tripal_chado.cv.api.inc 52 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574
  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. $result = tripal_update_cvtermpath_loop($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 = $cterm->fetchAssoc()) {
  371. if ($result === 'LOOP'){
  372. watchdog('debug', '<pre>$result: ' . print_r($result, TRUE) . '</pre>');
  373. continue;
  374. }
  375. else {
  376. tripal_update_cvtermpath_root_loop($cterm_result['subject_id'], $cvid);
  377. }
  378. }
  379. }
  380. /**
  381. *
  382. * @param $origin
  383. * @param $child_id
  384. * @param $cv_id
  385. * @param $type_id
  386. * @param $depth
  387. * @return multitype:
  388. */
  389. function tripal_update_cvtermpath_loop($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, $array_of_children){
  390. // Check to see if a row with these values already exists.
  391. chado_set_active('chado');
  392. $count = db_query(
  393. 'SELECT *
  394. FROM cvtermpath
  395. WHERE cv_id = :cvid
  396. AND object_id = :origin
  397. AND subject_id = :child_id
  398. AND pathdistance = :depth
  399. ',
  400. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  401. );
  402. $count_total = $count->rowCount();
  403. try{
  404. if($count_total == 0) {
  405. // If row with values does not already exist write to table.
  406. chado_set_active('chado');
  407. $query = db_insert('cvtermpath')
  408. ->fields(array(
  409. 'object_id' => $origin,
  410. 'subject_id' => $child_id,
  411. 'cv_id' => $cv_id,
  412. 'type_id' => $type_id,
  413. 'pathdistance' => $depth,
  414. ));
  415. $rows = $query->execute();
  416. //watchdog('debug', '<pre>$rows: ' . print_r($rows, TRUE) . '</pre>');
  417. // if($rows) {
  418. // Build the ID.
  419. $children_id = $origin . '|' . $child_id . '|' . $cv_id . '|' . $type_id;
  420. // Now check if the most recent entry already exists in the array.
  421. if ($increment_of_depth != 0) {
  422. // Search the $array_of_children for the new $child_id in the build_id column.
  423. foreach ($array_of_children as $key => $val) {
  424. if ($val['build_id'] == $children_id) {
  425. // If the search returns something check for a possible loop.
  426. $result_of_loop_checker = tripal_update_cvtermpath_loop_checker($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, 0, $val, [], $depth);
  427. watchdog('debug', '<pre>tripal_update_cvtermpath_loop $result_of_loop_checker: ' . print_r($result_of_loop_checker, TRUE) . '</pre>');
  428. if ($result_of_loop_checker) {
  429. watchdog('debug', 'Loop found exit the loop function');
  430. return 'LOOP';
  431. }
  432. }
  433. }
  434. }
  435. // Then add that new entry to the $array_of_children.
  436. $array_of_children[$increment_of_depth] = [
  437. 'build_id' => $children_id,
  438. 'depth' => $depth
  439. ];
  440. $query = db_select('cvterm_relationship', 'cvtr')
  441. ->fields('cvtr')
  442. ->condition('cvtr.object_id', $child_id, '=')
  443. ->execute();
  444. $cterm = $query->fetchAll();
  445. foreach ($cterm as $item) {
  446. $increment_of_depth++;
  447. tripal_update_cvtermpath_loop($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $increment_of_depth, $array_of_children);
  448. }
  449. // }
  450. }
  451. }
  452. catch(Exception $e){
  453. watchdog_exception('tripal_ds', $e);
  454. return FALSE;
  455. }
  456. }
  457. /**
  458. * @param $origin
  459. * @param $child_id
  460. * @param $cv_id
  461. * @param $type_id
  462. * @param $depth
  463. * @param $increment_of_depth
  464. * @param $distance_between_parent_child
  465. * @param $possible_start_of_loop
  466. * @param $array_of_possible_loop
  467. * @param $depth_at_start_of_loop
  468. *
  469. * @return bool
  470. */
  471. function tripal_update_cvtermpath_loop_checker($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){
  472. chado_set_active('chado');
  473. $query = db_select('cvterm_relationship', 'cvtr')
  474. ->fields('cvtr')
  475. ->condition('cvtr.object_id', $child_id, '=')
  476. ->execute();
  477. $cterm = $query->fetchAll();
  478. foreach ($cterm as $item){
  479. if ($distance_between_parent_child > 0){
  480. // Search the $array_of_children for the new $child_id in the build_id column.
  481. foreach ($array_of_possible_loop as $key => $val) {
  482. if ($val['build_id'] === $possible_start_of_loop['build_id']) {
  483. $possible_loop_starts = $val;
  484. // If the search returns something check for a possible loop.
  485. if (!empty($possible_loop_starts)) {
  486. $result = tripal_update_cvtermpath_loop_checker_traverse($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, $possible_start_of_loop, $array_of_possible_loop, array(), 0);
  487. watchdog('debug', '<pre>tripal_update_cvtermpath_loop_checker $result: '. print_r($result, TRUE) .'</pre>');
  488. if(!empty($result)){
  489. return $result;
  490. }
  491. }
  492. }
  493. }
  494. }
  495. elseif($loop_found !== TRUE){}
  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);
  500. $result = tripal_update_cvtermpath_loop_checker($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, $depth_at_start_of_loop);
  501. if($result !== FALSE){
  502. return $result;
  503. }
  504. }
  505. return FALSE;
  506. }
  507. function tripal_update_cvtermpath_loop_checker_traverse($origin, $child_id, $cv_id, $type_id, $depth, $increment_of_depth, $possible_start_of_loop, $array_of_possible_loop, $traverse_of_loop, $increment){
  508. if($increment > 10){
  509. die();
  510. }
  511. chado_set_active('chado');
  512. $query = db_select('cvterm_relationship', 'cvtr')
  513. ->fields('cvtr')
  514. ->condition('cvtr.object_id', $child_id, '=')
  515. ->execute();
  516. $cterm = $query->fetchAll();
  517. foreach ($cterm as $item){
  518. if($array_of_possible_loop === $traverse_of_loop){
  519. watchdog('debug', '<pre>$array_of_possible_loop: '. print_r($array_of_possible_loop, TRUE) .'</pre>');
  520. watchdog('debug', '<pre>$traverse_of_loop: '. print_r($traverse_of_loop, TRUE) .'</pre>');
  521. watchdog('debug', 'LOOP');
  522. //Report the loop.
