r58402 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r58401‎ \| r58402 \| r58403 >
Date:	14:12, 1 November 2009
Author:	daniel
Status:	deferred
Tags:
Comment:	wikiword thesaurus api
Modified paths:	/trunk/WikiWord/WikiWord/src/main/php/api.php (modified) (history) /trunk/WikiWord/WikiWord/src/main/php/wwclient.php (modified) (history) /trunk/WikiWord/WikiWord/src/main/php/wwthesaurus.php (modified) (history) /trunk/WikiWord/WikiWord/src/main/php/wwutils.php (modified) (history)

Diff [purge]

Index: trunk/WikiWord/WikiWord/src/main/php/api.php
—	—	@@ -12,8 +12,12 @@
13	13	$format = @$_REQUEST['format'];
14	14	if ( !$format ) $format = 'phps';
15	15
	16	+ if ($lang) {
	17	+ $lang = preg_replace('[^\\w\\d_]', '', $lang);
	18	+ }
	19	+
16	20	$result = array( 'query' => $query );
17		~~- $start = microtime();~~
	21	+ $start = microtime(true);
18	22
19	23	try {
20	24	$thesaurus = new WWThesaurus();
—	—	@@ -26,12 +30,12 @@
27	31	$page = @$_REQUEST['page'];
28	32
29	33	if ( $lang === null ) $result['error'] = array('code' => 150, 'message' => "missing parameter lang");
30		~~- else if ( $term !=== null ) {~~
	34	+ else if ( $term !== null ) {
31	35	$result['concepts'] = $thesaurus->getConceptsForTerm($lang, $term);
32	36	if ( $result['concepts'] === false \|\| $result['concepts'] === null ) {
33	37	$result['error'] = array('code' => 210, 'message' => "failed to retrieve concepts for term $langt:$term");
34	38	}
35		~~- } else if ( $page !=== null ) {~~
	39	+ } else if ( $page !== null ) {
36	40	$result['concepts'] = $thesaurus->getConceptsForPage($lang, $page);
37	41	if ( $result['concepts'] === false \|\| $result['concepts'] === null ) {
38	42	$result['error'] = array('code' => 250, 'message' => "failed to retrieve concepts for page $langt:$page");
—	—	@@ -39,7 +43,7 @@
40	44	} else {
41	45	$result['error'] = array('code' => 110, 'message' => "missing parameter term");
42	46	}
43		~~- } if ($query == 'properties') else {~~
	47	+ } else if ($query == 'properties') {
44	48	$gcid = @$_REQUEST['gcid'];
45	49	$props = @$_REQUEST['props'];
46	50
—	—	@@ -49,7 +53,7 @@
50	54	$props = preg_split('![\\s,;\|/:]\\s*!', $props);
51	55
52	56	foreach ( $props as $p ) {
53		~~- $m = "get" . ucfist($p) . "ForConcept";~~
	57	+ $m = "get" . ucfirst($p) . "ForConcept";
54	58	if ( !method_exists($thesaurus, $m) ) {
55	59	$result['error'] = array('code' => 190, 'message' => "unknown property: $p");
56	60	break;
—	—	@@ -66,11 +70,11 @@
67	71	} else {
68	72	$result['error'] = array('code' => 10, 'message' => "bad query: $query");
69	73	}
70		~~- } catch (Exception e) {~~
	74	+ } catch (Exception $e) {
71	75	$result['error'] = array('code' => 1000, 'message' => "unexpected exception: " . $e->getMessage());
72	76	}
73	77
74		~~- $result['time'] = (microtime() - $start) . "ms";~~
	78	+ $result['time'] = (microtime(true) - $start) . " sec";
75	79
76	80	if ( isset($result['error']) ) {
77	81	#TODO: HTTP error codce would be nice, but causes file_get_contents to swallow the data.
