More ECOD format changes

sbliven · sbliven · commit ae8ca434f54e · 2015-03-16T10:17:28.000+01:00
All files now have the proper 14-field format
diff --git a/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/domain/EcodInstallationTest.java b/biojava-integrationtest/src/test/java/org/biojava/nbio/structure/test/domain/EcodInstallationTest.java
@@ -35,12 +35,15 @@
 import java.util.concurrent.Future;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.TimeoutException;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import org.biojava.nbio.core.util.ConcurrencyTools;
 import org.biojava.nbio.structure.ecod.EcodDatabase;
 import org.biojava.nbio.structure.ecod.EcodDomain;
 import org.biojava.nbio.structure.ecod.EcodFactory;
 import org.biojava.nbio.structure.ecod.EcodInstallation;
+import org.junit.Ignore;
 import org.junit.Rule;
 import org.junit.Test;
 import org.junit.rules.TemporaryFolder;
@@ -123,7 +126,7 @@ public void testParsing() throws IOException {
 		domain = ecod.getDomainsById(ecodId);
 		expected = new EcodDomain(
 				//				Long uid, String domainId, Boolean manual,
-				20669l, "e1lyw.1", null,
+				20669l, "e1lyw.1", false,
 				//				Integer xGroup, Integer hGroup, Integer tGroup, Integer fGroup, String pdbId,
 				1,1,1,2,"1lyw",
 				//				String chainId, String range, String architectureName,
@@ -224,17 +227,33 @@ public void testVersion() throws IOException {
 		assertNotNull(version);
 		assertNotEquals("latest", version);
 	}
-	
+
+	/**
+	 * Parses all known versions. Only fails due to exceptions, so manually check for warnings.
+	 * Hierarchical field warnings are expected for versions prior to develop68.
+	 * @throws IOException
+	 */
+	//@Ignore
 	@Test
 	public void testAllVersions() throws IOException {
+		// Fetch latest version
+		EcodDatabase latest = EcodFactory.getEcodDatabase("latest");
+		String latestVersionStr = latest.getVersion();
+		int latestVersion = 0;
+		Matcher match = Pattern.compile("develop([0-9]+)",Pattern.CASE_INSENSITIVE).matcher(latestVersionStr);
+		if(match.matches())
+			latestVersion = Integer.parseInt(match.group(1));
+
+		// List all versions
 		int firstVersion = 45;
-		int lastVersion = 78;
+		int lastVersion = Math.max(78,latestVersion);
 		List<String> versions = new ArrayList<String>();
-		
+		versions.add("latest");
 		for(int version = firstVersion; version<= lastVersion;version++) {
 			versions.add("develop"+version);
 		}
-		versions.add("latest");
+		
+		// Parse all versions
 		for(String version : versions) {
 			EcodInstallation ecod = (EcodInstallation)EcodFactory.getEcodDatabase(version);
 			ecod.getAllDomains();
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/ecod/EcodInstallation.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/ecod/EcodInstallation.java
@@ -506,6 +506,7 @@ private void parse(BufferedReader in) throws IOException {
 				// prevent too many warnings; negative numbers print all warnings
 				int warnIsDomainAssembly = 1;
 				int warnHierarchicalFormat = 5;
+				int warnNumberOfFields = 10;
 
 				String line = in.readLine();
 				int lineNum = 1;
@@ -604,7 +605,13 @@ private void parse(BufferedReader in) throws IOException {
 									logger.warn("Error in ECOD parsing at line "+lineNum,e);
 								}
 							} else {
-								logger.warn("Unexpected number of fields in line {}",lineNum);
+								if(warnNumberOfFields > 1) {
+									logger.warn("Unexpected number of fields in line {}.",lineNum);
+									warnNumberOfFields--;
+								} else if(warnNumberOfFields == 0) {
+									logger.warn("Unexpected number of fields in line {}. Not printing future similar warnings",lineNum);
+									warnIsDomainAssembly--;
+								}
 							}
 						}
 					}

Original file line number	Diff line number	Diff line change
`@@ -506,6 +506,7 @@ private void parse(BufferedReader in) throws IOException {`
`506`	`506`	`// prevent too many warnings; negative numbers print all warnings`
`507`	`507`	`int warnIsDomainAssembly = 1;`
`508`	`508`	`int warnHierarchicalFormat = 5;`
	`509`	`+ int warnNumberOfFields = 10;`
`509`	`510`
`510`	`511`	`String line = in.readLine();`
`511`	`512`	`int lineNum = 1;`
`@@ -604,7 +605,13 @@ private void parse(BufferedReader in) throws IOException {`
`604`	`605`	`logger.warn("Error in ECOD parsing at line "+lineNum,e);`
`605`	`606`	`}`
`606`	`607`	`} else {`
`607`		`- logger.warn("Unexpected number of fields in line {}",lineNum);`
	`608`	`+ if(warnNumberOfFields > 1) {`
	`609`	`+ logger.warn("Unexpected number of fields in line {}.",lineNum);`
	`610`	`+ warnNumberOfFields--;`
	`611`	`+ } else if(warnNumberOfFields == 0) {`
	`612`	`+ logger.warn("Unexpected number of fields in line {}. Not printing future similar warnings",lineNum);`
	`613`	`+ warnIsDomainAssembly--;`
	`614`	`+ }`
`608`	`615`	`}`
`609`	`616`	`}`
`610`	`617`	`}`