r65690 MediaWiki - Code Review archive

Repository:MediaWiki
Revision:< r65689‎ | r65690 | r65691 >
Date:14:38, 30 April 2010
Author:daniel
Status:deferred
Tags:
Comment:
WikiWordWeb structure
Modified paths:
  • /trunk/WikiWord/WikiWord/src/main/php/build-concept-info.sh (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/build-search-index.sh (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/concept-info-local.sql (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/concept-info.sql (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/config.sample.php (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/search-index-local.sql (deleted) (history)
  • /trunk/WikiWord/WikiWord/src/main/php/search-index.sql (deleted) (history)
  • /trunk/WikiWord/WikiWordWeb/src/main/sql (deleted) (history)
  • /trunk/WikiWord/WikiWordWeb/src/main/www/config.sample.php (added) (history)

Diff [purge]

Index: trunk/WikiWord/WikiWord/src/main/php/build-search-index.sh
——@@ -1,15 +0,0 @@
2 -#!/bin/bash
3 -set -e
4 -
5 -db="$1"
6 -collection="$2"
7 -thesaurus="$3"
8 -languages="en de fr nl it es pt pl"
9 -
10 -echo "preparing search index"
11 -replace '{collection}' "$collection" '{thesaurus}' "$thesaurus" < search-index.sql | mysql "$db"
12 -
13 -for n in $languages; do
14 - echo "collection search index: $n"
15 - replace '{collection}' "$collection" '{thesaurus}' "$thesaurus" '{lang}' "$n" < search-index-local.sql | mysql "$db"
16 -done
Index: trunk/WikiWord/WikiWord/src/main/php/search-index-local.sql
——@@ -1,16 +0,0 @@
2 -insert into {collection}_{thesaurus}_search_index (
3 - concept, concept_name, `type`,
4 - `lang`, `term`, `score`, `norm` )
5 -select O.global_concept, M.concept_name, C.type, "{lang}",
6 - REPLACE( LCASE( CAST(M.term_text as CHAR CHARACTER SET utf8) COLLATE utf8_general_ci ), "-", "" ),
7 - M.rule * M.freq, 1
8 -from {collection}_{lang}_meaning as M
9 -join {collection}_{thesaurus}_origin as O on O.lang = "{lang}" and O.local_concept = M.concept
10 -join {collection}_{thesaurus}_concept as C on C.id = O.global_concept
11 -where (M.rule not in (10, 30) OR M.freq > 1) and C.type > 0
12 -on duplicate key update
13 - score = if (score > values(score), score, values(score)),
14 - norm = if (norm < values(norm), score, values(norm));
15 -
\ No newline at end of file
Index: trunk/WikiWord/WikiWord/src/main/php/concept-info.sql
——@@ -1,20 +0,0 @@
2 -SET SESSION group_concat_max_len = 262144; -- 1024*256
3 -
4 -create table if not exists {collection}_{thesaurus}_concept_info (
5 - concept int(11) NOT NULL,
6 - `lang` varbinary(10) NOT NULL,
7 - `name` varbinary(255) NOT NULL,
8 - `pages` MEDIUMBLOB DEFAULT NULL,
9 - `definition` MEDIUMBLOB DEFAULT NULL,
10 - `broader` MEDIUMBLOB DEFAULT NULL,
11 - `narrower` MEDIUMBLOB DEFAULT NULL,
12 - `similar` MEDIUMBLOB DEFAULT NULL,
13 - `related` MEDIUMBLOB DEFAULT NULL,
14 - PRIMARY KEY ( concept, lang )
15 - ) ENGINE=MyISAM DEFAULT CHARSET=utf8 COLLATE=utf8_bin;
16 -
17 -truncate {collection}_{thesaurus}_concept_info;
18 -
19 -insert into {collection}_{thesaurus}_concept_info ( concept, lang, name )
20 -select global_concept, lang, local_concept_name
21 -from {collection}_{thesaurus}_origin;
Index: trunk/WikiWord/WikiWord/src/main/php/search-index.sql
——@@ -1,13 +0,0 @@
2 -create table if not exists {collection}_{thesaurus}_search_index (
3 - concept int(11) NOT NULL,
4 - concept_name varbinary(255) NOT NULL,
5 - type int(11) NOT NULL,
6 - `lang` varbinary(10) NOT NULL,
7 - `term` varchar(255) character set utf8 collate utf8_general_ci NOT NULL,
8 - `score` int NOT NULL,
9 - `norm` int NOT NULL,
10 - PRIMARY KEY ( lang, term, concept ),
11 - KEY ( concept, lang )
12 - ) ENGINE=MyISAM DEFAULT CHARSET=utf8 COLLATE=utf8_bin;
13 -
14 -truncate {collection}_{thesaurus}_search_index;
Index: trunk/WikiWord/WikiWord/src/main/php/build-concept-info.sh
