upon multiple calls to `parse-query` with a failure on AST creation, it causes a...

align with s3select API
adding the JSON output format, otherwise it cause some s3test to fail.
remove obsolete initialization. the init is done by the constructor
s3select submodule

Signed-off-by: Gal Salomon <gal.salomon@gmail.com>

diff --git a/src/rgw/rgw_s3select.cc b/src/rgw/rgw_s3select.cc
index 800d276a6aab1c75c4ade9ef8d73e21dbd771d2f..73b55ebbf20d8b42bd5bb57f446f9d7d96afca12 100644 (file)
--- a/src/rgw/rgw_s3select.cc
+++ b/src/rgw/rgw_s3select.cc
@@ -287,6 +287,7 @@ RGWSelectObj_ObjStore_S3::RGWSelectObj_ObjStore_S3():
   m_object_size_for_processing(0),
   m_parquet_type(false),
   m_json_type(false),
+  m_outputFormat(OutputFormat::CSV),
   chunk_number(0),
   m_requested_range(0),
   m_scan_offset(1024),
@@ -426,7 +427,9 @@ int RGWSelectObj_ObjStore_S3::run_s3select_on_csv(const char* query, const char*
   } else if(m_header_info.compare("USE")==0) {
     csv.use_header_info=true;
   }
-
+  if (m_outputFormat == OutputFormat::JSON) {
+    csv.output_json_format = true;
+  }
   m_s3_csv_object.set_csv_query(&s3select_syntax, csv);
 
   m_s3_csv_object.set_external_system_functions(fp_s3select_continue,
@@ -478,6 +481,7 @@ int RGWSelectObj_ObjStore_S3::run_s3select_on_parquet(const char* query)
   if (!m_s3_parquet_object.is_set()) {
     //parsing the SQL statement.
     s3select_syntax.parse_query(m_sql_query.c_str());
+    parquet_object::csv_definitions parquet;
 
   m_s3_parquet_object.set_external_system_functions(fp_s3select_continue,
                                                fp_s3select_result_format,
@@ -486,7 +490,7 @@ int RGWSelectObj_ObjStore_S3::run_s3select_on_parquet(const char* query)
 
     try {
       //at this stage the Parquet-processing requires for the meta-data that reside on Parquet object 
-      m_s3_parquet_object.set_parquet_object(std::string("s3object"), &s3select_syntax, &m_rgw_api);
+      m_s3_parquet_object.set_parquet_object(std::string("s3object"), &s3select_syntax, &m_rgw_api, parquet);
     } catch(base_s3select_exception& e) {
       ldpp_dout(this, 10) << "S3select: failed upon parquet-reader construction: " << e.what() << dendl;
       fp_result_header_format(m_aws_response_handler.get_sql_result());
@@ -618,6 +622,10 @@ int RGWSelectObj_ObjStore_S3::handle_aws_cli_parameters(std::string& sql_query)
     ldpp_dout(this, 10) << "s3select: engine is set to process Parquet objects" << dendl;
   }
 
+  if (m_s3select_query.find(output_tag+"><JSON") != std::string::npos) {
+    m_outputFormat = OutputFormat::JSON;
+  }
+
   extract_by_tag(m_s3select_query, "Expression", sql_query);
   extract_by_tag(m_s3select_query, "Enabled", m_enable_progress);
   size_t _qi = m_s3select_query.find("<" + input_tag + ">", 0);
@@ -748,7 +756,6 @@ void RGWSelectObj_ObjStore_S3::execute(optional_yield y)
       op_ret = -ERR_INVALID_REQUEST;
       return;
     }
-    s3select_syntax.parse_query(m_sql_query.c_str());
     status = run_s3select_on_parquet(m_sql_query.c_str());
     if (status) {
       ldout(s->cct, 10) << "S3select: failed to process query <" << m_sql_query << "> on object " << s->object->get_name() << dendl;

diff --git a/src/rgw/rgw_s3select_private.h b/src/rgw/rgw_s3select_private.h
index 7beac4f4a5d881e60d6f93c5f7634875db6dfd5c..87d180acebc0f32f844c88d72b6ab952e6787d64 100644 (file)
--- a/src/rgw/rgw_s3select_private.h
+++ b/src/rgw/rgw_s3select_private.h
@@ -241,6 +241,11 @@ private:
   const char* s3select_json_error = "InvalidJsonType";
 
 public:
+  enum class OutputFormat {
+            CSV,
+            JSON
+        };
+  OutputFormat m_outputFormat;
   unsigned int chunk_number;
   size_t m_requested_range;
   size_t m_scan_offset;

diff --git a/src/s3select b/src/s3select
index f333ec82e6e8a3f7eb9ba1041d1442b2c7cd0f05..a6168396d246184ea4aadfb8ab3b9ab33848294c 160000 (submodule)
--- a/src/s3select
+++ b/src/s3select
@@ -1 +1 @@
-Subproject commit f333ec82e6e8a3f7eb9ba1041d1442b2c7cd0f05
+Subproject commit a6168396d246184ea4aadfb8ab3b9ab33848294c