std::search question...

[Dave Townsend]

Hi,

Can anybody help me with the following piece of code?

The purpose behind the code is to parse HTML files, strip out
the tags and return the text between tags. This is part of a larger
application which will perform "searches" for text values in a
directory of html files, trying to match only the non-tagged text
in the documents.

My approach was to customize a istreambuf_iterator which would
iterate over the text characters in the HTML bodies, and ignore the html
tags. I did
this by deriving from the std istreambuf_iterator, overriding the ++
operators to parse the underlying input to screen the data and return the
next valid character. The iteration seems to work ok on some simple
HTML test data - I've seen the tags stripped away and the text body get
through.

Now here's the clever part . I thought I could use the std algorithm
search
directly, plug in the range of the stream and the range of the text string
to
look for, and there you go. Unfortunately, this doesn't work ! The search
always fails to find the search string.

Am I missing something about the search? I thought the requirements for
search only required the iterators be forward iterators - is there something
I've
misunderstood here ? I'd like to use this code, it strikes me as a nice
solution
utilizing the power of STL and the stream classes.

Many thanks for any help!

dave.

#include <algorithm>
#include <fstream>
#include <istream>
#include <iostream>
#include <algorithm>
#include <vector>
using namespace std;

class htmlstreambuf_iterator : public istreambuf_iterator<char>
{
public:
htmlstreambuf_iterator( ifstream& stream)
:istreambuf_iterator<char>(stream),_intag(false)
{
// advance to first "legal" character in input.
while(
!istreambuf_iterator<char>::equal(istreambuf_iterator<char>()) )
{

char c = istreambuf_iterator<char>:perator*();
if ( c == '<' )
{
_intag=true;
incr();
break;

}
else if (!_intag )
{

break;

}

}

}
htmlstreambuf_iterator( )
:istreambuf_iterator<char>(), _intag(false)
{}
htmlstreambuf_iterator( const htmlstreambuf_iterator& iter)
:istreambuf_iterator<char>( iter), _intag(false)
{}

char operator*()
{
printf("operator*() %c\n", istreambuf_iterator<char>:perator*() );
return istreambuf_iterator<char>:perator*();

}

const char* operator->()
{
return istreambuf_iterator<char>:perator->();
}




htmlstreambuf_iterator& operator++()
{
incr();
return *this;
}
htmlstreambuf_iterator operator++(int)
{
htmlstreambuf_iterator tmp=*this;
incr();
return tmp;

}

bool equal( htmlstreambuf_iterator& rhs )
{
return istreambuf_iterator<char>::equal(rhs);
}

private:
void incr()
{
while( !istreambuf_iterator<char>::equal(istreambuf_iterator<char>()) )
{
// advance and skip over any tag data to next legal character.
istreambuf_iterator<char>:perator++();
char c = istreambuf_iterator<char>:perator*();
if ( c == '<' && !_intag )
{
_intag = true;

}
else if ( c == '>' && _intag )
{
_intag = false;
}
else if (!_intag )
{
break;
}

}

}
private:

bool _intag;

};



int main(int argc, char* argv[])
{
ifstream input( argv[1] );

htmlstreambuf_iterator iter(input);
htmlstreambuf_iterator iterend;

string searchtext("D");
htmlstreambuf_iterator pos = search( iter, iterend,

searchtext.begin(), searchtext.end() );

// pos always points to end of input. !!!
bool found = pos != iterend ;


return 0;
}

 

[Dietmar Kuehl]

My approach was to customize a istreambuf_iterator which would
iterate over the text characters in the HTML bodies, and ignore the html
tags. I did
this by deriving from the std istreambuf_iterator, overriding the ++
operators to parse the underlying input to screen the data and return the
next valid character.

You would be better off wrapping 'istreambuf_iterator' into a custom
iterator class as this class is not intended to be derived from. In
particular, you did not override 'operator++()' but overload this
operator: 'std::istreambuf_iterator' does not have any virtual
functions
and thus you cannot override anything. A class using public inheritance
from 'std::istreambuf_iterator' is rather error prone and subject e.g.
to slicing (e.g. when being passed to a function taking an
'std::istreambuf_iterator' as argument).

Am I missing something about the search?

Yes, you do: 'std::search()' takes forward iterators, not input
iterators. However, 'std::istreambuf_iterator' is an input iterator.
It is not too hard to create a wrapper for 'std::istreambuf_iterator'
which allows forward iteration (in particular allows the possibility
to recover to some position).

I thought the requirements for
search only required the iterators be forward iterators - is there something
I've misunderstood here?

Your misunderstanding seems to be with 'std::istreambuf_iterator':
this is not a forward iterator. It is an input iterator. The key
difference is that input iterators only support single pass while
forward iterators allow multiple passed (plus the possibility to
write to the sequence which is not needed in this case).