function get_all_taxa()
{
$this->institution_uris = self::get_institution_uris();
self::load_zip_contents();
print_r($this->text_path);
require_library('connectors/FishBaseAPI');
$func = new FishBaseAPI();
$fields = array("lngSpeciesSenior_ID", "lngSpecies_ID", "lngRepository_ID", "strCatNr", "lngPrepMeth_ID", "lngPersPrep_ID", "lngPersID2_ID", "lngPersID3_ID", "bytCountPrep", "lngDocuTypeSpecimen", "lngPrep_ID", "txtPrepNotes", "lngSpecimen_ID", "strTypeStat", "lngLoc_ID", "strLocName", "txtLocDescr", "sngElevation", "sngLatitudeWGS84", "sngLongitudeWGS84", "lngMacrohabitat_ID", "lngPermanency_ID", "strDepth", "lngGeology_ID", "lngSpecLocal_ID");
$records_specimen = $func->make_array($this->text_path["sd_specimen"], $fields, "", array());
array_shift($records_specimen);
$fields = array("lngPersIdent_ID", "lngSpecimen_ID");
$records_specimen_identifier = $func->make_array($this->text_path["sd_specimen_identifier"], $fields, "", array());
array_shift($records_specimen_identifier);
$fields = array("lngSpeciesSenior_ID", "lngMacroMicro_ID", "lngSpecies_ID");
$records_habitat = $func->make_array($this->text_path["sd_macro_locality"], $fields, "", array());
array_shift($records_habitat);
/*
$fields = array("lngLoc_ID", "strLocName", "txtLocDescr", "sngElevation", "sngLatitudeWGS84", "sngLongitudeWGS84", "lngMacrohabitat_ID", "lngPermanency_ID", "strDepth", "lngGeology_ID", "lngSpecLocal_ID", "lngSpecimen_ID", "lngSpeciesSenior_ID", "lngSpecies_ID");
$records_locality = $func->make_array($this->text_path["sd_locality"], $fields, "", array());
array_shift($records_locality);
*/
$this->assemble_identified_by($records_specimen_identifier);
$this->create_instances_from_taxon_object($records_specimen, "specimen");
// $this->create_instances_from_taxon_object($records_locality, "locality");
$this->create_instances_from_taxon_object($records_habitat, "habitat");
$this->create_archive();
// remove temp dir
$path = $this->text_path["sd_specimen"];
$parts = pathinfo($path);
$parts["dirname"] = str_ireplace("/type_specimen", "", $parts["dirname"]);
recursive_rmdir($parts["dirname"]);
debug("\n temporary directory removed: " . $parts["dirname"]);
}
private function get_worms_taxa()
{
require_library('connectors/FishBaseAPI');
$func = new FishBaseAPI();
$fields = array();
$excluded_fields = array("taxonID", "furtherInformationURL", "referenceID", "namePublishedIn", "taxonomicStatus", "taxonRemarks", "rightsHolder", "parentNameUsageID");
$taxa = $func->make_array($this->text_path["worms"]["worms_taxon"], $fields, "", $excluded_fields);
array_shift($taxa);
foreach ($taxa as $taxon) {
$final[Functions::canonical_form($taxon["scientificName"])] = $taxon["taxonRank"];
}
unset($taxa);
return $final;
}
private function process_text_files()
{
if (!self::load_zip_contents()) {
return false;
}
print_r($this->text_path);
require_library('connectors/FishBaseAPI');
$func = new FishBaseAPI();
$fields = array("IDNUM", "TAXONID", "TAXID1", "SUBORDER", "TAXID2", "DIVISION", "TAXID5", "SUPERFAM", "TAXID9", "FAMILY", "TAXID13", "SUBFAM", "TAXID17", "TRIBE", "TAXID20", "SUBTRIBE", "GENUS", "SUBGENUS", "SPECIES", "SUBSPECS", "RECTYPCD", "RECTYPE", "UPDATE", "UPDUSER", "LOCGENUS", "LOCSUBGN", "SPCGROUP", "LOCSPECS", "LOCSUBSP", "INFRASUB", "CAUTHOR", "ORIGAUTH", "AUTHOR", "PUBYEAR", "JOURNAL", "COLLATN", "INTROADV", "TYPEDEP", "TYPMATTX", "TYPELOC", "SEXTX", "TSGENUS", "TSSUBGEN", "TSSPECS", "TSSUBSP", "TSAUTHOR", "TSKIND", "INFRATYP", "LOCVAR", "SEQUENCE", "UPSTAT");
