int
main(int argc,char **argv)
{
/* read query from stdin, expect name of indexes in argv[1] */
static FILE *fa, *fb, *fc;
char nma[PATH_MAX], nmb[PATH_MAX], nmc[PATH_MAX],
*qitem[100], *rprog = NULL;
char nmd[PATH_MAX], grepquery[256];
static char oldname[30] ;
static int was =0;
/* these pointers are unions of pointer to int and pointer to long */
long *hpt = 0;
unsigned *master =0;
int falseflg, nhash, nitem, nfound = 0, frtbl, kk;
/* special wart for refpart: default is tags only */
falseflg = 0;
while (argc > 1 && argv[1][0] == '-')
{
switch(argv[1][1])
{
case 'a': /* all output, incl. false drops */
falseflg = 1;
break;
case 'r':
argc--;
argv++;
rprog = argv[1];
break;
case 'F': /* put out full text */
full = setfrom(argv[1][2]);
break;
case 'T': /* put out tags */
tags = setfrom(argv[1][2]);
break;
case 'i': /* input in argument string */
argc--;
argv++;
sinput = argv[1];
break;
case 's': /*text output to string */
case 'o':
argc--;
argv++;
soutput = argv[1];
if ((intptr_t) argv[2]<16000)
{
soutlen = (intptr_t)argv[2];
argc--;
argv++;
}
break;
case 't': /*tag output to string */
argc--;
argv++;
tagout = argv[1];
break;
case 'l': /* length of internal lists */
argc--;
argv++;
lmaster = atoi(argv[1]);
break;
case 'g': /* suppress fgrep search on old files */
keepold = 0;
break;
case 'C': /* coordination level */
colevel = atoi(argv[1]+2);
# if D1
fprintf(stderr, "colevel set to %d\n",colevel);
# endif
break;
case 'P': /* print term freqs */
prfreqs=1;
break;
case 'm':
measure=1;
break;
}
argc--;
argv++;
}
if(argc < 2)
exit(1);
strcpy (nma, todir(argv[1]));
if (was == 0 || strcmp (oldname, nma) !=0)
{
strcpy (oldname,nma);
strcpy (nmb, nma);
strcpy (nmc, nmb);
strcpy(nmd,nma);
strcat (nma, ".ia");
strcat (nmb, ".ib");
strcat (nmc, ".ic");
strcat (nmd, ".id");
if (was)
{
fclose(fa);
fclose(fb);
fclose(fc);
}
fa = fopen(nma, "r");
if (fa==NULL)
{
strcpy(*fgnamp++ = calloc(strlen(oldname)+2,1), oldname);
fb=NULL;
goto search;
}
fb = fopen(nmb, "r");
fc = fopen(nmc, "r");
was =1;
if (fb== NULL || fc ==NULL)
{
err("Index incomplete %s", nmb);
exit(1);
}
indexdate = gdate(fb);
fd = fopen(nmd, "r");
}
fseek (fa, 0, SEEK_SET);
fread (&nhash, sizeof(nhash), 1, fa);
fread (&iflong, sizeof(iflong), 1, fa);
if(master==0)
master = calloc (lmaster, iflong? sizeof(long): sizeof(unsigned));
hpt = calloc(nhash, sizeof(*hpt));
kk=fread( hpt, sizeof(*hpt), nhash, fa);
# if D1
fprintf(stderr,"read %d hashes, iflong %d, nhash %d\n", kk, iflong, nhash);
# endif
assert (kk==nhash);
hfreq = calloc(nhash, sizeof(*hfreq));
assert (hfreq != NULL);
frtbl = fread(hfreq, sizeof(*hfreq), nhash, fa);
hfrflg = (frtbl == nhash);
# if D1
fprintf(stderr, "read freqs %d\n", frtbl);
# endif
search:
while (1)
{
nitem = getq(qitem);
if (measure) tick();
if (nitem==0) continue;
if (nitem < 0) break;
if (tagout) tagout[0]=0;
if (fb!=NULL)
{
