| 1 | #include <string>
|
|---|
| 2 | #include <iostream>
|
|---|
| 3 | #include <cstring>
|
|---|
| 4 |
|
|---|
| 5 | using namespace std;
|
|---|
| 6 |
|
|---|
| 7 | #include <boost/regex/icu.hpp>
|
|---|
| 8 | using namespace boost;
|
|---|
| 9 |
|
|---|
| 10 |
|
|---|
| 11 | void do_stuff(const string & in, const string & re, bool unicode)
|
|---|
| 12 | {
|
|---|
| 13 | if(unicode) {
|
|---|
| 14 | std::cout << "DOING UNICODE:" << std::endl;
|
|---|
| 15 |
|
|---|
| 16 | UnicodeString str(in.c_str());
|
|---|
| 17 | u32regex reg = make_u32regex(re);
|
|---|
| 18 |
|
|---|
| 19 |
|
|---|
| 20 | u32regex_iterator<const UChar*>
|
|---|
| 21 | i(make_u32regex_iterator(str, reg, match_partial | match_perl)),
|
|---|
| 22 | j;
|
|---|
| 23 |
|
|---|
| 24 | while(i != j) {
|
|---|
| 25 | std::cout << "got a " << ((*i)[0].matched ? "" : "partial ")
|
|---|
| 26 | << "match legth: " << (*i)[0].length() << " match position: " << i->position()
|
|---|
| 27 | << " matched: " << string( (*i)[0].first, (*i)[0].second ) << std::endl;
|
|---|
| 28 | ++i;
|
|---|
| 29 | }
|
|---|
| 30 | } else {
|
|---|
| 31 | std::cout << "DOING ASCII:" << std::endl;
|
|---|
| 32 |
|
|---|
| 33 | string str = in;
|
|---|
| 34 | regex reg(re);
|
|---|
| 35 |
|
|---|
| 36 | regex_iterator<string::const_iterator>
|
|---|
| 37 | i(make_regex_iterator(str, reg, match_partial | match_perl)),
|
|---|
| 38 | j;
|
|---|
| 39 |
|
|---|
| 40 | while(i != j) {
|
|---|
| 41 | std::cout << "got a " << ((*i)[0].matched ? "" : "partial ")
|
|---|
| 42 | << "match legth: " << (*i)[0].length() << " match position: " << i->position()
|
|---|
| 43 | << " matched: " << string( (*i)[0].first, (*i)[0].second ) << std::endl;
|
|---|
| 44 | ++i;
|
|---|
| 45 | }
|
|---|
| 46 | }
|
|---|
| 47 | }
|
|---|
| 48 |
|
|---|
| 49 |
|
|---|
| 50 | int main(int argc, char * argv[]) {
|
|---|
| 51 | do_stuff("in summary in math we are using sum", "summary", true);
|
|---|
| 52 | do_stuff("in summary in math we are using sum", "summary", false);
|
|---|
| 53 |
|
|---|
| 54 |
|
|---|
| 55 | return 0;
|
|---|
| 56 | }
|
|---|