Escaping a C++ string

Question

What's the easiest way to convert a C++ std::string to another std::string, which has all the unprintable characters escaped?

For example, for the string of two characters [0x61,0x01], the result string might be "a\x01" or "a%01".

The easiest way to escape is through the emergency hatch right beside the `\0`. — FrustratedWithFormsDesigner, Mar 10 '10 at 14:31

score 11 · Accepted Answer · answered Mar 10 '10 at 14:47

Take a look at the Boost's String Algorithm Library. You can use its is_print classifier (together with its operator! overload) to pick out nonprintable characters, and its find_format() functions can replace those with whatever formatting you wish.

#include <iostream>
#include <boost/format.hpp>
#include <boost/algorithm/string.hpp>

struct character_escaper
{
    template<typename FindResultT>
    std::string operator()(const FindResultT& Match) const
    {
        std::string s;
        for (typename FindResultT::const_iterator i = Match.begin();
             i != Match.end();
             i++) {
            s += str(boost::format("\\x%02x") % static_cast<int>(*i));
        }
        return s;
    }
};

int main (int argc, char **argv)
{
    std::string s("a\x01");
    boost::find_format_all(s, boost::token_finder(!boost::is_print()), character_escaper());
    std::cout << s << std::endl;
    return 0;
}

score 9 · Answer 2 · answered Mar 10 '10 at 15:01

Assumes the execution character set is a superset of ASCII and CHAR_BIT is 8. For the OutIter pass a back_inserter (e.g. to a vector<char> or another string), ostream_iterator, or any other suitable output iterator.

template<class OutIter>
OutIter write_escaped(std::string const& s, OutIter out) {
  *out++ = '"';
  for (std::string::const_iterator i = s.begin(), end = s.end(); i != end; ++i) {
    unsigned char c = *i;
    if (' ' <= c and c <= '~' and c != '\\' and c != '"') {
      *out++ = c;
    }
    else {
      *out++ = '\\';
      switch(c) {
      case '"':  *out++ = '"';  break;
      case '\\': *out++ = '\\'; break;
      case '\t': *out++ = 't';  break;
      case '\r': *out++ = 'r';  break;
      case '\n': *out++ = 'n';  break;
      default:
        char const* const hexdig = "0123456789ABCDEF";
        *out++ = 'x';
        *out++ = hexdig[c >> 4];
        *out++ = hexdig[c & 0xF];
      }
    }
  }
  *out++ = '"';
  return out;
}

I thought && was a perfectly good operator. You can even use it without needing an extra header file. — Ben Voigt, Mar 10 '10 at 15:10
You can use *and* without a header in standard C++ too. This was copied from another project and I forgot to change those to make up for MSVC's deficiencies. — , Mar 10 '10 at 15:21
Why did you write in such a way that it requires a `back_inserter` to be passed in? Isn't simply returning a string by value (which means moving it anyway) just fine? — notadam, May 01 '16 at 14:11

Scindix · Answer 3 · 2019-03-21T13:42:59.557

7

Assuming that "easiest way" means short and yet easily understandable while not depending on any other resources (like libs) I would go this way:

#include <cctype>
#include <sstream>

// s is our escaped output string
std::string s = "";
// loop through all characters
for(char c : your_string)
{
    // check if a given character is printable
    // the cast is necessary to avoid undefined behaviour
    if(isprint((unsigned char)c))
        s += c;
    else
    {
        std::stringstream stream;
        // if the character is not printable
        // we'll convert it to a hex string using a stringstream
        // note that since char is signed we have to cast it to unsigned first
        stream << std::hex << (unsigned int)(unsigned char)(c);
        std::string code = stream.str();
        s += std::string("\\x")+(code.size()<2?"0":"")+code;
        // alternatively for URL encodings:
        //s += std::string("%")+(code.size()<2?"0":"")+code;
    }
}

edited Mar 21 '19 at 13:42

answered Jul 08 '17 at 17:41

Scindix

1,022
12
27

I like this answer a lot, despite the fiddling with stringstream, std::hex, and multiple casts. Would something like `char hex[5] = ""; ssize_t len = snprintf(hex, 5, "\\x%02x", c); s += std::string(hex, len);` in the `else` block work, too, or is there some gotcha I'm not seeing? – joshtch Jun 21 '18 at 18:33
1

The OP didn't specifically ask for a pure C++ solution, but I thought it sounded like he preferred C++. And so I limited myself to that. But yes, as far as I'm concerned your code would work just the same. (And would be a bit shorter.) – Scindix Jun 24 '18 at 15:06
1

There is a bug in the solution. The isprint method expects an int >-1 and <= 255, you run into problems for characters with ASCII code >127, meaning a negative `char c` – Tom Mar 16 '19 at 23:24
1

Here is my version: https://gist.github.com/timmi-on-rails/173c496a9c5a33ad9df7c6428b9a077b – Tom Mar 17 '19 at 09:55
1

@Tom You are right. I correctly casted the character in the stringstream section, but not in inside `isprint`. I missed that despite testing higher ASCII codes due to the fact that gcc's implementation always seems to return false unless it's a printable character. Nevertheless undefined behavior is evil and so I corrected my original code. – Scindix Mar 21 '19 at 13:54
@Scindix In windows with MSVC2017 an assertion window pops up. – Tom Mar 22 '19 at 14:59

score 3 · Answer 4 · answered Mar 10 '10 at 14:29

3

One person's unprintable character is another's multi-byte character. So you'll have to define the encoding before you can work out what bytes map to what characters, and which of those is unprintable.

answered Mar 10 '10 at 14:29

Douglas Leeder

49,001
8
86
133

score 2 · Answer 5 · answered Mar 11 '10 at 00:17

2

Have you seen the article about how to Generate Escaped String Output Using Spirit.Karma?

answered Mar 11 '10 at 00:17

hkaiser

11,092
1
28
33

Escaping a C++ string

5 Answers5

Linked

Related