private function get_page_count_from_BHL_csv($contents)
{
$temp_path = temp_filepath();
if ($contents) {
if (!($file = Functions::file_open($temp_path, "w"))) {
return;
}
fwrite($file, $contents);
fclose($file);
}
$page_ids = array();
$i = 0;
if (!($file = Functions::file_open($temp_path, "r"))) {
return;
}
while (!feof($file)) {
$i++;
if ($i == 1) {
$fields = fgetcsv($file);
} else {
$rec = array();
$temp = fgetcsv($file);
$k = 0;
if (!$temp) {
continue;
}
foreach ($temp as $t) {
$rec[$fields[$k]] = $t;
$k++;
}
$parts = pathinfo($rec["Url"]);
$page_ids[$parts["filename"]] = '';
}
}
fclose($file);
unlink($temp_path);
return count(array_keys($page_ids));
}
public static function update_taxon_concept_names($taxon_concept_ids)
{
if (!$taxon_concept_ids) {
return false;
}
if (is_numeric($taxon_concept_ids)) {
$taxon_concept_ids = array($taxon_concept_ids);
}
$mysqli =& $GLOBALS['db_connection'];
$started_new_transaction = false;
if (!$mysqli->in_transaction()) {
$mysqli->begin_transaction();
$started_new_transaction = true;
}
$batches = array_chunk($taxon_concept_ids, 500);
foreach ($batches as $batch_ids) {
usleep(500000);
$name_ids = array();
$matching_ids = array();
$query = "\n (SELECT he.taxon_concept_id, he.id, he.name_id, 'preferred' as type FROM hierarchy_entries he WHERE taxon_concept_id IN (" . implode(",", $batch_ids) . ") AND ((he.published=1 AND he.visibility_id=" . Visibility::visible()->id . ") OR (he.published=0 AND he.visibility_id=" . Visibility::preview()->id . ")))\n UNION\n (SELECT he.taxon_concept_id, s.hierarchy_entry_id, s.name_id, 'synonym' as type\n FROM hierarchy_entries he\n JOIN synonyms s ON (he.id=s.hierarchy_entry_id)\n WHERE he.taxon_concept_id IN (" . implode(",", $batch_ids) . ")\n AND s.language_id=0\n AND s.synonym_relation_id!=" . SynonymRelation::find_or_create_by_translated_label('genbank common name')->id . "\n AND s.synonym_relation_id!=" . SynonymRelation::find_or_create_by_translated_label('common name')->id . "\n AND s.synonym_relation_id!=" . SynonymRelation::find_or_create_by_translated_label('blast name')->id . "\n AND s.synonym_relation_id!=" . SynonymRelation::find_or_create_by_translated_label('genbank acronym')->id . "\n AND s.synonym_relation_id!=" . SynonymRelation::find_or_create_by_translated_label('acronym')->id . "\n AND ((he.published=1 AND he.visibility_id=" . Visibility::visible()->id . ") OR (he.published=0 AND he.visibility_id=" . Visibility::preview()->id . ")))";
foreach ($mysqli->iterate_file($query) as $row_num => $row) {
$taxon_concept_id = $row[0];
$hierarchy_entry_id = $row[1];
$name_id = $row[2];
$name_type = $row[3];
$name_ids[$name_id][$taxon_concept_id] = 1;
$matching_ids[$taxon_concept_id][$name_id][$hierarchy_entry_id] = $name_type;
}
if ($name_ids) {
//This makes sure we have a scientific name, gets the canonicalFormID
$query = "SELECT n.id, n_match.id FROM names n JOIN canonical_forms cf ON (n.canonical_form_id=cf.id) JOIN names n_match ON (cf.id=n_match.canonical_form_id) WHERE n.id IN (" . implode(",", array_keys($name_ids)) . ") AND n_match.string=cf.string";
foreach ($mysqli->iterate_file($query) as $row_num => $row) {
$original_name_id = $row[0];
$canonical_name_id = $row[1];
if ($original_name_id != $canonical_name_id) {
foreach ($name_ids[$original_name_id] as $taxon_concept_id => $junk) {
$matching_ids[$taxon_concept_id][$canonical_name_id][0] = 1;
}
}
}
}
$common_names = array();
$preferred_in_language = array();
$query = "SELECT he.taxon_concept_id, he.published, he.visibility_id, s.id, s.hierarchy_id, s.hierarchy_entry_id, s.name_id, s.language_id, s.preferred, s.vetted_id FROM hierarchy_entries he JOIN synonyms s ON (he.id=s.hierarchy_entry_id) JOIN vetted v ON (s.vetted_id=v.id) WHERE he.taxon_concept_id IN (" . implode(",", $batch_ids) . ") AND s.language_id!=0 AND (s.synonym_relation_id=" . SynonymRelation::genbank_common_name()->id . " OR s.synonym_relation_id=" . SynonymRelation::common_name()->id . ") ORDER BY s.language_id, (s.hierarchy_id=" . Hierarchy::contributors()->id . ") DESC, v.view_order ASC, s.preferred DESC, s.id DESC";
foreach ($mysqli->iterate_file($query) as $row_num => $row) {
$taxon_concept_id = $row[0];
$published = $row[1];
$visibility_id = $row[2];
$synonym_id = $row[3];
