add if-match to auto-range get (#226)

*Issue #, if available:*

- If the object was updated when we get some range. The other range of the object could be from another version and we don't error out.

*Description of changes:*

- Add if-match header to make sure error out when the corner case happens.

include/aws/s3/private/s3_auto_ranged_get.h

@@ -45,6 +45,9 @@ struct aws_s3_auto_ranged_get {
     } synced_data;
 
     uint32_t initial_message_has_range_header : 1;
+    uint32_t initial_message_has_if_match_header : 1;
+
+    struct aws_string *etag;
 };
 
 /* Creates a new auto-ranged get meta request.  This will do multiple parallel ranged-gets when appropriate. */

include/aws/s3/private/s3_util.h

@@ -143,6 +143,8 @@ extern const struct aws_byte_cursor g_s3_service_name;
 AWS_S3_API
 extern const struct aws_byte_cursor g_range_header_name;
 
+extern const struct aws_byte_cursor g_if_match_header_name;
+
 AWS_S3_API
 extern const struct aws_byte_cursor g_content_range_header_name;
 

include/aws/s3/s3.h

@@ -33,6 +33,7 @@ enum aws_s3_errors {
     AWS_ERROR_S3_LIST_PARTS_PARSE_FAILED,
     AWS_ERROR_S3_RESUMED_PART_CHECKSUM_MISMATCH,
     AWS_ERROR_S3_RESUME_FAILED,
+    AWS_ERROR_S3_OBJECT_MODIFIED,
     AWS_ERROR_S3_END_RANGE = AWS_ERROR_ENUM_END_RANGE(AWS_C_S3_PACKAGE_ID)
 };
 

source/s3.c

@@ -34,6 +34,7 @@ static struct aws_error_info s_errors[] = {
     AWS_DEFINE_ERROR_INFO_S3(AWS_ERROR_S3_LIST_PARTS_PARSE_FAILED, "Failed to parse result from list parts"),
     AWS_DEFINE_ERROR_INFO_S3(AWS_ERROR_S3_RESUMED_PART_CHECKSUM_MISMATCH, "Checksum does not match previously uploaded part"),
     AWS_DEFINE_ERROR_INFO_S3(AWS_ERROR_S3_RESUME_FAILED, "Resuming request failed"),
+    AWS_DEFINE_ERROR_INFO_S3(AWS_ERROR_S3_OBJECT_MODIFIED, "The object modifed during download.")
 };
 /* clang-format on */
 

source/s3_auto_ranged_get.c

@@ -96,6 +96,7 @@ struct aws_s3_meta_request *aws_s3_meta_request_auto_ranged_get_new(
     AWS_ASSERT(headers != NULL);
 
     auto_ranged_get->initial_message_has_range_header = aws_http_headers_has(headers, g_range_header_name);
+    auto_ranged_get->initial_message_has_if_match_header = aws_http_headers_has(headers, g_if_match_header_name);
 
     AWS_LOGF_DEBUG(
         AWS_LS_S3_META_REQUEST, "id=%p Created new Auto-Ranged Get Meta Request.", (void *)&auto_ranged_get->base);
@@ -115,6 +116,7 @@ static void s_s3_meta_request_auto_ranged_get_destroy(struct aws_s3_meta_request
     AWS_PRECONDITION(meta_request->impl);
 
     struct aws_s3_auto_ranged_get *auto_ranged_get = meta_request->impl;
+    aws_string_destroy(auto_ranged_get->etag);
     aws_mem_release(meta_request->allocator, auto_ranged_get);
 }
 
@@ -350,6 +352,7 @@ static int s_s3_auto_ranged_get_prepare_request(
 
     /* Generate a new ranged get request based on the original message. */
     struct aws_http_message *message = NULL;
+    struct aws_s3_auto_ranged_get *auto_ranged_get = meta_request->impl;
 
     switch (request->request_tag) {
         case AWS_S3_AUTO_RANGE_GET_REQUEST_TYPE_HEAD_OBJECT:
@@ -382,6 +385,19 @@ static int s_s3_auto_ranged_get_prepare_request(
     if (meta_request->checksum_config.validate_response_checksum) {
         aws_http_headers_set(aws_http_message_get_headers(message), g_request_validation_mode, g_enabled);
     }
+    if (!auto_ranged_get->initial_message_has_if_match_header && auto_ranged_get->etag) {
+        /* Add the if_match to the request */
+        AWS_LOGF_DEBUG(
+            AWS_LS_S3_META_REQUEST,
+            "id=%p: Added the If-Match header to request %p for part %d",
+            (void *)meta_request,
+            (void *)request,
+            request->part_number);
+        aws_http_headers_set(
+            aws_http_message_get_headers(message),
+            g_if_match_header_name,
+            aws_byte_cursor_from_string(auto_ranged_get->etag));
+    }
 
     aws_s3_request_setup_send_data(request, message);
     aws_http_message_release(message);
@@ -574,6 +590,24 @@ static void s_s3_auto_ranged_get_request_finished(
             goto update_synced_data;
         }
 
