tripal_chado.cv.api.inc 51 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543
  1. <?php
  2. /**
  3. * @file
  4. * This module provides a set of functions to simplify working with
  5. * controlled vocabularies.
  6. */
  7. /**
  8. * @defgroup tripal_chado_api Controlled Vocabulary API
  9. * @ingroup tripal_api
  10. * This module provides a set of functions to simplify working with
  11. * controlled vocabularies. Most of the API functions deal with retrieving
  12. * terms or their parent vocabularies.
  13. *
  14. * However, the API also supports
  15. * generation of trees for browsing a vocabulary as well as generation of
  16. * pie graphs for display of hierarchical counts of terms. Version 0.3b of
  17. * Tripal provides a feature browser and a feature summary chart uses
  18. * the API functions provided here. But in general charts and trees can be
  19. * created for any controlled vocabulary.
  20. *
  21. */
  22. /**
  23. * Retrieves a chado controlled vocabulary variable
  24. *
  25. * @param $identifier
  26. * An array with the key stating what the identifier is. Supported keys (only on of the
  27. * following unique keys is required):
  28. * - cv_id: the chado cv.cv_id primary key
  29. * - name: the chado cv.name field (assume unique)
  30. * @param $options
  31. * An array of options. Supported keys include:
  32. * - Any keys supported by chado_generate_var(). See that function definition for
  33. * additional details.
  34. *
  35. * NOTE: the $identifier parameter can really be any array similar to $values passed into
  36. * chado_select_record(). It should fully specify the cv record to be returned.
  37. *
  38. * @return
  39. * If unique values were passed in as an identifier then an object describing the cv
  40. * will be returned (will be a chado variable from chado_generate_var()). Otherwise,
  41. * FALSE will be returned.
  42. *
  43. * @ingroup tripal_chado_api
  44. */
  45. function tripal_get_cv($identifiers, $options = array()) {
  46. // Set Defaults
  47. if (!isset($options['include_fk'])) {
  48. // Tells chado_generate_var not to follow any foreign keys
  49. $options['include_fk'] = array();
  50. }
  51. // Error Checking of parameters
  52. if (!is_array($identifiers)) {
  53. tripal_report_error(
  54. 'tripal_chado_api',
  55. TRIPAL_ERROR,
  56. "tripal_get_cv: The identifier passed in is expected to be an array with the key
  57. matching a column name in the cv table (ie: cv_id or name). You passed in %identifier.",
  58. array(
  59. '%identifier'=> print_r($identifiers, TRUE)
  60. )
  61. );
  62. }
  63. elseif (empty($identifiers)) {
  64. tripal_report_error(
  65. 'tripal_chado_api',
  66. TRIPAL_ERROR,
  67. "tripal_get_cv: You did not pass in anything to identify the cv you want. The identifier
  68. is expected to be an array with the key matching a column name in the cv table
  69. (ie: cv_id or name). You passed in %identifier.",
  70. array(
  71. '%identifier'=> print_r($identifiers, TRUE)
  72. )
  73. );
  74. }
  75. // Try to get the cv
  76. $cv = chado_generate_var(
  77. 'cv',
  78. $identifiers,
  79. $options
  80. );
  81. // Ensure the cv is singular. If it's an array then it is not singular
  82. if (is_array($cv)) {
  83. tripal_report_error(
  84. 'tripal_chado_api',
  85. TRIPAL_ERROR,
  86. "tripal_get_cv: The identifiers you passed in were not unique. You passed in %identifier.",
  87. array(
  88. '%identifier'=> print_r($identifiers, TRUE)
  89. )
  90. );
  91. }
  92. // Report an error if $cv is FALSE since then chado_generate_var has failed
  93. elseif ($cv === FALSE) {
  94. tripal_report_error(
  95. 'tripal_chado_api',
  96. TRIPAL_ERROR,
  97. "tripal_get_cv: chado_generate_var() failed to return a cv based on the identifiers
  98. you passed in. You should check that your identifiers are correct, as well as, look
  99. for a chado_generate_var error for additional clues. You passed in %identifier.",
  100. array(
  101. '%identifier'=> print_r($identifiers, TRUE)
  102. )
  103. );
  104. }
  105. // Else, as far we know, everything is fine so give them their cv :)
  106. else {
  107. return $cv;
  108. }
  109. }
  110. /**
  111. * Create an options array to be used in a form element which provides a
  112. * list of all chado cvs
  113. *
  114. * @return
  115. * An array(cv_id => name) for each cv in the chado cv table
  116. *
  117. * @ingroup tripal_chado_api
  118. */
  119. function tripal_get_cv_select_options() {
  120. $results = chado_select_record('cv', array('cv_id', 'name'), array(), array('order_by' => array('name' => 'ASC')));
  121. $options = array();
  122. $options[] = 'Select a Vocabulary';
  123. foreach ($results as $r) {
  124. $options[$r->cv_id] = $r->name;
  125. }
  126. return $options;
  127. }
  128. /**
  129. * Retrieves a chado controlled vocabulary term variable
  130. *
  131. * @param $identifier
  132. * An array apropriate for use with the chado_generate_var for uniquely
  133. * identifying a cvterm record. Alternativley, there are also some specially
  134. * handled keys. They are:
  135. * - id: an ID for the term of the for [dbname]:[accession], where [dbname]
  136. * is the short name of the vocabulary and accession is the unique ID.
  137. * - cv_id: an integer indicating the cv_id or an array with 'name' => the
  138. * name of the cv.
  139. * - synonym: an array with 'name' => the name of the synonym of the cvterm
  140. * you want returned; 'cv_id' => the cv_id of the synonym; 'cv_name' =>
  141. * the name of the cv of the synonym
  142. * - property: An array/object describing the property to select records
  143. * for. It should at least have either a type_name (if unique across cvs)
  144. * or type_id. Other supported keys include: cv_id/cv_name (of the type),
  145. * value and rank
  146. * @param $options
  147. * An array of options. Supported keys include:
  148. * - Any keys supported by chado_generate_var(). See that function
  149. * definition for additional details.
  150. *
  151. * NOTE: the $identifier parameter can really be any array similar to $values
  152. * passed into chado_select_record(). It should fully specify the cvterm
  153. * record to be returned.
  154. *
  155. * @return
  156. * If unique values were passed in as an identifier then an object describing
  157. * the cvterm will be returned (will be a chado variable from
  158. * chado_generate_var()). Otherwise, FALSE will be returned.
