diff --git a/tools/datafile-generate.sh b/tools/datafile-generate.sh
new file mode 100755
index 00000000..6cc99f87
--- /dev/null
+++ b/tools/datafile-generate.sh
@@ -0,0 +1,40 @@
+#!/bin/bash
+data_dir=$1
+if [ $# -ne 2 ]
+then
+  echo "Usage: $0 {DATA_DIR}"
+  exit 1
+fi
+dir_output
+{
+  if test -d "$1"
+  then
+    echo "directory"
+  else
+    echo "file"
+  fi
+}
+rm dataset.txt
+cat_txt=$$
+find $data_dir | grep uuid.txt > $cat_txt
+cat $cat_txt | while read oneline;
+do
+echo $oneline >> dataset.txt;
+cat "$oneline" >> dataset.txt;
+uuid=`cat "$oneline"`;
+if [ -e $uuid.txt ]
+then
+  echo "do nothing! uuid重复？"
+else
+find -L "${oneline%uuid.txt}" -exec du -b -s {} \; -exec dir_output "{}" \; >$uuid.txt
+echo $uuid
+sed -i '/uuid.txt$/d' $uuid.txt
+fi
+echo -e \\n >> dataset.txt;
+done
+# 删除空行
+ sed -i '/^$/d' dataset.txt 
+rm $cat_txt
+sed -i 's/uuid.txt//' dataset.txt
+sed -i 's/^ //' dataset.txt
+exit 0