Blame


1 bc7cb1a1 2003-11-23 devnull /*
2 bc7cb1a1 2003-11-23 devnull * wc -- count things in utf-encoded text files
3 bc7cb1a1 2003-11-23 devnull * Bugs:
4 bc7cb1a1 2003-11-23 devnull * The only white space characters recognized are ' ', '\t' and '\n', even though
5 bc7cb1a1 2003-11-23 devnull * ISO 10646 has many more blanks scattered through it.
6 bc7cb1a1 2003-11-23 devnull * Should count characters that cannot occur in any rune (hex f0-ff) separately.
7 bc7cb1a1 2003-11-23 devnull * Should count non-canonical runes (e.g. hex c1,80 instead of hex 40).
8 bc7cb1a1 2003-11-23 devnull */
9 bc7cb1a1 2003-11-23 devnull #include <u.h>
10 bc7cb1a1 2003-11-23 devnull #include <libc.h>
11 bc7cb1a1 2003-11-23 devnull #define NBUF (8*1024)
12 bc7cb1a1 2003-11-23 devnull uvlong nline, tnline, pline;
13 bc7cb1a1 2003-11-23 devnull uvlong nword, tnword, pword;
14 bc7cb1a1 2003-11-23 devnull uvlong nrune, tnrune, prune;
15 bc7cb1a1 2003-11-23 devnull uvlong nbadr, tnbadr, pbadr;
16 bc7cb1a1 2003-11-23 devnull uvlong nchar, tnchar, pchar;
17 bc7cb1a1 2003-11-23 devnull void count(int, char *);
18 bc7cb1a1 2003-11-23 devnull void report(uvlong, uvlong, uvlong, uvlong, uvlong, char *);
19 bc7cb1a1 2003-11-23 devnull void
20 bc7cb1a1 2003-11-23 devnull main(int argc, char *argv[])
21 bc7cb1a1 2003-11-23 devnull {
22 bc7cb1a1 2003-11-23 devnull char *status="";
23 bc7cb1a1 2003-11-23 devnull int i, f;
24 bc7cb1a1 2003-11-23 devnull ARGBEGIN {
25 bc7cb1a1 2003-11-23 devnull case 'l': pline++; break;
26 bc7cb1a1 2003-11-23 devnull case 'w': pword++; break;
27 bc7cb1a1 2003-11-23 devnull case 'r': prune++; break;
28 bc7cb1a1 2003-11-23 devnull case 'b': pbadr++; break;
29 bc7cb1a1 2003-11-23 devnull case 'c': pchar++; break;
30 bc7cb1a1 2003-11-23 devnull default:
31 bc7cb1a1 2003-11-23 devnull fprint(2, "Usage: %s [-lwrbc] [file ...]\n", argv0);
32 bc7cb1a1 2003-11-23 devnull exits("usage");
33 bc7cb1a1 2003-11-23 devnull } ARGEND
34 bc7cb1a1 2003-11-23 devnull if(pline+pword+prune+pbadr+pchar == 0) {
35 bc7cb1a1 2003-11-23 devnull pline = 1;
36 bc7cb1a1 2003-11-23 devnull pword = 1;
37 bc7cb1a1 2003-11-23 devnull pchar = 1;
38 bc7cb1a1 2003-11-23 devnull }
39 bc7cb1a1 2003-11-23 devnull if(argc==0)
40 bc7cb1a1 2003-11-23 devnull count(0, 0);
41 bc7cb1a1 2003-11-23 devnull else{
42 bc7cb1a1 2003-11-23 devnull for(i=0;i<argc;i++){
43 bc7cb1a1 2003-11-23 devnull f=open(argv[i], OREAD);
44 bc7cb1a1 2003-11-23 devnull if(f<0){
45 bc7cb1a1 2003-11-23 devnull perror(argv[i]);
46 bc7cb1a1 2003-11-23 devnull status="can't open";
47 bc7cb1a1 2003-11-23 devnull }
