sys/src/cmd/join.c

   1 /*      join F1 F2 on stuff */
   2 #include <u.h>
   3 #include <libc.h>
   4 #include <stdio.h>
   5 #include <ctype.h>
   6 #define F1 0
   7 #define F2 1
   8 #define F0 3
   9 #define NFLD    100     /* max field per line */
  10 #define comp() runecmp(ppi[F1][j1],ppi[F2][j2])
  11 FILE *f[2];
  12 Rune buf[2][BUFSIZ];    /*input lines */
  13 Rune *ppi[2][NFLD+1];   /* pointers to fields in lines */
  14 Rune *s1,*s2;
  15 int     j1      = 1;    /* join of this field of file 1 */
  16 int     j2      = 1;    /* join of this field of file 2 */
  17 int     olist[2*NFLD];  /* output these fields */
  18 int     olistf[2*NFLD]; /* from these files */
  19 int     no;             /* number of entries in olist */
  20 Rune    sep1    = ' ';  /* default field separator */
  21 Rune    sep2    = '\t';
  22 char *sepstr=" ";
  23 int     discard;        /* count of truncated lines */
  24 Rune    null[BUFSIZ]    = L"";
  25 int     a1;
  26 int     a2;
  27
  28 char *getoptarg(int*, char***);
  29 void output(int, int);
  30 int input(int);
  31 void oparse(char*);
  32 void error(char*, char*);
  33 void seek1(void), seek2(void);
  34 Rune *strtorune(Rune *, char *);
  35
  36
  37 void
  38 main(int argc, char **argv)
  39 {
  40         int i;
  41
  42         while (argc > 1 && argv[1][0] == '-') {
  43                 if (argv[1][1] == '\0')
  44                         break;
  45                 switch (argv[1][1]) {
  46                 case '-':
  47                         argc--;
  48                         argv++;
  49                         goto proceed;
  50                 case 'a':
  51                         switch(*getoptarg(&argc, &argv)) {
  52                         case '1':
  53                                 a1++;
  54                                 break;
  55                         case '2':
  56                                 a2++;
  57                                 break;
  58                         default:
  59                                 error("incomplete option -a","");
  60                         }
  61                         break;
  62                 case 'e':
  63                         strtorune(null, getoptarg(&argc, &argv));
  64                         break;
  65                 case 't':
  66                         sepstr=getoptarg(&argc, &argv);
  67                         chartorune(&sep1, sepstr);
  68                         sep2 = sep1;
  69                         break;
  70                 case 'o':
  71                         if(argv[1][2]!=0 ||
  72                            argc>2 && strchr(argv[2],',')!=0)
  73                                 oparse(getoptarg(&argc, &argv));
  74                         else for (no = 0; no<2*NFLD && argc>2; no++){
  75                                 if (argv[2][0] == '1' && argv[2][1] == '.') {
  76                                         olistf[no] = F1;
  77                                         olist[no] = atoi(&argv[2][2]);
  78                                 } else if (argv[2][0] == '2' && argv[2][1] == '.') {
  79                                         olist[no] = atoi(&argv[2][2]);
  80                                         olistf[no] = F2;
  81                                 } else if (argv[2][0] == '0')
  82                                         olistf[no] = F0;
  83                                 else
  84                                         break;
  85                                 argc--;
  86                                 argv++;
  87                         }
  88                         break;
  89                 case 'j':
  90                         if(argc <= 2)
  91                                 break;
  92                         if (argv[1][2] == '1')
  93                                 j1 = atoi(argv[2]);
  94                         else if (argv[1][2] == '2')
  95                                 j2 = atoi(argv[2]);
  96                         else
  97                                 j1 = j2 = atoi(argv[2]);
  98                         argc--;
  99                         argv++;
 100                         break;
 101                 case '1':
 102                         j1 = atoi(getoptarg(&argc, &argv));
 103                         break;
 104                 case '2':
 105                         j2 = atoi(getoptarg(&argc, &argv));
 106                         break;
 107                 }
 108                 argc--;
 109                 argv++;
 110         }
 111 proceed:
 112         for (i = 0; i < no; i++)
 113                 if (olist[i]-- > NFLD)  /* 0 origin */
 114                         error("field number too big in -o","");
 115         if (argc != 3)
 116                 error("usage: join [-1 x -2 y] [-o list] file1 file2","");
 117         if (j1 < 1  || j2 < 1)
