1 /* $NetBSD: fields.c,v 1.12 2003/04/09 09:30:40 jdolecek Exp $ */ 2 3 /*- 4 * Copyright (c) 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Peter McIlroy. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the University of 21 * California, Berkeley and its contributors. 22 * 4. Neither the name of the University nor the names of its contributors 23 * may be used to endorse or promote products derived from this software 24 * without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 36 * SUCH DAMAGE. 37 */ 38 39 /* Subroutines to generate sort keys. */ 40 41 #include "sort.h" 42 43 #ifndef lint 44 __RCSID("$NetBSD: fields.c,v 1.12 2003/04/09 09:30:40 jdolecek Exp $"); 45 __SCCSID("@(#)fields.c 8.1 (Berkeley) 6/6/93"); 46 #endif /* not lint */ 47 48 #define SKIP_BLANKS(ptr) { \ 49 if (BLANK & d_mask[*(ptr)]) \ 50 while (BLANK & d_mask[*(++(ptr))]); \ 51 } 52 53 #define NEXTCOL(pos) { \ 54 if (!SEP_FLAG) \ 55 while (BLANK & l_d_mask[*(++pos)]); \ 56 while (!((FLD_D | REC_D_F) & l_d_mask[*++pos])); \ 57 } 58 59 static u_char *enterfield __P((u_char *, u_char *, struct field *, int)); 60 static u_char *number __P((u_char *, u_char *, u_char *, u_char *, int)); 61 62 #define DECIMAL '.' 63 #define OFFSET 128 64 65 u_char TENS[10]; /* TENS[0] = REC_D <= 128 ? 130 - '0' : 2 -'0'... */ 66 u_char NEGTENS[10]; /* NEGTENS[0] = REC_D <= 128 ? 126 + '0' : 252 +'0' */ 67 u_char *OFF_TENS, *OFF_NTENS; /* TENS - '0', NEGTENS - '0' */ 68 u_char fnum[NBINS], rnum[NBINS]; 69 70 /* 71 * constructs sort key with leading recheader, followed by the key, 72 * followed by the original line. 73 */ 74 length_t 75 enterkey(keybuf, line, size, fieldtable) 76 RECHEADER *keybuf; /* pointer to start of key */ 77 DBT *line; 78 int size; 79 struct field fieldtable[]; 80 { 81 int i; 82 u_char *l_d_mask; 83 u_char *lineend, *pos; 84 u_char *endkey, *keypos; 85 struct coldesc *clpos; 86 int col = 1; 87 struct field *ftpos; 88 l_d_mask = d_mask; 89 pos = (u_char *) line->data - 1; 90 lineend = (u_char *) line->data + line->size-1; 91 /* don't include rec_delimiter */ 92 93 for (i = 0; i < ncols; i++) { 94 clpos = clist + i; 95 for (; (col < clpos->num) && (pos < lineend); col++) { 96 NEXTCOL(pos); 97 } 98 if (pos >= lineend) 99 break; 100 clpos->start = SEP_FLAG ? pos + 1 : pos; 101 NEXTCOL(pos); 102 clpos->end = pos; 103 col++; 104 if (pos >= lineend) { 105 clpos->end = lineend; 106 i++; 107 break; 108 } 109 } 110 for (; i <= ncols; i++) 111 clist[i].start = clist[i].end = lineend; 112 if (clist[0].start < (u_char *) line->data) 113 clist[0].start++; 114 115 keypos = keybuf->data; 116 endkey = (u_char *) keybuf + size - line->size; 117 for (ftpos = fieldtable + 1; ftpos->icol.num; ftpos++) 118 if ((keypos = enterfield(keypos, endkey, ftpos, 119 fieldtable->flags)) == NULL) 120 return (1); 121 122 keybuf->offset = keypos - keybuf->data; 123 keybuf->length = keybuf->offset + line->size; 124 if (keybuf->length + sizeof(TRECHEADER) > size) { 125 /* line too long for buffer */ 126 return (1); 127 } 128 129 /* 130 * Make [s]radixsort() only sort by relevant part of key if: 131 * 1. we want to choose unique items by relevant field[s] 132 * 2. we want stable sort and so the items should be sorted only by 133 * the relevant field[s] 134 */ 135 if (UNIQUE || (stable_sort && keybuf->offset < line->size)) 136 keypos[-1] = REC_D; 137 138 memcpy(keybuf->data + keybuf->offset, line->data, line->size); 139 return (0); 140 } 141 142 /* 143 * constructs a field (as defined by -k) within a key 144 */ 145 static u_char * 146 enterfield(tablepos, endkey, cur_fld, gflags) 147 struct field *cur_fld; 148 u_char *tablepos, *endkey; 149 int gflags; 150 { 151 u_char *start, *end, *lineend, *mask, *lweight; 152 struct column icol, tcol; 153 u_int flags; 154 u_int Rflag; 155 156 icol = cur_fld->icol; 157 tcol = cur_fld->tcol; 158 flags = cur_fld->flags; 159 start = icol.p->start; 160 lineend = clist[ncols].end; 161 if (flags & BI) 162 SKIP_BLANKS(start); 163 start += icol.indent; 164 start = min(start, lineend); 165 166 if (!tcol.num) 167 end = lineend; 168 else { 169 if (tcol.indent) { 170 end = tcol.p->start; 171 if (flags & BT) 172 SKIP_BLANKS(end); 173 end += tcol.indent; 174 end = min(end, lineend); 175 } else 176 end = tcol.p->end; 177 } 178 179 if (flags & N) { 180 Rflag = (gflags & R ) ^ (flags & R) ? 