PHP Functions::canonical_form方法代码示例

 public function check_st_john($start, $limit)
 {
     $query = "SELECT id, string FROM names WHERE string REGEXP BINARY 'st\\\\\\.-[a-z]'\n            AND id BETWEEN {$start} AND " . ($start + $limit - 1);
     foreach ($this->mysqli->iterate_file($query) as $row) {
         $id = $row[0];
         $string = $row[1];
         $canonical_form_string = Functions::canonical_form($string);
         if ($canonical_form = CanonicalForm::find_or_create_by_string($canonical_form_string)) {
             echo "UPDATE names SET canonical_form_id={$canonical_form->id} WHERE id={$id}\n";
             $this->mysqli->update("UPDATE names SET canonical_form_id={$canonical_form->id}, ranked_canonical_form_id={$canonical_form->id} WHERE id={$id}");
         }
     }
     $this->mysqli->commit();
 }

 private function get_details($xml, $orig_sciname, $strict)
 {
     $taxa = array();
     foreach ($xml->entry as $species) {
         if ($strict) {
             if (strtolower(trim($orig_sciname)) == strtolower(trim(Functions::canonical_form(trim($species->title))))) {
                 $taxon_do = self::get_objects_info($species->id, $species->title, $orig_sciname);
                 $taxa[] = $taxon_do;
             }
         } else {
             $taxon_do = self::get_objects_info($species->id, $species->title, $orig_sciname);
             $taxa[] = $taxon_do;
         }
     }
     return $taxa;
 }

 function canonical_forms()
 {
     $mysqli =& $GLOBALS['mysqli_connection'];
     $mysqli->begin_transaction();
     $mysqli->delete("DELETE FROM canonical_forms");
     $mysqli->update("UPDATE names SET canonical_form_id=0");
     $result = $mysqli->query("SELECT MAX(id) as max FROM names");
     $row = $result->fetch_assoc();
     $max = $row["max"];
     $start = 1;
     $interval = 50000;
     while ($start < $max) {
         debug($start);
         $result = $mysqli->query("SELECT id, string FROM names WHERE id BETWEEN {$start} AND " . ($start + $interval - 2) . " AND canonical_form_verified!=0");
         while ($result && ($row = $result->fetch_assoc())) {
             $id = $row["id"];
             $string = $row["string"];
             $canonical_form_id = $row["id"];
             $canonical_form_verified = $row["id"];
             $canonical_form = "";
             if ($canonical_form_verified) {
                 $result2 = $mysqli->query("SELECT string FROM canonical_forms WHERE id={$canonical_form_id}");
                 if ($result2 && ($row2 = $result2->fetch_assoc())) {
                     $canonical_form = $row2["string"];
                 }
             }
             if (!$canonical_form) {
                 $canonical_form = Functions::canonical_form($string);
             }
             if (@(!$canonical_form_ids[$canonical_form])) {
                 $result2 = $mysqli->query("SELECT id FROM canonical_forms WHERE string='" . $mysqli->escape($canonical_form) . "'");
                 if ($result2 && ($row2 = $result2->fetch_assoc())) {
                     $canonical_form_ids[$canonical_form] = $row2["id"];
                 } else {
                     $result2 = $mysqli->insert("INSERT INTO canonical_forms VALUES (NULL,'" . $mysqli->escape($canonical_form) . "')");
                     $canonical_form_ids[$canonical_form] = $mysqli->insert_id;
                 }
             }
             $query = "UPDATE names SET canonical_form_id ({$id},'" . $mysqli->escape($canonical_form) . "')";
             $mysqli->update($query);
         }
         flush();
         $start += $interval;
     }
     $mysqli->end_transaction();
 }

 function get_details($xml, $orig_sciname, $strict)
 {
     $taxa = array();
     foreach ($xml->entry as $species) {
         if ($strict == 'canonical_match') {
             if (strtolower(trim($orig_sciname)) == strtolower(trim(Functions::canonical_form(trim($species->title))))) {
                 print "<br>" . strtolower(trim($orig_sciname)) . " == " . strtolower(trim(Functions::canonical_form(trim($species->title)))) . " == " . $species->title . "<br>";
                 $taxon_do = self::get_objects_info($species->id, $species->title, $orig_sciname);
                 $taxa[] = $taxon_do;
             }
         } elseif ($strict == 'exact_string') {
             if (strtolower(trim($orig_sciname)) == strtolower(trim($species->title))) {
                 print "<br>" . strtolower(trim($orig_sciname)) . " == " . $species->title . "<br>";
                 $taxon_do = self::get_objects_info($species->id, $species->title, $orig_sciname);
                 $taxa[] = $taxon_do;
             }
         } else {
             $taxon_do = self::get_objects_info($species->id, $species->title, $orig_sciname);
             $taxa[] = $taxon_do;
         }
     }
     return $taxa;
 }

