL);
             $mr->furtherInformationURL = (string) self::clean_str(trim($rec['url']));
             $mr->CVterm = "";
             $mr->Owner = "IUCN/SSC Tortoise and Freshwater Turtle Specialist Group";
             $mr->rights = "Copyright 2009 Chelonian Research Foundation";
             $mr->title = "Distribution";
             $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
             $mr->audience = 'Everyone';
             $mr->description = (string) $description;
             $mr->subtype = "Map";
             $mr->accessURI = $mediaURL;
             // $mr->bibliographicCitation = '';
             $this->archive_builder->write_object_to_file($mr);
         }
     }
 }

 private function process_record($rec)
 {
     $rec["sciname"] = self::valid_sciname($rec);
     if ($rec["sciname"]) {
         $taxon = new \eol_schema\Taxon();
         $taxon->taxonID = str_replace(" ", "_", $rec["sciname"]);
         $taxon->scientificName = $rec["sciname"];
         $taxon->furtherInformationURL = $rec["page"];
         if (!isset($this->taxa[$taxon->taxonID])) {
             $this->taxa[$taxon->taxonID] = '';
             $this->archive_builder->write_object_to_file($taxon);
         }
     } else {
         return;
     }
     if ($rec["src"]) {
         $agent_ids = self::process_agent();
         $mr = new \eol_schema\MediaResource();
         // if($reference_ids) $mr->referenceID = implode("; ", $reference_ids); //not used at the moment...
         if ($agent_ids) {
             $mr->agentID = implode("; ", $agent_ids);
         }
         $mr->taxonID = $taxon->taxonID;
         if ($rec["image_id"] == "258161") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Cistus creticus");
         } elseif ($rec["image_id"] == "238990") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Euphorbia antilibanotica");
         } elseif ($rec["image_id"] == "213710") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Ziziphora clinopodioides");
         } elseif ($rec["image_id"] == "212647") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Melanargia galathea");
         } elseif ($rec["image_id"] == "212559") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Galanthus");
         } elseif ($rec["image_id"] == "120397") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Plantago ciliata");
         } elseif ($rec["image_id"] == "48044") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Ixia sorrel");
         } elseif ($rec["image_id"] == "231718") {
             $mr->taxonID .= ";" . str_replace(" ", "_", "Ranunculus asiaticus");
         }
         $mr->identifier = $rec["image_id"];
         $mr->type = 'http://purl.org/dc/dcmitype/StillImage';
         $mr->format = Functions::get_mimetype($rec["src"]);
         $mr->furtherInformationURL = $rec["source"];
         $mr->UsageTerms = 'http://creativecommons.org/licenses/by-nc-sa/3.0/';
         $mr->accessURI = $rec["src"];
         $mr->description = utf8_encode($rec["caption"]);
         $mr->Owner = '';
         $mr->publisher = '';
         $mr->CVterm = '';
         $mr->title = '';
         $mr->spatial = $rec["location"];
         if (!isset($this->object_ids[$mr->identifier])) {
             $this->object_ids[$mr->identifier] = '';
             $this->archive_builder->write_object_to_file($mr);
         }
     }
 }

 private function process_images($rec, $url)
 {
     if ($html = self::get_html($url)) {
         // <img src="thb/11395a_thb.gif" alt="fig Calviria solaris">
         $source_urls = self::get_source_urls($html);
         if (preg_match_all("/src=\"thb\\/(.*?)\"/ims", $html, $arr)) {
             // http://turbellaria.umaine.edu/gif/11397a.gif
             $index = 0;
             foreach ($arr[1] as $image) {
                 // from: http://turbellaria.umaine.edu/thb/12223a_thb.gif to: http://turbellaria.umaine.edu/gif/12223a.gif
                 $info = self::get_folder_of_image($image);
                 $folder = $info["folder"];
                 $extension = $info["extension"];
                 if (!$folder) {
                     // e.g <img src="thb/13158jpg_thb.158" alt="fig Kuma asilhas"> http://turbellaria.umaine.edu/turb3.php?action=2&code=13158&smk=0
                     if (is_numeric($extension)) {
                         echo "\n [{$image}] \n";
                         $parts = explode(".", $image);
                         $image = $parts[0];
                         $image = str_ireplace("jpg_thb", "_thb.jpg", $image);
                         $image = str_ireplace("gif_thb", "_thb.gif", $image);
                         $image = str_ireplace("png_thb", "_thb.png", $image);
                         echo "\n [{$image}] \n";
                         $info = self::get_folder_of_image($image);
                         $folder = $info["folder"];
                         $extension = $info["extension"];
                         if (!$folder) {
                             echo "\n investigate no image 2 [{$url}][{$extension}] \n";
                         }
                     } else {
                         echo "\n investigate no image [{$url}][{$extension}] \n";
                         print_r($rec);
                         continue;
                     }
                 }
                 $mediaURL = $this->domain . $folder . str_ireplace("_thb", "", $image);
                 $mr = new \eol_schema\MediaResource();
                 if ($this->agent_ids) {
                     $mr->agentID = implode("; ", $this->agent_ids);
                 }
                 $mr->taxonID = (string) $rec["taxon_id"];
                 $mr->identifier = (string) $mediaURL;
                 $mr->type = "http://purl.org/dc/dcmitype/StillImage";
                 $mr->language = 'en';
                 $mr->format = Functions::get_mimetype($mediaURL);
                 $mr->title = "";
                 $mr->CreateDate = "";
                 $mr->Owner = $this->rights_holder;
                 $mr->rights = "";
                 $mr->UsageTerms = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
                 $mr->audience = 'Everyone';
                 $mr->description = "";
                 $mr->accessURI = $mediaURL;
                 if (@$source_urls[$index]) {
                     $mr->furtherInformationURL = @$source_urls[$index];
                 }
                 if (!in_array($mr->identifier, $this->object_ids)) {
                     $this->object_ids[] = $mr->identifier;
                     $this->archive_builder->write_object_to_file($mr);
                 }
                 $index++;
             }
         }
     }
 }

