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 <ctype.h>
  16 #include <math.h>
  17 #include <stdio.h>
  18 #include <stdlib.h>
  19 #include <string.h>
  20 #include <time.h>
  21 #include <unistd.h>
  22 #include "busybox.h"
  23
  24 static int global_flags;
  25
  26 /*
  27         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  28         sort -c [-bdfinru][-t char][-k keydef][file]
  29 */
  30
  31 /* These are sort types */
  32 #define FLAG_n                  1               /* Numeric sort */
  33 #define FLAG_g                  2               /* Sort using strtod() */
  34 #define FLAG_M                  4               /* Sort date */
  35 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  36 #define FLAG_u                  8               /* Unique */
  37 #define FLAG_c                  16              /* Check: no output, exit(!ordered) */
  38 #define FLAG_s                  32              /* Stable sort, no ascii fallback at end */
  39 #define FLAG_z                  64              /* Input is null terminated, not \n */
  40 /* These can be applied to search keys, the previous four can't */
  41 #define FLAG_b                  128             /* Ignore leading blanks */
  42 #define FLAG_r                  256             /* Reverse */
  43 #define FLAG_d                  512             /* Ignore !(isalnum()|isspace()) */
  44 #define FLAG_f                  1024    /* Force uppercase */
  45 #define FLAG_i                  2048    /* Ignore !isprint() */
  46 #define FLAG_bb                 32768   /* Ignore trailing blanks  */
  47
  48
  49 #ifdef CONFIG_FEATURE_SORT_BIG
  50 static char key_separator;
  51
  52 static struct sort_key
  53 {
  54         struct sort_key *next_key;      /* linked list */
  55         unsigned short range[4];        /* start word, start char, end word, end char */
  56         int flags;
  57 } *key_list;
  58
  59 static char *get_key(char *str, struct sort_key *key, int flags)
  60 {
  61         int start=0,end,len,i,j;
  62
  63         /* Special case whole string, so we don't have to make a copy */
  64         if(key->range[0]==1 && !key->range[1] && !key->range[2] && !key->range[3]
  65                 && !(flags&(FLAG_b&FLAG_d&FLAG_f&FLAG_i&FLAG_bb))) return str;
  66         /* Find start of key on first pass, end on second pass*/
  67         len=strlen(str);
  68
  69         for(j=0;j<2;j++) {
  70                 if(!key->range[2*j]) end=len;
  71                 /* Loop through fields */
  72                 else {
  73                         end=0;
  74                         for(i=1;i<key->range[2*j]+j;i++) {
  75                                 /* Skip leading blanks or first separator */
  76                                 if(str[end]) {
  77                                         if(key_separator) {
  78                                                 if(str[end]==key_separator) end++;
  79                                         } else if(isspace(str[end]))
  80                                                 while(isspace(str[end])) end++;
  81                                 }
  82                                 /* Skip body of key */
  83                                 for(;str[end];end++) {
  84                                         if(key_separator) {
  85                                                 if(str[end]==key_separator) break;
  86                                         } else if(isspace(str[end])) break;
  87                                 }
  88                         }
  89                 }
  90                 if(!j) start=end;
  91         }
  92         /* Key with explicit separator starts after separator */
  93         if(key_separator && str[start]==key_separator) start++;
  94         /* Strip leading whitespace if necessary */
  95         if(flags&FLAG_b) while(isspace(str[start])) start++;
  96         /* Strip trailing whitespace if necessary */
  97         if(flags&FLAG_bb) while(end>start && isspace(str[end-1])) end--;
  98         /* Handle offsets on start and end */
  99         if(key->range[3]) {
 100                 end+=key->range[3]-1;
 101                 if(end>len) end=len;
 102         }
 103         if(key->range[1]) {
 104                 start+=key->range[1]-1;
 105                 if(start>len) start=len;
 106         }
 107         /* Make the copy */
 108         if(end<start) end=start;
 109         str=bb_xstrndup(str+start,end-start);
 110         /* Handle -d */
 111         if(flags&FLAG_d) {
 112                 for(start=end=0;str[end];end++)
 113                         if(isspace(str[end]) || isalnum(str[end])) str[start++]=str[end];
 114                 str[start]=0;
 115         }
 116         /* Handle -i */
 117         if(flags&FLAG_i) {
 118                 for(start=end=0;str[end];end++)
 119                         if(isprint(str[end])) str[start++]=str[end];
 120                 str[start]=0;
 121         }
 122         /* Handle -f */
 123         if(flags*FLAG_f) for(i=0;str[i];i++) str[i]=toupper(str[i]);
