xref: /original-bsd/contrib/sort/init.c (revision c3e32dec)
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 sccsid[] = "@(#)init.c	8.1 (Berkeley) 06/06/93";
13 #endif /* not lint */
14 
15 #include "sort.h"
16 
17 #include <ctype.h>
18 #include <string.h>
19 
20 extern struct coldesc clist[(ND+1)*2];
21 extern int ncols;
22 u_char gweights[NBINS];
23 
24 /*
25  * clist (list of columns which correspond to one or more icol or tcol)
26  * is in increasing order of columns.
27  * Fields are kept in increasing order of fields.
28  */
29 
30 /*
31  * keep clist in order--inserts a column in a sorted array
32  */
33 static void
34 insertcol(field)
35 	struct field *field;
36 {
37 	int i;
38 	for (i = 0; i < ncols; i++)
39 		if (field->icol.num <= clist[i].num)
40 			break;
41 	if (field->icol.num != clist[i].num) {
42 		memmove(clist+i+1, clist+i, sizeof(COLDESC)*(ncols-i));
43 		clist[i].num = field->icol.num;
44 		ncols++;
45 	}
46 	if (field->tcol.num && field->tcol.num != field->icol.num) {
47 		for (i = 0; i < ncols; i++)
48 			if (field->tcol.num <= clist[i].num)
49 				break;
50 		if (field->tcol.num != clist[i].num) {
51 			memmove(clist+i+1, clist+i,sizeof(COLDESC)*(ncols-i));
52 			clist[i].num = field->tcol.num;
53 			ncols++;
54 		}
55 	}
56 }
57 
58 /*
59  * matches fields with the appropriate columns--n^2 but who cares?
60  */
61 void
62 fldreset(fldtab)
63 	struct field *fldtab;
64 {
65 	int i;
66 	fldtab[0].tcol.p = clist+ncols-1;
67 	for (++fldtab; fldtab->icol.num; ++fldtab) {
68 		for (i = 0; fldtab->icol.num != clist[i].num; i++);
69 		fldtab->icol.p = clist + i;
70 		if (!fldtab->tcol.num)
71 			continue;
72 		for (i = 0; fldtab->tcol.num != clist[i].num; i++);
73 		fldtab->tcol.p = clist + i;
74 	}
75 }
76 
77 /*
78  * interprets a column in a -k field
79  */
80 char *
81 setcolumn(pos, cur_fld, gflag)
82 	char *pos;
83 	struct field *cur_fld;
84 	int gflag;
85 {
86 	struct column *col;
87 	int tmp;
88 	col = cur_fld->icol.num ? (&(*cur_fld).tcol) : (&(*cur_fld).icol);
89 	pos += sscanf(pos, "%d", &(col->num));
90 	while (isdigit(*pos))
91 		pos++;
92 	if (col->num <= 0 && !(col->num == 0 && col == &(cur_fld->tcol)))
93 		errx(2, "field numbers must be positive");
94 	if (*pos == '.') {
95 		if (!col->num)
96 			errx(2, "cannot indent end of line");
97 		pos += sscanf(++pos, "%d", &(col->indent));
98 		while (isdigit(*pos))
99 			pos++;
100 		if (&cur_fld->icol == col)
101 			col->indent--;
102 		if (col->indent < 0)
103 			errx(2, "illegal offset");
104 	}
105 	if (optval(*pos, cur_fld->tcol.num))
106 		while (tmp = optval(*pos, cur_fld->tcol.num)) {
107 			cur_fld->flags |= tmp;
108 			pos++;
109 	}
110 	if (cur_fld->icol.num == 0)
111 		cur_fld->icol.num = 1;
112 	return (pos);
113 }
114 
115 int
116 setfield(pos, cur_fld, gflag)
117 	char *pos;
118 	struct field *cur_fld;
119 	int gflag;
120 {
121 	static int nfields = 0;
122 	int tmp;
123 	char *setcolumn();
124 	if (++nfields == ND)
125 		errx(2, "too many sort keys. (Limit is %d)", ND-1);
126 	cur_fld->weights = ascii;
127 	cur_fld->mask = alltable;
128 	pos = setcolumn(pos, cur_fld, gflag);
129 	if (*pos == '\0')			/* key extends to EOL. */
130 		cur_fld->tcol.num = 0;
131 	else {
132 		if (*pos != ',')
133 			errx(2, "illegal field descriptor");
134 		setcolumn((++pos), cur_fld, gflag);
135 	}
136 	if (!cur_fld->flags)
137 		cur_fld->flags = gflag;
138 	tmp = cur_fld->flags;
139 
140 	/*
141 	 * Assign appropriate mask table and weight table.
142 	 * If the global weights are reversed, the local field
143 	 * must be "re-reversed".
