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 bc7cb1a1 2003-11-23 devnull * state |09,20| 0a |00-7f|80-bf|c0-df|e0-ef|f0-ff
101 bc7cb1a1 2003-11-23 devnull * -------+-----+-----+-----+-----+-----+-----+-----
102 bc7cb1a1 2003-11-23 devnull * statesp|ASP |ASPN |AWDW |AWDWX|AC2W |AC3W |AWDWX
103 bc7cb1a1 2003-11-23 devnull * statewd|ASP |ASPN |AWD |AWDX |AC2 |AC3 |AWDX
104 bc7cb1a1 2003-11-23 devnull * statec2|ASPX |ASPNX|AWDX |AWDR |AC2X |AC3X |AWDX
105 bc7cb1a1 2003-11-23 devnull * statec3|ASPX |ASPNX|AWDX |AC2R |AC2X |AC3X |AWDX
106 bc7cb1a1 2003-11-23 devnull */
107 bc7cb1a1 2003-11-23 devnull enum{ /* actions */
108 bc7cb1a1 2003-11-23 devnull AC2, /* enter statec2 */
109 bc7cb1a1 2003-11-23 devnull AC2R, /* enter statec2, don't count a rune */
110 bc7cb1a1 2003-11-23 devnull AC2W, /* enter statec2, count a word */
111 bc7cb1a1 2003-11-23 devnull AC2X, /* enter statec2, count a bad rune */
112 bc7cb1a1 2003-11-23 devnull AC3, /* enter statec3 */
113 bc7cb1a1 2003-11-23 devnull AC3W, /* enter statec3, count a word */
114 bc7cb1a1 2003-11-23 devnull AC3X, /* enter statec3, count a bad rune */
115 bc7cb1a1 2003-11-23 devnull ASP, /* enter statesp */
116 bc7cb1a1 2003-11-23 devnull ASPN, /* enter statesp, count a newline */
117 bc7cb1a1 2003-11-23 devnull ASPNX, /* enter statesp, count a newline, count a bad rune */
118 bc7cb1a1 2003-11-23 devnull ASPX, /* enter statesp, count a bad rune */
119 bc7cb1a1 2003-11-23 devnull AWD, /* enter statewd */
120 bc7cb1a1 2003-11-23 devnull AWDR, /* enter statewd, don't count a rune */
121 bc7cb1a1 2003-11-23 devnull AWDW, /* enter statewd, count a word */
122 bc7cb1a1 2003-11-23 devnull AWDWX, /* enter statewd, count a word, count a bad rune */
123 cbeb0b26 2006-04-01 devnull AWDX /* enter statewd, count a bad rune */
124 bc7cb1a1 2003-11-23 devnull };
125 bc7cb1a1 2003-11-23 devnull uchar statesp[256]={ /* looking for the start of a word */
126 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 00-07 */
127 bc7cb1a1 2003-11-23 devnull AWDW, ASP, ASPN, AWDW, AWDW, AWDW, AWDW, AWDW, /* 08-0f */
128 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 10-17 */
129 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 18-1f */
130 bc7cb1a1 2003-11-23 devnull ASP, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 20-27 */
131 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 28-2f */
132 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 30-37 */
133 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 38-3f */
134 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 40-47 */
135 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 48-4f */
136 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 50-57 */
137 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 58-5f */
138 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 60-67 */
139 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 68-6f */
140 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 70-77 */
141 bc7cb1a1 2003-11-23 devnull AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, AWDW, /* 78-7f */
142 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 80-87 */
143 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 88-8f */
144 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 90-97 */
145 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* 98-9f */
146 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* a0-a7 */
147 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* a8-af */
148 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* b0-b7 */
149 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* b8-bf */
150 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* c0-c7 */
151 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* c8-cf */
152 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* d0-d7 */
153 bc7cb1a1 2003-11-23 devnull AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, AC2W, /* d8-df */
154 bc7cb1a1 2003-11-23 devnull AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, /* e0-e7 */
155 bc7cb1a1 2003-11-23 devnull AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, AC3W, /* e8-ef */
156 bc7cb1a1 2003-11-23 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,/* f0-f7 */
157 cbeb0b26 2006-04-01 devnull AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX,AWDWX/* f8-ff */
158 bc7cb1a1 2003-11-23 devnull };
159 bc7cb1a1 2003-11-23 devnull uchar statewd[256]={ /* looking for the next character in a word */
160 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 00-07 */
161 bc7cb1a1 2003-11-23 devnull AWD, ASP, ASPN, AWD, AWD, AWD, AWD, AWD, /* 08-0f */
162 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 10-17 */
163 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 18-1f */
