coreutils/sort.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * SuS3 compliant sort implementation for busybox
   4  *
   5  * Copyright (C) 2004 by Rob Landley <rob@landley.net>
   6  *
   7  * MAINTAINER: Rob Landley <rob@landley.net>
   8  *
   9  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
  10  *
  11  * See SuS3 sort standard at:
  12  * http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
  13  */
  14
  15 #include "busybox.h"
  16
  17 static int global_flags;
  18
  19 /*
  20         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  21         sort -c [-bdfinru][-t char][-k keydef][file]
  22 */
  23
  24 /* These are sort types */
  25 #define FLAG_n                  1               /* Numeric sort */
  26 #define FLAG_g                  2               /* Sort using strtod() */
  27 #define FLAG_M                  4               /* Sort date */
  28 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  29 #define FLAG_u                  8               /* Unique */
  30 #define FLAG_c                  16              /* Check: no output, exit(!ordered) */
  31 #define FLAG_s                  32              /* Stable sort, no ascii fallback at end */
  32 #define FLAG_z                  64              /* Input is null terminated, not \n */
  33 /* These can be applied to search keys, the previous four can't */
  34 #define FLAG_b                  128             /* Ignore leading blanks */
  35 #define FLAG_r                  256             /* Reverse */
  36 #define FLAG_d                  512             /* Ignore !(isalnum()|isspace()) */
  37 #define FLAG_f                  1024    /* Force uppercase */
  38 #define FLAG_i                  2048    /* Ignore !isprint() */
  39 #define FLAG_bb                 32768   /* Ignore trailing blanks  */
  40
  41
  42 #ifdef CONFIG_FEATURE_SORT_BIG
  43 static char key_separator;
  44
  45 static struct sort_key
  46 {
  47         struct sort_key *next_key;      /* linked list */
  48         unsigned short range[4];        /* start word, start char, end word, end char */
  49         int flags;
  50 } *key_list;
  51
  52 static char *get_key(char *str, struct sort_key *key, int flags)
  53 {
  54         int start=0,end,len,i,j;
  55
  56         /* Special case whole string, so we don't have to make a copy */
  57         if(key->range[0]==1 && !key->range[1] && !key->range[2] && !key->range[3]
  58                 && !(flags&(FLAG_b&FLAG_d&FLAG_f&FLAG_i&FLAG_bb))) return str;
  59         /* Find start of key on first pass, end on second pass*/
  60         len=strlen(str);
  61
  62         for(j=0;j<2;j++) {
  63                 if(!key->range[2*j]) end=len;
  64                 /* Loop through fields */
  65                 else {
  66                         end=0;
  67                         for(i=1;i<key->range[2*j]+j;i++) {
  68                                 /* Skip leading blanks or first separator */
  69                                 if(str[end]) {
  70                                         if(!key_separator && isspace(str[end]))
  71                                                 while(isspace(str[end])) end++;
  72                                 }
  73                                 /* Skip body of key */
  74                                 for(;str[end];end++) {
  75                                         if(key_separator) {
  76                                                 if(str[end]==key_separator) break;
  77                                         } else if(isspace(str[end])) break;
  78                                 }
  79                         }
  80                 }
  81                 if(!j) start=end;
  82         }
  83         /* Key with explicit separator starts after separator */
  84         if(key_separator && str[start]==key_separator) start++;
  85         /* Strip leading whitespace if necessary */
  86         if(flags&FLAG_b) while(isspace(str[start])) start++;
  87         /* Strip trailing whitespace if necessary */
  88         if(flags&FLAG_bb) while(end>start && isspace(str[end-1])) end--;
  89         /* Handle offsets on start and end */
  90         if(key->range[3]) {
  91                 end+=key->range[3]-1;
  92                 if(end>len) end=len;
  93         }
  94         if(key->range[1]) {
  95                 start+=key->range[1]-1;
  96                 if(start>len) start=len;
  97         }
  98         /* Make the copy */
  99         if(end<start) end=start;
 100         str=xstrndup(str+start,end-start);
 101         /* Handle -d */
 102         if(flags&FLAG_d) {
 103                 for(start=end=0;str[end];end++)
 104                         if(isspace(str[end]) || isalnum(str[end])) str[start++]=str[end];
 105                 str[start]=0;
 106         }
 107         /* Handle -i */
 108         if(flags&FLAG_i) {
 109                 for(start=end=0;str[end];end++)
 110                         if(isprint(str[end])) str[start++]=str[end];
 111                 str[start]=0;
 112         }
 113         /* Handle -f */
 114         if(flags*FLAG_f) for(i=0;str[i];i++) str[i]=toupper(str[i]);