  159. *
  160. * @ingroup tripal_cv_api
  161. */
  162. function tripal_get_cvterm($identifiers, $options = array()) {
  163. // Set Defaults
  164. if (!isset($options['include_fk'])) {
  165. // Tells chado_generate_var to only get the cv
  166. $options['include_fk'] = array('cv_id' => TRUE);
  167. }
  168. // Error Checking of parameters
  169. if (!is_array($identifiers)) {
  170. tripal_report_error('tripal_cv_api', TRIPAL_ERROR,
  171. "tripal_get_cvterm: The identifier passed in is expected to be an array with the key
  172. matching a column name in the cvterm table (ie: cvterm_id or name). You passed in %identifier.",
  173. array('%identifier'=> print_r($identifiers, TRUE))
  174. );
  175. }
  176. elseif (empty($identifiers)) {
  177. tripal_report_error('tripal_cv_api', TRIPAL_ERROR,
  178. "tripal_get_cvterm: You did not pass in anything to identify the cvterm you want. The identifier
  179. is expected to be an array with the key matching a column name in the cvterm table
  180. (ie: cvterm_id or name). You passed in %identifier.",
  181. array('%identifier'=> print_r($identifiers, TRUE))
  182. );
  183. }
  184. // If synonym was passed in, then process this first before calling chado_generate_var()
  185. if (isset($identifiers['synonym'])) {
  186. $synonym = $identifiers['synonym']['name'];
  187. $values = array('synonym' => $synonym);
  188. if (isset($identifiers['synonym']['cv_id'])) {
  189. $values['cvterm_id'] = array('cv_id' => $identifiers['synonym']['cv_id']);
  190. }
  191. if (isset($identifiers['synonym']['cv_name'])) {
  192. $values['cvterm_id'] = array('cv_id' => array('name' => $identifiers['synonym']['cv_name']));
  193. }
  194. $options = array(
  195. 'case_insensitive_columns' => array('name')
  196. );
  197. $result = chado_select_record('cvtermsynonym', array('cvterm_id'), $values, $options);
  198. // if the synonym doens't exist or more than one record is returned then return false
  199. if (count($result) == 0) {
  200. return FALSE;
  201. }
  202. if (count($result) > 1) {
  203. return FALSE;
  204. }
  205. $identifiers = array('cvterm_id' => $result[0]->cvterm_id);
  206. }
  207. // If one of the identifiers is property then use chado_get_record_with_property()
  208. if (isset($identifiers['property'])) {
  209. $property = $identifiers['property'];
  210. unset($identifiers['property']);
  211. $cvterm = chado_get_record_with_property(
  212. array('table' => 'cvterm', 'base_records' => $identifiers),
  213. array('type_name' => $property),
  214. $options
  215. );
  216. }
  217. if (isset($identifiers['id'])) {
  218. list($db_name, $accession) = preg_split('/:/', $identifiers['id']);
  219. $cvterm = chado_generate_var('cvterm',array(
  220. 'dbxref_id' => array(
  221. 'db_id' => array(
  222. 'name' => $db_name,
  223. ),
  224. 'accession' => $accession,
  225. )
  226. ));
  227. }
  228. // Else we have a simple case and we can just use chado_generate_var to get the cvterm
  229. else {
  230. // Try to get the cvterm
  231. $cvterm = chado_generate_var('cvterm', $identifiers, $options);
  232. }
  233. // Ensure the cvterm is singular. If it's an array then it is not singular
  234. if (is_array($cvterm)) {
  235. tripal_report_error(
  236. 'tripal_cv_api',
  237. TRIPAL_ERROR,
  238. "tripal_get_cvterm: The identifiers you passed in were not unique. You passed in %identifier.",
  239. array(
  240. '%identifier'=> print_r($identifiers, TRUE)
  241. )
  242. );
  243. }
  244. // Report an error if $cvterm is FALSE since then chado_generate_var has failed
  245. elseif ($cvterm === FALSE) {
  246. tripal_report_error(
  247. 'tripal_cv_api',
  248. TRIPAL_ERROR,
  249. "tripal_get_cvterm: chado_generate_var() failed to return a cvterm based on the identifiers
  250. you passed in. You should check that your identifiers are correct, as well as, look
  251. for a chado_generate_var error for additional clues. You passed in %identifier.",
  252. array(
  253. '%identifier'=> print_r($identifiers, TRUE)
  254. )
  255. );
  256. }
  257. // Else, as far we know, everything is fine so give them their cvterm :)
  258. else {
  259. return $cvterm;
  260. }
  261. }
  262. /**
  263. * Create an options array to be used in a form element
  264. * which provides a list of all chado cvterms
  265. *
  266. * @param $cv_id
  267. * The chado cv_id; only cvterms with the supplied cv_id will be returned
  268. * @param $rel_type
  269. * Set to TRUE if the terms returned should only be relationship types in
  270. * the vocabulary. This is useful for creating drop-downs of terms
  271. * used for relationship linker tables.
  272. *
  273. * @return
  274. * An associative array with the cvterm_id's as keys. The first
  275. * element in the array has a key of '0' and a value of 'Select a Type'
  276. *
  277. * @ingroup tripal_chado_api
  278. */
  279. function tripal_get_cvterm_select_options($cv_id, $rel_type = FALSE) {
  280. $columns = array('cvterm_id', 'name');
  281. $values = array('cv_id' => $cv_id);
  282. if ($rel_type) {
  283. $values['is_relationshiptype'] = 1;
  284. }
  285. $s_options = array('order_by' => array('name' => 'ASC'));
  286. $cvterms = chado_select_record('cvterm', $columns, $values, $s_options);
  287. $options = array();
  288. $options[0] = 'Select a Type';
  289. foreach ($cvterms as $cvterm) {
  290. $options[$cvterm->cvterm_id] = $cvterm->name;
  291. }
  292. return $options;
  293. }
  294. /**
  295. * Updates the cvtermpath table of Chado for the specified CV.
  296. *
  297. * @param $cv_id
  298. * The chado cv_id;
  299. * @param $job_id
  300. * This function is intended to be used with the Tripal Jobs API.
  301. * When this function is called as a job the $job_id is automatically
  302. * passed to this function.
  303. * @return
  304. * TRUE on success FALSE on failure
  305. *
  306. * @ingroup tripal_chado_api
  307. */
  308. function tripal_update_cvtermpath($cv_id, $job_id = NULL) {
  309. // TODO: need better error checking in this function
  310. // first get the controlled vocabulary name:
  311. $sql = "SELECT * FROM {cv} WHERE cv_id = :cv_id";
  312. $cv = chado_query($sql, array(':cv_id' => $cv_id))->fetchObject();
  313. print "\nUpdating cvtermpath for $cv->name...\n";
  314. // We need to set the chado schema as active because some of the
  315. // functions call other functions which would not be in scope.
