Can read Config

dev/src/dev.cpp

@@ -10,6 +10,8 @@ int main(int argc, char * argv[])
 
   Config config(argv[3], argv[1], argv[2]);
 
+  config.print(stdout);
+
   return 0;
 }
 

reading_machine/include/Config.hpp

@@ -21,10 +21,16 @@
 
 class Config
 {
+  public :
+
+  static constexpr const char * EOSColName = "EOS";
+  static constexpr const char * EOSSymbol1 = "1";
+  static constexpr const char * EOSSymbol0 = "0";
+
   private :
 
-  std::unordered_map<int, std::string> colIndex2Name;
-  std::unordered_map<int, std::string> colName2Index;
+  std::vector<std::string> colIndex2Name;
+  std::unordered_map<std::string, int> colName2Index;
 
   std::string rawInput;
   std::vector<std::string_view> rawInputUtf8;
@@ -33,9 +39,16 @@ class Config
   using Line = std::vector<ReferenceAndHypotheses>;
   std::vector<Line> lines;
 
+  private :
+
+  void readMCD(std::string_view mcdFilename);
+  void readRawInput(std::string_view rawFilename);
+  void readTSVInput(std::string_view tsvFilename);
+
   public :
 
   Config(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename);
+  void print(FILE * dest) const;
 };
 
 #endif

reading_machine/src/Config.cpp

@@ -9,49 +9,127 @@
 #include "Config.hpp"
 #include "util.hpp"
 
-Config::Config(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename)
+void Config::readMCD(std::string_view mcdFilename)
 {
-  if (tsvFilename.empty() and rawFilename.empty())
-    util::myThrow("tsvFilename and rawFilenames can't be both empty");
-  if (mcdFilename.empty())
-    util::myThrow("mcdFilename can't be empty");
+  if (!colIndex2Name.empty())
+    util::myThrow("a mcd has already been read for this Config");
+
+  std::FILE * file = std::fopen(mcdFilename.data(), "r");
 
+  if (not file)
+    util::myThrow(fmt::format("Cannot open file '{}'", mcdFilename));
+
+  char lineBuffer[1024];
+  while (std::fscanf(file, "%1023[^\n]\n", lineBuffer) == 1)
   {
-    std::FILE * file = std::fopen(mcdFilename.data(), "r");
+    colIndex2Name.emplace_back(lineBuffer);
+    colName2Index.emplace(lineBuffer, colIndex2Name.size()-1);
+  }
 
-    if (not file)
-      util::myThrow(fmt::format("Cannot open file '{}'", rawFilename));
+  std::fclose(file);
 
-    char lineBuffer[1024];
-    while (std::fscanf(file, "%1023[^\n]\n", lineBuffer) == 1)
-    {
-      auto splited = util::split(lineBuffer, ' ');
+  if (colName2Index.count(EOSColName))
+    util::myThrow(fmt::format("mcd '{}' must not contain column '{}'", mcdFilename, EOSColName));
+  colIndex2Name.emplace_back(EOSColName);
+  colName2Index.emplace(EOSColName, colIndex2Name.size()-1);
+}
 
-      if (splited.size() != 2)
-        util::myThrow(fmt::format("Invalid line '{}' for mcd '{}'", lineBuffer, mcdFilename));
-    }
+void Config::readRawInput(std::string_view rawFilename)
+{
+  std::FILE * file = std::fopen(rawFilename.data(), "r");
 
-    std::fclose(file);
-  }
+  if (not file)
+    util::myThrow(fmt::format("Cannot open file '{}'", rawFilename));
 
-  if (not rawFilename.empty())
+  while (not std::feof(file))
+    rawInput.push_back(std::fgetc(file));
+
+  std::fclose(file);
+
+  rawInputUtf8 = util::splitAsUtf8(rawInput);
+}
+
+void Config::readTSVInput(std::string_view tsvFilename)
+{
+  std::FILE * file = std::fopen(tsvFilename.data(), "r");
+
+  if (not file)
+    util::myThrow(fmt::format("Cannot open file '{}'", tsvFilename));
+
+  char lineBuffer[100000];
+  int inputLineIndex = 0;
+  bool inputHasBeenRead = false;
+  int usualNbCol = -1;
+  while (!std::feof(file))
   {
-    std::FILE * file = std::fopen(rawFilename.data(), "r");
+    if (lineBuffer != std::fgets(lineBuffer, 100000, file))
+      break;
+
+    std::string_view line(lineBuffer);
+    inputLineIndex++;
 
-    if (not file)
-      util::myThrow(fmt::format("Cannot open file '{}'", rawFilename));
+    if (line.size() < 3)
+    {
+      if (!inputHasBeenRead)
+        continue;
+
+      lines.back()[colName2Index[EOSColName]][0] = EOSSymbol1;
+
+      continue;
+    }
+    else if (line[0] == '#')
+      continue;
+
+    if (line.back() == '\n')
+      line.remove_suffix(1);
 
-    while (not std::feof(file))
-      rawInput.push_back(std::fgetc(file));
+    inputHasBeenRead = true;
 
-    std::fclose(file);
+    auto splited = util::split(line, '\t');
+    if (usualNbCol == -1)
+      usualNbCol = splited.size();
+    if ((int)splited.size() != usualNbCol)
+      util::myThrow(fmt::format("in file {} line {} is invalid, it shoud have {} columns", tsvFilename, line, usualNbCol));
 
-    rawInputUtf8 = util::splitAsUtf8(rawInput);
+    lines.emplace_back();
+    for (unsigned int i = 0; i < colIndex2Name.size(); i++)
+    {
+      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];
   }
 
+  std::fclose(file);
+}
+
+Config::Config(std::string_view mcdFilename, std::string_view tsvFilename, std::string_view rawFilename)
+{
+  if (tsvFilename.empty() and rawFilename.empty())
+    util::myThrow("tsvFilename and rawFilenames can't be both empty");
+  if (mcdFilename.empty())
+    util::myThrow("mcdFilename can't be empty");
+
+  readMCD(mcdFilename);
+
+  if (not rawFilename.empty())
+    readRawInput(rawFilename);
+
   if (not tsvFilename.empty())
+    readTSVInput(tsvFilename);
+}
+
+void Config::print(FILE * dest) const
+{
+  for (auto & line : lines)
   {
-  
+    for (unsigned int i = 0; i < line.size()-1; i++)
+      fmt::print(dest, "{}{}", line[i].back(), i < line.size()-2 ? "\t" : "\n");
+    if (line[colName2Index.at(EOSColName)].back() == EOSSymbol1)
+      fmt::print(dest, "\n");
   }
 }