 115
 116         return str;
 117 }
 118
 119 static struct sort_key *add_key(void)
 120 {
 121         struct sort_key **pkey=&key_list;
 122         while(*pkey) pkey=&((*pkey)->next_key);
 123         return *pkey = xzalloc(sizeof(struct sort_key));
 124 }
 125
 126 #define GET_LINE(fp) (global_flags&FLAG_z) ? bb_get_chunk_from_file(fp,NULL) \
 127                                                                                    : bb_get_chomped_line_from_file(fp)
 128 #else
 129 #define GET_LINE(fp)            bb_get_chomped_line_from_file(fp)
 130 #endif
 131
 132 /* Iterate through keys list and perform comparisons */
 133 static int compare_keys(const void *xarg, const void *yarg)
 134 {
 135         int flags=global_flags,retval=0;
 136         char *x,*y;
 137
 138 #ifdef CONFIG_FEATURE_SORT_BIG
 139         struct sort_key *key;
 140
 141         for(key=key_list;!retval && key;key=key->next_key) {
 142                 flags=(key->flags) ? key->flags : global_flags;
 143                 /* Chop out and modify key chunks, handling -dfib */
 144                 x=get_key(*(char **)xarg,key,flags);
 145                 y=get_key(*(char **)yarg,key,flags);
 146 #else
 147         /* This curly bracket serves no purpose but to match the nesting
 148            level of the for() loop we're not using */
 149         {
 150                 x=*(char **)xarg;
 151                 y=*(char **)yarg;
 152 #endif
 153                 /* Perform actual comparison */
 154                 switch(flags&7) {
 155                         default:
 156                                 bb_error_msg_and_die("Unknown sort type.");
 157                                 break;
 158                         /* Ascii sort */
 159                         case 0:
 160                                 retval=strcmp(x,y);
 161                                 break;
 162 #ifdef CONFIG_FEATURE_SORT_BIG
 163                         case FLAG_g:
 164                         {
 165                                 char *xx,*yy;
 166                                 double dx=strtod(x,&xx), dy=strtod(y,&yy);
 167                                 /* not numbers < NaN < -infinity < numbers < +infinity) */
 168                                 if(x==xx) retval=(y==yy ? 0 : -1);
 169                                 else if(y==yy) retval=1;
 170                                 /* Check for isnan */
 171                                 else if(dx != dx) retval = (dy != dy) ? 0 : -1;
 172                                 else if(dy != dy) retval = 1;
 173                                 /* Check for infinity.  Could underflow, but it avoids libm. */
 174                                 else if(1.0/dx == 0.0) {
 175                                         if(dx<0) retval=((1.0/dy == 0.0 && dy<0) ? 0 : -1);
 176                                         else retval=((1.0/dy == 0.0 && dy>0) ? 0 : 1);
 177                                 } else if(1.0/dy == 0.0) retval=dy<0 ? 1 : -1;
 178                                 else retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 179                                 break;
 180                         }
 181                         case FLAG_M:
 182                         {
 183                                 struct tm thyme;
 184                                 int dx;
 185                                 char *xx,*yy;
 186
 187                                 xx=strptime(x,"%b",&thyme);
 188                                 dx=thyme.tm_mon;
 189                                 yy=strptime(y,"%b",&thyme);
 190                                 if(!xx) retval=(!yy ? 0 : -1);
 191                                 else if(!yy) retval=1;
 192                                 else retval=(dx==thyme.tm_mon ? 0 : dx-thyme.tm_mon);
 193                                 break;
 194                         }
 195                         /* Full floating point version of -n */
 196                         case FLAG_n:
 197                         {
 198                                 double dx=atof(x),dy=atof(y);
 199                                 retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 200                                 break;
 201                         }
 202                 }
 203                 /* Free key copies. */
 204                 if(x!=*(char **)xarg) free(x);
 205                 if(y!=*(char **)yarg) free(y);
 206                 if(retval) break;
 207 #else
 208                         /* Integer version of -n for tiny systems */
 209                         case FLAG_n:
 210                                 retval=atoi(x)-atoi(y);
 211                                 break;
 212                 }
 213 #endif
 214         }
 215         /* Perform fallback sort if necessary */
 216         if(!retval && !(global_flags&FLAG_s))
 217                         retval=strcmp(*(char **)xarg, *(char **)yarg);
 218         return ((flags&FLAG_r)?-1:1)*retval;
 219 }
 220
 221 int sort_main(int argc, char **argv)
 222 {
 223         FILE *fp,*outfile=NULL;
 224         int linecount=0,i,flag;