  523. return $item;
  524. }
  525. elseif($array_of_possible_loop != $traverse_of_loop) {
  526. $increment_of_depth++;
  527. $increment++;
  528. $children_id = $origin . '|' . $item->subject_id . '|' . $cv_id . '|' . $item->type_id;
  529. $traverse_of_loop[$increment] = ['build_id' => $children_id];
  530. $result = tripal_update_cvtermpath_loop_checker_traverse($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $increment_of_depth, $possible_start_of_loop, $array_of_possible_loop, $traverse_of_loop, $increment);
  531. if($result !== FALSE){
  532. return $result;
  533. }
  534. }
  535. }
  536. return FALSE;
  537. }
  538. /*
  539. *
  540. * @param $origin
  541. * @param $subject_id
  542. * @param $cv_id
  543. * @param $type_id
  544. * @param $depth
  545. * @return multitype:
  546. function tripal_update_cvtermpath_loop($origin, $child_id, $cv_id, $type_id, $depth){
  547. // Variables and arrays needed for loop checking.
  548. $array_of_children;
  549. $array_of_possible_loop;
  550. $possible_start_of_loop;
  551. $distance_between_parent_child;
  552. $increment_of_depth;
  553. chado_set_active('chado');
  554. $count = db_query(
  555. 'SELECT *
  556. FROM cvtermpath
  557. WHERE cv_id = :cvid
  558. AND object_id = :origin
  559. AND subject_id = :child_id
  560. AND pathdistance = :depth
  561. ',
  562. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  563. );
  564. $count_total = $count->rowCount();
  565. //Loop check
  566. chado_set_active('chado');
  567. $loop = db_query(
  568. 'SELECT *
  569. FROM cvtermpath
  570. WHERE cv_id = :cvid
  571. AND object_id = :origin
  572. AND subject_id = :child_id
  573. AND type_id = :type_id
  574. ',
  575. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':type_id' => $type_id,)
  576. );
  577. $loop_check = $loop->rowCount();
  578. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $rows ' . print_r($rows, TRUE) . '</pre>');
  579. /*if(!empty($rows)){
  580. foreach($rows as $row){
  581. tripal_update_cvtermpath_loop_check($origin, $child_id, $cv_id, $type_id, $depth, $row->cvtermpath_id, 0);
  582. }
  583. }
  584. else {*/
  585. //If no loop proceed.
  586. /*try{
  587. if($count_total == 0) {
  588. chado_set_active('chado');
  589. $query = db_insert('cvtermpath')
  590. ->fields(array(
  591. 'object_id' => $origin,
  592. 'subject_id' => $child_id,
  593. 'cv_id' => $cv_id,
  594. 'type_id' => $type_id,
  595. 'pathdistance' => $depth,
  596. ));
  597. $rows = $query->execute();
  598. }
  599. if ($loop_check == 0) {
  600. chado_set_active('chado');
  601. $query = db_select('cvterm_relationship', 'cvtr')
  602. ->fields('cvtr')
  603. ->condition('cvtr.object_id', $child_id, '=')
  604. ->execute();
  605. $cterm = $query->fetchAll();
  606. foreach ($cterm as $item) {
  607. //watchdog('debug', '<pre>tripal_ds_preprocess_TripalEntity $item ' . print_r($item, TRUE) . '</pre>');
  608. tripal_update_cvtermpath_loop($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1);
  609. };
  610. //}
  611. }
  612. }
  613. catch(Exception $e){
  614. watchdog_exception('tripal_ds', $e);
  615. return FALSE;
  616. }
  617. return 1;
  618. }
  619. */
  620. /**
  621. *
  622. * @param $origin
  623. * @param $subject_id
  624. * @param $cv_id
  625. * @param $type_id
  626. * @param $depth
  627. * @return multitype:
  628. function tripal_update_cvtermpath_loop_check($origin, $child_id, $cv_id, $type_id, $depth, $cvtermpath_id, $loop_count, $loop_check, $object_id){
  629. //Store the
  630. //Check if the passed parameters match any of the items in the loop_check array.
