1 | module namespace cmd-model = "http://spraakbanken.gu.se/clarin/xquery/model"; |
---|
2 | |
---|
3 | (: |
---|
4 | $Id: cmd-model.xqm 248 2010-03-17 23:50:05Z ljo $ |
---|
5 | :) |
---|
6 | |
---|
7 | import module namespace xdb="http://exist-db.org/xquery/xmldb"; |
---|
8 | import module namespace util="http://exist-db.org/xquery/util"; |
---|
9 | |
---|
10 | declare variable $cmd-model:cmdiDatabaseURI as xs:string := "xmldb:exist:///db"; |
---|
11 | |
---|
12 | declare variable $cmd-model:commonFreqsPath as xs:string := "/db/common/clarin/freqs"; |
---|
13 | declare variable $cmd-model:cmdiMirrorPath as xs:string := "/db/cmdi-mirror"; |
---|
14 | |
---|
15 | declare variable $cmd-model:typeActorPath as xs:string := "MDGroup/Actors/Actor"; |
---|
16 | declare variable $cmd-model:typeActorPath0 as xs:string := "Actor"; |
---|
17 | declare variable $cmd-model:typeActorRolePath as xs:string := "MDGroup/Actors/Actor/Role"; |
---|
18 | |
---|
19 | declare variable $cmd-model:docTypeModelTrees as xs:string := "ModelTrees"; |
---|
20 | declare variable $cmd-model:docTypeTerms as xs:string := "Terms"; |
---|
21 | declare variable $cmd-model:docTypeSuffix as xs:string := "Values"; |
---|
22 | |
---|
23 | declare variable $cmd-model:xmlExt as xs:string := ".xml"; |
---|
24 | |
---|
25 | declare function cmd-model:elem($collection as xs:string, $path as xs:string, $depth as xs:integer) as element() { |
---|
26 | let $path-nodes := util:eval(fn:concat("collection('", $collection, "')//", $path)) |
---|
27 | let $path-count := count($path-nodes) |
---|
28 | |
---|
29 | let $subs := distinct-values($path-nodes/child::element()/name()) |
---|
30 | let $text-nodes := $path-nodes//text() |
---|
31 | let $text-count := count($text-nodes) |
---|
32 | let $text-count-distinct := count(distinct-values($text-nodes)) |
---|
33 | return |
---|
34 | <elem path="{fn:concat("//", $path)}" count="{$path-count}" count_text="{$text-count}" count_distinct_text="{$text-count-distinct}" > |
---|
35 | { if ($depth > 0) then |
---|
36 | for $elname in $subs[. != ''] |
---|
37 | return |
---|
38 | cmd-model:elem($collection, concat($path, '/', $elname), $depth - 1) |
---|
39 | else 'maxdepth' |
---|
40 | }</elem> |
---|
41 | }; |
---|
42 | |
---|
43 | declare function cmd-model:paths($n) { |
---|
44 | for $el in $n |
---|
45 | return <elem name="{$el/name()}"> { |
---|
46 | for $anc in $el/parent::element() |
---|
47 | return util:node-xpath($anc) |
---|
48 | }</elem> |
---|
49 | }; |
---|
50 | |
---|
51 | (: |
---|
52 | |
---|
53 | :) |
---|
54 | declare function cmd-model:recurse-collections-model($collection as xs:string, $type-name as xs:string, $depth as xs:integer) as item()* { |
---|
55 | let $children := xdb:get-child-collections($collection) |
---|
56 | return |
---|
57 | if (fn:exists($children)) then |
---|
58 | let $child-results := |
---|
59 | for $child in $children |
---|
60 | return |
---|
61 | cmd-model:recurse-collections-model(fn:concat($collection, '/', xs:string($child)), $type-name, $depth), |
---|
62 | $current := cmd-model:create-doc($collection, $type-name, $depth) |
---|
63 | return ($current, $child-results) |
---|
64 | else |
---|
65 | cmd-model:create-doc($collection, $type-name, $depth) |
---|
66 | }; |
---|
67 | |
---|
68 | (: |
---|
69 | |
---|
70 | :) |
---|
71 | declare function cmd-model:create-doc($collection as xs:string, $type-name as xs:string, $depth as xs:integer) as xs:string* { |
---|
72 | (: if newer data available :) |
---|
73 | cmd-model:store-result($collection, cmd-model:elem($collection, $type-name, $depth), $type-name, $depth) |
---|
74 | (:else () :) |
---|
75 | }; |
---|
76 | |
---|
77 | (: |
---|
78 | |
---|
79 | :) |
---|
80 | declare function cmd-model:get-result-doc($collection as xs:string, $type-name as xs:string, $depth as xs:integer) as item()* { |
---|
81 | let $name-last := text:groups($type-name, "/(\w+)$")[last()], |
---|
82 | $new-name := if (fn:empty($name-last)) then $type-name else $name-last, |
---|
83 | $dummy := if (cmd-model:is-result-available($collection, fn:concat("/", $new-name, xs:string($depth)))) then |
---|
84 | () |
---|
85 | else |
---|
86 | cmd-model:create-doc($collection, $type-name, $depth) |
---|
87 | return |
---|
88 | fn:doc(fn:concat($collection, "/", $new-name, xs:string($depth), $cmd-model:xmlExt)) |
---|
89 | }; |
---|