  316. $previous = chado_set_active('chado');
  317. try {
  318. $sql = "SELECT * FROM fill_cvtermpath(:name)";
  319. db_query($sql, array(':name' => $cv->name));
  320. chado_set_active($previous);
  321. }
  322. catch (Exception $e) {
  323. chado_set_active($previous);
  324. $error = $e->getMessage();
  325. tripal_report_error('tripal_chado', TRIPAL_ERROR, "Could not fill cvtermpath table: @error", array('@error' => $error));
  326. return FALSE;
  327. }
  328. return TRUE;
  329. }
  330. /**
  331. *
  332. * @param unknown $cv_id
  333. * @param string $job_id
  334. */
  335. function tripal_update_cvtermpath_bak($cv_id, $job_id = NULL){
  336. chado_set_active('chado');
  337. $result = db_query('
  338. SELECT DISTINCT t.*
  339. FROM cvterm t
  340. LEFT JOIN cvterm_relationship r ON (t.cvterm_id = r.subject_id)
  341. INNER JOIN cvterm_relationship r2 ON (t.cvterm_id = r2.object_id)
  342. WHERE t.cv_id = :cvid AND r.subject_id is null',
  343. array(':cvid' => $cv_id)
  344. );
  345. $record = $result->fetchObject();
  346. tripal_update_cvtermpath_root_loop($record->cvterm_id, $record->cv_id);
  347. }
  348. /**
  349. *
  350. * @param unknown $rootid
  351. * @param unknown $cvid
  352. */
  353. function tripal_update_cvtermpath_root_loop($rootid, $cvid){
  354. chado_set_active('chado');
  355. $ttype = db_select('cvterm', 'cv')
  356. ->fields('cv', array('cvterm_id'));
  357. $db_or = db_or();
  358. $db_or->condition('cv.name', "isa", '=' );
  359. $db_or->condition('cv.name', "is_a", '=' );
  360. $ttype->condition($db_or);
  361. $result = $ttype->execute()->fetchObject();
  362. tripal_update_cvtermpath_loop($rootid, $rootid, $cvid, $result->cvterm_id, 0, 0, array());
  363. $cterm = db_query(
  364. 'SELECT *
  365. FROM cvterm_relationship
  366. WHERE object_id = :rootid
  367. ',
  368. array(':rootid' => $rootid)
  369. );
  370. while($cterm_result = $cterm->fetchAssoc()) {
  371. tripal_update_cvtermpath_root_loop($cterm_result['subject_id'], $cvid);
  372. };
  373. }
  374. /**
  375. *
  376. * @param $origin
  377. * @param $subject_id
  378. * @param $cv_id
  379. * @param $type_id
  380. * @param $depth
  381. * @return multitype:
  382. */
  383. function tripal_update_cvtermpath_loop($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, $array_of_children){
  384. // Check to see if a row with these values already exists.
  385. chado_set_active('chado');
  386. $count = db_query(
  387. 'SELECT *
  388. FROM cvtermpath
  389. WHERE cv_id = :cvid
  390. AND object_id = :origin
  391. AND subject_id = :child_id
  392. AND pathdistance = :depth
  393. ',
  394. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  395. );
  396. $count_total = $count->rowCount();
  397. try{
  398. if($count_total == 0) {
  399. // If row with values does not already exist write to table.
  400. chado_set_active('chado');
  401. $query = db_insert('cvtermpath')
  402. ->fields(array(
  403. 'object_id' => $origin,
  404. 'subject_id' => $child_id,
  405. 'cv_id' => $cv_id,
  406. 'type_id' => $type_id,
  407. 'pathdistance' => $depth,
  408. ));
  409. $rows = $query->execute();
  410. }
  411. // Build the ID.
  412. $children_id = $origin.$child_id.$cv_id.$type_id;
  413. // Now check if the most recent entry already exists in the array.
  414. if($increment_of_depth != 0){
  415. // Search the $array_of_children for the new $child_id in the build_id column.
  416. foreach ($array_of_children as $key => $val) {
  417. if ($val['build_id'] === $children_id) {
  418. $possible_loop_start = $val;
  419. watchdog('debug', '<pre>tripal_update_cvtermpath_loop_checker $val: '. print_r($val, TRUE) .'</pre>');
  420. // If the search returns something check for a possible loop.
  421. //if(!empty($possible_loop_starts)){
  422. //foreach($possible_loop_starts as $possible_loop_start){
  423. // Call the loop checker function.
  424. $result_of_loop_checker = tripal_update_cvtermpath_loop_checker($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, 0, $possible_loop_start, array(), $depth);
  425. if($result_of_loop_checker == TRUE){
  426. continue;
  427. // }
  428. //}
  429. }
  430. }
  431. }
  432. }
  433. // Then add that new entry to the $array_of_children.
  434. $array_of_children[$increment_of_depth] = array('build_id'=>$children_id, 'depth' => $depth);
  435. //watchdog('debug', '<pre>tripal_update_cvtermpath_loop_checker $array_of_children: '. print_r($array_of_children, TRUE) .'</pre>');
  436. if ($count_total == 0) {
  437. chado_set_active('chado');
  438. $query = db_select('cvterm_relationship', 'cvtr')
  439. ->fields('cvtr')
  440. ->condition('cvtr.object_id', $child_id, '=')
  441. ->execute();
  442. $cterm = $query->fetchAll();
  443. foreach ($cterm as $item) {
  444. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $item ' . print_r($item, TRUE) . '</pre>');
  445. $increment_of_depth++;
  446. tripal_update_cvtermpath_loop($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $increment_of_depth, $array_of_children);
  447. };
  448. //}
  449. }
  450. }
  451. catch(Exception $e){
  452. watchdog_exception('tripal_ds', $e);
  453. return FALSE;
  454. }
  455. return 1;
  456. }
  457. /**
  458. *
  459. * @param $origin
  460. * @param $subject_id
  461. * @param $cv_id
  462. * @param $type_id
  463. * @param $depth
  464. * @return bool
  465. */
  466. function tripal_update_cvtermpath_loop_checker($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, $distance_between_parent_child, $possible_start_of_loop, $array_of_possible_loop, $depth_at_start_of_loop){
  467. if ($distance_between_parent_child > 1) {
  468. // Search the $array_of_children for the new $child_id in the build_id column.
  469. //$possible_loop_starts = array_keys(array_column($array_of_possible_loop, 'build_id'), $possible_start_of_loop);
  470. foreach ($array_of_possible_loop as $key => $val) {
  471. watchdog('debug', '<pre>tripal_update_cvtermpath_loop_checker $possible_start_of_loop[\'build_id\']: '. print_r($possible_start_of_loop['build_id'], TRUE) .'</pre>');
  472. if ($val['build_id'] === $possible_start_of_loop['build_id']) {
  473. $possible_loop_starts = $val;
  474. // If the search returns something check for a possible loop.
  475. if (!empty($possible_loop_starts)) {
  476. foreach ($possible_loop_starts as $possible_loop_start) {
  477. watchdog('debug', '<pre>tripal_update_cvtermpath_loop_checker $possible_loop_start[\'depth\']: '. print_r($possible_loop_start['depth'], TRUE) .'</pre>');
  478. // Check the depth measurements, if the depth of the passed relationships minus the depth of the
  479. // possible_start of loop is equal to the $distance_between_parent_child it's a loop.
  480. if (($depth_at_start_of_loop - $possible_loop_start['depth']) == $distance_between_parent_child) {
  481. //This is a loop so it needs to be terminated.
  482. return TRUE;
  483. }
  484. }
  485. }
  486. }
  487. }
  488. chado_set_active('chado');
  489. $query = db_select('cvterm_relationship', 'cvtr')
  490. ->fields('cvtr')
  491. ->condition('cvtr.object_id', $child_id, '=')
  492. ->execute();
  493. $cterm = $query->fetchAll();
  494. foreach ($cterm as $item) {
  495. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $item ' . print_r($item, TRUE) . '</pre>');
  496. $increment_of_depth++;
  497. $distance_between_parent_child++;
  498. $children_id = $origin . $item->subject_id . $cv_id . $item->type_id;
  499. $array_of_possible_loop[ $distance_between_parent_child ] = array('build_id' => $children_id, 'depth' => $depth);
  500. tripal_update_cvtermpath_loop_checker($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $increment_of_depth, $distance_between_parent_child, $possible_start_of_loop, $array_of_possible_loop, $depth_at_start_of_loop);
  501. };
  502. return FALSE;
  503. }
  504. }
  505. /*
  506. *
  507. * @param $origin
  508. * @param $subject_id
  509. * @param $cv_id
  510. * @param $type_id
  511. * @param $depth
  512. * @return multitype:
  513. function tripal_update_cvtermpath_loop($origin, $child_id, $cv_id, $type_id, $depth){
  514. // Variables and arrays needed for loop checking.