—	—	@@ -84,13 +88,13 @@
85	89	$data = serialize($result);
86	90	echo $data;
87	91	} else if ($format == 'php') {
88		~~- header("Content-Type: text/php");~~
	92	+ header("Content-Type: text/php; charset=UTF-8");
89	93	var_export($result);
90	94	} else if ($format == 'text') {
91		~~- header("Content-Type: text/plain");~~
	95	+ header("Content-Type: text/plain; charset=UTF-8");
92	96	print_r($result);
93	97	} else {
94		~~- header("Content-Type: text/plain");~~
	98	+ header("Content-Type: text/plain; charset=UTF-8");
95	99	header("Status: 400 Bad Request", true, 400);
96	100	echo "Bad format: $format";
97	101	}
—	—	@@ -98,6 +102,6 @@
99	103	exit();
100	104	}
101	105
102		~~-header("Content-Type: text/plain");~~
	106	+header("Content-Type: text/plain; charset=UTF-8");
103	107	?>
104	108	WikiWord REST API
\ No newline at end of file
Index: trunk/WikiWord/WikiWord/src/main/php/wwclient.php
—	—	@@ -28,13 +28,52 @@
29	29	return $data;
30	30	}
31	31
32		~~- function getWikiPages( $id ) {~~
33		~~- $p = $this->getConceptProperties( $id, 'pages' );~~
34		-
	32	+ function getPagesForConcept( $id, $lang = null ) {
	33	+ $p = $this->getConceptProperties( $id, 'pages', $lang );
35	34	return $p['pages'];
36	35	}
37	36
38		~~- function getConceptProperties( $id, $props, $lang = NUL L) {~~
	37	+ function getRelatedForConcept( $id, $lang = null ) {
	38	+ $p = $this->getConceptProperties( $id, 'related', $lang );
	39	+ return $p['related'];
	40	+ }
	41	+
	42	+ function getBroaderForConcept( $id, $lang = null ) {
	43	+ $p = $this->getConceptProperties( $id, 'broader', $lang );
	44	+ return $p['broader'];
	45	+ }
	46	+
	47	+ function getNarrowerForConcept( $id, $lang = null ) {
	48	+ $p = $this->getConceptProperties( $id, 'narrower', $lang );
	49	+ return $p['narrower'];
	50	+ }
	51	+
	52	+ function getTermsForConcept( $id, $lang = null ) {
	53	+ $p = $this->getConceptProperties( $id, 'terms', $lang );
	54	+ return $p['terms'];
	55	+ }
	56	+
	57	+ function getDefinitionForConcept( $id, $lang = null ) {
	58	+ $p = $this->getConceptProperties( $id, 'definition', $lang );
	59	+ return $p['definition'];
	60	+ }
	61	+
	62	+ function getReferencesForConcept( $id, $lang = null ) {
	63	+ $p = $this->getConceptProperties( $id, 'links', $lang );
	64	+ return $p['references'];
	65	+ }
	66	+
	67	+ function getLinksForConcept( $id, $lang = null ) {
	68	+ $p = $this->getConceptProperties( $id, 'links', $lang );
	69	+ return $p['links'];
	70	+ }
	71	+
	72	+ function getScoresForConcept( $id, $lang = null ) {
	73	+ $p = $this->getConceptProperties( $id, 'scores', $lang );
	74	+ return $p['scores'];
	75	+ }
	76	+
	77	+ function getConceptProperties( $id, $props, $lang = null ) {
39	78	$param = array(
40	79	'query' => 'properties',
41	80	'props' => ( is_array($props) ? join('\|', $props) : $props ),
—	—	@@ -60,4 +99,16 @@
61	100	return $rs['concepts'];
62	101	}
63	102
	103	+ function getConceptsForPage( $lang, $page ) {
	104	+ $param = array(
	105	+ 'query' => 'concepts',
	106	+ 'lang' => $lang,
	107	+ 'page' => $page,
	108	+ );
	109	+
	110	+ $rs = $this->query( $param );
	111	+
	112	+ return $rs['concepts'];
	113	+ }
	114	+
64	115	}
Index: trunk/WikiWord/WikiWord/src/main/php/wwutils.php
—	—	@@ -56,17 +56,53 @@
57	57	$this->wikidbs = array();
58	58	}
59	59
60		~~- static function slurpList($rs, $field) {~~
	60	+ function getRows($sql, $key = NULL) {
	61	+ $rs = $this->query($sql);
	62	+ $list = WWUtils::slurpRows($rs, $key);
	63	+ mysql_free_result($rs);
	64	+ return $list;
	65	+ }
	66	+
	67	+ function getList($sql, $valueField, $key = NULL) {
