commands/join/join.c

   1 /* join - relation data base operator   Author:  Saeko Hirabayashi */
   2
   3 /* Written by Saeko Hirabayashi, 1989.
   4  * 1992-01-28 Modified by Kouichi Hirabayashi to add some POSIX1003.2 options.
   5  *
   6  * This a free program.
   7  */
   8
   9 #include <string.h>
  10 #include <stdio.h>
  11
  12 #define MAXFLD  200             /* maximum # of fields to accept */
  13
  14 int main(int argc, char **argv);
  15 void error(char *s, char *t);
  16 void usage(void);
  17 void match(void);
  18 void f1_only(void);
  19 void f2_only(void);
  20 void output(int flag);
  21 void outfld(int file);
  22 void outputf(int flag);
  23 int compare(void);
  24 int get1(void);
  25 int get2(int back);
  26 int getrec(int file);
  27 int split(int file);
  28 int atoi(char *str);
  29 int exit(int val);
  30 FILE * efopen(char *file, char *mode);
  31 void(*outfun) (int file);
  32
  33 #define F1      1
  34 #define F2      2
  35 #define SEP     (sep ? sep : ' ')
  36
  37 FILE *fp[2];                    /* file pointer for file1 and file2 */
  38 long head;                      /* head of the current (same)key group of the
  39                                  * file2 */
  40
  41 char buf[2][BUFSIZ];            /* input buffer for file1 and file2 */
  42 char *fld[2][MAXFLD];           /* field vector for file1 and file2 */
  43 int nfld[2];                    /* # of fields for file1 and file2 */
  44
  45 int kpos[2];                    /* key field position for file1 and file2
  46                                  * (from 0) */
  47 char oldkey[BUFSIZ];            /* previous key of the file1 */
  48
  49 struct {                        /* output list by -o option */
  50   int o_file;                   /* file #: 0 or 1 */
  51   int o_field;                  /* field #: 0, 1, 2, .. */
  52 } olist[MAXFLD];
  53 int nout;                       /* # of output filed */
  54
  55 int aflag;                      /* n for '-an': F1 or F2 or both */
  56 int vflag;                      /* n for '-vn': F1 or F2 or both */
  57 char *es;                       /* s for '-e s' */
  58 char sep;                       /* c for -tc: filed separator */
  59 char *cmd;                      /* name of this program */
  60
  61 int main(argc, argv)
  62 int argc;
  63 char **argv;
  64 {
  65   register char *s;
  66   int c, i, j;
  67
  68   cmd = argv[0];
  69   outfun = output;              /* default output form */
  70
  71   while (--argc > 0 && (*++argv)[0] == '-' && (*argv)[1]) {
  72         /* "-" is a file name (stdin) */
  73         s = argv[0] + 1;
  74         if ((c = *s) == '-' && !s[1]) {
  75                 ++argv;
  76                 --argc;
  77                 break;          /* -- */
  78         }
  79         if (*++s == '\0') {
  80                 s = *++argv;
  81                 --argc;
  82         }
  83         switch (c) {
  84             case 'a':           /* add unpairable line to output */
  85                 vflag = 0;
  86                 switch (*s) {
  87                     case '1':   aflag |= F1;    break;
  88                     case '2':   aflag |= F2;    break;
  89                     default:    aflag |= (F1 | F2);     break;
  90                 }
  91                 break;
  92
  93             case 'e':           /* replace empty field by es */
  94                 es = s;
  95                 break;
  96
  97             case 'j':           /* key field (obsolute) */
  98                 c = *s++;
  99                 if (*s == '\0') {
 100                         s = *++argv;
 101                         --argc;
 102                 }
 103
 104             case '1':           /* key field of file1 */
 105             case '2':           /* key field of file2 */
 106                 i = atoi(s) - 1;
 107
