tripal/tripal_chado/api/modules/tripal_chado.cv.api.inc

<?php
/**
 * @file
 * Provides API functions specificially for managing controlled vocabulary
 * records in Chado.
 *
 * @ingroup tripal_chado
 */

/**
 * @defgroup tripal_chado_cv_api Chado CV
 * @ingroup tripal_chado_api
 * @{
 * Provides API functions specificially for managing controlled vocabulary
 * records in Chado. Please note that Tripal v3 provides a generic set of
 * API functions for working with controlled vocabularies (CVs). This allows for
 * CVs to be stored using any back-end.  By default CV's continue to be housed
 * in Chado.  Therefore, if you are working directly with controlled vocabulary
 * records inside of a Chado-aware module then these functions can be used.
 * @}
 */

/**
 * Retrieves a chado controlled vocabulary variable
 *
 * @param $identifier
 *   An array with the key stating what the identifier is. Supported keys (only
 *   on of the following unique keys is required):
 *    - cv_id: the chado cv.cv_id primary key.
 *    - name: the chado cv.name field (assume unique).
 * @param $options
 *   An array of options. Supported keys include:
 *     - Any keys supported by chado_generate_var(). See that function
 *       definition fot additional details.
 *
 * NOTE: the $identifier parameter can really be any array similar to $values
 * passed into chado_select_record(). It should fully specify the cv record to
 * be returned.
 *
 * @return
 *   If unique values were passed in as an identifier then an object describing
 *   the cv will be returned (will be a chado variable from
 *   chado_generate_var()). Otherwise, FALSE will be returned.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_get_cv($identifiers, $options = array()) {

  // Set Defaults.
  if (!isset($options['include_fk'])) {
    // Tells chado_generate_var not to follow any foreign keys.
    $options['include_fk'] = array();
  }

  // Error Checking of parameters.
  if (!is_array($identifiers)) {
    tripal_report_error(
      'tripal_chado_api',
      TRIPAL_ERROR,
      "chado_get_cv: The identifier passed in is expected to be an array with the key
        matching a column name in the cv table (ie: cv_id or name). You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }
  elseif (empty($identifiers)) {
    tripal_report_error(
      'tripal_chado_api',
      TRIPAL_ERROR,
      "chado_get_cv: You did not pass in anything to identify the cv you want. The identifier
        is expected to be an array with the key matching a column name in the cv table
        (ie: cv_id or name). You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }

  // Try to get the cv.
  $cv = chado_generate_var(
    'cv',
    $identifiers,
    $options
  );

  // Ensure the cv is singular. If it's an array then it is not singular.
  if (is_array($cv)) {
    tripal_report_error(
      'tripal_chado_api',
      TRIPAL_ERROR,
      "chado_get_cv: The identifiers you passed in were not unique. You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }

  // Report an error if $cv is FALSE since then chado_generate_var has failed.
  elseif ($cv === FALSE) {
    tripal_report_error(
      'tripal_chado_api',
      TRIPAL_ERROR,
      "chado_get_cv: chado_generate_var() failed to return a cv based on the identifiers
        you passed in. You should check that your identifiers are correct, as well as, look
        for a chado_generate_var error for additional clues. You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }

  // Else, as far we know, everything is fine so give them their cv :)
  else {
    return $cv;
  }
}

/**
 * Create an options array to be used in a form element which provides a
 * list of all chado cvs.
 *
 * @return
 *   An array(cv_id => name) for each cv in the chado cv table.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_get_cv_select_options() {

  $results = chado_select_record('cv', array('cv_id', 'name'), array(), array('order_by' => array('name' => 'ASC')));

  $options = array();
  $options[] = 'Select a Vocabulary';

  foreach ($results as $r) {
    $options[$r->cv_id] = $r->name;
  }

  return $options;

}

/**
 * Retrieves a chado controlled vocabulary term variable.
 *
 * @param $identifier
 *   An array apropriate for use with the chado_generate_var for uniquely
 *   identifying a cvterm record. Alternativley, there are also some specially
 *   handled keys. They are:
 *    - id: an ID for the term of the for [dbname]:[accession], where [dbname]
 *      is the short name of the vocabulary and accession is the unique ID.
 *    - cv_id:  an integer indicating the cv_id or an array with 'name' => the
 *      name of the cv.
 *    - synonym: an array with 'name' => the name of the synonym of the cvterm
 *      you want returned; 'cv_id' => the cv_id of the synonym; 'cv_name' =>
 *      the name of the cv of the synonym.
 *    - property: An array/object describing the property to select records
 *      for. It should at least have either a type_name (if unique across cvs)
 *      or type_id. Other supported keys include: cv_id/cv_name (of the type),
 *      value and rank.
 * @param $options
 *   An array of options. Supported keys include:
 *     - Any keys supported by chado_generate_var(). See that function
 *       definition for additional details.
 *
 * NOTE: the $identifier parameter can really be any array similar to $values
 *   passed into chado_select_record(). It should fully specify the cvterm
 *   record to be returned.
 *
 * @return
 *   If unique values were passed in as an identifier then an object describing
 *   the cvterm will be returned (will be a chado variable from
 *   chado_generate_var()). Otherwise, FALSE will be returned.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_get_cvterm($identifiers, $options = array()) {

  // Set Defaults.
  if (!isset($options['include_fk'])) {
    // Tells chado_generate_var to only get the cv.
    $options['include_fk'] = array('cv_id' => TRUE);
  }

  // Error Checking of parameters.
  if (!is_array($identifiers)) {
    tripal_report_error('tripal_cv_api', TRIPAL_ERROR,
      "chado_get_cvterm: The identifier passed in is expected to be an array with the key
        matching a column name in the cvterm table (ie: cvterm_id or name). You passed in %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }
  elseif (empty($identifiers)) {
    tripal_report_error('tripal_cv_api', TRIPAL_ERROR,
      "chado_get_cvterm: You did not pass in anything to identify the cvterm you want. The identifier
        is expected to be an array with the key matching a column name in the cvterm table
        (ie: cvterm_id or name). You passed in %identifier.",
      array('%identifier'=> print_r($identifiers, TRUE))
    );
  }

