Override getJson() and getCitation() in MockNcbiPublicationSearchService

Author: vagisha

panoramapublic/src/org/labkey/panoramapublic/ncbi/MockNcbiPublicationSearchService.java

@@ -1,25 +1,25 @@
 package org.labkey.panoramapublic.ncbi;
 
-import org.apache.logging.log4j.Logger;
-import org.jetbrains.annotations.NotNull;
 import org.jetbrains.annotations.Nullable;
-import org.labkey.api.util.Pair;
-import org.labkey.panoramapublic.model.ExperimentAnnotations;
+import org.json.JSONArray;
+import org.json.JSONObject;
 import org.labkey.panoramapublic.ncbi.NcbiConstants.DB;
 
-import java.text.ParseException;
-import java.text.SimpleDateFormat;
-import java.util.Collections;
-import java.util.Date;
-import java.util.List;
+import java.io.IOException;
 
 /**
  * Mock implementation of {@link NcbiPublicationSearchService} that returns canned data for PMID 23689285
  * (Abbatiello et al., Mol Cell Proteomics 2013). Used by Selenium tests when NCBI is not reachable.
+ *
+ * Extends {@link NcbiPublicationSearchServiceImpl} and only overrides the two methods that make HTTP calls
+ * to NCBI: {@link #getJson(String)} (used by ESearch/ESummary) and {@link #getCitation(String, DB)}
+ * (used by the Citation Exporter API). All search logic, filtering, author/title verification, and
+ * priority filtering run through the real implementation code.
  */
-public class MockNcbiPublicationSearchService implements NcbiPublicationSearchService
+public class MockNcbiPublicationSearchService extends NcbiPublicationSearchServiceImpl
 {
     private static final String PMID = "23689285";
+    private static final String PMC_ID = "3769335";
 
     private static final String CITATION = "Abbatiello SE, Mani DR, Schilling B, Maclean B, Zimmerman LJ, " +
             "Feng X, Cusack MP, Sedransk N, Hall SC, Addona T, Allen S, Dodder NG, Ghosh M, Held JM, Hedrick V, " +
@@ -32,19 +32,37 @@ public class MockNcbiPublicationSearchService implements NcbiPublicationSearchSe
             "Mol Cell Proteomics. 2013 Sep;12(9):2623-39. doi: 10.1074/mcp.M112.027078. Epub 2013 May 20. " +
             "PMID: 23689285; PMCID: PMC3769335.";
 
-    private static final Date PUBLISHED_DATE;
-    static
+    private static final String ARTICLE_TITLE = "Design, implementation and multisite evaluation of a system suitability " +
+            "protocol for the quantitative assessment of instrument performance in liquid chromatography-multiple " +
+            "reaction monitoring-MS (LC-MRM-MS).";
+
+    /**
+     * Returns canned JSON for NCBI ESearch and ESummary API requests.
+     * <ul>
+     *   <li>ESearch for PMC with a query containing "PXD010535" returns PMC ID 3769335</li>
+     *   <li>ESummary for PMC ID 3769335 returns article metadata (authors, title, pubdate, PMID)</li>
+     *   <li>All other requests return empty results</li>
+     * </ul>
+     * This allows the real search logic in {@link NcbiPublicationSearchServiceImpl} to run against mock data.
+     */
+    @Override
+    protected JSONObject getJson(String url) throws IOException
     {
-        try
+        if (url.contains("esearch.fcgi"))
         {
-            PUBLISHED_DATE = new SimpleDateFormat("yyyy MMM").parse("2013 Sep");
+            return handleESearch(url);
         }
-        catch (ParseException e)
+        else if (url.contains("esummary.fcgi"))
         {
-            throw new RuntimeException(e);
+            return handleESummary(url);
         }
+        throw new IOException("MockNcbiPublicationSearchService: unexpected URL: " + url);
     }
 
+    /**
+     * Returns the canned NLM citation for PMID 23689285. Returns null for any other publication ID.
+     * Overrides the real implementation which makes an HTTP call to the NCBI Citation Exporter API.
+     */
     @Override
     public @Nullable String getCitation(String publicationId, DB database)
     {
@@ -55,42 +73,66 @@ public class MockNcbiPublicationSearchService implements NcbiPublicationSearchSe
         return null;
     }
 