48 bc7cb1a1 2003-11-23 devnull else{
49 bc7cb1a1 2003-11-23 devnull count(f, argv[i]);
50 bc7cb1a1 2003-11-23 devnull tnline+=nline;
51 bc7cb1a1 2003-11-23 devnull tnword+=nword;
52 bc7cb1a1 2003-11-23 devnull tnrune+=nrune;
53 bc7cb1a1 2003-11-23 devnull tnbadr+=nbadr;
54 bc7cb1a1 2003-11-23 devnull tnchar+=nchar;
55 bc7cb1a1 2003-11-23 devnull close(f);
56 bc7cb1a1 2003-11-23 devnull }
57 bc7cb1a1 2003-11-23 devnull }
58 bc7cb1a1 2003-11-23 devnull if(argc>1)
59 bc7cb1a1 2003-11-23 devnull report(tnline, tnword, tnrune, tnbadr, tnchar, "total");
60 bc7cb1a1 2003-11-23 devnull }
61 bc7cb1a1 2003-11-23 devnull exits(status);
62 bc7cb1a1 2003-11-23 devnull }
63 bc7cb1a1 2003-11-23 devnull void
64 bc7cb1a1 2003-11-23 devnull report(uvlong nline, uvlong nword, uvlong nrune, uvlong nbadr, uvlong nchar, char *fname)
65 bc7cb1a1 2003-11-23 devnull {
66 bc7cb1a1 2003-11-23 devnull char line[1024], word[128];
67 bc7cb1a1 2003-11-23 devnull line[0] = '\0';
68 bc7cb1a1 2003-11-23 devnull if(pline){
69 bc7cb1a1 2003-11-23 devnull sprint(word, " %7llud", nline);
70 bc7cb1a1 2003-11-23 devnull strcat(line, word);
71 bc7cb1a1 2003-11-23 devnull }
72 bc7cb1a1 2003-11-23 devnull if(pword){
73 bc7cb1a1 2003-11-23 devnull sprint(word, " %7llud", nword);
74 bc7cb1a1 2003-11-23 devnull strcat(line, word);
75 bc7cb1a1 2003-11-23 devnull }
76 bc7cb1a1 2003-11-23 devnull if(prune){
77 bc7cb1a1 2003-11-23 devnull sprint(word, " %7llud", nrune);
78 bc7cb1a1 2003-11-23 devnull strcat(line, word);
79 bc7cb1a1 2003-11-23 devnull }
80 bc7cb1a1 2003-11-23 devnull if(pbadr){
81 bc7cb1a1 2003-11-23 devnull sprint(word, " %7llud", nbadr);
82 bc7cb1a1 2003-11-23 devnull strcat(line, word);
83 bc7cb1a1 2003-11-23 devnull }
84 bc7cb1a1 2003-11-23 devnull if(pchar){
85 bc7cb1a1 2003-11-23 devnull sprint(word, " %7llud", nchar);
86 bc7cb1a1 2003-11-23 devnull strcat(line, word);
87 bc7cb1a1 2003-11-23 devnull }
88 bc7cb1a1 2003-11-23 devnull if(fname){
89 bc7cb1a1 2003-11-23 devnull sprint(word, " %s", fname);
90 bc7cb1a1 2003-11-23 devnull strcat(line, word);
91 bc7cb1a1 2003-11-23 devnull }
92 bc7cb1a1 2003-11-23 devnull print("%s\n", line+1);
93 bc7cb1a1 2003-11-23 devnull }
94 bc7cb1a1 2003-11-23 devnull /*
95 bc7cb1a1 2003-11-23 devnull * How it works. Start in statesp. Each time we read a character,
96 bc7cb1a1 2003-11-23 devnull * increment various counts, and do state transitions according to the
97 bc7cb1a1 2003-11-23 devnull * following table. If we're not in statesp or statewd when done, the
98 bc7cb1a1 2003-11-23 devnull * file ends with a partial rune.
