Changeset: e7e309a77c0e for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=e7e309a77c0e
Modified Files:
clients/Tests/exports.stable.out
common/stream/stream.c
common/stream/stream.h
gdk/gdk_batop.c
gdk/gdk_select.c
monetdb5/mal/mal_import.c
monetdb5/modules/mal/tokenizer.c
sql/backends/monet5/rel_bin.c
sql/backends/monet5/sql.c
sql/backends/monet5/sql_gencode.c
sql/backends/monet5/sql_scenario.c
sql/backends/monet5/sql_statement.c
sql/backends/monet5/sql_statement.h
Branch: default
Log Message:
Merge with Oct2014 branch.
diffs (truncated from 977 to 300 lines):
diff --git a/clients/Tests/exports.stable.out b/clients/Tests/exports.stable.out
--- a/clients/Tests/exports.stable.out
+++ b/clients/Tests/exports.stable.out
@@ -2835,6 +2835,7 @@ stream *file_rstream(FILE *fp, const cha
stream *file_wastream(FILE *fp, const char *name);
stream *file_wstream(FILE *fp, const char *name);
FILE *getFile(stream *s);
+size_t getFileSize(stream *s);
stream *iconv_rstream(stream *ss, const char *charset, const char *name);
stream *iconv_wstream(stream *ss, const char *charset, const char *name);
int isa_block_stream(stream *s);
diff --git a/common/stream/stream.c b/common/stream/stream.c
--- a/common/stream/stream.c
+++ b/common/stream/stream.c
@@ -71,8 +71,16 @@
#include <stdarg.h> /* va_alist.. */
#include <assert.h>
+#ifdef HAVE_SYS_TYPES_H
+# include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+# include <sys/stat.h>
+#endif
+#ifdef HAVE_UNISTD_H
+# include <unistd.h>
+#endif
#ifdef HAVE_NETDB_H
-# include <sys/types.h>
# include <netinet/in_systm.h>
# include <netinet/in.h>
# include <netinet/ip.h>
@@ -577,7 +585,7 @@ file_read(stream *s, void *buf, size_t e
return -1;
}
- if (!feof(fp)) {
+ if (elmsize && cnt && !feof(fp)) {
if (ferror(fp) ||
((rc = fread(buf, elmsize, cnt, fp)) == 0 &&
ferror(fp))) {
@@ -724,6 +732,18 @@ getFile(stream *s)
return (FILE *) s->stream_data.p;
}
+size_t
+getFileSize(stream *s)
+{
+ if (s->read == file_read) {
+ struct stat stb;
+
+ fstat(fileno((FILE *) s->stream_data.p), &stb);
+ return (size_t) stb.st_size;
+ }
+ return 0; /* unknown */
+}
+
static stream *
open_stream(const char *filename, const char *flags)
{
@@ -781,7 +801,7 @@ stream_gzread(stream *s, void *buf, size
return -1;
}
- if (!gzeof(fp)) {
+ if (size && !gzeof(fp)) {
size = gzread(fp, buf, size);
if (gzerror(fp, &err) != NULL && err < 0) {
s->errnr = MNSTR_READ_ERROR;
@@ -1003,6 +1023,8 @@ stream_bzread(stream *s, void *buf, size
s->errnr = MNSTR_READ_ERROR;
return -1;
}
+ if (size == 0)
+ return 0;
size = BZ2_bzRead(&err, bzp->b, buf, size);
if (err == BZ_STREAM_END) {
/* end of stream, but not necessarily end of file: get
@@ -1450,20 +1472,23 @@ static ssize_t
curl_read(stream *s, void *buf, size_t elmsize, size_t cnt)
{
struct curl_data *c = (struct curl_data *) s->stream_data.p;
- size_t size;
+ size_t size = cnt * elmsize;
if (c == NULL) {
s->errnr = MNSTR_READ_ERROR;
return -1;
}
+ if (size == 0)
+ return 0;
