|
|
1.1 root 1: #ifndef lint
2: static char *sccsid = "@(#)inv1.c 4.3 (Berkeley) 5/11/89";
3: #endif
4:
5: #include <stdio.h>
6: #include <assert.h>
7: #include "pathnames.h"
8:
9: main(argc, argv)
10: char *argv[];
11: {
12: /* Make inverted file indexes. Reads a stream from mkey which
13: * gives record pointer items and keys. Generates set of files
14: * a. NHASH pointers to file b.
15: * b. lists of record numbers.
16: * c. record pointer items.
17: *
18: * these files are named xxx.ia, xxx.ib, xxx.ic;
19: * where xxx is taken from arg1.
20: * If the files exist they are updated.
21: */
22:
23: FILE *fa, *fb, *fc, *fta, *ftb, *ftc, *fd;
24: int nhash = 256;
25: int appflg = 1;
26: int keepkey = 0, pipein = 0;
27: char nma[100], nmb[100], nmc[100], com[100], nmd[100];
28: char tmpa[20], tmpb[20], tmpc[20];
29: char *remove = NULL;
30: int chatty = 0, docs, hashes, fp[2], fr, fw, pfork, pwait, status;
31: int i,j,k;
32: long keys;
33: int iflong =0;
34: char *sortdir;
35:
36: sortdir = _PATH_USRTMP;
37: while (argv[1][0] == '-')
38: {
39: switch(argv[1][1])
40: {
41: case 'h': /* size of hash table */
42: nhash = atoi (argv[1]+2);
43: break;
44: case 'n': /* new, don't append */
45: appflg=0;
46: break;
47: case 'a': /* append to old file */
48: appflg=1;
49: break;
50: case 'v': /* verbose output */
51: chatty=1;
52: break;
53: case 'd': /* keep keys on file .id for check on searching */
54: keepkey=1;
55: break;
56: case 'p': /* pipe into sort (saves space, costs time)*/
57: pipein = 1;
58: break;
59: case 'i': /* input is on file, not stdin */
60: close(0);
61: if (open(argv[2], 0) != 0)
62: err("Can't read input %s", argv[2]);
63: if (argv[1][2]=='u') /* unlink */
64: remove = argv[2];
65: argc--;
66: argv++;
67: break;
68: }
69: argc--;
70: argv++;
71: }
72: strcpy (nma, argc >= 2 ? argv[1] : "Index");
73: strcpy (nmb, nma);
74: strcpy (nmc, nma);
75: strcpy (nmd, nma);
76: strcat (nma, ".ia");
77: strcat (nmb, ".ib");
78: strcat (nmc, ".ic");
79: strcat (nmd, ".id");
80:
81: sprintf(tmpa, "junk%di", getpid());
82: if (pipein)
83: {
84: pipe(fp);
85: fr=fp[0];
86: fw=fp[1];
87: if ( (pfork=fork()) == 0)
88: {
89: close(fw);
90: close(0);
91: _assert(dup(fr)==0);
92: close(fr);
93: execl(_PATH_SORT, "sort", "-T", sortdir, "-o", tmpa, 0);
94: _assert(0);
95: }
96: _assert(pfork!= -1);
97: close(fr);
98: fta = fopen(_PATH_DEVNULL, "w");
99: close(fta->_file);
100: fta->_file = fw;
101: }
102: else /* use tmp file */
103: {
104: fta = fopen(tmpa, "w");
105: _assert (fta != NULL);
106: }
107: fb = 0;
108: if (appflg )
109: {
110: if (fb = fopen(nmb, "r"))
111: {
112: sprintf(tmpb, "junk%dj", getpid());
113: ftb = fopen(tmpb, "w");
114: if (ftb==NULL)
115: err("Can't get scratch file %s",tmpb);
116: nhash = recopy(ftb, fb, fopen(nma, "r"));
117: fclose(ftb);
118: }
119: else
120: appflg=0;
121: }
122: fc = fopen(nmc, appflg ? "a" : "w");
123: fd = keepkey ? fopen(nmd, "w") : 0;
124: docs = newkeys(fta, stdin, fc, nhash, fd, &iflong);
125: fclose(stdin);
126: if (remove != NULL)
127: unlink(remove);
128: fclose(fta);
129: if (pipein)
130: {
131: pwait = wait(&status);
132: printf("pfork %o pwait %o status %d\n",pfork,pwait,status);
133: _assert(pwait==pfork);
134: _assert(status==0);
135: }
136: else
137: {
138: sprintf(com, "sort -T %s %s -o %s", sortdir, tmpa, tmpa);
139: system(com);
140: }
141: if (appflg)
142: {
143: sprintf(tmpc, "junk%dk", getpid());
144: sprintf(com, "mv %s %s", tmpa, tmpc);
145: system(com);
146: sprintf(com, "sort -T %s -m %s %s -o %s", sortdir,
147: tmpb, tmpc, tmpa);
148: system(com);
149: }
150: fta = fopen(tmpa, "r");
151: fa = fopen(nma, "w");
152: fb = fopen(nmb, "w");
153: whash(fta, fa, fb, nhash, iflong, &keys, &hashes);
154: fclose(fta);
155: # ifndef D1
156: unlink(tmpa);
157: # endif
158: if (appflg)
159: {
160: unlink(tmpb);
161: unlink(tmpc);
162: }
163: if (chatty)
164:
165: printf ("%ld key occurrences, %d hashes, %d docs\n",
166: keys, hashes, docs);
167: }
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.