DD-174 Signposting (#46)

vicding-mi · web-flow · commit 2e6deaec7bcf · 2021-02-18T11:08:41.000+01:00
* set default when no config is set for signposting
* modification according to reviews
* move long json string from code to bunddle
* allow empty config on the level 2 profile
* revision based on Herbert feedback
* coding style cleanup SignpostingResources
* remove leading comma
* fix capitalize with header name
diff --git a/src/main/java/edu/harvard/iq/dataverse/util/SignpostingResources.java b/src/main/java/edu/harvard/iq/dataverse/util/SignpostingResources.java
@@ -1,27 +1,19 @@
 package edu.harvard.iq.dataverse.util;
 
-/**
- * Eko Indarto, DANS
- * Vic Ding, DANS
- *
- * This file prepares the resources used in Signposting
- *
- * It requires correspondence configuration to function well.
- * The configuration key used is SignpostingConf.
- * It is a json structure shown below
- *
- * useDefaultFileType is an on/off switch during linkset creating time, it controls whether the default type is
- * used, which is always Dataset
- * {
- *   "indetifier-schema": {"ORCID":"https://orcid.org/", "ISNI":"https://isni.org/isni/", "ScopusID":"https://www.scopus.com/authid/detail.uri?authorId="},
- *   "license": {"CC0":"https://creativecommons.org/licenses/by/4.0/", "MIT": "https://url", "APACHE":"https://url"},
- *   "cite-as": {"doi":"https://citation.crosscite.org/format?style=bibtex&doi=", "type":"application/vnd.datacite.datacite+json"},
- *   "useDefaultFileType": true,
- *   "defaultFileTypeValue": "https://schema.org/Dataset"
- * }
- *
- * The configuration can be modified during run time by the administrator.
- *
+/*
+  Eko Indarto, DANS
+  Vic Ding, DANS
+
+  This file prepares the resources used in Signposting
+
+  It requires correspondence configuration to function well.
+  The configuration key used is SignpostingConf.
+  It is a json structure shown below
+
+  useDefaultFileType is an on/off switch during linkset creating time, it controls whether the default type is
+  used, which is always Dataset
+
+  The configuration can be modified during run time by the administrator.
  */
 
 import edu.harvard.iq.dataverse.*;
@@ -48,13 +40,14 @@ public class SignpostingResources {
     private static final Logger logger = Logger.getLogger(SignpostingResources.class.getCanonicalName());
     SystemConfig systemConfig;
     DatasetVersion workingDatasetVersion;
-    JsonObject idschemaJsonObj;
     JsonObject licJsonObj;
-    JsonObject citeAsJsonObj;
+    JsonObject describedByJsonObj;
     Boolean useDefaultFileType;
     String defaultFileTypeValue;
+    int maxAuthors;
+    int maxItems;
 
-    public SignpostingResources(SystemConfig systemConfig, DatasetVersion workingDatasetVersion, String jsonSetting){
+    public SignpostingResources(SystemConfig systemConfig, DatasetVersion workingDatasetVersion, String jsonSetting) {
         this.systemConfig = systemConfig;
         this.workingDatasetVersion = workingDatasetVersion;
         if (jsonSetting == null) {
@@ -63,46 +56,47 @@ public SignpostingResources(SystemConfig systemConfig, DatasetVersion workingDat
         JsonReader jsonReader = Json.createReader(new StringReader(jsonSetting));
         JsonObject spJsonSetting = jsonReader.readObject();
         jsonReader.close();
-        idschemaJsonObj = spJsonSetting.getJsonObject("indetifier-schema");
         licJsonObj = spJsonSetting.getJsonObject("license");
-        citeAsJsonObj = spJsonSetting.getJsonObject("cite-as");
+        describedByJsonObj = spJsonSetting.getJsonObject("describedby");
         useDefaultFileType = spJsonSetting.getBoolean("useDefaultFileType", true);
         defaultFileTypeValue = spJsonSetting.getString("defaultFileTypeValue", "https://schema.org/Dataset");
+        maxAuthors = spJsonSetting.getInt("maxAuthors", 5);
+        maxItems = spJsonSetting.getInt("maxItems", 5);
     }
 