 function prepare_species_page($taxon, $taxon_images, $taxon_comnames, $taxon_syn)
 {
     $arr_scraped = array();
     $arr_photos = array();
     $arr_sciname = array();
     $taxon_id = $taxon["SId"];
     $sciname = $taxon['GenusSpecies'];
     $agent = array();
     $agent[] = array("role" => "project", "homepage" => "http://www.fishwisepro.com", "name" => "FishWise Professional");
     $rights_holder = "FishWise Professional";
     $mimeType = "text/html";
     $dataType = "http://purl.org/dc/dcmitype/Text";
     $mediaURL = "";
     $reference = array();
     $dc_source = FWP_SPECIES_PAGE_URL . $taxon['SId'];
     if (trim(@$taxon['DistributionT']) != "---") {
         $desc = $taxon['DistributionT'];
         $desc = utf8_decode($desc);
         $arr_texts[$sciname][] = self::fill_data_object($desc, "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Distribution", "", $agent, $rights_holder, $mediaURL, $mimeType, $dataType, $dc_source);
     }
     if (trim(@$taxon['LengthMax']) > 0) {
         $desc = "Maximum size: {$taxon['LengthMax']} {$taxon['LengthMaxSuffix']} {$taxon['LengthMaxType']}";
         $desc = utf8_decode($desc);
         $arr_texts[$sciname][] = self::fill_data_object($desc, "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Size", "", $agent, $rights_holder, $mediaURL, $mimeType, $dataType, $dc_source);
     }
     if (trim(@$taxon['DepthRangeDeep']) > 0) {
         $desc = "Depth: {$taxon['DepthRangeShallow']} - {$taxon['DepthRangeDeep']}m. ";
         if (trim(@$taxon['DepthRange']) != "---") {
             $desc .= "<br>{$taxon['DepthRange']}";
         }
         if (trim(@$taxon['Habitat']) != "---") {
             $desc .= "<br><br>Habitat: {$taxon['Habitat']}.";
         }
         if (trim(@$taxon['HabitatNotes']) != "---") {
             $desc .= " {$taxon['HabitatNotes']}";
         }
         $desc = utf8_decode($desc);
         $arr_texts[$sciname][] = self::fill_data_object($desc, "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Habitat", "", $agent, $rights_holder, $mediaURL, $mimeType, $dataType, $dc_source);
     }
     $arr_photos = array();
     if ($arr = @$taxon_images) {
         foreach ($arr as $r) {
             if (trim(@$r['FileName']) != "---") {
                 $path_info = pathinfo($r['FileName']);
                 $extension = strtolower($path_info['extension']);
                 $mediaURL = FWP_IMAGE_URL . "{$extension}/" . trim(@$r['FileName']);
                 $mimeType = Functions::get_mimetype($mediaURL);
                 $dataType = "http://purl.org/dc/dcmitype/StillImage";
                 $desc = "";
                 if (trim(@$r['LifeStage']) != "---") {
                     $desc .= "Life stage: " . $r['LifeStage'];
                 }
                 if (trim(@$r['dbo_Picture_PictureNote']) != "---") {
                     $desc .= "<br>Note: " . $r['dbo_Picture_PictureNote'];
                 }
                 if (trim(@$r['PictureType']) != "---") {
                     $desc .= "<br>Type: " . $r['PictureType'];
                 }
                 if (trim(@$r['Location']) != "---") {
                     $desc .= "<br>Location: " . $r['Location'];
                 }
                 $desc = utf8_decode(trim($desc));
                 $agent = array();
                 $agent[] = array("role" => "photographer", "homepage" => "", "name" => @$r['DisplayName']);
                 $dc_source = FWP_IMAGE_PAGE_URL . "&SId=" . $r['SId'] . "&PictureId=" . $r['PictureId'];
                 $arr_photos[$sciname][] = self::fill_data_object($desc, "", "", $agent, $rights_holder, $mediaURL, $mimeType, $dataType, $dc_source);
             }
         }
     }
     if (@$arr_texts[$sciname] || @$arr_photos[$sciname]) {
         $dc_source = FWP_SPECIES_PAGE_URL . $taxon['SId'];
         $arr_scraped[] = array("identifier" => "SId-" . $taxon['SId'], "kingdom" => "", "phylum" => "", "class" => "", "order" => $taxon['OrderName'], "family" => $taxon['Family'], "sciname" => $sciname, "dc_source" => $dc_source, "texts" => @$arr_texts[$sciname], "photos" => @$arr_photos[$sciname], "references" => $reference, "comnames" => $taxon_comnames, "synonyms" => $taxon_syn);
     }
     return $arr_scraped;
 }