Case insensitive std::string.find()

I am using std::string's find() method to test if a string is a substring of another. Now I need case insensitive version of the same thing. For string comparison I c开发者_JAVA百科an always turn to stricmp() but there doesn't seem to be a stristr().

I have found various answers and most suggest using Boost which is not an option in my case. Additionally, I need to support std::wstring/wchar_t. Any ideas?

You could use std::search with a custom predicate.

#include <locale>
#include <iostream>
#include <algorithm>
using namespace std;

// templated version of my_equal so it could work with both char and wchar_t
template<typename charT>
struct my_equal {
    my_equal( const std::locale& loc ) : loc_(loc) {}
    bool operator()(charT ch1, charT ch2) {
        return std::toupper(ch1, loc_) == std::toupper(ch2, loc_);
    }
private:
    const std::locale& loc_;
};

// find substring (case insensitive)
template<typename T>
int ci_find_substr( const T& str1, const T& str2, const std::locale& loc = std::locale() )
{
    typename T::const_iterator it = std::search( str1.begin(), str1.end(), 
        str2.begin(), str2.end(), my_equal<typename T::value_type>(loc) );
    if ( it != str1.end() ) return it - str1.begin();
    else return -1; // not found
}

int main(int arc, char *argv[]) 
{
    // string test
    std::string str1 = "FIRST HELLO";
    std::string str2 = "hello";
    int f1 = ci_find_substr( str1, str2 );

    // wstring test
    std::wstring wstr1 = L"ОПЯТЬ ПРИВЕТ";
    std::wstring wstr2 = L"привет";
    int f2 = ci_find_substr( wstr1, wstr2 );

    return 0;
}

The new C++11 style:

#include <algorithm>
#include <string>
#include <cctype>

/// Try to find in the Haystack the Needle - ignore case
bool findStringIC(const std::string & strHaystack, const std::string & strNeedle)
{
  auto it = std::search(
    strHaystack.begin(), strHaystack.end(),
    strNeedle.begin(),   strNeedle.end(),
    [](unsigned char ch1, unsigned char ch2) { return std::toupper(ch1) == std::toupper(ch2); }
  );
  return (it != strHaystack.end() );
}

Explanation of the std::search can be found on cplusplus.com.

why not use Boost.StringAlgo:

#include <boost/algorithm/string/find.hpp>

bool Foo()
{
   //case insensitive find

   std::string str("Hello");

   boost::iterator_range<std::string::const_iterator> rng;

   rng = boost::ifind_first(str, std::string("EL"));

   return rng;
}

Why not just convert both strings to lowercase before you call find()?

tolower

Notice:

• Inefficient for long strings.
• Beware of internationalization issues.

Since you're doing substring searches (std::string) and not element (character) searches, there's unfortunately no existing solution I'm aware of that's immediately accessible in the standard library to do this.

Nevertheless, it's easy enough to do: simply convert both strings to upper case (or both to lower case - I chose upper in this example).

std::string upper_string(const std::string& str)
{
    string upper;
    transform(str.begin(), str.end(), std::back_inserter(upper), toupper);
    return upper;
}

std::string::size_type find_str_ci(const std::string& str, const std::string& substr)
{
    return upper(str).find(upper(substr) );
}

This is not a fast solution (bordering into pessimization territory) but it's the only one I know of off-hand. It's also not that hard to implement your own case-insensitive substring finder if you are worried about efficiency.

Additionally, I need to support std::wstring/wchar_t. Any ideas?

tolower/toupper in locale will work on wide-strings as well, so the solution above should be just as applicable (simple change std::string to std::wstring).

[Edit] An alternative, as pointed out, is to adapt your own case-insensitive string type from basic_string by specifying your own character traits. This works if you can accept all string searches, comparisons, etc. to be case-insensitive for a given string type.

If you want “real” comparison according to Unicode and locale rules, use ICU’s Collator class.

Also make sense to provide Boost version: This will modify original strings.

