Handle surrogate pair characters substring (#955)

Author: XingY

flow/src/org/labkey/flow/ScriptParser.java

@@ -21,6 +21,7 @@
 import org.apache.xmlbeans.XmlOptions;
 import org.fhcrc.cpas.flow.script.xml.ScriptDef;
 import org.fhcrc.cpas.flow.script.xml.ScriptDocument;
+import org.labkey.api.util.StringUtilsLabKey;
 import org.xml.sax.SAXParseException;
 
 import java.io.StringReader;
@@ -94,7 +95,7 @@ public void parse(String script)
                     message = StringUtils.replace(message, "@" + ScriptDocument.type.getContentModel().getName().getNamespaceURI(), "");
                     String location = xmlError.getCursorLocation().xmlText();
                     if (location.length() > 100)
-                        location = location.substring(0, 100);
+                        location = StringUtilsLabKey.leftSurrogatePairFriendly(location, 100);
                     addError(new Error("Schema Validation Error: " + message + "\nLocation of invalid XML: " + location));
                 }
             }

protein/api-src/org/labkey/api/protein/ProteinManager.java

@@ -20,6 +20,7 @@
 import org.labkey.api.util.HashHelpers;
 import org.labkey.api.util.HtmlString;
 import org.labkey.api.util.LinkBuilder;
+import org.labkey.api.util.StringUtilsLabKey;
 import org.labkey.api.view.NotFoundException;
 
 import java.io.ByteArrayOutputStream;
@@ -100,7 +101,7 @@ private static SimpleProtein ensureProteinInDatabase(String sequence, Organism o
             map.put("Mass", PeptideHelpers.computeMass(sequenceBytes, 0, sequenceBytes.length, PeptideHelpers.AMINO_ACID_AVERAGE_MASSES));
             map.put("OrgId", organism.getOrgId());
             map.put("Hash", hashSequence(sequence));
-            map.put("Description", description == null ? null : (description.length() > 200 ? description.substring(0, 196) + "..." : description));
+            map.put("Description", description == null ? null : (description.length() > 200 ? StringUtilsLabKey.leftSurrogatePairFriendly(description, 196) + "..." : description));
             map.put("BestName", name);
             map.put("Length", sequence.length());
             map.put("InsertDate", new Date());

protein/api-src/org/labkey/api/protein/ProteinPlus.java

@@ -18,6 +18,7 @@
 
 import org.labkey.api.protein.fasta.FastaProtein;
 import org.labkey.api.util.HashHelpers;
+import org.labkey.api.util.StringUtilsLabKey;
 
 public class ProteinPlus
 {
@@ -111,7 +112,7 @@ public String getBestName()
         {
             result = getProtein().getHeader();
         }
-        if (result.length() > 500) result = result.substring(0, 499);
+        if (result.length() > 500) result = StringUtilsLabKey.leftSurrogatePairFriendly(result, 499);
         return result;
     }
 }

protein/api-src/org/labkey/api/protein/fasta/FastaDbLoader.java

@@ -37,6 +37,7 @@
 import org.labkey.api.protein.organism.OrganismGuessStrategy;
 import org.labkey.api.util.HashHelpers;
 import org.labkey.api.util.NetworkDrive;
+import org.labkey.api.util.StringUtilsLabKey;
 import org.labkey.api.view.ViewBackgroundInfo;
 
 import java.io.File;
@@ -259,7 +260,7 @@ protected void preProcessSequences(List<ProteinPlus> mouthful, Connection c, Log
             }
             else
             {
-                if (desc.length() >= 200) desc = desc.substring(0, 195) + "...";
+                if (desc.length() >= 200) desc = StringUtilsLabKey.leftSurrogatePairFriendly(desc, 195) + "...";
                 fdbu._addSeqStmt.setString(3, desc);
             }
             fdbu._addSeqStmt.setDouble(4, curSeq.getProtein().getMass());

protein/api-src/org/labkey/api/protein/fasta/IdPattern.java

@@ -27,6 +27,7 @@
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 import java.util.regex.PatternSyntaxException;
+import org.labkey.api.util.StringUtilsLabKey;
 
 /**
  * this class implements a regular expression-based recognition of identifiers parsed from the fasta files.
@@ -178,7 +179,7 @@ public static Map<String, Set<String>> createIdMap(String key, String value)
             {
                 v = v.trim();
                 if (v.length() > 50)
-                    v = v.substring(0, 50);
+                    v = StringUtilsLabKey.leftSurrogatePairFriendly(v, 50);
                 if (!v.isEmpty())
                     vals.add(v);
             }

protein/api-src/org/labkey/api/protein/uniprot/uniprot.java

@@ -605,7 +605,7 @@ public int insertSequences(ParseContext context, Connection conn) throws SQLExce
             else
             {
                 String tmp = curSeq.getDescription();
-                if (tmp.length() >= 200) tmp = tmp.substring(0, 190) + "...";
+                if (tmp.length() >= 200) tmp = StringUtilsLabKey.leftSurrogatePairFriendly(tmp, 190) + "...";
                 _addSeq.setString(3, tmp);
             }
             if (curSeq.getSourceChangeDate() == null)
@@ -657,7 +657,7 @@ public int insertSequences(ParseContext context, Connection conn) throws SQLExce
             else
             {
                 String tmp = curSeq.getBestName();
-                if (tmp.length() >= 50) tmp = tmp.substring(0, 45) + "...";
+                if (tmp.length() >= 50) tmp = StringUtilsLabKey.leftSurrogatePairFriendly(tmp, 45) + "...";
                 _addSeq.setString(11, tmp);
             }
             if (curSeq.getBestGeneName() == null)
@@ -667,7 +667,7 @@ public int insertSequences(ParseContext context, Connection conn) throws SQLExce
             else
             {
                 String tmp = curSeq.getBestGeneName();
-                if (tmp.length() >= 50) tmp = tmp.substring(0, 45) + "...";
+                if (tmp.length() >= 50) tmp = StringUtilsLabKey.leftSurrogatePairFriendly(tmp, 45) + "...";
                 _addSeq.setString(12, tmp);
             }
             // Timestamp at index 13 is set once for the whole prepared statement
@@ -709,7 +709,7 @@ public int insertIdentifiers(ParseContext context, Connection conn) throws SQLEx
         {
             transactionCount++;
             String curIdentVal = curIdent.getIdentifier();
-            if (curIdentVal.length() > 50) curIdentVal = curIdentVal.substring(0, 45) + "...";
+            if (curIdentVal.length() > 50) curIdentVal = StringUtilsLabKey.leftSurrogatePairFriendly(curIdentVal, 45) + "...";
             _addIdent.setString(1, curIdentVal);
             _addIdent.setString(2, curIdent.getIdentType());
             UniprotSequence curSeq = curIdent.getSequence();