* Sebastian Hammer, Adam Dickmeiss
*
* $Log: insert.c,v $
- * Revision 1.5 1994-09-01 17:49:39 adam
+ * Revision 1.7 1994-09-12 08:06:42 adam
+ * Futher development of insert.c
+ *
+ * Revision 1.6 1994/09/06 13:05:15 adam
+ * Further development of insertion. Some special cases are
+ * not properly handled yet! assert(0) are put here. The
+ * binary search in each page definitely reduce usr CPU.
+ *
+ * Revision 1.5 1994/09/01 17:49:39 adam
* Removed stupid line. Work on insertion in dictionary. Not finished yet.
*
* Revision 1.4 1994/09/01 17:44:09 adam
#include <dict.h>
+#define CHECK 0
+
static int dict_ins (Dict dict, const Dict_char *str,
Dict_ptr back_ptr, int userlen, void *userinfo);
DICT_nextptr(p) = 0;
DICT_nodir(p) = 0;
DICT_size(p) = DICT_infoffset;
- *pp = p;
+ if (pp)
+ *pp = p;
return ptr;
}
Dict_ptr subptr;
int i, need;
short *indxp, *best_indxp;
- Dict_char best_char;
- Dict_char prev_char;
- int best_no = -1, no_current;
+ Dict_char best_char = 0;
+ Dict_char prev_char = 0;
+ int best_no = -1, no_current = 1;
indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short));
for (i = DICT_nodir (p); --i >= 0; --indxp)
if (best_no < 0)
{ /* first entry met */
best_char = prev_char = dc;
- no_current = best_no = 1;
+ best_no = 1;
}
else if (prev_char == dc)
{ /* same char prefix. update */
assert (*indxp > 0);
info = (char*) p + *indxp; /* entry start */
+ assert (*info == best_char);
slen = dict_strlen(info);
assert (slen > 0);
assert (*indxp > 0);
info = (char*) p + *indxp; /* entry start */
+ assert (*info == best_char);
slen = dict_strlen(info);
if (slen > 1)
if (need <= (1+slen)*sizeof(Dict_char) + 1 + *info1)
best_indxp = indxp; /* space for entry */
dict_ins (dict, info+sizeof(Dict_char), subptr, *info1, info1+1);
+ dict_bf_readp (dict->dbf, ptr, &p);
}
}
if (best_indxp)
memcpy (info, info_here, *info_here+1); /* with information */
else
*info = 0; /* without info */
+#if CHECK
+ best_indxp = NULL;
+ prev_char = 0;
+ indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short));
+ for (i = DICT_nodir (p); --i >= 0; --indxp)
+ {
+ if (*indxp > 0) /* tail string here! */
+ {
+ Dict_char dc;
+
+ memcpy (&dc, (char*) p + *indxp, sizeof(dc));
+ assert (dc != best_char);
+ assert (dc >= prev_char);
+ prev_char = dc;
+ }
+ else
+ {
+ Dict_char dc;
+ memcpy (&dc, (char*)p - *indxp+sizeof(Dict_ptr),
+ sizeof(dc));
+ assert (dc > prev_char);
+ if (dc == best_char)
+ {
+ assert (best_indxp == NULL);
+ best_indxp = indxp;
+ }
+ prev_char = dc;
+ }
+ }
+ assert (best_indxp);
+#endif
}
else
{
return 0;
}
-static void clean_page (Dict dict, void *p)
+static void clean_page (Dict dict, Dict_ptr ptr, void *p, Dict_char *out)
{
char *np = xmalloc (dict->head.page_size);
- int i, slen;
+ int i, slen, no = 0;
short *indxp1, *indxp2;
char *info1, *info2;
/* unsigned char length of information */
/* char * information */
- *--indxp2 = info2 - np;
info1 = (char*) p + *indxp1;
+ if (out && memcmp (out, info1, sizeof(Dict_char)) == 0)
+ continue;
+ *--indxp2 = info2 - np;
slen = (dict_strlen(info1)+1)*sizeof(Dict_char);
memcpy (info2, info1, slen);
info2 += slen;
info1 += slen;
- slen = *info1+1;
- memcpy (info2, info1, slen);
- info2 += slen;
- info1 += slen;
}
else
{
