xref: /original-bsd/old/refer/hunt/hunt1.c (revision e59fb703)
1 /*-
2  * %sccs.include.proprietary.c%
3  */
4 
5 #ifndef lint
6 static char sccsid[] = "@(#)hunt1.c	4.4 (Berkeley) 04/18/91";
7 #endif /* not lint */
8 
9 # include <stdio.h>
10 # include <assert.h>
11 extern char refdir[];
12 extern int keepold;
13 extern char *fgnames[];
14 extern char **fgnamp;
15 FILE *fd =NULL;
16 int lmaster =500;
17 int *hfreq, hfrflg;
18 int colevel =0;
19 int measure=0;
20 int soutlen =1000;
21 int reached =0;
22 int iflong =0;
23 int prfreqs =0;
24 char usedir[100];
25 char * calloc();
26 char * todir();
27 char gfile[50];
28 static int full =1000;
29 static int tags =0;
30 char *sinput, *soutput, *tagout;
31 long indexdate =0, gdate();
32 
33 main(argc,argv)
34 char *argv[];
35 {
36 	/* read query from stdin, expect name of indexes in argv[1] */
37 	static FILE *fa, *fb, *fc;
38 	char nma[100], nmb[100], nmc[100], *qitem[100], *rprog = NULL;
39 	char nmd[100], grepquery[256];
40 	static char oldname[30] ;
41 	static int was =0;
42 	/* these pointers are unions of pointer to int and pointer to long */
43 	long *hpt;
44 	unsigned *master =0;
45 	int falseflg, nhash, nitem, nfound, frtbl, kk;
46 
47 	/* special wart for refpart: default is tags only */
48 
49 	while (argv[1][0] == '-')
50 	{
51 		switch(argv[1][1])
52 		{
53 		case 'a': /* all output, incl. false drops */
54 			falseflg = 1;
55 			break;
56 		case 'r':
57 			argc--;
58 			argv++;
59 			rprog = argv[1];
60 			break;
61 		case 'F': /* put out full text */
62 			full = setfrom(argv[1][2]);
63 			break;
64 		case 'T': /* put out tags */
65 			tags = setfrom(argv[1][2]);
66 			break;
67 		case 'i': /* input in argument string */
68 			argc--;
69 			argv++;
70 			sinput = argv[1];
71 			break;
72 		case 's': /*text output to string */
73 		case 'o':
74 			argc--;
75 			argv++;
76 			soutput = argv[1];
77 			if ((int) argv[2]<16000)
78 			{
79 				soutlen = (int)argv[2];
80 				argc--;
81 				argv++;
82 			}
83 			break;
84 		case 't': /*tag output to string */
85 			argc--;
86 			argv++;
87 			tagout = argv[1];
88 			break;
89 		case 'l': /* length of internal lists */
90 			argc--;
91 			argv++;
92 			lmaster = atoi(argv[1]);
93 			break;
94 		case 'g': /* suppress fgrep search on old files */
95 			keepold = 0;
96 			break;
97 		case 'C': /* coordination level */
98 			colevel = atoi(argv[1]+2);
99 # if D1
100 			fprintf(stderr, "colevel set to %d\n",colevel);
101 # endif
102 			break;
103 		case 'P': /* print term freqs */
104 			prfreqs=1;
105 			break;
106 		case 'm':
107 			measure=1;
108 			break;
109 		}
110 		argc--;
111 		argv++;
112 	}
113 	strcpy (nma, todir(argv[1]));
114 	if (was == 0 || strcmp (oldname, nma) !=0)
115 	{
116 		strcpy (oldname,nma);
117 		strcpy (nmb, nma);
118 		strcpy (nmc, nmb);
119 		strcpy(nmd,nma);
120 		strcat (nma, ".ia");
121 		strcat (nmb, ".ib");
122 		strcat (nmc, ".ic");
123 		strcat (nmd, ".id");
124 		if (was)
125 		{
126 			fclose(fa);
127 			fclose(fb);
128 			fclose(fc);
129 		}
130 
131 		fa = fopen(nma, "r");
132 		if (fa==NULL)
133 		{
134 			strcpy(*fgnamp++ = calloc(strlen(oldname)+2,1), oldname);
135 			fb=NULL;
136 			goto search;
137 		}
138 		fb = fopen(nmb, "r");
139 		fc = fopen(nmc, "r");
140 		was =1;
