sys/src/libc/port/tokenize.c

   1 #include <u.h>
   2 #include <libc.h>
   3
   4 static char qsep[] = " \t\r\n";
   5
   6 static char*
   7 qtoken(char *s, char *sep)
   8 {
   9         int quoting;
  10         char *t;
  11
  12         quoting = 0;
  13         t = s;  /* s is output string, t is input string */
  14         while(*t!='\0' && (quoting || utfrune(sep, *t)==nil)){
  15                 if(*t != '\''){
  16                         *s++ = *t++;
  17                         continue;
  18                 }
  19                 /* *t is a quote */
  20                 if(!quoting){
  21                         quoting = 1;
  22                         t++;
  23                         continue;
  24                 }
  25                 /* quoting and we're on a quote */
  26                 if(t[1] != '\''){
  27                         /* end of quoted section; absorb closing quote */
  28                         t++;
  29                         quoting = 0;
  30                         continue;
  31                 }
  32                 /* doubled quote; fold one quote into two */
  33                 t++;
  34                 *s++ = *t++;
  35         }
  36         if(*s != '\0'){
  37                 *s = '\0';
  38                 if(t == s)
  39                         t++;
  40         }
  41         return t;
  42 }
  43
  44 static char*
  45 etoken(char *t, char *sep)
  46 {
  47         int quoting;
  48
  49         /* move to end of next token */
  50         quoting = 0;
  51         while(*t!='\0' && (quoting || utfrune(sep, *t)==nil)){
  52                 if(*t != '\''){
  53                         t++;
  54                         continue;
  55                 }
  56                 /* *t is a quote */
  57                 if(!quoting){
  58                         quoting = 1;
  59                         t++;
  60                         continue;
  61                 }
  62                 /* quoting and we're on a quote */
  63                 if(t[1] != '\''){
  64                         /* end of quoted section; absorb closing quote */
  65                         t++;
  66                         quoting = 0;
  67                         continue;
  68                 }
  69                 /* doubled quote; fold one quote into two */
  70                 t += 2;
  71         }
  72         return t;
  73 }
  74
  75 int
  76 gettokens(char *s, char **args, int maxargs, char *sep)
  77 {
  78         int nargs;
  79
  80         for(nargs=0; nargs<maxargs; nargs++){
  81                 while(*s!='\0' && utfrune(sep, *s)!=nil)
  82                         *s++ = '\0';
  83                 if(*s == '\0')
  84                         break;
  85                 args[nargs] = s;
  86                 s = etoken(s, sep);
  87         }
  88
  89         return nargs;
  90 }
  91
  92 int
  93 tokenize(char *s, char **args, int maxargs)
  94 {
  95         int nargs;
  96
  97         for(nargs=0; nargs<maxargs; nargs++){
  98                 while(*s!='\0' && utfrune(qsep, *s)!=nil)
  99                         s++;
 100                 if(*s == '\0')
 101                         break;
 102                 args[nargs] = s;
 103                 s = qtoken(s, qsep);
 104         }
 105
 106         return nargs;
 107 }