99 bc7cb1a1 2003-11-23 devnull * | character
100 02d8e824 2010-12-08 rsc * state |09,20| 0a |00-7f|80-bf|c0-df|e0-ef|f0-f7|f8-ff
101 02d8e824 2010-12-08 rsc * -------+-----+-----+-----+-----+-----+-----+-----+-----
102 02d8e824 2010-12-08 rsc * statesp|ASP |ASPN |AWDW |AWDWX|AC2W |AC3W |AC4W |AWDWX
103 02d8e824 2010-12-08 rsc * statewd|ASP |ASPN |AWD |AWDX |AC2 |AC3 |AC4 |AWDX
104 02d8e824 2010-12-08 rsc * statec2|ASPX |ASPNX|AWDX |AWDR |AC2X |AC3X |AC4X |AWDX
105 02d8e824 2010-12-08 rsc * statec3|ASPX |ASPNX|AWDX |AC2R |AC2X |AC3X |AC4X |AWDX
106 02d8e824 2010-12-08 rsc * statec4|ASPX |ASPNX|AWDX |AC3R |AC2X |AC3X |AC4X |AWDX f4 8f bf bf
107 bc7cb1a1 2003-11-23 devnull */
108 bc7cb1a1 2003-11-23 devnull enum{ /* actions */
109 bc7cb1a1 2003-11-23 devnull AC2, /* enter statec2 */
110 bc7cb1a1 2003-11-23 devnull AC2R, /* enter statec2, don't count a rune */
111 bc7cb1a1 2003-11-23 devnull AC2W, /* enter statec2, count a word */
112 bc7cb1a1 2003-11-23 devnull AC2X, /* enter statec2, count a bad rune */
113 bc7cb1a1 2003-11-23 devnull AC3, /* enter statec3 */
114 02d8e824 2010-12-08 rsc AC3R, /* enter statec3, don't count a rune */
115 bc7cb1a1 2003-11-23 devnull AC3W, /* enter statec3, count a word */
116 bc7cb1a1 2003-11-23 devnull AC3X, /* enter statec3, count a bad rune */
117 02d8e824 2010-12-08 rsc AC4, /* enter statec4 */
118 02d8e824 2010-12-08 rsc AC4W, /* enter statec4, count a word */
119 02d8e824 2010-12-08 rsc AC4X, /* enter statec4, count a bad rune */
120 bc7cb1a1 2003-11-23 devnull ASP, /* enter statesp */
121 bc7cb1a1 2003-11-23 devnull ASPN, /* enter statesp, count a newline */
122 bc7cb1a1 2003-11-23 devnull ASPNX, /* enter statesp, count a newline, count a bad rune */
123 bc7cb1a1 2003-11-23 devnull ASPX, /* enter statesp, count a bad rune */
124 bc7cb1a1 2003-11-23 devnull AWD, /* enter statewd */
125 bc7cb1a1 2003-11-23 devnull AWDR, /* enter statewd, don't count a rune */
126 bc7cb1a1 2003-11-23 devnull AWDW, /* enter statewd, count a word */
127 bc7cb1a1 2003-11-23 devnull AWDWX, /* enter statewd, count a word, count a bad rune */
128 02d8e824 2010-12-08 rsc AWDX, /* enter statewd, count a bad rune */
129 bc7cb1a1 2003-11-23 devnull };
130 bc7cb1a1 2003-11-23 devnull uchar statesp[256]={ /* looking for the start of a word */
131 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 00-07 */
132 bc7cb1a1 2003-11-23 devnull AWDW, ASP, ASPN, AWDW, AWDW, AWDW, AWDW, AWDW, /* 08-0f */
133 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 10-17 */
134 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 18-1f */
135 bc7cb1a1 2003-11-23 devnull ASP, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 20-27 */
136 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 28-2f */
137 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 30-37 */
138 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 38-3f */
139 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 40-47 */
140 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 48-4f */
141 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 50-57 */
142 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 58-5f */
143 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 60-67 */
144 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 68-6f */
145 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 70-77 */
146 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 78-7f */
147 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 80-87 */
148 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 88-8f */
149 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 90-97 */
150 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 98-9f */
151 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* a0-a7 */
152 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* a8-af */
153 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* b0-b7 */
154 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* b8-bf */
155 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* c0-c7 */
156 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* c8-cf */
157 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* d0-d7 */
158 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* d8-df */
159 bc7cb1a1 2003-11-23 devnull AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, /* e0-e7 */
160 bc7cb1a1 2003-11-23 devnull AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, /* e8-ef */
161 02d8e824 2010-12-08 rsc AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, AC4W, /* f0-f7 */
162 02d8e824 2010-12-08 rsc AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* f8-ff */
163 bc7cb1a1 2003-11-23 devnull };
164 bc7cb1a1 2003-11-23 devnull uchar statewd[256]={ /* looking for the next character in a word */
165 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 00-07 */
166 bc7cb1a1 2003-11-23 devnull AWD, ASP, ASPN, AWD, AWD, AWD, AWD, AWD, /* 08-0f */
167 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 10-17 */
168 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 18-1f */
169 bc7cb1a1 2003-11-23 devnull ASP, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 20-27 */
170 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 28-2f */
171 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 30-37 */
172 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 38-3f */
173 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 40-47 */
174 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 48-4f */
175 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 50-57 */
176 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 58-5f */
177 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 60-67 */
178 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 68-6f */
179 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 70-77 */
