From 118a3888db7f5095ce8f9d96644131c5a8caa12b Mon Sep 17 00:00:00 2001
From: Danny Colligan <danny.colligan@gmail.com>
Date: Wed, 30 Mar 2016 08:13:59 -0400
Subject: [PATCH] Fix repo manager choking on traililng path slashes

Issue #1032
---
 ga4gh/repo_manager.py                 | 23 ++++++++++++++++-------
 tests/end_to_end/test_repo_manager.py |  4 +++-
 2 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/ga4gh/repo_manager.py b/ga4gh/repo_manager.py
index 25dbf2b73..a2b583a7a 100644
--- a/ga4gh/repo_manager.py
+++ b/ga4gh/repo_manager.py
@@ -37,11 +37,20 @@ def filenameWithoutExtension(filepath, extension):
     (os.path.splitext(filename)[0] messes up
     on filenames with more than one period)
     """
-    filename = os.path.basename(filepath)
+    filename = basename(filepath)
     index = filename.index(extension)
     return filename[:index]
 
 
+def basename(path):
+    """
+    Return the final component of a pathname, after normalizing
+    the path (os.path.basename returns a empty string when given
+    a path with a trailing slash)
+    """
+    return os.path.basename(os.path.normpath(path))
+
+
 def runCommandSplits(splits, silent=False):
     """
     Run a shell command given the command's parsed command line
@@ -319,7 +328,7 @@ def addReferenceSet(self, filePath, moveMode, metadata):
             self._repoPath, self.referenceSetsDirName, fileName)
         self._assertPathEmpty(destPath, inRepo=True)
         os.mkdir(destPath)
-        fileDestPath = os.path.join(destPath, os.path.basename(filePath))
+        fileDestPath = os.path.join(destPath, basename(filePath))
         self._moveFile(filePath, fileDestPath, moveMode)
 
         # move the index files if they exist, otherwise do indexing
@@ -333,11 +342,11 @@ def addReferenceSet(self, filePath, moveMode, metadata):
         if os.path.exists(indexPathFai) and os.path.exists(indexPathGzi):
             self._moveFile(
                 indexPathFai,
-                os.path.join(destPath, os.path.basename(indexPathFai)),
+                os.path.join(destPath, basename(indexPathFai)),
                 moveMode)
             self._moveFile(
                 indexPathGzi,
-                os.path.join(destPath, os.path.basename(indexPathGzi)),
+                os.path.join(destPath, basename(indexPathGzi)),
                 moveMode)
         else:
             runCommand("samtools faidx {}".format(fileDestPath))
@@ -394,7 +403,7 @@ def addReadGroupSet(self, datasetName, filePath, moveMode):
         self._check()
         self._checkDataset(datasetName)
         self._checkFile(filePath, self.bamExtension)
-        fileName = os.path.basename(filePath)
+        fileName = basename(filePath)
         readGroupSetName = filenameWithoutExtension(
             fileName, self.bamExtension)
         destPath = os.path.join(
@@ -412,7 +421,7 @@ def addReadGroupSet(self, datasetName, filePath, moveMode):
             dstDir = os.path.split(destPath)[0]
             self._moveFile(
                 indexPath,
-                os.path.join(dstDir, os.path.basename(indexPath)),
+                os.path.join(dstDir, basename(indexPath)),
                 moveMode)
         else:
             pysam.index(destPath.encode('utf-8'))
@@ -447,7 +456,7 @@ def addVariantSet(self, datasetName, filePath, moveMode):
         self._check()
         self._checkDataset(datasetName)
         self._checkFolder(filePath, self.vcfExtension)
-        dirName = os.path.basename(filePath)
+        dirName = basename(filePath)
         destPath = os.path.join(
             self._repoPath, self.datasetsDirName, datasetName,
             self.variantsDirName, dirName)
diff --git a/tests/end_to_end/test_repo_manager.py b/tests/end_to_end/test_repo_manager.py
index 9861fc148..478d9315d 100644
--- a/tests/end_to_end/test_repo_manager.py
+++ b/tests/end_to_end/test_repo_manager.py
@@ -38,7 +38,9 @@ def testEndToEnd(self):
         self._runCmd("add-referenceset", paths.faPath)
         self._runCmd("add-dataset", self.datasetName)
         self._runCmd("add-readgroupset", self.datasetName, paths.bamPath)
-        self._runCmd("add-variantset", self.datasetName, paths.vcfDirPath)
+        self._runCmd(
+            "add-variantset", self.datasetName,
+            paths.vcfDirPath + '/')  # ensure we can handle trailing slashes
         self._runCmd("check", "--skipConsistencyCheck")
         self._runCmd("list")
         self._runCmd(