  515. $array_of_children;
  516. $array_of_possible_loop;
  517. $possible_start_of_loop;
  518. $distance_between_parent_child;
  519. $increment_of_depth;
  520. chado_set_active('chado');
  521. $count = db_query(
  522. 'SELECT *
  523. FROM cvtermpath
  524. WHERE cv_id = :cvid
  525. AND object_id = :origin
  526. AND subject_id = :child_id
  527. AND pathdistance = :depth
  528. ',
  529. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  530. );
  531. $count_total = $count->rowCount();
  532. //Loop check
  533. chado_set_active('chado');
  534. $loop = db_query(
  535. 'SELECT *
  536. FROM cvtermpath
  537. WHERE cv_id = :cvid
  538. AND object_id = :origin
  539. AND subject_id = :child_id
  540. AND type_id = :type_id
  541. ',
  542. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':type_id' => $type_id,)
  543. );
  544. $loop_check = $loop->rowCount();
  545. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $rows ' . print_r($rows, TRUE) . '</pre>');
  546. /*if(!empty($rows)){
  547. foreach($rows as $row){
  548. tripal_update_cvtermpath_loop_check($origin, $child_id, $cv_id, $type_id, $depth, $row->cvtermpath_id, 0);
  549. }
  550. }
  551. else {*/
  552. //If no loop proceed.
  553. /*try{
  554. if($count_total == 0) {
  555. chado_set_active('chado');
  556. $query = db_insert('cvtermpath')
  557. ->fields(array(
  558. 'object_id' => $origin,
  559. 'subject_id' => $child_id,
  560. 'cv_id' => $cv_id,
  561. 'type_id' => $type_id,
  562. 'pathdistance' => $depth,
  563. ));
  564. $rows = $query->execute();
  565. }
  566. if ($loop_check == 0) {
  567. chado_set_active('chado');
  568. $query = db_select('cvterm_relationship', 'cvtr')
  569. ->fields('cvtr')
  570. ->condition('cvtr.object_id', $child_id, '=')
  571. ->execute();
  572. $cterm = $query->fetchAll();
  573. foreach ($cterm as $item) {
  574. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $item ' . print_r($item, TRUE) . '</pre>');
  575. tripal_update_cvtermpath_loop($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1);
  576. };
  577. //}
  578. }
  579. }
  580. catch(Exception $e){
  581. watchdog_exception('tripal_ds', $e);
  582. return FALSE;
  583. }
  584. return 1;
  585. }
  586. */
  587. /**
  588. *
  589. * @param $origin
  590. * @param $subject_id
  591. * @param $cv_id
  592. * @param $type_id
  593. * @param $depth
  594. * @return multitype:
  595. function tripal_update_cvtermpath_loop_check($origin, $child_id, $cv_id, $type_id, $depth, $cvtermpath_id, $loop_count, $loop_check, $object_id){
  596. //Store the
  597. //Check if the passed parameters match any of the items in the loop_check array.
  598. if(!empty($loop_check)){
  599. foreach($loop_check as $item){
  600. if ($item['type_id'] = $type_id){
  601. if($item['subject_id'] = $child_id){
  602. if($item['object_id'] = $object_id){
  603. //Loop found, roll back all rows until $cvtermpath_id-1 (last correct entry)
  604. // and step into the next loop
  605. }
  606. }
  607. }
  608. }
  609. }
  610. $loop_count + 1;
  611. chado_set_active('chado');
  612. $count = db_query(
  613. 'SELECT *
  614. FROM cvtermpath
  615. WHERE cv_id = :cvid AND object_id = :origin
  616. AND subject_id = :child_id
  617. AND pathdistance = :depth
  618. ',
  619. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  620. );
  621. $count_total = $count->rowCount();
  622. if ($count_total == 0) {
  623. chado_set_active('chado');
  624. $query = db_insert('cvtermpath')
  625. ->fields(array(
  626. 'object_id' => $origin,
  627. 'subject_id' => $child_id,
  628. 'cv_id' => $cv_id,
  629. 'type_id' => $type_id,
  630. 'pathdistance' => $depth,
  631. ));
  632. $rows = $query->execute();
  633. $cterm = array();
  634. $query = db_select('cvterm_relationship', 'cvtr')
  635. ->fields('cvtr')
  636. ->condition('cvtr.object_id', $child_id, '=' )
  637. ->execute();
  638. $cterm = $query->fetchAll();
  639. foreach ($cterm as $item) {
  640. $loop_check[$loop_count]= $item;
  641. tripal_update_cvtermpath_loop_check($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $loop_count, $loop_check, $item->object_id);
  642. };
  643. }
  644. return 1;
  645. }
  646. */
  647. /**
  648. * Adds a controlled vocabular to the CV table of Chado.
  649. *
  650. * @param $name
  651. * The name of the controlled vocabulary. These are typically all lower case
  652. * with no special characters other than an undrescore (for spaces).
  653. * @param $comment
  654. * A description or definition of the vocabulary.
  655. *
  656. * @return
  657. * An object populated with fields from the newly added database.
  658. *
  659. * @ingroup tripal_chado_api
  660. */
  661. function tripal_insert_cv($name, $definition) {
  662. // insert/update values
  663. $ins_values = array(
  664. 'name' => $name,
  665. 'definition' => $definition
  666. );
  667. // see if the CV default exists already in the database
  668. $sel_values = array('name' => $name);
  669. $results = chado_select_record('cv', array('*'), $sel_values);
  670. // if it does not exists then add it
  671. if (count($results) == 0) {
  672. $success = chado_insert_record('cv', $ins_values);
  673. if (!$success) {
  674. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to create the CV record", NULL);
  675. return FALSE;
  676. }
  677. $results = chado_select_record('cv', array('*'), $sel_values);
  678. }
  679. // if it already exists then do an update
  680. else {
  681. $success = chado_update_record('cv', $sel_values, $ins_values);
  682. if (!$success) {
  683. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the CV record", NULL);
  684. return FALSE;
  685. }
  686. $results = chado_select_record('cv', array('*'), $sel_values);
  687. }
  688. // return the cv object
  689. return $results[0];
  690. }
  691. /**
  692. * Add's a controlled vocabulary term to Chado.
  693. *
  694. * This function will add a cvterm record (and a dbxref record if appropriate
  695. * values are provided). If the parent vocabulary does not exist then
  696. * that also is added to the cv table. If the cvterm is a relationship term
  697. * then the 'is_relationship' value should be set. All
  698. * terms must also have a corresponding database. This is specified in the
  699. * term's ID just before the colon (e.g. GO:003824). If the database does not
  700. * exist in the DB table then it will be added automatically. The accession
  701. * (the value just after the colon in the term's ID) will be added to the
  702. * dbxref table. If the CVterm already exists and $update is set (default)
  703. * then the cvterm is updated. If the CVTerm already exists and $update is
  704. * not set, then no changes are made and the CVTerm object is returned.