——@@ -1,15 +0,0 @@
2 -#!/bin/bash
3 -set -e
4 -
5 -db="$1"
6 -collection="$2"
7 -thesaurus="$3"
8 -languages="commons en de fr nl it es pt pl"
9 -
10 -echo "preparing concept info"
11 -replace '{collection}' "$collection" '{thesaurus}' "$thesaurus" < concept-info.sql | mysql "$db"
12 -
13 -for n in $languages; do
14 - echo "collection concept info: $n"
15 - replace '{collection}' "$collection" '{thesaurus}' "$thesaurus" '{lang}' "$n" < concept-info-local.sql | mysql "$db"
16 -done
Index: trunk/WikiWord/WikiWord/src/main/php/concept-info-local.sql
——@@ -1,61 +0,0 @@
2 -SET SESSION group_concat_max_len = 262144; -- 1024*256
3 -
4 -update {collection}_{thesaurus}_concept_info as I
5 -join {collection}_{thesaurus}_origin as O on I.lang = O.lang and I.concept = O.global_concept
6 -join {collection}_{lang}_definition as D on O.local_concept = D.concept and O.lang = "{lang}"
7 -set I.definition = D.definition
8 -where I.lang = "{lang}";
9 -
10 -update {collection}_{thesaurus}_concept_info as I
11 -join ( select O.global_concept as concept, O.lang as lang,
12 - group_concat(distinct concat(R.type, ":", R.name) separator "|" ) as pages
13 - from {collection}_{thesaurus}_origin as O
14 - join {collection}_{lang}_about as A on A.concept = O.local_concept and O.lang = "{lang}"
15 - join {collection}_{lang}_resource as R on R.id = A.resource
16 - where O.lang = "{lang}" and R.type IN (10, 50)
17 - group by O.global_concept, O.lang
18 - ) as X
19 -on I.concept = X.concept and I.lang = X.lang
20 -set I.pages = X.pages
21 -where I.lang = "{lang}";
22 -
23 -update {collection}_{thesaurus}_concept_info as I
24 -join ( select narrow as concept, group_concat(distinct concat(broad, ":", if (local_concept_name is null, "", local_concept_name)) separator "|") as broader
25 -from {collection}_{thesaurus}_broader
26 - left join {collection}_{thesaurus}_origin as O on O.global_concept = broad and O.lang = "{lang}"
27 - group by narrow ) as X
28 -on X.concept = I.concept and I.lang = "{lang}"
29 -set I.broader = X.broader;
30 -
31 -update {collection}_{thesaurus}_concept_info as I
32 -join ( select broad as concept, group_concat(distinct concat(narrow, ":", if (local_concept_name is null, "", local_concept_name)) separator "|") as narrower
33 -from {collection}_{thesaurus}_broader
34 - left join {collection}_{thesaurus}_origin as O on O.global_concept = narrow and O.lang = "{lang}"
35 - group by broad ) as X
36 -on X.concept = I.concept and I.lang = "{lang}"
37 -set I.narrower = X.narrower;
38 -
39 -update {collection}_{thesaurus}_concept_info as I
40 -join ( select concept1 as concept, group_concat(distinct concat(concept2, ":", if (local_concept_name is null, "", local_concept_name)) separator "|") as similar
41 -from {collection}_{thesaurus}_relation
42 - left join {collection}_{thesaurus}_origin as O on O.global_concept = concept2 and O.lang = "{lang}"
43 - where langmatch >= 1 or langref >= 1
44 - group by concept1 ) as X
45 -on X.concept = I.concept and I.lang = "{lang}"
46 -set I.similar = X.similar;
47 -
48 -update {collection}_{thesaurus}_concept_info as I
49 -join ( select concept1 as concept, group_concat(distinct concat(concept2, ":", if (local_concept_name is null, "", local_concept_name)) separator "|") as related
50 -from {collection}_{thesaurus}_relation
51 - left join {collection}_{thesaurus}_origin as O on O.global_concept = concept2 and O.lang = "{lang}"
52 - where bilink >= 1
53 - group by concept1 ) as X
54 -on X.concept = I.concept and I.lang = "{lang}"
55 -set I.related = X.related;
56 -
Index: trunk/WikiWord/WikiWord/src/main/php/config.sample.php
——@@ -1,81 +0,0 @@
2 -<?php
3 -
4 -#$wwAPI = "http://wikiword.wikimedia.de/api.php";
5 -$wwAPI = false;
6 -$wwAllowTranslate = false;
7 -$wwImageSearch = null; //NOTE: trinary: true forces image searche, false disables it. null makes it optional.
