name_list --> name_char_data and name_offsets

-Use stxxl vectors with char and unsigned int containers
-Write out the entire character vector to fil
-Cap the names at length 255 during the parsing so we reduce
 the amount of memory used by stxxl vectors and we can do a
 direct writing of the character vector to .names

extractor/extraction_containers.cpp

@@ -55,7 +55,8 @@ ExtractionContainers::ExtractionContainers()
 {
     // Check if stxxl can be instantiated
     stxxl::vector<unsigned> dummy_vector;
-    name_list.push_back("");
+    name_char_data.push_back('\0');
+    name_offsets.push_back(1);
 }
 
 ExtractionContainers::~ExtractionContainers()
@@ -64,7 +65,8 @@ ExtractionContainers::~ExtractionContainers()
     used_node_id_list.clear();
     all_nodes_list.clear();
     all_edges_list.clear();
-    name_list.clear();
+    name_char_data.clear();
+    name_offsets.clear();
     restrictions_list.clear();
     way_start_end_id_list.clear();
 }
@@ -116,27 +118,20 @@ void ExtractionContainers::WriteNames(const std::string& names_file_name) const
     boost::filesystem::ofstream name_file_stream(names_file_name, std::ios::binary);
 
     unsigned total_length = 0;
-    std::vector<unsigned> name_lengths;
+
-    for (const std::string &temp_string : name_list)
+    for (const unsigned &name_offset : name_offsets)
     {
-        const unsigned string_length =
+        total_length += name_offset;
-            std::min(static_cast<unsigned>(temp_string.length()), 255u);
-        name_lengths.push_back(string_length);
-        total_length += string_length;
     }
 
     // builds and writes the index
-    RangeTable<> name_index_range(name_lengths);
+    RangeTable<> name_index_range(name_offsets);
     name_file_stream << name_index_range;
 
     name_file_stream.write((char *)&total_length, sizeof(unsigned));
+
     // write all chars consecutively
-    for (const std::string &temp_string : name_list)
+    name_file_stream.write((const char *)&name_char_data[0], name_char_data.size());
-    {
-        const unsigned string_length =
-            std::min(static_cast<unsigned>(temp_string.length()), 255u);
-        name_file_stream.write(temp_string.c_str(), string_length);
-    }
 
     name_file_stream.close();
     TIMER_STOP(write_name_index);

extractor/extraction_containers.hpp

@@ -70,7 +70,8 @@ class ExtractionContainers
     STXXLNodeIDVector used_node_id_list;
     STXXLNodeVector all_nodes_list;
     STXXLEdgeVector all_edges_list;
-    std::vector<std::string> name_list;
+    stxxl::vector<char> name_char_data;
+    std::vector<unsigned> name_offsets;
     STXXLRestrictionsVector restrictions_list;
     STXXLWayIDStartEndVector way_start_end_id_list;
     std::unordered_map<OSMNodeID, NodeID> external_to_internal_node_id_map;

extractor/extractor_callbacks.cpp

@@ -153,10 +153,21 @@ void ExtractorCallbacks::ProcessWay(const osmium::Way &input_way, const Extracti
 
     // Get the unique identifier for the street name
     const auto &string_map_iterator = string_map.find(parsed_way.name);
-    unsigned name_id = external_memory.name_list.size();
+    unsigned name_id = external_memory.name_offsets.size();
     if (string_map.end() == string_map_iterator)
     {
-        external_memory.name_list.push_back(parsed_way.name);
+        unsigned name_length = 0;
+        for (const char &c : parsed_way.name)
+        {
+            // Cap name length at 255 characters
+            if (name_length == 255u)
+            {
+                break;
+            }
+            external_memory.name_char_data.push_back(c);
+            name_length++;
+        }
+        external_memory.name_offsets.push_back(name_length);
         string_map.insert(std::make_pair(parsed_way.name, name_id));
     }
     else