diplodocus.org Git - nmh/blob - docs/historical/mh-nov-1983/cmds/grep.c

   1 #include <stdio.h>
   2 #define CCHR    2
   3 #define CDOT    4
   4 #define CCL     6
   5 #define NCCL    8
   6 #define CDOL    10
   7 #define CEOF    11
   8
   9 #define STAR    01
  10
  11 #define LBSIZE  1024
  12 #define ESIZE   256
  13
  14 char    ibuf[BUFSIZ];
  15 char    expbuf[ESIZE];
  16 char    linebuf[LBSIZE+1];
  17 int     circf;
  18
  19 char cc[] = {
  20         0000,0001,0002,0003,0004,0005,0006,0007,
  21         0010,0011,0012,0013,0014,0015,0016,0017,
  22         0020,0021,0022,0023,0024,0025,0026,0027,
  23         0030,0031,0032,0033,0034,0035,0036,0037,
  24         0040,0041,0042,0043,0044,0045,0046,0047,
  25         0050,0051,0052,0053,0054,0055,0056,0057,
  26         0060,0061,0062,0063,0064,0065,0066,0067,
  27         0070,0071,0072,0073,0074,0075,0076,0077,
  28         0100,0141,0142,0143,0144,0145,0146,0147,
  29         0150,0151,0152,0153,0154,0155,0156,0157,
  30         0160,0161,0162,0163,0164,0165,0166,0167,
  31         0170,0171,0172,0133,0134,0135,0136,0137,
  32         0140,0141,0142,0143,0144,0145,0146,0147,
  33         0150,0151,0152,0153,0154,0155,0156,0157,
  34         0160,0161,0162,0163,0164,0165,0166,0167,
  35         0170,0171,0172,0173,0174,0175,0176,0177,
  36 };
  37
  38
  39 compile(astr)
  40 char *astr;
  41 {
  42         register c;
  43         register char *ep, *sp;
  44         char *lastep;
  45         int cclcnt;
  46
  47         ep = expbuf;
  48         sp = astr;
  49         if (*sp == '^') {
  50                 circf++;
  51                 sp++;
  52         }
  53         for (;;) {
  54                 if (ep >= &expbuf[ESIZE])
  55                         goto cerror;
  56                 if ((c = *sp++) != '*')
  57                         lastep = ep;
  58                 switch (c) {
  59
  60                 case '\0':
  61                         *ep++ = CEOF;
  62                         return(1);
  63
  64                 case '.':
  65                         *ep++ = CDOT;
  66                         continue;
  67
  68                 case '*':
  69                         if (lastep==0)
  70                                 goto defchar;
  71                         *lastep |= STAR;
  72                         continue;
  73
  74                 case '$':
  75                         if (*sp != '\0')
  76                                 goto defchar;
  77                         *ep++ = CDOL;
  78                         continue;
  79
  80                 case '[':
  81                         *ep++ = CCL;
  82                         *ep++ = 0;
  83                         cclcnt = 1;
  84                         if ((c = *sp++) == '^') {
  85                                 c = *sp++;
  86                                 ep[-2] = NCCL;
  87                         }
  88                         do {
  89                                 *ep++ = c;
  90                                 cclcnt++;
  91                                 if (c=='\0' || ep >= &expbuf[ESIZE])
  92                                         goto cerror;
  93                         } while ((c = *sp++) != ']');
  94                         lastep[1] = cclcnt;
  95                         continue;
  96
  97                 case '\\':
  98                         if ((c = *sp++) == '\0')
  99                                 goto cerror;
 100                 defchar:
 101                 default:
 102                         *ep++ = CCHR;
 103                         *ep++ = c;
 104                 }
 105         }
 106     cerror:
 107         return(0);
 108 }
 109
 110
 111 execute(file)
 112 char *file;
 113 {
 114         register char *p1, *p2;
 115         register c;
 116         int f, body, lf;
 117         char *ebp, *cbp;
 118
 119         if ((f = open(file, 0)) < 0) {
 120                 fprintf(stderr, "Grep: Can't open %s\n", file);
 121                 return(0);
 122         }
 123         body = 0;
 124         ebp = ibuf;
 125         cbp = ibuf;
 126         for (;;) {
 127                 p1 = linebuf;
 128                 p2 = cbp;
 129                 lf = 0;
 130                 for (;;) {
 131                         if (p2 >= ebp) {
 132                                 if ((c = read(f, ibuf, sizeof ibuf)) <= 0) {
 133                                         close(f);
 134                                         if(lf) break; /* bodyless comp! */
 135                                         return(0);
 136                                 }
 137                                 p2 = ibuf;