1 : 0; 181 return number(tablepos, endkey, start, end, Rflag); 182 } 183 184 mask = cur_fld->mask; 185 lweight = cur_fld->weights; 186 for (; start < end; start++) 187 if (mask[*start]) { 188 if (*start <= 1) { 189 if (tablepos+2 >= endkey) 190 return (NULL); 191 *tablepos++ = lweight[1]; 192 *tablepos++ = lweight[*start ? 2 : 1]; 193 } else { 194 if (tablepos+1 >= endkey) 195 return (NULL); 196 *tablepos++ = lweight[*start]; 197 } 198 } 199 *tablepos++ = lweight[0]; 200 return (tablepos == endkey ? NULL : tablepos); 201 } 202 203 /* Uses the first bin to assign sign, expsign, 0, and the first 204 * 61 out of the exponent ( (254 - 3 origins - 4 over/underflows)/4 = 61 ). 205 * When sorting in forward order: 206 * use (0-99) -> (130->240) for sorting the mantissa if REC_D <=128; 207 * else use (0-99)->(2-102). 208 * If the exponent is >=61, use another byte for each additional 253 209 * in the exponent. Cutoff is at 567. 210 * To avoid confusing the exponent and the mantissa, use a field delimiter 211 * if the exponent is exactly 61, 61+252, etc--this is ok, since it's the 212 * only time a field delimiter can come in that position. 213 * Reverse order is done analagously. 214 */ 215 216 static u_char * 217 number(pos, bufend, line, lineend, Rflag) 218 u_char *line, *pos, *bufend, *lineend; 219 int Rflag; 220 { 221 int or_sign, parity = 0; 222 int expincr = 1, exponent = -1; 223 int bite, expsign = 1, sign = 1; 224 u_char lastvalue, *nonzero, *tline, *C_TENS; 225 u_char *nweights; 226 227 if (Rflag) 228 nweights = rnum; 229 else 230 nweights = fnum; 231 if (pos > bufend - 8) 232 return (NULL); 233 /* 234 * or_sign sets the sort direction: 235 * (-r: +/-)(sign: +/-)(expsign: +/-) 236 */ 237 or_sign = sign ^ expsign ^ Rflag; 238 SKIP_BLANKS(line); 239 if (*line == '-') { /* set the sign */ 240 or_sign ^= 1; 241 sign = 0; 242 line++; 243 } 244 /* eat initial zeroes */ 245 for (; *line == '0' && line < lineend; line++) 246 ; 247 /* calculate exponents < 0 */ 248 if (*line == DECIMAL) { 249 exponent = 1; 250 while (*++line == '0' && line < lineend) 251 exponent++; 252 expincr = 0; 253 expsign = 0; 254 } 255 /* next character better be a digit */ 256 if (*line < '1' || *line > '9' || line >= lineend) { 257 *pos++ = nweights[127]; 258 return (pos); 259 } 260 if (expincr) { 261 for (tline = line-1; *++tline >= '0' && 262 *tline <= '9' && tline < lineend;) 263 exponent++; 264 } 265 if (exponent > 567) { 266 *pos++ = nweights[sign ? (expsign ? 254 : 128) 267 : (expsign ? 0 : 126)]; 268 warnx("exponent out of bounds"); 269 return (pos); 270 } 271 bite = min(exponent, 61); 272 *pos++ = nweights[(sign) ? (expsign ? 189+bite : 189-bite) 273 : (expsign ? 64-bite : 64+bite)]; 274 if (bite >= 61) { 275 do { 276 exponent -= bite; 277 bite = min(exponent, 254); 278 *pos++ = nweights[or_sign ? 254-bite : bite]; 279 } while (bite == 254); 280 } 281 C_TENS = or_sign ? OFF_NTENS : OFF_TENS; 282 for (; line < lineend; line++) { 283 if (*line >= '0' && *line <= '9') { 284 if (parity) { 285 *pos++ = C_TENS[lastvalue] + (or_sign ? - *line 286 : *line); 287 if (pos == bufend) 288 return (NULL); 289 if (*line != '0' || lastvalue != '0') 290 nonzero = pos; 291 } else 292 lastvalue = *line; 293 parity ^= 1; 294 } else if(*line == DECIMAL) { 295 if(!expincr) /* a decimal already occurred once */ 296 break; 297 expincr = 0; 298 } else 299 break; 300 } 301 if (parity && lastvalue != '0') { 302 *pos++ = or_sign ? OFF_NTENS[lastvalue] - '0' : 303 OFF_TENS[lastvalue] + '0'; 304 } else 305 pos = nonzero; 306 if (pos > bufend-1) 307 return (NULL); 308 *pos++ = or_sign ? nweights[254] : nweights[0]; 309 return (pos); 310 } 311 312 /* This forces a gap around the record delimiter 313 * Thus fnum has vaues over (0,254) -> ((0,REC_D-1),(REC_D+1,255)); 314 * rnum over (0,254) -> (255,REC_D+1),(REC_D-1,0)) 315 */ 316 void 317 num_init() 318 { 319 int i; 320 TENS[0] = REC_D <=128 ? 130 - '0' : 2 - '0'; 321 NEGTENS[0] = REC_D <=128 ? 126 + '0' : 254 + '0'; 322 OFF_TENS = TENS - '0'; 323 OFF_NTENS = NEGTENS - '0'; 324 for (i = 1; i < 10; i++) { 325 TENS[i] = TENS[i - 1] + 10; 326 NEGTENS[i] = NEGTENS[i - 1] - 10; 327 } 328 for (i = 0; i < REC_D; i++) { 329 fnum[i] = i; 330 rnum[255 - i] = i; 331 } 332 for (i = REC_D; i <255; i++) { 333 fnum[i] = i + 1; 334 rnum[255 - i] = i - 1; 335 } 336 } 337