BH_4879: stop search based on max results arg

BigBasket · Feb 18, 2024 · d12c807 · d12c807
1 parent fd2647e
commit d12c807
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 3 deletions.
diff --git a/components/core/src/clp_s/clp-s.cpp b/components/core/src/clp_s/clp-s.cpp
@@ -223,7 +223,8 @@ bool search_archive(
             archive_dir,
             timestamp_dict,
             std::move(output_handler),
-            command_line_arguments.get_ignore_case()
+            command_line_arguments.get_ignore_case(),
+            command_line_arguments.get_max_num_results()
     );
     output.filter();
 

diff --git a/components/core/src/clp_s/search/Output.cpp b/components/core/src/clp_s/search/Output.cpp
@@ -19,6 +19,7 @@
 namespace clp_s::search {
 void Output::filter() {
     auto top_level_expr = m_expr;
+    auto total_matched_results = 0;
 
     for (auto const& archive : ReaderUtils::get_archives(m_archives_dir)) {
         std::vector<int32_t> matched_schemas;
@@ -113,7 +114,12 @@ void Output::filter() {
                 }
             } else {
                 while (reader.get_next_message(message, this)) {
+                    if (total_matched_results >= m_max_num_results) {
+
+                        break;
+                    }
                     m_output_handler->write(message);
+                    total_matched_results = total_matched_results + 1;
                 }
             }
 

diff --git a/components/core/src/clp_s/search/Output.hpp b/components/core/src/clp_s/search/Output.hpp
@@ -31,15 +31,17 @@ class Output : public FilterClass {
            std::string archives_dir,
            std::shared_ptr<TimestampDictionaryReader> timestamp_dict,
            std::unique_ptr<OutputHandler> output_handler,
-           bool ignore_case)
+           bool ignore_case,
+           uint64_t max_num_results)
             : m_schema_tree(std::move(tree)),
               m_schemas(std::move(schemas)),
               m_match(match),
               m_expr(std::move(expr)),
               m_archives_dir(std::move(archives_dir)),
               m_timestamp_dict(std::move(timestamp_dict)),
               m_output_handler(std::move(output_handler)),
-              m_ignore_case(ignore_case) {}
+              m_ignore_case(ignore_case),
+              m_max_num_results{max_num_results} {}
 
     /**
      * Filters messages from all archives
@@ -52,6 +54,7 @@ class Output : public FilterClass {
     std::string m_archives_dir;
     std::unique_ptr<OutputHandler> m_output_handler;
     bool m_ignore_case;
+    uint64_t m_max_num_results;
 
     // variables for the current schema being filtered
     std::vector<BaseColumnReader*> m_searched_columns;