#include <boost/algorithm/string.hpp>

string str1 = "hello world!!!";
string str2 = "HELLO";
boost::algorithm::to_lower(str1)
boost::algorithm::to_lower(str2)

if (str1.find(str2) != std::string::npos)
{
    // str1 contains str2
}

or using perfect boost xpression library

#include <boost/xpressive/xpressive.hpp>
using namespace boost::xpressive;
....
std::string long_string( "very LonG string" );
std::string word("long");
smatch what;
sregex re = sregex::compile(word, boost::xpressive::icase);
if( regex_match( long_string, what, re ) )
{
    cout << word << " found!" << endl;
}

In this example you should pay attention that your search word don't have any regex special characters.

#include <iostream>
using namespace std;

template <typename charT>
struct ichar {
    operator charT() const { return toupper(x); }
    charT x;
};
template <typename charT>
static basic_string<ichar<charT> > *istring(basic_string<charT> &s) { return (basic_string<ichar<charT> > *)&s; }
template <typename charT>
static ichar<charT> *istring(const charT *s) { return (ichar<charT> *)s; }

int main()
{
    string s = "The STRING";
    wstring ws = L"The WSTRING";
    cout << istring(s)->find(istring("str")) << " " << istring(ws)->find(istring(L"wstr"))  << endl;
}

A little bit dirty, but short & fast.

I love the answers from Kiril V. Lyadvinsky and CC. but my problem was a little more specific than just case-insensitivity; I needed a lazy Unicode-supported command-line argument parser that could eliminate false-positives/negatives when dealing with alphanumeric string searches that could have special characters in the base string used to format alphanum keywords I was searching against, e.g., Wolfjäger shouldn't match jäger but <jäger> should.

It's basically just Kiril/CC's answer with extra handling for alphanumeric exact-length matches.

/* Undefined behavior when a non-alpha-num substring parameter is used. */
bool find_alphanum_string_CI(const std::wstring& baseString, const std::wstring& subString)
{
    /* Fail fast if the base string was smaller than what we're looking for */
    if (subString.length() > baseString.length()) 
        return false;

    auto it = std::search(
        baseString.begin(), baseString.end(), subString.begin(), subString.end(),
        [](char ch1, char ch2)
        {
            return std::toupper(ch1) == std::toupper(ch2);
        }
    );

    if(it == baseString.end())
        return false;

    size_t match_start_offset = it - baseString.begin();

    std::wstring match_start = baseString.substr(match_start_offset, std::wstring::npos);

    /* Typical special characters and whitespace to split the substring up. */
    size_t match_end_pos = match_start.find_first_of(L" ,<.>;:/?\'\"[{]}=+-_)(*&^%$#@!~`");

    /* Pass fast if the remainder of the base string where
       the match started is the same length as the substring. */
    if (match_end_pos == std::wstring::npos && match_start.length() == subString.length()) 
        return true;

    std::wstring extracted_match = match_start.substr(0, match_end_pos);

    return (extracted_match.length() == subString.length());
}

The Most Efficient Way

Simple and Fast.

Performance is guaranteed to be linear, with an initialization cost of 2 * NEEDLE_LEN comparisons. (glic)

#include <cstring>
#include <string>
#include <iostream>

int main(void) {

    std::string s1{"abc de fGH"};
    std::string s2{"DE"};

    auto pos = strcasestr(s1.c_str(), s2.c_str());

    if(pos != nullptr)
        std::cout << pos - s1.c_str() << std::endl;

    return 0;
}

wxWidgets has a very rich string API wxString

it can be done with (using the case conversion way)

int Contains(const wxString& SpecProgramName, const wxString& str)
{
  wxString SpecProgramName_ = SpecProgramName.Upper();
  wxString str_ = str.Upper();
  int found = SpecProgramName.Find(str_);
  if (wxNOT_FOUND == found)
  {
    return 0;
  }
  return 1;
}