|
1
|
<RESOURCE_PROFILE>
|
|
2
|
<HEADER>
|
|
3
|
<RESOURCE_IDENTIFIER value="b844840b-4efa-40a1-a1ce-90b4b63972c2_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
|
|
4
|
<RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
|
|
5
|
<RESOURCE_KIND value="TransformationRuleDSResources"/>
|
|
6
|
<RESOURCE_URI value=""/>
|
|
7
|
<DATE_OF_CREATION value="2013-02-28T12:14:22+01:00"/>
|
|
8
|
</HEADER>
|
|
9
|
<BODY>
|
|
10
|
<CONFIGURATION>
|
|
11
|
<IMPORTED/>
|
|
12
|
<SCRIPT>
|
|
13
|
<TITLE>dc_cleaning_OPENAIREplus_compliant</TITLE>
|
|
14
|
<CODE><![CDATA[
|
|
15
|
declare_script "dc_cleaning_OpenAIREplus_compliant";
|
|
16
|
declare_ns oaf = "http://namespace.openaire.eu/oaf";
|
|
17
|
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
|
|
18
|
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
|
|
19
|
declare_ns dc = "http://purl.org/dc/elements/1.1/";
|
|
20
|
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
|
|
21
|
$var0 = "''";
|
|
22
|
$varFP7 = "'corda_______::'";
|
|
23
|
$varH2020 = "'corda__h2020::'";
|
|
24
|
$varDummy = "''";
|
|
25
|
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
|
26
|
static $varRepoid = xpath:"//dri:repositoryId";
|
|
27
|
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//OFFICIAL_NAME"]);
|
|
28
|
dri:objIdentifier = xpath:"//dri:objIdentifier";
|
|
29
|
dri:repositoryId = $varRepoid;
|
|
30
|
dri:recordIdentifier = RegExpr(xpath:"//dri:recordIdentifier", $var0, "s/^(.*)(::)/$2/");
|
|
31
|
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0 and not(contains(., 'US National Cancer Institute'))" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
|
|
32
|
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
|
|
33
|
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
|
|
34
|
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
|
|
35
|
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
|
|
36
|
dc:contributor = xpath:"//dc:contributor";
|
|
37
|
dc:description = xpath:"//dc:description";
|
|
38
|
$varHttpTest = "''";
|
|
39
|
if xpath:"//dc:identifier[starts-with(., 'http')]" $varHttpTest = "true"; else dc:identifier = skipRecord();
|
|
40
|
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
|
|
41
|
static dr:dateOfCollection = getValue(CURRENTDATE, []);
|
|
42
|
dc:type = xpath:"//dc:type";
|
|
43
|
dc:format = xpath:"//dc:format";
|
|
44
|
dc:date = xpath:"//dc:date";
|
|
45
|
dc:language = Convert(xpath:"//dc:language", Languages);
|
|
46
|
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
|
|
47
|
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
|
|
48
|
apply xpath:"//dc:date" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/date')" oaf:embargoenddate = RegExpr(xpath:"normalize-space(.)", $var0, "s/^(.*info:eu-repo\/date\/embargoEnd\/)//gmi"); else $var0 = "''";
|
|
49
|
// FP7
|
|
50
|
oaf:projectid = xpath:"//dc:relation[matches(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', 'i')]/concat($varFP7, replace(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', '$2', 'i'))";
|
|
51
|
// H2020
|
|
52
|
oaf:projectid = xpath:"//dc:relation[matches(normalize-space(.), '(info:eu-repo/grantagreement/ec/h2020/)(\d\d\d\d\d\d)(.*)', 'i')]/concat($varH2020, replace(normalize-space(.), '(info:eu-repo/grantagreement/ec/h2020/)(\d\d\d\d\d\d)(.*)', '$2', 'i'))";
|
|
53
|
dc:relation = xpath:"//dc:relation";
|
|
54
|
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
|
|
55
|
// static oaf:datasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
|
56
|
//
|
|
57
|
//
|
|
58
|
static $varDsType = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//CONFIGURATION/DATASOURCE_TYPE"]);
|
|
59
|
//if xpath:"$varDsType='aggregator'" oaf:hostingDatasourceid = xpath:"//prov:baseURL"; else oaf:hostingDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
|
60
|
oaf:collectedDatasourceid = getValue(PROFILEFIELD, [xpath:"//dri:repositoryId", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
|
61
|
//
|
|
62
|
apply xpath:"//dc:type" if xpath:"." dr:CobjCategory = Convert(xpath:"normalize-space(.)", TextTypologies); else dc:type = xpath:".";
|
|
63
|
apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" oaf:accessrights = Convert(xpath:"normalize-space(.)", AccessRights); else dc:rights = xpath:".";
|
|
64
|
if xpath:"//dc:rights[starts-with(normalize-space(.), 'info:eu-repo/semantics')]" $var0 = "''"; else oaf:accessrights = "OPEN";
|
|
65
|
//
|
|
66
|
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
|
|
67
|
static oaf:hostedBy = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
|
|
68
|
//
|
|
69
|
$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/<>]*/[^\s"<>]+)');
|
|
70
|
// 1st param: list of xpath expresssions to be applied on the metadata in json syntax; 2nd param: xpath expression for the metadata record; 3rd param reg expr that matches with a negative lookahead for the first group and extracts digits of the second group
|
|
71
|
$varPmId = identifierExtract('["//dc:relation[starts-with(normalize-space(.), \"info:eu-repo/semantics/altIdentifier/pmid/\")]"]' , xpath:"./*[local-name()='record']" , '(?!info:eu-repo/semantics/altIdentifier/pmid/)(\d+)');
|
|
72
|
$varUrn = identifierExtract('["//dc:relation[starts-with(normalize-space(.), \"info:eu-repo/semantics/altIdentifier/urn/\")]"]' , xpath:"./*[local-name()='record']" , '(?!info:eu-repo/semantics/altIdentifier/urn/)(urn:nbn:.*)');
|
|
73
|
oaf:identifier = set(xpath:"$varId//value", @identifierType = "doi";);
|
|
74
|
oaf:identifier = set(xpath:"$varPmId//value", @identifierType = "pmid";);
|
|
75
|
oaf:identifier = set(xpath:"$varUrn//value", @identifierType = "urn";);
|
|
76
|
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
|
|
77
|
end
|
|
78
|
]]></CODE>
|
|
79
|
</SCRIPT>
|
|
80
|
</CONFIGURATION>
|
|
81
|
<STATUS/>
|
|
82
|
<SECURITY_PARAMETERS/>
|
|
83
|
</BODY>
|
|
84
|
</RESOURCE_PROFILE>
|