Re-download empty structure or chemical component structures

sbliven · sbliven · commit 6ea9ae762946 · 2018-06-08T12:27:41.000+02:00
Files less than 40 bytes are deleted to allow for gzip headers. This addresses #703
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/io/LocalPDBDirectory.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/io/LocalPDBDirectory.java
@@ -127,6 +127,9 @@ public static enum FetchBehavior {
 
 	protected static final String lineSplit = System.getProperty("file.separator");
 
+	/** Minimum size for a valid file, in bytes */
+	public static final long MIN_PDB_FILE_SIZE = 40;
+
 	private File path;
 	private List<String> extensions;
 
@@ -687,6 +690,14 @@ public File getLocalFile(String pdbId) {
 				for(String ex : getExtensions() ){
 					File f = new File(searchdir,prefix + pdbId.toLowerCase() + ex) ;
 					if ( f.exists()) {
+						// delete files that are too short to have contents
+						if( f.length() < MIN_PDB_FILE_SIZE ) {
+							boolean success = f.delete();
+							if( ! success) {
+								return null;
+							}
+							assert(!f.exists());
+						}
 						return f;
 					}
 				}
diff --git a/biojava-structure/src/main/java/org/biojava/nbio/structure/io/mmcif/DownloadChemCompProvider.java b/biojava-structure/src/main/java/org/biojava/nbio/structure/io/mmcif/DownloadChemCompProvider.java
@@ -42,6 +42,7 @@
 import org.biojava.nbio.core.util.InputStreamProvider;
 import org.biojava.nbio.structure.align.util.HTTPConnectionTools;
 import org.biojava.nbio.structure.align.util.UserConfiguration;
+import org.biojava.nbio.structure.io.LocalPDBDirectory;
 import org.biojava.nbio.structure.io.mmcif.model.ChemComp;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -344,6 +345,12 @@ private static  boolean fileExists(String recordName){
 
 		File f = new File(fileName);
 
+		// delete files that are too short to have contents
+		if( f.length() < LocalPDBDirectory.MIN_PDB_FILE_SIZE ) {
+			f.delete();
+			return false;
+		}
+
 		return f.exists();
 
 	}