 private function is_sciname_synonym($sciname)
 {
     /*
     Squatarola squatarola
     http://eol.org/api/search/1.0.xml?q=Xanthopsar+flavus&page=1&exact=false&filter_by_taxon_concept_id=&filter_by_hierarchy_entry_id=&filter_by_string=&cache_ttl=
     http://eol.org/api/pages/1.0.xml?batch=false&id=686274&images_per_page=0&images_page=0&videos_per_page=0&videos_page=0&sounds_per_page=0&sounds_page=0&maps_per_page=0&maps_page=0&texts_per_page=0&texts_page=0&iucn=false&subjects=overview&licenses=all&details=false&common_names=false&synonyms=true&references=false&taxonomy=false&vetted=0&cache_ttl=&language=en
     */
     $search_call = "http://eol.org/api/search/1.0.xml?q=" . $sciname . "&page=1&exact=false&filter_by_taxon_concept_id=&filter_by_hierarchy_entry_id=&filter_by_string=&cache_ttl=";
     if ($xml = Functions::lookup_with_cache($search_call, array('timeout' => 30, 'expire_seconds' => false, 'resource_id' => 'eol_api'))) {
         $xml = simplexml_load_string($xml);
         $sciname = Functions::canonical_form($sciname);
         if ($sciname == Functions::canonical_form($xml->entry[0]->title)) {
             return false;
         } else {
             $titles = array();
             echo "\n" . $xml->entry[0]->id . "\n";
             foreach ($xml->entry as $entry) {
                 $titles[] = Functions::canonical_form($entry->title);
             }
             // print_r($titles);
             if (in_array($sciname, $titles)) {
                 return false;
             } else {
                 return true;
             }
         }
     }
     return false;
 }

 private function process_mushroom_observer_list($wrong_urls)
 {
     if ($file = Functions::lookup_with_cache($this->mushroom_observer_eol, $this->download_options)) {
         $xml = simplexml_load_string($file);
         $i = 0;
         $total = count($xml->taxon);
         foreach ($xml->taxon as $t) {
             $i++;
             // if($i > 40) break; //debug
             $t_dwc = $t->children("http://rs.tdwg.org/dwc/dwcore/");
             $t_dc = $t->children("http://purl.org/dc/elements/1.1/");
             $sciname = Functions::import_decode($t_dwc->ScientificName);
             $sciname = Functions::canonical_form($sciname);
             echo "\n{$i} of {$total}: {$sciname}";
             $url = "http://en.wikipedia.org/wiki/" . str_replace(" ", "_", $sciname);
             self::get_triple($url, $wrong_urls);
         }
     }
 }

 private function get_vernacular_names()
 {
     $temp_filepath = Functions::save_remote_file_to_local($this->vernacular_path, array('timeout' => 4800, 'download_attempts' => 5));
     foreach (new FileIterator($temp_filepath, true) as $line_number => $line) {
         if ($line) {
             $fields = explode("\t", trim($line));
             $fields = array_map('trim', $fields);
             //trims all array values in the array
             $common_name = @$fields[1];
             $sciname = Functions::canonical_form(trim(@$fields[0]));
             $taxon_id = @$this->taxa_all[$sciname]['Identifier'];
             if ($common_name == '' || $taxon_id == '' || $sciname == '') {
                 continue;
             }
             $language = self::get_language(@$fields[3]);
             $vernacular = new \eol_schema\VernacularName();
             $vernacular->taxonID = $taxon_id;
             $vernacular->vernacularName = (string) $common_name;
             $vernacular->language = $language;
             $vernacular_id = md5("{$vernacular->taxonID}|{$vernacular->vernacularName}|{$vernacular->language}");
             if (!isset($this->vernacular_name_ids[$vernacular_id])) {
                 $this->archive_builder->write_object_to_file($vernacular);
                 $this->vernacular_name_ids[$vernacular_id] = 1;
             }
         }
     }
 }

