/*
*	uniqx ver.1.0
*	date: 2005/02/03
*
*	Kenar (Kenji Arisaw)
*	E-mail: arisawa@aichi-u.ac.jp
*/

#include <u.h>
#include <libc.h>
#include <bio.h>
#include <ctype.h>

#define gline() Brdstr(&in, '\n', 1)

#define MFIELD 128

typedef struct Ranges Ranges;
struct Ranges{
	int	n;
	int m;
	Ranges	*next;
};

Ranges *ranges;

Biobuf	in;
Biobuf	out;

int count=0;
int cflag=0;
int uflag=0;
int dflag=0;
int rflag=0;
int fflag=0;

void pline(char *bp);
int	C_equal(Ranges *ranges, char *b1, char *b2);
int	F_equal(Ranges *ranges, char *b1, char *b2);
int	(*equal)(Ranges *ranges, char *b1, char *b2);

void
appendRange(Ranges **r, int n, int m)
{
	Ranges *p, *q;
	q = nil;
	p = *r;
	while(p){
		q = p;
		p = p->next;
	}
	// then p is nil
	p = malloc(sizeof(Ranges));
	if(q)
		q ->next = p;
	else
		*r = p;
	p->n = n;
	p->m = m;
	p->next = nil;
}


void
usage(void)
{
	fprint(2,"usage:\n\
	uniq [-udcrf] [-k key,key,...] [file]\n\
	-udc flag is same as that of uniq.\n\
	-r flag is `print last line of duplicated line'\n\
	-f flag means field wise comparison.\n\
	key follows Python syntax of range;\n\
	that is, one of `n:m' `:m' `n:' `:'\n\
	n begins with 0\n\
	defaut is: uniqx -k 0:\n");
	exits("usage");
}

void
main(int argc, char *argv[])
{
	int fd,n,m;
	char *file;
	char *keys=nil;
	char *s,c;
	char *range[2];
	int nrange;
	char *b1;
	char *b2;

	ARGBEGIN{
	case 'c':
		cflag = 1;
		break;
	case 'd':
		dflag = 1;
		break;
	case 'f':
		fflag = 1;
		break;
	case 'k':
		keys=ARGF();
		if(keys == nil)
			usage();
		break;
	case 'r':
		rflag = 1;
		break;
	case 'u':
		uflag = 1;
		break;
	default:
		usage();
	}ARGEND

	if(uflag && dflag)
		sysfatal("# incompatible flag -u and -d");

	if(keys == nil)
		keys = "0:";

	equal = C_equal;
	if(fflag)
		equal = F_equal;

	for(;;){
		for(s = keys; *s && *s != ','; s++);
		c = *s;
		if(*s == ',')
			*s = 0;
		nrange = getfields(keys,range,2,0,":");
		n = 0;
		m = 0;
		if(nrange == 1){
			n = atoi(range[0]);
			m = n + 1;
		}
		if(nrange == 2){
				n = atoi(range[0]);
				m = atoi(range[1]);
		}
		if(n < 0)
			sysfatal("# field out of range");
		appendRange(&ranges, n,m);
		if(c == 0)
			break;
		keys = ++s;
	}

	file = *argv++;
	if(file && *argv != nil)
		usage();
	fd = 0;
	if(file){
		fd = open(file, OREAD);
		if(fd < 0)
			sysfatal("# cannot open %s: %r\n", *argv);
	}

	Binit(&in, fd, OREAD);
	Binit(&out, 1, OWRITE);

	if((b1 = gline()) == nil)
		exits(0);
	b1 = strdup(b1);
	count = 1;
	for(;;){
		if((b2 = gline()) == nil) {
			pline(b1);
			exits(0);
		}
		if(!equal(ranges, b1, b2)){
			pline(b1);
			free(b1);
			b1 = strdup(b2);
			count = 1;
			continue;
		}
		if(rflag){
			free(b1);
			b1 = strdup(b2);
		}
		count++;
	}
}

void
pline(char *bp)
{
	if(uflag && count != 1)
		return;
	if(dflag && count == 1)
		return;
	if(cflag)
		Bprint(&out, "%4d ", count);
	Bprint(&out, "%s\n", bp);
}

int
F_equal(Ranges *ranges, char *b1, char *b2)
{
	char *tb1[MFIELD], *tb2[MFIELD];
	int n1,n2,i,n,m,min,max;
	char *bb1;
	char *bb2;
	int status;

	bb1 = strdup(b1);
	bb2 = strdup(b2);
	n1 = tokenize(bb1, tb1, MFIELD);
	n2 = tokenize(bb2, tb2, MFIELD);
	min = (n1<n2)?n1:n2;
	max = (n1<n2)?n2:n1;
	if(max == MFIELD)
		sysfatal("# fields over limit %d", MFIELD);
	while(ranges){
		n = ranges->n;
		m = ranges->m;
		if(m == 0 || m > max)
			m = max;
		if(n < min){
			for(i=n; i < m; i++){
				if(strcmp(tb1[i], tb2[i]) != 0){
					status = 0;
					goto L1;
				}
			}
		}
		else if(n < max){
			status = 0;
			goto L1;
		}
		ranges = ranges->next;
	}
	status = 1;
L1:	free(bb1); free(bb2);
	return status;
}

int
C_equal(Ranges *ranges, char *b1, char *b2)
{
	int n1,n2,min,max, n, m;
	n1 = strlen(b1);
	n2 = strlen(b2);
	min = (n1<n2)?n1:n2;
	max = (n1<n2)?n2:n1;
	while(ranges){
		n = ranges->n;
		m = ranges->m;
		if(m == 0 || m > max)
			m = max;
		if(n < min){
			if(strncmp(b1+n,b2+n,m-n) != 0)
				return 0;
		}
		else if(n < max)
			return 0;
		ranges = ranges->next;
	}
	return 1;
}