  631. if(!empty($loop_check)){
  632. foreach($loop_check as $item){
  633. if ($item['type_id'] = $type_id){
  634. if($item['subject_id'] = $child_id){
  635. if($item['object_id'] = $object_id){
  636. //Loop found, roll back all rows until $cvtermpath_id-1 (last correct entry)
  637. // and step into the next loop
  638. }
  639. }
  640. }
  641. }
  642. }
  643. $loop_count + 1;
  644. chado_set_active('chado');
  645. $count = db_query(
  646. 'SELECT *
  647. FROM cvtermpath
  648. WHERE cv_id = :cvid AND object_id = :origin
  649. AND subject_id = :child_id
  650. AND pathdistance = :depth
  651. ',
  652. array(':cvid' => $cv_id, ':origin' => $origin, ':child_id' => $child_id, ':depth' => $depth)
  653. );
  654. $count_total = $count->rowCount();
  655. if ($count_total == 0) {
  656. chado_set_active('chado');
  657. $query = db_insert('cvtermpath')
  658. ->fields(array(
  659. 'object_id' => $origin,
  660. 'subject_id' => $child_id,
  661. 'cv_id' => $cv_id,
  662. 'type_id' => $type_id,
  663. 'pathdistance' => $depth,
  664. ));
  665. $rows = $query->execute();
  666. $cterm = array();
  667. $query = db_select('cvterm_relationship', 'cvtr')
  668. ->fields('cvtr')
  669. ->condition('cvtr.object_id', $child_id, '=' )
  670. ->execute();
  671. $cterm = $query->fetchAll();
  672. foreach ($cterm as $item) {
  673. $loop_check[$loop_count]= $item;
  674. tripal_update_cvtermpath_loop_check($origin, $item->subject_id, $cv_id, $item->type_id, $depth + 1, $loop_count, $loop_check, $item->object_id);
  675. };
  676. }
  677. return 1;
  678. }
  679. */
  680. /**
  681. * Adds a controlled vocabular to the CV table of Chado.
  682. *
  683. * @param $name
  684. * The name of the controlled vocabulary. These are typically all lower case
  685. * with no special characters other than an undrescore (for spaces).
  686. * @param $comment
  687. * A description or definition of the vocabulary.
  688. *
  689. * @return
  690. * An object populated with fields from the newly added database.
  691. *
  692. * @ingroup tripal_chado_api
  693. */
  694. function tripal_insert_cv($name, $definition) {
  695. // insert/update values
  696. $ins_values = array(
  697. 'name' => $name,
  698. 'definition' => $definition
  699. );
  700. // see if the CV default exists already in the database
  701. $sel_values = array('name' => $name);
  702. $results = chado_select_record('cv', array('*'), $sel_values);
  703. // if it does not exists then add it
  704. if (count($results) == 0) {
  705. $success = chado_insert_record('cv', $ins_values);
  706. if (!$success) {
  707. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to create the CV record", NULL);
  708. return FALSE;
  709. }
  710. $results = chado_select_record('cv', array('*'), $sel_values);
  711. }
  712. // if it already exists then do an update
  713. else {
  714. $success = chado_update_record('cv', $sel_values, $ins_values);
  715. if (!$success) {
  716. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the CV record", NULL);
  717. return FALSE;
  718. }
  719. $results = chado_select_record('cv', array('*'), $sel_values);
  720. }
  721. // return the cv object
  722. return $results[0];
  723. }
  724. /**
  725. * Add's a controlled vocabulary term to Chado.
  726. *
  727. * This function will add a cvterm record (and a dbxref record if appropriate
  728. * values are provided). If the parent vocabulary does not exist then
  729. * that also is added to the cv table. If the cvterm is a relationship term
  730. * then the 'is_relationship' value should be set. All
  731. * terms must also have a corresponding database. This is specified in the
  732. * term's ID just before the colon (e.g. GO:003824). If the database does not
  733. * exist in the DB table then it will be added automatically. The accession
  734. * (the value just after the colon in the term's ID) will be added to the
  735. * dbxref table. If the CVterm already exists and $update is set (default)
  736. * then the cvterm is updated. If the CVTerm already exists and $update is
  737. * not set, then no changes are made and the CVTerm object is returned.
  738. *
  739. * @param $term
  740. * An associative array with the following keys:
  741. * - id: the term accession. must be of the form <DB>:<ACCESSION>, where
  742. * <DB> is the name of the database to which the cvterm belongs and the
  743. * <ACCESSION> is the term's accession number in the database.
  744. * - name: the name of the term. usually meant to be human-readable.
  745. * - is_obsolete: is present and set to 1 if the term is defunct
  746. * - definition: the definition of the term
  747. * - cv_name: The CV name to which the term belongs. If this arugment is
  748. * null or not provided then the function tries to find a record in the
  749. * CV table with the same name provided in the $term[namespace]. If
  750. * this field is provided then it overrides what the value in
  751. * $term[namespace]
  752. * - is_relationship: If this term is a relationship term then this value
  753. * should be 1.
  754. * - db_name: In some cases the database name will not be part of the
  755. * $term['id'] and it needs to be explicitly set. Use this argument
  756. * only if the database name cannot be specififed in the term ID
  757. * (e.g. <DB>:<ACCESSION>).
  758. * @param $options
  759. * An associative array with the following keys:
  760. * - update_existing: By default this is TRUE. If the term exists it is
  761. * automatically updated.
  762. * - force_db_change: Sometimes a term may need to switch from one
  763. * database to another. If the term already exists, but associated
  764. * with another term the insert (or update, rather) will fail. Set
  765. * this variable to TRUE to force the change to occur.