 private function generate_taxon_extension_for_dwca($rec)
 {
     /*
     <field index="0" term="http://rs.tdwg.org/dwc/terms/taxonID"/>
     <field index="2" term="http://rs.tdwg.org/dwc/terms/scientificName"/>
     <field index="3" term="http://rs.tdwg.org/dwc/terms/taxonRank"/>
     http://rs.tdwg.org/ontology/voc/TaxonRank#Subfamily
     http://rs.tdwg.org/ontology/voc/TaxonRank#Tribe
     http://rs.tdwg.org/ontology/voc/TaxonRank#Subtribe
     <field index="4" term="http://rs.tdwg.org/dwc/terms/genus"/>
     <field index="5" term="http://rs.tdwg.org/dwc/terms/subgenus"/>
     http://rs.tdwg.org/ontology/voc/TaxonRank#Species
     */
     // taxonID  scientificName  taxonRank   Subfamily   Tribe   Subtribe    genus   subgenus    Species
     if ($OUT = Functions::file_open($this->taxon_tab_file, "a")) {
         if ($val = @$rec['sciname2']) {
             $scientificName = $val;
         } elseif ($val = @$rec['sciname']) {
             $scientificName = $val;
         }
         fwrite($OUT, $rec['taxon_id'] . "\t");
         fwrite($OUT, $scientificName . "\t");
         fwrite($OUT, $rec['rank'] . "\t");
         fwrite($OUT, @$rec['ancestry']['subfamily']['name'] . "\t");
         fwrite($OUT, @$rec['ancestry']['tribe']['name'] . "\t");
         fwrite($OUT, @$rec['ancestry']['subtribe']['name'] . "\t");
         fwrite($OUT, @$rec['ancestry']['genus']['name'] . "\t");
         fwrite($OUT, @$rec['ancestry']['subgenus']['name'] . "\t");
         $final_species = "";
         if ($species = @$rec['ancestry']['species']['name']) {
             $final_species = Functions::canonical_form($rec['ancestry']['genus']['name']);
             if ($val = @$rec['ancestry']['subgenus']['name']) {
                 $final_species .= " (" . Functions::canonical_form($val) . ")";
             }
             $final_species .= " {$species}";
         }
         fwrite($OUT, $final_species . "\n");
         fclose($OUT);
     }
 }

 private function check_sciname_ancestry_values($taxon)
 {
     //scientificname should not be equal to any of the ancestry
     $canonical = Functions::canonical_form($taxon->scientificName);
     if ($taxon->kingdom == $canonical) {
         $taxon->kingdom = '';
     }
     if ($taxon->phylum == $canonical) {
         $taxon->phylum = '';
     }
     if ($taxon->class == $canonical) {
         $taxon->class = '';
     }
     if ($taxon->order == $canonical) {
         $taxon->order = '';
     }
     if ($taxon->family == $canonical) {
         $taxon->family = '';
     }
     if ($taxon->genus == $canonical) {
         $taxon->genus = '';
     }
     return $taxon;
 }

 private function get_taxon_id($rec)
 {
     $taxon_id = trim((string) $rec["http://rs.tdwg.org/dwc/terms/taxonID"]);
     if (!$taxon_id) {
         if ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/scientificName"])) {
             $taxon_id = str_replace(" ", "_", Functions::canonical_form($val));
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/genus"])) {
             $taxon_id = md5($val);
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/family"])) {
             $taxon_id = md5($val);
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/order"])) {
             $taxon_id = md5($val);
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/class"])) {
             $taxon_id = md5($val);
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/phylum"])) {
             $taxon_id = md5($val);
         } elseif ($val = trim((string) $rec["http://rs.tdwg.org/dwc/terms/kingdom"])) {
             $taxon_id = md5($val);
         } else {
             exit("\n got it \n");
         }
     }
     return $taxon_id;
 }

