D-Net
/
dnet-hadoop
13
4
Fork 10
Code Issues 14 Pull Requests 8 Projects 2 Releases 1 Wiki Activity

[EOSC TAG] before adding the tag added a step to verify the same tag is not already present

This commit is contained in:
Miriam Baglioni 2022-07-21 14:36:48 +02:00
parent 5143a80232
commit 56d09e6348
1 changed files with 14 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
dhp-workflows/dhp-enrichment/src/main/java/eu/dnetlib/dhp/bulktag/SparkEoscTag.java
View File

@ -80,13 +80,14 @@ public class SparkEoscTag {
if (containsCriteriaNotebook(s)) { if (containsCriteriaNotebook(s)) {
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent()) if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>()); s.setEoscifguidelines(new ArrayList<>());
s.getEoscifguidelines().add(newInstance("EOSC::Jupyter Notebook", "EOSC::Jupyter Notebook", "", "compliesWith")); addEIG(s.getEoscifguidelines(), "EOSC::Jupyter Notebook", "EOSC::Jupyter Notebook", "", "compliesWith");
} }
if (containsCriteriaGalaxy(s)) { if (containsCriteriaGalaxy(s)) {
if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent()) if (!Optional.ofNullable(s.getEoscifguidelines()).isPresent())
s.setEoscifguidelines(new ArrayList<>()); s.setEoscifguidelines(new ArrayList<>());
s.getEoscifguidelines().add(newInstance("EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith")); addEIG(s.getEoscifguidelines(),"EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith");
} }
return s; return s;
}, Encoders.bean(Software.class)) }, Encoders.bean(Software.class))
@ -103,15 +104,15 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/otherresearchproduct", OtherResearchProduct.class) readPath(spark, inputPath + "/otherresearchproduct", OtherResearchProduct.class)
.map((MapFunction<OtherResearchProduct, OtherResearchProduct>) orp -> { .map((MapFunction<OtherResearchProduct, OtherResearchProduct>) orp -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(orp.getEoscifguidelines()).isPresent()) if (!Optional.ofNullable(orp.getEoscifguidelines()).isPresent())
orp.setEoscifguidelines(new ArrayList<>()); orp.setEoscifguidelines(new ArrayList<>());
if (containsCriteriaGalaxy(orp)) { if (containsCriteriaGalaxy(orp)) {
orp.getEoscifguidelines().add(newInstance("EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith")); addEIG(orp.getEoscifguidelines(),"EOSC::Galaxy Workflow", "EOSC::Galaxy Workflow", "", "compliesWith");
} }
if (containscriteriaTwitter(orp)) { if (containscriteriaTwitter(orp)) {
orp.getEoscifguidelines().add(newInstance("EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith")); addEIG(orp.getEoscifguidelines(),"EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith");
} }
return orp; return orp;
}, Encoders.bean(OtherResearchProduct.class)) }, Encoders.bean(OtherResearchProduct.class))
@ -128,11 +129,11 @@ public class SparkEoscTag {
readPath(spark, inputPath + "/dataset", Dataset.class) readPath(spark, inputPath + "/dataset", Dataset.class)
.map((MapFunction<Dataset, Dataset>) d -> { .map((MapFunction<Dataset, Dataset>) d -> {
List<StructuredProperty> sbject;
if (!Optional.ofNullable(d.getEoscifguidelines()).isPresent()) if (!Optional.ofNullable(d.getEoscifguidelines()).isPresent())
d.setEoscifguidelines(new ArrayList<>()); d.setEoscifguidelines(new ArrayList<>());
if (containscriteriaTwitter(d)) { if (containscriteriaTwitter(d)) {
d.getEoscifguidelines().add(newInstance("EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith")); addEIG(d.getEoscifguidelines(),"EOSC::Twitter Data", "EOSC::Twitter Data", "", "compliesWith");
} }
return d; return d;
}, Encoders.bean(Dataset.class)) }, Encoders.bean(Dataset.class))
@ -148,6 +149,12 @@ public class SparkEoscTag {
.json(inputPath + "/dataset"); .json(inputPath + "/dataset");
} }
private static void addEIG(List<EoscIfGuidelines> eoscifguidelines, String code, String label, String url, String sem) {
if (!eoscifguidelines.stream().anyMatch(eig -> eig.getCode().equals(code)))
eoscifguidelines.add(newInstance(code, label, url, sem));
}
private static boolean containscriteriaTwitter(Result r) { private static boolean containscriteriaTwitter(Result r) {
Set<String> words = getWordsSP(r.getTitle()); Set<String> words = getWordsSP(r.getTitle());
words.addAll(getWordsF(r.getDescription())); words.addAll(getWordsF(r.getDescription()));