The R Project SVN R

Rev

Rev 25281 | Rev 29410 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 25281 Rev 25282
Line 257... Line 257...
257 
	v[i] = isDuplicated(x, i, &data);
 257 
	v[i] = isDuplicated(x, i, &data);
258 
 
 258 
 
259 
    return ans;
 259 
    return ans;
260 
}
 260 
}
261 
 
 261 
 
262 
static SEXP duplicated2(SEXP x, HashData *d)
 - 
 
263 
{
 - 
 
264 
    SEXP ans;
 - 
 
265 
    int *h, *v;
 - 
 
266 
    int i, n;
 - 
 
267 
 
 - 
 
268 
    n = LENGTH(x);
 - 
 
269 
    HashTableSetup(x, d);
 - 
 
270 
    PROTECT(d->HashTable);
 - 
 
271 
    ans = allocVector(LGLSXP, n);
 - 
 
272 
    UNPROTECT(1);
 - 
 
273 
    h = INTEGER(d->HashTable);
 - 
 
274 
    v = LOGICAL(ans);
 - 
 
275 
 
 - 
 
276 
    for (i = 0; i < d->M; i++)
 - 
 
277 
	h[i] = NIL;
 - 
 
278 
 
 - 
 
279 
    for (i = 0; i < n; i++)
 - 
 
280 
	v[i] = isDuplicated(x, i, d);
 - 
 
281 
 
 - 
 
282 
    return ans;
 - 
 
283 
}
 - 
 
284 
 
 - 
 
285 
 
 - 
 
286 
/* .Internal(duplicated(x)) [op=0]  and
 262 
/* .Internal(duplicated(x)) [op=0]  and
287 
   .Internal(unique(x))	    [op=1] :
 263 
   .Internal(unique(x))	    [op=1] :
288 
*/
 264 
*/
289 
SEXP do_duplicated(SEXP call, SEXP op, SEXP args, SEXP env)
 265 
SEXP do_duplicated(SEXP call, SEXP op, SEXP args, SEXP env)
290 
{
 266 
{
Line 927... Line 903...
927 
    UNPROTECT(2); /*HashTable, matches*/
 903 
    UNPROTECT(2); /*HashTable, matches*/
928 
    UNPROTECT(1); /*ans*/
 904 
    UNPROTECT(1); /*ans*/
929 
    return ans;
 905 
    return ans;
930 
}
 906 
}
931 
 
 907 
 
- 
 
 908 
/* returns 1-based duplicate no */
- 
 
 909 
static int isDuplicated2(SEXP x, int indx, HashData *d)
- 
 
 910 
{
- 
 
 911 
    int i, *h;
- 
 
 912 
 
- 
 
 913 
    h = INTEGER(d->HashTable);
- 
 
 914 
    i = d->hash(x, indx, d);
- 
 
 915 
    while (h[i] != NIL) {
- 
 
 916 
	if (d->equal(x, h[i], x, indx))
- 
 
 917 
	    return h[i] + 1;
932 
#include <R_ext/RS.h>
 918 
	i = (i + 1) % d->M;
- 
 
