TreeUtils.cxx

Go to the documentation of this file.

// @(#)root/tree:$Id$
// Author: Timur Pocheptsov   30/01/2014
 
/*************************************************************************
 * Copyright (C) 1995-2014, Rene Brun and Fons Rademakers.               *
 * All rights reserved.                                                  *
 *                                                                       *
 * For the licensing terms see $ROOTSYS/LICENSE.                         *
 * For the list of contributors see $ROOTSYS/README/CREDITS.             *
 *************************************************************************/
 
/** \class TreeUtils
\ingroup tree
 
Different standalone functions to work with trees and tuples,
not reqiuired to be a member of any class.
*/
 
#include <istream>
#include <cassert>
#include <cctype>
 
#include "TreeUtils.h"
#include "TNtupleD.h"
#include "TNtuple.h"
#include "TError.h"
#include "TTree.h"
 
namespace ROOT {
namespace TreeUtils {
 
//Some aux. functions to read tuple from a text file. No reason to make them memeber-functions.
void SkipEmptyLines(std::istream &input);
void SkipWSCharacters(std::istream &input);
bool NextCharacterIsEOL(std::istream &input);
 
//Enforce/limit what can be a Tuple.
//Actually, at the moment you can only use
//the fill function for TNtuple/TNtupleD
//(enforced by hidden definition and explicit instantiations).
//But in future this can potentially change.
 
//TODO: there is no line number in any of error messages.
//It can be improved, though, we can have mixed line endings
//so I can not rely on this numbering (for example, my vim shows these lines:
//aaaa\r\r\nbbb as
//aaaa
//bbbb
//Though it can be also treated as
//aaaa
//
//bbb
//or even as
//aaaa
//
//
//bbb - so line numbers can be useless and misleading.
 
template<class> struct InvalidTupleType;
 
template<>
struct InvalidTupleType<TNtuple>
{
   InvalidTupleType()
   {
   }
};
 
template<>
struct InvalidTupleType<TNtupleD>
{
   InvalidTupleType()
   {
   }
};
 
////////////////////////////////////////////////////////////////////////////////
 
template<class DataType, class Tuple>
Long64_t FillNtupleFromStream(std::istream &inputStream, Tuple &tuple, char delimiter, bool strictMode)
{
   InvalidTupleType<Tuple> typeChecker;
 
   if (delimiter == '\r' || delimiter == '\n') {
      ::Error("FillNtupleFromStream", "invalid delimiter - newline character");
      return 0;
   }
 
   if (delimiter == '#') {
      ::Error("FillNtuplesFromStream", "invalid delimiter, '#' symbols can only start a comment");
      return 0;
   }
 
   const Int_t nVars = tuple.GetNvar();
   if (nVars <= 0) {
      ::Error("FillNtupleFromStream", "invalid number of elements");
      return 0;
   }
 
   DataType *args = tuple.GetArgs();
   assert(args != nullptr && "FillNtupleFromStream, args buffer is a null");
 
   Long64_t nLines = 0;
 
   if (strictMode) {
      while (true) {
         //Skip empty-lines (containing only newlines, comments, whitespaces + newlines
         //and combinations).
         SkipEmptyLines(inputStream);
 
         if (!inputStream.good()) {
            if (!nLines)
               ::Error("FillNtupleFromStream", "no data read");
            return nLines;
         }
 
         //Now, we have to be able to read _the_ _required_ number of entires.
         for (Int_t i = 0; i < nVars; ++i) {
            SkipWSCharacters(inputStream);//skip all wses except newlines.
            if (!inputStream.good()) {
               ::Error("FillNtupleFromStream", "failed to read a tuple (not enough values found)");
               return nLines;
            }
 
            if (i > 0 && !std::isspace(delimiter)) {
               const char test = inputStream.peek();
               if (!inputStream.good() || test != delimiter) {
                  ::Error("FillNtupleFromStream", "delimiter expected");
                  return nLines;
               }
 
               inputStream.get();//we skip a dilimiter whatever it is.
               SkipWSCharacters(inputStream);
            }
 
            if (NextCharacterIsEOL(inputStream)) {
               //This is unexpected!
               ::Error("FillNtupleFromStream", "unexpected character or eof found");
               return nLines;
            }
 
            inputStream>>args[i];
 
            if (!(inputStream.eof() && i + 1 == nVars) && !inputStream.good()){
               ::Error("FillNtupleFromStream", "error while reading a value");
               return nLines;
            }
         }
 