+        if (!request_failed && !auto_ranged_get->initial_message_has_if_match_header) {
+            AWS_ASSERT(auto_ranged_get->etag == NULL);
+            struct aws_byte_cursor etag_header_value;
+
+            if (aws_http_headers_get(request->send_data.response_headers, g_etag_header_name, &etag_header_value)) {
+                aws_raise_error(AWS_ERROR_S3_MISSING_ETAG);
+                error_code = AWS_ERROR_S3_MISSING_ETAG;
+                goto update_synced_data;
+            }
+
+            AWS_LOGF_TRACE(
+                AWS_LS_S3_META_REQUEST,
+                "id=%p Etag received for the meta request. value is: " PRInSTR "",
+                (void *)meta_request,
+                AWS_BYTE_CURSOR_PRI(etag_header_value));
+            auto_ranged_get->etag = aws_string_new_from_cursor(auto_ranged_get->base.allocator, &etag_header_value);
+        }
+
         /* If we were able to discover the object-range/content length successfully, then any error code that was passed
          * into this function is being handled and does not indicate an overall failure.*/
         error_code = AWS_ERROR_SUCCESS;
@@ -671,6 +705,12 @@ static void s_s3_auto_ranged_get_request_finished(
         }
 
         if (error_code != AWS_ERROR_SUCCESS) {
+            if (error_code == AWS_ERROR_S3_INVALID_RESPONSE_STATUS &&
+                request->send_data.response_status == AWS_HTTP_STATUS_CODE_412_PRECONDITION_FAILED &&
+                !auto_ranged_get->initial_message_has_if_match_header) {
+                /* Use more clear error code as we added the if-match header under the hood. */
+                error_code = AWS_ERROR_S3_OBJECT_MODIFIED;
+            }
             aws_s3_meta_request_set_fail_synced(meta_request, request, error_code);
             if (error_code == AWS_ERROR_S3_RESPONSE_CHECKSUM_MISMATCH) {
                 /* It's a mismatch of checksum, tell user that we validated the checksum and the algorithm we validated

source/s3_request_messages.c

@@ -220,7 +220,7 @@ static const struct aws_byte_cursor s_x_amz_meta_prefix = AWS_BYTE_CUR_INIT_FROM
 const size_t g_s3_abort_multipart_upload_excluded_headers_count =
     AWS_ARRAY_SIZE(g_s3_abort_multipart_upload_excluded_headers);
 
-static int s_s3_message_util_add_content_range_header(
+static int s_s3_message_util_add_range_header(
     uint64_t part_range_start,
     uint64_t part_range_end,
     struct aws_http_message *out_message);
@@ -242,7 +242,7 @@ struct aws_http_message *aws_s3_ranged_get_object_message_new(
         return NULL;
     }
 
-    if (s_s3_message_util_add_content_range_header(range_start, range_end, message)) {
+    if (s_s3_message_util_add_range_header(range_start, range_end, message)) {
         goto error_clean_up;
     }
 
@@ -995,8 +995,8 @@ int aws_s3_message_util_copy_headers(
     return AWS_OP_SUCCESS;
 }
 
-/* Add a content-range header.*/
-static int s_s3_message_util_add_content_range_header(
+/* Add a range header.*/
+static int s_s3_message_util_add_range_header(
     uint64_t part_range_start,
     uint64_t part_range_end,
     struct aws_http_message *out_message) {

source/s3_util.c

@@ -22,6 +22,7 @@ const struct aws_byte_cursor g_s3_client_version = AWS_BYTE_CUR_INIT_FROM_STRING
 const struct aws_byte_cursor g_s3_service_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("s3");
 const struct aws_byte_cursor g_host_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("Host");
 const struct aws_byte_cursor g_range_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("Range");
+const struct aws_byte_cursor g_if_match_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("If-Match");
 const struct aws_byte_cursor g_etag_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("ETag");
 const struct aws_byte_cursor g_content_range_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("Content-Range");
 const struct aws_byte_cursor g_content_type_header_name = AWS_BYTE_CUR_INIT_FROM_STRING_LITERAL("Content-Type");