  // If synonym was passed in, then process this first before calling
  // chado_generate_var().
  if (isset($identifiers['synonym'])) {
    $synonym = $identifiers['synonym']['name'];

    $values = array('synonym' => $synonym);
    if (isset($identifiers['synonym']['cv_id'])) {
      $values['cvterm_id'] = array('cv_id' => $identifiers['synonym']['cv_id']);
    }
    if (isset($identifiers['synonym']['cv_name'])) {
      $values['cvterm_id'] = array('cv_id' => array('name' => $identifiers['synonym']['cv_name']));
    }
    $options = array(
      'case_insensitive_columns' => array('name')
    );
    $result = chado_select_record('cvtermsynonym', array('cvterm_id'), $values, $options);

    // if the synonym doens't exist or more than one record is returned then
    // return false.
    if (count($result) == 0) {
      return FALSE;
    }
    if (count($result) > 1) {
      return FALSE;
    }

    $identifiers = array('cvterm_id' => $result[0]->cvterm_id);
  }

  // If one of the identifiers is property then use chado_get_record_with_property().
  if (isset($identifiers['property'])) {
    $property = $identifiers['property'];
    unset($identifiers['property']);
    $cvterm = chado_get_record_with_property(
      array('table' => 'cvterm', 'base_records' => $identifiers),
      array('type_name' => $property),
      $options
    );
  }
  if (isset($identifiers['id'])) {
    list($db_name, $accession) = preg_split('/:/', $identifiers['id']);
    $cvterm = chado_generate_var('cvterm',array(
      'dbxref_id' => array(
        'db_id' => array(
          'name' => $db_name,
        ),
        'accession' => $accession,
      )
    ));
  }

  // Else we have a simple case and we can just use chado_generate_var to get
  // the cvterm.
  else {
    // Try to get the cvterm.
    $cvterm = chado_generate_var('cvterm', $identifiers, $options);
  }

  // Ensure the cvterm is singular. If it's an array then it is not singular.
  if (is_array($cvterm)) {
    tripal_report_error(
      'tripal_cv_api',
      TRIPAL_ERROR,
      "chado_get_cvterm: The identifiers you passed in were not unique. You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }

  // Report an error if $cvterm is FALSE since then chado_generate_var has
  // failed.
  elseif ($cvterm === FALSE) {
    tripal_report_error(
      'tripal_cv_api',
      TRIPAL_ERROR,
      "chado_get_cvterm: chado_generate_var() failed to return a cvterm based on the identifiers
        you passed in. You should check that your identifiers are correct, as well as, look
        for a chado_generate_var error for additional clues. You passed in %identifier.",
      array(
        '%identifier'=> print_r($identifiers, TRUE)
      )
    );
  }

  // Else, as far we know, everything is fine so give them their cvterm :)
  else {
    return $cvterm;
  }

}

/**
 * Create an options array to be used in a form element
 *   which provides a list of all chado cvterms.
 *
 * @param $cv_id
 *   The chado cv_id; only cvterms with the supplied cv_id will be returnedl.
 * @param $rel_type
 *   Set to TRUE if the terms returned should only be relationship types in
 *   the vocabulary.  This is useful for creating drop-downs of terms
 *   used for relationship linker tables.
 *
 * @return
 *   An associative array with the cvterm_id's as keys. The first
 *   element in the array has a key of '0' and a value of 'Select a Type'.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_get_cvterm_select_options($cv_id, $rel_type = FALSE) {
  $columns = array('cvterm_id', 'name');
  $values = array('cv_id' => $cv_id);
  if ($rel_type) {
    $values['is_relationshiptype'] = 1;
  }
  $s_options = array('order_by' => array('name' => 'ASC'));

  $cvterms = chado_select_record('cvterm', $columns, $values, $s_options);

  $options = array();
  $options[0] = 'Select a Type';
  foreach ($cvterms as $cvterm) {
    $options[$cvterm->cvterm_id] = $cvterm->name;
  }

  return $options;

}

/**
 * Updates the cvtermpath table of Chado for the specified CV.
 *
 * @param $cv_id
 *   The chado cv_id.
 * @param $job_id
 *   This function is intended to be used with the Tripal Jobs API.
 *   When this function is called as a job the $job_id is automatically
 *   passed to this function.
 * @return
 *   TRUE on success FALSE on failure.
 *
 * @ingroup tripal_chado_cv_api
 */
function tripal_update_cvtermpath_old($cv_id, $job_id = NULL) {
  // TODO: need better error checking in this function

  // First get the controlled vocabulary name:
  $sql = "SELECT * FROM {cv} WHERE cv_id = :cv_id";
  $cv = chado_query($sql, array(':cv_id' => $cv_id))->fetchObject();

  print "\nUpdating cvtermpath for $cv->name...\n";

  // We need to set the chado schema as active because some of the
  // functions call other functions which would not be in scope.
  $previous = chado_set_active('chado');
  try {
    $sql = "SELECT * FROM fill_cvtermpath(:name)";

    db_query($sql, array(':name' => $cv->name));
    chado_set_active($previous);
  }
  catch (Exception $e) {
    chado_set_active($previous);
    $error = $e->getMessage();
    tripal_report_error('tripal_chado', TRIPAL_ERROR, "Could not fill cvtermpath table: @error", array('@error' => $error));
    return FALSE;
  }
  return TRUE;
}

/**
 * 
 * @param unknown $cv_id
 */
function chado_clear_cvtermpath($cv_id) {
  $sql = "DELETE FROM {cvtermpath} WHERE cv_id = :cv_id";
  chado_query($sql, [':cv_id' => $cv_id]);
}

