1 /*-
2 * Copyright (c) 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * Peter McIlroy.
7 *
8 * %sccs.include.redist.c%
9 */
10
11 #ifndef lint
12 static char copyright[] =
13 "@(#) Copyright (c) 1993\n\
14 The Regents of the University of California. All rights reserved.\n";
15 #endif /* not lint */
16
17 #ifndef lint
18 static char sccsid[] = "@(#)sort.c 8.1 (Berkeley) 06/06/93";
19 #endif /* not lint */
20
21 /* Sort sorts a file using an optional user-defined key.
22 * Sort uses radix sort for internal sorting, and allows
23 * a choice of merge sort and radix sort for external sorting.
24 */
25
26 #include "sort.h"
27 #include "fsort.h"
28 #include "pathnames.h"
29
30 #include <paths.h>
31 #include <signal.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <unistd.h>
35
36 int REC_D = '\n';
37 u_char d_mask[NBINS]; /* flags for rec_d, field_d, <blank> */
38 /*
39 * weight tables. Gweights is one of ascii, Rascii..
40 * modified to weight rec_d = 0 (or 255)
41 */
42 extern u_char gweights[NBINS];
43 u_char ascii[NBINS], Rascii[NBINS], RFtable[NBINS], Ftable[NBINS];
44 /*
45 * masks of ignored characters. Alltable is 256 ones
46 */
47 u_char dtable[NBINS], itable[NBINS], alltable[NBINS];
48 int SINGL_FLD = 0, SEP_FLAG = 0, UNIQUE = 0;
49 struct coldesc clist[(ND+1)*2];
50 int ncols = 0;
51 extern struct coldesc clist[(ND+1)*2];
52 extern int ncols;
53
54 char devstdin[] = _PATH_STDIN;
55 char toutpath[_POSIX_PATH_MAX];
56
57 static void cleanup __P((void));
58 static void onsig __P((int));
59 static void usage __P((char *));
60
61 int
main(argc,argv)62 main(argc, argv)
63 int argc;
64 char *argv[];
65 {
66 extern int optind;
67 extern char *optarg;
68 int (*get)();
69 int ch, i, stdinflag = 0, tmp = 0;
70 char cflag = 0, mflag = 0, nflag = 0;
71 char *outfile, *outpath = 0;
72 struct field fldtab[ND+2], *ftpos;
73 union f_handle filelist;
74 FILE *outfd;
75 memset(fldtab, 0, (ND+2)*sizeof(struct field));
76 memset(d_mask, 0, NBINS);
77 d_mask[REC_D = '\n'] = REC_D_F;
78 SINGL_FLD = SEP_FLAG = 0;
79 d_mask['\t'] = d_mask[' '] = BLANK | FLD_D;
80 ftpos = fldtab;
81 fixit(&argc, argv);
82 while ((ch = getopt(argc, argv, "bcdfik:mHno:rt:T:ux")) != EOF) {
83 switch (ch) {
84 case 'b': fldtab->flags |= BI | BT;
85 break;
86 case 'd':
87 case 'i':
88 case 'f':
89 case 'r': tmp |= optval(ch, 0);
90 if (tmp & R && tmp & F)
91 fldtab->weights = RFtable;
92 else if (tmp & F)
93 fldtab->weights = Ftable;
94 else if(tmp & R)
95 fldtab->weights = Rascii;
96 fldtab->flags |= tmp;
97 break;
98 case 'o':
99 outpath = optarg;
100 break;
101 case 'n':
102 nflag = 1;
103 setfield("1n", ++ftpos, fldtab->flags&(~R));
104 break;
105 case 'k':
106 setfield(optarg, ++ftpos, fldtab->flags);
107 break;
108 case 't':
109 if (SEP_FLAG)
110 usage("multiple field delimiters");
111 SEP_FLAG = 1;
112 d_mask[' '] &= ~FLD_D;
113 d_mask['\t'] &= ~FLD_D;
114 d_mask[*optarg] |= FLD_D;
115 if (d_mask[*optarg] & REC_D_F)
116 err(2, "record/field delimiter clash");
117 break;
118 case 'T':
119 if (REC_D != '\n')
120 usage("multiple record delimiters");
121 if ('\n' == (REC_D = *optarg))
122 break;
123 d_mask['\n'] = d_mask[' '];
124 d_mask[REC_D] = REC_D_F;
125 break;
126 case 'u':
127 UNIQUE = 1;
128 break;