nfound = doquery(hpt, nhash, fb, nitem, qitem, master);
# if D1
fprintf(stderr,"after doquery nfound %d\n", nfound);
# endif
fgnamp=fgnames;
if (falseflg == 0)
nfound = baddrop(master, nfound, fc, nitem, qitem, rprog, full);
# if D1
fprintf(stderr,"after baddrop nfound %d\n", nfound);
# endif
}
if (fgnamp>fgnames)
{
char **fgp, tgbuff[100];
int k;
# if D1
fprintf(stderr, "were %d bad files\n", fgnamp-fgnames);
# endif
memset(tgbuff, 0, sizeof (tgbuff));
grepquery[0]=0;
for(k=0; k<nitem; k++)
{
strcat(grepquery, " ");
strcat(grepquery, qitem[k]);
}
# if D1
fprintf(stderr, "grepquery %s\n",grepquery);
# endif
for(fgp=fgnames; fgp<fgnamp; fgp++)
{
# if D1
fprintf(stderr, "Now on %s query /%s/\n", *fgp, grepquery);
# endif
makefgrep(*fgp);
# if D1
fprintf(stderr, "grepmade\n");
# endif
if (tagout==0)
tagout=tgbuff;
grepcall(grepquery, tagout, *fgp);
# if D1
fprintf(stderr, "tagout now /%s/\n", tagout);
# endif
if (full)
{
int nout;
char *bout;
char *tagp;
char *oldtagp;
tagp = tagout;
while (*tagp) {
oldtagp = tagp;
while (*tagp && (*tagp != '\n'))
tagp++;
if (*tagp)
tagp++;
nout = findline(oldtagp, &bout, 1000, 0L);
if (nout > 0)
{
fputs(bout, stdout);
free(bout);
}
}
}
}
}
if (tags)
result (master, nfound >tags ? tags: nfound, fc);
if (measure) tock();
}
/* NOTREACHED */
return 0;
}
void
huntmain(int argc,char **argv)
{
/* read query from stdin, expect name of indexes in argv[1] */
static FILE *fa, *fb, *fc;
char indexname[PATH_MAX], *qitem[100], *rprog = 0;
char grepquery[200];
static char oldname[30] ;
static int nhash = 0;
static int maxhash = 0;
int falseflg = 0, nitem, nfound, frtbl;
static long *hpt = 0;
unsigned *masterp;
# if D1
fprintf(stderr, "in glue1 argc %d argv %o %o\n", argc, argv[0],argv[1]);
# endif
savedir();
while (argv[1][0] == '-')
{
# if D1
fprintf(stderr, "argv.1 is %s\n",argv[1]);
# endif
switch(argv[1][1])
{
case 'a': /* all output, incl. false drops */
falseflg = 1;
break;
case 'r':
argc--;
argv++;
rprog = argv[1];
break;
case 'F': /* put out full text */
full = setfrom(argv[1][2]);
break;
case 'T': /* put out tags */
tags = setfrom(argv[1][2]);
break;
case 'i': /* input in argument string */
argc--;
argv++;
sinput = argv[1];
break;
case 's': /*text output to string */
case 'o':
argc--;
argv++;
soutput = argv[1];
if ((intptr_t) argv[2]<16000)
{
soutlen = (intptr_t) argv[2];
argc--;
argv++;
}
break;
case 't': /*tag output to string */
argc--;
argv++;
tagout = argv[1];
if ((intptr_t)argv[2]<16000)
{
taglen = (intptr_t)argv[2];
argc--;
argv++;
}
break;
case 'l': /* specify length of lists */
argc--;
argv++;
lmaster = atoi(argv[1]);
# if D1
fprintf(stderr, "lmaster now %d\n",lmaster);
# endif
break;
case 'C':
argc--;
argv++;
colevel = atoi(argv[1]);
break;
}
argc--;
argv++;
}
n_strcpy (indexname, todir(argv[1]), sizeof(indexname));
# if D1
fprintf(stderr, "in huntmain indexname %s typeindex %d\n", indexname, typeindex);