 118                 error("invalid field indices", "");
 119         j1--;
 120         j2--;   /* everyone else believes in 0 origin */
 121         s1 = ppi[F1][j1];
 122         s2 = ppi[F2][j2];
 123         if (strcmp(argv[1], "-") == 0)
 124                 f[F1] = stdin;
 125         else if ((f[F1] = fopen(argv[1], "r")) == 0)
 126                 error("can't open %s", argv[1]);
 127         if(strcmp(argv[2], "-") == 0) {
 128                 f[F2] = stdin;
 129         } else if ((f[F2] = fopen(argv[2], "r")) == 0)
 130                 error("can't open %s", argv[2]);
 131
 132         if(ftell(f[F2]) >= 0)
 133                 seek2();
 134         else if(ftell(f[F1]) >= 0)
 135                 seek1();
 136         else
 137                 error("neither file is randomly accessible","");
 138         if (discard)
 139                 error("some input line was truncated", "");
 140         exits("");
 141 }
 142 int runecmp(Rune *a, Rune *b){
 143         while(*a==*b){
 144                 if(*a=='\0') return 0;
 145                 a++;
 146                 b++;
 147         }
 148         if(*a<*b) return -1;
 149         return 1;
 150 }
 151 char *runetostr(char *buf, Rune *r){
 152         char *s;
 153         for(s=buf;*r;r++) s+=runetochar(s, r);
 154         *s='\0';
 155         return buf;
 156 }
 157 Rune *strtorune(Rune *buf, char *s){
 158         Rune *r;
 159         for(r=buf;*s;r++) s+=chartorune(r, s);
 160         *r='\0';
 161         return buf;
 162 }
 163 /* lazy.  there ought to be a clean way to combine seek1 & seek2 */
 164 #define get1() n1=input(F1)
 165 #define get2() n2=input(F2)
 166 void
 167 seek2()
 168 {
 169         int n1, n2;
 170         int top2=0;
 171         int bot2 = ftell(f[F2]);
 172         get1();
 173         get2();
 174         while(n1>0 && n2>0 || (a1||a2) && n1+n2>0) {
 175                 if(n1>0 && n2>0 && comp()>0 || n1==0) {
 176                         if(a2) output(0, n2);
 177                         bot2 = ftell(f[F2]);
 178                         get2();
 179                 } else if(n1>0 && n2>0 && comp()<0 || n2==0) {
 180                         if(a1) output(n1, 0);
 181                         get1();
 182                 } else /*(n1>0 && n2>0 && comp()==0)*/ {
 183                         while(n2>0 && comp()==0) {
 184                                 output(n1, n2);
 185                                 top2 = ftell(f[F2]);
 186                                 get2();
 187                         }
 188                         fseek(f[F2], bot2, 0);
 189                         get2();
 190                         get1();
 191                         for(;;) {
 192                                 if(n1>0 && n2>0 && comp()==0) {
 193                                         output(n1, n2);
 194                                         get2();
 195                                 } else if(n1>0 && n2>0 && comp()<0 || n2==0) {
 196                                         fseek(f[F2], bot2, 0);
 197                                         get2();
 198                                         get1();
 199                                 } else /*(n1>0 && n2>0 && comp()>0 || n1==0)*/{
 200                                         fseek(f[F2], top2, 0);
 201                                         bot2 = top2;
 202                                         get2();
 203                                         break;
 204                                 }
 205                         }
 206                 }
 207         }
 208 }
 209 void
 210 seek1()
 211 {
 212         int n1, n2;
 213         int top1=0;
 214         int bot1 = ftell(f[F1]);
 215         get1();
 216         get2();
 217         while(n1>0 && n2>0 || (a1||a2) && n1+n2>0) {
 218                 if(n1>0 && n2>0 && comp()>0 || n1==0) {
 219                         if(a2) output(0, n2);
 220                         get2();
 221                 } else if(n1>0 && n2>0 && comp()<0 || n2==0) {
 222                         if(a1) output(n1, 0);
 223                         bot1 = ftell(f[F1]);
 224                         get1();
 225                 } else /*(n1>0 && n2>0 && comp()==0)*/ {
 226                         while(n2>0 && comp()==0) {
 227                                 output(n1, n2);
 228                                 top1 = ftell(f[F1]);
 229                                 get1();
 230                         }
 231                         fseek(f[F1], bot1, 0);
 232                         get2();
 233                         get1();
 234                         for(;;) {
 235                                 if(n1>0 && n2>0 && comp()==0) {
 236                                         output(n1, n2);
 237                                         get1();
 238                                 } else if(n1>0 && n2>0 && comp()>0 || n1==0) {