	68	+ $rs = $this->query($sql);
	69	+ $list = WWUtils::slurpList($rs, $valueField, $key);
	70	+ mysql_free_result($rs);
	71	+ return $list;
	72	+ }
	73	+
	74	+ static function slurpList($rs, $field, $key = null) {
61	75	if (is_string($rs)) $rs = $this->query($rs);
62	76
63	77	$list = array();
64	78	while ($row = mysql_fetch_assoc($rs)) {
65		~~- $list[] = $row[$field];~~
	79	+ $v = $row[$field];
	80	+ if ($key) {
	81	+ $k = $row[$key];
	82	+ $list[$k] = $v;
	83	+ } else {
	84	+ $list[] = $v;
	85	+ }
66	86	}
67	87
68	88	return $list;
69	89	}
70	90
	91	+ static function slurpRows($rs, $key = null) {
	92	+ if (is_string($rs)) $rs = $this->query($rs);
	93	+
	94	+ $list = array();
	95	+ while ($row = mysql_fetch_assoc($rs)) {
	96	+ if ($key) {
	97	+ $k = $row[$key];
	98	+ $list[$k] = $row;
	99	+ } else {
	100	+ $list[] = $row;
	101	+ }
	102	+ }
	103	+
	104	+ return $list;
	105	+ }
	106	+
71	107	static function slurpAssoc($rs, $keyField, $valueField) {
72	108	if (is_string($rs)) $rs = $this->query($rs);
73	109
Index: trunk/WikiWord/WikiWord/src/main/php/wwthesaurus.php
—	—	@@ -3,21 +3,49 @@
4	4
5	5	class WWThesaurus extends WWUTils {
6	6
7		~~- function queryConceptsForTerm($lang, $term) {~~
	7	+ function queryConceptsForTerm($lang, $term, $limit = 100) {
8	8	global $wwTablePrefix, $wwThesaurusDataset;
9	9
10	10	$term = trim($term);
11	11
12		~~- $sql = "SELECT M., O., definition FROM {$wwTablePrefix}_{$lang}_meaning as M"~~
	12	+ $sql = "SELECT O.global_concept as id, M., O., definition FROM {$wwTablePrefix}_{$lang}_meaning as M"
13	13	. " LEFT JOIN {$wwTablePrefix}_{$lang}_definition as D ON M.concept = D.concept "
14	14	. " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ON O.lang = \"" . mysql_real_escape_string($lang) . "\" AND M.concept = O.local_concept "
15	15	. " WHERE term_text = \"" . mysql_real_escape_string($term) . "\""
16	16	. " ORDER BY freq DESC "
17		~~- . " LIMIT 100";~~
	17	+ . " LIMIT $limit";
18	18
19	19	return $this->query($sql);
20	20	}
21	21
	22	+ function getConceptsForTerm($lang, $term, $limit = 100) {
	23	+ $rs = $this->queryConceptsForTerm($lang, $term);
	24	+ $list = WWUtils::slurpRows($rs);
	25	+ mysql_free_result($rs);
	26	+ return $list;
	27	+ }
	28	+
	29	+ function queryConceptsForPage($lang, $page, $limit = 100) {
	30	+ global $wwTablePrefix, $wwThesaurusDataset;
	31	+
	32	+ $page = trim($page);
	33	+
	34	+ $sql = "SELECT O.global_concept as id, O.* FROM {$wwTablePrefix}_{$lang}_resource as R "
	35	+ . " JOIN {$wwTablePrefix}_{$lang}_about as A ON A.resource = R.id "
	36	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ON O.lang = \"" . mysql_real_escape_string($lang) . "\" AND A.concept = O.local_concept "
	37	+ . " WHERE R.name = \"" . mysql_real_escape_string($page) . "\""
	38	+ . " LIMIT $limit";
	39	+
	40	+ return $this->query($sql);
	41	+ }
	42	+
	43	+ function getConceptsForPage($lang, $page, $limit = 100) {
	44	+ $rs = $this->queryConceptsForPage($lang, $page);
	45	+ $list = WWUtils::slurpRows($rs);
	46	+ mysql_free_result($rs);
	47	+ return $list;
	48	+ }
	49	+
22	50	function queryLocalConcepts($id) {
23	51	global $wwTablePrefix, $wwThesaurusDataset;
24	52	$sql = "SELECT O.lang, O.local_concept_name from {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ";
—	—	@@ -119,4 +147,121 @@
120	148	return $names;
121	149	}
122	150
	151	+ /////////////////////////////////////////////////////////
	152	+ function getPagesForConcept( $id, $lang, $limit = 100 ) {
	153	+ global $wwTablePrefix, $wwThesaurusDataset;
	154	+
	155	+ #FIXME: if $lang is not given, collect all languages!