 124
 125         return str;
 126 }
 127
 128 static struct sort_key *add_key(void)
 129 {
 130         struct sort_key **pkey=&key_list;
 131         while(*pkey) pkey=&((*pkey)->next_key);
 132         return *pkey=xcalloc(1,sizeof(struct sort_key));
 133 }
 134
 135 #define GET_LINE(fp) (global_flags&FLAG_z) ? bb_get_chunk_from_file(fp,NULL) \
 136                                                                                    : bb_get_chomped_line_from_file(fp)
 137 #else
 138 #define GET_LINE(fp)            bb_get_chomped_line_from_file(fp)
 139 #endif
 140
 141 /* Iterate through keys list and perform comparisons */
 142 static int compare_keys(const void *xarg, const void *yarg)
 143 {
 144         int flags=global_flags,retval=0;
 145         char *x,*y;
 146
 147 #ifdef CONFIG_FEATURE_SORT_BIG
 148         struct sort_key *key;
 149
 150         for(key=key_list;!retval && key;key=key->next_key) {
 151                 flags=(key->flags) ? key->flags : global_flags;
 152                 /* Chop out and modify key chunks, handling -dfib */
 153                 x=get_key(*(char **)xarg,key,flags);
 154                 y=get_key(*(char **)yarg,key,flags);
 155 #else
 156         /* This curly bracket serves no purpose but to match the nesting
 157            level of the for() loop we're not using */
 158         {
 159                 x=*(char **)xarg;
 160                 y=*(char **)yarg;
 161 #endif
 162                 /* Perform actual comparison */
 163                 switch(flags&7) {
 164                         default:
 165                                 bb_error_msg_and_die("Unknown sort type.");
 166                                 break;
 167                         /* Ascii sort */
 168                         case 0:
 169                                 retval=strcmp(x,y);
 170                                 break;
 171 #ifdef CONFIG_FEATURE_SORT_BIG
 172                         case FLAG_g:
 173                         {
 174                                 char *xx,*yy;
 175                                 double dx=strtod(x,&xx), dy=strtod(y,&yy);
 176                                 /* not numbers < NaN < -infinity < numbers < +infinity) */
 177                                 if(x==xx) retval=(y==yy ? 0 : -1);
 178                                 else if(y==yy) retval=1;
 179                                 else if(isnan(dx)) retval=isnan(dy) ? 0 : -1;
 180                                 else if(isnan(dy)) retval=1;
 181                                 else if(isinf(dx)) {
 182                                         if(dx<0) retval=((isinf(dy) && dy<0) ? 0 : -1);
 183                                         else retval=((isinf(dy) && dy>0) ? 0 : 1);
 184                                 } else if(isinf(dy)) retval=dy<0 ? 1 : -1;
 185                                 else retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 186                                 break;
 187                         }
 188                         case FLAG_M:
 189                         {
 190                                 struct tm thyme;
 191                                 int dx;
 192                                 char *xx,*yy;
 193
 194                                 xx=strptime(x,"%b",&thyme);
 195                                 dx=thyme.tm_mon;
 196                                 yy=strptime(y,"%b",&thyme);
 197                                 if(!xx) retval=(!yy ? 0 : -1);
 198                                 else if(!yy) retval=1;
 199                                 else retval=(dx==thyme.tm_mon ? 0 : dx-thyme.tm_mon);
 200                                 break;
 201                         }
 202                         /* Full floating point version of -n */
 203                         case FLAG_n:
 204                         {
 205                                 double dx=atof(x),dy=atof(y);
 206                                 retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 207                                 break;
 208                         }
 209                 }
 210                 /* Free key copies. */
 211                 if(x!=*(char **)xarg) free(x);
 212                 if(y!=*(char **)yarg) free(y);
 213                 if(retval) break;
 214 #else
 215                         /* Integer version of -n for tiny systems */
 216                         case FLAG_n:
 217                                 retval=atoi(x)-atoi(y);
 218                                 break;
 219                 }
 220 #endif
 221         }
 222         /* Perform fallback sort if necessary */
 223         if(!retval && !(global_flags&FLAG_s))
 224                         retval=strcmp(*(char **)xarg, *(char **)yarg);
 225 //dprintf(2,"reverse=%d\n",flags&FLAG_r);
 226         return ((flags&FLAG_r)?-1:1)*retval;
 227 }
 228
 229 int sort_main(int argc, char **argv)
 230 {
 231         FILE *fp,*outfile=NULL;
 232         int linecount=0,i,flag;
 233         char *line,**lines=NULL,*optlist="ngMucszbrdfimS:T:o:k:t:";
 234         int c;
 235