-    @Override
-    public @Nullable Pair<String, String> getPubMedLinkAndCitation(String pubmedId)
+    /**
+     * Handle mock ESearch requests. Returns PMC ID 3769335 when query contains "PXD010535",
+     * empty results otherwise.
+     */
+    private JSONObject handleESearch(String url)
     {
-        if (PMID.equals(pubmedId))
+        JSONObject result = new JSONObject();
+        JSONObject esearchResult = new JSONObject();
+
+        if (url.contains("PXD010535"))
         {
-            return new Pair<>(NcbiConstants.getPubmedLink(pubmedId), CITATION);
+            esearchResult.put("idlist", new JSONArray().put(PMC_ID));
+        }
+        else
+        {
+            esearchResult.put("idlist", new JSONArray());
         }
-        return null;
-    }
 
-    @Override
-    public @Nullable PublicationMatch searchForPublication(@NotNull ExperimentAnnotations expAnnotations, @Nullable Logger logger)
-    {
-        List<PublicationMatch> matches = searchForPublication(expAnnotations, 1, logger, true);
-        return matches.isEmpty() ? null : matches.get(0);
+        result.put("esearchresult", esearchResult);
+        return result;
     }
 
-    @Override
-    public List<PublicationMatch> searchForPublication(@NotNull ExperimentAnnotations expAnnotations, int maxResults, @Nullable Logger logger, boolean getCitations)
+    /**
+     * Handle mock ESummary requests. Returns article metadata for PMC ID 3769335.
+     */
+    private JSONObject handleESummary(String url)
     {
-        PublicationMatch match = new PublicationMatch(
-                PMID,
-                DB.PubMed,
-                false,   // foundByPxId (simulates finding via PXD010535)
-                false,  // foundByUrl
-                false,  // foundByDoi
-                true,   // authorMatch
-                true,   // titleMatch
-                PUBLISHED_DATE
-        );
-
-        if (getCitations)
+        JSONObject response = new JSONObject();
+        JSONObject result = new JSONObject();
+
+        if (url.contains(PMC_ID))
         {
-            match.setCitation(CITATION);
+            JSONObject articleData = new JSONObject();
+            articleData.put("title", ARTICLE_TITLE);
+            articleData.put("sorttitle", ARTICLE_TITLE.toLowerCase());
+            articleData.put("source", "Mol Cell Proteomics");
+            articleData.put("fulljournalname", "Molecular & cellular proteomics : MCP");
+            articleData.put("pubdate", "2013 Sep");
+            articleData.put("epubdate", "2013 May 20");
+
+            // Authors — include first 3 plus a few more for realism
+            JSONArray authors = new JSONArray();
+            authors.put(new JSONObject().put("name", "Abbatiello SE"));
+            authors.put(new JSONObject().put("name", "Mani DR"));
+            authors.put(new JSONObject().put("name", "Schilling B"));
+            authors.put(new JSONObject().put("name", "Maclean B"));
+            authors.put(new JSONObject().put("name", "Zimmerman LJ"));
+            authors.put(new JSONObject().put("name", "Carr SA"));
+            articleData.put("authors", authors);
+
+            // Article IDs — PMC ID and PubMed ID
+            JSONArray articleIds = new JSONArray();
+            articleIds.put(new JSONObject().put("idtype", "pmcid").put("value", "PMC" + PMC_ID));
+            articleIds.put(new JSONObject().put("idtype", "pmid").put("value", PMID));
+            articleData.put("articleids", articleIds);
+
+            result.put(PMC_ID, articleData);
         }
 
-        return Collections.singletonList(match);
+        response.put("result", result);
+        return response;
     }
 }

panoramapublic/src/org/labkey/panoramapublic/ncbi/NcbiPublicationSearchServiceImpl.java