     /**
      * Get identifier schema for each author
      *
-     * Author may have identifiers from different providers
-     * ORCID: the url format is https://orcid.org/:id
-     * ISNI: the url format is https://isni.org/isni/:id
-     * ScopusID: the url format is https://www.scopus.com/authid/detail.uri?authorId=:id
-     * VIAF: the url format is http://viaf.org/viaf/:id
-     *
      * For example:
-     *      if author has VIAF
-    *       Link: <http://viaf.org/viaf/:id/>; rel="author"
+     * if author has VIAF
+     * Link: <http://viaf.org/viaf/:id/>; rel="author"
      *
-     * @param datasetAuthors
-     * @return
+     * @param datasetAuthors list of all DatasetAuthor object
+     * @return all the non empty author links in a string
      */
     private String getIdentifierSchema(List<DatasetAuthor> datasetAuthors) {
-        StringBuilder sb = new StringBuilder();
+        String singleAuthorString;
         String identifierSchema = "";
 
-        for (DatasetAuthor da:datasetAuthors){
-            if (da.getIdValue() != null && !da.getIdValue().isEmpty()) {
-                sb.append("<").
-                        append(idschemaJsonObj.getString(da.getIdType())).
-                        append(da.getIdValue()).
-                        append(">;rel=\"author\"");
-                if (identifierSchema == "") {
-                    identifierSchema = sb.toString();
+        for (DatasetAuthor da : datasetAuthors) {
+            logger.info(String.format(
+                    "idtype: %s; idvalue: %s, affiliation: %s; identifierUrl: %s",
+                    da.getIdType(),
+                    da.getIdValue(),
+                    da.getAffiliation(),
+                    da.getIdentifierAsUrl()
+            ));
+            if (da.getIdentifierAsUrl() != null && !da.getIdentifierAsUrl().trim().isEmpty()) {
+                singleAuthorString = "<" + da.getIdentifierAsUrl() + ">;rel=\"author\"";
+                if (Objects.equals(identifierSchema, "")) {
+                    identifierSchema = singleAuthorString;
                 } else {
-                    identifierSchema = String.join(",", identifierSchema, sb.toString());
+                    identifierSchema = String.join(",", identifierSchema, singleAuthorString);
                 }
             }
         }
+
+        logger.info(String.format("identifierSchema: %s", identifierSchema));
         return identifierSchema;
     }
 
@@ -111,7 +105,7 @@ private String getIdentifierSchema(List<DatasetAuthor> datasetAuthors) {
      *
      * @return comma delimited string
      */
-    public String getLinks(){
+    public String getLinks() {
         List<String> valueList = new LinkedList<>();
         Dataset ds = workingDatasetVersion.getDataset();
 
@@ -125,8 +119,8 @@ public String getLinks(){
             valueList.add(citeAs);
         }
 
-        String describedby = "<" + citeAsJsonObj.getString(ds.getProtocol()) + ds.getAuthority() + "/"
-                + ds.getIdentifier() + ">;rel=\"describedby\"" + "; type=\""+ citeAsJsonObj.getString("type") + "\"";
+        String describedby = "<" + describedByJsonObj.getString(ds.getProtocol()) + ds.getAuthority() + "/"
+                + ds.getIdentifier() + ">;rel=\"describedby\"" + ";type=\"" + describedByJsonObj.getString("type") + "\"";
         describedby += ",<" + systemConfig.getDataverseSiteUrl() + "/api/datasets/export?exporter=schema.org&persistentId="
                 + ds.getProtocol() + ":" + ds.getAuthority() + "/" + ds.getIdentifier() + ">;rel=\"describedby\"" + ";type=\"application/json+ld\"";
         valueList.add(describedby);
@@ -135,8 +129,8 @@ public String getLinks(){
         valueList.add(type);
 
         // TODO: support only CC0 now, should add flexible license support when flex-terms is ready
-        String license = "";
-        if (workingDatasetVersion.getTermsOfUseAndAccess().getLicense() == TermsOfUseAndAccess.License.CC0){
+        String license;
+        if (workingDatasetVersion.getTermsOfUseAndAccess().getLicense() == TermsOfUseAndAccess.License.CC0) {
             // On the current Dataverse, only None and CC0. In the signposting protocol: cardinality is 1
             license = "<https://creativecommons.org/publicdomain/zero/1.0/>;rel=\"license\"";
             valueList.add(license);
@@ -151,42 +145,52 @@ public String getLinks(){
         return String.join(", ", valueList);
     }
 
