How to read tsv file?

Hi,

I have a tab separated value table like this:
header1 header2 header3
13.455 55.3 A string
4.55 5.66 Another string

I want to load this guy into a vector of vectors, since I do not know how
long it may be. I think I have to have a vector of vectors of strings, and
then extract the doubles later(?):
std::vector<std::vector<std::string> > m_data_vec;

I started off with this skeletal function, but Im not sure how to parse the
line for tabs and newlines, and stuff the elements into the vector. Is it
better to read in the whole line then parse it? Can I parse it on the fly?
How?

void MyClass::ReadTSV(const* filename)
{
using namespace std;

ifstream infile(filename);
if (!infile) {
cout << "unable to load file" << endl;
}

// Now what?
}

Thanks,
Bryan

re: How to read tsv file?

"BCC" <a@b.c> wrote...[color=blue]
> I have a tab separated value table like this:
> header1 header2 header3
> 13.455 55.3 A string
> 4.55 5.66 Another string
>
> I want to load this guy into a vector of vectors, since I do not know how
> long it may be. I think I have to have a vector of vectors of strings,[/color]
and[color=blue]
> then extract the doubles later(?):
> std::vector<std::vector<std::string> > m_data_vec;
>
> I started off with this skeletal function, but Im not sure how to parse[/color]
the[color=blue]
> line for tabs and newlines, and stuff the elements into the vector. Is it
> better to read in the whole line then parse it?[/color]

Oh, so much better...
[color=blue]
> Can I parse it on the fly?[/color]

I don't know. Can you?
[color=blue]
> How?
>
> void MyClass::ReadTSV(const* filename)
> {
> using namespace std;
>
> ifstream infile(filename);
> if (!infile) {
> cout << "unable to load file" << endl;
> }
>
> // Now what?[/color]

If you know how many fields to expect, you could use get( ... , '\t') N-1
times and then get( ... , '\n') and then again and again.

Easier still to get one by one character and watch for '\t' and '\n'. But
I would still do the "get the whole line and then parse it" thing.
[color=blue]
> }[/color]

V

re: How to read tsv file?

"BCC" <a@b.c> wrote in message
news:p1lSb.7633$uM.3791@newssvr29.news.prodigy.com ...[color=blue]
> Hi,
>
> I have a tab separated value table like this:
> header1 header2 header3
> 13.455 55.3 A string
> 4.55 5.66 Another string
>
> I want to load this guy into a vector of vectors, since I do not know how
> long it may be. I think I have to have a vector of vectors of strings, and
> then extract the doubles later(?):
> std::vector<std::vector<std::string> > m_data_vec;
>
> I started off with this skeletal function, but Im not sure how to parse the
> line for tabs and newlines, and stuff the elements into the vector. Is it
> better to read in the whole line then parse it? Can I parse it on the fly?
> How?
>
> void MyClass::ReadTSV(const* filename)
> {
> using namespace std;
>
> ifstream infile(filename);
> if (!infile) {
> cout << "unable to load file" << endl;
> }
>
> // Now what?
> }[/color]
May be this gives you the basic idea.
I haven't tested it. Also no checks for errors etc.

<UNTESTED CODE>

#include <fstream>
#include <string>
#include <vector>
using namespace std;

void ReadTSV(const char* filename)
{
using namespace std;

ifstream infile(filename);
if (!infile) {
cout << "unable to load file" << endl;
}
string str;

vector<vector<string> > vvStr;
vector<string> vStr;
int pos1, pos2;
while (getline(infile, str))
{
pos1 = 0;
while((pos2 = str.find('\t'))!= string::npos)
{
vStr.push_back(str.substr(pos1, pos2));
pos1 = pos2++;
}
vStr.push_back(str.substr(pos1, string::npos));
vvStr.push_back(vStr);
}

}

</UNTESTED CODE>

Best wishes,
Sharad

re: How to read tsv file?

