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  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  15  * General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License
  18  * along with this program; if not, write to the Free Software
  19  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  20  *
  21  * See SuS3 sort standard at:
  22  * http://www.opengroup.org/onlinepubs/007904975/utilities/sort.html
  23  */
  24
  25 #include <ctype.h>
  26 #include <math.h>
  27 #include <stdio.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <time.h>
  31 #include <unistd.h>
  32 #include "busybox.h"
  33
  34 static int global_flags;
  35
  36 /*
  37         sort [-m][-o output][-bdfinru][-t char][-k keydef]... [file...]
  38         sort -c [-bdfinru][-t char][-k keydef][file]
  39 */
  40
  41 /* These are sort types */
  42 #define FLAG_n                  1               /* Numeric sort */
  43 #define FLAG_g                  2               /* Sort using strtod() */
  44 #define FLAG_M                  4               /* Sort date */
  45 /* ucsz apply to root level only, not keys.  b at root level implies bb */
  46 #define FLAG_u                  8               /* Unique */
  47 #define FLAG_c                  16              /* Check: no output, exit(!ordered) */
  48 #define FLAG_s                  32              /* Stable sort, no ascii fallback at end */
  49 #define FLAG_z                  64              /* Input is null terminated, not \n */
  50 /* These can be applied to search keys, the previous four can't */
  51 #define FLAG_b                  128             /* Ignore leading blanks */
  52 #define FLAG_r                  256             /* Reverse */
  53 #define FLAG_d                  512             /* Ignore !(isalnum()|isspace()) */
  54 #define FLAG_f                  1024    /* Force uppercase */
  55 #define FLAG_i                  2048    /* Ignore !isprint() */
  56 #define FLAG_bb                 32768   /* Ignore trailing blanks  */
  57
  58
  59 #ifdef CONFIG_FEATURE_SORT_BIG
  60 static char key_separator;
  61
  62 static struct sort_key
  63 {
  64         struct sort_key *next_key;      /* linked list */
  65         unsigned short range[4];        /* start word, start char, end word, end char */
  66         int flags;
  67 } *key_list;
  68
  69 static char *get_key(char *str, struct sort_key *key, int flags)
  70 {
  71         int start=0,end,len,i,j;
  72
  73         /* Special case whole string, so we don't have to make a copy */
  74         if(key->range[0]==1 && !key->range[1] && !key->range[2] && !key->range[3]
  75                 && !(flags&(FLAG_b&FLAG_d&FLAG_f&FLAG_i&FLAG_bb))) return str;
  76         /* Find start of key on first pass, end on second pass*/
  77         len=strlen(str);
  78
  79         for(j=0;j<2;j++) {
  80                 if(!key->range[2*j]) end=len;
  81                 /* Loop through fields */
  82                 else {
  83                         end=0;
  84                         for(i=1;i<key->range[2*j]+j;i++) {
  85                                 /* Skip leading blanks or first separator */
  86                                 if(str[end]) {
  87                                         if(key_separator) {
  88                                                 if(str[end]==key_separator) end++;
  89                                         } else if(isspace(str[end]))
  90                                                 while(isspace(str[end])) end++;
  91                                 }
  92                                 /* Skip body of key */
  93                                 for(;str[end];end++) {
  94                                         if(key_separator) {
  95                                                 if(str[end]==key_separator) break;
  96                                         } else if(isspace(str[end])) break;
  97                                 }
  98                         }
  99                 }
 100                 if(!j) start=end;
 101         }
 102         /* Key with explicit separator starts after separator */
 103         if(key_separator && str[start]==key_separator) start++;
 104         /* Strip leading whitespace if necessary */
 105         if(flags&FLAG_b) while(isspace(str[start])) start++;
 106         /* Strip trailing whitespace if necessary */
 107         if(flags&FLAG_bb) while(end>start && isspace(str[end-1])) end--;
 108         /* Handle offsets on start and end */
 109         if(key->range[3]) {
 110                 end+=key->range[3]-1;
 111                 if(end>len) end=len;
 112         }
 113         if(key->range[1]) {
 114                 start+=key->range[1]-1;
 115                 if(start>len) start=len;
 116         }
 117         /* Make the copy */
 118         if(end<start) end=start;
 119         str=bb_xstrndup(str+start,end-start);
 120         /* Handle -d */
 121         if(flags&FLAG_d) {
 122                 for(start=end=0;str[end];end++)
 123                         if(isspace(str[end]) || isalnum(str[end])) str[start++]=str[end];
 124                 str[start]=0;
 125         }
 126         /* Handle -i */
 127         if(flags&FLAG_i) {
