CharAE *_new_CharAE_from_string(const char *string)
{
CharAE *ae;
ae = _new_CharAE(strlen(string));
_CharAE_set_nelt(ae, ae->_buflength);
memcpy(ae->elts, string, ae->_buflength);
return ae;
}
/* --- .Call ENTRY POINT ---/
* Same arguments as RangesList_encode_overlaps() plus:
* 'query_hits', 'subject_hits': integer vectors of the same length.
* 'flip_query': logical vector of the same length as 'query_hits'.
*/
SEXP Hits_encode_overlaps(SEXP query_starts, SEXP query_widths,
SEXP query_spaces, SEXP query_breaks,
SEXP subject_starts, SEXP subject_widths,
SEXP subject_spaces,
SEXP query_hits, SEXP subject_hits, SEXP flip_query)
{
int q_len, s_len, ans_len, i, j, k;
const int *q_hits, *s_hits;
SEXP ans_Loffset, ans_Roffset, ans_encoding, ans_encoding_elt, ans;
CharAE buf;
/* TODO: Add some basic checking of the input values. */
q_len = LENGTH(query_starts);
s_len = LENGTH(subject_starts);
ans_len = _check_integer_pairs(query_hits, subject_hits,
&q_hits, &s_hits,
"queryHits(hits)", "subjectHits(hits)");
PROTECT(ans_Loffset = NEW_INTEGER(ans_len));
PROTECT(ans_Roffset = NEW_INTEGER(ans_len));
PROTECT(ans_encoding = NEW_CHARACTER(ans_len));
buf = _new_CharAE(0);
for (k = 0; k < ans_len; k++) {
i = q_hits[k];
j = s_hits[k];
if (i == NA_INTEGER || i < 1 || i > q_len ||
j == NA_INTEGER || j < 1 || j > s_len) {
UNPROTECT(3);
error("'queryHits(hits)' or 'subjectHits(hits)' "
"contain invalid indices");
}
i--;
j--;
PROTECT(ans_encoding_elt = RangesList_encode_overlaps_ij(
query_starts, query_widths,
query_spaces, query_breaks,
subject_starts, subject_widths, subject_spaces,
i, j, LOGICAL(flip_query)[k],
INTEGER(ans_Loffset) + k,
INTEGER(ans_Roffset) + k,
&buf));
SET_STRING_ELT(ans_encoding, k, ans_encoding_elt);
UNPROTECT(1);
_CharAE_set_nelt(&buf, 0);
}
PROTECT(ans = make_LIST_from_ovenc_parts(ans_Loffset, ans_Roffset,
ans_encoding));
UNPROTECT(4);
return ans;
}
void _append_string_to_CharAE(CharAE *ae, const char *string)
{
int nnewval, nelt, new_nelt;
char *dest;
nnewval = strlen(string);
nelt = _CharAE_get_nelt(ae);
new_nelt = nelt + nnewval;
if (new_nelt > ae->_buflength)
CharAE_extend(ae, new_nelt);
dest = ae->elts + nelt;
memcpy(dest, string, sizeof(char) * nnewval);
_CharAE_set_nelt(ae, new_nelt);
return;
}
/*
* Delete 'nelt' elements, starting at position 'at'.
* Doing _CharAE_delete_at(x, at, nelt) is equivalent to doing
* _CharAE_delete_at(x, at, 1) 'nelt' times.
*/
void _CharAE_delete_at(CharAE *ae, int at, int nelt)
{
char *c1_p;
const char *c2_p;
int nelt0, i2;
if (nelt == 0)
return;
c1_p = ae->elts + at;
c2_p = c1_p + nelt;
nelt0 = _CharAE_get_nelt(ae);
for (i2 = at + nelt; i2 < nelt0; i2++)
*(c1_p++) = *(c2_p++);
_CharAE_set_nelt(ae, nelt0 - nelt);
return;
}
void _CharAE_insert_at(CharAE *ae, int at, char c)
{
int nelt, i;
char *elt1_p;
const char *elt2_p;
nelt = _CharAE_get_nelt(ae);
if (nelt >= ae->_buflength)
CharAE_extend(ae, -1);
elt1_p = ae->elts + nelt;
elt2_p = elt1_p - 1;
for (i = nelt; i > at; i--)
*(elt1_p--) = *(elt2_p--);
*elt1_p = c;
_CharAE_set_nelt(ae, nelt + 1);
return;
}
/* --- .Call ENTRY POINT ---/
* 'query_starts', 'query_widths', 'query_spaces': lists of integer vectors.
* The 3 lists are assumed to have the same length (M) and shape.
* 'query_breaks': NULL or integer vector of length M.
* 'subject_starts', 'subject_widths', 'subject_spaces': lists of integer
* vectors. The 3 lists are assumed to have the same length (N) and shape.
* Return a named list with the 3 following components (all of the same
* length):
* 1. Loffset: integer vector;
* 2. Roffset: integer vector;
* 3. encoding: character vector containing the compact encodings (type
* II).
