// Verify that the index is set up correctly
// by comparing it to the on-disk version.
// This is SLOW
void FMIndex::verify(const std::string& filename)
{
SGABWTReader* p_reader = new SGABWTReader(filename);
// Discard header for now
size_t n1, n2;
BWFlag flag;
p_reader->readHeader(n1, n2, flag);
AlphaCount64 running_count;
// Read one symbol from the bwt at a time
size_t i = 0;
char b;
while((b = p_reader->readChar()) != '\n')
{
// Verify that the symbol at position i matches the symbol
// read from the disk
char s = getChar(i);
assert(s == b);
// Verifiy that the counts interpolated from the markers
// are correct
running_count.increment(b);
// single symbol count
size_t occ = getOcc(s, i);
assert(occ == running_count.get(b));
// full count
AlphaCount64 full_occ = getFullOcc(i);
assert(full_occ == running_count);
i++;
}
printf("Verified all %zu match expected\n", i);
delete p_reader;
}
char *dumpRecord1(RECSTRU *crecp,
RECSTRU *recp,
char *areap,
char *area2p) {
if ((crecp != NULL) && (recp != NULL) &&
(areap != NULL) && (area2p != NULL) && (MFRstatus == ACTIVE)) {
int xdir = 0;
int loop = 0;
int xocc = 0;
int xtag = 0;
FFI xlen = 0;
char stag = 0;
int socc = 0;
int spos = 0;
FFI slen = 0;
char *p = areap;
char *q = area2p;
char *cur = NULL;
char *f = NULL;
char *xxconvent[256];
RECSTRU *arecp = recp;
OccLst *ftop = NULL;
OccLst *stop = NULL;
*p = '\0';
/* chars to entities */
memset(xxconvent, 0x00, sizeof(xxconvent));
xxconvent['&']="&";
xxconvent['<']="<";
xxconvent['>']=">";
xxconvent['"']=""";
xxconvent['\'']="'";
xxconvent['|']="%7C";
// registro de controle
recp = crecp;
sprintf(p, "<masterfile name=\"%s\" nxtmfn=\"%"_LD_"\" mftype=\"%d\" ewlock=\"%"_LD_"\">\n",
RDBname, MF0nxtmfn, MF0mftype, MF0mfcxx3);
p += strlen(p);
recp = arecp;
sprintf(p, "<record mfn=\"%"_LD_"\" nvf=\"%d\" base=\"%"_LD_"\" len=\"%"_LD_"\" status=\"%s\" rclock=\"%d\">\n",
MFRmfn, MFRnvf, (LONGX)MFRbase, (LONGX)MFRmfrl,
(MFRstatus == ACTIVE) ? "active" : "deleted", REClock);
p += strlen(p);
for (xdir = 0, loop = MFRnvf; loop--; xdir++) {
xtag = DIRtag(xdir);
xlen = DIRlen(xdir);
xocc = getOcc(&ftop, xtag);
stag = 0;
spos = 0;
stop = NULL;
sprintf(p, "<field iocc=\"%d\" tag=\"%u\" occ=\"%d\" len=\"%"_LD_"\">",
xdir+1, xtag, xocc, (LONGX)xlen);
p += strlen(p);
cur = f = FIELDP(xdir);
for (; xlen--; cur++) {
if (*cur == '^') {
if (stag > 0) {
socc = getOcc(&stop, stag);
slen = (int)((cur - f) - spos - 1);
sprintf(p, "<subfield tag=\"%c\" occ=\"%d\" pos=\"%d\" len=\"%"_LD_"\">",
stag, socc, spos, slen);
p += strlen(p);
area2p[slen] = 0;
strcpy(p, area2p);
p += strlen(p);
strcpy(p, "</subfield>");
p += strlen(p);
}
stag = *++cur;
xlen--;
spos = (int)(cur - f);
q = area2p;
} else {
if (stag > 0) {
if (xxconvent[*cur]) {
strcpy(q, xxconvent[*cur]);
q += strlen(q);
} else {
*q++ = (*cur);
}
} else {
if (xxconvent[*cur]) {
strcpy(p,xxconvent[*cur]);
p += strlen(p);
} else {
*p++ = (*cur);
}
}
}
}
if (stag > 0) {
socc = getOcc(&stop, stag);
slen = (int)((cur - f) - spos - 1);
sprintf(p, "<subfield tag=\"%c\" occ=\"%d\" pos=\"%d\" len=\"%"_LD_"\">",
stag, socc, spos, slen);
p += strlen(p);
area2p[slen] = 0;
strcpy(p, area2p);
p += strlen(p);
strcpy(p, "</subfield>");
p += strlen(p);
}
sprintf(p, "</field>\n");
p += strlen(p);
}
strcpy(p,"</record>\n");
p += strlen(p);
strcpy(p, "</masterfile>");
deleteList(&stop);
deleteList(&ftop);
}
return areap;
}