 919 
    }
- 
 
 920 
    h[i] = indx;
- 
 
 921 
    return 0;
- 
 
 922 
}
- 
 
 923 
 
- 
 
 924 
static SEXP duplicated2(SEXP x, HashData *d)
- 
 
 925 
{
- 
 
 926 
    SEXP ans;
- 
 
 927 
    int *h, *v;
- 
 
 928 
    int i, n;
- 
 
 929 
 
- 
 
 930 
    n = LENGTH(x);
- 
 
 931 
    HashTableSetup(x, d);
- 
 
 932 
    PROTECT(d->HashTable);
- 
 
 933 
    ans = allocVector(INTSXP, n);
- 
 
 934 
    UNPROTECT(1);
- 
 
 935 
    h = INTEGER(d->HashTable);
- 
 
 936 
    v = INTEGER(ans);
- 
 
 937 
    for (i = 0; i < d->M; i++) h[i] = NIL;
- 
 
 938 
    for (i = 0; i < n; i++) v[i] = isDuplicated2(x, i, d);
- 
 
 939 
    return ans;
- 
 
 940 
}
933 
 
 941 
 
- 
 
 942 
#include <R_ext/RS.h> /* for Calloc, free */
934 
 
 943 
 
935 
SEXP do_makeunique(SEXP call, SEXP op, SEXP args, SEXP env)
 944 
SEXP do_makeunique(SEXP call, SEXP op, SEXP args, SEXP env)
936 
{
 945 
{
937 
    SEXP names, sep, ans, dup, newx;
 946 
    SEXP names, sep, ans, dup, newx;
938 
    int i, n, cnt, len, maxlen = 0;
 947 
    int i, n, cnt, len, maxlen = 0, *cnts, dp;
939 
    HashData data;
 948 
    HashData data;
940 
    char *csep, *buf;
 949 
    char *csep, *buf;
941
950
942 
    checkArity(op, args);
 951 
    checkArity(op, args);
943 
    names = CAR(args);
 952 
    names = CAR(args);
Line 955... Line 964...
955 
    }
 964 
    }
956 
    if(n > 1) {
 965 
    if(n > 1) {
957 
	/* +2 for terminator and rounding error */
 966 
	/* +2 for terminator and rounding error */
958 
	buf = Calloc(maxlen + strlen(csep) + log((double)n)/log(10.0) + 2,
967 
	buf = Calloc(maxlen + strlen(csep) + log((double)n)/log(10.0) + 2,
959 
		     char);
 968 
		     char);
- 
 
 969 
	cnts = Calloc(n, int);
- 
 
 970 
	for(i = 0; i < n; i++) cnts[i] = 1;
960 
	data.nomatch = 0;
 971 
	data.nomatch = 0;
961 
	PROTECT(newx = allocVector(STRSXP, 1));
 972 
	PROTECT(newx = allocVector(STRSXP, 1));
962 
	PROTECT(dup = duplicated2(names, &data));
 973 
	PROTECT(dup = duplicated2(names, &data));
963 
	PROTECT(data.HashTable);
 974 
	PROTECT(data.HashTable);
964 
	for(i = 1; i < n; i++) { /* first cannot be a duplicate */
 975 
	for(i = 1; i < n; i++) { /* first cannot be a duplicate */
- 
 
 976 
	    dp = INTEGER(dup)[i]; /* 1-based number of first occurrence */
965 
	    if(!LOGICAL(dup)[i]) continue;
 977 
	    if(dp == 0) continue;
966 
	    /* Try appending 1,2,3, ..., n-1 until it is not already in use */
 978 
	    /* Try appending 1,2,3, ..., n-1 until it is not already in use */
967 
	    for(cnt = 1; cnt < n; cnt++) {
 979 
	    for(cnt = cnts[dp-1]; cnt < n; cnt++) {
968 
		sprintf(buf, "%s%s%d", CHAR(STRING_ELT(names, i)), csep, cnt);
 980 
		sprintf(buf, "%s%s%d", CHAR(STRING_ELT(names, i)), csep, cnt);
969 
		SET_STRING_ELT(newx, 0, mkChar(buf));
 981 
		SET_STRING_ELT(newx, 0, mkChar(buf));
970 
		if(Lookup(ans, newx, 0, &data) == data.nomatch) break;
 982 
		if(Lookup(ans, newx, 0, &data) == data.nomatch) break;
971 
	    }
 983 
	    }
972 
	    SET_STRING_ELT(ans, i, STRING_ELT(newx, 0));
 984 
	    SET_STRING_ELT(ans, i, STRING_ELT(newx, 0));
973 
	    /* insert it */ (void) isDuplicated(ans, i, &data);
 985 
	    /* insert it */ (void) isDuplicated(ans, i, &data);
- 
 
 986 
	    cnts[dp-1] = cnt+1; /* cache the first unused cnt value */
974 
	}
 987 
	}
- 
 
 988 
	Free(cnts);
975 
	Free(buf);
 989 
	Free(buf);
976 
	UNPROTECT(3);
 990 
	UNPROTECT(3);
977 
    }
 991 
    }
978 
    UNPROTECT(1);
 992 
    UNPROTECT(1);
979 
    return ans;
 993 
    return ans;