#include <stdio.h>
#include <string.h>
#include <stdlib.h>

// http://www.cplusplus.com/reference/cstdio/scanf/

// POSIX regex, http://stackoverflow.com/questions/1085083/regular-expressions-in-c-examples
// GNU, http://www.gnu.org/software/libc/manual/html_node/Regular-Expressions.html
// PCRE - Perl Compatible Regular Expressions, http://www.pcre.org/
// Super Light Regular Expression library, http://slre.sourceforge.net/


int main()
{
    unsigned int uivalue;
    int i, ivalue, x, y;
    double dvalue;
    char c, buf[100]="zz", buf1[100], buf2[100];

// skip to the end of the line
    strcpy(buf, "1234 pqewiurqweir adsfads adskmcxvnz\nsecond line");
    sscanf(buf, "%d%*[^\n]\n%s", &ivalue, buf1); printf("1: [%d][%s]\n", ivalue, buf1);

    strcpy(buf, "   123 /* hello 976d/asfhvk */ 456 ");
    sscanf(buf, "%d /* %*[^*]*/%d", &x, &y);  // do not accept    /*    *    */
    printf("2: x=[%d] y=[%d]\n", x, y);

// equivalent to %c, equivalent to space (skip all white spaces)
    strcpy(buf, "  a");
    sscanf(buf, " %[^ \t\n]", &c); printf("3: c=[%c]\n", c);
    sscanf(buf, "%*[ \t\n]%[^ \t\n]", &c); printf("3: c=[%c]\n", c);

// equivalent to %s
    strcpy(buf, "   hello123?-& world\n   again\n");
    sscanf(buf, " %[^ \t\n]", buf1); printf("4: buf1=[%s]\n", buf1);
    sscanf(buf, "%*[ \t\n]%[^ \t\n]", buf1); printf("5: buf1=[%s]\n", buf1);

// equivalent to gets()
    sscanf(buf, "%[^\n]\n%s", buf1, buf2); printf("6: buf1=[%s] buf2=[%s]\n", buf1, buf2);

// equivalent to fgets()
    sscanf(buf, "%[^\n]\n%s", buf1, buf2); strcat(buf1, "\n");
    printf("7: buf1=[%s] buf2=[%s]\n", buf1, buf2);

// equivalent to %d  note that - and + is not restricted as the first char and occurs once
    strcpy(buf, "   -1234    ");
    sscanf(buf, " %[-+0-9]", buf1); 
    ivalue = atoi(buf1); printf("8: buf1=[%s] ivalue = %d\n", buf1, ivalue);

    // " %1[-+1-9]%[0-9]", buf1, &buf1[1]

    strcpy(buf, "   +1234    ");
    sscanf(buf, "%*[ \t\n]%[-+0-9]", buf1); 
    ivalue = atoi(buf1); printf("9: buf1=[%s] ivalue = %d\n", buf1, ivalue);

// equivalent to %o
    strcpy(buf, "   012345670888   ");
    sscanf(buf, "%*[ \t\n]%[0-7]", buf1); 
    for (uivalue=i=0; i<strlen(buf1); i++)
        uivalue = uivalue * 8 + (buf1[i]-'0');
    printf("10: buf1=[%s] uivalue = %u\n", buf1, uivalue);
    
// equivalent to %x
    strcpy(buf, "   a1790dfGG   ");
    sscanf(buf, "%*[ \t\n]%[0-9A-Fa-f]", buf1); 
    for (uivalue=i=0; i<strlen(buf1); i++)
        uivalue = uivalue * 16 + 
                  (buf1[i]>='A' ? 
                      (buf1[i]>='a'? buf1[i]-'a' : buf1[i]-'A')+10 : 
                      buf1[i]-'0');
    printf("11: buf1=[%s] uivalue = %d (%x)\n", buf1, uivalue, uivalue);
    