"BCC" <a@b.c> wrote in message
news:p1lSb.7633$uM.3791@newssvr29.news.prodigy.com ...[color=blue]
> Hi,
>
> I have a tab separated value table like this:
> header1 header2 header3
> 13.455 55.3 A string
> 4.55 5.66 Another string
>
> I want to load this guy into a vector of vectors, since I do not[/color]
know how[color=blue]
> long it may be. I think I have to have a vector of vectors of[/color]
strings, and[color=blue]
> then extract the doubles later(?):
> std::vector<std::vector<std::string> > m_data_vec;
>
> I started off with this skeletal function, but Im not sure ho to[/color]
parse the[color=blue]
> line for tabs and newlines, and stuff the elements into the vector.[/color]
Is it[color=blue]
> better to read in the whole line then parse it? Can I parse it on[/color]
the fly?[color=blue]
> How?[/color]

Here's some code I wrote some time ago for splitting sequences of
characters and adding them to lists. I have used it a lot with Visual
C++. I don''t guarantee its portability or efficiency, but I looks
generally okay.

Usage:

struct is_tab {
bool operator(char c) const { return c == '\t'; }
};

// Split s using tab as a separator character,
// adding segments to the end of a vector.
string s;
vector<string> vec;
split(s.begin(), s.end(), back_inserter(vec), is_tab(), false);

Here you could use any input iterators for the first and second
arguments; in particular, you should be able to use istream_iterators
or istreambuf_iterators.

Jonathan


---------------------
//
// File name: split.h
//
// Descriptions: Contains template functions for splitting a string
into
// a list.
//
// Author: Jonathan Turkanis
//
// Copyright: Jonathan Turkanis, July 29, 2002. See Readme.txt for
// license information.
//

#ifndef UT_SPLIT_H_INCLUDED
#define UT_SPLIT_H_INCLUDED

#include <iterator>
#include <locale>
#include <string>
#include <boost/bind.hpp>
#include <boost/ref.hpp>

namespace Utility {

//
// Function name: split.
//
// Description: Splits the given string into components.
//
// Template paramters:
// InIt - An input iterator type with any value type Elem.
// OutIt - An output iterator type with value type equal to
// std::basic_string<Elem>.
// Pred - A predicate with argument type Elem.
// Parameters:
// first - The beginning of the input sequence.
// last - The end of the input sequence.
// dest - Receives the terms in the generated list.
// sep - Determines where to split the input sequence.
// coalesce - true if sequences of consecutive elements satisfying
sep
// should be treated as one. Defaults to true.
//
template<class InIt, class OutIt, class Pred>
void split(InIt first, InIt last, OutIt dest, Pred sep, bool coalesce
= true);

//
// Function name: split_by_whitespace.
//
// Description: Splits the given string into components.
//
// Template paramters:
// InIt - An input iterator type with any value type Elem.
// OutIt - An output iterator type with value type equal to
// std::basic_string<Elem>.
// Pred - A predicate with argument type Elem.
// Parameters:
// first - The begiining of the input sequence.
// last - The end of the input sequence.
// dest - Receives the terms in the generated list.
//
template<class InIt, class OutIt>
void split_by_whitespace(InIt first, InIt last, OutIt dest)
{
using namespace std;
typedef iterator_traits<InIt>::value_type char_type;
locale loc;
split(first, last, dest, boost::bind(isspace<char_type>, _1,
boost::ref(loc)));
}

template<class InIt, class OutIt, class Pred>
void split(InIt first, InIt last, OutIt dest, Pred sep, bool coalesce)
{
using namespace std;
typedef iterator_traits<InIt>::value_type char_type;
typedef basic_string<char_type> string_type;

bool prev = true; // True if prev char was a separator.
string_type term;
while (first != last) {
char_type c = *first++;
bool is_sep = sep(c);
if (is_sep && (!coalesce || coalesce && !prev)) {
*dest++ = term;
term.clear();
}
if (!is_sep)
term += c;
prev = is_sep;
}
if (!term.empty() && !coalesce || coalesce && !prev)
*dest++ = term;
}
}

#endif // #ifndef UT_SPLIT_H_INCLUDED

re: How to read tsv file?

"Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
news:bvcrv5$qomb8$1@ID-221354.news.uni-berlin.de...[color=blue]
>
> "BCC" <a@b.c> wrote in message
> news:p1lSb.7633$uM.3791@newssvr29.news.prodigy.com ...[color=green]
> > Hi,
> >
> > I have a tab separated value table like this:
> > header1 header2 header3
> > 13.455 55.3 A string
> > 4.55 5.66 Another string
> >
> > I want to load this guy into a vector of vectors, since I do not know how
> > long it may be. I think I have to have a vector of vectors of strings, and
> > then extract the doubles later(?):
> > std::vector<std::vector<std::string> > m_data_vec;
> >
> > I started off with this skeletal function, but Im not sure how to parse the
> > line for tabs and newlines, and stuff the elements into the vector. Is it
> > better to read in the whole line then parse it? Can I parse it on the fly?
> > How?
> >
> > void MyClass::ReadTSV(const* filename)
> > {
> > using namespace std;
> >
> > ifstream infile(filename);
> > if (!infile) {
> > cout << "unable to load file" << endl;
> > }
> >
> > // Now what?
> > }[/color]
> May be this gives you the basic idea.
> I haven't tested it. Also no checks for errors etc.
>
> <UNTESTED CODE>
>
> #include <fstream>
> #include <string>
> #include <vector>
> using namespace std;
>
> void ReadTSV(const char* filename)
> {
> using namespace std;
>
> ifstream infile(filename);
> if (!infile) {
> cout << "unable to load file" << endl;
> }
> string str;
>
> vector<vector<string> > vvStr;
> vector<string> vStr;
> int pos1, pos2;
> while (getline(infile, str))
> {
> pos1 = 0;
> while((pos2 = str.find('\t'))!= string::npos)
> {
> vStr.push_back(str.substr(pos1, pos2));[/color]

oops..second parameter should be pos2-pos1+1 i guess.

re: How to read tsv file?

In article <p1lSb.7633$uM.3791@newssvr29.news.prodigy.com>, BCC <a@b.c> wrote:[color=blue]
>Hi,
>
>I have a tab separated value table like this:
>header1 header2 header3
>13.455 55.3 A string
>4.55 5.66 Another string
>
>I want to load this guy into a vector of vectors,[/color]

Use getline() to read one line at a time, then use a stringstream to split
the line into tokens. Note you can specify some other line terminator
than '\n', for getline().

std::vector<std::vector<std::string> > m_data_vec;
std::string line;
while (std::getline (infile, line))
{
std::istringstream linestream (line);
std::string token;
std::vector<std::string> row;
while (std::getline (linestream, token, '\t')
{
row.push_back (token);
}
m_data_vec.push_back (row);
}

Actually, your example is easy to parse without a stringstream, if you
use a struct to represent a line, with appropriate member data types:

struct data_rec
{
double foo, bar;
std::string baz;
};

std::vector<data_rec> m_data_vec;
data_rec linedata;
while ((infile >> linedata.foo >> linedata.bar))
&& std::getline (input, linedata.baz))
{
m_data_vec.push_back (linedata);
}

--
Jon Bell <jtbellm4h@presby.edu> Presbyterian College
Dept. of Physics and Computer Science Clinton, South Carolina USA

re: How to read tsv file?

In article <p1lSb.7633$uM.3791@newssvr29.news.prodigy.com>, BCC <a@b.c> wrote:[color=blue]
>Hi,
>
>I have a tab separated value table like this:
>header1 header2 header3
>13.455 55.3 A string
>4.55 5.66 Another string
>
>I want to load this guy into a vector of vectors,[/color]

Use getline() to read one line at a time, then use a stringstream to split
the line into tokens. Note you can specify some other line terminator
than '\n', for getline().

std::vector<std::vector<std::string> > m_data_vec;
std::string line;
while (std::getline (infile, line))
{
std::istringstream linestream (line);
std::string token;
std::vector<std::string> row;
while (std::getline (linestream, token, '\t')
{
row.push_back (token);
}
m_data_vec.push_back (row);
}

Actually, your example is easy to parse without a stringstream, if you
use a struct to represent a line, with appropriate member data types:

struct data_rec
{
double foo, bar;
std::string baz;
};

std::vector<data_rec> m_data_vec;
data_rec linedata;
while ((infile >> linedata.foo >> linedata.bar)
&& std::getline (infile, linedata.baz))
{
m_data_vec.push_back (linedata);
}

--
Jon Bell <jtbellm4h@presby.edu> Presbyterian College
Dept. of Physics and Computer Science Clinton, South Carolina USA

re: How to read tsv file?

"Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
news:bvct2d$r15ue$1@ID-221354.news.uni-berlin.de...[color=blue]
>
> "Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
> news:bvcrv5$qomb8$1@ID-221354.news.uni-berlin.de...[color=green]
> >
> > "BCC" <a@b.c> wrote in message
> > news:p1lSb.7633$uM.3791@newssvr29.news.prodigy.com ...[color=darkred]
> > > Hi,
> > >
> > > I have a tab separated value table like this:
> > > header1 header2 header3
> > > 13.455 55.3 A string
> > > 4.55 5.66 Another string
> > >
> > > I want to load this guy into a vector of vectors, since I do not know[/color][/color][/color]
how[color=blue][color=green][color=darkred]
> > > long it may be. I think I have to have a vector of vectors of[/color][/color][/color]
strings, and[color=blue][color=green][color=darkred]
> > > then extract the doubles later(?):
> > > std::vector<std::vector<std::string> > m_data_vec;
> > >
> > > I started off with this skeletal function, but Im not sure how to[/color][/color][/color]
parse the[color=blue][color=green][color=darkred]
> > > line for tabs and newlines, and stuff the elements into the vector.[/color][/color][/color]
Is it[color=blue][color=green][color=darkred]
> > > better to read in the whole line then parse it? Can I parse it on the[/color][/color][/color]
fly?[color=blue][color=green][color=darkred]
> > > How?
> > >
> > > void MyClass::ReadTSV(const* filename)
> > > {
> > > using namespace std;
> > >
> > > ifstream infile(filename);
> > > if (!infile) {
> > > cout << "unable to load file" << endl;
> > > }
> > >
> > > // Now what?
> > > }[/color]
> > May be this gives you the basic idea.
> > I haven't tested it. Also no checks for errors etc.
> >
> > <UNTESTED CODE>
> >
> > #include <fstream>
> > #include <string>
> > #include <vector>
> > using namespace std;
> >
> > void ReadTSV(const char* filename)
> > {
> > using namespace std;
> >
> > ifstream infile(filename);
> > if (!infile) {
> > cout << "unable to load file" << endl;
> > }
> > string str;
> >
> > vector<vector<string> > vvStr;
> > vector<string> vStr;
> > int pos1, pos2;
> > while (getline(infile, str))
> > {
> > pos1 = 0;
> > while((pos2 = str.find('\t'))!= string::npos)
> > {
> > vStr.push_back(str.substr(pos1, pos2));[/color]
>
> oops..second parameter should be pos2-pos1+1 i guess.
>[/color]

There is even an easier way to obtain the vStr vector using stringstreams:


template <class T>
std::vector<T> StringToVector( const std::string& Str )
{
std::istringstream iss( Str );
return std::vector<T>( std::istream_iterator<T>(iss),
std::istream_iterator<T>() );
}

[OT]
Using VC++ 6.0 this solution has to be altered a little bit using copy and a
back_inserter 'cause the appropriate ctor of vector is not yet available in
that compiler version.

Regards
Chris

re: How to read tsv file?