8 -
9 -$wwDBServer = "localhost";
10 -$wwDBUser = "wikiword";
11 -$wwDBPassword = "xxxxxxxxxx";
12 -$wwDBDatabase = "wikiword";
13 -
14 -$wwLanguages = array("en" => "Englisch", "de" => "German", "fr" => "French");
15 -$wwTablePrefix = "full";
16 -$wwThesaurusDataset = "thesaurus";
17 -
18 -#error_reporting(E_ALL);
19 -
20 -$wwMaxPreviewImages = 5;
21 -$wwMaxPreviewLinks = 20;
22 -$wwGalleryColumns = 5;
23 -$wwMaxGalleryImages = 200;
24 -$wwMaxDetailLinks = 1000;
25 -$wwThumbSize = 120;
26 -$wwThumbnailURL = "http://toolserver.org/tsthumb/tsthumb?f={name}&domain=commons.wikimedia.org&w={width}&h={height}";
27 -$wwImagePageURL = "http://commons.wikimedia.org/wiki/File:{name}";
28 -
29 -#$wwFakeCommonsConcepts = false;
30 -#$wwFakeCommonsPlural = false;
31 -$wwCommonsTablePrefix = "commonswiki_p.";
32 -
33 -$wwWikiInfoTable = "toolserver.wiki";
34 -$wwWikiDbName = "{lang}wiki_p";
35 -$wwWikiServerName = "sql-s{num}";
36 -
37 -$wwCommonsServerName = null;
38 -
39 -$wwTagsTable = null;
40 -
41 -#$wwGilNoiceTable = null;
42 -#$wwGilNoiceThreshold = 10;
43 -$wwFrequentImageThreshold = 10;
44 -
45 -$wwTagScores = array(
46 - 'Category:Featured_pictures_on_Wikimedia_Commons' => 10,
47 - 'Category:Featured_pictures_on_Wikipedia,_German' => 10,
48 -
49 - 'Template:Former_featured_picture' => 6,
50 -
51 - 'Template:Media_of_the_day' => 10,
52 - 'Template:Picture_of_the_day' => 10,
53 -
54 - 'Category:Quality_images' => 8,
55 -
56 - 'Category:Valued_image' => 8,
57 - 'Category:Former_valued_images' => 6,
58 - 'Category:Images_used_in_valued_image_sets' => 7,
59 -
60 - 'assessment:Media_of_the_day' => 10,
61 - 'assessment:Picture_of_the_day' => 10,
62 - 'assessment:Picture_of_the_week' => 15,
63 - 'assessment:Picture_of_the_month' => 20,
64 - 'assessment:Picture_of_the_year' => 25,
65 - 'assessment:Quality_image' => 8,
66 - 'assessment:Featured_picture' => 15,
67 -);
68 -
69 -
70 -$wwLabelPatterns = array(
71 - '/^assessment:.*_of_the_day([-_].*|$)/' => "PotD",
72 - '/^assessment:.*_of_the_week([-_].*|$)/' => "PotW",
73 - '/^assessment:.*_of_the_month([-_].*|$)/' => "PotM",
74 - '/^assessment:.*_of_the_year([-_].*|$)/' => "PotY",
75 - '/^assessment:(Featured|Former_featured_picture).*/' => 'FP',
76 - '/^assessment:Quiality.*/' => 'QI',
77 - '/^assessment:(Valued|Former_valued_images|Images_used_in_valued_image_sets).*/' => 'VI',
78 - '/^license:PD([-_].*|$)/' => 'PD',
79 - '/^license:GFDL([-_].*|$)/' => 'GFDL',
80 - '/^license:CC-BY-SA([-_].*|$)/i' => 'CC-BY-SA',
81 - '/^license:CC-BY([-_].*|$)/i' => 'BY',
82 -);
\ No newline at end of file
Index: trunk/WikiWord/WikiWordWeb/src/main/www/config.sample.php
——@@ -0,0 +1,81 @@
 2+<?php
 3+
 4+#$wwAPI = "http://wikiword.wikimedia.de/api.php";
 5+$wwAPI = false;
 6+$wwAllowTranslate = false;
 7+$wwImageSearch = null; //NOTE: trinary: true forces image searche, false disables it. null makes it optional.