  766. *
  767. * @return
  768. * A cvterm object
  769. *
  770. * @ingroup tripal_chado_api
  771. */
  772. function tripal_insert_cvterm($term, $options = array()) {
  773. // get the term properties
  774. $id = (isset($term['id'])) ? $term['id'] : '';
  775. $name = '';
  776. $cvname = '';
  777. $definition = '';
  778. $is_obsolete = 0;
  779. $accession = '';
  780. // Set Defaults
  781. if (isset($term['cv_name'])) {
  782. $cvname = $term['cv_name'];
  783. }
  784. else {
  785. $cvname = 'local';
  786. }
  787. // Namespace is deprecated but must be supported for backwards
  788. // compatability
  789. if (array_key_exists('namespace', $term)) {
  790. $cvname = $term['namespace'];
  791. }
  792. if (isset($term['is_relationship'])) {
  793. $is_relationship = $term['is_relationship'];
  794. }
  795. else {
  796. $is_relationship = 0;
  797. }
  798. if (isset($term['db_name'])) {
  799. $dbname = $term['db_name'];
  800. }
  801. else {
  802. $dbname = 'local';
  803. }
  804. if (isset($options['update_existing'])) {
  805. $update = $options['update_existing'];
  806. }
  807. else {
  808. $update = 1;
  809. }
  810. if (array_key_exists('name', $term)) {
  811. $name = $term['name'];
  812. }
  813. else {
  814. $name = $id;
  815. }
  816. if (array_key_exists('definition', $term)) {
  817. $definition = preg_replace('/^\"(.*)\"/', '\1', $term['definition']);
  818. }
  819. else {
  820. $definition = '';
  821. }
  822. if (array_key_exists('is_obsolete', $term)) {
  823. $is_obsolete = $term['is_obsolete'];
  824. if (strcmp($is_obsolete, 'true') == 0) {
  825. $is_obsolete = 1;
  826. }
  827. }
  828. if (!$name and !$id) {
  829. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Cannot find cvterm without 'id' or 'name'", NULL);
  830. return 0;
  831. }
  832. if (!$id) {
  833. $id = $name;
  834. }
  835. // Get the accession and the database from the cvterm id.
  836. if ($dbname) {
  837. $accession = $id;
  838. }
  839. if (preg_match('/^.+?:.*$/', $id)) {
  840. $accession = preg_replace('/^.+?:(.*)$/', '\1', $id);
  841. $dbname = preg_replace('/^(.+?):.*$/', '\1', $id);
  842. }
  843. // Check that we have a database name, give a different message if it's a
  844. // relationship.
  845. if ($is_relationship and !$dbname) {
  846. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  847. "A database name is not provided for this relationship term: $id", NULL);
  848. return 0;
  849. }
  850. if (!$is_relationship and !$dbname) {
  851. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  852. "A database identifier is missing from the term: $id", NULL);
  853. return 0;
  854. }
  855. // Make sure the CV name exists
  856. $cv = tripal_get_cv(array('name' => $cvname));
  857. if (!$cv) {
  858. $cv = tripal_insert_cv($cvname, '');
  859. }
  860. if (!$cv) {
  861. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  862. "Cannot find namespace '$cvname' when adding/updating $id", NULL);
  863. return 0;
  864. }
  865. // This SQL statement will be used a lot to find a cvterm so just set it
  866. // here for easy reference below. Because CV terms can change their names
  867. // but accessions don't change, the following SQL finds cvterms based on
  868. // their accession rather than the name.
  869. $cvtermsql = "
  870. SELECT CVT.name, CVT.cvterm_id, CV.cv_id, CV.name as cvname,
  871. DB.name as dbname, DB.db_id, DBX.accession
  872. FROM {cvterm} CVT
  873. INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
  874. INNER JOIN {db} DB on DBX.db_id = DB.db_id
  875. INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
  876. WHERE DBX.accession = :accession and DB.name = :name
  877. ";
  878. // Add the database. The function will just return the DB object if the
  879. // database already exists.
  880. $db = tripal_get_db(array('name' => $dbname));
  881. if (!$db) {
  882. $db = tripal_insert_db(array('name' => $dbname));
  883. }
  884. if (!$db) {
  885. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Cannot find database '$dbname' in Chado.", NULL);
  886. return 0;
  887. }
  888. // The cvterm table has two unique dependencies. We need to check both.
  889. // first check the (name, cv_id, is_obsolete) constraint.
  890. $values = array(
  891. 'name' => $name,
  892. 'is_obsolete' => $is_obsolete,
  893. 'cv_id' => array(
  894. 'name' => $cvname,
  895. ),
  896. );
  897. $result = chado_select_record('cvterm', array('*'), $values);
  898. if (count($result) == 1) {
  899. $cvterm = $result[0];
  900. // Get the dbxref record.
  901. $values = array('dbxref_id' => $cvterm->dbxref_id);
  902. $result = chado_select_record('dbxref', array('*'), $values);
  903. $dbxref = $result[0];
  904. if (!$dbxref) {
  905. tripal_report_error('tripal_cv', TRIPAL_ERROR,
  906. 'Unable to access the dbxref record for the :term cvterm. Term Record: !record',
  907. array(':term' => $name, '!record' => print_r($cvterm, TRUE))
  908. );
  909. return FALSE;
  910. }
  911. // Get the db.
  912. $values = array('db_id' => $dbxref->db_id);
  913. $result = chado_select_record('db', array('*'), $values);
  914. $db_check = $result[0];
  915. // If the database from the existing record doesn't match the one that
  916. // has been provided then we have a problem. The term already exists
  917. // but we don't want to just switch the term, the callee must force it.
  918. if ($db_check->name != $db->name) {
  919. if (array_key_exists('force_db_change', $options) and
  920. $options['force_db_change'] != TRUE) {
  921. tripal_report_error('tripal_cv', TRIPAL_ERROR,
  922. 'The term already exists, but associated with a different database record: \'!db\'. It cannot be added: !record',
  923. array(':term' => $name, '!db' => $db_check->name, '!record' => print_r($cvterm, TRUE))
  924. );
  925. return FALSE;
  926. }
  927. // Look to see if the correct dbxref record already exists for this
  928. // database.