"Chris Theis" <Christian.Theis@nospam.cern.ch> wrote in message
news:bvd698$1k9$1@sunnews.cern.ch...[color=blue]
>
> "Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
> news:bvct2d$r15ue$1@ID-221354.news.uni-berlin.de...[color=green]
> >
> > "Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
> > news:bvcrv5$qomb8$1@ID-221354.news.uni-berlin.de...[color=darkred]
> > >
> > > "BCC" <a@b.c> wrote in message
> > > news:p1lSb.7633$uM.3791@newssvr29.news.prodigy.com ...
> > > > Hi,
> > > >
> > > > I have a tab separated value table like this:
> > > > header1 header2 header3
> > > > 13.455 55.3 A string
> > > > 4.55 5.66 Another string
> > > >
> > > > I want to load this guy into a vector of vectors, since I do not know[/color][/color]
> how[color=green][color=darkred]
> > > > long it may be. I think I have to have a vector of vectors of[/color][/color]
> strings, and[color=green][color=darkred]
> > > > then extract the doubles later(?):
> > > > std::vector<std::vector<std::string> > m_data_vec;
> > > >
> > > > I started off with this skeletal function, but Im not sure how to[/color][/color]
> parse the[color=green][color=darkred]
> > > > line for tabs and newlines, and stuff the elements into the vector.[/color][/color]
> Is it[color=green][color=darkred]
> > > > better to read in the whole line then parse it? Can I parse it on the[/color][/color]
> fly?[color=green][color=darkred]
> > > > How?
> > > >
> > > > void MyClass::ReadTSV(const* filename)
> > > > {
> > > > using namespace std;
> > > >
> > > > ifstream infile(filename);
> > > > if (!infile) {
> > > > cout << "unable to load file" << endl;
> > > > }
> > > >
> > > > // Now what?
> > > > }
> > > May be this gives you the basic idea.
> > > I haven't tested it. Also no checks for errors etc.
> > >
> > > <UNTESTED CODE>
> > >
> > > #include <fstream>
> > > #include <string>
> > > #include <vector>
> > > using namespace std;
> > >
> > > void ReadTSV(const char* filename)
> > > {
> > > using namespace std;
> > >
> > > ifstream infile(filename);
> > > if (!infile) {
> > > cout << "unable to load file" << endl;
> > > }
> > > string str;
> > >
> > > vector<vector<string> > vvStr;
> > > vector<string> vStr;
> > > int pos1, pos2;
> > > while (getline(infile, str))
> > > {
> > > pos1 = 0;
> > > while((pos2 = str.find('\t'))!= string::npos)
> > > {
> > > vStr.push_back(str.substr(pos1, pos2));[/color]
> >
> > oops..second parameter should be pos2-pos1+1 i guess.
> >[/color]
>
> There is even an easier way to obtain the vStr vector using stringstreams:
>
>
> template <class T>
> std::vector<T> StringToVector( const std::string& Str )
> {
> std::istringstream iss( Str );
> return std::vector<T>( std::istream_iterator<T>(iss),
> std::istream_iterator<T>() );
> }[/color]

How do you take care of the '\t' in the string?

re: How to read tsv file?

On Fri, 30 Jan 2004 16:20:30 +0530 in comp.lang.c++, "Sharad Kala"
<no.spam_sharadk_ind@yahoo.com> was alleged to have written:[color=blue][color=green]
>> template <class T>
>> std::vector<T> StringToVector( const std::string& Str )
>> {
>> std::istringstream iss( Str );
>> return std::vector<T>( std::istream_iterator<T>(iss),
>> std::istream_iterator<T>() );
>> }[/color]
>
>How do you take care of the '\t' in the string?[/color]

istream_iterator<T> uses T's operator>> which in turn recognizes any
kind of whitespace as a delimiter.

re: How to read tsv file?

"Sharad Kala" <no.spam_sharadk_ind@yahoo.com> wrote in message
news:bvdcko$r7lmj$1@ID-221354.news.uni-berlin.de...
[SNIP]> >[color=blue][color=green]
> > There is even an easier way to obtain the vStr vector using[/color][/color]
stringstreams:[color=blue][color=green]
> >
> >
> > template <class T>
> > std::vector<T> StringToVector( const std::string& Str )
> > {
> > std::istringstream iss( Str );
> > return std::vector<T>( std::istream_iterator<T>(iss),
> > std::istream_iterator<T>() );
> > }[/color]
>
> How do you take care of the '\t' in the string?
>[/color]

This should be done by the istream_iterators (at least in the Dinkumware
implementation used under VC++). However, I did not yet try it under another
compiler like g++.

Cheers
Chris