Ensure we never add pre-existing input as intermediate file

Author: bbimber

SequenceAnalysis/src/org/labkey/sequenceanalysis/SequenceRemoteIntegrationTests.java

@@ -57,7 +57,17 @@ private File setupConfigDir(File outDir) throws IOException
 
         baseDir.mkdirs();
 
+        if (_sampleData == null)
+        {
+            throw new IOException("_sampleData was null");
+        }
+
         File source = new File(_sampleData, "remotePipeline");
+        if (!source.exists())
+        {
+            throw new IOException("Unable to find file: " + source.getPath());
+        }
+
         FileUtils.copyFile(new File(source, "sequenceanalysisConfig.xml"), new File(baseDir, "sequenceanalysisConfig.xml"));
 
         try (PrintWriter writer = PrintWriters.getPrintWriter(new File(baseDir, "pipelineConfig.xml")); BufferedReader reader = Readers.getReader(new File(source, "pipelineConfig.xml")))
@@ -179,6 +189,7 @@ protected void executeJobRemote(File workDir, @Nullable File jobJson) throws IOE
         args.add("org.labkey.bootstrap.ClusterBootstrap");
 
         File webappDir = new File(ModuleLoader.getServletContext().getRealPath(""));
+
         args.add("-modulesdir=" + new File(webappDir.getParentFile(), "modules").getPath());
         args.add("-webappdir=" + webappDir.getPath());
 

SequenceAnalysis/src/org/labkey/sequenceanalysis/pipeline/ProcessVariantsHandler.java

@@ -1,10 +1,8 @@
 package org.labkey.sequenceanalysis.pipeline;
 
-import htsjdk.samtools.SAMSequenceDictionary;
 import htsjdk.samtools.util.Interval;
 import htsjdk.tribble.AbstractFeatureReader;
 import htsjdk.tribble.FeatureReader;
-import htsjdk.variant.utils.SAMSequenceDictionaryExtractor;
 import htsjdk.variant.vcf.VCFCodec;
 import htsjdk.variant.vcf.VCFFileReader;
 import htsjdk.variant.vcf.VCFHeader;
@@ -38,6 +36,7 @@
 import org.labkey.api.sequenceanalysis.run.SimpleScriptWrapper;
 import org.labkey.api.util.FileType;
 import org.labkey.api.util.FileUtil;
+import org.labkey.api.util.URIUtil;
 import org.labkey.api.view.ActionURL;
 import org.labkey.api.writer.PrintWriters;
 import org.labkey.sequenceanalysis.SequenceAnalysisModule;
@@ -403,9 +402,22 @@ public static File processVCF(File input, Integer libraryId, JobContext ctx, Res
                 action.addInput(vcfIdx, "Input VCF Index");
             }
 
-            resumer.getFileManager().addIntermediateFile(currentVCF);
-            resumer.getFileManager().addIntermediateFile(vcfIdx);
-
+            if (!currentVCF.equals(input))
+            {
+                if (!URIUtil.isDescendant(ctx.getOutputDir().toURI(), currentVCF.toURI()))
+                {
+                    ctx.getLogger().info("VCF is not a descendent of the output directory, will not add as intermediate file: " + currentVCF.getPath());
+                }
+                else
+                {
+                    resumer.getFileManager().addIntermediateFile(currentVCF);
+                    resumer.getFileManager().addIntermediateFile(vcfIdx);
+                }
+            }
+            else
+            {
+                ctx.getLogger().info("ending VCF same as input, will not add as intermediate file: " + input.getPath());
+            }
             ReferenceGenome genome = ctx.getSequenceSupport().getCachedGenome(libraryId);
             action.addInput(genome.getSourceFastaFile(), "Reference FASTA");