 236         bb_default_error_retval = 2;
 237         /* Parse command line options */
 238         while((c=getopt(argc,argv,optlist))>0) {
 239                 line=strchr(optlist,c);
 240                 if(!line) bb_show_usage();
 241                 switch(*line) {
 242 #ifdef CONFIG_FEATURE_SORT_BIG
 243                         case 'o':
 244                                 if(outfile) bb_error_msg_and_die("Too many -o.");
 245                                 outfile=bb_xfopen(optarg,"w");
 246                                 break;
 247                         case 't':
 248                                 if(key_separator || optarg[1])
 249                                         bb_error_msg_and_die("Too many -t.");
 250                                 key_separator=*optarg;
 251                                 break;
 252                         /* parse sort key */
 253                         case 'k':
 254                         {
 255                                 struct sort_key *key=add_key();
 256                                 char *temp, *temp2;
 257
 258                                 temp=optarg;
 259                                 for(i=0;*temp;) {
 260                                         /* Start of range */
 261                                         key->range[2*i]=(unsigned short)strtol(temp,&temp,10);
 262                                         if(*temp=='.')
 263                                                 key->range[(2*i)+1]=(unsigned short)strtol(temp+1,&temp,10);
 264                                         for(;*temp;temp++) {
 265                                                 if(*temp==',' && !i++) {
 266                                                         temp++;
 267                                                         break;
 268                                                 } /* no else needed: fall through to syntax error
 269                                                          because comma isn't in optlist */
 270                                                 temp2=strchr(optlist,*temp);
 271                                                 flag=(1<<(temp2-optlist));
 272                                                 if(!temp2 || (flag>FLAG_M && flag<FLAG_b))
 273                                                         bb_error_msg_and_die("Unknown key option.");
 274                                                 /* b after , means strip _trailing_ space */
 275                                                 if(i && flag==FLAG_b) flag=FLAG_bb;
 276                                                 key->flags|=flag;
 277                                         }
 278                                 }
 279                                 break;
 280                         }
 281 #endif
 282                         default:
 283                                 global_flags|=(1<<(line-optlist));
 284                                 /* global b strips leading and trailing spaces */
 285                                 if(global_flags&FLAG_b) global_flags|=FLAG_bb;
 286                                 break;
 287                 }
 288         }
 289         /* Open input files and read data */
 290         for(i=argv[optind] ? optind : optind-1;argv[i];i++) {
 291                 if(i<optind || (*argv[i]=='-' && !argv[i][1])) fp=stdin;
 292                 else fp=bb_xfopen(argv[i],"r");
 293                 for(;;) {
 294                         line=GET_LINE(fp);
 295                         if(!line) break;
 296                         if(!(linecount&63))
 297                                 lines=xrealloc(lines, sizeof(char *)*(linecount+64));
 298                         lines[linecount++]=line;
 299                 }
 300                 fclose(fp);
 301         }
 302 #ifdef CONFIG_FEATURE_SORT_BIG
 303         /* if no key, perform alphabetic sort */
 304     if(!key_list) add_key()->range[0]=1;
 305         /* handle -c */
 306         if(global_flags&FLAG_c) {
 307                 int j=(global_flags&FLAG_u) ? -1 : 0;
 308                 for(i=1;i<linecount;i++)
 309                         if(compare_keys(&lines[i-1],&lines[i])>j) {
 310                                 fprintf(stderr,"Check line %d\n",i);
 311                                 return 1;
 312                         }
 313                 return 0;
 314         }
 315 #endif
 316         /* Perform the actual sort */
 317         qsort(lines,linecount,sizeof(char *),compare_keys);
 318         /* handle -u */
 319         if(global_flags&FLAG_u) {
 320                 for(flag=0,i=1;i<linecount;i++) {
 321                         if(!compare_keys(&lines[flag],&lines[i])) free(lines[i]);
 322                         else lines[++flag]=lines[i];
 323                 }
 324                 if(linecount) linecount=flag+1;
 325         }
 326         /* Print it */
 327         if(!outfile) outfile=stdout;
 328         for(i=0;i<linecount;i++) fprintf(outfile,"%s\n",lines[i]);
 329         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 330 }