Improved size computation

common/include/util.hpp

@@ -17,11 +17,11 @@
 #include <string>
 #include <vector>
 #include <array>
+#include <unordered_map>
 #include <fmt/core.h>
 #include <experimental/source_location>
 #include "utf8.hpp"
 
-
 namespace util
 {
 typedef std::array<char, 4> utf8char;
@@ -39,6 +39,35 @@ void myThrow(std::string_view message, const std::experimental::source_location
 
 std::string int2HumanStr(int number);
 
+template<typename T>
+std::size_t memorySize(const T & val)
+{
+  myThrow("Type not yet supported");
+  return sizeof val;
+}
+
+inline std::size_t memorySize(int val)
+{
+  return sizeof val;
+}
+
+template<typename T>
+std::size_t memorySize(const std::basic_string<T> & val)
+{
+  return sizeof val + val.capacity() * sizeof (T);
+}
+
+template<typename T>
+std::size_t memorySize(const std::vector<T> & vec)
+{
+  std::size_t result = sizeof vec + sizeof (T) * (vec.capacity()-vec.size());
+
+  for (auto & elem : vec)
+    result += memorySize(elem);
+
+  return result;
+}
+
 };
 
 template <>

reading_machine/include/Config.hpp

@@ -19,6 +19,13 @@
 #include <unordered_map>
 #include "util.hpp"
 
+class Config;
+
+namespace util
+{
+std::size_t memorySize(const Config & c);
+};
+
 class Config
 {
   public :
@@ -45,6 +52,8 @@ class Config
   void readRawInput(std::string_view rawFilename);
   void readTSVInput(std::string_view tsvFilename);
 
+  friend std::size_t util::memorySize(const Config &);
+
   public :
 
   Config(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename);

reading_machine/src/Config.cpp

@@ -97,7 +97,9 @@ void Config::readTSVInput(std::string_view tsvFilename)
       lines.back().emplace_back();
       lines.back().back().emplace_back("");
     }
+
     lines.back()[colName2Index[EOSColName]][0] = EOSSymbol0;
+
     for (unsigned int i = 0; i < splited.size(); i++)
       if (i < colIndex2Name.size())
         lines.back()[i][0] = splited[i];
@@ -108,21 +110,20 @@ void Config::readTSVInput(std::string_view tsvFilename)
 
 void Config::printSize(FILE * dest)
 {
-  int rawInputNbElements = rawInput.size();
-  int rawInputSize = sizeof rawInput + rawInput.capacity() * sizeof rawInput[0];
-
-  int rawInputUtf8NbElements = 4*rawInputUtf8.size();
-  int rawInputUtf8Size = sizeof rawInputUtf8 + rawInputUtf8.capacity()* sizeof rawInputUtf8[0];
+  int rawInputSize = util::memorySize(rawInput);
+  int rawInputUtf8Size = util::memorySize(rawInputUtf8);
+  int linesSize = util::memorySize(lines);
 
-  int totalSize = rawInputSize + rawInputUtf8Size;
+  int totalSize = rawInputSize + rawInputUtf8Size + linesSize;
 
   std::string unit = "Mo";
   int unitPower = 6;
   float unitMultiplier = std::stof(fmt::format("0.{:0^{}}1","",unitPower-1));
 
-  fmt::print(dest, "{:<43} : {:<{}.2f} {}\n", fmt::format("{:<20} {:>12} elements", "rawInput", util::int2HumanStr(rawInputNbElements)), unitMultiplier*rawInputSize, 2+11-unitPower, unit);
-  fmt::print(dest, "{:<43} : {:<{}.2f} {}\n", fmt::format("{:<20} {:>12} elements", "rawInputUtf8", util::int2HumanStr(rawInputUtf8NbElements)), unitMultiplier*rawInputUtf8Size, 2+11-unitPower, unit);
-  fmt::print(dest, "{:<43} : {:<{}.2f} {}\n", "Total", unitMultiplier*totalSize, 2+11-unitPower, unit);
+  fmt::print(dest, "{:<15} : {:<{}.2f} {}\n", "rawInput", unitMultiplier*rawInputSize, 2+11-unitPower, unit);
+  fmt::print(dest, "{:<15} : {:<{}.2f} {}\n", "rawInputUtf8", unitMultiplier*rawInputUtf8Size, 2+11-unitPower, unit);
+  fmt::print(dest, "{:<15} : {:<{}.2f} {}\n", "lines", unitMultiplier*linesSize, 2+11-unitPower, unit);
+  fmt::print(dest, "{:<15} : {:<{}.2f} {}\n", "Total", unitMultiplier*totalSize, 2+11-unitPower, unit);
 }
 
 Config::Config(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename)
@@ -152,3 +153,8 @@ void Config::print(FILE * dest) const
   }
 }
 
+std::size_t util::memorySize(const Config & c)
+{
+  return sizeof c + memorySize(c.rawInput) + memorySize(c.rawInputUtf8) + memorySize(c.lines) + memorySize(c.colIndex2Name) + memorySize(c.colName2Index);
+}
+