$hierarchy_id = $row[4];
$hierarchy_entry_id = $row[5];
$name_id = $row[6];
$language_id = $row[7];
$preferred = $row[8];
$vetted_id = $row[9];
// skipping Wikipedia common names entirely
if ($hierarchy_id == @Hierarchy::wikipedia()->id) {
continue;
}
$curator_name = $hierarchy_id == @Hierarchy::contributors()->id;
$ubio_name = $hierarchy_id == @Hierarchy::ubio()->id;
if ($curator_name || $ubio_name || $curator_name || $published == 1 && $visibility_id == Visibility::visible()->id) {
if (isset($preferred_in_language[$taxon_concept_id][$language_id])) {
$preferred = 0;
}
if ($preferred && $curator_name && ($vetted_id == Vetted::trusted()->id || $vetted_id == Vetted::unknown()->id)) {
$preferred_in_language[$taxon_concept_id][$language_id] = 1;
} else {
$preferred = 0;
}
if (!isset($common_names[$taxon_concept_id])) {
$common_names[$taxon_concept_id] = array();
}
$common_names[$taxon_concept_id][] = array('synonym_id' => $synonym_id, 'language_id' => $language_id, 'name_id' => $name_id, 'hierarchy_entry_id' => $hierarchy_entry_id, 'preferred' => $preferred, 'vetted_id' => $vetted_id, 'is_curator_name' => $curator_name);
}
}
// if there was no preferred name
foreach ($common_names as $taxon_concept_id => $arr) {
foreach ($arr as $key => $arr2) {
if (@(!$preferred_in_language[$taxon_concept_id][$arr2['language_id']]) && ($arr2['vetted_id'] == Vetted::trusted()->id || $arr2['vetted_id'] == Vetted::unknown()->id)) {
$common_names[$taxon_concept_id][$key]['preferred'] = 1;
$preferred_in_language[$taxon_concept_id][$arr2['language_id']] = 1;
}
}
}
$mysqli->delete("DELETE FROM taxon_concept_names WHERE taxon_concept_id IN (" . implode(",", $batch_ids) . ")");
$tmp_file_path = temp_filepath();
if (!($LOAD_DATA_TEMP = fopen($tmp_file_path, "w+"))) {
debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $tmp_file_path);
return;
}
/* Insert the scientific names */
foreach ($matching_ids as $taxon_concept_id => $arr) {
foreach ($arr as $name_id => $arr2) {
foreach ($arr2 as $hierarchy_entry_id => $type) {
$preferred = 0;
if ($hierarchy_entry_id && $type == "preferred") {
$preferred = 1;
}
fwrite($LOAD_DATA_TEMP, "{$taxon_concept_id}\t{$name_id}\t{$hierarchy_entry_id}\t0\t0\t{$preferred}\n");
}
}
}
$mysqli->load_data_infile($tmp_file_path, 'taxon_concept_names');
unlink($tmp_file_path);
$tmp_file_path = temp_filepath();
if (!($LOAD_DATA_TEMP = fopen($tmp_file_path, "w+"))) {
debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $tmp_file_path);
return;
}
/* Insert the common names */
foreach ($common_names as $taxon_concept_id => $arr) {
foreach ($arr as $key => $arr2) {
$synonym_id = $arr2['synonym_id'];
$language_id = $arr2['language_id'];
$name_id = $arr2['name_id'];
$hierarchy_entry_id = $arr2['hierarchy_entry_id'];
$preferred = $arr2['preferred'];
$vetted_id = $arr2['vetted_id'];
fwrite($LOAD_DATA_TEMP, "{$taxon_concept_id}\t{$name_id}\t{$hierarchy_entry_id}\t{$language_id}\t1\t{$preferred}\t{$synonym_id}\t{$vetted_id}\n");
}
}
$mysqli->load_data_infile($tmp_file_path, 'taxon_concept_names');
unlink($tmp_file_path);
unset($matching_ids);
unset($common_names);
unset($name_ids);
unset($preferred_in_language);
$mysqli->commit();
}
if ($started_new_transaction) {
$mysqli->end_transaction();
}
}
function write_partner_summaries($stats)
{
$outfile = temp_filepath();
if (!($OUT = fopen($outfile, "w+"))) {
debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $outfile);
return;
}
fwrite($OUT, implode("\t", $stats) . "\n");
fclose($OUT);
$this->mysqli->load_data_infile($outfile, 'google_analytics_partner_summaries');
unlink($outfile);
}
public static function save_remote_file_to_local($url, $options = array())
{
if (!isset($options['download_wait_time'])) {
$options['download_wait_time'] = DOWNLOAD_WAIT_TIME;
}
if (!isset($options['timeout'])) {
$options['timeout'] = DOWNLOAD_TIMEOUT_SECONDS;
}
if (!isset($options['download_attempts'])) {
$options['download_attempts'] = DOWNLOAD_ATTEMPTS;
}
$temp_path = temp_filepath();
if (isset($options['file_extension'])) {
$temp_path .= "." . $options['file_extension'];
}
debug("\n\n Saving remote file: " . $url);
debug("\n\n Temporary file: " . $temp_path);
if (@$options['cache']) {
$file_contents = self::lookup_with_cache($url, $options);
} else {