  705. *
  706. * @param $term
  707. * An associative array with the following keys:
  708. * - id: the term accession. must be of the form <DB>:<ACCESSION>, where
  709. * <DB> is the name of the database to which the cvterm belongs and the
  710. * <ACCESSION> is the term's accession number in the database.
  711. * - name: the name of the term. usually meant to be human-readable.
  712. * - is_obsolete: is present and set to 1 if the term is defunct
  713. * - definition: the definition of the term
  714. * - cv_name: The CV name to which the term belongs. If this arugment is
  715. * null or not provided then the function tries to find a record in the
  716. * CV table with the same name provided in the $term[namespace]. If
  717. * this field is provided then it overrides what the value in
  718. * $term[namespace]
  719. * - is_relationship: If this term is a relationship term then this value
  720. * should be 1.
  721. * - db_name: In some cases the database name will not be part of the
  722. * $term['id'] and it needs to be explicitly set. Use this argument
  723. * only if the database name cannot be specififed in the term ID
  724. * (e.g. <DB>:<ACCESSION>).
  725. * @param $options
  726. * An associative array with the following keys:
  727. * - update_existing: By default this is TRUE. If the term exists it is
  728. * automatically updated.
  729. * - force_db_change: Sometimes a term may need to switch from one
  730. * database to another. If the term already exists, but associated
  731. * with another term the insert (or update, rather) will fail. Set
  732. * this variable to TRUE to force the change to occur.
  733. *
  734. * @return
  735. * A cvterm object
  736. *
  737. * @ingroup tripal_chado_api
  738. */
  739. function tripal_insert_cvterm($term, $options = array()) {
  740. // get the term properties
  741. $id = (isset($term['id'])) ? $term['id'] : '';
  742. $name = '';
  743. $cvname = '';
  744. $definition = '';
  745. $is_obsolete = 0;
  746. $accession = '';
  747. // Set Defaults
  748. if (isset($term['cv_name'])) {
  749. $cvname = $term['cv_name'];
  750. }
  751. else {
  752. $cvname = 'local';
  753. }
  754. // Namespace is deprecated but must be supported for backwards
  755. // compatability
  756. if (array_key_exists('namespace', $term)) {
  757. $cvname = $term['namespace'];
  758. }
  759. if (isset($term['is_relationship'])) {
  760. $is_relationship = $term['is_relationship'];
  761. }
  762. else {
  763. $is_relationship = 0;
  764. }
  765. if (isset($term['db_name'])) {
  766. $dbname = $term['db_name'];
  767. }
  768. else {
  769. $dbname = 'local';
  770. }
  771. if (isset($options['update_existing'])) {
  772. $update = $options['update_existing'];
  773. }
  774. else {
  775. $update = 1;
  776. }
  777. if (array_key_exists('name', $term)) {
  778. $name = $term['name'];
  779. }
  780. else {
  781. $name = $id;
  782. }
  783. if (array_key_exists('definition', $term)) {
  784. $definition = preg_replace('/^\"(.*)\"/', '\1', $term['definition']);
  785. }
  786. else {
  787. $definition = '';
  788. }
  789. if (array_key_exists('is_obsolete', $term)) {
  790. $is_obsolete = $term['is_obsolete'];
  791. if (strcmp($is_obsolete, 'true') == 0) {
  792. $is_obsolete = 1;
  793. }
  794. }
  795. if (!$name and !$id) {
  796. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Cannot find cvterm without 'id' or 'name'", NULL);
  797. return 0;
  798. }
  799. if (!$id) {
  800. $id = $name;
  801. }
  802. // Get the accession and the database from the cvterm id.
  803. if ($dbname) {
  804. $accession = $id;
  805. }
  806. if (preg_match('/^.+?:.*$/', $id)) {
  807. $accession = preg_replace('/^.+?:(.*)$/', '\1', $id);
  808. $dbname = preg_replace('/^(.+?):.*$/', '\1', $id);
  809. }
  810. // Check that we have a database name, give a different message if it's a
  811. // relationship.
  812. if ($is_relationship and !$dbname) {
  813. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  814. "A database name is not provided for this relationship term: $id", NULL);
  815. return 0;
  816. }
  817. if (!$is_relationship and !$dbname) {
  818. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  819. "A database identifier is missing from the term: $id", NULL);
  820. return 0;
  821. }
  822. // Make sure the CV name exists
  823. $cv = tripal_get_cv(array('name' => $cvname));
  824. if (!$cv) {
  825. $cv = tripal_insert_cv($cvname, '');
  826. }
  827. if (!$cv) {
  828. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  829. "Cannot find namespace '$cvname' when adding/updating $id", NULL);
  830. return 0;
  831. }
  832. // This SQL statement will be used a lot to find a cvterm so just set it
  833. // here for easy reference below. Because CV terms can change their names
  834. // but accessions don't change, the following SQL finds cvterms based on
  835. // their accession rather than the name.
  836. $cvtermsql = "
  837. SELECT CVT.name, CVT.cvterm_id, CV.cv_id, CV.name as cvname,
  838. DB.name as dbname, DB.db_id, DBX.accession
  839. FROM {cvterm} CVT
  840. INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
  841. INNER JOIN {db} DB on DBX.db_id = DB.db_id
  842. INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
  843. WHERE DBX.accession = :accession and DB.name = :name
  844. ";
  845. // Add the database. The function will just return the DB object if the
  846. // database already exists.
  847. $db = tripal_get_db(array('name' => $dbname));
  848. if (!$db) {
  849. $db = tripal_insert_db(array('name' => $dbname));
  850. }
  851. if (!$db) {
  852. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Cannot find database '$dbname' in Chado.", NULL);
  853. return 0;
  854. }
  855. // The cvterm table has two unique dependencies. We need to check both.
  856. // first check the (name, cv_id, is_obsolete) constraint.
  857. $values = array(
  858. 'name' => $name,
  859. 'is_obsolete' => $is_obsolete,
  860. 'cv_id' => array(
  861. 'name' => $cvname,
  862. ),
  863. );
  864. $result = chado_select_record('cvterm', array('*'), $values);
  865. if (count($result) == 1) {
  866. $cvterm = $result[0];
  867. // Get the dbxref record.
  868. $values = array('dbxref_id' => $cvterm->dbxref_id);
  869. $result = chado_select_record('dbxref', array('*'), $values);
  870. $dbxref = $result[0];
  871. if (!$dbxref) {
  872. tripal_report_error('tripal_cv', TRIPAL_ERROR,
  873. 'Unable to access the dbxref record for the :term cvterm. Term Record: !record',
  874. array(':term' => $name, '!record' => print_r($cvterm, TRUE))
  875. );
  876. return FALSE;
  877. }
  878. // Get the db.
  879. $values = array('db_id' => $dbxref->db_id);
  880. $result = chado_select_record('db', array('*'), $values);
  881. $db_check = $result[0];
  882. // If the database from the existing record doesn't match the one that
  883. // has been provided then we have a problem. The term already exists
  884. // but we don't want to just switch the term, the callee must force it.
