use correct content-type when overwriting and copying files

src/curl.cpp

@@ -2961,13 +2961,14 @@ int S3fsCurl::PutHeadRequest(const char* tpath, headers_t& meta, bool is_copy)
   responseHeaders.clear();
   bodydata.Clear();
 
+  string contype       = S3fsCurl::LookupMimeType(string(tpath));
+  requestHeaders = curl_slist_sort_insert(requestHeaders, "Content-Type", contype.c_str());
+
   // Make request headers
   for(headers_t::iterator iter = meta.begin(); iter != meta.end(); ++iter){
     string key   = lower(iter->first);
     string value = iter->second;
-    if(key == "content-type"){
+    if(key.substr(0, 9) == "x-amz-acl"){
-      requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
-    }else if(key.substr(0, 9) == "x-amz-acl"){
       // not set value, but after set it.
     }else if(key.substr(0, 10) == "x-amz-meta"){
       requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
@@ -3106,12 +3107,13 @@ int S3fsCurl::PutRequest(const char* tpath, headers_t& meta, int fd)
     requestHeaders = curl_slist_sort_insert(requestHeaders, "Content-MD5", strMD5.c_str());
   }
 
+  string contype       = S3fsCurl::LookupMimeType(string(tpath));
+  requestHeaders = curl_slist_sort_insert(requestHeaders, "Content-Type", contype.c_str());
+
   for(headers_t::iterator iter = meta.begin(); iter != meta.end(); ++iter){
     string key   = lower(iter->first);
     string value = iter->second;
-    if(key == "content-type"){
+    if(key.substr(0, 9) == "x-amz-acl"){
-      requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
-    }else if(key.substr(0, 9) == "x-amz-acl"){
       // not set value, but after set it.
     }else if(key.substr(0, 10) == "x-amz-meta"){
       requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
@@ -3714,13 +3716,14 @@ int S3fsCurl::CopyMultipartPostSetup(const char* from, const char* to, int part_
   bodydata.Clear();
   headdata.Clear();
 
+  string contype       = S3fsCurl::LookupMimeType(string(to));
+  requestHeaders = curl_slist_sort_insert(requestHeaders, "Content-Type", contype.c_str());
+
   // Make request headers
   for(headers_t::iterator iter = meta.begin(); iter != meta.end(); ++iter){
     string key   = lower(iter->first);
     string value = iter->second;
-    if(key == "content-type"){
+    if(key == "x-amz-copy-source"){
-      requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
-    }else if(key == "x-amz-copy-source"){
       requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());
     }else if(key == "x-amz-copy-source-range"){
       requestHeaders = curl_slist_sort_insert(requestHeaders, iter->first.c_str(), value.c_str());

test/integration-test-main.sh

@@ -75,6 +75,12 @@ function test_mv_file {
        return 1
     fi
     
+    #check the renamed file content-type
+    if [ -f "/etc/mime.types" ]
+    then
+      check_content_type "$1/$ALT_TEST_TEXT_FILE" "text/plain"
+    fi
+
     # Check the contents of the alt file
     ALT_FILE_LENGTH=`wc -c $ALT_TEST_TEXT_FILE | awk '{print $1}'`
     if [ "$ALT_FILE_LENGTH" -ne "$ALT_TEXT_LENGTH" ]
@@ -282,7 +288,7 @@ function test_external_modification {
     echo "old" > ${TEST_TEXT_FILE}
     OBJECT_NAME="$(basename $PWD)/${TEST_TEXT_FILE}"
     sleep 2
-    echo "new new" | aws_cli cp - "s3://${TEST_BUCKET_1}/${OBJECT_NAME}"
+    echo "new new" | aws_cli s3 cp - "s3://${TEST_BUCKET_1}/${OBJECT_NAME}"
     cmp ${TEST_TEXT_FILE} <(echo "new new")
     rm -f ${TEST_TEXT_FILE}
 }
@@ -291,7 +297,7 @@ function test_read_external_object() {
     describe "create objects via aws CLI and read via s3fs"
     OBJECT_NAME="$(basename $PWD)/${TEST_TEXT_FILE}"
     sleep 3
-    echo "test" | aws_cli cp - "s3://${TEST_BUCKET_1}/${OBJECT_NAME}"
+    echo "test" | aws_cli s3 cp - "s3://${TEST_BUCKET_1}/${OBJECT_NAME}"
     cmp ${TEST_TEXT_FILE} <(echo "test")
     rm -f ${TEST_TEXT_FILE}
 }
@@ -343,6 +349,9 @@ function test_multipart_copy {
        return 1
     fi
 
+    #check the renamed file content-type
+    check_content_type "$1/${BIG_FILE}-copy" "application/octet-stream"
+
     rm -f "/tmp/${BIG_FILE}"
     rm_test_file "${BIG_FILE}-copy"
 }

test/test-utils.sh

@@ -142,7 +142,8 @@ function cd_run_dir {
         echo "TEST_BUCKET_MOUNT_POINT variable not set"
         exit 1
     fi
-    RUN_DIR=$(mktemp -d ${TEST_BUCKET_MOUNT_POINT_1}/testrun-XXXXXX)
+    RUN_DIR=${TEST_BUCKET_MOUNT_POINT_1}/${1}
+    mkdir -p ${RUN_DIR}
     cd ${RUN_DIR}
 }
 
@@ -191,7 +192,8 @@ function describe {
 # made after the test run.  
 function run_suite {
    orig_dir=$PWD
-   cd_run_dir
+   key_prefix="testrun-$RANDOM"
+   cd_run_dir $key_prefix
    for t in "${TEST_LIST[@]}"; do
        # The following sequence runs tests in a subshell to allow continuation
        # on test failure, but still allowing errexit to be in effect during
@@ -202,7 +204,7 @@ function run_suite {
        # Other ways of trying to capture the return value will also disable
        # errexit in the function due to bash... compliance with POSIX? 
        set +o errexit
-       (set -o errexit; $t)
+       (set -o errexit; $t $key_prefix)
        if [[ $? == 0 ]]; then
            report_pass $t
        else
@@ -247,7 +249,15 @@ function get_mtime() {
         stat -c %Y "$1"
     fi
 }
+function check_content_type() {
+    INFO_STR=`aws_cli s3api head-object --bucket ${TEST_BUCKET_1} --key $1`
+    if [[ "${INFO_STR}" != *"$2"* ]]
+    then
+        echo "moved file content-type is not as expected expected:$2 got:${INFO_STR}"
+        exit 1
+    fi
+}
 
 function aws_cli() {
-    AWS_ACCESS_KEY_ID=local-identity AWS_SECRET_ACCESS_KEY=local-credential aws s3 --endpoint-url "${S3_URL}" --no-verify-ssl $*
+    AWS_ACCESS_KEY_ID=local-identity AWS_SECRET_ACCESS_KEY=local-credential aws $* --endpoint-url "${S3_URL}" --no-verify-ssl 
 }