| 1 | # include "stdio.h" |
| 2 | # include "assert.h" |
| 3 | extern char refdir[]; |
| 4 | extern int keepold; |
| 5 | extern char *fgnames[]; |
| 6 | extern char **fgnamp; |
| 7 | FILE *fd = 0; |
| 8 | int lmaster = 500; |
| 9 | int *hfreq, hfrflg; |
| 10 | int colevel = 0; |
| 11 | int soutlen = 1000; |
| 12 | int reached = 0; |
| 13 | int iflong = 0; |
| 14 | int prfreqs = 0; |
| 15 | char usedir[100]; |
| 16 | char gfile[50]; |
| 17 | static int full = 1000; |
| 18 | static int tags = 0; |
| 19 | char *sinput, *soutput, *tagout; |
| 20 | long indexdate = 0, gdate(); |
| 21 | extern char *todir(); |
| 22 | |
| 23 | main(argc,argv) |
| 24 | char *argv[]; |
| 25 | { |
| 26 | /* read query from stdin, expect name of indexes in argv[1] */ |
| 27 | static FILE *fa, *fb, *fc; |
| 28 | char nma[100], nmb[100], nmc[100], *qitem[100], *rprog = 0; |
| 29 | char nmd[100], grepquery[256]; |
| 30 | static char oldname[30] ; |
| 31 | static int was = 0; |
| 32 | /* these pointers are unions of pointer to int and pointer to long */ |
| 33 | long *hpt; |
| 34 | unsigned *master = 0; |
| 35 | int falseflg, nhash, nitem, nfound, frtbl, kk; |
| 36 | |
| 37 | /* special wart for refpart: default is tags only */ |
| 38 | |
| 39 | while (argv[1][0] == '-') |
| 40 | { |
| 41 | switch(argv[1][1]) |
| 42 | { |
| 43 | case 'a': /* all output, incl. false drops */ |
| 44 | falseflg = 1; break; |
| 45 | case 'r': |
| 46 | argc--; argv++; |
| 47 | rprog = argv[1]; |
| 48 | break; |
| 49 | case 'F': /* put out full text */ |
| 50 | full = setfrom(argv[1][2]); |
| 51 | break; |
| 52 | case 'T': /* put out tags */ |
| 53 | tags = setfrom(argv[1][2]); |
| 54 | break; |
| 55 | case 'i': /* input in argument string */ |
| 56 | argc--; argv++; |
| 57 | sinput = argv[1]; |
| 58 | break; |
| 59 | case 's': /*text output to string */ |
| 60 | case 'o': |
| 61 | argc--; argv++; |
| 62 | soutput = argv[1]; |
| 63 | if (argv[2]<16000) |
| 64 | { |
| 65 | soutlen = argv[2]; |
| 66 | argc--; argv++; |
| 67 | } |
| 68 | break; |
| 69 | case 't': /*tag output to string */ |
| 70 | argc--; argv++; |
| 71 | tagout = argv[1]; |
| 72 | break; |
| 73 | case 'l': /* length of internal lists */ |
| 74 | argc--; argv++; |
| 75 | lmaster = atoi(argv[1]); |
| 76 | break; |
| 77 | case 'g': /* suppress fgrep search on old files */ |
| 78 | keepold = 0; |
| 79 | break; |
| 80 | case 'C': /* coordination level */ |
| 81 | colevel = atoi(argv[1]+2); |
| 82 | # if D1 |
| 83 | fprintf(stderr, "colevel set to %d\n",colevel); |
| 84 | # endif |
| 85 | break; |
| 86 | case 'P': /* print term freqs */ |
| 87 | prfreqs=1; break; |
| 88 | } |
| 89 | argc--; argv++; |
| 90 | } |
| 91 | strcpy (nma, todir(argv[1])); |
| 92 | if (was == 0 || strcmp (oldname, nma) !=0) |
| 93 | { |
| 94 | strcpy (oldname,nma); |
| 95 | strcpy (nmb, nma); strcpy (nmc, nmb); strcpy(nmd,nma); |
| 96 | strcat (nma, ".ia"); |
| 97 | strcat (nmb, ".ib"); |
| 98 | strcat (nmc, ".ic"); |
| 99 | strcat (nmd, ".id"); |
| 100 | if (was) |
| 101 | {fclose(fa); fclose(fb); fclose(fc);} |
| 102 | |
| 103 | fa = fopen(nma, "r"); |
| 104 | if (fa==NULL) |
| 105 | { |
| 106 | err("No files %s",nma); |
| 107 | exit(1); |
| 108 | } |
| 109 | fb = fopen(nmb, "r"); |
| 110 | fc = fopen(nmc, "r"); |
| 111 | was =1; |
| 112 | if (fb== NULL || fc ==NULL) |