@@ -264,7 +264,7 @@ public List<PublicationMatch> searchForPublication(@NotNull ExperimentAnnotation
     /**
      * Search PubMed Central
      */
-    private static @NotNull List<PublicationMatch> searchPmc(@NotNull ExperimentAnnotations expAnnotations, Logger log)
+    private @NotNull List<PublicationMatch> searchPmc(@NotNull ExperimentAnnotations expAnnotations, Logger log)
     {
         // Track PMC IDs found by each strategy
         Map<String, List<String>> pmcIdsByStrategy = new HashMap<>();
@@ -338,23 +338,23 @@ public List<PublicationMatch> searchForPublication(@NotNull ExperimentAnnotation
     /**
      * Search PubMed Central with the given query
      */
-    private static List<String> searchPmc(String query, Logger log)
+    private List<String> searchPmc(String query, Logger log)
     {
         return executeSearch(query, "pmc", log);
     }
 
     /**
      * Search PubMed with the given query
      */
-    private static List<String> searchPubMed(String query, Logger log)
+    private List<String> searchPubMed(String query, Logger log)
     {
         return executeSearch(query, "pubmed", log);
     }
 
     /**
      * Execute search using NCBI ESearch API
      */
-    private static List<String> executeSearch(String query, String database, Logger log)
+    private List<String> executeSearch(String query, String database, Logger log)
     {
         String encodedQuery = URLEncoder.encode(query, StandardCharsets.UTF_8);
         String url = ESEARCH_URL +
@@ -389,7 +389,7 @@ private static List<String> executeSearch(String query, String database, Logger
      * @throws IOException if the request fails or the server returns a non-2xx response
      * @throws JSONException if the response body is not valid JSON
      */
-    private static JSONObject getJson(String url) throws IOException
+    protected JSONObject getJson(String url) throws IOException
     {
         ConnectionConfig connectionConfig = ConnectionConfig.custom()
             .setConnectTimeout(Timeout.ofMilliseconds(TIMEOUT_MS))
@@ -424,23 +424,23 @@ private static JSONObject getJson(String url) throws IOException
     /**
      * Fetch metadata for PMC articles using ESummary API
      */
-    private static Map<String, JSONObject> fetchPmcMetadata(Collection<String> pmcIds, Logger log)
+    private Map<String, JSONObject> fetchPmcMetadata(Collection<String> pmcIds, Logger log)
     {
         return fetchMetadata(pmcIds, "pmc", log);
     }
 
     /**
      * Fetch metadata for PubMed articles using ESummary API
      */
-    private static Map<String, JSONObject> fetchPubMedMetadata(Collection<String> pmids, Logger log)
+    private Map<String, JSONObject> fetchPubMedMetadata(Collection<String> pmids, Logger log)
     {
         return fetchMetadata(pmids, "pubmed", log);
     }
 
     /**
      * Fetch metadata using NCBI ESummary API (batch request)
      */
-    private static Map<String, JSONObject> fetchMetadata(Collection<String> ids, String database, Logger log)
+    private Map<String, JSONObject> fetchMetadata(Collection<String> ids, String database, Logger log)
     {
         if (ids.isEmpty()) return Collections.emptyMap();
 
@@ -479,7 +479,7 @@ private static Map<String, JSONObject> fetchMetadata(Collection<String> ids, Str
     /**
      * Fetch and verify PMC articles (filter preprints, check author/title matches)
      */
-    private static List<PublicationMatch> fetchAndVerifyPmcArticles(
+    private List<PublicationMatch> fetchAndVerifyPmcArticles(
         Set<String> pmcIds,
         Map<String, List<String>> idToStrategies,
         ExperimentAnnotations expAnnotations,
@@ -612,7 +612,7 @@ private static int countDataIdMatches(PublicationMatch a)
     /**
      * Fall back to PubMed search if PMC finds nothing
      */
-    private static List<PublicationMatch> searchPubMed(ExperimentAnnotations expAnnotations, Logger log)
+    private List<PublicationMatch> searchPubMed(ExperimentAnnotations expAnnotations, Logger log)
     {
         String firstName = expAnnotations.getSubmitterUser() != null
             ? expAnnotations.getSubmitterUser().getFirstName() : null;

panoramapublic/test/src/org/labkey/test/tests/panoramapublic/PublicationSearchTest.java

@@ -236,6 +236,8 @@ protected void doCleanup(boolean afterTest) throws TestTimeoutException
             restoreNcbiService();
         }
 
+        // TODO: Reset the reminder settings to not search for publications.
+
         _userHelper.deleteUsers(false, SUBMITTER_USER, ADMIN_USER);
         super.doCleanup(afterTest);
     }