	156	+ $sql = "SELECT R.name FROM {$wwTablePrefix}_{$lang}_resource as R "
	157	+ . " JOIN {$wwTablePrefix}_{$lang}_about as A ON A.resource = R.id "
	158	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ON O.lang = \"" . mysql_real_escape_string($lang) . "\" AND A.concept = O.local_concept "
	159	+ . " WHERE O.global_concept = " . (int)$id
	160	+ . " LIMIT $limit";
	161	+
	162	+ $pages = $this->getList($sql, "name");
	163	+ if ( $pages === false \|\| $pages === null ) return false;
	164	+
	165	+ return array( $lang => $pages );
	166	+ }
	167	+
	168	+ function getRelatedForConcept( $id, $lang = null, $limit = 100 ) {
	169	+ global $wwTablePrefix, $wwThesaurusDataset;
	170	+
	171	+ $sql = "SELECT C.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept as C "
	172	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_relation as R ON R.concept2 = C.id "
	173	+ . " WHERE R.concept1 = ".(int)$id
	174	+ . " AND ( R.bilink > 0 OR R.langref > 0 OR R.langmatch > 0 )"
	175	+ . " LIMIT $limit";
	176	+
	177	+ return $this->getRows($sql);
	178	+ }
	179	+
	180	+ function getBroaderForConcept( $id, $lang = null, $limit = 100 ) {
	181	+ global $wwTablePrefix, $wwThesaurusDataset;
	182	+
	183	+ $sql = "SELECT C.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept as C "
	184	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_broader as R ON R.broad = C.id "
	185	+ . " WHERE R.narrow = ".(int)$id
	186	+ . " LIMIT $limit";
	187	+
	188	+ return $this->getRows($sql);
	189	+ }
	190	+
	191	+ function getNarrowerForConcept( $id, $lang = null, $limit = 100 ) {
	192	+ global $wwTablePrefix, $wwThesaurusDataset;
	193	+
	194	+ $sql = "SELECT C.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept as C "
	195	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_broader as R ON R.narrow = C.id "
	196	+ . " WHERE R.broad = ".(int)$id
	197	+ . " LIMIT $limit";
	198	+
	199	+ return $this->getRows($sql);
	200	+ }
	201	+
	202	+ function getTermsForConcept( $id, $lang , $limit = 100 ) {
	203	+ global $wwTablePrefix, $wwThesaurusDataset;
	204	+
	205	+ #FIXME: if $lang is not given, collect all languages!
	206	+ $sql = "SELECT M.term_text FROM {$wwTablePrefix}_{$lang}_meaning as M"
	207	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ON O.lang = \"" . mysql_real_escape_string($lang) . "\" AND M.concept = O.local_concept "
	208	+ . " WHERE O.global_concept = " . (int)$id
	209	+ . " ORDER BY freq DESC "
	210	+ . " LIMIT $limit";
	211	+
	212	+ $terms = $this->getList($sql, "term_text");
	213	+ if ( $terms === false \|\| $terms === null ) return false;
	214	+
	215	+ return array( $lang => $terms );
	216	+ }
	217	+
	218	+ function getDefinitionForConcept( $id, $lang, $limit = 100 ) {
	219	+ global $wwTablePrefix, $wwThesaurusDataset;
	220	+
	221	+ #FIXME: if $lang is not given, collect all languages!
	222	+ $sql = "SELECT D.definition FROM {$wwTablePrefix}_{$lang}_definition as D"
	223	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_origin as O ON O.lang = \"" . mysql_real_escape_string($lang) . "\" AND D.concept = O.local_concept "
	224	+ . " WHERE O.global_concept = " . (int)$id
	225	+ . " LIMIT $limit";
	226	+
	227	+ $definitions = $this->getList($sql, "definition");
	228	+ if ( $definitions === false \|\| $definitions === null ) return false;
	229	+
	230	+ return array( $lang => $definitions );
	231	+ }
	232	+
	233	+ function getLinksForConcept( $id, $lang = null, $limit = 100 ) {
	234	+ global $wwTablePrefix, $wwThesaurusDataset;
	235	+
	236	+ $sql = "SELECT C.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept as C "
	237	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_link as L ON L.target = C.id "
	238	+ . " WHERE L.anchor = ".(int)$id
	239	+ . " LIMIT $limit";
	240	+
	241	+ return $this->getRows($sql);
	242	+ }
	243	+
	244	+ function getReferencesForConcept( $id, $lang = null, $limit = 100 ) {
	245	+ global $wwTablePrefix, $wwThesaurusDataset;
	246	+
	247	+ $sql = "SELECT C.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept as C "
	248	+ . " JOIN {$wwTablePrefix}_{$wwThesaurusDataset}_link as L ON L.anchor = C.id "
	249	+ . " WHERE L.target = ".(int)$id
	250	+ . " LIMIT $limit";
	251	+
	252	+ return $this->getRows($sql);
	253	+ }
	254	+
	255	+ function getScoresForConcept( $id, $lang = null ) {
	256	+ global $wwTablePrefix, $wwThesaurusDataset;
	257	+
	258	+ $sql = "SELECT S.* FROM {$wwTablePrefix}_{$wwThesaurusDataset}_concept_stats as S "
	259	+ . " WHERE S.concept = ".(int)$id
	260	+ ;
	261	+
	262	+ $r = $this->getRows($sql);
	263	+ if ( !$r ) return false;
	264	+
	265	+ return $r;
	266	+ }
	267	+
123	268	}

Status & tagging log

14:13, 2 December 2009 😂 (talk | contribs) changed the status of r58402 [removed: new added: deferred]