public static function begin_concept_assignment($hierarchy_id = null, $use_synonyms_for_merging = false)
{
$mysqli =& $GLOBALS['mysqli_connection'];
if (!defined('SOLR_SERVER') || !SolrAPI::ping(SOLR_SERVER, 'hierarchy_entries')) {
debug("ERROR: No Solr server; cannot begin_concept_assignment");
return false;
}
// looking up which hierarchies might have nodes in preview mode
// this will save time later on when we need to check published vs preview taxa
self::lookup_preview_harvests();
$confirmed_exclusions = self::check_curator_assertions();
$hierarchies_compared = array();
if ($default_id = Hierarchy::default_id()) {
$default_hierarchy = Hierarchy::find($default_id);
// TODO: This should NOT be hard-coded. ...particularly without an
// explanation. :| ...And isn't this just ... undone as soon as we
// leave this block by "$hierarchy_lookup_ids2 = array();"? ...This
// seems stupid.
if (@$default_hierarchy->id) {
$hierarchy_lookup_ids2 = array($default_id => 1475377);
}
}
$hierarchy_lookup_ids2 = array();
// TODO: This is a SLOW, huge query. If we had a denormalized HE count
// on each hierarchy (and, truly, we should), it would be super-fast. Do
// this.
$result = $mysqli->query("SELECT id ,hierarchy_entries_count as count FROM hierarchies ");
while ($result && ($row = $result->fetch_assoc())) {
// TODO: DON'T hard-code this (this is GBIF Nub Taxonomy).
// Instead, add an attribute to hierarchies called "no_comparisons"
// and check that, and make sure the migration for that sets the
// value of that field to true for all hierarchies with a ).
// Also make sure curators can set that value from the resource
// page.
if ($row['id'] == 129) {
continue;
}
$hierarchy_lookup_ids2[$row['id']] = $row['count'];
}
// TODO: This is, again, a GBIF Nub taxonomy:
$hierarchy_lookup_ids2[800] = 1;
arsort($hierarchy_lookup_ids2);
// if the function is passed a hierarchy_id then make the first loop
// just that hierarchy otherwise make the first loop the same as the
// inner loop - compare everything with everything else
if ($hierarchy_id) {
$hierarchy1 = Hierarchy::find($hierarchy_id);
// TODO: Don't we have this in $hierarchy_lookup_ids2 ?
$count1 = $hierarchy1->count_entries();
$hierarchy_lookup_ids1[$hierarchy_id] = $count1;
} else {
$hierarchy_lookup_ids1 = $hierarchy_lookup_ids2;
}
foreach ($hierarchy_lookup_ids1 as $id1 => $count1) {
$hierarchy1 = Hierarchy::find($id1);
if (@(!$hierarchy1->id)) {
debug("ERROR: Attempt to compare hierarchy {$id1}, but it was missing");
continue;
}
foreach ($hierarchy_lookup_ids2 as $id2 => $count2) {
$hierarchy2 = Hierarchy::find($id2);
if (@(!$hierarchy2->id)) {
debug("WARNING: Skipping compare of missing hierarchy {$id2}");
continue;
}
// already compared - skip
if (isset($hierarchies_compared[$hierarchy1->id][$hierarchy2->id])) {
continue;
}
debug("Comparing hierarchy {$id1} ({$hierarchy1->label}; {$count1} entries) to {$id2} ({$hierarchy2->label}; {$count2} entries)");
// have the smaller hierarchy as the first parameter so the
// comparison will be quicker
if ($count1 < $count2) {
self::assign_concepts_across_hierarchies($hierarchy1, $hierarchy2, $confirmed_exclusions, $use_synonyms_for_merging);
} else {
self::assign_concepts_across_hierarchies($hierarchy2, $hierarchy1, $confirmed_exclusions, $use_synonyms_for_merging);
}
$hierarchies_compared[$hierarchy1->id][$hierarchy2->id] = 1;
$hierarchies_compared[$hierarchy2->id][$hierarchy1->id] = 1;
}
}
}
function get_count_of_taxa_pages_per_partner($user_id)
{
$totals = array();
if ($user_id == $this->bhl_user_id) {
$bhl_file_path = PAGE_METRICS_TEXT_PATH . "taxon_concept_with_bhl_links.txt";
$bhl_concept_ids = array();
foreach (new FileIterator($bhl_file_path) as $line_number => $line) {
$fields = explode("\t", $line);
$taxon_concept_id = trim($fields[0]);
$bhl_concept_ids[$taxon_concept_id] = 1;
}
$totals[] = count($bhl_concept_ids);
} elseif ($user_id == $this->col_user_id) {
$result = $this->mysqli->query("SELECT COUNT(taxon_concept_id) count FROM hierarchy_entries WHERE hierarchy_id = " . Hierarchy::default_id());
$row = $result->fetch_row();
$totals[] = $row[0];
} else {
$concept_ids = array();
$query = "SELECT r.* FROM resources r JOIN content_partners cp ON r.content_partner_id = cp.id WHERE cp.user_id = {$user_id}";
$resources = $this->mysqli->query($query);
while ($resources && ($resource_row = $resources->fetch_assoc())) {
$query = "SELECT MAX(he.id) latest_harvest_event_id FROM harvest_events he\n WHERE he.resource_id = " . $resource_row['id'] . " AND he.published_at IS NOT NULL AND he.published_at < '" . $this->year . "-" . ($this->month + 1) . "-01'";
$result = $this->mysqli->query($query);
$row = $result->fetch_row();
$latest_harvest_event_id = $row[0];
if (!$latest_harvest_event_id) {
continue;
}
$query = "SELECT tc.id taxon_concept_id\n FROM harvest_events_hierarchy_entries hehe\n STRAIGHT_JOIN hierarchy_entries he ON hehe.hierarchy_entry_id = he.id\n STRAIGHT_JOIN taxon_concepts tc ON he.taxon_concept_id = tc.id\n WHERE tc.published=1 AND tc.supercedure_id=0 AND hehe.harvest_event_id={$latest_harvest_event_id}";
$result = $this->mysqli->query($query);
while ($result && ($row = $result->fetch_assoc())) {
$taxon_concept_id = $row['taxon_concept_id'];
$concept_ids[$taxon_concept_id] = '';
}
}
$totals[] = count($concept_ids);
}
$query = "SELECT COUNT(gapt.taxon_concept_id) FROM google_analytics_partner_taxa gapt WHERE gapt.user_id = {$user_id} AND gapt.`year` = {$this->year} AND gapt.`month` = {$this->month}";
$result = $this->mysqli->query($query);
$row = $result->fetch_row();
$totals[] = $row[0];
// $totals[0] => total taxa for user_id
// $totals[1] => taxa viewed for user_id this month
return $totals;
}