/**
 * Replacement for the fill_cvtermpath() stored procedure in Chado.
 *
 * Fills the cvtermpath table of Chado with relationships between every
 * node in the ontology graph and all of it's descendents.  This was 
 * previously performed using the fill_cvtermpath() stored procedure of Chado
 * but that function cannot handle loops in the ontology graphs and results
 * in stack depth errors in PostgreSQL.
 *
 * @param $cv_id
 *   The controlled vocabulary ID from the cv table of Chado (i.e. cv.cv_id).
 * @param $job
 *   An instance of a TripalJob.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_update_cvtermpath($cv_id, $job = NULL) {
  
  $cv = new ChadoRecord('cv', $cv_id);
  print "Building cvterm paths for vocabulary: " . $cv->getValue('name') ."\n"; 
  print "Clearing the cvtermpath table for this vocabulary...\n";
  chado_clear_cvtermpath($cv_id);  
  print "Clearing completed.\n";
  
  // The cache is used to limit repetitive queries by storing known data.
  $cache = [
    // Stores all relationships in the cvtermpath table.
    'rels' => [],
    // Stores all nodes that were visited when processing all nodes as roots.
    'roots_processed' => [],
    // Stores all nodes that were visited when prociessing all children of 
    // a single root. It gets emptied at each root.
    'nodes_processed' => [],
    // For easy lookup stores the is_a relationship.
    'is_a' => NULL,
  ];
  
  // TODO: there's a function to determine the current Chado instance.
  // we should use that.
  $prev_db = chado_set_active('chado');
  $transaction = db_transaction();
  
  try {
    
    // Get the is_a term. The OBO importer adds this for evey vocabulary.
    $sql = "SELECT * FROM cvterm WHERE name = :is_a and cv_id = :cv_id";
    $args = [':is_a' => 'is_a', ':cv_id' => $cv_id];
    $cache['$is_a'] = chado_query($sql, $args)->fetchObject();
    
    // First cache all the relationships for this vocaublary so that we
    // don't have to do repetitive queries to Chado.
    print "Retrieving relationships...\n";
    $sql = "
      SELECT CVTR.subject_id, CVTR.type_id, CVTR.object_id, CVTS.name
      FROM {cvterm_relationship} CVTR
        INNER JOIN {cvterm} CVTO on CVTO.cvterm_id = CVTR.object_id
        INNER JOIN {cvterm} CVTS on CVTS.cvterm_id = CVTR.subject_id
      WHERE CVTO.cv_id = :cv_id 
      ORDER BY CVTR.object_id, CVTR.subject_id
    ";
    $rels = chado_query($sql, [':cv_id' => $cv_id]);
    $total_items;
    while ($rel = $rels->fetchObject()) {
      $cache['rels'][$rel->object_id][] = [$rel->subject_id, $rel->type_id, $rel->name];
    }
    $total_items = count(array_keys($cache['rels']));
    if ($job) {
      $job->logMessage('Found !total relationships in this vocabulary.', ['!total' => $total_items]);
      $job->setTotalItems($total_items);
      $job->logMessage('Note: Progress updates occur as each term is processed and ' . 
        'some terms take longer than others.');
      $job->setProgress(0);
      $job->setInterval(1);
    }
    
    // Next get the tree roots. These are terms that are in relationships as
    // an object but never as a subject.
    $sql = "
      SELECT DISTINCT CVT.cvterm_id, CVT.name
      FROM {cvterm} CVT
        LEFT JOIN {cvterm_relationship} CVTR ON CVT.cvterm_id = CVTR.subject_id
        INNER JOIN {cvterm_relationship} CVTR2 ON CVT.cvterm_id = CVTR2.object_id
      WHERE CVT.cv_id = :cvid AND CVTR.subject_id is NULL and 
        CVT.is_relationshiptype = 0 and CVT.is_obsolete = 0
    ";
    $roots = chado_query($sql, [':cvid' => $cv_id]);
     
    // Iterate through the tree roots.
    print "Processing terms...\n";    
    while ($root = $roots->fetchObject()) {
      $root_id =  $root->cvterm_id;
      $root_name = $root->name;
      
      if ($job) {
        $job->logMessage('Processing tree root: ' . $root_name . '...');
      }
      
      // Now start descending through the tree and add the relationships
      // to the cvtermpath table.
      $num_handled = 0;
      $depth = 0;
      _chado_update_cvtermpath_root($cv_id, $root_id, $root_name, $cache, $job, 
        $num_handled, $depth);
    }
  }
  catch (Exception $e) {
    $transaction->rollback();
    chado_set_active($prev_db);
    throw $e;
  }
}

/**
 * Treats a term within the ontology as a root. 
 * 
 * In order to add all relationships between a term and it's descendents each
 * term gets it's turn as a "root".  The first time this function is called
 * it should be called with the actual root's of the ontology.  This function
 * will then recursively treat each child term within the tree as a root in
 * order to find all relationships.
 * 
 * @param $cv_id
 *   The vocaulary Id
 * @param $root_id
 *   This root term's cvterm Id.
 * @param $root_name
 *   The name of this root term.
 * @param $cache
 *   The cache used for lookups.
 * @param $job
 *   The TripalJob instance.
 * @param $num_handled
 *   Used to keep track of the number of nodes that have been handled for
 *   progress reporting.
 * @param $root_depth
 *   The current depth in the tree of this term.
 */
function _chado_update_cvtermpath_root($cv_id, $root_id, $root_name, &$cache, 
  $job, &$num_handled, $root_depth = 0) {
    
    
  // Don't use a node as a root if we've already used it once before.
  if (in_array($root_id, $cache['roots_processed'])) {
    return;
  }
    
  // Mark this node as having been processed as a root node.
  $cache['roots_processed'][] = $root_id;
  
  // For the actual tree roots we need to add a relatioship to themselves.
  if ($root_depth == 0) {
    $is_a = $cache['$is_a'];
    $type_id = $is_a->cvterm_id;
    $depth = 1;
    _chado_update_cvtermpath_add_relationship($type_id, $root_id, $root_id, $cv_id, $depth);
  }
  