 128                 for(start=end=0;str[end];end++)
 129                         if(isprint(str[end])) str[start++]=str[end];
 130                 str[start]=0;
 131         }
 132         /* Handle -f */
 133         if(flags*FLAG_f) for(i=0;str[i];i++) str[i]=toupper(str[i]);
 134
 135         return str;
 136 }
 137
 138 static struct sort_key *add_key(void)
 139 {
 140         struct sort_key **pkey=&key_list;
 141         while(*pkey) pkey=&((*pkey)->next_key);
 142         return *pkey=xcalloc(1,sizeof(struct sort_key));
 143 }
 144
 145 #define GET_LINE(fp) (global_flags&FLAG_z) ? bb_get_chunk_from_file(fp) \
 146                                                                                    : bb_get_chomped_line_from_file(fp)
 147 #else
 148 #define GET_LINE(fp)            bb_get_chomped_line_from_file(fp)
 149 #endif
 150
 151 /* Iterate through keys list and perform comparisons */
 152 static int compare_keys(const void *xarg, const void *yarg)
 153 {
 154         int flags=global_flags,retval=0;
 155         char *x,*y;
 156
 157 #ifdef CONFIG_FEATURE_SORT_BIG
 158         struct sort_key *key;
 159
 160         for(key=key_list;!retval && key;key=key->next_key) {
 161                 flags=(key->flags) ? key->flags : global_flags;
 162                 /* Chop out and modify key chunks, handling -dfib */
 163                 x=get_key(*(char **)xarg,key,flags);
 164                 y=get_key(*(char **)yarg,key,flags);
 165 #else
 166         /* This curly bracket serves no purpose but to match the nesting
 167            level of the for() loop we're not using */
 168         {
 169                 x=*(char **)xarg;
 170                 y=*(char **)yarg;
 171 #endif
 172                 /* Perform actual comparison */
 173                 switch(flags&7) {
 174                         default:
 175                                 bb_error_msg_and_die("Unknown sort type.");
 176                                 break;
 177                         /* Ascii sort */
 178                         case 0:
 179                                 retval=strcmp(x,y);
 180                                 break;
 181 #ifdef CONFIG_FEATURE_SORT_BIG
 182                         case FLAG_g:
 183                         {
 184                                 char *xx,*yy;
 185                                 double dx=strtod(x,&xx), dy=strtod(y,&yy);
 186                                 /* not numbers < NaN < -infinity < numbers < +infinity) */
 187                                 if(x==xx) retval=(y==yy ? 0 : -1);
 188                                 else if(y==yy) retval=1;
 189                                 else if(isnan(dx)) retval=isnan(dy) ? 0 : -1;
 190                                 else if(isnan(dy)) retval=1;
 191                                 else if(isinf(dx)) {
 192                                         if(dx<0) retval=((isinf(dy) && dy<0) ? 0 : -1);
 193                                         else retval=((isinf(dy) && dy>0) ? 0 : 1);
 194                                 } else if(isinf(dy)) retval=dy<0 ? 1 : -1;
 195                                 else retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 196                                 break;
 197                         }
 198                         case FLAG_M:
 199                         {
 200                                 struct tm thyme;
 201                                 int dx;
 202                                 char *xx,*yy;
 203
 204                                 xx=strptime(x,"%b",&thyme);
 205                                 dx=thyme.tm_mon;
 206                                 yy=strptime(y,"%b",&thyme);
 207                                 if(!xx) retval=(!yy ? 0 : -1);
 208                                 else if(!yy) retval=1;
 209                                 else retval=(dx==thyme.tm_mon ? 0 : dx-thyme.tm_mon);
 210                                 break;
 211                         }
 212                         /* Full floating point version of -n */
 213                         case FLAG_n:
 214                         {
 215                                 double dx=atof(x),dy=atof(y);
 216                                 retval=dx>dy ? 1 : (dx<dy ? -1 : 0);
 217                                 break;
 218                         }
 219                 }
 220                 /* Free key copies. */
 221                 if(x!=*(char **)xarg) free(x);
 222                 if(y!=*(char **)yarg) free(y);
 223                 if(retval) break;
 224 #else
 225                         /* Integer version of -n for tiny systems */
 226                         case FLAG_n:
 227                                 retval=atoi(x)-atoi(y);
 228                                 break;
 229                 }
 230 #endif
 231         }
 232         /* Perform fallback sort if necessary */
 233         if(!retval && !(global_flags&FLAG_s))
 234                         retval=strcmp(*(char **)xarg, *(char **)yarg);
 235 //dprintf(2,"reverse=%d\n",flags&FLAG_r);
 236         return ((flags&FLAG_r)?-1:1)*retval;
 237 }
 238
 239 int sort_main(int argc, char **argv)
 240 {
 241         FILE *fp,*outfile=NULL;