  885. if ($db_check->name != $db->name) {
  886. if (array_key_exists('force_db_change', $options) and
  887. $options['force_db_change'] != TRUE) {
  888. tripal_report_error('tripal_cv', TRIPAL_ERROR,
  889. 'The term already exists, but associated with a different database record: \'!db\'. It cannot be added: !record',
  890. array(':term' => $name, '!db' => $db_check->name, '!record' => print_r($cvterm, TRUE))
  891. );
  892. return FALSE;
  893. }
  894. // Look to see if the correct dbxref record already exists for this
  895. // database.
  896. $values = array(
  897. 'db_id' => $db->db_id,
  898. 'accession' => $accession,
  899. );
  900. $result = chado_select_record('dbxref', array('*'), $values);
  901. // If we already have a good dbxref then we want to update our cvterm
  902. // to use this dbxref.
  903. if (count($result) > 0) {
  904. $dbxref = $result[0];
  905. $match = array('cvterm_id' => $cvterm->cvterm_id);
  906. $values = array('dbxref_id' => $dbxref->dbxref_id);
  907. $success = chado_update_record('cvterm', $match, $values);
  908. if (!$success) {
  909. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to correct the dbxref id for the cvterm " .
  910. "'$name' (id: $accession), for database $dbname", NULL);
  911. return 0;
  912. }
  913. }
  914. // If we don't have the dbxref then we want to delete our cvterm and let
  915. // the code below recreate it with the correct info.
  916. else {
  917. $match = array('cvterm_id' => $cvterm->cvterm_id);
  918. chado_delete_record('cvterm', $match);
  919. }
  920. }
  921. // Check that the accession matches. Sometimes an OBO can define a term
  922. // multiple times but with different accessions. If this is the case we
  923. // can't do an insert or it will violate the constraint in the cvterm table.
  924. // So we'll need to add the record to the cvterm_dbxref table instead.
  925. if ($dbxref->accession != $accession) {
  926. // Get/add the dbxref for his term.
  927. $dbxref_new = tripal_insert_dbxref(array(
  928. 'db_id' => $db->db_id,
  929. 'accession' => $accession
  930. ));
  931. if (!$dbxref_new) {
  932. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  933. "$name (id: $accession), for database $dbname", NULL);
  934. return 0;
  935. }
  936. // Check to see if the cvterm_dbxref record already exists.
  937. $values = array(
  938. 'cvterm_id' => $cvterm->cvterm_id,
  939. 'dbxref_id' => $dbxref_new->dbxref_id,
  940. 'is_for_definition' => 1,
  941. );
  942. $result = chado_select_record('cvterm_dbxref', array('*'), $values);
  943. // if the cvterm_dbxref record does not exists then add it
  944. if (count($result)==0) {
  945. $options = array(
  946. 'return_record' => FALSE,
  947. );
  948. $success = chado_insert_record('cvterm_dbxref', $values, $options);
  949. if (!$success) {
  950. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the cvterm_dbxref record for a " .
  951. "duplicated cvterm: $name (id: $accession), for database $dbname", NULL);
  952. return 0;
  953. }
  954. }
  955. // get the original cvterm with the same name and return that.
  956. $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
  957. $cvterm = $result->fetchObject();
  958. return $cvterm;
  959. }
  960. // Continue on, we've fixed the record if the db_id did not match.
  961. // We can now perform and updated if we need to.
  962. }
  963. // get the CVterm record
  964. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  965. $cvterm = $result->fetchObject();
  966. if (!$cvterm) {
  967. // Check to see if the dbxref exists if not, add it.
  968. $dbxref = tripal_insert_dbxref(array(
  969. 'db_id' => $db->db_id,
  970. 'accession' => $accession
  971. ));
  972. if (!$dbxref) {
  973. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  974. "$name (id: $accession), for database $dbname", NULL);
  975. return 0;
  976. }
  977. // Check to see if the dbxref already has an entry in the cvterm table.
  978. $values = array('dbxref_id' => $dbxref->dbxref_id);
  979. $check = chado_select_record('cvterm', array('cvterm_id'), $values);
  980. if (count($check) == 0) {
  981. // now add the cvterm
  982. $ins_values = array(
  983. 'cv_id' => $cv->cv_id,
  984. 'name' => $name,
  985. 'definition' => $definition,
  986. 'dbxref_id' => $dbxref->dbxref_id,
  987. 'is_obsolete' => $is_obsolete,
  988. 'is_relationshiptype' => $is_relationship,
  989. );
  990. $success = chado_insert_record('cvterm', $ins_values);
  991. if (!$success) {
  992. if (!$is_relationship) {
  993. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to insert the term: $name ($dbname)", NULL);
  994. return 0;
  995. }
  996. else {
  997. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to insert the relationship term: $name (cv: " . $cvname . " db: $dbname)", NULL);
  998. return 0;
  999. }
  1000. }
  1001. }
  1002. // This dbxref already exists in the cvterm table.
  1003. else {
  1004. tripal_report_error('tripal_chado', TRIPAL_WARNING, "The dbxref already exists for another cvterm record: $name (cv: " . $cvname . " db: $dbname)", NULL);
  1005. return 0;
  1006. }
  1007. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  1008. $cvterm = $result->fetchObject();
  1009. }
  1010. // Update the cvterm.
  1011. elseif ($update) {
  1012. // First, basic update of the term.
  1013. $match = array('cvterm_id' => $cvterm->cvterm_id);
  1014. $upd_values = array(
  1015. 'name' => $name,
  1016. 'definition' => $definition,
  1017. 'is_obsolete' => $is_obsolete,
  1018. 'is_relationshiptype' => $is_relationship,
  1019. );
  1020. $success = chado_update_record('cvterm', $match, $upd_values);
  1021. if (!$success) {
  1022. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the term: $name", NULL);
  1023. return 0;
  1024. }
  1025. // Second, check that the dbxref has not changed and if it has then update it.
  1026. $checksql = "
  1027. SELECT cvterm_id
  1028. FROM {cvterm} CVT
  1029. INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
  1030. INNER JOIN {db} DB on DBX.db_id = DB.db_id
  1031. INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
  1032. WHERE DBX.accession = :accession and DB.name = :dbname and CVT.name = :term and CV.name = :cvname
  1033. ";
  1034. $check = chado_query($checksql, array(':accession' => $dbxref->accession, ':dbname' => $dbname, ':term' => $name, ':cvname' => $cvname))->fetchObject();
  1035. if (!$check) {
  1036. // check to see if the dbxref exists if not, add it.
  1037. $dbxref = tripal_insert_dbxref(array(
  1038. 'db_id' => $db->db_id,
  1039. 'accession' => $accession
  1040. ));
  1041. if (!$dbxref) {
  1042. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  1043. "$name (id: $accession), for database $dbname", NULL);
  1044. return 0;
  1045. }
  1046. $match = array('cvterm_id' => $cvterm->cvterm_id);
  1047. $upd_values = array(
  1048. 'dbxref_id' => $dbxref->dbxref_id,
  1049. );
  1050. $success = chado_update_record('cvterm', $match, $upd_values);
  1051. if (!$success) {
  1052. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the term $name with new accession $db:$accession", NULL);
  1053. return 0;
  1054. }
  1055. }
  1056. // Finally grab the updated details.
  1057. $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
  1058. $cvterm = $result->fetchObject();
  1059. }
  1060. else {
  1061. // do nothing, we have the cvterm but we don't want to update
  1062. }
  1063. // return the cvterm
  1064. return $cvterm;
  1065. }
  1066. /**
  1067. * This function allows other modules to programatically
  1068. * submit an ontology for loading into Chado.