 108                 switch (c) {
 109                     case '1':   kpos[0] = i;    break;
 110                     case '2':   kpos[1] = i;    break;
 111                     default:    kpos[0] = kpos[1] = i;
 112                                 break;
 113                 }
 114                 break;
 115
 116             case 'o':           /* specify output format */
 117                 do {
 118                         i = j = 0;
 119                         sscanf(s, "%d.%d", &i, &j);
 120                         if (i < 1 || j < 1 || i > 2) usage();
 121                         olist[nout].o_file = i - 1;
 122                         olist[nout].o_field = j - 1;
 123                         nout++;
 124                         if ((s = strchr(s, ',')) != (char *) 0)
 125                                 s++;
 126                         else {
 127                                 s = *++argv;
 128                                 --argc;
 129                         }
 130                 } while (argc > 2 && *s != '-');
 131                 ++argc;
 132                 --argv;         /* compensation */
 133                 outfun = outputf;
 134                 break;
 135
 136             case 't':           /* tab char */
 137                 sep = *s;
 138                 break;
 139
 140             case 'v':           /* output unpairable line only */
 141                 aflag = 0;
 142                 switch (*s) {
 143                     case '1':   vflag |= F1;    break;
 144                     case '2':   vflag |= F2;    break;
 145                     default:    vflag |= (F1 | F2);     break;
 146                 }
 147                 break;
 148
 149             default:    usage();
 150         }
 151   }
 152   if (argc != 2) usage();
 153
 154   fp[0] = strcmp(argv[0], "-") ? efopen(argv[0], "r") : stdin;
 155   fp[1] = efopen(argv[1], "r");
 156
 157   nfld[0] = get1();             /* read file1 */
 158   nfld[1] = get2(0);            /* read file2 */
 159
 160   while (nfld[0] || nfld[1]) {
 161         if ((i = compare()) == 0)
 162                 match();
 163         else if (i < 0)
 164                 f1_only();
 165         else
 166                 f2_only();
 167   }
 168   fflush(stdout);
 169
 170   exit(0);
 171 }
 172
 173 void usage()
 174 {
 175   fprintf(stderr,
 176     "Usage: %s [-an|-vn] [-e str] [-o list] [-tc] [-1 f] [-2 f] file1 file2\n",
 177     cmd);
 178   exit(1);
 179 }
 180
 181 int compare()
 182 {                               /* compare key field */
 183   register int r;
 184
 185   if (nfld[1] == 0)             /* file2 EOF */
 186         r = -1;
 187   else if (nfld[0] == 0)        /* file1 EOF */
 188         r = 1;
 189   else {
 190         if (nfld[0] <= kpos[0])
 191                 error("missing key field in file1", (char *) 0);
 192         if (nfld[1] <= kpos[1])
 193                 error("missing key field in file2", (char *) 0);
 194
 195         r = strcmp(fld[0][kpos[0]], fld[1][kpos[1]]);
 196   }
 197   return r;
 198 }
 199
 200 void match()
 201 {
 202   long p;
 203
 204   if (!vflag) (*outfun) (F1 | F2);
 205
 206   p = ftell(fp[1]);
 207   nfld[1] = get2(0);            /* check key order */
 208   if (nfld[1] == 0 || strcmp(fld[0][kpos[0]], fld[1][kpos[1]])) {
 209         nfld[0] = get1();
 210         if (strcmp(fld[0][kpos[0]], oldkey) == 0) {
 211                 fseek(fp[1], head, 0);  /* re-do from head */
 212                 nfld[1] = get2(1);      /* don't check key order */
 213         } else
 214                 head = p;       /* mark here */
 215   }
 216 }
 217
 218 void f1_only()
 219 {
 220   if ((aflag & F1) || (vflag & F1)) (*outfun) (F1);
 221   nfld[0] = get1();
 222 }
 223
 224 void f2_only()
 225 {
 226   if ((aflag & F2) || (vflag & F2)) (*outfun) (F2);
 227   head = ftell(fp[1]);          /* mark */