  929. $values = array(
  930. 'db_id' => $db->db_id,
  931. 'accession' => $accession,
  932. );
  933. $result = chado_select_record('dbxref', array('*'), $values);
  934. // If we already have a good dbxref then we want to update our cvterm
  935. // to use this dbxref.
  936. if (count($result) > 0) {
  937. $dbxref = $result[0];
  938. $match = array('cvterm_id' => $cvterm->cvterm_id);
  939. $values = array('dbxref_id' => $dbxref->dbxref_id);
  940. $success = chado_update_record('cvterm', $match, $values);
  941. if (!$success) {
  942. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to correct the dbxref id for the cvterm " .
  943. "'$name' (id: $accession), for database $dbname", NULL);
  944. return 0;
  945. }
  946. }
  947. // If we don't have the dbxref then we want to delete our cvterm and let
  948. // the code below recreate it with the correct info.
  949. else {
  950. $match = array('cvterm_id' => $cvterm->cvterm_id);
  951. chado_delete_record('cvterm', $match);
  952. }
  953. }
  954. // Check that the accession matches. Sometimes an OBO can define a term
  955. // multiple times but with different accessions. If this is the case we
  956. // can't do an insert or it will violate the constraint in the cvterm table.
  957. // So we'll need to add the record to the cvterm_dbxref table instead.
  958. if ($dbxref->accession != $accession) {
  959. // Get/add the dbxref for his term.
  960. $dbxref_new = tripal_insert_dbxref(array(
  961. 'db_id' => $db->db_id,
  962. 'accession' => $accession
  963. ));
  964. if (!$dbxref_new) {
  965. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  966. "$name (id: $accession), for database $dbname", NULL);
  967. return 0;
  968. }
  969. // Check to see if the cvterm_dbxref record already exists.
  970. $values = array(
  971. 'cvterm_id' => $cvterm->cvterm_id,
  972. 'dbxref_id' => $dbxref_new->dbxref_id,
  973. 'is_for_definition' => 1,
  974. );
  975. $result = chado_select_record('cvterm_dbxref', array('*'), $values);
  976. // if the cvterm_dbxref record does not exists then add it
  977. if (count($result)==0) {
  978. $options = array(
  979. 'return_record' => FALSE,
  980. );
  981. $success = chado_insert_record('cvterm_dbxref', $values, $options);
  982. if (!$success) {
  983. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the cvterm_dbxref record for a " .
  984. "duplicated cvterm: $name (id: $accession), for database $dbname", NULL);
  985. return 0;
  986. }
  987. }
  988. // get the original cvterm with the same name and return that.
  989. $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
  990. $cvterm = $result->fetchObject();
  991. return $cvterm;
  992. }
  993. // Continue on, we've fixed the record if the db_id did not match.
  994. // We can now perform and updated if we need to.
  995. }
  996. // get the CVterm record
  997. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  998. $cvterm = $result->fetchObject();
  999. if (!$cvterm) {
  1000. // Check to see if the dbxref exists if not, add it.
  1001. $dbxref = tripal_insert_dbxref(array(
  1002. 'db_id' => $db->db_id,
  1003. 'accession' => $accession
  1004. ));
  1005. if (!$dbxref) {
  1006. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  1007. "$name (id: $accession), for database $dbname", NULL);
  1008. return 0;
  1009. }
  1010. // Check to see if the dbxref already has an entry in the cvterm table.
  1011. $values = array('dbxref_id' => $dbxref->dbxref_id);
  1012. $check = chado_select_record('cvterm', array('cvterm_id'), $values);
  1013. if (count($check) == 0) {
  1014. // now add the cvterm
  1015. $ins_values = array(
  1016. 'cv_id' => $cv->cv_id,
  1017. 'name' => $name,
  1018. 'definition' => $definition,
  1019. 'dbxref_id' => $dbxref->dbxref_id,
  1020. 'is_obsolete' => $is_obsolete,
  1021. 'is_relationshiptype' => $is_relationship,
  1022. );
  1023. $success = chado_insert_record('cvterm', $ins_values);
  1024. if (!$success) {
  1025. if (!$is_relationship) {
  1026. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to insert the term: $name ($dbname)", NULL);
  1027. return 0;
  1028. }
  1029. else {
  1030. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to insert the relationship term: $name (cv: " . $cvname . " db: $dbname)", NULL);
  1031. return 0;
  1032. }
  1033. }
  1034. }
  1035. // This dbxref already exists in the cvterm table.
  1036. else {
  1037. tripal_report_error('tripal_chado', TRIPAL_WARNING, "The dbxref already exists for another cvterm record: $name (cv: " . $cvname . " db: $dbname)", NULL);
  1038. return 0;
  1039. }
  1040. $result = chado_query($cvtermsql, array(':accession' => $accession, ':name' => $dbname));
  1041. $cvterm = $result->fetchObject();
  1042. }
  1043. // Update the cvterm.
  1044. elseif ($update) {
  1045. // First, basic update of the term.
  1046. $match = array('cvterm_id' => $cvterm->cvterm_id);
  1047. $upd_values = array(
  1048. 'name' => $name,
  1049. 'definition' => $definition,
  1050. 'is_obsolete' => $is_obsolete,
  1051. 'is_relationshiptype' => $is_relationship,
  1052. );
  1053. $success = chado_update_record('cvterm', $match, $upd_values);
  1054. if (!$success) {
  1055. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the term: $name", NULL);
  1056. return 0;
  1057. }
  1058. // Second, check that the dbxref has not changed and if it has then update it.