  1069. *
  1070. * This function will add a job to the Jobs subsystem for parsing the ontology.
  1071. * You can either pass a known OBO ID to the function or the URL
  1072. * or full path the the ontology file. If a URL or file name is
  1073. * passed then the $obo_name argument must also be provided. If
  1074. * this is the first time the ontology has been provided to Tripal
  1075. * then it will be added to the database and will be assigned a
  1076. * unique OBO ID.
  1077. *
  1078. * @param $obo_id
  1079. * If the ontology is already loaded into the Tripal tables then
  1080. * use this argument to specify the unique ID for the ontology
  1081. * that will be loaded
  1082. * @param $obo_name
  1083. * If the OBO has not been added before then use this argument
  1084. * to specify the human readable name of the ontology.
  1085. * @param $obo_url
  1086. * If the OBO to be loaded is located on a remote server then
  1087. * use this argument to provide the URL.
  1088. * @param $obo_file
  1089. * If the OBO is housed on the local file system of the server then
  1090. * use this argument to specify the full path.
  1091. *
  1092. * @return
  1093. * returns the job_id of the submitted job or FALSE if the job was not added
  1094. *
  1095. * @ingroup tripal_chado_api
  1096. */
  1097. function tripal_submit_obo_job($obo) {
  1098. global $user;
  1099. // Set Defaults
  1100. $obo['obo_id'] = (isset($obo['obo_id'])) ? $obo['obo_id'] : NULL;
  1101. $obo['name'] = (isset($obo['name'])) ? $obo['name'] : NULL;
  1102. $obo['url'] = (isset($obo['url'])) ? $obo['url'] : NULL;
  1103. $obo['file'] = (isset($obo['file'])) ? $obo['file'] : NULL;
  1104. $includes = array(
  1105. module_load_include('inc', 'tripal_chado', 'includes/loaders/tripal_chado.obo_loader'),
  1106. );
  1107. if ($obo['obo_id']) {
  1108. $sql = "SELECT * FROM {tripal_cv_obo} WHERE obo_id = :obo_id";
  1109. $result = db_query($sql, array(':obo_id' => $obo['obo_id']))->fetchObject();
  1110. $args = array($result->obo_id);
  1111. return tripal_add_job("Load OBO " . $result->name, 'tripal_chado',
  1112. "tripal_chado_load_obo_v1_2_id", $args, $user->uid, 10, $includes);
  1113. }
  1114. else {
  1115. if ($obo['url']) {
  1116. $args = array($obo['name'], $obo['url']);
  1117. return tripal_add_job("Load OBO " . $obo['name'], 'tripal_chado',
  1118. "tripal_chado_load_obo_v1_2_url", $args, $user->uid, 10, $includes);
  1119. }
  1120. elseif ($obo['file']) {
  1121. $args = array($obo['name'], $obo['file']);
  1122. return tripal_add_job("Load OBO " . $obo['name'], 'tripal_chado',
  1123. "tripal_chado_load_obo_v1_2_file", $args, $user->uid, 10, $includes);
  1124. }
  1125. }
  1126. return FALSE;
  1127. }
  1128. /**
  1129. * Add the OBO to the tripal_cv_obo table in the Drupal database.
  1130. *
  1131. * If the OBO name already exists in the table then the path is updated.
  1132. *
  1133. * @param $name
  1134. * The human readable name of this ontology
  1135. * @param $path
  1136. * The file path or URL of the ontology
  1137. *
  1138. * @return
  1139. * Returns the ontology ID
  1140. *
  1141. * @ingroup tripal_chado_api
  1142. */
  1143. function tripal_insert_obo($name, $path) {
  1144. // make sure an OBO with the same name doesn't already exist
  1145. $obo_id = db_select('tripal_cv_obo', 'tco')
  1146. ->fields('tco', array('obo_id'))
  1147. ->condition('name', $name)
  1148. ->execute()
  1149. ->fetchField();
  1150. if ($obo_id) {
  1151. db_update('tripal_cv_obo')
  1152. ->fields(array(
  1153. 'path' => $path,
  1154. ))
  1155. ->condition('name', $name)
  1156. ->execute();
  1157. return $obo_id;
  1158. }
  1159. else {
  1160. $obo_id = db_insert('tripal_cv_obo')
  1161. ->fields(array(
  1162. 'name' => $name,
  1163. 'path' => $path,
  1164. ))
  1165. ->execute();
  1166. return $obo_id;
  1167. }
  1168. }
  1169. /**
  1170. * Retrieves an OBO record.
  1171. *
  1172. * @param $values
  1173. * An associate array with the following allowed keys: obo_id, name
  1174. *
  1175. * @return
  1176. * An instance of an OBO record object.
  1177. */
  1178. function tripal_get_obo($values) {
  1179. $query = db_select('tripal_cv_obo', 'tco')
  1180. ->fields('tco');
  1181. if (array_key_exists('name', $values)) {
  1182. $query->condition('tco.name', $values['name']);
  1183. }
  1184. if (array_key_exists('obo_id', $values)) {
  1185. $query->condition('tco.obo_id', $values['obo_id']);
  1186. }
  1187. return $query->execute()->fetchObject();
  1188. }
  1189. /**
  1190. * This function is intended to be used in autocomplete forms
  1191. * for searching for CV terms that begin with the provided string
  1192. *
  1193. * @param $cv_id
  1194. * The CV ID in which to search for the term
  1195. * @param $string
  1196. * The string to search for
  1197. *
  1198. * @return
  1199. * A json array of terms that begin with the provided string
  1200. *
  1201. * @ingroup tripal_chado_api
  1202. */
  1203. function tripal_autocomplete_cvterm($cv_id, $string = '') {
  1204. if ($cv_id) {
  1205. $sql = "
  1206. SELECT CVT.cvterm_id, CVT.name
  1207. FROM {cvterm} CVT
  1208. WHERE CVT.cv_id = :cv_id and lower(CVT.name) like lower(:name)
  1209. UNION
  1210. SELECT CVT2.cvterm_id, CVTS.synonym as name
  1211. FROM {cvterm} CVT2
  1212. INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
  1213. WHERE CVT2.cv_id = :cv_id and lower(CVTS.synonym) like lower(:name)
  1214. ORDER by name
  1215. LIMIT 25 OFFSET 0
  1216. ";
  1217. $results = chado_query($sql, array(':cv_id' => $cv_id, ':name' => $string . '%'));
  1218. $items = array();
  1219. foreach ($results as $term) {
  1220. $items[$term->name] = $term->name;
  1221. }
  1222. }
  1223. // If a CV wasn't provided then search all of them, and include the cv
  1224. // in the results.