 228   nfld[1] = get2(0);            /* check key order */
 229 }
 230
 231 void output(f)
 232 {                               /* default output form */
 233   if (f & F1)
 234         fputs(fld[0][kpos[0]], stdout);
 235   else
 236         fputs(fld[1][kpos[1]], stdout);
 237   if (f & F1) outfld(0);
 238   if (f & F2) outfld(1);
 239   fputc('\n', stdout);
 240 }
 241
 242 void outfld(file)
 243 {                               /* output all fields except key_field */
 244   register int i;
 245   int k, n;
 246
 247   k = kpos[file];
 248   n = nfld[file];
 249   for (i = 0; i < n; i++)
 250         if (i != k) {
 251                 fputc(SEP, stdout);
 252                 fputs(fld[file][i], stdout);
 253         }
 254 }
 255
 256 void outputf(f)
 257 {                               /* output by '-o list' */
 258   int i, j, k;
 259   register char *s;
 260
 261   for (i = k = 0; i < nout; i++) {
 262         j = olist[i].o_file;
 263         if ((f & (j + 1)) && (olist[i].o_field < nfld[j]))
 264                 s = fld[j][olist[i].o_field];
 265         else
 266                 s = es;
 267         if (s) {
 268                 if (k++) fputc(SEP, stdout);
 269                 fputs(s, stdout);
 270         }
 271   }
 272   fputc('\n', stdout);
 273 }
 274
 275 int get1()
 276 {                               /* read file1 */
 277   int r;
 278   static char oldkey1[BUFSIZ];
 279
 280   if (fld[0][kpos[0]]) {
 281         strcpy(oldkey, fld[0][kpos[0]]);  /* save previous key for control */
 282   }
 283   r = getrec(0);
 284
 285   if (r) {
 286         if (strcmp(oldkey1, fld[0][kpos[0]]) > 0)
 287               error("file1 is not sorted", (char *) 0);
 288         strcpy(oldkey1, fld[0][kpos[0]]);  /* save prev key for sort check */
 289   }
 290   return r;
 291 }
 292
 293 int get2(back)
 294 {                               /* read file2 */
 295   static char oldkey2[BUFSIZ];
 296   int r;
 297
 298   r = getrec(1);
 299
 300   if (r) {
 301         if (!back && strcmp(oldkey2, fld[1][kpos[1]]) > 0)
 302               error("file2 is not sorted", (char *) 0);
 303         strcpy(oldkey2, fld[1][kpos[1]]);  /* save prev key for sort check */
 304   }
 305   return r;
 306 }
 307
 308 int getrec(file)
 309 {                               /* read one line to split it */
 310   if (fgets(buf[file], BUFSIZ, fp[file]) == (char *) 0)
 311         *buf[file] = '\0';
 312   else if (*buf[file] == '\n' || *buf[file] == '\r')
 313         error("null line in file%s", file ? "1" : "0");
 314
 315   return split(file);
 316 }
 317
 318 int split(file)
 319 {                               /* setup fields */
 320   register int n;
 321   register char *s, *t;
 322
 323   for (n = 0, s = buf[file]; *s && *s != '\n' && *s != '\r';) {
 324         if (sep) {
 325                 for (t = s; *s && *s != sep && *s != '\n' && *s != '\r'; s++);
 326         } else {
 327                 while (*s == ' ' || *s == '\t')
 328                         s++;    /* skip leading white space */
 329                 for (t = s; *s && *s != ' ' && *s != '\t'
 330                      && *s != '\n' && *s != '\r'; s++);
 331                 /* We will treat trailing white space as NULL field */
 332         }
 333         if (*s) *s++ = '\0';
 334         fld[file][n++] = t;
 335         if (n == MAXFLD) error("too many filed in file%s", file ? "1" : "0");
 336   }
 337   fld[file][n] = (char *) 0;
 338
 339   return n;
 340 }
 341
 342 FILE *efopen(file, mode)
 343 char *file, *mode;
 344 {
 345   FILE *fp;
 346
 347   if ((fp = fopen(file, mode)) == (FILE *) 0) error("can't open %s", file);
 348
 349   return fp;
 350 }
 351
 352 void error(s, t)
 353 char *s, *t;
 354 {
 355   fprintf(stderr, "%s: ", cmd);
 356   fprintf(stderr, s, t);
 357   fprintf(stderr, "\n");
 358
 359   exit(1);
 360 }