$taxa = $func->make_array($this->text_path[1], $fields, "", array(), "^");
/* to get the breakdown of taxa levels
foreach($taxa as $rec)
{
if(@$temp[$rec["RECTYPE"]]) $temp[$rec["RECTYPE"]]++;
else $temp[$rec["RECTYPE"]] = 1;
}
print_r($temp);
// [Super-Generic ] => 614 xxx
// [Genus/Subgenus ] => 2540 xxx
// [Genus/Subgenus Synon] => 3459
// [Species ] => 19970 xxx
// [Species Synonym ] => 11083
// [Unplaced Taxon ] => 445 xxx
// [Nomen Nudum ] => 177
// [Species Group ] => 440 xxx
*/
$i = 0;
$link = array();
foreach ($taxa as $rec) {
if (in_array($rec["RECTYPE"], array("Species", "Genus/Subgenus", "Species Group", "Super-Generic"))) {
$i++;
$sciname = $rec["GENUS"] . " " . $rec["SPECIES"];
$sciname = trim($sciname);
if ($rec["RECTYPE"] == "Species Group") {
$sciname = $rec["SPCGROUP"];
}
$temp[$sciname] = 1;
$link = $this->create_instances_from_taxon_object($rec, $link);
}
}
echo "\n\n total rows: " . count($taxa);
echo "\n total taxa: " . count($temp);
echo "\n link: " . count($link) . "\n";
$free_text = self::process_free_text($func, $link);
$geo_data = self::process_geo_data($link);
$ecology = self::process_ecology($func, $link);
$pollen_visitation = self::process_pollen_visitation($func, $link);
$predation = self::process_predation($func, $link);
return true;
/* Waiting from SPG on items below this line */
// some sort of references
// $fields = array("IDNUM", "REVSNTX", "TAXNTX");
// $references = $func->make_array($this->text_path[4], $fields, "", array(), "^");
// references for biology and morphology
// $fields = array("IDNUM", "BIOLGYTX", "MORPHTX");
// $texts = $func->make_array($this->text_path[5], $fields, "", array(), "^");
// $ref_ids = array();
// $agent_ids = array();
// foreach($texts as $text)
// {
// $biology = $text["BIOLGYTX"];
// $morphology = $text["MORPHTX"];
// $taxon_id = @$link[$text["IDNUM"]];
// if($biology) $ref_ids = array_merge($ref_ids, get_object_reference_ids($biology));
// if($morphology) $ref_ids = array_merge($ref_ids, get_object_reference_ids($morphology));
// }
// comma-separated taxon names
// $fields = array("IDNUM", "AHOSTTX", "PHOSTTX");
// $taxon_remarks = $func->make_array($this->text_path[7], $fields, "", array(), "^");
// comma-separated taxon names
// $fields = array("IDNUM", "PARATX", "SECHOSTX");
// $taxon_remarks2 = $func->make_array($this->text_path[8], $fields, "", array(), "^");
// wait for SPG to classify
// $fields = array("IDNUM", "HYPATX", "SYNTX");
// $comments = $func->make_array($this->text_path[10], $fields, "", array(), "^");
}
cs.conservation_status_title, cs.conservation_status_title_sp, cs.conservation_status_abbrev
From nam_species s
LEFT JOIN nam_genus g ON s.genus_id = g.genus_id
LEFT Join nam_family f ON g.family_id = f.Family_ID
LEFT Join nam_orders o ON f.order_id = o.order_id
LEFT Join nam_conservation_status cs ON s.conservation_status_id = cs.id";
We've requested the partner to provide us with just a text dump of the result of the query above, but it seems they'll just
continue providing us with the Access MDB.