  1059. $checksql = "
  1060. SELECT cvterm_id
  1061. FROM {cvterm} CVT
  1062. INNER JOIN {dbxref} DBX on CVT.dbxref_id = DBX.dbxref_id
  1063. INNER JOIN {db} DB on DBX.db_id = DB.db_id
  1064. INNER JOIN {cv} CV on CV.cv_id = CVT.cv_id
  1065. WHERE DBX.accession = :accession and DB.name = :dbname and CVT.name = :term and CV.name = :cvname
  1066. ";
  1067. $check = chado_query($checksql, array(':accession' => $dbxref->accession, ':dbname' => $dbname, ':term' => $name, ':cvname' => $cvname))->fetchObject();
  1068. if (!$check) {
  1069. // check to see if the dbxref exists if not, add it.
  1070. $dbxref = tripal_insert_dbxref(array(
  1071. 'db_id' => $db->db_id,
  1072. 'accession' => $accession
  1073. ));
  1074. if (!$dbxref) {
  1075. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to find or insert the dbxref record for cvterm, " .
  1076. "$name (id: $accession), for database $dbname", NULL);
  1077. return 0;
  1078. }
  1079. $match = array('cvterm_id' => $cvterm->cvterm_id);
  1080. $upd_values = array(
  1081. 'dbxref_id' => $dbxref->dbxref_id,
  1082. );
  1083. $success = chado_update_record('cvterm', $match, $upd_values);
  1084. if (!$success) {
  1085. tripal_report_error('tripal_chado', TRIPAL_WARNING, "Failed to update the term $name with new accession $db:$accession", NULL);
  1086. return 0;
  1087. }
  1088. }
  1089. // Finally grab the updated details.
  1090. $result = chado_query($cvtermsql, array(':accession' => $dbxref->accession, ':name' => $dbname));
  1091. $cvterm = $result->fetchObject();
  1092. }
  1093. else {
  1094. // do nothing, we have the cvterm but we don't want to update
  1095. }
  1096. // return the cvterm
  1097. return $cvterm;
  1098. }
  1099. /**
  1100. * TODO: deprecate this function
  1101. *
  1102. * This function allows other modules to programatically
  1103. * submit an ontology for loading into Chado.
  1104. *
  1105. * This function will add a job to the Jobs subsystem for parsing the ontology.
  1106. * You can either pass a known OBO ID to the function or the URL
  1107. * or full path the the ontology file. If a URL or file name is
  1108. * passed then the $obo_name argument must also be provided. If
  1109. * this is the first time the ontology has been provided to Tripal
  1110. * then it will be added to the database and will be assigned a
  1111. * unique OBO ID.
  1112. *
  1113. * @param $obo_id
  1114. * If the ontology is already loaded into the Tripal tables then
  1115. * use this argument to specify the unique ID for the ontology
  1116. * that will be loaded
  1117. * @param $obo_name
  1118. * If the OBO has not been added before then use this argument
  1119. * to specify the human readable name of the ontology.
  1120. * @param $obo_url
  1121. * If the OBO to be loaded is located on a remote server then
  1122. * use this argument to provide the URL.
  1123. * @param $obo_file
  1124. * If the OBO is housed on the local file system of the server then
  1125. * use this argument to specify the full path.
  1126. *
  1127. * @return
  1128. * returns the job_id of the submitted job or FALSE if the job was not added
  1129. *
  1130. * @ingroup tripal_chado_api
  1131. */
  1132. function tripal_submit_obo_job($obo) {
  1133. global $user;
  1134. // Set Defaults
  1135. $obo['obo_id'] = (isset($obo['obo_id'])) ? $obo['obo_id'] : NULL;
  1136. $obo['name'] = (isset($obo['name'])) ? $obo['name'] : NULL;
  1137. $obo['url'] = (isset($obo['url'])) ? $obo['url'] : NULL;
  1138. $obo['file'] = (isset($obo['file'])) ? $obo['file'] : NULL;
  1139. $includes = array(
  1140. module_load_include('inc', 'tripal_chado', 'includes/loaders/tripal_chado.obo_loader'),
  1141. );
  1142. if ($obo['obo_id']) {
  1143. $sql = "SELECT * FROM {tripal_cv_obo} WHERE obo_id = :obo_id";
  1144. $result = db_query($sql, array(':obo_id' => $obo['obo_id']))->fetchObject();
  1145. $args = array($result->obo_id);
  1146. return tripal_add_job("Load OBO " . $result->name, 'tripal_chado',
  1147. "tripal_chado_load_obo_v1_2_id", $args, $user->uid, 10, $includes);
  1148. }
  1149. else {
  1150. if ($obo['url']) {
  1151. $args = array($obo['name'], $obo['url']);
  1152. return tripal_add_job("Load OBO " . $obo['name'], 'tripal_chado',
  1153. "tripal_chado_load_obo_v1_2_url", $args, $user->uid, 10, $includes);
  1154. }
  1155. elseif ($obo['file']) {
  1156. $args = array($obo['name'], $obo['file']);
  1157. return tripal_add_job("Load OBO " . $obo['name'], 'tripal_chado',
  1158. "tripal_chado_load_obo_v1_2_file", $args, $user->uid, 10, $includes);
  1159. }
  1160. }
  1161. return FALSE;
  1162. }
  1163. /**
  1164. * Add the OBO to the tripal_cv_obo table in the Drupal database.
  1165. *
  1166. * If the OBO name already exists in the table then the path is updated.