 239                                         fseek(f[F1], bot1, 0);
 240                                         get2();
 241                                         get1();
 242                                 } else /*(n1>0 && n2>0 && comp()<0 || n2==0)*/{
 243                                         fseek(f[F1], top1, 0);
 244                                         bot1 = top1;
 245                                         get1();
 246                                         break;
 247                                 }
 248                         }
 249                 }
 250         }
 251 }
 252
 253 int
 254 input(int n)            /* get input line and split into fields */
 255 {
 256         register int i, c;
 257         Rune *bp;
 258         Rune **pp;
 259         char line[BUFSIZ];
 260
 261         bp = buf[n];
 262         pp = ppi[n];
 263         if (fgets(line, BUFSIZ, f[n]) == 0)
 264                 return(0);
 265         strtorune(bp, line);
 266         i = 0;
 267         do {
 268                 i++;
 269                 if (sep1 == ' ')        /* strip multiples */
 270                         while ((c = *bp) == sep1 || c == sep2)
 271                                 bp++;   /* skip blanks */
 272                 *pp++ = bp;     /* record beginning */
 273                 while ((c = *bp) != sep1 && c != '\n' && c != sep2 && c != '\0')
 274                         bp++;
 275                 *bp++ = '\0';   /* mark end by overwriting blank */
 276         } while (c != '\n' && c != '\0' && i < NFLD-1);
 277         if (c != '\n')
 278                 discard++;
 279
 280         *pp = 0;
 281         return(i);
 282 }
 283
 284 void
 285 output(int on1, int on2)        /* print items from olist */
 286 {
 287         int i;
 288         Rune *temp;
 289         char buf[BUFSIZ*UTFmax+1];
 290
 291         if (no <= 0) {  /* default case */
 292                 printf("%s", runetostr(buf, on1? ppi[F1][j1]: ppi[F2][j2]));
 293                 for (i = 0; i < on1; i++)
 294                         if (i != j1)
 295                                 printf("%s%s", sepstr, runetostr(buf, ppi[F1][i]));
 296                 for (i = 0; i < on2; i++)
 297                         if (i != j2)
 298                                 printf("%s%s", sepstr, runetostr(buf, ppi[F2][i]));
 299                 printf("\n");
 300         } else {
 301                 for (i = 0; i < no; i++) {
 302                         if (olistf[i]==F0 && on1>j1)
 303                                 temp = ppi[F1][j1];
 304                         else if (olistf[i]==F0 && on2>j2)
 305                                 temp = ppi[F2][j2];
 306                         else {
 307                                 temp = ppi[olistf[i]][olist[i]];
 308                                 if(olistf[i]==F1 && on1<=olist[i] ||
 309                                    olistf[i]==F2 && on2<=olist[i] ||
 310                                    *temp==0)
 311                                         temp = null;
 312                         }
 313                         printf("%s", runetostr(buf, temp));
 314                         if (i == no - 1)
 315                                 printf("\n");
 316                         else
 317                                 printf("%s", sepstr);
 318                 }
 319         }
 320 }
 321
 322 void
 323 error(char *s1, char *s2)
 324 {
 325         fprintf(stderr, "join: ");
 326         fprintf(stderr, s1, s2);
 327         fprintf(stderr, "\n");
 328         exits(s1);
 329 }
 330
 331 char *
 332 getoptarg(int *argcp, char ***argvp)
 333 {
 334         int argc = *argcp;
 335         char **argv = *argvp;
 336         if(argv[1][2] != 0)
 337                 return &argv[1][2];
 338         if(argc<=2 || argv[2][0]=='-')
 339                 error("incomplete option %s", argv[1]);
 340         *argcp = argc-1;
 341         *argvp = ++argv;
 342         return argv[1];
 343 }
 344
 345 void
 346 oparse(char *s)
 347 {
 348         for (no = 0; no<2*NFLD && *s; no++, s++) {
 349                 switch(*s) {
 350                 case 0:
 351                         return;
 352                 case '0':
 353                         olistf[no] = F0;
 354                         break;
 355                 case '1':
 356                 case '2':
 357                         if(s[1] == '.' && isdigit(s[2])) {
 358                                 olistf[no] = *s=='1'? F1: F2;
 359                                 olist[no] = atoi(s += 2);
 360                                 break;
 361                         } /* fall thru */
 362                 default:
 363                         error("invalid -o list", "");
 364                 }
 365                 if(s[1] == ',')
 366                         s++;
 367         }
 368 }