-    private JsonArrayBuilder getIdentifiersSchema(List<DatasetAuthor> datasetAuthors){
+    private JsonArrayBuilder getIdentifiersSchema(List<DatasetAuthor> datasetAuthors) {
+        if (datasetAuthors.size() > maxAuthors) return null;
         JsonArrayBuilder authors = Json.createArrayBuilder();
-        for (DatasetAuthor da:datasetAuthors){
-            if (da.getIdValue() != null && !da.getIdValue().isEmpty()) {
-                authors.add(jsonObjectBuilder().add("href", idschemaJsonObj.getString(da.getIdType())  + da.getIdValue()));
+        boolean returnNull = true;
+        for (DatasetAuthor da : datasetAuthors) {
+            if (da.getIdentifierAsUrl() != null && !da.getIdentifierAsUrl().trim().isEmpty()) {
+                authors.add(jsonObjectBuilder().add("href", da.getIdentifierAsUrl()));
+                returnNull = false;
             }
         }
-        return authors;
+        return returnNull ? null : authors;
+    }
+
+    private JsonArrayBuilder getJsonItems(List<FileMetadata> fms) {
+        if (fms.size() > maxItems) return null;
+        JsonArrayBuilder items = Json.createArrayBuilder();
+        for (FileMetadata fm : fms) {
+            DataFile df = fm.getDataFile();
+            items.add(jsonObjectBuilder().add("href", getPublicDownloadUrl(df)).add("type", df.getContentType()));
+        }
+
+        return items;
     }
 
     public JsonArrayBuilder getJsonLinkset() {
         Dataset ds = workingDatasetVersion.getDataset();
-        String landingPage = systemConfig.getDataverseSiteUrl() + "/dataset.xhtml?persistentId=" + ds.getProtocol() + ":" + ds.getAuthority() + "/" + ds.getIdentifier() ;
+        String landingPage = systemConfig.getDataverseSiteUrl() + "/dataset.xhtml?persistentId=" + ds.getProtocol() + ":" + ds.getAuthority() + "/" + ds.getIdentifier();
         JsonArrayBuilder authors = getIdentifiersSchema(workingDatasetVersion.getDatasetAuthors());
 
-        JsonArrayBuilder items = Json.createArrayBuilder();
-
         List<FileMetadata> fms = workingDatasetVersion.getFileMetadatas();
-        for (FileMetadata fm:fms){
-            DataFile df = fm.getDataFile();
-            items.add(jsonObjectBuilder().add("href", getPublicDownloadUrl(df)).add("type",df.getContentType()));
-        }
+        JsonArrayBuilder items = getJsonItems(fms);
 
         String license = "";
-        if (workingDatasetVersion.getTermsOfUseAndAccess().getLicense() == TermsOfUseAndAccess.License.CC0){
+        if (workingDatasetVersion.getTermsOfUseAndAccess().getLicense() == TermsOfUseAndAccess.License.CC0) {
             license = licJsonObj.getString(TermsOfUseAndAccess.License.CC0.name());
         }
 
         JsonArrayBuilder mediaTypes = Json.createArrayBuilder();
         mediaTypes.add(
                 jsonObjectBuilder().add(
                         "href",
-                        citeAsJsonObj.getJsonObject(ds.getProtocol() + ds.getAuthority() + "/" + ds.getIdentifier())
+                        describedByJsonObj.getString(ds.getProtocol()) + ds.getAuthority() + "/"
+                                + ds.getIdentifier()
                 ).add(
                         "type",
-                        citeAsJsonObj.getString("type")
+                        describedByJsonObj.getString("type")
                 )
         );
 
@@ -199,35 +203,45 @@ public JsonArrayBuilder getJsonLinkset() {
                         "application/json+ld"
                 )
         );
-        JsonArrayBuilder linkset = Json.createArrayBuilder();
+        JsonArrayBuilder linksetJsonObj = Json.createArrayBuilder();
         JsonObjectBuilder mandatory = jsonObjectBuilder()
                 .add("anchor", landingPage)
                 .add("cite-as", Json.createArrayBuilder().add(jsonObjectBuilder().add("href", ds.getPersistentURL())))
-                .add("type", Json.createArrayBuilder().add(jsonObjectBuilder().add("href", "https://schema.org/AboutPage")))
-                .add("author", authors)
-                .add("license", jsonObjectBuilder().add("href", license))
-                .add("item", items).add("describedby", mediaTypes);
-        linkset.add(mandatory);
+                .add("type", Json.createArrayBuilder().add(jsonObjectBuilder().add("href", "https://schema.org/AboutPage")));
+
+        if (authors != null) {
+            mandatory.add("author", authors);
+        }
+        if (license != null && !license.trim().isEmpty()) {
+            mandatory.add("license", jsonObjectBuilder().add("href", license));
+        }
+        if (!mediaTypes.toString().trim().isEmpty()) {
+            mandatory.add("describedby", mediaTypes);
+        }
+        if (items != null) {
+            mandatory.add("item", items);
+        }
+        linksetJsonObj.add(mandatory);
 