 225         char *line,**lines=NULL,*optlist="ngMucszbrdfimS:T:o:k:t:";
 226         int c;
 227
 228         bb_default_error_retval = 2;
 229         /* Parse command line options */
 230         while((c=getopt(argc,argv,optlist))>0) {
 231                 line=strchr(optlist,c);
 232                 if(!line) bb_show_usage();
 233                 switch(*line) {
 234 #ifdef CONFIG_FEATURE_SORT_BIG
 235                         case 'o':
 236                                 if(outfile) bb_error_msg_and_die("Too many -o.");
 237                                 outfile=xfopen(optarg,"w");
 238                                 break;
 239                         case 't':
 240                                 if(key_separator || optarg[1])
 241                                         bb_error_msg_and_die("Too many -t.");
 242                                 key_separator=*optarg;
 243                                 break;
 244                         /* parse sort key */
 245                         case 'k':
 246                         {
 247                                 struct sort_key *key=add_key();
 248                                 char *temp, *temp2;
 249
 250                                 temp=optarg;
 251                                 for(i=0;*temp;) {
 252                                         /* Start of range */
 253                                         key->range[2*i]=(unsigned short)strtol(temp,&temp,10);
 254                                         if(*temp=='.')
 255                                                 key->range[(2*i)+1]=(unsigned short)strtol(temp+1,&temp,10);
 256                                         for(;*temp;temp++) {
 257                                                 if(*temp==',' && !i++) {
 258                                                         temp++;
 259                                                         break;
 260                                                 } /* no else needed: fall through to syntax error
 261                                                          because comma isn't in optlist */
 262                                                 temp2=strchr(optlist,*temp);
 263                                                 flag=(1<<(temp2-optlist));
 264                                                 if(!temp2 || (flag>FLAG_M && flag<FLAG_b))
 265                                                         bb_error_msg_and_die("Unknown key option.");
 266                                                 /* b after , means strip _trailing_ space */
 267                                                 if(i && flag==FLAG_b) flag=FLAG_bb;
 268                                                 key->flags|=flag;
 269                                         }
 270                                 }
 271                                 break;
 272                         }
 273 #endif
 274                         default:
 275                                 global_flags|=(1<<(line-optlist));
 276                                 /* global b strips leading and trailing spaces */
 277                                 if(global_flags&FLAG_b) global_flags|=FLAG_bb;
 278                                 break;
 279                 }
 280         }
 281         /* Open input files and read data */
 282         for(i=argv[optind] ? optind : optind-1;argv[i];i++) {
 283                 if(i<optind || (*argv[i]=='-' && !argv[i][1])) fp=stdin;
 284                 else fp=xfopen(argv[i],"r");
 285                 for(;;) {
 286                         line=GET_LINE(fp);
 287                         if(!line) break;
 288                         if(!(linecount&63))
 289                                 lines=xrealloc(lines, sizeof(char *)*(linecount+64));
 290                         lines[linecount++]=line;
 291                 }
 292                 fclose(fp);
 293         }
 294 #ifdef CONFIG_FEATURE_SORT_BIG
 295         /* if no key, perform alphabetic sort */
 296     if(!key_list) add_key()->range[0]=1;
 297         /* handle -c */
 298         if(global_flags&FLAG_c) {
 299                 int j=(global_flags&FLAG_u) ? -1 : 0;
 300                 for(i=1;i<linecount;i++)
 301                         if(compare_keys(&lines[i-1],&lines[i])>j) {
 302                                 fprintf(stderr,"Check line %d\n",i);
 303                                 return 1;
 304                         }
 305                 return 0;
 306         }
 307 #endif
 308         /* Perform the actual sort */
 309         qsort(lines,linecount,sizeof(char *),compare_keys);
 310         /* handle -u */
 311         if(global_flags&FLAG_u) {
 312                 for(flag=0,i=1;i<linecount;i++) {
 313                         if(!compare_keys(&lines[flag],&lines[i])) free(lines[i]);
 314                         else lines[++flag]=lines[i];
 315                 }
 316                 if(linecount) linecount=flag+1;
 317         }
 318         /* Print it */
 319         if(!outfile) outfile=stdout;
 320         for(i=0;i<linecount;i++) fprintf(outfile,"%s\n",lines[i]);
 321         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 322 }