  // Get this term's children and recurse.
  $children = $cache['rels'][$root_id];
  
  // If there are no children do nothing.
  if (!$children) {
    return;
  }
    
  // Set the job progress.
  if ($job) {
    $job->setItemsHandled($num_handled);
  }
  $num_handled++;
  
  // The $path variable contains only the current path on the descent. This
  // is used for detecting loops in the graph. If we encounter a node a
  // second time while traversing a single path of the tree then we've hit
  // a loop.
  $path = [];
 
  // Process the children of this term.  
  $cache['nodes_processed'] = [];
  _chado_update_cvtermpath_process_children($cv_id, $root_id, $root_id, $path, $cache, $next_depth);
  
  // Next make each child of this node a root and recurse again.
  foreach ($children as $child) {
    $child_id = $child[0];
    $child_type_id = $child[1];
    $child_name = $child[2];

    // Process this child as a root. 
    $next_depth = $root_depth + 1;
    _chado_update_cvtermpath_root($cv_id, $child_id, $child_name, $cache, 
      $job, $num_handled, $next_depth);
  }
}  
/**
 * Handles a single node in the tree.
 * 
 * This is a recursive function which calls itself as the tree is descended. It
 * performs a depth-first search of the tree.
 * 
 * @param $cv_id
 *   The vocaulary Id
 * @param $root_id
 *   This root term's cvterm Id.
 * @param $cvterm_id
 *   This term's cvterm Id.
 * @param $path
 *   An array used for storing the current path down the tree. This is the
 *   sequence of nodes visited to this point down a single branch.
 * @param $cache
 *   The cache used for lookups.
 * @param $depth
 *   The current depth in the tree. 
 */
function _chado_update_cvtermpath_process_children($cv_id, $root_id, $cvterm_id, 
  $path, &$cache, $depth = 1) {
  
  $cache['nodes_processed'][$cvterm_id] = TRUE;
      
  // Have we visited this node before while on this path then we won't
  // descend further as this means we've hit a loop.
  if (in_array($cvterm_id, $path)) {
    return;
  }
  
  // Add this term to the path.
  $path[] = $cvterm_id;
  //print implode('-', $path) . "\n";
  
  // Get this term's children and recurse.
  $children = $cache['rels'][$cvterm_id];
  
  // If this term does not have children then return.
  if (!$children) {
    return;
  }
  
  // If the term has children then recurse on those.
  $next_depth = $depth + 1;
  foreach ($children as $child) {
    $child_id = $child[0];
    $child_type_id = $child[1];
         
    // Don't descend for children we've already seen.
    if (array_key_exists($child_id, $cache['nodes_processed'])) {
      continue;
    }
    
    _chado_update_cvtermpath_add_relationship($child_type_id, $child_id, $root_id,
      $cv_id, $next_depth);
    
    _chado_update_cvtermpath_process_children($cv_id, $root_id, $child_id, $path, $cache, $next_depth);
  }
}

/**
 * Inserts values into the cvtermpath table.
 * 
 * After the entire tree below the current root term is traversed, this
 * function is called and inserts all of the relationships that were found
 * into the cvtermpath table. 
 * 
 * @param $visited
 *   The array contaiing relationships for all visited nodes in the tree. These
 *   elements will become the entries in the cvtermpath table.
 * @param $job
 *   The TripalJob instance.
 */
function _chado_update_cvtermpath_add_relationship($type_id, $cvterm_id, 
  $root_id, $cv_id, $depth) {
  
  $cvtermpath = new ChadoRecord('cvtermpath');
  $cvtermpath->setValues([
    'type_id' =>  $type_id,
    'subject_id' => $cvterm_id,
    'object_id' => $root_id,
    'cv_id'  => $cv_id,
    'pathdistance'  => $depth,
  ]);
  $cvtermpath->insert();
}


/**
 * Adds a controlled vocabulary to the CV table of Chado.
 *
 * @param $name
 *   The name of the controlled vocabulary. These are typically all lower case
 *   with no special characters other than an undrescore (for spaces).
 * @param $comment
 *   A description or definition of the vocabulary.
 *
 * @return
 *   An object populated with fields from the newly added database.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_insert_cv($name, $definition) {

  // Insert/update values.
  $ins_values = array(
    'name'       => $name,
    'definition' => $definition
  );

  // See if the CV default exists already in the database.
  $sel_values = array('name' => $name);
  $results = chado_select_record('cv', array('*'), $sel_values);

  // If it does not exists then add it.
  if (count($results) == 0) {
    $success = chado_insert_record('cv', $ins_values);
    if (!$success) {
      tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to create the CV record", NULL);
      return FALSE;
    }
    $results = chado_select_record('cv', array('*'), $sel_values);
  }
  // If it already exists then do an update.
  else {
    $success = chado_update_record('cv', $sel_values, $ins_values);
    if (!$success) {
      tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the CV record", NULL);
      return FALSE;
    }
    $results = chado_select_record('cv', array('*'), $sel_values);
  }

