# Determine number, i, of nonmatches, n[1..n[0]], and matches, m[1..m[0]], 
# in line, l, using criteria, k, 0 maximal, 1 trim end, 2 trim start, 3 minimal 
# nonmatches and matches are alternating and nonoverlapping 
# the first nonmatch or last match may be null

function mmatch(n,m,l,k,regexp) 
{
    i = 1;
    l1 = l;
    n[0] = 1;
    m[0] = 0;
    while (match(l1,regexp)!=0)
    { 
        # Recurse to handle minimal matches
        rs = RSTART;
        rl = RLENGTH;
        if (k%2) 
        {
            # Trim end
            l0 = substr(l1,rs,rl-1);
            while (match(l0,regexp)!=0) 
            {
                rs += RSTART - 1;
                rl = RLENGTH;
                l0 = substr(l1,rs,rl-1);
                #printf "%d %d %d: %s\n", j, RSTART, RLENGTH, l0
            }
        }
        if (k%4 > 1)
        {
            # Trim start
            l0 = substr(l1,rs+1,rl-1);
            while (match(l0,regexp)!=0) 
            {
                rs += RSTART;
                rl = RLENGTH;
                l0 = substr(l1,rs+1,rl-1);
                #printf "%d %d %d: %s\n", j, RSTART, RLENGTH, l0
            }
        }
        n[i] = substr(l1,0,rs-1); 
        m[i] = substr(l1,rs,rl);
        l1 = substr(l1,rs+rl);
        printf "%d %d: n)%s\tm)%s\n", FNR, i, n[i], m[i]
        i++;
    }
    if (length(l1)) 
    {
        n[i] = l1;
        m[i] = NULL;
 	  printf "%d %d: n)%s\tm)%s \n", FNR, i, n[i], m[i]
    }
    m[0] = i;
    n[0] = i;
    return (0);
}

# Determine number, i, and indicies of 
# nonmatches, l[n[i-1]+m[i-1]..n[i]-1], and matches, l[n[i]..n[i]+m[i]-1], 
# in line, l, using criteria, k, 0 maximal, 1 trim end, 2 trim start, 3 minimal 
# nonmatches and matches are alternating and nonoverlapping 
# the first nonmatch or last match may be null
# this allows partitioning of the line into string before, of, and after the nth match

function imatch(n,m,l,k,regexp) 
{
    i = 1;
    l1 = l;
    n[0] = 1;
    m[0] = 0;
    while (match(l1,regexp)!=0)
    { 
        # Recurse to handle minimal matches
        rs = RSTART;
        rl = RLENGTH;
        if (k%2) 
        {
            # Trim end
            l0 = substr(l1,rs,rl-1);
            while (match(l0,regexp)!=0) 
            {
                rs += RSTART - 1;
                rl = RLENGTH;
                l0 = substr(l1,rs,rl-1);
                #printf "%d %d %d: %s\n", j, RSTART, RLENGTH, l0
            }
        }
        if (k%4 > 1)
        {
            # Trim start
            l0 = substr(l1,rs+1,rl-1);
            while (match(l0,regexp)!=0) 
            {
                rs += RSTART;
                rl = RLENGTH;
                l0 = substr(l1,rs+1,rl-1);
                #printf "%d %d %d: %s\n", j, RSTART, RLENGTH, l0
            }
        }
        n[i] = n[i-1] + m[i-1] + rs - 1; 
        m[i] = rl;
        l1 = substr(l1,rs+rl);
        # indicies
        #printf "%d %d: n)%d\tm)%d\n", FNR, i, n[i], m[i]
        # nonmatches and matches
        #printf "%d %d: n)%s\tm)%s\n", FNR, i, substr(l,n[i-1]+m[i-1],n[i]-(n[i-1]+m[i-1])), substr(l,n[i],m[i])
        # nth matches
        ##printf "%d %d: n)%s\tm)%s l)%s\n", FNR, i, substr(l,1,n[i]-1), substr(l,n[i],m[i]), substr(l,n[i]+m[i])
        i++;
    }
    if (length(l1)) 
    {
        n[i] = n[i-1] + m[i-1] + length(l1);
        m[i] = 0;
        # indicies
 	  #printf "%d %d: n)%d\tm)%d \n", FNR, i, n[i], m[i]
        # nonmatches and matches
        #printf "%d %d: n)%s\tm)%s\n", FNR, i, substr(l,n[i-1]+m[i-1],n[i]-(n[i-1]+m[i-1])), substr(l,n[i],m[i])
        # nth matches
        ##printf "%d %d: n)%s\tm)%s l)%s\n", FNR, i, substr(l,1,n[i]-1), substr(l,n[i],m[i]), substr(l,n[i]+m[i])
    }
    m[0] = i;
    n[0] = i;
    return (0);
}

# Process the file
{ 
    i = 1;
    l = $0;
    k = 3;
    # Determine number, i, of nonmatches, n[1..n[0]], and matches, m[1..m[0]], 
    # in line, l, using criteria, k 0 maximal, 1 trim end, 2 trim start, 3 minimal 
    # Need string "m.*n" instead of pattern /m.*n/ for call
    #i = mmatch(n,m,l,k,"m.*n");
 
    # Lower case
    i = imatch(n,m,l,k,"<a href=\".*\">"); 
    i1 = 1;
    while (i1 < m[0] && m[i1] > 0)
    {
        # Print enclosed matches
        printf "%s %d %d: %s\n", FILENAME, FNR, i1, substr(l,n[i1]+9,m[i1]-11)
        i1 = i1 + 1;
    } 
    # Upper case
    i = imatch(n,m,l,k,"<A HREF=\".*\">"); 
    i1 = 1;
    while (i1 < m[0] && m[i1] > 0)
    {
        # Print enclosed matches
        printf "%s %d %d: %s\n", FILENAME, FNR, i1, substr(l,n[i1]+9,m[i1]-11)
        i1 = i1 + 1;
    } 
}