if (c->usesize - c->offset >= elmsize || !c->running) {
/* there is at least one element's worth of data
* available, or we have reached the end: return as
* much as we have, but no more than requested */
- if (cnt * elmsize > c->usesize - c->offset)
+ if (size > c->usesize - c->offset) {
cnt = (c->usesize - c->offset) / elmsize;
- size = cnt * elmsize;
+ size = cnt * elmsize;
+ }
memcpy(buf, c->buffer + c->offset, size);
c->offset += size;
if (c->offset == c->usesize)
@@ -1645,6 +1670,8 @@ socket_read(stream *s, void *buf, size_t
if (s->errnr || size == 0)
return -1;
+ if (size == 0)
+ return 0;
do {
errno = 0;
#ifdef NATIVE_WIN32
@@ -1910,6 +1937,8 @@ udp_read(stream *s, void *buf, size_t el
if (s->errnr || udp == NULL)
return -1;
+ if (size == 0)
+ return 0;
errno = 0;
if ((res = recvfrom(udp->s, buf,
#ifdef NATIVE_WIN32
@@ -2255,6 +2284,8 @@ ic_read(stream *s, void *buf, size_t elm
inbytesleft = ic->buflen;
outbuf = (char *) buf;
outbytesleft = elmsize * cnt;
+ if (outbytesleft == 0)
+ return 0;
while (outbytesleft > 0 && !ic->eof) {
if (ic->buflen == sizeof(ic->buffer)) {
/* ridiculously long multibyte sequence, return error */
@@ -2560,7 +2591,7 @@ buffer_read(stream *s, void *buf, size_t
b = (buffer *) s->stream_data.p;
assert(b);
- if (b && b->pos + size <= b->len) {
+ if (size && b && b->pos + size <= b->len) {
memcpy(buf, b->buf + b->pos, size);
b->pos += size;
return (ssize_t) (size / elmsize);
@@ -2946,7 +2977,7 @@ bs_read(stream *ss, void *buf, size_t el
* empty read */
if (todo > 0 && cnt == 0)
s->nr = 0;
- return (ssize_t) (cnt / elmsize);
+ return (ssize_t) (elmsize > 0 ? cnt / elmsize : 0);
}
static void
diff --git a/common/stream/stream.h b/common/stream/stream.h
--- a/common/stream/stream.h
+++ b/common/stream/stream.h
@@ -192,6 +192,7 @@ stream_export stream *file_rastream(FILE
stream_export stream *file_wastream(FILE *fp, const char *name);
stream_export FILE *getFile(stream *s);
+stream_export size_t getFileSize(stream *s);
stream_export stream *iconv_rstream(stream *ss, const char *charset, const
char *name);
stream_export stream *iconv_wstream(stream *ss, const char *charset, const
char *name);
diff --git a/gdk/gdk_batop.c b/gdk/gdk_batop.c
--- a/gdk/gdk_batop.c
+++ b/gdk/gdk_batop.c
@@ -1897,6 +1897,21 @@ BATcount_no_nil(BAT *b)
return cnt;
}
+static BAT *
+newdensecand(oid first, oid last)
+{
+ BAT *bn;
+
+ if ((bn = BATnew(TYPE_void, TYPE_void, 0, TRANSIENT)) == NULL)
+ return NULL;
+ if (last < first)
+ first = last = 0; /* empty range */
+ BATsetcount(bn, last - first + 1);
+ BATseqbase(bn, 0);
+ BATseqbase(BATmirror(bn), first);
+ return bn;
+}
+
/* merge two candidate lists and produce a new one
*
* candidate lists are VOID-headed BATs with an OID tail which is
@@ -1926,10 +1941,10 @@ BATmergecand(BAT *a, BAT *b)
assert(b->T->nonil);
/* we can return a if b is empty (and v.v.) */
- if ( BATcount(a) == 0){
+ if (BATcount(a) == 0) {
return BATcopy(b, b->htype, b->ttype, 0, TRANSIENT);
}
- if ( BATcount(b) == 0){
+ if (BATcount(b) == 0) {