         SkipWSCharacters(inputStream);
         if (!NextCharacterIsEOL(inputStream)) {
            ::Error("FillNtupleFromStream",
                    "only whitespace and new line can follow the last number on the line");
            return nLines;
         }
 
         //Only God forgives :) Ugly but ...
         //for TNtuple it's protected :)
         static_cast<TTree &>(tuple).Fill();
         ++nLines;
      }
   } else {
      Int_t i = 0;//how many values we found for a given tuple's entry.
      while (true) {
         //Skip empty lines, comments and whitespaces before
         //the first 'non-ws' symbol:
         //it can be a delimiter/a number (depends on a context) or an invalid symbol.
         SkipEmptyLines(inputStream);
 
         if (!inputStream.good()) {
            //No data to read, check what we read by this moment:
            if (!nLines)
               ::Error("FillNtupleFromStream", "no data read");
            else if (i > 0)//we've read only a part of a row.
               ::Error("FillNtupleFromStream", "unexpected character or eof found");
            return nLines;
         }
 
         if (i > 0 && !std::isspace(delimiter)) {
            //The next one must be a delimiter.
            const char test = inputStream.peek();
            if (!inputStream.good() || test != delimiter) {
               ::Error("FillNtupleFromStream", "delimiter expected (non-strict mode)");
               return nLines;
            }
 
            inputStream.get();//skip the delimiter.
            SkipEmptyLines(inputStream);//probably, read till eof.
         }
 
         //Here must be a number.
         inputStream>>args[i];
 
         if (!(inputStream.eof() && i + 1 == nVars) && !inputStream.good()){
            ::Error("FillNtupleFromStream", "error while reading a value");
            return nLines;
         }
 
         if (i + 1 == nVars) {
            //We god the row, can fill now and continue.
            static_cast<TTree &>(tuple).Fill();
            ++nLines;
            i = 0;
         } else
            ++i;
      }
   }
 
   return nLines;
}
 
template Long64_t FillNtupleFromStream<Float_t, TNtuple>(std::istream &, TNtuple &, char, bool);
template Long64_t FillNtupleFromStream<Double_t, TNtupleD>(std::istream &, TNtupleD &, char, bool);
 
//Aux. functions to read tuples from text files.
 
//file:
//    lines
//
//lines:
//    line
//    line lines
//
//line:
//    comment
//    tuple
//    empty-line
 
//comment:
// '#' non-newline-character-sequence newline-character
//
//non-newline-character-sequence:
// any symbol except '\r' or '\n'
//
//newline-character:
// '\r' | '\n'
////////////////////////////////////////////////////////////////////////////////
/// Skips everything from '#' to (including) '\\r' or '\\n'.
 
void SkipComment(std::istream &input)
{
   while (input.good()) {
      const char next = input.peek();
      if (input.good()) {
         input.get();
         if (next == '\r' || next == '\n')
            break;
      }
   }
}
 
//empty-line:
//    newline-character
//    ws-sequence newline-character
//    ws-sequence comment
////////////////////////////////////////////////////////////////////////////////
/// Skips empty lines (newline-characters), ws-lines (consisting only of whitespace characters + newline-characters).
 
void SkipEmptyLines(std::istream &input)
{
   while (input.good()) {
      const char c = input.peek();
      if (!input.good())
         break;
 
      if (c == '#')
         SkipComment(input);
      else if (!std::isspace(c))//'\r' and '\n' are also 'isspaces'.
         break;
      else
         input.get();
   }
}
 
//ws-sequence:
//    c such that isspace(c) and c is not a newline-character.
////////////////////////////////////////////////////////////////////////////////
/// Skip whitespace characters, but not newline-characters we support ('\\r' or '\\n').
 
void SkipWSCharacters(std::istream &input)
{
   while (input.good()) {
      const char next = input.peek();
      if (input.good()) {
         if (std::isspace(next) && next != '\n' && next != '\r')
            input.get();
         else
            break;
      }
   }
}
 
//Next character is either newline-character, eof or we have some problems reading
//the next symbol.
////////////////////////////////////////////////////////////////////////////////
/// Either '\\r' | '\\n' or eof of some problem.
 
bool NextCharacterIsEOL(std::istream &input)
{
   if (!input.good())
      return true;
 
   const char next = input.peek();
   if (!input.good())
      return true;
 
   return next == '\r' || next == '\n';
}
 
}//TreeUtils
}//ROOT