From: Ben Soares <b.soares@dundee.ac.uk>
Date: Fri, 17 Sep 2021 17:11:05 +0000 (+0100)
Subject: JAL-1260 rudimentary parsing of LOCUS line to at least get accession id
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;ds=sidebyside;h=refs%2Fheads%2Ffeature%2FJAL-1260_genbank_embl_flatfile_import_reloaded;p=jalview.git

JAL-1260 rudimentary parsing of LOCUS line to at least get accession id
---

diff --git a/src/jalview/io/GenBankFile.java b/src/jalview/io/GenBankFile.java
index 7988764..ba7b4b4 100644
--- a/src/jalview/io/GenBankFile.java
+++ b/src/jalview/io/GenBankFile.java
@@ -2,8 +2,6 @@ package jalview.io;
 
 import java.io.IOException;
 
-import jalview.bin.Cache;
-
 /**
  * A class that provides selective parsing of the GenBank flatfile format.
  * <p>
@@ -47,7 +45,11 @@ public class GenBankFile extends FlatFile
     String line = nextLine();
     while (line != null)
     {
-      if (line.startsWith(DEFINITION))
+      if (line.startsWith("LOCUS"))
+      {
+        line = parseLocus(line);
+      }
+      else if (line.startsWith(DEFINITION))
       {
         line = parseDefinition(line);
       }
@@ -89,45 +91,25 @@ public class GenBankFile extends FlatFile
    */
   String parseLocus(String line) throws IOException
   {
-    String[] tokens = line.substring(2).split(";");
+    String[] tokens = line.split(WHITESPACE);
 
     /*
-     * first is primary accession
+     * first should be "LOCUS"
      */
-    String token = tokens[0].trim();
-    if (!token.isEmpty())
+    if (tokens.length < 2 || !"LOCUS".equals(tokens[0]))
     {
-      this.accession = token;
+      return nextLine();
     }
-
     /*
-     * second token is 'SV versionNo'
+     * second is primary accession
      */
-    if (tokens.length > 1)
+    String token = tokens[1].trim();
+    if (!token.isEmpty())
     {
-      token = tokens[1].trim();
-      if (token.startsWith("SV"))
-      {
-        String[] bits = token.trim().split(WHITESPACE);
-        this.version = bits[bits.length - 1];
-      }
+      this.accession = token;
     }
 
-    /*
-     * seventh token is 'length BP'
-     */
-    if (tokens.length > 6)
-    {
-      token = tokens[6].trim();
-      String[] bits = token.trim().split(WHITESPACE);
-      try
-      {
-        this.length = Integer.valueOf(bits[0]);
-      } catch (NumberFormatException e)
-      {
-        Cache.log.error("bad length read in flatfile, line: " + line);
-      }
-    }
+    // not going to guess the rest just yet, but third is length with unit (bp)
 
     return nextLine();
   }