  // Return the cv object.
  return $results[0];
}

/**
 *  Add's a controlled vocabulary term to Chado.
 *
 *  This function will add a cvterm record (and a dbxref record if appropriate
 *  values are provided). If the parent vocabulary does not exist then
 *  that also is added to the cv table.  If the cvterm is a relationship term
 *  then the 'is_relationship' value should be set.  All
 *  terms must also have a corresponding database.  This is specified in the
 *  term's ID just before the colon (e.g. GO:003824).  If the database does not
 *  exist in the DB table then it will be added automatically.  The accession
 *  (the value just after the colon in the term's ID) will be added to the
 *  dbxref table.  If the CVterm already exists and $update is set (default)
 *  then the cvterm is updated.  If the CVTerm already exists and $update is
 *  not set, then no changes are made and the CVTerm object is returned.
 *
 * @param $term
 *   An associative array with the following keys:
 *    - id: the term accession. must be of the form <DB>:<ACCESSION>, where
 *      <DB> is the name of the database to which the cvterm belongs and the
 *      <ACCESSION> is the term's accession number in the database.
 *    - name: the name of the term. usually meant to be human-readable.
 *    - is_obsolete: is present and set to 1 if the term is defunct.
 *    - definition: the definition of the term.
 *    - cv_name: The CV name to which the term belongs.  If this arugment is
 *        null or not provided then the function tries to find a record in the
 *        CV table with the same name provided in the $term[namespace].  If
 *        this field is provided then it overrides what the value in
 *        $term[namespace].
 *    - is_relationship: If this term is a relationship term then this value
 *        should be 1.
 *    - db_name: In some cases the database name will not be part of the
 *        $term['id'] and it needs to be explicitly set.  Use this argument
 *        only if the database name cannot be specififed in the term ID
 *        (e.g. <DB>:<ACCESSION>).
 * @param $options
 *   An associative array with the following keys:
 *    - update_existing: By default this is TRUE.  If the term exists it is
 *      automatically updated.
 *
 * @return
 *   A cvterm object
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_insert_cvterm($term, $options = array()) {
  // Get the term properties.
  $id = (isset($term['id'])) ? $term['id'] : '';
  $name = '';
  $cvname = '';
  $definition = '';
  $is_obsolete = 0;
  $accession = '';
  // Set Defaults.
  if (isset($term['cv_name'])) {
    $cvname = $term['cv_name'];
  }
  else {
    $cvname = 'local';
  }
  // Namespace is deprecated but must be supported for backwards
  // compatability.
  if (array_key_exists('namespace', $term)) {
    $cvname = $term['namespace'];
  }
  if (isset($term['is_relationship'])) {
    $is_relationship = $term['is_relationship'];
  }
  else {
    $is_relationship = 0;
  }
  if (isset($term['db_name'])) {
    $dbname = $term['db_name'];
  }
  else {
    $dbname = 'internal';
  }
  if (isset($options['update_existing'])) {
    $update = $options['update_existing'];
  }
  else {
    $update = 1;
  }
  if (array_key_exists('name', $term)) {
    $name = $term['name'];
  }
  else {
    $name = $id;
  }
  if (array_key_exists('definition', $term)) {
    $definition = preg_replace('/^\"(.*)\"/', '\1', $term['definition']);
  }
  else {
    $definition = '';
  }
  if (array_key_exists('is_obsolete', $term)) {
    $is_obsolete = $term['is_obsolete'];
    if (strcmp($is_obsolete, 'true') == 0) {
      $is_obsolete = 1;
    }
  }
  if (!$name and !$id) {
    tripal_report_error('tripal_cv', TRIPAL_WARNING, "Cannot find cvterm without 'id' or 'name'", NULL);
    return 0;
  }
  if (!$id) {
    $id = $name;
  }
  // Get the accession and the database from the cvterm id.
  if ($dbname) {
    $accession = $id;
  }
  if (preg_match('/^.+?:.*$/', $id)) {
    $accession = preg_replace('/^.+?:(.*)$/', '\1', $id);
    $dbname = preg_replace('/^(.+?):.*$/', '\1', $id);
  }
  // Check that we have a database name, give a different message if it's a
  // relationship.
  if ($is_relationship and !$dbname) {
    tripal_report_error('tripal_cv', TRIPAL_WARNING, "A database name is not provided for this relationship term: $id", NULL);
    return 0;
  }
  if (!$is_relationship and !$dbname) {
    tripal_report_error('tripal_cv', TRIPAL_WARNING, "A database identifier is missing from the term: $id", NULL);
    return 0;
  }
  // Make sure the CV name exists.
  $cv = chado_get_cv(array('name' => $cvname));
  if (!$cv) {
    $cv = chado_insert_cv($cvname, '');
  }
  if (!$cv) {
    tripal_report_error('tripal_cv', TRIPAL_WARNING, "Cannot find namespace '$cvname' when adding/updating $id", NULL);
    return 0;
  }
  // This SQL statement will be used a lot to find a cvterm so just set it
  // here for easy reference below.  Because CV terms can change their names
  // but accessions don't change, the following SQL finds cvterms based on
  // their accession rather than the name.
  $cvtermsql = "
    SELECT CVT.name, CVT.cvterm_id, CV.cv_id, CV.name as cvname,
      DB.name as dbname, DB.db_id, DBX.accession
    FROM {cvterm} CVT
      INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
      INNER JOIN {db} DB on DBX.db_id = DB.db_id
      INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
    WHERE DBX.accession = :accession and DB.name = :name
  ";
  // Add the database. The function will just return the DB object if the
  // database already exists.
  $db = chado_get_db(array('name' => $dbname));
  if (!$db) {
    $db = chado_insert_db(array('name' => $dbname));
  }
  if (!$db) {
    tripal_report_error('tripal_cv', TRIPAL_WARNING, "Cannot find database '$dbname' in Chado.", NULL);
    return 0;
  }
  // The cvterm table has two unique dependencies. We need to check both.
  // first check the (name, cv_id, is_obsolete) constraint.
  $values = array(
    'name' => $name,
    'is_obsolete' => $is_obsolete,
    'cv_id' => array(
      'name' => $cvname,
    ),
  );
  $result = chado_select_record('cvterm', array('*'), $values);
  if (count($result) == 1) {
    $cvterm = $result[0];
    // Get the dbxref record.
    $values = array('dbxref_id' => $cvterm->dbxref_id);
    $result = chado_select_record('dbxref', array('*'), $values);
    $dbxref = $result[0];
    if (!$dbxref) {
      tripal_report_error('tripal_cv', TRIPAL_ERROR,
          'Unable to access the dbxref record for the :term cvterm. Term Record: !record',
          array(':term' => $name, '!record' => print_r($cvterm, TRUE))
      );
      return FALSE;
    }
    // Get the db.
    $values = array('db_id' => $dbxref->db_id);
    $result = chado_select_record('db', array('*'), $values);
    $db_check = $result[0];
    //     // The database name for this existing term does not match that of the
    //     // one provided to this function.  The CV name matches otherwise we
    //     // wouldn't have made it this far. So, let's swap the database for
    //     // this term.
    //     if ($db_check->name != $db->name) {