return BATcopy(a, a->htype, a->ttype, 0, TRANSIENT);
}
/* we can return a if a fully covers b (and v.v) */
@@ -1941,11 +1956,24 @@ BATmergecand(BAT *a, BAT *b)
bl = *(oid*) BUNtail(bi, BUNlast(b) - 1);
ad = (af + BATcount(a) - 1 == al); /* i.e., dense */
bd = (bf + BATcount(b) - 1 == bl); /* i.e., dense */
+ if (ad && bd) {
+ /* both are dense */
+ if (af <= bf && bf <= al + 1) {
+ /* partial overlap starting with a, or b is
+ * smack bang after a */
+ return newdensecand(af, al < bl ? bl : al);
+ }
+ if (bf <= af && af <= bl + 1) {
+ /* partial overlap starting with b, or a is
+ * smack bang after b */
+ return newdensecand(bf, al < bl ? bl : al);
+ }
+ }
if (ad && af <= bf && al >= bl) {
- return BATcopy(a, a->htype, a->ttype, 0, TRANSIENT);
+ return newdensecand(af, al);
}
if (bd && bf <= af && bl >= al) {
- return BATcopy(b, b->htype, b->ttype, 0, TRANSIENT);
+ return newdensecand(bf, bl);
}
bn = BATnew(TYPE_void, TYPE_oid, BATcount(a) + BATcount(b), TRANSIENT);
@@ -2010,11 +2038,22 @@ BATmergecand(BAT *a, BAT *b)
/* properties */
BATsetcount(bn, (BUN) (p - (oid *) Tloc(bn, BUNfirst(bn))));
BATseqbase(bn, 0);
- bn->trevsorted = 0;
+ bn->trevsorted = BATcount(bn) <= 1;
bn->tsorted = 1;
bn->tkey = 1;
bn->T->nil = 0;
bn->T->nonil = 1;
+ af = * (const oid *) Tloc(bn, BUNfirst(bn));
+ if (af + BATcount(bn) - 1 == *(const oid *) Tloc(bn, BUNlast(bn) - 1)) {
+ /* new bat is in fact dense, replace by void column */
+ bn->tseqbase = af;
+ bn->tdense = 1;
+ HEAPfree(&bn->T->heap);
+ bn->ttype = TYPE_void;
+ bn->tvarsized = 1;
+ bn->T->width = 0;
+ bn->T->shift = 0;
+ }
return bn;
}
@@ -2028,7 +2067,9 @@ BATintersectcand(BAT *a, BAT *b)
{
BAT *bn;
const oid *ap, *bp, *ape, *bpe;
- oid *p, i;
+ oid *p;
+ oid af, al, bf, bl;
+ BATiter ai, bi;
BATcheck(a, "BATintersectcand");
BATcheck(b, "BATintersectcand");
@@ -2044,30 +2085,19 @@ BATintersectcand(BAT *a, BAT *b)
assert(b->T->nonil);
if (BATcount(a) == 0 || BATcount(b) == 0) {
- bn = BATnew(TYPE_void, TYPE_void, 0, TRANSIENT);
- if (bn) {
- BATseqbase(bn, 0);
- BATseqbase(BATmirror(bn), 0);
- }
- return bn;
+ return newdensecand(0, 0);
}
- if (a->ttype == TYPE_void && b->ttype == TYPE_void) {
+ ai = bat_iterator(a);
+ bi = bat_iterator(b);
+ af = *(oid*) BUNtail(ai, BUNfirst(a));
+ bf = *(oid*) BUNtail(bi, BUNfirst(b));
+ al = *(oid*) BUNtail(ai, BUNlast(a) - 1);
+ bl = *(oid*) BUNtail(bi, BUNlast(b) - 1);
+
+ if ((af + BATcount(a) - 1 == al) && (bf + BATcount(b) - 1 == bl)) {
/* both lists are VOID */
- bn = BATnew(TYPE_void, TYPE_void, 0, TRANSIENT);
- if (bn == NULL)
- return NULL;
- i = MAX(a->tseqbase, b->tseqbase);
- if (a->tseqbase + BATcount(a) <= b->tseqbase ||
- b->tseqbase + BATcount(b) <= a->tseqbase) {
- /* no overlap */
- BATsetcount(bn, 0);
- } else {
- BATsetcount(bn, MIN(a->tseqbase + BATcount(a) - i,
- b->tseqbase + BATcount(b) - i));
- }
- BATseqbase(BATmirror(bn), i);
- return bn;
+ return newdensecand(MAX(af, bf), MIN(al, bl));
}
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