129 case 'c':
130 cflag = 1;
131 break;
132 case 'm':
133 mflag = 1;
134 break;
135 case 'H':
136 PANIC = 0;
137 break;
138 case '?':
139 default: usage("");
140 }
141 }
142 if (cflag && argc > optind+1)
143 errx(2, "too many input files for -c option");
144 if (argc - 2 > optind && !strcmp(argv[argc-2], "-o")) {
145 outpath = argv[argc-1];
146 argc -= 2;
147 }
148 if (mflag && argc - optind > (MAXFCT - (16+1))*16)
149 errx(2, "too many input files for -m option");
150 for (i = optind; i < argc; i++) {
151 /* allow one occurrence of /dev/stdin */
152 if (!strcmp(argv[i], "-") || !strcmp(argv[i], devstdin)) {
153 if (stdinflag)
154 warnx("ignoring extra \"%s\" in file list",
155 argv[i]);
156 else {
157 stdinflag = 1;
158 argv[i] = devstdin;
159 }
160 } else if (ch = access(argv[i], R_OK))
161 err(2, "%s", argv[i]);
162 }
163 if (!(fldtab->flags & (I|D) || fldtab[1].icol.num)) {
164 SINGL_FLD = 1;
165 fldtab[0].icol.num = 1;
166 } else {
167 if (!fldtab[1].icol.num) {
168 fldtab[0].flags &= ~(BI|BT);
169 setfield("1", ++ftpos, fldtab->flags);
170 }
171 if (nflag)
172 fldtab[1].flags |= fldtab->flags;
173 fldreset(fldtab);
174 fldtab[0].flags &= ~F;
175 }
176 settables(fldtab[0].flags);
177 num_init();
178 fldtab->weights = gweights;
179 if (optind == argc)
180 argv[--optind] = devstdin;
181 filelist.names = argv+optind;
182 if (SINGL_FLD)
183 get = makeline;
184 else
185 get = makekey;
186 if (cflag) {
187 order(filelist, get, fldtab);
188 /* NOT REACHED */
189 }
190 if (!outpath) {
191 (void)snprintf(toutpath,
192 sizeof(toutpath), "%sstdout", _PATH_DEV);
193 outfile = outpath = toutpath;
194 } else if (!(ch = access(outpath, 0)) &&
195 strncmp(_PATH_DEV, outpath, 5)) {
196 struct sigaction act = {0, SIG_BLOCK, 6};
197 int sigtable[] = {SIGHUP, SIGINT, SIGPIPE, SIGXCPU, SIGXFSZ,
198 SIGVTALRM, SIGPROF, 0};
199 errno = 0;
200 if (access(outpath, W_OK))
201 err(2, "%s", outpath);
202 act.sa_handler = cleanup;
203 (void)snprintf(toutpath, sizeof(toutpath), "%sXXXX", outpath);
204 outfile = mktemp(toutpath);
205 if (!outfile)
206 err(2, "%s", toutpath);
207 (void)atexit(cleanup);
208 for (i = 0; sigtable[i]; ++i) /* always unlink toutpath */
209 sigaction(sigtable[i], &act, 0);
210 } else outfile = outpath;
211 if (!(outfd = fopen(outfile, "w")))
212 err(2, "%s", outfile);
213 if (mflag)
214 fmerge(-1, filelist, argc-optind, get, outfd, putline, fldtab);
215 else
216 fsort(-1, 0, filelist, argc-optind, outfd, fldtab);
217 if (outfile != outpath) {
218 if (access(outfile, 0))
219 err(2, "%s", outfile);
220 (void)unlink(outpath);
221 if (link(outfile, outpath))
222 err(2, "cannot link %s: output left in %s",
223 outpath, outfile);
224 (void)unlink(outfile);
225 }
226 exit(0);
227 }
228
229 static void
onsig(s)230 onsig(s)
231 int s;
232 {
233 cleanup();
234 exit(2); /* return 2 on error/interrupt */
235 }
236
237 static void
cleanup()238 cleanup()
239 {
240 if (toutpath[0])
241 (void)unlink(toutpath);
242 }
243
244 static void
usage(msg)245 usage(msg)
246 char *msg;
247 {
248 if (msg)
249 (void)fprintf(stderr, "sort: %s\n", msg);
250 (void)fprintf(stderr, "usage: [-o output] [-cmubdfinr] [-t char] ");
251 (void)fprintf(stderr, "[-T char] [-k keydef] ... [files]\n");
252 exit(2);
253 }
254