Changeset: 0dd7b7ee1154 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/0dd7b7ee1154
Modified Files:
gdk/gdk_strimps.c
Branch: string_imprints
Log Message:
Remove unused code
diffs (192 lines):
diff --git a/gdk/gdk_strimps.c b/gdk/gdk_strimps.c
--- a/gdk/gdk_strimps.c
+++ b/gdk/gdk_strimps.c
@@ -737,188 +737,3 @@ STRMPcreate(BAT *b, BAT *s)
TRC_DEBUG(ACCELERATOR, "strimp creation took " LLFMT " usec\n",
GDKusec()-t0);
return GDK_SUCCEED;
}
-
-/* Left over code */
-#if 0
-/* This counts how many unicode codepoints the given string
- * contains.
- */
-static size_t
-STRMP_utf8_strlen(const uint8_t *s)
-{
- size_t ret = 0;
- size_t i;
- int m,n;
- uint8_t c;
-
- i = 0;
- while((c = *(s + i)) != 0) {
- if (c < 0x80)
- i++;
- else {
- for (n = 0, m=0x40; c & m; n++, m >>= 1)
- ;
- /* n is now the number of 10xxxxxx bytes that should
- follow. */
- if (n == 0 || n >= 4)
- /* TODO: handle invalid utf-8 */
- {}
- i += n+1;
- }
- ret++;
- }
-
- return ret;
-}
-
-/* Construct a histogram of pairs of bytes in the input BAT.
- *
- * Return the histogram in hist and the number of non-zero bins in
- * count.
- */
-static gdk_return
-STRMPmakehistogramBP(BAT *b, uint64_t *hist, size_t hist_size, size_t *nbins)
-{
- lng t0=0;
- size_t hi;
- BUN i;
- BATiter bi;
- char *ptr, *s;
- /* uint64_t cur_min = 0; */
-
- TRC_DEBUG_IF(ACCELERATOR) t0 = GDKusec();
- assert(b->ttype == TYPE_str);
-
- for(hi = 0; hi < hist_size; hi++)
- hist[hi] = 0;
-
- bi = bat_iterator(b);
- *nbins = 0;
- for(i = 0; i < b->batCount; i++) {
- s = (char *)BUNtvar(bi, i);
- if (!strNil(s)) {
- for(ptr = s; *ptr != 0 && *(ptr + 1) != 0; ptr++) {
- if (isIgnored(*(ptr+1))) {
- /* Skip this and the next pair
- * if the next char is ignored.
- */
- ptr++;
- }
- else if (isIgnored(*ptr)) {
- /* Skip this pair if the current
- * char is ignored. This should
- * only happen at the beginnig
- * of a string.
- */
- ;
- }
- else {
- hi = pairToIndex(*(ptr), *(ptr+1));
- assert(hi < hist_size);
- if (hist[hi] == 0)
- (*nbins)++;
- hist[hi]++;
- /* if (hist[hi] > cur_min) */
- /* cur_min = add_to_header(hi,
hist[hi]); */
- }
- }
- }
- }
-
- TRC_DEBUG(ACCELERATOR, LLFMT " usec\n", GDKusec() - t0);
- GDKtracer_flush_buffer();
- return GDK_SUCCEED;
-}
-
-static bool
-create_header(BAT *b)
-{
- uint64_t hist[STRIMP_HISTSIZE] = {0};
- size_t nbins = 0;
- StrimpHeader *header;
- if ((header = (StrimpHeader*)GDKmalloc(sizeof(StrimpHeader))) == NULL)
- return false;
-
- if(STRMPmakehistogramBP(b, hist, STRIMP_HISTSIZE, &nbins) !=
GDK_SUCCEED) {
- GDKfree(header);
- return NULL;
- }
-
- make_header(header, hist, STRIMP_HISTSIZE);
-
- return header;
-}
-
-/* Given a strimp h and a pair p, return the index i for which
- *
- * h[i] == p
- *
- * Returns -1 if p is not in h.
- *
- * TODO: Should this be inlined somehow? (probably yes)
- */
-static int8_t
-lookup_index(BAT *b, uint8_t *pair, uint8_t psize)
-{
- size_t i,j;
- size_t idx = 0;
- Heap strimp = b->tstrimps->strimps;
- uint64_t desc = (uint64_t)strimp.base[0];
- uint8_t npairs = NPAIRS(desc);
- uint8_t *pair_sizes = b->tstrimps->sizes_base;
- uint8_t *pairs = b->tstrimps->pairs_base;
-
- for(i = 0; i < npairs; i++) {
- if (psize == pair_sizes[i]) {
- uint8_t *h = pairs + idx;
- for (j = 0; j < psize; j++) {
- if(pair[j] != h[j])
- break;
- }
- if (j == psize)
- return i;
- }
- idx += pair_sizes[i];
- }
-
- return -1;
-}
-
-/* Given a BAT return the number of digrams in it. The observation is
- * that the number of digrams is the number of characters - 1:
- *
- * 1 digram starting at character 1
- * 1 digram starting at character 2
- * [...]
- * 1 digram starting at character n - 1
- */
-gdk_return
-STRMPndigrams(BAT *b, size_t *n)
-{
- // lng t0;
- BUN i;
- BATiter bi;
- char *s;
- // GDKtracer_set_component_level("ALGO", "DEBUG");
- // struct canditer ci;
-
- // t0 = GDKusec();
- // BATcheck(b, NULL);
- assert(b->ttype == TYPE_str);
-
- bi = bat_iterator(b);
- *n = 0;
- for (i = 0; i < b->batCount; i++) {
- s = (char *)BUNtail(bi, i);
- // *n += STRMP_strlen(s) - 1;
- *n += strlen(s) - 1;
- // TRC_DEBUG(ACCELERATOR, "s["LLFMT"]=%s\n", i, s);
- }
-
- // TRC_DEBUG(ACCELERATOR, LLFMT "usec\n", GDKusec() - t0);
- // GDKtracer_flush_buffer();
-
- return GDK_SUCCEED;
-}
-
-#endif
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