*/
SEXP RangesList_encode_overlaps(SEXP query_starts, SEXP query_widths,
SEXP query_spaces, SEXP query_breaks,
SEXP subject_starts, SEXP subject_widths,
SEXP subject_spaces)
{
int q_len, s_len, ans_len, i, j, k;
SEXP ans_Loffset, ans_Roffset, ans_encoding, ans_encoding_elt, ans;
CharAE buf;
/* TODO: Add some basic checking of the input values. */
q_len = LENGTH(query_starts);
s_len = LENGTH(subject_starts);
if (q_len == 0 || s_len == 0)
ans_len = 0;
else
ans_len = q_len >= s_len ? q_len : s_len;
PROTECT(ans_Loffset = NEW_INTEGER(ans_len));
PROTECT(ans_Roffset = NEW_INTEGER(ans_len));
PROTECT(ans_encoding = NEW_CHARACTER(ans_len));
buf = _new_CharAE(0);
for (i = j = k = 0; k < ans_len; i++, j++, k++) {
if (i >= q_len)
i = 0; /* recycle i */
if (j >= s_len)
j = 0; /* recycle j */
PROTECT(ans_encoding_elt = RangesList_encode_overlaps_ij(
query_starts, query_widths,
query_spaces, query_breaks,
subject_starts, subject_widths, subject_spaces,
i, j, 0,
INTEGER(ans_Loffset) + k,
INTEGER(ans_Roffset) + k,
&buf));
SET_STRING_ELT(ans_encoding, k, ans_encoding_elt);
UNPROTECT(1);
_CharAE_set_nelt(&buf, 0);
}
if (ans_len != 0 && (i != q_len || j != s_len))
warning("longer object length is not a multiple "
"of shorter object length");
PROTECT(ans = make_LIST_from_ovenc_parts(ans_Loffset, ans_Roffset,
ans_encoding));
UNPROTECT(4);
return ans;
}
/*
* q_start, q_width: int arrays of length q_len. No NAs.
* q_space: NULL or an int array of length q_len. No NAs.
* q_len: nb of ranges in the query.
* q_break: 0 if all the ranges in the query are coming from the same
* segment (single-end read), or, an int >= 1 and < q_len specifying
* the position of the break between the ranges coming from one
* segment and the ranges coming from the other if the query is a
* paired-end read.
* flip_query: if non-zero, then the query is "flipped" before the encoding is
* computed.
* s_start, s_width: int arrays of length s_len. No NAs.
* s_space: NULL or an int array of length s_len. No NAs.
* s_len: nb of ranges in the subject.
* as_matrix, Loffset, Roffset: if as_matrix, then the full matrix of codes
* is returned and the returned values for Loffset and Roffset are
* undefined. Otherwise, the matrix is trimmed and the returned values
* for Loffset and Roffset are the number of cols removed on the left
* and right sides of the matrix, respectively.
* out: character array containing the matrix of codes (possibly trimmed)
*/
static void unsafe_overlap_encoding(
const int *q_start, const int *q_width, const int *q_space,
int q_len,
int q_break, int flip_query,
const int *s_start, const int *s_width, const int *s_space,
int s_len,
int as_matrix, int *Loffset, int *Roffset, CharAE *out)
{
int out_nelt0, i, starti, widthi, spacei, j, startj, widthj, spacej,
j1, j2, nrow;
char letter;
if (!as_matrix) {
if (q_break != 0) {
if (flip_query) {
CharAE_append_int(out, q_len - q_break);
CharAE_append_char(out, '-', 2);
CharAE_append_int(out, q_break);
} else {
CharAE_append_int(out, q_break);
CharAE_append_char(out, '-', 2);
CharAE_append_int(out, q_len - q_break);
}
} else {
CharAE_append_int(out, q_len);
}
CharAE_append_char(out, ':', 1);
out_nelt0 = _CharAE_get_nelt(out);
}
/* j1: 0-based index of first (i.e. leftmost) col with a non-"m",
or 's_len' if there is no such col.
j2: 0-based index of last (i.e. rightmost) col with a non-"a",
or -1 if there is no such col. */
j1 = s_len;
j2 = -1;
/* Walk col by col. */
for (j = 0; j < s_len; j++) {
startj = s_start[j];
widthj = s_width[j];
spacej = s_space == NULL ? 0 : s_space[j];
if (flip_query) {
for (i = q_len - 1; i >= 0; i--) {
starti = q_start[i];
widthi = q_width[i];
spacei = q_space == NULL ? 0 : - q_space[i];
letter = overlap_letter(starti, widthi, spacei,
startj, widthj, spacej);
CharAE_append_char(out, letter, 1);
if (j1 == s_len && letter != 'm')
j1 = j;
if (letter != 'a')
j2 = j;
if (q_break != 0 && i == q_break)
CharAE_append_char(out, '-', 2);
}
} else {
for (i = 0; i < q_len; i++) {
if (q_break != 0 && i == q_break)
CharAE_append_char(out, '-', 2);
starti = q_start[i];
widthi = q_width[i];
spacei = q_space == NULL ? 0 : q_space[i];
letter = overlap_letter(starti, widthi, spacei,
startj, widthj, spacej);
CharAE_append_char(out, letter, 1);
if (j1 == s_len && letter != 'm')
j1 = j;
if (letter != 'a')
j2 = j;
}
}
}
if (as_matrix)
return;
/* By making 'j2' a 1-based index we will then have
0 <= j1 <= j2 <= s_len, which will simplify further
arithmetic/logic. */
if (q_len == 0) {
/* A 0-row matrix needs special treatment. */
j2 = s_len;
} else {
j2++;
}
*Loffset = j1;
*Roffset = s_len - j2;
nrow = q_len;
if (q_break != 0)
nrow += 2;
/* Remove "a"-cols on the right. */
_CharAE_set_nelt(out, out_nelt0 + j2 * nrow);
/* Remove "m"-cols on the left. */
_CharAE_delete_at(out, out_nelt0, j1 * nrow);
/* Insert ":" at the end of each remaining col. */
for (j = j2 - j1; j >= 1; j--)
_CharAE_insert_at(out, out_nelt0 + j * nrow, ':');
return;
}