| 113 | { |
| 114 | err("Index incomplete %s", nmb); |
| 115 | exit(1); |
| 116 | } |
| 117 | indexdate = gdate(fb); |
| 118 | fd = fopen(nmd, "r"); |
| 119 | } |
| 120 | fseek (fa, 0L, 0); |
| 121 | fread (&nhash, sizeof(nhash), 1, fa); |
| 122 | fread (&iflong, sizeof(iflong), 1, fa); |
| 123 | if(master==0) |
| 124 | master = calloc (lmaster, iflong? sizeof(long): sizeof(int)); |
| 125 | hpt = calloc(nhash, sizeof(*hpt)); |
| 126 | kk=fread( hpt, sizeof(*hpt), nhash, fa); |
| 127 | # if D1 |
| 128 | fprintf(stderr,"read %d hashes, iflong %d, nhash %d\n", kk, iflong, nhash); |
| 129 | # endif |
| 130 | _assert (kk==nhash); |
| 131 | hfreq = calloc(nhash, sizeof(*hfreq)); |
| 132 | _assert (hfreq != NULL); |
| 133 | frtbl = fread(hfreq, sizeof(*hfreq), nhash, fa); |
| 134 | hfrflg = (frtbl == nhash); |
| 135 | # if D1 |
| 136 | fprintf(stderr, "read freqs %d\n", frtbl); |
| 137 | # endif |
| 138 | |
| 139 | while (1) |
| 140 | { |
| 141 | nitem = getq(qitem); |
| 142 | if (nitem==0) continue; |
| 143 | if (nitem < 0) break; |
| 144 | nfound = doquery(hpt, nhash, fb, nitem, qitem, master); |
| 145 | # if D1 |
| 146 | fprintf(stderr,"after doquery nfound %d\n", nfound); |
| 147 | # endif |
| 148 | fgnamp=fgnames; |
| 149 | if (falseflg == 0) |
| 150 | nfound = baddrop(master, nfound, fc, nitem, qitem, rprog, full); |
| 151 | # if D1 |
| 152 | fprintf(stderr,"after baddrop nfound %d\n", nfound); |
| 153 | # endif |
| 154 | if (fgnamp>fgnames) |
| 155 | { |
| 156 | char **fgp, tgbuff[100]; |
| 157 | int k; |
| 158 | # if D1 |
| 159 | fprintf(stderr, "were %d bad files\n", fgnamp-fgnames); |
| 160 | # endif |
| 161 | grepquery[0]=0; |
| 162 | for(k=0; k<nitem; k++) |
| 163 | { |
| 164 | strcat(grepquery, " "); |
| 165 | strcat(grepquery, qitem[k]); |
| 166 | } |
| 167 | # if D1 |
| 168 | fprintf(stderr, "grepquery %s\n",grepquery); |
| 169 | # endif |
| 170 | for(fgp=fgnames; fgp<fgnamp; fgp++) |
| 171 | { |
| 172 | # if D1 |
| 173 | fprintf(stderr, "Now on %s query /%s/\n", *fgp, grepquery); |
| 174 | # endif |
| 175 | makefgrep(*fgp); |
| 176 | # if D1 |
| 177 | fprintf(stderr, "grepmade\n"); |
| 178 | # endif |
| 179 | if (tagout==0) |
| 180 | tagout=tgbuff; |
| 181 | grepcall(grepquery, tagout, *fgp); |
| 182 | # if D1 |
| 183 | fprintf(stderr, "tagout now /%s/\n", tagout); |
| 184 | # endif |
| 185 | if (full) |
| 186 | { |
| 187 | char bout[1000]; |
| 188 | findline(tagout, bout, 1000); |
| 189 | fputs(bout,stdout); |
| 190 | } |
| 191 | } |
| 192 | } |
| 193 | if (tags) |
| 194 | result (master, nfound >tags ? tags: nfound, fc); |
| 195 | } |
| 196 | } |
| 197 | |
| 198 | char * |
| 199 | todir(t) |
| 200 | char *t; |
| 201 | { |
| 202 | char *s; |
| 203 | s=t; |
| 204 | while (*s) s++; |
| 205 | while (s>=t && *s != '/') s--; |
| 206 | if (s<t) return(t); |
| 207 | *s++ = 0; |
| 208 | t = (*t ? t : "/"); |
| 209 | chdir (t); |
| 210 | strcpy (usedir,t); |
| 211 | return(s); |
| 212 | } |
| 213 | setfrom(c) |
| 214 | { |
| 215 | switch(c) |
| 216 | { |
| 217 | case 'y': case '\0': |
| 218 | default: |
| 219 | return(1000); |
| 220 | case '1': |
| 221 | case '2': case '3': case '4': case '5': |
| 222 | case '6': case '7': case '8': case '9': |
| 223 | return(c-'0'); |
| 224 | case 'n': case '0': |
| 225 | return(0); |
| 226 | } |
| 227 | } |