 private function manuall_add_taxon()
 {
     $records[] = array("sciname" => "Cordulegaster diastatops", "length" => "60-65", "url" => "http://www.americaninsects.net/d/cordulegaster-diastatops.html");
     $records[] = array("sciname" => "Cordulegaster bilineata", "length" => "60-65", "url" => "http://www.americaninsects.net/d/cordulegaster-diastatops.html");
     $records[] = array("sciname" => "Enallagma cyathigerum", "length" => "29-40", "url" => "http://americaninsects.net//d/enallagma-cyathigerum.html");
     $records[] = array("sciname" => "Enallagma boreale", "length" => "28-36", "url" => "http://americaninsects.net//d/enallagma-cyathigerum.html");
     foreach ($records as $rec) {
         $r = array();
         $r["sciname"] = Functions::canonical_form($rec["sciname"]);
         $r["taxon_id"] = str_replace(" ", "_", $r["sciname"]);
         $r["source"] = $rec["url"];
         self::create_instances_from_taxon_object($r);
         $r["lengths"] = array($rec["length"]);
         self::prepare_length_structured_data($r);
     }
 }

//.........这里部分代码省略.........
         return "Annelida";
     } elseif ($string == "Other Octocorals") {
         return "Octocorallia";
     } elseif ($string == "Other Sea Cucumbers") {
         return "Echinodermata";
     } elseif ($string == "Checkered Snapper Lutjanus decussatus") {
         return "Lutjanus decussatus";
     } elseif ($string == "Other Sea Anemones") {
         return "Actiniaria";
     } elseif ($string == "Blotched Goby Coryphopterus inframaculatus") {
         return "Coryphopterus inframaculatus";
     } elseif ($string == "One-spot Snapper Lutjanus monostigma") {
         return "Lutjanus monostigma";
     } elseif ($string == "Lined Fang Blenny; Meiacanthus-lineatus") {
         return "Meiacanthus lineatus";
     } elseif ($string == "Reef Octopus encounter May 2010") {
         return "Octopus";
     } elseif ($string == "Comb Jellys; Phyllum Ctenophora") {
         return "Ctenophora";
     } elseif ($string == "Stony Coral Heteropsammia eupammides") {
         return "Heteropsammia eupammides";
     } elseif ($string == "Stony Coral Balanophyllia elegans") {
         return "Balanophyllia elegans";
     } elseif ($string == "Giant clam with Montipora coral surrounding.") {
         return "Acropora";
     } elseif ($string == "bakaruda-n-bigeyes") {
         return "Sphyraena";
     } elseif ($string == "Bandfish Acanthocepola species") {
         return "Acanthocepola";
     } elseif ($string == "Snakefish Trachinocephalus myops") {
         return "Trachinocephalus myops";
     } elseif ($string == "Deep bodied Silverbelly Gerres erythrourus") {
         return "Gerres erythrourus";
     } elseif ($string == "Ponyfish Leiognathus fasciatus") {
         return "Leiognathus fasciatus";
     } elseif ($string == "Parapercis species we think") {
         return "Parapercis";
     } elseif ($string == "Moorish Idol Zanclus cornutus") {
         return "Zanclus cornutus";
     } elseif ($string == "Large Shrimp Goby Cryptocentrus species") {
         return "Cryptocentrus";
     } elseif ($string == "Green Shrimp Goby Cryptocentrus octafasciatus") {
         return "Cryptocentrus octafasciatus";
     } elseif (in_array($string, array("Flagfin Shrimp Goby Mahidolia mystacina with Shrimp", "Flagfin Shrimp Goby Mahidolia mystacina"))) {
         return "Mahidolia mystacina";
     } elseif (in_array($string, array("blenny-head", "blenny-head-2", "blenny-head-3", "blenny-upper-body"))) {
         return "Blenniidae";
     } elseif (in_array($string, array("false-cleanerfish-2", "false-cleanerfish-aspidontus-taeniatus"))) {
         return "Aspidontus taeniatus";
     } elseif ($string == "blenny-omobranchus-sp-uuh") {
         return "Omobranchus";
     } elseif (in_array($string, array("rabbitfish-juve-demise", "rabbitfish-juves", "rabbitfish-juves-2", "rabbitfish-juves-en-masse"))) {
         return "Siganus";
     } elseif (in_array($string, array("Unidentified Goby", "Unidentified Goby tail", "Shrimp Gobies unidentified", "Shrimp Goby unidentified", "Drab Shrimp Goby with Shrimp"))) {
         return "Gobiidae";
     } elseif (in_array($string, array("pallid-pipefish-uuh", "slender-pipefish-2"))) {
         return "Syngnathinae";
     } elseif (in_array($string, array("squirrelfish-red-1", "squirrelfish-red-3"))) {
         return "Holocentrus";
     } elseif ($string == "giant-trevally-caranx-ignobilis") {
         return "Caranx ignobilis";
     } elseif ($string == "trevally-fringe-finned-pantolobus-radiatus") {
         return "Pantolobus radiatus";
     } elseif ($string == "trevally-golden-juv-gnathanodon-speciosus") {
         return "Gnathanodon speciosus";
     } elseif ($string == "trevally-fringe-finned-2") {
         return "Caranx";
     } elseif ($string == "slender-pipefish-trachyrhampus-longirostris-1") {
         return "Trachyrhampus longirostris";
     } elseif ($string == "Platycephalus-caeruleopunctatus") {
         return "Platycephalus caeruleopunctatus";
     } elseif (in_array($string, array("Unidentified Juvenile", "Small juvenile fish with Sea Pen"))) {
         return "";
     } elseif ($string == "Unidentified Hard Coral") {
         return "Scleractinia";
     } elseif (in_array($string, array("Unidentified bivalve Mollusc", "Bivalve with encrusting Sponge and Barnacles", "Large Mussel, free standing out on the sand.", "Unidentified Clam; small", "Unidentified Clam"))) {
         return "Bivalvia";
     } elseif (in_array($string, array("Scythozoa; other Jellyfish", "Pelagidae; Sea nettles", "Ellisellidae fam; Wire Coral", "Pennatulacea; Sea Pens", "Gorgonacea; Sea Fans/Whip Corals", "Ellisellidae; Sea Fans.", "Echinigorgia; Sea Fans", "Nephtheidae; more Soft corals", "Alcyoniidae; some Soft Corals", "Heteractis Anemone, unknown species", "Anthopleura species, perhaps", "Dardanus lagapodes; Hermit Crab", "Dardanus megistos; Hermit Crab", "Oceanapia sagittaria; Porifera."))) {
         $temp = explode(";", $string);
         if (count($temp) == 1) {
             $temp = explode(",", $string);
         }
         if ($val = @$temp[0]) {
             return Functions::canonical_form($val);
         } else {
             return "";
         }
     } else {
         $string = trim(str_ireplace(" family", "", $string));
         $temp = explode(";", $string);
         if (count($temp) == 1) {
             $temp = explode(",", $string);
         }
         if ($val = @$temp[1]) {
             return Functions::canonical_form($val);
         } else {
             return Functions::canonical_form($temp[0]);
         }
     }
 }