  1167. *
  1168. * @param $name
  1169. * The human readable name of this ontology
  1170. * @param $path
  1171. * The file path or URL of the ontology
  1172. *
  1173. * @return
  1174. * Returns the ontology ID
  1175. *
  1176. * @ingroup tripal_chado_api
  1177. */
  1178. function tripal_insert_obo($name, $path) {
  1179. // make sure an OBO with the same name doesn't already exist
  1180. $obo_id = db_select('tripal_cv_obo', 'tco')
  1181. ->fields('tco', array('obo_id'))
  1182. ->condition('name', $name)
  1183. ->execute()
  1184. ->fetchField();
  1185. if ($obo_id) {
  1186. db_update('tripal_cv_obo')
  1187. ->fields(array(
  1188. 'path' => $path,
  1189. ))
  1190. ->condition('name', $name)
  1191. ->execute();
  1192. return $obo_id;
  1193. }
  1194. else {
  1195. $obo_id = db_insert('tripal_cv_obo')
  1196. ->fields(array(
  1197. 'name' => $name,
  1198. 'path' => $path,
  1199. ))
  1200. ->execute();
  1201. return $obo_id;
  1202. }
  1203. }
  1204. /**
  1205. * Retrieves an OBO record.
  1206. *
  1207. * @param $values
  1208. * An associate array with the following allowed keys: obo_id, name
  1209. *
  1210. * @return
  1211. * An instance of an OBO record object.
  1212. */
  1213. function tripal_get_obo($values) {
  1214. $query = db_select('tripal_cv_obo', 'tco')
  1215. ->fields('tco');
  1216. if (array_key_exists('name', $values)) {
  1217. $query->condition('tco.name', $values['name']);
  1218. }
  1219. if (array_key_exists('obo_id', $values)) {
  1220. $query->condition('tco.obo_id', $values['obo_id']);
  1221. }
  1222. return $query->execute()->fetchObject();
  1223. }
  1224. /**
  1225. * This function is intended to be used in autocomplete forms
  1226. * for searching for CV terms that begin with the provided string
  1227. *
  1228. * @param $cv_id
  1229. * The CV ID in which to search for the term
  1230. * @param $string
  1231. * The string to search for
  1232. *
  1233. * @return
  1234. * A json array of terms that begin with the provided string
  1235. *
  1236. * @ingroup tripal_chado_api
  1237. */
  1238. function tripal_autocomplete_cvterm($cv_id, $string = '') {
  1239. if ($cv_id) {
  1240. $sql = "
  1241. SELECT CVT.cvterm_id, CVT.name
  1242. FROM {cvterm} CVT
  1243. WHERE CVT.cv_id = :cv_id and lower(CVT.name) like lower(:name)
  1244. UNION
  1245. SELECT CVT2.cvterm_id, CVTS.synonym as name
  1246. FROM {cvterm} CVT2
  1247. INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
  1248. WHERE CVT2.cv_id = :cv_id and lower(CVTS.synonym) like lower(:name)
  1249. ORDER by name
  1250. LIMIT 25 OFFSET 0
  1251. ";
  1252. $results = chado_query($sql, array(':cv_id' => $cv_id, ':name' => $string . '%'));
  1253. $items = array();
  1254. foreach ($results as $term) {
  1255. $items[$term->name] = $term->name;
  1256. }
  1257. }
  1258. // If a CV wasn't provided then search all of them, and include the cv
  1259. // in the results.
  1260. else {
  1261. $sql = "
  1262. SELECT CVT.cvterm_id, CVT.name, CV.name as cvname, CVT.cv_id
  1263. FROM {cvterm} CVT
  1264. INNER JOIN {cv} CV on CVT.cv_id = CV.cv_id
  1265. WHERE lower(CVT.name) like lower(:name)
  1266. UNION
  1267. SELECT CVT2.cvterm_id, CVTS.synonym as name, CV2.name as cvname, CVT2.cv_id
  1268. FROM {cvterm} CVT2
  1269. INNER JOIN {cv} CV2 on CVT2.cv_id = CV2.cv_id
  1270. INNER JOIN {cvtermsynonym} CVTS ON CVTS.cvterm_id = CVT2.cvterm_id
  1271. WHERE lower(CVTS.synonym) like lower(:name)
  1272. ORDER by name
  1273. LIMIT 25 OFFSET 0
  1274. ";
  1275. $results = chado_query($sql, array(':name' => $string . '%'));
  1276. $items = array();
  1277. foreach ($results as $term) {
  1278. $items[$term->name] = $term->name;
  1279. }
  1280. }
  1281. drupal_json_output($items);
  1282. }
  1283. /**
  1284. * Add a record to a cvterm linking table (ie: feature_cvterm)
  1285. *
  1286. * @param $basetable
  1287. * The base table to which the cvterm should be linked/associated. Thus to associate a
  1288. * cvterm to a feature the basetable=feature and cvterm_id is added to the feature_cvterm table.
  1289. * @param $record_id
  1290. * The primary key of the basetable to associate the cvterm with. This should be in integer.
  1291. * @param $cvterm
  1292. * An associative array describing the cvterm. Valid keys include:
  1293. * - name: the name for the cvterm,
  1294. * - cv_name: the name of the cv the cvterm belongs to.
  1295. * - cv_id: the primary key of the cv the cvterm belongs to.