180 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 78-7f */
181 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 80-87 */
182 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 88-8f */
183 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 90-97 */
184 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 98-9f */
185 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a0-a7 */
186 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a8-af */
187 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b0-b7 */
188 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b8-bf */
189 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* c0-c7 */
190 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* c8-cf */
191 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* d0-d7 */
192 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* d8-df */
193 bc7cb1a1 2003-11-23 devnull AC3, AC3, AC3, AC3, AC3, AC3, AC3, AC3, /* e0-e7 */
194 bc7cb1a1 2003-11-23 devnull AC3, AC3, AC3, AC3, AC3, AC3, AC3, AC3, /* e8-ef */
195 02d8e824 2010-12-08 rsc AC4, AC4, AC4, AC4, AC4, AC4, AC4, AC4, /* f0-f7 */
196 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f8-ff */
197 bc7cb1a1 2003-11-23 devnull };
198 bc7cb1a1 2003-11-23 devnull uchar statec2[256]={ /* looking for 10xxxxxx to complete a rune */
199 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 00-07 */
200 bc7cb1a1 2003-11-23 devnull AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX, /* 08-0f */
201 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 10-17 */
202 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 18-1f */
203 bc7cb1a1 2003-11-23 devnull ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 20-27 */
204 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 28-2f */
205 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 30-37 */
206 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 38-3f */
207 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 40-47 */
208 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 48-4f */
209 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 50-57 */
210 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 58-5f */
211 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 60-67 */
212 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 68-6f */
213 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 70-77 */
214 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 78-7f */
215 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 80-87 */
216 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 88-8f */
217 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 90-97 */
218 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 98-9f */
219 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* a0-a7 */
220 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* a8-af */
221 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* b0-b7 */
222 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* b8-bf */
223 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c0-c7 */
224 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c8-cf */
225 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d0-d7 */
226 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d8-df */
227 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e0-e7 */
228 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e8-ef */
229 02d8e824 2010-12-08 rsc AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, /* f0-f7 */
230 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f8-ff */
231 bc7cb1a1 2003-11-23 devnull };
232 bc7cb1a1 2003-11-23 devnull uchar statec3[256]={ /* looking for 10xxxxxx,10xxxxxx to complete a rune */
233 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 00-07 */
234 bc7cb1a1 2003-11-23 devnull AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX, /* 08-0f */
235 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 10-17 */
236 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 18-1f */
237 bc7cb1a1 2003-11-23 devnull ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 20-27 */
238 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 28-2f */
239 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 30-37 */
240 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 38-3f */
241 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 40-47 */
242 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 48-4f */
243 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 50-57 */
244 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 58-5f */
245 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 60-67 */
246 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 68-6f */
247 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 70-77 */
248 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 78-7f */
249 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 80-87 */
250 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 88-8f */
251 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 90-97 */
252 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 98-9f */
253 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* a0-a7 */
254 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* a8-af */
255 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* b0-b7 */
256 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* b8-bf */
257 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c0-c7 */
258 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c8-cf */
259 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d0-d7 */
260 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d8-df */
261 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e0-e7 */
262 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e8-ef */