90 | |
---|
91 | (: |
---|
92 | Function for telling wether the result is already available or not. |
---|
93 | :) |
---|
94 | declare function cmd-model:is-result-available($collection as xs:string, $result-ref as xs:string) as xs:boolean { |
---|
95 | fn:doc-available(fn:concat($collection, $result-ref, $cmd-model:xmlExt)) |
---|
96 | }; |
---|
97 | |
---|
98 | (: |
---|
99 | Store the calculated frequencies for reuse. |
---|
100 | If more than one collection is given the result is stored in the common |
---|
101 | collection for reuse. |
---|
102 | :) |
---|
103 | declare function cmd-model:store-result($coll-names as xs:string+, $entries as element()*, $type-name as xs:string, $depth as xs:integer) as xs:string { |
---|
104 | let $clarin-writer := fn:doc("/db/clarin/writer.xml"), |
---|
105 | $dummy := xdb:login($cmd-model:cmdiDatabaseURI, $clarin-writer//write-user/text(), $clarin-writer//write-user-cred/text()) |
---|
106 | return |
---|
107 | |
---|
108 | if (fn:exists($coll-names[2])) then |
---|
109 | (: Det gÀller fler Àn en samling. :) |
---|
110 | xdb:store($cmd-model:commonFreqsPath, cmd-model:make-compound-doc-name($coll-names, $type-name, $depth), cmd-model:make-doc-element-of-type($type-name, $coll-names, $entries, $depth)) |
---|
111 | else |
---|
112 | (: Det gÀller endast en samling. :) |
---|
113 | let $dummy := util:log('debug', fn:concat('Stores ', $type-name, ' in ', $coll-names)) |
---|
114 | return xdb:store($coll-names, cmd-model:make-doc-name($coll-names, $type-name, xs:string($depth), fn:false()), cmd-model:make-doc-element-of-type($type-name, (), $entries, xs:string($depth))) |
---|
115 | }; |
---|
116 | |
---|
117 | (: |
---|
118 | Create document name for type () with or without collection path. |
---|
119 | :) |
---|
120 | declare function cmd-model:make-doc-name($coll-name as xs:string?, $type-name as xs:string, $depth as xs:string, $incl-path as xs:boolean) as xs:string { |
---|
121 | let $doc-name := fn:concat($type-name, $depth, $cmd-model:xmlExt) |
---|
122 | return |
---|
123 | if ($incl-path) then |
---|
124 | fn:concat($coll-name, "/", $doc-name) |
---|
125 | else |
---|
126 | $doc-name |
---|
127 | }; |
---|
128 | |
---|
129 | (: |
---|
130 | Create document name with md5-hash for selected collections (or types) for reuse. |
---|
131 | :) |
---|
132 | declare function cmd-model:make-compound-doc-name($coll-names as xs:string+, $type-name as xs:string, $depth as xs:string) as xs:string { |
---|
133 | let $name-prefix := fn:concat($type-name, $depth) |
---|
134 | return |
---|
135 | fn:concat($name-prefix, "-", util:hash(string-join($coll-names, ""), "MD5"), $cmd-model:xmlExt) |
---|
136 | }; |
---|
137 | |
---|
138 | (: |
---|
139 | Skapa ett element av angiven typ. |
---|
140 | :) |
---|
141 | declare function cmd-model:make-element-of-type($type-name as xs:string, $count as xs:string, $text-count as xs:string, $text-types-count as xs:string, $value as xs:string) as element() { |
---|
142 | element {$type-name} { |
---|
143 | |
---|
144 | attribute count {$freq}, |
---|
145 | attribute text-count {$rank}, |
---|
146 | attribute text-types-count {$text-types}, |
---|
147 | text {$value} |
---|
148 | } |
---|
149 | }; |
---|
150 | |
---|
151 | (: |
---|
152 | Skapa ett dokumentelement av angiven typ. |
---|
153 | :) |
---|
154 | declare function cmd-model:make-doc-element-of-type($type-name as xs:string, $coll-names as xs:string*, $entries as element()*, $depth as xs:string) as element() { |
---|
155 | let $depth-value := attribute depth {$depth}, |
---|
156 | $coll-names-value := if (fn:empty($coll-names)) then () else attribute colls {fn:string-join($coll-names, ",")} |
---|
157 | return |
---|
158 | element {cmd-model:get-doc-type-element-name($type-name)} { |
---|
159 | $depth-value, |
---|
160 | $coll-names-value, |
---|
161 | attribute created {fn:current-dateTime()}, |
---|
162 | $entries |
---|
163 | } |
---|
164 | }; |
---|
165 | |
---|
166 | (: |
---|
167 | Skapa elementnamn för dokumentet av typ. |
---|
168 | :) |
---|
169 | declare function cmd-model:get-doc-type-element-name($type-name as xs:string) as xs:string { |
---|
170 | let $name-last := text:groups($type-name, "/(\w+)$")[last()] |
---|
171 | return |
---|
172 | if (fn:empty($name-last)) then $cmd-model:ModelTrees else $cmd-model:Terms |
---|
173 | |
---|
174 | }; |
---|