164 bc7cb1a1 2003-11-23 devnull ASP, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 20-27 */
165 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 28-2f */
166 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 30-37 */
167 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 38-3f */
168 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 40-47 */
169 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 48-4f */
170 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 50-57 */
171 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 58-5f */
172 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 60-67 */
173 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 68-6f */
174 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 70-77 */
175 bc7cb1a1 2003-11-23 devnull AWD, AWD, AWD, AWD, AWD, AWD, AWD, AWD, /* 78-7f */
176 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 80-87 */
177 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 88-8f */
178 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 90-97 */
179 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 98-9f */
180 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a0-a7 */
181 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* a8-af */
182 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b0-b7 */
183 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* b8-bf */
184 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* c0-c7 */
185 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* c8-cf */
186 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* d0-d7 */
187 bc7cb1a1 2003-11-23 devnull AC2, AC2, AC2, AC2, AC2, AC2, AC2, AC2, /* d8-df */
188 bc7cb1a1 2003-11-23 devnull AC3, AC3, AC3, AC3, AC3, AC3, AC3, AC3, /* e0-e7 */
189 bc7cb1a1 2003-11-23 devnull AC3, AC3, AC3, AC3, AC3, AC3, AC3, AC3, /* e8-ef */
190 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f0-f7 */
191 cbeb0b26 2006-04-01 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX /* f8-ff */
192 bc7cb1a1 2003-11-23 devnull };
193 bc7cb1a1 2003-11-23 devnull uchar statec2[256]={ /* looking for 10xxxxxx to complete a rune */
194 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 00-07 */
195 bc7cb1a1 2003-11-23 devnull AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX, /* 08-0f */
196 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 10-17 */
197 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 18-1f */
198 bc7cb1a1 2003-11-23 devnull ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 20-27 */
199 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 28-2f */
200 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 30-37 */
201 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 38-3f */
202 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 40-47 */
203 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 48-4f */
204 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 50-57 */
205 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 58-5f */
206 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 60-67 */
207 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 68-6f */
208 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 70-77 */
209 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 78-7f */
210 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 80-87 */
211 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 88-8f */
212 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 90-97 */
213 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* 98-9f */
214 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* a0-a7 */
215 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* a8-af */
216 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* b0-b7 */
217 bc7cb1a1 2003-11-23 devnull AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, AWDR, /* b8-bf */
218 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c0-c7 */
219 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c8-cf */
220 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d0-d7 */
221 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d8-df */
222 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e0-e7 */
223 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e8-ef */
224 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f0-f7 */
225 cbeb0b26 2006-04-01 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX /* f8-ff */
226 bc7cb1a1 2003-11-23 devnull };
227 bc7cb1a1 2003-11-23 devnull uchar statec3[256]={ /* looking for 10xxxxxx,10xxxxxx to complete a rune */
228 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 00-07 */
229 bc7cb1a1 2003-11-23 devnull AWDX, ASPX, ASPNX,AWDX, AWDX, AWDX, AWDX, AWDX, /* 08-0f */
230 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 10-17 */
231 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 18-1f */