144 	 */
145 	if (((tmp & R) ^ (gflag & R)) && tmp & F)
146 		cur_fld->weights = RFtable;
147 	else if (tmp & F)
148 		cur_fld->weights = Ftable;
149 	else if ((tmp & R) ^ (gflag & R))
150 		cur_fld->weights = Rascii;
151 	if (tmp & I)
152 		cur_fld->mask = itable;
153 	else if (tmp & D)
154 		cur_fld->mask = dtable;
155 	cur_fld->flags |= (gflag & (BI | BT));
156 	if (!cur_fld->tcol.indent)	/* BT has no meaning at end of field */
157 		cur_fld->flags &= (D|F|I|N|R|BI);
158 	if (cur_fld->tcol.num && !(!(cur_fld->flags & BI)
159 	    && cur_fld->flags & BT) && (cur_fld->tcol.num <= cur_fld->icol.num
160 	    && cur_fld->tcol.indent < cur_fld->icol.indent))
161 		errx(2, "fields out of order");
162 	insertcol(cur_fld);
163 	return (cur_fld->tcol.num);
164 }
165 
166 int
167 optval(desc, tcolflag)
168 	int desc, tcolflag;
169 {
170 	switch(desc) {
171 		case 'b':
172 			if (!tcolflag)
173 				return(BI);
174 			else
175 				return(BT);
176 		case 'd': return(D);
177 		case 'f': return(F);
178 		case 'i': return(I);
179 		case 'n': return(N);
180 		case 'r': return(R);
181 		default:  return(0);
182 	}
183 }
184 
185 void
186 fixit(argc, argv)
187 	int *argc;
188 	char **argv;
189 {
190 	int i, j, v, w, x;
191 	static char vbuf[ND*20], *vpos, *tpos;
192 	vpos = vbuf;
193 
194 	for (i = 1; i < *argc; i++) {
195 		if (argv[i][0] == '+') {
196 			tpos = argv[i]+1;
197 			argv[i] = vpos;
198 			vpos += sprintf(vpos, "-k");
199 			tpos += sscanf(tpos, "%d", &v);
200 			while (isdigit(*tpos))
201 				tpos++;
202 			vpos += sprintf(vpos, "%d", v+1);
203 			if (*tpos == '.') {
204 				tpos += sscanf(++tpos, "%d", &x);
205 				vpos += sprintf(vpos, ".%d", x+1);
206 			}
207 			while (*tpos)
208 				*vpos++ = *tpos++;
209 			vpos += sprintf(vpos, ",");
210 			if (argv[i+1] &&
211 			    argv[i+1][0] == '-' && isdigit(argv[i+1][1])) {
212 				tpos = argv[i+1] + 1;
213 				tpos += sscanf(tpos, "%d", &w);
214 				while (isdigit(*tpos))
215 					tpos++;
216 				x = 0;
217 				if (*tpos == '.') {
218 					tpos += sscanf(++tpos, "%d", &x);
219 					while (isdigit(*tpos))
220 						*tpos++;
221 				}
222 				if (x) {
223 					vpos += sprintf(vpos, "%d", w+1);
224 					vpos += sprintf(vpos, ".%d", x);
225 				} else
226 					vpos += sprintf(vpos, "%d", w);
227 				while (*tpos)
228 					*vpos++ = *tpos++;
229 				for (j= i+1; j < *argc; j++)
230 					argv[j] = argv[j+1];
231 				*argc -= 1;
232 			}
233 		}
234 	}
235 }
236 
237 /*
238  * ascii, Rascii, Ftable, and RFtable map
239  * REC_D -> REC_D;  {not REC_D} -> {not REC_D}.
240  * gweights maps REC_D -> (0 or 255); {not REC_D} -> {not gweights[REC_D]}.
241  * Note: when sorting in forward order, to encode character zero in a key,
242  * use \001\001; character 1 becomes \001\002.  In this case, character 0
243  * is reserved for the field delimiter.  Analagously for -r (fld_d = 255).
244  * Note: this is only good for ASCII sorting.  For different LC 's,
245  * all bets are off.  See also num_init in number.c
246  */
247 void
248 settables(gflags)
249 	int gflags;
250 {
251 	u_char *wts;
252 	int i, incr;
253 	for (i=0; i < 256; i++) {
254 		ascii[i] = i;
255 		if (i > REC_D && i < 255 - REC_D+1)
256 			Rascii[i] = 255 - i + 1;
257 		else
258 			Rascii[i] = 255 - i;
259 		if (islower(i)) {
260 			Ftable[i] = Ftable[i- ('a' -'A')];
261 			RFtable[i] = RFtable[i - ('a' - 'A')];
262 		} else if (REC_D>= 'A' && REC_D < 'Z' && i < 'a' && i > REC_D) {
263 			Ftable[i] = i + 1;
264 			RFtable[i] = Rascii[i] - 1;
265 		} else {
266 			Ftable[i] = i;
267 			RFtable[i] = Rascii[i];
268 		}
269 		alltable[i] = 1;
270 		if (i == '\n' || isprint(i))
271 			itable[i] = 1;
272 		else itable[i] = 0;
273 		if (i == '\n' || i == '\t' || i == ' ' || isalnum(i))
274 			dtable[i] = 1;
275 		else dtable[i] = 0;
276 	}
277 	Rascii[REC_D] = RFtable[REC_D] = REC_D;
278 	if (REC_D >= 'A' && REC_D < 'Z')
279 		++Ftable[REC_D + ('a' - 'A')];
280 	if (gflags & R && (!(gflags & F) || !SINGL_FLD))
281 		wts = Rascii;
282 	else if (!(gflags & F) || !SINGL_FLD)
283 		wts = ascii;
284 	else if (gflags & R)
285 		wts = RFtable;
286 	else
287 		wts = Ftable;
288 	memmove(gweights, wts, sizeof(gweights));
289 	incr = (gflags & R) ? -1 : 1;
290 	for (i = 0; i < REC_D; i++)
291 		gweights[i] += incr;
292 	gweights[REC_D] = ((gflags & R) ? 255 : 0);
293 	if (SINGL_FLD && gflags & F) {
294 		for (i = 0; i < REC_D; i++) {
295 			ascii[i] += incr;
296 			Rascii[i] += incr;
297 		}
298 		ascii[REC_D] = Rascii[REC_D] = gweights[REC_D];
299 	}
300 }
301