merge

[plan9front.git] / sys / src / cmd / wc.c

diff --git a/sys/src/cmd/wc.c b/sys/src/cmd/wc.c
index 83940880e3b169ed3e7583c75928fbd9db7daefe..c457fe477683c5827553b2ce16e8a665838a8f91 100644 (file)
--- a/sys/src/cmd/wc.c
+++ b/sys/src/cmd/wc.c
@@ -3,17 +3,17 @@
  * Bugs:
  *     The only white space characters recognized are ' ', '\t' and '\n', even though
  *     ISO 10646 has many more blanks scattered through it.
- *     Should count characters that cannot occur in any rune (hex f0-ff) separately.
+ *     Should count characters that cannot occur in any rune (hex f8-ff) separately.
  *     Should count non-canonical runes (e.g. hex c1,80 instead of hex 40).
  */
 #include <u.h>
 #include <libc.h>
 #define        NBUF    (8*1024)
-uvlong nline, tnline, pline;
-uvlong nword, tnword, pword;
-uvlong nrune, tnrune, prune;
-uvlong nbadr, tnbadr, pbadr;
-uvlong nchar, tnchar, pchar;
+uvlong nline, tnline; int pline;
+uvlong nword, tnword; int pword;
+uvlong nrune, tnrune; int prune;
+uvlong nbadr, tnbadr; int pbadr;
+uvlong nchar, tnchar; int pchar;
 void count(int, char *);
 void report(uvlong, uvlong, uvlong, uvlong, uvlong, char *);
 void
@@ -37,7 +37,7 @@ main(int argc, char *argv[])
                pchar = 1;
        }
        if(argc==0)
-               count(0, 0);
+               count(0, nil);
        else{
                for(i=0;i<argc;i++){
                        f=open(argv[i], OREAD);
@@ -63,33 +63,23 @@ main(int argc, char *argv[])
 void
 report(uvlong nline, uvlong nword, uvlong nrune, uvlong nbadr, uvlong nchar, char *fname)
 {
-       char line[1024], word[128];
+       char line[128], *e;
        line[0] = '\0';
-       if(pline){
-               sprint(word, " %7llud", nline);
-               strcat(line, word);
-       }
-       if(pword){
-               sprint(word, " %7llud", nword);
-               strcat(line, word);
-       }
-       if(prune){
-               sprint(word, " %7llud", nrune);
-               strcat(line, word);
-       }
-       if(pbadr){
-               sprint(word, " %7llud", nbadr);
-               strcat(line, word);
-       }
-       if(pchar){
-               sprint(word, " %7llud", nchar);
-               strcat(line, word);
-       }
-       if(fname){
-               sprint(word, " %s",   fname);
-               strcat(line, word);
-       }
-       print("%s\n", line+1);
+       e = line;
+       if(pline)
+               e += sprint(e, " %7llud", nline);
+       if(pword)
+               e += sprint(e, " %7llud", nword);
+       if(prune)
+               e += sprint(e, " %7llud", nrune);
+       if(pbadr)
+               e += sprint(e, " %7llud", nbadr);
+       if(pchar)
+               sprint(e, " %7llud", nchar);
+       if(fname)
+               print("%s %s\n", line+1, fname);
+       else
+               print("%s\n", line+1);
 }
 /*
  * How it works.  Start in statesp.  Each time we read a character,
@@ -97,12 +87,13 @@ report(uvlong nline, uvlong nword, uvlong nrune, uvlong nbadr, uvlong nchar, cha
  * following table.  If we're not in statesp or statewd when done, the
  * file ends with a partial rune.
  *        |                character
- *  state |09,20| 0a  |00-7f|80-bf|c0-df|e0-ef|f0-ff
- * -------+-----+-----+-----+-----+-----+-----+-----
- * statesp|ASP  |ASPN |AWDW |AWDWX|AC2W |AC3W |AWDWX
- * statewd|ASP  |ASPN |AWD  |AWDX |AC2  |AC3  |AWDX
- * statec2|ASPX |ASPNX|AWDX |AWDR |AC2X |AC3X |AWDX
- * statec3|ASPX |ASPNX|AWDX |AC2R |AC2X |AC3X |AWDX
+ *  state |09,20| 0a  |00-7f|80-bf|c0-df|e0-ef|f0-f7|f8-ff
+ * -------+-----+-----+-----+-----+-----+-----+-----+-----
+ * statesp|ASP  |ASPN |AWDW |AWDWX|AC2W |AC3W |AC4W |AWDWX
+ * statewd|ASP  |ASPN |AWD  |AWDX |AC2  |AC3  |AC4  |AWDX
+ * statec2|ASPX |ASPNX|AWDX |AWDR |AC2X |AC3X |AC4X |AWDX
+ * statec3|ASPX |ASPNX|AWDX |AC2R |AC2X |AC3X |AC4X |AWDX
+ * statec4|ASPX |ASPNX|AWDX |AC3R |AC2X |AC3X |AC4X |AWDX
  */
 enum{                  /* actions */
        AC2,            /* enter statec2 */
@@ -110,8 +101,12 @@ enum{                      /* actions */
        AC2W,           /* enter statec2, count a word */
        AC2X,           /* enter statec2, count a bad rune */
        AC3,            /* enter statec3 */
+       AC3R,           /* enter statec3, don't count a rune */
        AC3W,           /* enter statec3, count a word */
        AC3X,           /* enter statec3, count a bad rune */
+       AC4,            /* enter statec4 */
+       AC4W,           /* enter statec4, count a word */
+       AC4X,           /* enter statec4, count a bad rune */
        ASP,            /* enter statesp */
        ASPN,           /* enter statesp, count a newline */
        ASPNX,          /* enter statesp, count a newline, count a bad rune */
@@ -153,10 +148,10 @@ AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W,   /* d0-d7 */
 AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W,        /* d8-df */
 AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W,        /* e0-e7 */
 AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W,        /* e8-ef */
-AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* f0-f7 */
+AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, /* f0-f7 */
 AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* f8-ff */
 };
-uchar statewd[256]={   /* looking for the next character in a word */
+uchar statewd[256] = { /* looking for the next character in a word */
 AWD,  AWD,  AWD,  AWD,  AWD,  AWD,  AWD,  AWD, /* 00-07 */
 AWD,  ASP,  ASPN, AWD,  AWD,  AWD,  AWD,  AWD, /* 08-0f */
 AWD,  AWD,  AWD,  AWD,  AWD,  AWD,  AWD,  AWD, /* 10-17 */
@@ -187,10 +182,10 @@ AC2,  AC2,  AC2,  AC2,  AC2,  AC2,  AC2,  AC2,    /* d0-d7 */