// binary string
    strcpy(buf, "   01101011001   ");
    sscanf(buf, "%*[ \t\n]%[01]", buf1); 
    for (uivalue=i=0; i<strlen(buf1); i++)
        uivalue = uivalue * 2 + (buf1[i]-'0');
    printf("12: buf1=[%s] uivalue = %u\n", buf1, uivalue);
    
// double, float
    strcpy(buf, "   -1234.5678   ");
    sscanf(buf, "%*[ \t\n]%[-+0-9]", buf1); 
    printf("13: buf1=[%s] uivalue = %u\n", buf1, uivalue);

// scientific
    strcpy(buf, "   1234.5678e+10   ");
    printf("14: buf1=[%s] uivalue = %u\n", buf1, uivalue);

// field extration
    char name[20], tel[20];
    int age;
    strcpy(buf, "name:john age:40 tel:082-313530");
    sscanf(buf, "%*[^:]:%s %*[^:]:%d %*[^:]:%s", name, &age, tel);
    printf("15: %s %d %s\n", name, age, tel);

// URL parsing
    char protocol[10], site[50], path[50];
    sscanf("http://ccckmit.wikidot.com/cp/list/hello.txt", 
           "%[^:]:%*2[/]%[^/]/%[a-zA-Z0-9._/-]", 
           protocol, site, path);
    printf("16: protocol=%s site=%s path=%s\n", protocol, site, path);

// ID
    char ID_c[2], ID_numeric[20];
    i = sscanf("   Z1234567890123 ",
               "%*[ \t\n]%1[a-zA-Z]%9[0-9]", 
               ID_c, ID_numeric); // ID_c should be an array
    printf("17: ID is %s%s (%d)\n", ID_c, ID_numeric, i);

    strcpy(ID_numeric, "---------");
    i = sscanf("   ZA123456789 ",
               "%*[ \t\n]%1[a-zA-Z]%9[0-9]", 
               ID_c, ID_numeric);
    printf("17: ID is %s%s (%d)\n", ID_c, ID_numeric, i);

// all characters after 'a' before 'n'  (note. the first occurrence of 'n')
    strcpy(buf, "this is a test string extraction");
    sscanf(buf, "%*[^a]a%[^n]", buf1);
    printf("18: buf1=[%s]\n", buf1);

// extrating characters between '(' and ')' and skip all others
    strcpy(buf, "seven eight (nine) ten\n");
    i = sscanf(buf, "%*[^(](%[^)])%*[^\n]\n", buf1);
    printf("19a: buf1=[%s] (%d)\n", buf1, i);

    int start=0, end=0;
    i = sscanf(buf, "%*[^(](%n%*[^)]%n)%*[^\n]\n", &start, &end);
    strncpy(buf1, &buf[start], end-start); buf1[end-start] = 0;
    printf("19b: start=%d end=%d buf1=[%s] (%d)\n", start, end, buf1, i);

    strcpy(buf, "seven eight () nine");
    i = sscanf(buf, "%*[^(](%[^)])%*[^\n]\n", buf1); // fail on null string
    printf("20a: buf1=[%s] (%d) failed\n", buf1, i);

    start=end=0;
    i = sscanf(buf, "%*[^(](%n%*[^)]%n)%*[^\n]\n", &start, &end); // fail again
    if (end>=start)
    {
        strncpy(buf1, &buf[start], end-start); 
        buf1[end-start] = 0;
    }
    printf("20b: start=%d end=%d buf1=[%s] (%d) failed\n", start, end, buf1, i);

    char *startptr, *endptr;
    startptr = strchr(buf, '(');
    endptr = strchr(startptr, ')');
    for (i=0; i<endptr-startptr-1; i++)
        buf1[i] = (startptr+1)[i];
    buf1[endptr-startptr-1] = 0;
    printf("20c: start=%d end=%d buf1=[%s] success\n", startptr-buf, endptr-buf, buf1);

    return 0;
}