tripal_bibtex.module

<?php
module_load_include('inc', 'tripal_chado', 'includes/loaders/tripal_chado.pub_importers');
/**
 * @ file
 * Provides the main functionality for this module.
 */

/**
 * Import publications from a bibtex formatted file.
 */
function tripal_bibtex_import_bibtex($filename, $options = []) {
  print 'Filename: ' . $filename . "\n";

  $in = fopen($filename, 'r');
  if ($in) {
    $record = [];
    while (($line = fgets($in)) !== FALSE) {
      $line = trim($line);

      // Check if we are starting a new record.

      if (preg_match('/@(\w+)\s*{/', $line, $matches)) {
        $ptype = ($matches[1] == 'article' or $matches[1] == 'Article') ? 'Journal Article' : $matches[1];

        if (sizeof($record) > 1) {
          tripal_bibtext_import_publication($record);
        }

        // Start the new record.
        $record = [
          'Publication Type' => [ucwords($ptype)],
        ];
      }
      // Otherwise check which type of line we are on.
      // Bibtex format consists of tag = {value}
      // This is made slightly more complicated by the fact that value 
      // can span multiple lines.
      // First; Simple Case: All on one line
      // elseif (preg_match('/^(\w+) = ["{](.*)["}]/', $line, $matches)) {
      elseif (preg_match('/\s*(.*?)\s*=\s*["{](.*)["}]/', $line, $matches)) {

        $current_tag = ucwords($matches[1]);
        $record[$current_tag] = $matches[2];

        $current_tag = NULL;
        $record_ended = TRUE;
      }
      // Numeric values don't need to be in quotes or brackets:
      elseif (preg_match('/\s*(.*?)\s*=\s*(\d+\.*\d*)/', $line, $matches)) {
        $current_tag = ucwords($matches[1]);
        $record[$current_tag] = $matches[2];

        $current_tag = NULL;
        $record_ended = TRUE;
      }
      // Next up: the tag is starting.
      elseif (preg_match('/\s*(.*?)\s*=\s*["{](.*)/', $line, $matches)) {

        $current_tag = ucwords($matches[1]);
        $record[$current_tag] = $matches[2];

        $record_ended = FALSE;
      }
      // The tag is ending.
      elseif (preg_match('/(.*)}/', $line, $matches) AND $current_tag) {

        $record[$current_tag] .= $matches[1];

        $current_tag = NULL;
        $record_ended = TRUE;
      }
      // We are in the middle of a multi-line tag.
      elseif ($record_ended == FALSE AND $current_tag) {
        $record[$current_tag] .= $line;
      }
      // if we hit the ending bracket of the stanza do nothing.
      elseif (preg_match('/^\s*}\s*$/', $line) or preg_match('/^\s*$/', $line)) {
         // do nothing.
      }
      else {
        print("worst case: '$line'\n");
      }
    }
  }

  // Last record.
  if (sizeof($record) > 1) {
    tripal_bibtext_import_publication($record);
  }

  // Finally, sync publications & contacts.

  $pub_bundle = tripal_bibtex_get_chado_table_bundle('pub');
  $contact_bundle = tripal_bibtex_get_chado_table_bundle('contact');
  if ($pub_bundle) {
    chado_publish_records(['bundle_name' => $pub_bundle]);

  }
  if ($contact_bundle) {
    chado_publish_records(['bundle_name' => $contact_bundle]);
  }

  // else {
  //   print "ERROR: Unable to open file\n";
  // }
}

/**
 * Clean-up a bibtext pub array and insert it into chado using the tripal api.
 */
function tripal_bibtext_import_publication($record) {

  // Clean-up Record to match expectations.
  // Key Expectations:
  // Publication Type: an array of publication types. a publication can have more than one type
  // Authors: a string containing all of the authors of a publication
  // Journal Name: a string containing the journal name
  // Journal Abbreviation: a string containing the journal name abbreviation
  // Series Name: a string containing the series (e.g. conference proceedings) name
  // Series Abbreviation: a string containing the series name abbreviation
  // Volume: the serives volume number
  // Issue: the series issue number
  // Pages: the page numbers for the publication
  // Publication Date: A date in the

  if (isset($record['Journal'])) {
    $record['Journal Name'] = $record['Journal'];
  }
  unset($record['Journal']);
  if (isset($record['Author'])) {
    $record['Authors'] = $record['Author'];
  }
  unset($record['Author']);
  if (preg_match('/(.*)\.$/', $record['Title'], $matches)) {
    $record['Title'] = $matches[1];
  }
  if (isset($record['Note'])) {
    $record['Notes'] = $record['Note'];
    unset($record['Note']);
  }

  // Process the Authors.
  // The main goal is to process the bibtex format into the array that tripal_pub_add_authors() requires.
  // Additionally, the bibtex author list is ugly so we would like to re-format that based
  // on the format of each author name.
  $first_is_inital = [];
  $record['Author List'] = explode(' and ', $record['Authors']);
  foreach ($record['Author List'] as $k => $author) {
    $author = explode(', ', $author);
    $record['Author List'][$k] = [
      'Surname' => $author[0],
      'Given Name' => $author[1],
    ];

    // Check if the first name is condensed into initals.
    if (preg_match('/^[A-Z]+$/', $author[1])) {
      $first_is_inital[] = $k;
    }
  }
  // Now if all the authors are first initial formatted then remove comma's between name
  // name parts and replace ands with comma's
  // ie: "Sanderson, LA and Bett, KE and Caron, C" becomes "Sanderson LA, Bett KE, Caron C"
  if (sizeof($first_is_inital) == sizeof($record['Author List'])) {
    $record['Authors'] = str_replace(',', '', $record['Authors']);
    $record['Authors'] = str_replace(' and ', ', ', $record['Authors']);
  }
  // Otherwise, just replace ands with semi-colons.
  else {
    $record['Authors'] = str_replace(' and ', '; ', $record['Authors']);
  }

  // Add Citation.
  $record['Citation'] = chado_pub_create_citation($record);

  // Finally, the Tripal pub check for duplicate record doesn't seem to be working for
  // me. Thus check manually here.
  // @TODO: Investigate why Tripal pub's check is not working for me.
  // I'm just going to check title & year.
  $pub_id = chado_query('SELECT pub_id FROM {pub} WHERE title=:title AND pyear=:year',
    [':title' => $record['Title'], ':year' => $record['Year']])->fetchField();
  if (!$pub_id) {
    $action = NULL;
    $pub_id = tripal_pub_add_publication($record, $action, TRUE);
  }

}

function tripal_bibtex_get_chado_table_bundle($table_name) {

  $query = db_select('public.chado_bundle', 'cb');
  $query->fields('cb', ['bundle_id']);
  $query->condition('data_table', $table_name);
  $id = $query->execute()->fetchField();
  if (!$id) {
    return FALSE;
  }
  return 'bio_data_' . $id;
}