1 |
46285
|
aenne.loeh
|
<RESOURCE_PROFILE>
|
2 |
|
|
<HEADER>
|
3 |
|
|
<RESOURCE_IDENTIFIER value="6c9b4eea-45fc-4017-9fdb-8818afcd80ea_VHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZXMvVHJhbnNmb3JtYXRpb25SdWxlRFNSZXNvdXJjZVR5cGU="/>
|
4 |
|
|
<RESOURCE_TYPE value="TransformationRuleDSResourceType"/>
|
5 |
|
|
<RESOURCE_KIND value="TransformationRuleDSResources"/>
|
6 |
|
|
<RESOURCE_URI value=""/>
|
7 |
|
|
<DATE_OF_CREATION value="2016-07-13T11:51:55+00:00"/>
|
8 |
|
|
</HEADER>
|
9 |
|
|
<BODY>
|
10 |
|
|
<CONFIGURATION>
|
11 |
|
|
<IMPORTED/>
|
12 |
|
|
<SCRIPT>
|
13 |
|
|
<TITLE>dc_cleaning_OPENAIREplus_compliant_handle</TITLE>
|
14 |
|
|
<CODE>declare_script "dc_cleaning_OpenAIREplus_compliant_handle";
|
15 |
|
|
declare_ns oaf = "http://namespace.openaire.eu/oaf";
|
16 |
|
|
declare_ns dri = "http://www.driver-repository.eu/namespace/dri";
|
17 |
|
|
declare_ns dr = "http://www.driver-repository.eu/namespace/dr";
|
18 |
|
|
declare_ns dc = "http://purl.org/dc/elements/1.1/";
|
19 |
|
|
declare_ns prov = "http://www.openarchives.org/OAI/2.0/provenance";
|
20 |
|
|
$var0 = "''";
|
21 |
|
|
$varFP7 = "'corda_______::'";
|
22 |
|
|
$varH2020 = "'corda__h2020::'";
|
23 |
|
|
$varFCT="'fct_________::'";
|
24 |
|
|
$varDummy = "''";
|
25 |
|
|
static $varDatasourceid = getValue(PROFILEFIELD, [xpath:"concat('collection(&apos;/db/DRIVER/RepositoryServiceResources&apos;)//RESOURCE_PROFILE[.//EXTRA_FIELDS/FIELD[key=&quot;NamespacePrefix&quot;][value=&quot;', //oaf:datasourceprefix, '&quot;]]')", xpath:"//EXTRA_FIELDS/FIELD[key='OpenAireDataSourceId']/value"]);
|
26 |
|
|
static $varRepoid = xpath:"//dri:repositoryId";
|
27 |
|
|
static $varOfficialname = getValue(PROFILEFIELD, [xpath:"concat('collection(&apos;/db/DRIVER/RepositoryServiceResources&apos;)//RESOURCE_PROFILE[.//EXTRA_FIELDS/FIELD[key=&quot;NamespacePrefix&quot;][value=&quot;', //oaf:datasourceprefix, '&quot;]]')", xpath:"//CONFIGURATION/OFFICIAL_NAME"]);
|
28 |
|
|
dri:objIdentifier = xpath:"//dri:objIdentifier";
|
29 |
|
|
dri:repositoryId = $varRepoid;
|
30 |
|
|
dri:recordIdentifier = xpath:"//dri:recordIdentifier";
|
31 |
|
|
apply xpath:"//dc:creator" if xpath:"string-length(.) > 0" dc:creator = xpath:"normalize-space(.)"; else $varDummy = "''";
|
32 |
|
|
if xpath:"//dc:title[string-length(.)> 0]" $varDummy = "''"; else dc:coverage = skipRecord();
|
33 |
|
|
apply xpath:"//dc:title" if xpath:"string-length(.) > 0" dc:title = xpath:"normalize-space(.)"; else $varDummy = "''";
|
34 |
|
|
apply xpath:"//dc:subject" if xpath:"string-length(.) > 0" dc:subject = xpath:"normalize-space(.)"; else $varDummy = "''";
|
35 |
|
|
apply xpath:"//dc:publisher" if xpath:"string-length(.) > 0" dc:publisher = xpath:"normalize-space(.)"; else $varDummy = "''";
|
36 |
|
|
apply xpath:"//dc:source" if xpath:"string-length(.) > 0" dc:source = xpath:"normalize-space(.)"; else $varDummy = "''";
|
37 |
|
|
dc:contributor = xpath:"//dc:contributor";
|
38 |
|
|
dc:description = xpath:"string-join(//dc:description[concat(normalize-space(.), '')], codepoints-to-string(10))";
|
39 |
|
|
$varHttpTest = "''";
|
40 |
|
|
if xpath:"//dc:identifier[starts-with(., 'http://hdl.handle') or starts-with(., 'https://hdl.handle')][not(starts-with(., 'http://hdl.handle.net/123456789') or starts-with(., 'https://hdl.handle.net/123456789'))]" $varHttpTest = "true"; else dc:identifier = skipRecord();
|
41 |
|
|
apply xpath:"//dc:identifier" if xpath:"starts-with(normalize-space(.), 'http://hdl.handle') or starts-with(., 'https://hdl.handle')" dc:identifier = xpath:"normalize-space(.)"; else dr:CobjIdentifier = xpath:"normalize-space(.)";
|
42 |
|
|
dr:dateOfCollection = xpath:"//dri:dateOfCollection";
|
43 |
|
|
static dr:dateOfTransformation = xpath:"current-dateTime()";
|
44 |
|
|
dc:type = xpath:"//dc:type";
|
45 |
|
|
dc:format = xpath:"//dc:format";
|
46 |
|
|
dc:date = xpath:"//dc:date";
|
47 |
|
|
dc:language = Convert(xpath:"//dc:language", Languages);
|
48 |
|
|
//dc:language = "eng";
|
49 |
|
|