232 bc7cb1a1 2003-11-23 devnull ASPX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 20-27 */
233 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 28-2f */
234 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 30-37 */
235 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 38-3f */
236 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 40-47 */
237 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 48-4f */
238 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 50-57 */
239 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 58-5f */
240 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 60-67 */
241 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 68-6f */
242 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 70-77 */
243 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* 78-7f */
244 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 80-87 */
245 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 88-8f */
246 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 90-97 */
247 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* 98-9f */
248 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* a0-a7 */
249 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* a8-af */
250 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* b0-b7 */
251 bc7cb1a1 2003-11-23 devnull AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, AC2R, /* b8-bf */
252 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c0-c7 */
253 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* c8-cf */
254 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d0-d7 */
255 bc7cb1a1 2003-11-23 devnull AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, AC2X, /* d8-df */
256 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e0-e7 */
257 bc7cb1a1 2003-11-23 devnull AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, AC3X, /* e8-ef */
258 bc7cb1a1 2003-11-23 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, /* f0-f7 */
259 cbeb0b26 2006-04-01 devnull AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX, AWDX /* f8-ff */
260 bc7cb1a1 2003-11-23 devnull };
261 bc7cb1a1 2003-11-23 devnull void
262 bc7cb1a1 2003-11-23 devnull count(int f, char *name)
263 bc7cb1a1 2003-11-23 devnull {
264 bc7cb1a1 2003-11-23 devnull int n;
265 bc7cb1a1 2003-11-23 devnull uchar buf[NBUF];
266 bc7cb1a1 2003-11-23 devnull uchar *bufp, *ebuf;
267 bc7cb1a1 2003-11-23 devnull uchar *state=statesp;
268 bc7cb1a1 2003-11-23 devnull
269 bc7cb1a1 2003-11-23 devnull nline = 0;
270 bc7cb1a1 2003-11-23 devnull nword = 0;
271 bc7cb1a1 2003-11-23 devnull nrune = 0;
272 bc7cb1a1 2003-11-23 devnull nbadr = 0;
273 bc7cb1a1 2003-11-23 devnull nchar = 0;
274 bc7cb1a1 2003-11-23 devnull
275 bc7cb1a1 2003-11-23 devnull for(;;){
276 bc7cb1a1 2003-11-23 devnull n=read(f, buf, NBUF);
277 bc7cb1a1 2003-11-23 devnull if(n<=0)
278 bc7cb1a1 2003-11-23 devnull break;
279 bc7cb1a1 2003-11-23 devnull nchar+=n;
280 bc7cb1a1 2003-11-23 devnull nrune+=n; /* might be too large, gets decreased later */
281 bc7cb1a1 2003-11-23 devnull bufp=buf;
282 bc7cb1a1 2003-11-23 devnull ebuf=buf+n;
283 bc7cb1a1 2003-11-23 devnull do{
284 bc7cb1a1 2003-11-23 devnull switch(state[*bufp]){
285 bc7cb1a1 2003-11-23 devnull case AC2: state=statec2; break;
286 bc7cb1a1 2003-11-23 devnull case AC2R: state=statec2; --nrune; break;
287 bc7cb1a1 2003-11-23 devnull case AC2W: state=statec2; nword++; break;
288 bc7cb1a1 2003-11-23 devnull case AC2X: state=statec2; nbadr++; break;
289 bc7cb1a1 2003-11-23 devnull case AC3: state=statec3; break;
290 bc7cb1a1 2003-11-23 devnull case AC3W: state=statec3; nword++; break;
291 bc7cb1a1 2003-11-23 devnull case AC3X: state=statec3; nbadr++; break;
292 bc7cb1a1 2003-11-23 devnull case ASP: state=statesp; break;
293 bc7cb1a1 2003-11-23 devnull case ASPN: state=statesp; nline++; break;
294 bc7cb1a1 2003-11-23 devnull case ASPNX: state=statesp; nline++; nbadr++; break;
295 bc7cb1a1 2003-11-23 devnull case ASPX: state=statesp; nbadr++; break;
296 bc7cb1a1 2003-11-23 devnull case AWD: state=statewd; break;
297 bc7cb1a1 2003-11-23 devnull case AWDR: state=statewd; --nrune; break;
298 bc7cb1a1 2003-11-23 devnull case AWDW: state=statewd; nword++; break;
299 bc7cb1a1 2003-11-23 devnull case AWDWX: state=statewd; nword++; nbadr++; break;
300 bc7cb1a1 2003-11-23 devnull case AWDX: state=statewd; nbadr++; break;
301 bc7cb1a1 2003-11-23 devnull }
302 bc7cb1a1 2003-11-23 devnull }while(++bufp!=ebuf);
303 bc7cb1a1 2003-11-23 devnull }
304 bc7cb1a1 2003-11-23 devnull if(state!=statesp && state!=statewd)
305 bc7cb1a1 2003-11-23 devnull nbadr++;
306 bc7cb1a1 2003-11-23 devnull if(n<0)
307 bc7cb1a1 2003-11-23 devnull perror(name);
308 bc7cb1a1 2003-11-23 devnull report(nline, nword, nrune, nbadr, nchar, name);
309 bc7cb1a1 2003-11-23 devnull }