    //       // Look to see if the correct dbxref record already exists for this
    //       // database.
    //       $values = array(
    //         'db_id' => $db->db_id,
    //         'accession' => $accession,
    //       );
    //       $result = chado_select_record('dbxref', array('*'), $values);
    //       // If we already have a good dbxref then we want to update our cvterm
    //       // to use this dbxref.
    //       if (count($result) > 0) {
    //         $dbxref = $result[0];
    //         $match = array('cvterm_id' => $cvterm->cvterm_id);
    //         $values = array('dbxref_id' => $dbxref->dbxref_id);
    //         $success = chado_update_record('cvterm', $match, $values);
    //         if (!$success) {
    //           tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to correct the dbxref id for the cvterm " .
    //             "'$name' (id: $accession), for database $dbname", NULL);
    //           return 0;
    //         }
    //       }
    //       // If we don't have the dbxref then we want to delete our cvterm and let
    //       // the code below recreate it with the correct info.
    //       else {
    //         $match = array('cvterm_id' => $cvterm->cvterm_id);
    //         chado_delete_record('cvterm', $match);
    //       }
    //     }
    // Check that the accession matches.  Sometimes an OBO can define a term
    // multiple times but with different accessions.  If this is the case we
    // can't do an insert or it will violate the constraint in the cvterm table.
    // So we'll need to add the record to the cvterm_dbxref table instead.
    if ($dbxref->accession != $accession) {
      // Get/add the dbxref for his term.
      $dbxref_new = chado_insert_dbxref(array(
        'db_id' => $db->db_id,
        'accession' => $accession
      ));
      if (!$dbxref_new) {
        tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
            "$name (id: $accession), for database $dbname", NULL);
        return 0;
      }
      // Check to see if the cvterm_dbxref record already exists.
      $values = array(
        'cvterm_id' => $cvterm->cvterm_id,
        'dbxref_id' => $dbxref_new->dbxref_id,
        'is_for_definition' => 1,
      );
      $result = chado_select_record('cvterm_dbxref', array('*'), $values);
      // if the cvterm_dbxref record does not exists then add it
      if (count($result)==0) {
        $options = array(
          'return_record' => FALSE,
        );
        $success = chado_insert_record('cvterm_dbxref', $values, $options);
        if (!$success) {
          tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to find or insert the cvterm_dbxref record for a " .
              "duplicated cvterm:  $name (id: $accession), for database $dbname", NULL);
          return 0;
        }
      }
      // Get the original cvterm with the same name and return that.
      $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
      $cvterm = $result->fetchObject();
      return $cvterm;
    }
    // Continue on, we've fixed the record if the db_id did not match.
    // We can now perform and updated if we need to.
  }
  // Get the CVterm record.
  $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  $cvterm = $result->fetchObject();
  if (!$cvterm) {
    // Check to see if the dbxref exists if not, add it.
    $dbxref = chado_insert_dbxref(array(
      'db_id' => $db->db_id,
      'accession' => $accession
    ));
    if (!$dbxref) {
      tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
          "$name (id: $accession), for database $dbname", NULL);
      return 0;
    }
    // Check to see if the dbxref already has an entry in the cvterm table.
    $values = array('dbxref_id' => $dbxref->dbxref_id);
    $check = chado_select_record('cvterm', array('cvterm_id'), $values);
    if (count($check) == 0) {
      // now add the cvterm
      $ins_values = array(
        'cv_id'                => $cv->cv_id,
        'name'                 => $name,
        'definition'           => $definition,
        'dbxref_id'            => $dbxref->dbxref_id,
        'is_obsolete'          => $is_obsolete,
        'is_relationshiptype'  => $is_relationship,
      );
      $success = chado_insert_record('cvterm', $ins_values);
      if (!$success) {
        if (!$is_relationship) {
          tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to insert the term: $name ($dbname)", NULL);
          return 0;
        }
        else {
          tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to insert the relationship term: $name (cv: " . $cvname . " db: $dbname)", NULL);
          return 0;
        }
      }
    }
    // This dbxref already exists in the cvterm table.
    else {
      tripal_report_error('tripal_cv', TRIPAL_WARNING, "The dbxref already exists for another cvterm record: $name (cv: " . $cvname . " db: $dbname)", NULL);
      return 0;
    }
    $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
    $cvterm = $result->fetchObject();
  }
  // Update the cvterm.
  elseif ($update) {
    // First, basic update of the term.
    $match = array('cvterm_id' => $cvterm->cvterm_id);
    $upd_values = array(
      'name'                => $name,
      'definition'          => $definition,
      'is_obsolete'         => $is_obsolete,
      'is_relationshiptype' => $is_relationship,
    );
    $success = chado_update_record('cvterm', $match, $upd_values);
    if (!$success) {
      tripal_report_error('tripal_cv', TRIPAL_WARNING, "Failed to update the term: $name", NULL);
      return 0;
    }
    // Second, check that the dbxref has not changed and if it has then update
    // it.
    $checksql = "
      SELECT cvterm_id
      FROM {cvterm} CVT
        INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
        INNER JOIN {db} DB on DBX.db_id = DB.db_id
        INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
      WHERE DBX.accession = :accession and DB.name = :dbname and CVT.name = :term and CV.name = :cvname
    ";
    $check = chado_query($checksql, array(':accession' => $accession, ':dbname' => $dbname, ':term' => $name, ':cvname' => $cvname))->fetchObject();
    if (!$check) {
      // Check to see if the dbxref exists if not, add it.
      $dbxref = chado_insert_dbxref(array(
        'db_id' => $db->db_id,
        'accession' => $accession
      ));
      if (!$dbxref) {
        tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
            "$name (id: $accession), for database $dbname", NULL);
        return 0;
      }
      $match = array('cvterm_id' => $cvterm->cvterm_id);
      $upd_values = array(
        'dbxref_id' => $dbxref->dbxref_id,
      );
      $success = chado_update_record('cvterm', $match, $upd_values);
      if (!$success) {
        tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the term $name with new accession $db:$accession", NULL);
        return 0;
      }
    }
    // Finally grab the updated details.
    $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
    $cvterm = $result->fetchObject();
  }
  else {
    // Do nothing, we have the cvterm but we don't want to update.
  }
  // Return the cvterm.
  return $cvterm;
}