141 		if (fb== NULL || fc ==NULL)
142 		{
143 			err("Index incomplete %s", nmb);
144 			exit(1);
145 		}
146 		indexdate = gdate(fb);
147 		fd = fopen(nmd, "r");
148 	}
149 	fseek (fa, 0L, 0);
150 	fread (&nhash, sizeof(nhash), 1, fa);
151 	fread (&iflong, sizeof(iflong), 1, fa);
152 	if(master==0)
153 		master = (unsigned *) calloc (lmaster, iflong? sizeof(long): sizeof(unsigned));
154 	hpt = (long *) calloc(nhash, sizeof(*hpt));
155 	kk=fread( hpt, sizeof(*hpt), nhash, fa);
156 # if D1
157 	fprintf(stderr,"read %d hashes, iflong %d, nhash %d\n", kk, iflong, nhash);
158 # endif
159 	_assert (kk==nhash);
160 	hfreq = (int *) calloc(nhash, sizeof(*hfreq));
161 	_assert (hfreq != NULL);
162 	frtbl = fread(hfreq, sizeof(*hfreq), nhash, fa);
163 	hfrflg = (frtbl == nhash);
164 # if D1
165 	fprintf(stderr, "read freqs %d\n", frtbl);
166 # endif
167 
168 search:
169 	while (1)
170 	{
171 		nitem = getq(qitem);
172 		if (measure) tick();
173 		if (nitem==0) continue;
174 		if (nitem < 0) break;
175 		if (tagout) tagout[0]=0;
176 		if (fb!=NULL)
177 		{
178 			nfound = doquery(hpt, nhash, fb, nitem, qitem, master);
179 # if D1
180 			fprintf(stderr,"after doquery nfound %d\n", nfound);
181 # endif
182 			fgnamp=fgnames;
183 			if (falseflg == 0)
184 				nfound = baddrop(master, nfound, fc, nitem, qitem, rprog, full);
185 # if D1
186 			fprintf(stderr,"after baddrop nfound %d\n", nfound);
187 # endif
188 		}
189 		if (fgnamp>fgnames)
190 		{
191 			char **fgp, tgbuff[100];
192 			int k;
193 # if D1
194 			fprintf(stderr, "were %d bad files\n", fgnamp-fgnames);
195 # endif
196 			grepquery[0]=0;
197 			for(k=0; k<nitem; k++)
198 			{
199 				strcat(grepquery, " ");
200 				strcat(grepquery, qitem[k]);
201 			}
202 # if D1
203 			fprintf(stderr, "grepquery %s\n",grepquery);
204 # endif
205 			for(fgp=fgnames; fgp<fgnamp; fgp++)
206 			{
207 # if D1
208 				fprintf(stderr, "Now on %s query /%s/\n", *fgp, grepquery);
209 # endif
210 				makefgrep(*fgp);
211 # if D1
212 				fprintf(stderr, "grepmade\n");
213 # endif
214 				if (tagout==0)
215 					tagout=tgbuff;
216 				grepcall(grepquery, tagout, *fgp);
217 # if D1
218 				fprintf(stderr, "tagout now /%s/\n", tagout);
219 # endif
220 				if (full)
221 				{
222 					char bout[1000];
223 					char *tagp;
224 					char *oldtagp;
225 					tagp = tagout;
226 					while (*tagp) {
227 						oldtagp = tagp;
228 						while (*tagp && (*tagp != '\n'))
229 							tagp++;
230 						if (*tagp)
231 							tagp++;
232 						findline(oldtagp, bout, 1000);
233 						fputs(bout,stdout);
234 					}
235 				}
236 			}
237 		}
238 		if (tags)
239 			result (master, nfound >tags ? tags: nfound, fc);
240 		if (measure) tock();
241 	}
242 }
243 
244 char *
245 todir(t)
246 char *t;
247 {
248 	char *s;
249 	s=t;
250 	while (*s) s++;
251 	while (s>=t && *s != '/') s--;
252 	if (s<t) return(t);
253 	*s++ = 0;
254 	t = (*t ? t : "/");
255 	chdir (t);
256 	strcpy (usedir,t);
257 	return(s);
258 }
259 setfrom(c)
260 {
261 	switch(c)
262 	{
263 	case 'y':
264 	case '\0':
265 	default:
266 		return(1000);
267 	case '1':
268 	case '2':
269 	case '3':
270 	case '4':
271 	case '5':
272 	case '6':
273 	case '7':
274 	case '8':
275 	case '9':
276 		return(c-'0');
277 	case 'n':
278 	case '0':
279 		return(0);
280 	}
281 }
282