//if xpath:"//dc:rights[text()='info:eu-repo/semantics/openAccess']" dc:publisher = xpath:"//dc:publisher"; else dc:publisher = skipRecord();
|
50 |
|
|
oaf:dateAccepted = Convert(xpath:"descendant-or-self::dc:date", DateISO8601, "yyyy-MM-dd", "min()");
|
51 |
|
|
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 $var0 = "''";
|
52 |
|
|
// FP7
|
53 |
|
|
oaf:projectid = xpath:"//dc:relation[matches(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', 'i')]/concat($varFP7, replace(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', '$2', 'i'))";
|
54 |
|
|
// H2020
|
55 |
|
|
oaf:projectid = xpath:"//dc:relation[matches(normalize-space(.), '(info:eu-repo/grantagreement/ec/h2020/)(\d\d\d\d\d\d)(.*)', 'i')]/concat($varH2020, replace(normalize-space(.), '(info:eu-repo/grantagreement/ec/h2020/)(\d\d\d\d\d\d)(.*)', '$2', 'i'))";
|
56 |
|
|
// FCT
|
57 |
|
|
oaf:projectid = xpath:"//dc:relation[matches(normalize-space(.), '(info:eu-repo/grantagreement/fct)/(.+)/(\d+)(.*)', 'i')]/concat($varFCT, replace(normalize-space(.), '(info:eu-repo/grantagreement/fct)/(.+)/(\d+)(.*)', '$3', 'i'))";
|
58 |
|
|
dc:relation = xpath:"//dc:relation";
|
59 |
|
|
// apply xpath:"//dc:relation" if xpath:"matches(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', 'i')" oaf:projectid = xpath:"concat($var1, replace(normalize-space(.), '(info:eu-repo/grantagreement/ec/fp7/)(\d\d\d\d\d\d)(.*)', '$2', 'i'))"; else dc:relation = xpath:"normalize-space(.)";
|
60 |
|
|
//comment-js-09-10-2012 apply xpath:"//dc:rights" if xpath:"starts-with(normalize-space(.), 'info:eu-repo/semantics')" dc:rights = empty; else dc:rights = xpath:"normalize-space(.)";
|
61 |
|
|
//
|
62 |
|
|
oaf:collectedDatasourceid = xpath:"$varDatasourceid";
|
63 |
|
|
//
|
64 |
|
|
dr:CobjCategory = Convert(xpath:"//dc:type", TextTypologies);
|
65 |
|
|
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:".";
|
66 |
|
|
if xpath:"//dc:rights[starts-with(normalize-space(.), 'info:eu-repo/semantics')]" $var0 = "''"; else oaf:accessrights = "OPEN";
|
67 |
|
|
//if xpath:"count(//dc:rights) = 0" oaf:accessrights = "OPEN"; else $var0 = "''";
|
68 |
|
|
// oaf:accessrights = Convert(xpath:"normalize-space(//dc:rights)", AccessRights);
|
69 |
|
|
//
|
70 |
|
|
static oaf:collectedFrom = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
|
71 |
|
|
static oaf:hostedBy = set("''", @name = $varOfficialname; , @id = $varDatasourceid;);
|
72 |
|
|
//
|
73 |
|
|
// $varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '(10[.][0-9]{4,}[^\s"/<>]*/[^\s"<>]+)');
|
74 |
|
|
$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&<>])\S)+)\b');
|
75 |
|
|
//$varId = identifierExtract('["//dc:identifier", "//dc:relation"]' , xpath:"./*[local-name()='record']" , '\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&'<>])\S)+)\b');
|
76 |
|
|
// 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
|
77 |
|
|
$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+)');
|
78 |
|
|
// $varUrn = xpath:"substring-after(//dc:relation[starts-with(normalize-space(.), 'info:eu-repo/semantics/altIdentifier/urn/')], 'info:eu-repo/semantics/altIdentifier/urn/')";
|
79 |
|
|
$varUrn = identifierExtract('["//dc:relation[starts-with(normalize-space(.), \"info:eu-repo/semantics/altIdentifier/urn/\")]"]' , xpath:"./*[local-name()='record']" , '(?!info:eu-repo/semantics/altIdentifier/urn/)(urn:nbn:.*)');
|
80 |
|
|
oaf:identifier = set(xpath:"$varId//value", @identifierType = "doi";);
|
81 |
|
|
oaf:identifier = set(xpath:"$varPmId//value", @identifierType = "pmid";);
|
82 |
|
|
oaf:identifier = set(xpath:"$varUrn//value", @identifierType = "urn";);
|
83 |
|
|
oaf:datasourceprefix = xpath:"//oaf:datasourceprefix";
|
84 |
|
|
// specific to NORA (Norwegian Research Archive)
|
85 |
|
|
apply xpath:"//dc:identifier[starts-with(., 'Fulltext')]" if xpath:"contains(., 'http')" oaf:fulltext = xpath:"substring-after(., 'Fulltext ')"; else $varDummy = "''";
|
86 |
|
|
end</CODE>
|
87 |
|
|
</SCRIPT>
|
88 |
|
|
</CONFIGURATION>
|
89 |
|
|
<STATUS/>
|
90 |
|
|
<SECURITY_PARAMETERS/>
|
91 |
|
|
</BODY>
|
92 |
|
|
</RESOURCE_PROFILE>
|