diff --git a/functest/clean.sh b/functest/clean.sh
new file mode 100755
index 000000000..37e78239a
--- /dev/null
+++ b/functest/clean.sh
@@ -0,0 +1,5 @@
+
+source common.rc
+
+rm -rf $RAW_DATA_LOCAL $CODE_LOCAL
+
diff --git a/functest/common.rc b/functest/common.rc
new file mode 100644
index 000000000..ef1b8582f
--- /dev/null
+++ b/functest/common.rc
@@ -0,0 +1,27 @@
+
+BASE_DIR_S3=https://s3-us-west-2.amazonaws.com/dataversioncontrol
+RAW_DATA_S3=$BASE_DIR_S3/stackoverflow_raw_small
+
+RAW_DATA_LOCAL=sync_raw_data
+CODE_LOCAL=sync_code
+TEST_REPO=repo_clean
+
+TEST_REPO=repo_clean
+
+function _init_repo() {
+    git init
+    dvc-init
+    mkdir data/xml
+    dvc-data-import ../$RAW_DATA_LOCAL/* data/xml
+    mkdir code
+    cp ../$CODE_LOCAL/* code/
+    git add code/
+    git commit -m 'Add code'
+}
+
+function create_repo() {
+    rm -rf $TEST_REPO
+    mkdir $TEST_REPO
+    (cd $TEST_REPO && _init_repo)
+}
+
diff --git a/functest/init.sh b/functest/init.sh
new file mode 100755
index 000000000..c36f5d9bf
--- /dev/null
+++ b/functest/init.sh
@@ -0,0 +1,16 @@
+
+source common.rc
+
+./clean.sh
+
+rm -rf $RAW_DATA_LOCAL
+mkdir $RAW_DATA_LOCAL
+for file in "Badges.xml" "Tags.xml" "PostLinks.xml"
+do
+    (cd $RAW_DATA_LOCAL; wget $RAW_DATA_S3/$file)
+done
+
+rm -rf $CODE_LOCAL
+mkdir $CODE_LOCAL
+(cd $CODE_LOCAL; wget "$BASE_DIR_S3/functests/code/xmltotsv.py")
+
diff --git a/functest/test-repro-code.sh b/functest/test-repro-code.sh
new file mode 100755
index 000000000..657c276dd
--- /dev/null
+++ b/functest/test-repro-code.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+
+source common.rc
+
+function run_test() {
+    mkdir data/tsv
+    dvc-run --code code/xmltotsv.py python code/xmltotsv.py --extract row/@Id,row/@UserId,row/@Name,row/@Class,row/@TagBased,row/@Date data/xml/Badges.xml data/tsv/Badges.tsv
+    
+    echo " " >> code/xmltotsv.py 
+    git commit -am 'Change code'
+    dvc-repro data/tsv/Badges.tsv
+    
+}
+
+TEST_REPO=repo_repro_test
+create_repo
+(cd repo_repro_test; run_test)