Make ChatGPT specify publication links when mentioning publications.

server/catgenome/profiles/dev/catgenome.properties

@@ -183,7 +183,7 @@ llm.openai.api.key=${NGB_OPEN_AI_KEY:}
 
 llm.openai.chatgpt35.prompt.template=Generate brief gene summary as drug target based on the following articles:
 llm.openai.chatgpt35.first.message.prefix=Consider the following article abstracts: 
-llm.openai.chatgpt35.last.message.prefix=Based on previously provided articles
+llm.openai.chatgpt35.last.message.prefix=Based on previously provided articles. Share links when mentioning articles.
 
 llm.openai.chatgpt40.prompt.template=Generate brief gene summary as drug target based on the following articles:
 llm.openai.chatgpt40.first.message.prefix=Consider the following article abstracts: 

server/catgenome/src/main/java/com/epam/catgenome/controller/target/LaunchIdentificationController.java

@@ -286,8 +286,9 @@ public Result<SearchResult<NCBISummaryVO>> getPublications(@RequestBody final Pu
     @ApiResponses(
             value = {@ApiResponse(code = HTTP_STATUS_OK, message = API_STATUS_DESCRIPTION)
             })
-    public Result<String> getAbstracts(@RequestBody final PublicationSearchRequest request) {
-        return Result.success(launchIdentificationSecurityService.getArticleAbstracts(request));
+    public Result<String> getAbstracts(@RequestBody final PublicationSearchRequest request,
+                                       @RequestParam(required = false, defaultValue = "false") boolean getLinks) {
+        return Result.success(launchIdentificationSecurityService.getArticleAbstracts(request, getLinks));
     }
 
     @GetMapping(value = "/target/sequences")

server/catgenome/src/main/java/com/epam/catgenome/manager/externaldb/PubMedService.java

@@ -40,6 +40,7 @@
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.stereotype.Service;
 import org.w3c.dom.Document;
+import org.w3c.dom.Node;
 import org.w3c.dom.NodeList;
 import org.xml.sax.InputSource;
 import org.xml.sax.SAXException;
@@ -70,6 +71,7 @@ public class PubMedService {
     private Integer articlesMaxNumber;
 
     private static final Integer BATCH_SIZE = 500;
+    private static final String ARTICLE_XPATH = "MedlineCitation//Article";
 
     private final NCBIGeneIdsManager ncbiGeneIdsManager;
     private final NCBIGeneManager ncbiGeneManager;
@@ -115,21 +117,21 @@ public int getPublicationsCount(final List<String> entrezIds) {
     }
 
     @SneakyThrows
-    public String getArticleAbstracts(final List<String> pmcIds) {
+    public String getArticleAbstracts(final List<String> pmcIds, boolean getLinks) {
         final String xml = ncbiDataManager.fetchXmlById(NCBIDatabase.PUBMED.name(),
                 String.join(",", pmcIds), "Abstract");
-        return parseAbstract(xml);
+        return getLinks ? parseAbstractWithLink(xml) : parseAbstract(xml);
     }
 
-    public String getArticleAbstracts(final PublicationSearchRequest request) {
+    public String getArticleAbstracts(final PublicationSearchRequest request, boolean getLinks) {
         request.setPage(1);
         request.setPageSize(articlesMaxNumber);
         final SearchResult<NCBISummaryVO> result = fetchPubMedArticles(request);
         final List<String> ids = ListUtils.emptyIfNull(result.getItems())
                 .stream()
                 .map(NCBISummaryVO::getUid)
                 .collect(Collectors.toList());
-        return getArticleAbstracts(ids);
+        return getArticleAbstracts(ids, getLinks);
     }
 