 8+
 9+$wwDBServer = "localhost";
 10+$wwDBUser = "wikiword";
 11+$wwDBPassword = "xxxxxxxxxx";
 12+$wwDBDatabase = "wikiword";
 13+
 14+$wwLanguages = array("en" => "Englisch", "de" => "German", "fr" => "French");
 15+$wwTablePrefix = "full";
 16+$wwThesaurusDataset = "thesaurus";
 17+
 18+#error_reporting(E_ALL);
 19+
 20+$wwMaxPreviewImages = 5;
 21+$wwMaxPreviewLinks = 20;
 22+$wwGalleryColumns = 5;
 23+$wwMaxGalleryImages = 200;
 24+$wwMaxDetailLinks = 1000;
 25+$wwThumbSize = 120;
 26+$wwThumbnailURL = "http://toolserver.org/tsthumb/tsthumb?f={name}&domain=commons.wikimedia.org&w={width}&h={height}";
 27+$wwImagePageURL = "http://commons.wikimedia.org/wiki/File:{name}";
 28+
 29+#$wwFakeCommonsConcepts = false;
 30+#$wwFakeCommonsPlural = false;
 31+$wwCommonsTablePrefix = "commonswiki_p.";
 32+
 33+$wwWikiInfoTable = "toolserver.wiki";
 34+$wwWikiDbName = "{lang}wiki_p";
 35+$wwWikiServerName = "sql-s{num}";
 36+
 37+$wwCommonsServerName = null;
 38+
 39+$wwTagsTable = null;
 40+
 41+#$wwGilNoiceTable = null;
 42+#$wwGilNoiceThreshold = 10;
 43+$wwFrequentImageThreshold = 10;
 44+
 45+$wwTagScores = array(
 46+ 'Category:Featured_pictures_on_Wikimedia_Commons' => 10,
 47+ 'Category:Featured_pictures_on_Wikipedia,_German' => 10,
 48+
 49+ 'Template:Former_featured_picture' => 6,
 50+
 51+ 'Template:Media_of_the_day' => 10,
 52+ 'Template:Picture_of_the_day' => 10,
 53+
 54+ 'Category:Quality_images' => 8,
 55+
 56+ 'Category:Valued_image' => 8,
 57+ 'Category:Former_valued_images' => 6,
 58+ 'Category:Images_used_in_valued_image_sets' => 7,
 59+
 60+ 'assessment:Media_of_the_day' => 10,
 61+ 'assessment:Picture_of_the_day' => 10,
 62+ 'assessment:Picture_of_the_week' => 15,
 63+ 'assessment:Picture_of_the_month' => 20,
 64+ 'assessment:Picture_of_the_year' => 25,
 65+ 'assessment:Quality_image' => 8,
 66+ 'assessment:Featured_picture' => 15,
 67+);
 68+
 69+
 70+$wwLabelPatterns = array(
 71+ '/^assessment:.*_of_the_day([-_].*|$)/' => "PotD",
 72+ '/^assessment:.*_of_the_week([-_].*|$)/' => "PotW",
 73+ '/^assessment:.*_of_the_month([-_].*|$)/' => "PotM",
 74+ '/^assessment:.*_of_the_year([-_].*|$)/' => "PotY",
 75+ '/^assessment:(Featured|Former_featured_picture).*/' => 'FP',
 76+ '/^assessment:Quiality.*/' => 'QI',
 77+ '/^assessment:(Valued|Former_valued_images|Images_used_in_valued_image_sets).*/' => 'VI',
 78+ '/^license:PD([-_].*|$)/' => 'PD',
 79+ '/^license:GFDL([-_].*|$)/' => 'GFDL',
 80+ '/^license:CC-BY-SA([-_].*|$)/i' => 'CC-BY-SA',
 81+ '/^license:CC-BY([-_].*|$)/i' => 'BY',
 82+);
\ No newline at end of file

Status & tagging log