 private static function evaluate_scientific_name($scientific_name)
 {
     if (preg_match_all("/<i>(.*?)<\\/i>/i", trim($scientific_name), $matches, PREG_SET_ORDER)) {
         $canonical_form = trim($matches[0][1]);
         if (@$matches[1]) {
             $canonical_form .= " " . trim($matches[1][1]);
         }
         if (@$matches[2]) {
             $canonical_form .= " " . trim($matches[2][1]);
         }
         $canonical_form = str_replace("ssp. ", "", $canonical_form);
         $scientific_name = str_ireplace("<i>", "", $scientific_name);
         $scientific_name = str_ireplace("</i>", "", $scientific_name);
         $scientific_name = str_replace("  ", " ", $scientific_name);
         $scientific_name = str_ireplace("&rsquo;", "'", $scientific_name);
     }
     if (@(!$canonical_form)) {
         $canonical_form = Functions::canonical_form($scientific_name);
     }
     $taxon_id = str_replace(" ", "_", strtolower($canonical_form));
     return array($scientific_name, $canonical_form, $taxon_id);
 }

 private function get_taxon_id($name)
 {
     if (is_numeric(stripos($name, " sp"))) {
         return str_ireplace(" ", "_", $name);
     } else {
         return str_ireplace(" ", "_", Functions::canonical_form($name));
     }
 }

 private function name_exists_in_eol($name)
 {
     $eol_api = "http://eol.org/api/search/1.0.json?exact=true&q=";
     if ($json = Functions::lookup_with_cache($eol_api . $name, $this->download_options)) {
         $taxon = json_decode($json, true);
         if (intval($taxon["totalResults"]) > 0) {
             return Functions::canonical_form($taxon["results"][0]["title"]);
         } else {
             return false;
         }
     }
 }