 242         int linecount=0,i,flag;
 243         char *line,**lines=NULL,*optlist="ngMucszbrdfimS:T:o:k:t:";
 244         int c;
 245
 246         bb_default_error_retval = 2;
 247         /* Parse command line options */
 248         while((c=getopt(argc,argv,optlist))>0) {
 249                 line=index(optlist,c);
 250                 if(!line) bb_show_usage();
 251                 switch(*line) {
 252 #ifdef CONFIG_FEATURE_SORT_BIG
 253                         case 'o':
 254                                 if(outfile) bb_error_msg_and_die("Too many -o.");
 255                                 outfile=bb_xfopen(optarg,"w");
 256                                 break;
 257                         case 't':
 258                                 if(key_separator || optarg[1])
 259                                         bb_error_msg_and_die("Too many -t.");
 260                                 key_separator=*optarg;
 261                                 break;
 262                         /* parse sort key */
 263                         case 'k':
 264                         {
 265                                 struct sort_key *key=add_key();
 266                                 char *temp, *temp2;
 267
 268                                 temp=optarg;
 269                                 for(i=0;*temp;) {
 270                                         /* Start of range */
 271                                         key->range[2*i]=(unsigned short)strtol(temp,&temp,10);
 272                                         if(*temp=='.')
 273                                                 key->range[(2*i)+1]=(unsigned short)strtol(temp+1,&temp,10);
 274                                         for(;*temp;temp++) {
 275                                                 if(*temp==',' && !i++) {
 276                                                         temp++;
 277                                                         break;
 278                                                 } /* no else needed: fall through to syntax error
 279                                                          because comma isn't in optlist */
 280                                                 temp2=index(optlist,*temp);
 281                                                 flag=(1<<(temp2-optlist));
 282                                                 if(!temp2 || (flag>FLAG_M && flag<FLAG_b))
 283                                                         bb_error_msg_and_die("Unknown key option.");
 284                                                 /* b after , means strip _trailing_ space */
 285                                                 if(i && flag==FLAG_b) flag=FLAG_bb;
 286                                                 key->flags|=flag;
 287                                         }
 288                                 }
 289                                 break;
 290                         }
 291 #endif
 292                         default:
 293                                 global_flags|=(1<<(line-optlist));
 294                                 /* global b strips leading and trailing spaces */
 295                                 if(global_flags&FLAG_b) global_flags|=FLAG_bb;
 296                                 break;
 297                 }
 298         }
 299         /* Open input files and read data */
 300         for(i=argv[optind] ? optind : optind-1;argv[i];i++) {
 301                 if(i<optind || (*argv[i]=='-' && !argv[i][1])) fp=stdin;
 302                 else fp=bb_xfopen(argv[i],"r");
 303                 for(;;) {
 304                         line=GET_LINE(fp);
 305                         if(!line) break;
 306                         if(!(linecount&63))
 307                                 lines=xrealloc(lines, sizeof(char *)*(linecount+64));
 308                         lines[linecount++]=line;
 309                 }
 310                 fclose(fp);
 311         }
 312 #ifdef CONFIG_FEATURE_SORT_BIG
 313         /* if no key, perform alphabetic sort */
 314     if(!key_list) add_key()->range[0]=1;
 315         /* handle -c */
 316         if(global_flags&FLAG_c) {
 317                 int j=(global_flags&FLAG_u) ? -1 : 0;
 318                 for(i=1;i<linecount;i++)
 319                         if(compare_keys(&lines[i-1],&lines[i])>j) {
 320                                 fprintf(stderr,"Check line %d\n",i);
 321                                 return 1;
 322                         }
 323                 return 0;
 324         }
 325 #endif
 326         /* Perform the actual sort */
 327         qsort(lines,linecount,sizeof(char *),compare_keys);
 328         /* handle -u */
 329         if(global_flags&FLAG_u) {
 330                 for(flag=0,i=1;i<linecount;i++) {
 331                         if(!compare_keys(&lines[flag],&lines[i])) free(lines[i]);
 332                         else lines[++flag]=lines[i];
 333                 }
 334                 if(linecount) linecount=flag+1;
 335         }
 336         /* Print it */
 337         if(!outfile) outfile=stdout;
 338         for(i=0;i<linecount;i++) fprintf(outfile,"%s\n",lines[i]);
 339         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 340 }