/**
 * TODO: deprecate this function
 *
 * Avoid using this function as it will be deprecated in future releases.
 *
 * This function allows other modules to programatically
 * submit an ontology for loading into Chado.
 *
 * This function will add a job to the Jobs subsystem for parsing the ontology.
 * You can either pass a known OBO ID to the function or the URL
 * or full path the the ontology file.  If a URL or file name is
 * passed then the $obo_name argument must also be provided.  If
 * this is the first time the ontology has been provided to Tripal
 * then it will be added to the database and will be assigned a
 * unique OBO ID.
 *
 * @param $obo_id
 *   If the ontology is already loaded into the Tripal tables then
 *   use this argument to specify the unique ID for the ontology
 *   that will be loaded.
 * @param $obo_name
 *   If the OBO has not been added before then use this argument
 *   to specify the human readable name of the ontology.
 * @param $obo_url
 *   If the OBO to be loaded is located on a remote server then
 *   use this argument to provide the URL.
 * @param $obo_file
 *   If the OBO is housed on the local file system of the server then
 *   use this argument to specify the full path.
 *
 * @return
 *   returns the job_id of the submitted job or FALSE if the job was not added
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_submit_obo_job($obo) {
  global $user;

  // Set Defaults
  $obo['obo_id'] = (isset($obo['obo_id'])) ? $obo['obo_id'] : NULL;
  $obo['name']   = (isset($obo['name']))   ? $obo['name']   : NULL;
  $obo['url']    = (isset($obo['url']))    ? $obo['url']    : NULL;
  $obo['file']   = (isset($obo['file']))   ? $obo['file']   : NULL;

  $includes = array(
    drupal_get_path('module', 'tripal_chado') . '/includes/tripal_chado.cv.inc',
  );

  if ($obo['obo_id']) {
    $sql = "SELECT * FROM {tripal_cv_obo} WHERE obo_id = :obo_id";
    $result = db_query($sql, array(':obo_id' => $obo['obo_id']))->fetchObject();

    $args = array($result->obo_id);
    return tripal_add_job("Load OBO " . $result->name, 'tripal_chado',
       "tripal_cv_load_obo", $args, $user->uid, 10, $includes);
  }
  else {
    if ($obo['url']) {
      $sql = "SELECT * FROM {tripal_cv_obo} WHERE path = :url";
      $result = db_query($sql, array(':url' => $obo['url']))->fetchObject();

      $args = array($result->obo_id);
      return tripal_add_job("Load OBO " . $result->name, 'tripal_chado',
        "tripal_cv_load_obo", $args, $user->uid, 10, $includes);
    }
    elseif ($obo['file']) {
      $sql = "SELECT * FROM {tripal_cv_obo} WHERE path = :file";
      $result = db_query($sql, array(':url' => $obo['file']))->fetchObject();

      $args = array($result->obo_id);
      return tripal_add_job("Load OBO " . $result->name, 'tripal_chado',
        "tripal_cv_load_obo", $args, $user->uid, 10, $includes);
    }
  }
  return FALSE;
}

/**
 * Add the OBO to the tripal_cv_obo table in the Drupal database.
 *
 * If the OBO name already exists in the table then the path is updated.
 *
 * @param $name
 *   The human readable name of this ontology.
 * @param $path
 *   The file path or URL of the ontology.
 *
 * @return
 *   Returns the ontology ID.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_insert_obo($name, $path) {
  // Make sure an OBO with the same name doesn't already exist.
  $obo_id = db_select('tripal_cv_obo', 'tco')
    ->fields('tco', array('obo_id'))
    ->condition('name', $name)
    ->execute()
    ->fetchField();

  if ($obo_id) {
    db_update('tripal_cv_obo')
      ->fields(array(
        'path' => $path,
      ))
      ->condition('name', $name)
      ->execute();
    return $obo_id;
  }
  else {
    $obo_id = db_insert('tripal_cv_obo')
      ->fields(array(
        'name' => $name,
        'path' => $path,
      ))
      ->execute();
    return $obo_id;
  }
}

/**
 * Retrieves an OBO record.
 *
 * @param $values
 *   An associate array with the following allowed keys: obo_id, name.
 *
 * @return
 *   An instance of an OBO record object.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_get_obo($values) {
  $query = db_select('tripal_cv_obo', 'tco')
    ->fields('tco');

  if (array_key_exists('name', $values)) {
    $query->condition('tco.name', $values['name']);
  }
  if (array_key_exists('obo_id', $values)) {
    $query->condition('tco.obo_id', $values['obo_id']);
  }
  return $query->execute()->fetchObject();
}