  1296. * @param $options
  1297. * An associative array of options. Valid keys include:
  1298. * - insert_cvterm: Insert the cvterm if it doesn't already exist. FALSE is the default
  1299. *
  1300. * @ingroup tripal_chado_api
  1301. */
  1302. function tripal_associate_cvterm($basetable, $record_id, $cvterm, $options = array()) {
  1303. $linking_table = $basetable . '_cvterm';
  1304. $foreignkey_name = $basetable . '_id';
  1305. // Default Values
  1306. $options['insert_cvterm'] = (isset($options['insert_cvterm'])) ? $options['insert_cvterm'] : FALSE;
  1307. // If the cvterm_id is not set then find the cvterm record using the name and cv_id
  1308. if (!isset($cvterm['cvterm_id'])) {
  1309. $values = array(
  1310. 'name' => $cvterm['name'],
  1311. );
  1312. if (isset($cvterm['cv_id'])) {
  1313. $values['cv_id'] = $cvterm['cv_id'];
  1314. }
  1315. elseif (isset($cvterm['cv_name'])) {
  1316. $values['cv_id'] = array(
  1317. 'name' => $cvterm['cv_name']
  1318. );
  1319. }
  1320. else {
  1321. tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
  1322. "tripal_associate_cvterm: The cvterm needs to have either the cv_name or cv_id
  1323. supplied. You were trying to associate a cvterm with the %base %record_id
  1324. and supplied the cvterm values: %cvterm.",
  1325. array('%base' => $basetable, '%record_id' => $record_id, '%cvterm' => print_r($cvterm,TRUE))
  1326. );
  1327. return FALSE;
  1328. }
  1329. // Get the cvterm. If it doesn't exist then add it if the option
  1330. // 'insert_cvterm' is set.
  1331. $select = chado_select_record('cvterm', array('*'), $values);
  1332. if ($select) {
  1333. $cvterm['cvterm_id'] = $select[0]->cvterm_id;
  1334. }
  1335. elseif ($options['insert_cvterm']) {
  1336. // Insert the cvterm
  1337. $insert = tripal_insert_cvterm($values);
  1338. if (isset($insert->cvterm_id)) {
  1339. $cvterm['cvterm_id'] = $insert->cvterm_id;
  1340. }
  1341. else {
  1342. tripal_report_error('tripal_chado_api', TRIPAL_WARNING,
  1343. "tripal_associate_cvterm: Unable to insert the cvterm using the cvterm values: %cvterm.",
  1344. array('%cvterm' => print_r($cvterm,TRUE))
  1345. );
  1346. return FALSE;
  1347. }
  1348. }
  1349. else {
  1350. tripal_report_error('tripal_api', TRIPAL_WARNING,
  1351. "tripal_associate_cvterm: The cvterm doesn't already exist. You supplied the cvterm values: %cvterm.",
  1352. array('%cvterm' => print_r($cvterm,TRUE))
  1353. );
  1354. return FALSE;
  1355. }
  1356. }
  1357. // Now add the link between the record & cvterm
  1358. if ($cvterm['cvterm_id'] > 0) {
  1359. $values = array(
  1360. 'cvterm_id' => $cvterm['cvterm_id'],
  1361. $foreignkey_name => $record_id,
  1362. 'pub_id' => 1,
  1363. );
  1364. // Check if the cvterm is already associated. If so, don't re-add it.
  1365. $result = chado_select_record($linking_table, array('*'), $values);
  1366. if (!$result) {
  1367. $success = chado_insert_record($linking_table, $values);
  1368. if (!$success) {
  1369. tripal_report_error('tripal_api', TRIPAL_WARNING,
  1370. "Failed to insert the %base record %term",
  1371. array('%base' => $linking_table, '%term' => $cvterm['name'])
  1372. );
  1373. return FALSE;
  1374. }
  1375. $result = chado_select_record($linking_table, array('*'), $values);
  1376. }
  1377. if (isset($result[0])) {
  1378. return $result[0];
  1379. }
  1380. else {
  1381. return FALSE;
  1382. }
  1383. }
  1384. return FALSE;
  1385. }
  1386. /**
  1387. * This function sets the default vocabulary for a given table and field.
  1388. *
  1389. * @param $table
  1390. * The name of the table that contains a field with a foreign key
  1391. * relationship to the cvterm table
  1392. * @param $field
  1393. * The table field name that has the foreign key relationship to the
  1394. * cvterm table for which the default vocabulary will be set
  1395. * @param $cv_name
  1396. * The name of the vocabulary
  1397. *
  1398. * @return
  1399. * TRUE if set, FALSE if an error occured
  1400. */
  1401. function tripal_set_default_cv($table, $field, $cv_name, $cv_id = FALSE) {
  1402. // Get the CV object
  1403. if ($cv_id) {
  1404. $cv = tripal_get_cv(array('cv_id' => $cv_id));
  1405. }
  1406. else {
  1407. $cv = tripal_get_cv(array('name' => $cv_name));
  1408. }
  1409. if ($cv) {
  1410. // first delete any entries for this table and field
  1411. $num_deleted = db_delete('tripal_cv_defaults')
  1412. ->condition('table_name', $table)
  1413. ->condition('field_name', $field)
  1414. ->execute();
  1415. // now add the default value
  1416. $cv_default_id = db_insert('tripal_cv_defaults')
  1417. ->fields(array(
  1418. 'table_name' => $table,
  1419. 'field_name' => $field,
  1420. 'cv_id' => $cv->cv_id,
  1421. ))
  1422. ->execute();
  1423. if (!$cv_default_id) {
  1424. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  1425. "Cannot set default vocabulary for %table.%field. Check the error logs.",
  1426. array('%table' => $table, '%field' => $field));
  1427. return FALSE;
  1428. }
  1429. }
  1430. else {
  1431. tripal_report_error('tripal_chado', TRIPAL_WARNING,
  1432. "Cannot set default vocabulary for %table.%field. The vocabulary name, '%cvname', doesn't exist.",
  1433. array('%table' => $table, '%field' => $field, '%cvname' => $cv_name));
  1434. return FALSE;
  1435. }
  1436. }