 AC2,  AC2,  AC2,  AC2,  AC2,  AC2,  AC2,  AC2, /* d8-df */
 AC3,  AC3,  AC3,  AC3,  AC3,  AC3,  AC3,  AC3, /* e0-e7 */
 AC3,  AC3,  AC3,  AC3,  AC3,  AC3,  AC3,  AC3, /* e8-ef */
-AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f0-f7 */
+AC4,  AC4,  AC4,  AC4,  AC4,  AC4,  AC4,  AC4,  /* f0-f7 */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f8-ff */
 };
-uchar statec2[256]={   /* looking for 10xxxxxx to complete a rune */
+uchar statec2[256] = { /* looking for 10xxxxxx to complete a rune */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 00-07 */
 AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX,        /* 08-0f */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 10-17 */
@@ -221,10 +216,10 @@ AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,   /* d0-d7 */
 AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* d8-df */
 AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e0-e7 */
 AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e8-ef */
-AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f0-f7 */
+AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X,        /* f0-f7 */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f8-ff */
 };
-uchar statec3[256]={   /* looking for 10xxxxxx,10xxxxxx to complete a rune */
+uchar statec3[256] = { /* looking for 10xxxxxx,10xxxxxx to complete a rune */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 00-07 */
 AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX,        /* 08-0f */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 10-17 */
@@ -255,16 +250,51 @@ AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,   /* d0-d7 */
 AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* d8-df */
 AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e0-e7 */
 AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e8-ef */
-AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f0-f7 */
+AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X,        /* f0-f7 */
 AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f8-ff */
 };
+uchar statec4[256] = { /* looking for 10xxxxxx,10xxxxxx,10xxxxxx to complete a rune */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 00-07 */
+AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX,        /* 08-0f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 10-17 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 18-1f */
+ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 20-27 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 28-2f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 30-37 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 38-3f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 40-47 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 48-4f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 50-57 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 58-5f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 60-67 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 68-6f */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 70-77 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* 78-7f */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* 80-87 */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* 88-8f */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* 90-97 */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* 98-9f */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* a0-a7 */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* a8-af */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* b0-b7 */
+AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R,        /* b8-bf */
+AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* c0-c7 */
+AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* c8-cf */
+AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* d0-d7 */
+AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X,        /* d8-df */
+AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e0-e7 */
+AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X,        /* e8-ef */
+AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X,        /* f0-f7 */
+AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX,        /* f8-ff */
+};
+
 void
 count(int f, char *name)
 {
        int n;
        uchar buf[NBUF];
        uchar *bufp, *ebuf;
-       uchar *state=statesp;
+       uchar *state = statesp;
 
        nline = 0;
        nword = 0;
@@ -287,8 +317,12 @@ count(int f, char *name)
                        case AC2W:  state=statec2; nword++;          break;
                        case AC2X:  state=statec2;          nbadr++; break;
                        case AC3:   state=statec3;                   break;
+                       case AC3R:  state=statec3; --nrune;          break;
                        case AC3W:  state=statec3; nword++;          break;
                        case AC3X:  state=statec3;          nbadr++; break;
+                       case AC4:   state=statec4;                   break;
+                       case AC4W:  state=statec4; nword++;          break;
+                       case AC4X:  state=statec4;          nbadr++; break;
                        case ASP:   state=statesp;                   break;
                        case ASPN:  state=statesp; nline++;          break;
                        case ASPNX: state=statesp; nline++; nbadr++; break;
@@ -301,7 +335,7 @@ count(int f, char *name)
                        }
                }while(++bufp!=ebuf);
        }
-       if(state!=statesp && state!=statewd)
+       if(state != statesp && state != statewd)
                nbadr++;
        if(n<0)
                perror(name);