/**
 * This function is intended to be used in autocomplete forms.
 *
 * This function searches for a matching controlled vobulary name.
 *
 * @param $string
 * The string to search for.
 *
 * @return
 * A json array of terms that begin with the provided string.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_autocomplete_cv($string = '') {
  $sql = "
    SELECT CV.cv_id, CV.name
    FROM {cv} CV
    WHERE lower(CV.name) like lower(:name)
    ORDER by CV.name
    LIMIT 25 OFFSET 0
  ";
  $results = chado_query($sql, array(':name' => $string . '%'));
  $items = array();
  foreach ($results as $cv) {
    $items[$cv->name] = $cv->name;
  }

  drupal_json_output($items);
}

/**
 * This function is intended to be used in autocomplete forms
 * for searching for CV terms that begin with the provided string.
 *
 * @param $cv_id
 * The CV ID in which to search for the term.
 * @param $string
 * The string to search for.
 *
 * @return
 * A json array of terms that begin with the provided string.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_autocomplete_cvterm($cv_id, $string = '') {
  if ($cv_id) {
    $sql = "
      SELECT CVT.cvterm_id, CVT.name
      FROM {cvterm} CVT
      WHERE CVT.cv_id = :cv_id and lower(CVT.name) like lower(:name)
      UNION
      SELECT CVT2.cvterm_id, CVTS.synonym as name
      FROM {cvterm} CVT2
        INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
      WHERE CVT2.cv_id = :cv_id and lower(CVTS.synonym) like lower(:name)
      ORDER by name
      LIMIT 25 OFFSET 0
    ";
    $results = chado_query($sql, array(':cv_id' => $cv_id, ':name' => $string . '%'));
    $items = array();
    foreach ($results as $term) {
      $items[$term->name] = $term->name;
    }
  }
  // If a CV wasn't provided then search all of them, and include the cv
  // in the results.
  else {
    $sql = "
      SELECT CVT.cvterm_id, CVT.name, CV.name as cvname, CVT.cv_id
      FROM {cvterm} CVT
        INNER JOIN {cv} CV on CVT.cv_id = CV.cv_id
      WHERE lower(CVT.name) like lower(:name)
      UNION
      SELECT CVT2.cvterm_id, CVTS.synonym as name, CV2.name as cvname, CVT2.cv_id
      FROM {cvterm} CVT2
        INNER JOIN {cv} CV2 on CVT2.cv_id = CV2.cv_id
        INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
      WHERE lower(CVTS.synonym) like lower(:name)
      ORDER by name
      LIMIT 25 OFFSET 0
    ";
    $results = chado_query($sql, array(':name' => $string . '%'));
    $items = array();
    foreach ($results as $term) {
      $items[$term->name] = $term->name;
    }
  }

  drupal_json_output($items);
}

/**
 * Add a record to a cvterm linking table (ie: feature_cvterm).
 *
 * @param $basetable
 *   The base table to which the cvterm should be linked/associated. Thus to
 *   associate a cvterm to a feature the basetable=feature and cvterm_id is
 *   added to the feature_cvterm table.
 * @param $record_id
 *   The primary key of the basetable to associate the cvterm with. This should
 *   be in integer.
 * @param $cvterm
 *   An associative array describing the cvterm. Valid keys include:
 *     - name: the name for the cvterm,
 *     - cv_name: the name of the cv the cvterm belongs to.
 *     - cv_id: the primary key of the cv the cvterm belongs to.
 * @param $options
 *   An associative array of options. Valid keys include:
 *     - insert_cvterm: Insert the cvterm if it doesn't already exist. FALSE is
 *       the default.
 *
 * @ingroup tripal_chado_cv_api
 */
function chado_associate_cvterm($basetable, $record_id, $cvterm, $options = array()) {
  $linking_table = $basetable . '_cvterm';
  $foreignkey_name = $basetable . '_id';

  // Default Values
  $options['insert_cvterm'] = (isset($options['insert_cvterm'])) ? $options['insert_cvterm'] : FALSE;

  // If the cvterm_id is not set then find the cvterm record using the name and
  // cv_id.
  if (!isset($cvterm['cvterm_id'])) {
    $values = array(
      'name' => $cvterm['name'],
    );
    if (isset($cvterm['cv_id'])) {
      $values['cv_id'] = $cvterm['cv_id'];
    }
    elseif (isset($cvterm['cv_name'])) {
      $values['cv_id'] = array(
        'name' => $cvterm['cv_name']
      );
    }
    else {
      tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
        "chado_associate_cvterm: The cvterm needs to have either the cv_name or cv_id
          supplied. You were trying to associate a cvterm with the %base %record_id
          and supplied the cvterm values: %cvterm.",
        array('%base' => $basetable, '%record_id' => $record_id, '%cvterm' => print_r($cvterm,TRUE))
      );
      return FALSE;
    }

    // Get the cvterm. If it doesn't exist then add it if the option
    // 'insert_cvterm' is set.
    $select = chado_select_record('cvterm', array('*'), $values);
    if ($select) {
      $cvterm['cvterm_id'] = $select[0]->cvterm_id;
    }
    elseif ($options['insert_cvterm']) {
      // Insert the cvterm
      $insert = chado_insert_cvterm($values);
      if (isset($insert->cvterm_id)) {
        $cvterm['cvterm_id'] = $insert->cvterm_id;
      }
      else {
        tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
          "chado_associate_cvterm: Unable to insert the cvterm using the cvterm values: %cvterm.",
          array('%cvterm' => print_r($cvterm,TRUE))
        );
        return FALSE;
      }
    }
    else {
      tripal_report_error('tripal_api', TRIPAL_WARNING,
        "chado_associate_cvterm: The cvterm doesn't already exist. You supplied the cvterm values: %cvterm.",
        array('%cvterm' => print_r($cvterm,TRUE))
      );
      return FALSE;
    }
  }

  // Now add the link between the record & cvterm.
  if ($cvterm['cvterm_id'] > 0) {
    $values = array(
      'cvterm_id' => $cvterm['cvterm_id'],
      $foreignkey_name => $record_id,
      'pub_id' => 1,
    );

    // Check if the cvterm is already associated. If so, don't re-add it.
    $result = chado_select_record($linking_table, array('*'), $values);
    if (!$result) {
      $success = chado_insert_record($linking_table, $values);
      if (!$success) {
        tripal_report_error('tripal_api', TRIPAL_WARNING,
          "Failed to insert the %base record %term",
          array('%base' => $linking_table, '%term' => $cvterm['name'])
        );
        return FALSE;
      }
      $result = chado_select_record($linking_table, array('*'), $values);
    }

    if (isset($result[0])) {
      return $result[0];
    }
    else {
      return FALSE;
    }
  }

  return FALSE;
}