*/
$remote_file = "https://dl.dropboxusercontent.com/u/7597512/NorthAmericanMammals/data_from_sql_export.txt";
// $text_file = DOC_ROOT . "/update_resources/connectors/files/NorthAmericanMammals/data_from_sql_export.txt";
$text_file = Functions::save_remote_file_to_local($remote_file, array('download_wait_time' => 1000000, 'timeout' => 600));
require_library('connectors/FishBaseAPI');
$fields = array("species_id", "genus_name", "sci_name", "family_name", "order_name", "avg_length", "avg_length_sp", "range_length", "range_length_sp", "avg_weight", "avg_weight_sp", "range_weight", "range_weight_sp", "conservation_status_notes", "conservation_status_notes_sp", "common_name", "common_name_sp", "other_names", "other_names_sp", "refs", "refs_sp", "links", "links_sp", "dimorphism", "dimorphism_sp", "legend", "legend_sp", "refs(2)", "refs_sp(2)", "adaptation", "adaptation_sp", "conservation_status_id", "conservation_status_title", "conservation_status_title_sp", "conservation_status_abbrev");
$taxa = FishBaseAPI::make_array($text_file, $fields, "", array());
$resource_id = 85;
//for North American Mammals
$schema_taxa = array();
$used_taxa = array();
$ctr = 0;
foreach ($taxa as $row) {
$ctr++;
print "{$ctr} - ";
$dwc_Kingdom = "Animalia";
$dwc_Order = trim($row["order_name"]);
$dwc_Family = trim($row["family_name"]);
$dwc_Genus = trim($row["genus_name"]);
$dwc_ScientificName = trim($row["sci_name"]);
$taxon_identifier = "NAM_" . $row["species_id"];
if (@$used_taxa[$taxon_identifier]) {
function save_before_site_goes_dark()
{
if (!self::load_zip_contents()) {
return FALSE;
}
$names_no_entry_from_partner = self::get_names_no_entry_from_partner();
require_library('connectors/FishBaseAPI');
$func = new FishBaseAPI();
$fields = $this->media_fields;
$records = $func->make_array($this->text_path["media"], $fields);
$total = count($records);
$i = 0;
$folder = $this->local_destination . "_15";
if (!file_exists($folder)) {
mkdir($folder);
}
foreach ($records as $rec) {
$i++;
echo "\n {$i} of {$total} \n";
// if($i <= 88325) continue; //debug 4,709
$parts = pathinfo($rec["accessURI"]);
$destination = $folder . "/" . $parts["basename"];
// if(!file_exists($destination))
if (!self::image_exists($parts["basename"])) {
if (in_array($rec["accessURI"], $names_no_entry_from_partner)) {
continue;
}
// [$param] -> name_no_entry_from_partner";
echo "- does not exist ";
echo "\n -- " . $rec["accessURI"] . " processing... \n";
$filenames = glob($folder . "/" . "*.*");
if (count($filenames) >= 5000) {
$arr = explode("_", $folder);
$count = intval($arr[1]) + 1;
$folder = $this->local_destination . "_" . $count . "/";
if (!file_exists($folder)) {
mkdir($folder);
}
print "\n saving 111...";
echo "\n destination: [{$destination}] ";
self::save_big_file_to_local($rec["accessURI"], $destination);
} else {
print "\n saving 222...";
echo "\n destination: [{$destination}] ";
self::save_big_file_to_local($rec["accessURI"], $destination);
}
} else {
echo " - already exists";
}
}
}
function some_stats()
{
require_library('connectors/FishBaseAPI');
$func = new FishBaseAPI();
$fields = array("identifier", "full_reference");
$path = DOC_ROOT . "applications/content_server/resources/660/reference.tab";
$records = $func->make_array($path, $fields, "", array());
foreach ($records as $rec) {
print_r($rec);
}
}