     private String parseAbstract(final String xml) throws ParserConfigurationException, IOException,
@@ -150,4 +152,32 @@ private String parseAbstract(final String xml) throws ParserConfigurationExcepti
         }
         return result.toString();
     }
+
+    private String parseAbstractWithLink(final String xml) throws ParserConfigurationException, IOException,
+            SAXException, XPathExpressionException {
+        final XPath xPath = XPathFactory.newInstance().newXPath();
+        final DocumentBuilderFactory builderFactory = DocumentBuilderFactory.newInstance();
+        final DocumentBuilder builder = builderFactory.newDocumentBuilder();
+        final InputSource is = new InputSource(new StringReader(xml));
+        final Document document = builder.parse(is);
+        final XPathExpression exp = xPath.compile("//PubmedArticle");
+        final NodeList articles = (NodeList)exp.evaluate(document, XPathConstants.NODESET);
+        final StringBuilder result = new StringBuilder();
+        for (int i = 0; i < articles.getLength(); i++) {
+            if (result.length() != 0) {
+                result.append('\n');
+            }
+            Node item = articles.item(i).cloneNode(true);
+            String title = xPath.compile(ARTICLE_XPATH + "//ArticleTitle").evaluate(item);
+            result.append(String.format("Article: \"%s\" ", title));
+            String id = xPath.compile("MedlineCitation//PMID").evaluate(item);
+            result.append(String.format("Link: https://pubmed.ncbi.nlm.nih.gov/%s. ", id));
+            XPathExpression abstractExp = xPath.compile(ARTICLE_XPATH + "//Abstract//AbstractText");
+            NodeList abstracts = (NodeList)abstractExp.evaluate(item, XPathConstants.NODESET);
+            for (int j = 0; j < abstracts.getLength(); j++) {
+                result.append(abstracts.item(j).getTextContent());
+            }
+        }
+        return result.toString();
+    }
 }

server/catgenome/src/main/java/com/epam/catgenome/manager/llm/LLMService.java

@@ -57,7 +57,7 @@ public String getArticleSummary(final List<String> pubMedIDs,
                                     final int maxSize,
                                     final double temperature) {
         return getHandler(provider)
-                .getSummary(pubmedService.getArticleAbstracts(pubMedIDs), temperature);
+                .getSummary(pubmedService.getArticleAbstracts(pubMedIDs, false), temperature);
     }
 
     public String getGeneArticleSummary(final List<String> geneIds,
@@ -66,7 +66,7 @@ public String getGeneArticleSummary(final List<String> geneIds,
                                         final double temperature) {
         final PublicationSearchRequest request = new PublicationSearchRequest();
         request.setGeneIds(geneIds);
-        return getHandler(provider).getSummary(pubmedService.getArticleAbstracts(request), temperature);
+        return getHandler(provider).getSummary(pubmedService.getArticleAbstracts(request, false), temperature);
     }
 
     public String getChatResponse(final List<LLMMessage> messages,

server/catgenome/src/main/java/com/epam/catgenome/manager/target/LaunchIdentificationManager.java

@@ -223,8 +223,8 @@ public SearchResult<NCBISummaryVO> getPublications(final PublicationSearchReques
         return pubMedService.fetchPubMedArticles(request);
     }
 
-    public String getArticlesAbstracts(final PublicationSearchRequest request) {
-        return pubMedService.getArticleAbstracts(request);
+    public String getArticlesAbstracts(final PublicationSearchRequest request, boolean getLinks) {
+        return pubMedService.getArticleAbstracts(request, getLinks);
     }
 
     public List<GeneSequences> getGeneSequences(final List<String> geneIds) throws ParseException, IOException {

server/catgenome/src/main/java/com/epam/catgenome/manager/target/LaunchIdentificationSecurityService.java

@@ -148,8 +148,8 @@ public SearchResult<NCBISummaryVO> getPublications(final PublicationSearchReques
     }
 
     @PreAuthorize(ROLE_USER)
-    public String getArticleAbstracts(final PublicationSearchRequest request) {
-        return manager.getArticlesAbstracts(request);
+    public String getArticleAbstracts(final PublicationSearchRequest request, boolean getLinks) {
+        return manager.getArticlesAbstracts(request, getLinks);
     }
 
     @PreAuthorize(ROLE_USER)