/* unsigned char length of information */
/* char * information */
+ assert (*indxp1 < 0);
*--indxp2 = -(info2 - np);
info1 = (char*) p - *indxp1;
memcpy (info2, info1, sizeof(Dict_ptr)+sizeof(Dict_char));
info2 += sizeof(Dict_ptr)+sizeof(Dict_char);
info1 += sizeof(Dict_ptr)+sizeof(Dict_char);
- slen = *info1+1;
- memcpy (info2, info1, slen);
- info2 += slen;
- info1 += slen;
}
+ slen = *info1+1;
+ memcpy (info2, info1, slen);
+ info2 += slen;
+ ++no;
}
- memcpy ((char*) p + DICT_infoffset, (char*) np + DICT_infoffset,
+ memcpy ((char*)p+DICT_infoffset, (char*)np+DICT_infoffset,
DICT_PAGESIZE-DICT_infoffset);
DICT_size(p) = info2 - np;
DICT_type(p) = 0;
+ DICT_nodir(p) = no;
xfree (np);
}
+
+
+/* return 0 if new */
+/* return 1 if before but change of info */
+/* return 2 if same as before */
+
static int dict_ins (Dict dict, const Dict_char *str,
Dict_ptr back_ptr, int userlen, void *userinfo)
{
- int i, slen, cmp = 1;
+ int hi, lo, mid, i, slen, cmp = 1;
Dict_ptr ptr = back_ptr;
short *indxp;
char *info;
assert (p);
assert (ptr);
+ mid = lo = 0;
+ hi = DICT_nodir(p)-1;
indxp = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short));
- for (i = DICT_nodir (p); --i >= 0; --indxp)
+ while (lo <= hi)
{
- if (*indxp > 0) /* tail string here! */
+ mid = (lo+hi)/2;
+ if (indxp[-mid] > 0)
{
- info = (char*) p + *indxp;
/* string (Dict_char *) DICT_EOS terminated */
/* unsigned char length of information */
/* char * information */
- cmp = dict_strcmp ((Dict_char*) info, str);
+ info = (char*)p + indxp[-mid];
+ cmp = dict_strcmp((Dict_char*) info, str);
if (!cmp)
{
info += (dict_strlen(info)+1)*sizeof(Dict_char);
{
dict_bf_touch (dict->dbf, ptr);
memcpy (info+1, userinfo, userlen);
+ return 1;
}
+ return 2;
}
else if (*info > userlen)
{
*info = userlen;
dict_bf_touch (dict->dbf, ptr);
memcpy (info+1, userinfo, userlen);
+ return 1;
}
- else
- {
- DICT_type(p) = 1;
- break;
- }
- return 0;
- }
- else if(cmp > 0)
break;
+ }
}
- else /* tail of string in sub page */
+ else
{
Dict_char dc;
- assert (*indxp < 0);
- info = (char*) p - *indxp;
+ Dict_ptr subptr;
+
/* Dict_ptr subptr */
/* Dict_char sub char */
/* unsigned char length of information */
/* char * information */
+ info = (char*)p - indxp[-mid];
memcpy (&dc, info+sizeof(Dict_ptr), sizeof(Dict_char));
cmp = dc- *str;
if (!cmp)
{
- Dict_ptr subptr;
- void *pp;
+ memcpy (&subptr, info, sizeof(Dict_ptr));
if (*++str == DICT_EOS)
- { /* missing: consider change of userinfo length ... */
- if (memcmp (info+sizeof(Dict_char)+sizeof(Dict_ptr)+1,
- userinfo, userlen))
+ {
+ int xlen;
+
+ xlen = info[sizeof(Dict_ptr)+sizeof(Dict_char)];
+ if (xlen == userlen)
+ {
+ if (memcmp (info+sizeof(Dict_ptr)+sizeof(Dict_char)+1,
+ userinfo, userlen))
+ {
+ dict_bf_touch (dict->dbf, ptr);
+ memcpy (info+sizeof(Dict_ptr)+sizeof(Dict_char)+1,
+ userinfo, userlen);
+ return 1;
+ }
+ return 2;
+ }
+ else if (xlen > userlen)
+ {
+ DICT_type(p) = 1;
+ info[sizeof(Dict_ptr)+sizeof(Dict_char)] = userlen;
+ memcpy (info+sizeof(Dict_ptr)+sizeof(Dict_char)+1,
+ userinfo, userlen);
+ dict_bf_touch (dict->dbf, ptr);
+ return 1;
+ }
+ if (DICT_size(p)+sizeof(Dict_char)+sizeof(Dict_ptr)+
+ userlen >=
+ DICT_PAGESIZE - (1+DICT_nodir(p))*sizeof(short))