# endif
if (typeindex == 0 || strcmp (oldname, indexname) !=0)
{
n_strcpy (oldname, indexname, sizeof(oldname));
unopen(fa);
unopen(fb);
unopen(fc);
if (ckexist(indexname, ".ib"))
{
# if D1
fprintf(stderr, "found old index\n");
# endif
fa = iopen(indexname, ".ia");
fb = iopen(indexname, ".ib");
fc = iopen(indexname, ".ic");
typeindex =1;
# if D1
fprintf(stderr, "opened f's as %o %o %o\n",fa,fb,fc);
# endif
indexdate = gdate(fb);
fread (&nhash, sizeof(nhash), 1, fa);
fread (&iflong, sizeof(iflong), 1, fa);
if (nhash > maxhash)
{
if (hpt)
free (hpt);
hpt=0;
if (hfreq)
free(hfreq);
hfreq=0;
maxhash=nhash;
# if D1
fprintf(stderr, "Freed if needed maxhash %d\n",maxhash);
# endif
}
if (hpt==0)
hpt = zalloc(nhash, sizeof(*hpt));
# if D1
fprintf(stderr, "hpt now %o\n",hpt);
# endif
if (hpt == NULL)
err("No space for hash list (%d)", nhash);
fread( hpt, sizeof(*hpt), nhash, fa);
if (hfreq==0)
hfreq=zalloc(nhash, sizeof(*hfreq));
if (hfreq==NULL)
err("No space for hash frequencies (%d)",
nhash);
frtbl = fread(hfreq, sizeof(*hfreq), nhash, fa);
hfrflg = (frtbl == nhash);
# if D1
fprintf(stderr,"Read pointer files\n");
# endif
if (master.a == NULL)
{
if (iflong)
master.b = zalloc(lmaster, sizeof(long));
else
master.a = zalloc(lmaster, sizeof(int));
}
if (master.a == NULL)
err("no space for answer list", 0);
}
else
if (makefgrep(indexname))
typeindex=2;
else
{
err("No files %s\n", indexname);
exit(1);
}
}
if (iflong)
masterp = (unsigned *) master.b;
else
masterp = master.a;
# if D1
fprintf(stderr, "typeindex now %d\n",typeindex);
# endif
tagout[0]=0;
if (typeindex==2)
{
grepcall(sinput, tagout, indexname);
# if D1
fprintf(stderr, " back from grepcall\n");
# endif
restodir();
return;
}
nitem = getq(qitem);
# if D1
fprintf(stderr, "approaching doquery fb %o\n", fb);
# endif
nfound = doquery(hpt, nhash, fb, nitem, qitem, masterp);
# ifdef D1
fprintf(stderr, "return from doquery with nfound %d\n", nfound);
# endif
if (falseflg == 0)
nfound = baddrop(masterp, nfound, fc, nitem, qitem, rprog, full);
# ifdef D1
fprintf(stderr, "after baddrop with nfound %d\n",nfound);
fprintf(stderr, "tagout is /%s/, sout /%s/\n",tagout, soutput);
# endif
if (tags)
result (masterp, nfound >tags ? tags : nfound, fc);
# if D1
fprintf(stderr, "done with huntmain\n");
fprintf(stderr, "tagout is /%s/\n", tagout);
fprintf(stderr, "string out is /%s/\n", soutput);
# endif
if (fgnamp>fgnames)
{
char **fgp;
int k;
# if D1
fprintf(stderr, "were %d bad files\n", fgnamp-fgnames);
# endif
grepquery[0]=0;
for(k=0; k<nitem; k++)
{
n_strcat(grepquery, " ", sizeof(grepquery));
n_strcat(grepquery, qitem[k], sizeof(grepquery));
}
for(fgp=fgnames; fgp<fgnamp; fgp++)
{
# if D1
fprintf(stderr, "Now on %s query /%s/\n", *fgp, grepquery);
# endif
makefgrep(*fgp);
grepcall(grepquery, tagout, *fgp);
# if D1
fprintf(stderr, "tagout now /%s/\n", tagout);
# endif
}
}
restodir();
}