$file_contents = self::get_remote_file($url, $options);
}
if ($file_contents) {
if (!($file = Functions::file_open($temp_path, "w"))) {
return;
}
fwrite($file, $file_contents);
fclose($file);
return $temp_path;
}
return false;
}
public function convert_to_new_schema_archive()
{
// previous problems reading spreadsheet
if ($this->errors || @(!$this->spreadsheet_reader)) {
return false;
}
if (!$this->is_new_schema_spreadsheet()) {
return false;
}
$archive_temp_directory_path = $this->output_directory();
// fail if for some reason there is no valid output directory
if ($archive_temp_directory_path === null) {
return false;
}
$sheet_names = $this->spreadsheet_reader->getSheetNames();
$worksheet_fields = array();
// loop through all the worksheets in the file
foreach ($sheet_names as $sheet_index => $sheet_name) {
if ($sheet_name == "controlled terms") {
continue;
}
$worksheet_reader = $this->spreadsheet_reader->setActiveSheetIndex($sheet_index);
$worksheetTitle = $worksheet_reader->getTitle();
$highest_row = $worksheet_reader->getHighestRow();
// e.g. 10
$highest_column = $worksheet_reader->getHighestColumn();
// e.g 'F'
$highest_column_index = \PHPExcel_Cell::columnIndexFromString($highest_column);
$number_of_columns = ord($highest_column) - 64;
if (!($OUTFILE = fopen($archive_temp_directory_path . "/{$sheet_name}.txt", "w+"))) {
debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $archive_temp_directory_path . "/{$sheet_name}.txt");
return;
}
$worksheet_fields[$sheet_name] = array();
for ($row_index = 1; $row_index <= $highest_row; $row_index++) {
static $i = 0;
$i++;
// if($i % 100 == 0) echo "$i - ".time_elapsed()."\n";
$values = array();
for ($column_index = 0; $column_index < $highest_column_index; $column_index++) {
$cell = $worksheet_reader->getCellByColumnAndRow($column_index, $row_index, true);
if ($cell === null) {
$value = null;
} else {
$value = self::prepare_value($cell->getCalculatedValue());
}
/*
Row1: readable label
Row2: field type URI
Row3: required
Row4: foreign key worksheet
Row5: extension group label
Row6: extension thesaurus URI
Row7: definition
Row8: comment
Row9: extension thesaurus URI
*/
if ($row_index > 9) {
$value = self::fix_spreadsheet_shorthand($sheet_name, @$worksheet_fields[$sheet_name][$column_index]['uri'], $value);
$values[] = $value;
} elseif ($row_index == 1) {
$worksheet_fields[$sheet_name][$column_index]['label'] = $value;
$values[] = $value;
} elseif ($row_index == 2) {
$worksheet_fields[$sheet_name][$column_index]['uri'] = $value;
} elseif ($row_index == 3) {
$worksheet_fields[$sheet_name][$column_index]['required'] = strtolower($value);
} elseif ($row_index == 4) {
$worksheet_fields[$sheet_name][$column_index]['foreign_key'] = $value;
} elseif ($row_index == 5) {
$worksheet_fields[$sheet_name][$column_index]['group'] = $value;
} elseif ($row_index == 6) {
$worksheet_fields[$sheet_name][$column_index]['thesaurus'] = $value;
} elseif ($row_index == 7) {
$worksheet_fields[$sheet_name][$column_index]['definition'] = $value;
} elseif ($row_index == 8) {
$worksheet_fields[$sheet_name][$column_index]['comment'] = $value;
} elseif ($row_index == 9) {
$worksheet_fields[$sheet_name][$column_index]['example'] = $value;
}
}
if ($values) {
$all_empty_values = true;
foreach ($values as $value) {
if ($value) {
$all_empty_values = false;
break;
}
}
if (!$all_empty_values) {
$row = self::$field_enclosure . implode(self::$field_enclosure . self::$field_delimeter . self::$field_enclosure, $values) . self::$field_enclosure . self::$row_delimiter;
fwrite($OUTFILE, $row);
}
}
}
fclose($OUTFILE);
}
if (!($META = fopen($archive_temp_directory_path . "/meta.xml", "w+"))) {
debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $archive_temp_directory_path . "/meta.xml");
return;
}
fwrite($META, self::meta_xml_from_worksheets($worksheet_fields));
fclose($META);
$info = pathinfo($archive_temp_directory_path);
$temporary_tarball_path = temp_filepath();
$final_tarball_path = $archive_temp_directory_path . ".tar.gz";
shell_exec("tar -czf {$temporary_tarball_path} --directory=" . $info['dirname'] . "/" . $info['basename'] . " .");
@unlink($new_tarball_path);
if (copy($temporary_tarball_path, $final_tarball_path)) {
unlink($temporary_tarball_path);
}
return $archive_temp_directory_path;
}