+ {
+ if (DICT_type(p) == 1)
+ {
+ clean_page (dict, ptr, p, NULL);
+ dict_bf_touch (dict->dbf, ptr);
+ return dict_ins (dict, str-1, ptr,
+ userlen, userinfo);
+ }
+ if (split_page (dict, ptr, p))
+ {
+ log (LOG_FATAL, "Unable to split page %d\n", ptr);
+ abort ();
+ }
+ return dict_ins (dict, str-1, ptr, userlen, userinfo);
+ }
+ else
{
- memcpy (dict+sizeof(Dict_char)+sizeof(Dict_ptr)+1,
+ info = (char*)p + DICT_size(p);
+ memcpy (info, &subptr, sizeof(subptr));
+ memcpy (info+sizeof(Dict_ptr), &dc, sizeof(Dict_char));
+ info[sizeof(Dict_char)+sizeof(Dict_ptr)] = userlen;
+ memcpy (info+sizeof(Dict_char)+sizeof(Dict_ptr)+1,
userinfo, userlen);
+ indxp[-mid] = -DICT_size(p);
+ DICT_size(p) += sizeof(Dict_char)+sizeof(Dict_ptr)
+ +1+userlen;
+ DICT_type(p) = 1;
dict_bf_touch (dict->dbf, ptr);
}
+ if (xlen)
+ return 1;
return 0;
}
else
{
- memcpy (&subptr, info, sizeof(subptr));
if (subptr == 0)
{
- subptr = new_page (dict, ptr, &pp);
+ subptr = new_page (dict, ptr, NULL);
memcpy (info, &subptr, sizeof(subptr));
dict_bf_touch (dict->dbf, ptr);
}
- return dict_ins (dict, str, ptr, userlen, userinfo);
+ return dict_ins (dict, str, subptr, userlen, userinfo);
}
}
- else if(cmp > 0)
- break;
}
+ if (cmp < 0)
+ lo = mid+1;
+ else
+ hi = mid-1;
}
+ indxp = indxp-mid;
+ if (lo>hi && cmp < 0)
+ --indxp;
slen = (dict_strlen(str)+1)*sizeof(Dict_char);
if (DICT_size(p)+slen+userlen >=
DICT_PAGESIZE - (1+DICT_nodir(p))*sizeof(short)) /* overflow? */
{
if (DICT_type(p) == 1)
{
- clean_page (dict, p);
- dict_ins (dict, str, ptr, userlen, userinfo);
- return 0;
+ clean_page (dict, ptr, p, NULL);
+ dict_bf_touch (dict->dbf, ptr);
+ return dict_ins (dict, str, ptr, userlen, userinfo);
}
i = 0;
do
{
- if (i > 0)
- assert (0);
+ assert (i <= 1);
if (split_page (dict, ptr, p))
{
log (LOG_FATAL, "Unable to split page %d\n", ptr);
DICT_PAGESIZE - (1+DICT_nodir(p))*sizeof(short))
break;
i++;
- clean_page (dict, p);
+ clean_page (dict, ptr, p, NULL);
} while (DICT_size(p)+slen+userlen > DICT_PAGESIZE -
(1+DICT_nodir(p))*sizeof(short));
- dict_ins (dict, str, ptr, userlen, userinfo);
- return 0;
+ return dict_ins (dict, str, ptr, userlen, userinfo);
}
if (cmp)
{
- DICT_nodir(p)*sizeof(short));
for (; indxp1 != indxp; indxp1++)
indxp1[0] = indxp1[1];
+#if CHECK
+ indxp1 = (short*) ((char*) p+DICT_PAGESIZE-sizeof(short));
+ for (i = DICT_nodir (p); --i >= 0; --indxp1)
+ {
+ if (*indxp1 < 0)
+ {
+ info = (char*)p - *indxp1;
+ assert (info[sizeof(Dict_ptr)] > ' ');
+ }
+ }
+#endif
}
+ else
+ DICT_type(p) = 1;
info = (char*)p + DICT_size(p);
memcpy (info, str, slen);
info += slen;
info += userlen;
*indxp = DICT_size(p);
-#if 0
- printf ("indxp[%d]\n", (char*) indxp - (char*) p);
-#endif
-
DICT_size(p) = info- (char*) p;
dict_bf_touch (dict->dbf, ptr);
- return 0;
+ if (cmp)
+ return 0;
+ return 1;
}
int dict_insert (Dict dict, const Dict_char *str, int userlen, void *userinfo)
{
assert (dict->head.last > 0);
if (dict->head.last == 1)
- dict_ins (dict, str, 0, userlen, userinfo);
+ return dict_ins (dict, str, 0, userlen, userinfo);
else
- dict_ins (dict, str, 1, userlen, userinfo);
- return 0;
+ return dict_ins (dict, str, 1, userlen, userinfo);
}
+
projects / idzebra-moved-to-github.git / blobdiff