 138                                 ebp = ibuf+c;
 139                         }
 140                         c = *p2++;
 141                         if(lf) if(c != ' ' && c != '\t') {
 142                                 --p2;
 143                                 break;
 144                         } else
 145                                 lf = 0;
 146                         if (c == '\n')
 147                                 if(body)
 148                                         break;
 149                                 else {
 150                                         if(lf) {
 151                                                 body++;
 152                                                 break;
 153                                         }
 154                                         lf++;
 155                                         c = ' ';
 156                                 }
 157                         if(c && p1 < &linebuf[LBSIZE-1])
 158                                 *p1++ = c;
 159                 }
 160                 *p1++ = 0;
 161                 cbp = p2;
 162                 p1 = linebuf;
 163                 p2 = expbuf;
 164                 if (circf) {
 165                         if (advance(p1, p2))
 166                                 goto found;
 167                         continue;
 168                 }
 169                 /* fast check for first character */
 170                 if (*p2==CCHR) {
 171                         c = p2[1];
 172                         do {
 173                                 if(*p1==c || cc[*p1]==c)
 174                                         if (advance(p1, p2))
 175                                                 goto found;
 176                         } while (*p1++);
 177                         continue;
 178                 }
 179                 /* regular algorithm */
 180                 do {
 181                         if (advance(p1, p2))
 182                                 goto found;
 183                 } while (*p1++);
 184                 continue;
 185         found:
 186                 close(f);
 187                 return(1);
 188         }
 189 }
 190
 191
 192 advance(alp, aep)
 193 char *alp, *aep;
 194 {
 195         register char *lp, *ep, *curlp;
 196         char *nextep;
 197
 198         lp = alp;
 199         ep = aep;
 200         for (;;) switch (*ep++) {
 201
 202         case CCHR:
 203                 if (*ep++ == *lp++ || ep[-1] == cc[lp[-1]])
 204                         continue;
 205                 return(0);
 206
 207         case CDOT:
 208                 if (*lp++)
 209                         continue;
 210                 return(0);
 211
 212         case CDOL:
 213                 if (*lp==0)
 214                         continue;
 215                 return(0);
 216
 217         case CEOF:
 218                 return(1);
 219
 220         case CCL:
 221                 if (cclass(ep, *lp++, 1)) {
 222                         ep += *ep;
 223                         continue;
 224                 }
 225                 return(0);
 226
 227         case NCCL:
 228                 if (cclass(ep, *lp++, 0)) {
 229                         ep += *ep;
 230                         continue;
 231                 }
 232                 return(0);
 233
 234         case CDOT|STAR:
 235                 curlp = lp;
 236                 while (*lp++);
 237                 goto star;
 238
 239         case CCHR|STAR:
 240                 curlp = lp;
 241                 while (*lp++ == *ep || cc[lp[-1]] == *ep) ;
 242                 ep++;
 243                 goto star;
 244
 245         case CCL|STAR:
 246         case NCCL|STAR:
 247                 curlp = lp;
 248                 while (cclass(ep, *lp++, ep[-1]==(CCL|STAR)));
 249                 ep += *ep;
 250                 goto star;
 251
 252         star:
 253                 do {
 254                         lp--;
 255                         if (advance(lp, ep))
 256                                 return(1);
 257                 } while (lp > curlp);
 258                 return(0);
 259
 260         default:
 261                 printf("RE botch\n");
 262                 return(0);
 263         }
 264 }
 265
 266
 267 cclass(aset, ac, af)
 268 char *aset;
 269 {
 270         register char *set, c;
 271         register n;
 272
 273         set = aset;
 274         if ((c = ac) == 0)
 275                 return(0);
 276         n = *set++;
 277         while (--n)
 278                 if (*set++ == c)
 279                         return(af);
 280         return(!af);
 281 }