  1225. else {
  1226. $sql = "
  1227. SELECT CVT.cvterm_id, CVT.name, CV.name as cvname, CVT.cv_id
  1228. FROM {cvterm} CVT
  1229. INNER JOIN {cv} CV on CVT.cv_id = CV.cv_id
  1230. WHERE lower(CVT.name) like lower(:name)
  1231. UNION
  1232. SELECT CVT2.cvterm_id, CVTS.synonym as name, CV2.name as cvname, CVT2.cv_id
  1233. FROM {cvterm} CVT2
  1234. INNER JOIN {cv} CV2 on CVT2.cv_id = CV2.cv_id
  1235. INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
  1236. WHERE lower(CVTS.synonym) like lower(:name)
  1237. ORDER by name
  1238. LIMIT 25 OFFSET 0
  1239. ";
  1240. $results = chado_query($sql, array(':name' => $string . '%'));
  1241. $items = array();
  1242. foreach ($results as $term) {
  1243. $items[$term->name] = $term->name;
  1244. }
  1245. }
  1246. drupal_json_output($items);
  1247. }
  1248. /**
  1249. * Add a record to a cvterm linking table (ie: feature_cvterm)
  1250. *
  1251. * @param $basetable
  1252. * The base table to which the cvterm should be linked/associated. Thus to associate a
  1253. * cvterm to a feature the basetable=feature and cvterm_id is added to the feature_cvterm table.
  1254. * @param $record_id
  1255. * The primary key of the basetable to associate the cvterm with. This should be in integer.
  1256. * @param $cvterm
  1257. * An associative array describing the cvterm. Valid keys include:
  1258. * - name: the name for the cvterm,
  1259. * - cv_name: the name of the cv the cvterm belongs to.
  1260. * - cv_id: the primary key of the cv the cvterm belongs to.
  1261. * @param $options
  1262. * An associative array of options. Valid keys include:
  1263. * - insert_cvterm: Insert the cvterm if it doesn't already exist. FALSE is the default
  1264. *
  1265. * @ingroup tripal_chado_api
  1266. */
  1267. function tripal_associate_cvterm($basetable, $record_id, $cvterm, $options = array()) {
  1268. $linking_table = $basetable . '_cvterm';
  1269. $foreignkey_name = $basetable . '_id';
  1270. // Default Values
  1271. $options['insert_cvterm'] = (isset($options['insert_cvterm'])) ? $options['insert_cvterm'] : FALSE;
  1272. // If the cvterm_id is not set then find the cvterm record using the name and cv_id
  1273. if (!isset($cvterm['cvterm_id'])) {
  1274. $values = array(
  1275. 'name' => $cvterm['name'],
  1276. );
  1277. if (isset($cvterm['cv_id'])) {
  1278. $values['cv_id'] = $cvterm['cv_id'];
  1279. }
  1280. elseif (isset($cvterm['cv_name'])) {
  1281. $values['cv_id'] = array(
  1282. 'name' => $cvterm['cv_name']
  1283. );
  1284. }
  1285. else {
  1286. tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
  1287. "tripal_associate_cvterm: The cvterm needs to have either the cv_name or cv_id
  1288. supplied. You were trying to associate a cvterm with the %base %record_id
  1289. and supplied the cvterm values: %cvterm.",
  1290. array('%base' => $basetable, '%record_id' => $record_id, '%cvterm' => print_r($cvterm,TRUE))
  1291. );
  1292. return FALSE;
  1293. }
  1294. // Get the cvterm. If it doesn't exist then add it if the option
  1295. // 'insert_cvterm' is set.
  1296. $select = chado_select_record('cvterm', array('*'), $values);
  1297. if ($select) {
  1298. $cvterm['cvterm_id'] = $select[0]->cvterm_id;
  1299. }
  1300. elseif ($options['insert_cvterm']) {
  1301. // Insert the cvterm
  1302. $insert = tripal_insert_cvterm($values);
  1303. if (isset($insert->cvterm_id)) {
  1304. $cvterm['cvterm_id'] = $insert->cvterm_id;
  1305. }
  1306. else {
  1307. tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
  1308. "tripal_associate_cvterm: Unable to insert the cvterm using the cvterm values: %cvterm.",
  1309. array('%cvterm' => print_r($cvterm,TRUE))
  1310. );
  1311. return FALSE;
  1312. }
  1313. }
  1314. else {
  1315. tripal_report_error('tripal_api', TRIPAL_WARNING,
  1316. "tripal_associate_cvterm: The cvterm doesn't already exist. You supplied the cvterm values: %cvterm.",
  1317. array('%cvterm' => print_r($cvterm,TRUE))
  1318. );
  1319. return FALSE;
  1320. }
  1321. }
  1322. // Now add the link between the record & cvterm
  1323. if ($cvterm['cvterm_id'] > 0) {
  1324. $values = array(
  1325. 'cvterm_id' => $cvterm['cvterm_id'],
  1326. $foreignkey_name => $record_id,
  1327. 'pub_id' => 1,
  1328. );
  1329. // Check if the cvterm is already associated. If so, don't re-add it.
  1330. $result = chado_select_record($linking_table, array('*'), $values);
  1331. if (!$result) {
  1332. $success = chado_insert_record($linking_table, $values);
  1333. if (!$success) {
  1334. tripal_report_error('tripal_api', TRIPAL_WARNING,
  1335. "Failed to insert the %base record %term",
  1336. array('%base' => $linking_table, '%term' => $cvterm['name'])
  1337. );
  1338. return FALSE;
  1339. }
  1340. $result = chado_select_record($linking_table, array('*'), $values);
  1341. }
  1342. if (isset($result[0])) {
  1343. return $result[0];
  1344. }
  1345. else {
  1346. return FALSE;
  1347. }
  1348. }
  1349. return FALSE;
  1350. }
  1351. /**
  1352. * This function sets the default vocabulary for a given table and field.
  1353. *
  1354. * @param $table
  1355. * The name of the table that contains a field with a foreign key
  1356. * relationship to the cvterm table
  1357. * @param $field
  1358. * The table field name that has the foreign key relationship to the
  1359. * cvterm table for which the default vocabulary will be set
  1360. * @param $cv_name
  1361. * The name of the vocabulary
  1362. *
  1363. * @return
  1364. * TRUE if set, FALSE if an error occured
  1365. */
  1366. function tripal_set_default_cv($table, $field, $cv_name, $cv_id = FALSE) {
  1367. // Get the CV object
  1368. if ($cv_id) {
  1369. $cv = tripal_get_cv(array('cv_id' => $cv_id));
  1370. }
  1371. else {
  1372. $cv = tripal_get_cv(array('name' => $cv_name));
  1373. }
  1374. if ($cv) {
  1375. // first delete any entries for this table and field
  1376. $num_deleted = db_delete('tripal_cv_defaults')
  1377. ->condition('table_name', $table)
  1378. ->condition('field_name', $field)
  1379. ->execute();
  1380. // now add the default value
  1381. $cv_default_id = db_insert('tripal_cv_defaults')
  1382. ->fields(array(
  1383. 'table_name' => $table,
  1384. 'field_name' => $field,
  1385. 'cv_id' => $cv->cv_id,
  1386. ))
  1387. ->execute();
  1388. if (!$cv_default_id) {
  1389. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  1390. "Cannot set default vocabulary for %table.%field. Check the error logs.",
  1391. array('%table' => $table, '%field' => $field));
  1392. return FALSE;
  1393. }
  1394. }
  1395. else {
  1396. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  1397. "Cannot set default vocabulary for %table.%field. The vocabulary name, '%cvname', doesn't exist.",
  1398. array('%table' => $table, '%field' => $field, '%cvname' => $cv_name));
  1399. return FALSE;
  1400. }
  1401. }