263 02d8e824 2010-12-08 rsc AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, /* f0-f7 */
264 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f8-ff */
265 bc7cb1a1 2003-11-23 devnull };
266 02d8e824 2010-12-08 rsc uchar statec4[256]={ /* looking for 10xxxxxx,10xxxxxx,10xxxxxx to complete a rune */
267 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 00-07 */
268 02d8e824 2010-12-08 rsc AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX, /* 08-0f */
269 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 10-17 */
270 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 18-1f */
271 02d8e824 2010-12-08 rsc ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 20-27 */
272 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 28-2f */
273 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 30-37 */
274 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 38-3f */
275 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 40-47 */
276 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 48-4f */
277 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 50-57 */
278 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 58-5f */
279 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 60-67 */
280 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 68-6f */
281 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 70-77 */
282 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 78-7f */
283 02d8e824 2010-12-08 rsc AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, /* 80-87 */
284 02d8e824 2010-12-08 rsc AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, /* 88-8f */
285 02d8e824 2010-12-08 rsc AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, AC3R, /* 90-97 */
286 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 98-9f */
287 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a0-a7 */
288 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a8-af */
289 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b0-b7 */
290 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b8-bf */
291 02d8e824 2010-12-08 rsc AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c0-c7 */
292 02d8e824 2010-12-08 rsc AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c8-cf */
293 02d8e824 2010-12-08 rsc AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d0-d7 */
294 02d8e824 2010-12-08 rsc AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d8-df */
295 02d8e824 2010-12-08 rsc AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e0-e7 */
296 02d8e824 2010-12-08 rsc AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e8-ef */
297 02d8e824 2010-12-08 rsc AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, AC4X, /* f0-f7 */
298 02d8e824 2010-12-08 rsc AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f8-ff */
299 02d8e824 2010-12-08 rsc };
300 bc7cb1a1 2003-11-23 devnull void
301 bc7cb1a1 2003-11-23 devnull count(int f, char *name)
302 bc7cb1a1 2003-11-23 devnull {
303 bc7cb1a1 2003-11-23 devnull int n;
304 bc7cb1a1 2003-11-23 devnull uchar buf[NBUF];
305 bc7cb1a1 2003-11-23 devnull uchar *bufp, *ebuf;
306 bc7cb1a1 2003-11-23 devnull uchar *state=statesp;
307 bc7cb1a1 2003-11-23 devnull
308 bc7cb1a1 2003-11-23 devnull nline = 0;
309 bc7cb1a1 2003-11-23 devnull nword = 0;
310 bc7cb1a1 2003-11-23 devnull nrune = 0;
311 bc7cb1a1 2003-11-23 devnull nbadr = 0;
312 bc7cb1a1 2003-11-23 devnull nchar = 0;
313 bc7cb1a1 2003-11-23 devnull
314 bc7cb1a1 2003-11-23 devnull for(;;){
315 bc7cb1a1 2003-11-23 devnull n=read(f, buf, NBUF);
316 bc7cb1a1 2003-11-23 devnull if(n<=0)
317 bc7cb1a1 2003-11-23 devnull break;
318 bc7cb1a1 2003-11-23 devnull nchar+=n;
319 bc7cb1a1 2003-11-23 devnull nrune+=n; /* might be too large, gets decreased later */
320 bc7cb1a1 2003-11-23 devnull bufp=buf;
321 bc7cb1a1 2003-11-23 devnull ebuf=buf+n;
322 bc7cb1a1 2003-11-23 devnull do{
323 bc7cb1a1 2003-11-23 devnull switch(state[*bufp]){
324 bc7cb1a1 2003-11-23 devnull case AC2: state=statec2; break;
325 bc7cb1a1 2003-11-23 devnull case AC2R: state=statec2; --nrune; break;
326 bc7cb1a1 2003-11-23 devnull case AC2W: state=statec2; nword++; break;
327 bc7cb1a1 2003-11-23 devnull case AC2X: state=statec2; nbadr++; break;
328 bc7cb1a1 2003-11-23 devnull case AC3: state=statec3; break;
329 02d8e824 2010-12-08 rsc case AC3R: state=statec3; --nrune; break;
330 bc7cb1a1 2003-11-23 devnull case AC3W: state=statec3; nword++; break;
331 bc7cb1a1 2003-11-23 devnull case AC3X: state=statec3; nbadr++; break;
332 02d8e824 2010-12-08 rsc case AC4: state=statec4; break;
333 02d8e824 2010-12-08 rsc case AC4W: state=statec4; nword++; break;
334 02d8e824 2010-12-08 rsc case AC4X: state=statec4; nbadr++; break;
335 bc7cb1a1 2003-11-23 devnull case ASP: state=statesp; break;
336 bc7cb1a1 2003-11-23 devnull case ASPN: state=statesp; nline++; break;
337 bc7cb1a1 2003-11-23 devnull case ASPNX: state=statesp; nline++; nbadr++; break;
338 bc7cb1a1 2003-11-23 devnull case ASPX: state=statesp; nbadr++; break;
339 bc7cb1a1 2003-11-23 devnull case AWD: state=statewd; break;
340 bc7cb1a1 2003-11-23 devnull case AWDR: state=statewd; --nrune; break;
341 bc7cb1a1 2003-11-23 devnull case AWDW: state=statewd; nword++; break;
342 bc7cb1a1 2003-11-23 devnull case AWDWX: state=statewd; nword++; nbadr++; break;
343 bc7cb1a1 2003-11-23 devnull case AWDX: state=statewd; nbadr++; break;
344 bc7cb1a1 2003-11-23 devnull }
345 bc7cb1a1 2003-11-23 devnull }while(++bufp!=ebuf);
346 bc7cb1a1 2003-11-23 devnull }
347 bc7cb1a1 2003-11-23 devnull if(state!=statesp && state!=statewd)
348 bc7cb1a1 2003-11-23 devnull nbadr++;
349 bc7cb1a1 2003-11-23 devnull if(n<0)
350 02d8e824 2010-12-08 rsc fprint(2, "%s: %r\n", name);
351 bc7cb1a1 2003-11-23 devnull report(nline, nword, nrune, nbadr, nchar, name);
352 bc7cb1a1 2003-11-23 devnull }