         if (useDefaultFileType) {
-            for (FileMetadata fm:fms){
+            for (FileMetadata fm : fms) {
                 DataFile df = fm.getDataFile();
                 JsonObjectBuilder itemAnchor = jsonObjectBuilder().add("anchor", getPublicDownloadUrl(df));
                 itemAnchor.add("collection", Json.createArrayBuilder().add(jsonObjectBuilder()
                         .add("href", landingPage)).add(jsonObjectBuilder().add("type", defaultFileTypeValue)));
-                linkset.add(itemAnchor);
+                linksetJsonObj.add(itemAnchor);
             }
         } else {
-            for (FileMetadata fm:fms){
+            for (FileMetadata fm : fms) {
                 DataFile df = fm.getDataFile();
                 JsonObjectBuilder itemAnchor = jsonObjectBuilder().add("anchor", getPublicDownloadUrl(df));
                 itemAnchor.add("collection", Json.createArrayBuilder().add(jsonObjectBuilder()
                         .add("href", landingPage)));
-                linkset.add(itemAnchor);
+                linksetJsonObj.add(itemAnchor);
             }
         }
 
-        return linkset;
+        return linksetJsonObj;
     }
 
 
@@ -240,7 +254,7 @@ private String getPublicDownloadUrl(DataFile dataFile) {
         }
 
         if (storageIO instanceof SwiftAccessIO) {
-            String fileDownloadUrl = null;
+            String fileDownloadUrl;
             SwiftAccessIO<DataFile> swiftIO = (SwiftAccessIO<DataFile>) storageIO;
             try {
                 swiftIO.open();
@@ -249,7 +263,7 @@ private String getPublicDownloadUrl(DataFile dataFile) {
             }
 
             //if its a public install, lets just give users the permanent URL!
-            if (systemConfig.isPublicInstall()){
+            if (systemConfig.isPublicInstall()) {
                 fileDownloadUrl = swiftIO.getRemoteUrl();
             } else {
                 //TODO: if a user has access to this file, they should be given the swift url
diff --git a/src/main/java/propertyFiles/Bundle.properties b/src/main/java/propertyFiles/Bundle.properties
@@ -2570,4 +2570,4 @@ publishDatasetCommand.pidNotReserved=Cannot publish dataset because its persiste
 api.errors.invalidApiToken=Invalid API token.
 
 # Signposting configuration
-signposting.configuration.SignpostingConf = {"indetifier-schema": {"ORCID":"https://orcid.org/", "ISNI":"https://isni.org/isni/", "ScopusID":"https://www.scopus.com/authid/detail.uri?authorId="}, "license": {"CC0":"https://creativecommons.org/licenses/by/4.0/", "MIT": "https://url", "APACHE":"https://url"},"cite-as": {"doi":"https://citation.crosscite.org/format?style=bibtex&doi=", "type":"application/vnd.datacite.datacite+json"},"useDefaultFileType": true,"defaultFileTypeValue": "https://schema.org/Dataset"}
+signposting.configuration.SignpostingConf={"license": {"CC0": "https://creativecommons.org/licenses/cc0/"},"describedby": {"doi": "https://doi.org/","type": "application/vnd.citationstyles.csl+json"},"useDefaultFileType": true,"defaultFileTypeValue": "https://schema.org/Dataset","maxItems": 5,"maxAuthors": 5}
diff --git a/src/main/webapp/dataset.xhtml b/src/main/webapp/dataset.xhtml
@@ -81,7 +81,7 @@
                 <o:importFunctions type="edu.harvard.iq.dataverse.util.MarkupChecker" />
                 <f:metadata>
                     <!-- Add Signposting-->
-                    <f:event type="preRenderView" listener="#{facesContext.externalContext.response.setHeader('link', DatasetPage.getSignpostingLinkHeader())}" />
+                    <f:event type="preRenderView" listener="#{facesContext.externalContext.response.setHeader('Link', DatasetPage.getSignpostingLinkHeader())}" />
                     <!-- End add Signposting-->
                     <f:viewParam name="id" value="#{DatasetPage.id}"/>
                     <o:viewParam name="ownerId" value="#{DatasetPage.ownerId}"/>
