1 /*
2  * Deal with duplicated lines in a file
3  */
4 #include <u.h>
5 #include <libc.h>
6 #include <bio.h>
7 #include <ctype.h>
8 
9 #define	SIZE	8000
10 
11 int	fields	= 0;
12 int	letters	= 0;
13 int	linec	= 0;
14 char	mode;
15 int	uniq;
16 char	*b1, *b2;
17 long	bsize;
18 Biobuf	fin;
19 Biobuf	fout;
20 
21 int	gline(char *buf);
22 void	pline(char *buf);
23 int	equal(char *b1, char *b2);
24 char*	skip(char *s);
25 
26 void
27 main(int argc, char *argv[])
28 {
29 	int f;
30 
31 	bsize = SIZE;
32 	b1 = malloc(bsize);
33 	b2 = malloc(bsize);
34 	f = 0;
35 	while(argc > 1) {
36 		if(*argv[1] == '-') {
37 			if(isdigit((uchar)argv[1][1]))
38 				fields = atoi(&argv[1][1]);
39 			else
40 				mode = argv[1][1];
41 			argc--;
42 			argv++;
43 			continue;
44 		}
45 		if(*argv[1] == '+') {
46 			letters = atoi(&argv[1][1]);
47 			argc--;
48 			argv++;
49 			continue;
50 		}
51 		f = open(argv[1], 0);
52 		if(f < 0) {
53 			fprint(2, "cannot open %s\n", argv[1]);
54 			exits("open");
55 		}
56 		break;
57 	}
58 	if(argc > 2) {
59 		fprint(2, "unexpected argument %s\n", argv[2]);
60 		exits("arg");
61 	}
62 	Binit(&fin, f, OREAD);
63 	Binit(&fout, 1, OWRITE);
64 
65 	if(gline(b1))
66 		exits(0);
67 	for(;;) {
68 		linec++;
69 		if(gline(b2)) {
70 			pline(b1);
71 			exits(0);
72 		}
73 		if(!equal(b1, b2)) {
74 			pline(b1);
75 			linec = 0;
76 			do {
77 				linec++;
78 				if(gline(b1)) {
79 					pline(b2);
80 					exits(0);
81 				}
82 			} while(equal(b2, b1));
83 			pline(b2);
84 			linec = 0;
85 		}
86 	}
87 }
88 
89 int
90 gline(char *buf)
91 {
92 	char *p;
93 
94 	p = Brdline(&fin, '\n');
95 	if(p == 0)
96 		return 1;
97 	if(fin.rdline >= bsize-1) {
98 		fprint(2, "line too long\n");
99 		exits("too long");
100 	}
101 	memmove(buf, p, fin.rdline);
102 	buf[fin.rdline-1] = 0;
103 	return 0;
104 }
105 
106 void
107 pline(char *buf)
108 {
109 
110 	switch(mode) {
111 
112 	case 'u':
113 		if(uniq) {
114 			uniq = 0;
115 			return;
116 		}
117 		break;
118 
119 	case 'd':
120 		if(uniq)
121 			break;
122 		return;
123 
124 	case 'c':
125 		Bprint(&fout, "%4d ", linec);
126 	}
127 	uniq = 0;
128 	Bprint(&fout, "%s\n", buf);
129 }
130 
131 int
132 equal(char *b1, char *b2)
133 {
134 	char c;
135 
136 	if(fields || letters) {
137 		b1 = skip(b1);
138 		b2 = skip(b2);
139 	}
140 	for(;;) {
141 		c = *b1++;
142 		if(c != *b2++) {
143 			if(c == 0 && mode == 's')
144 				return 1;
145 			return 0;
146 		}
147 		if(c == 0) {
148 			uniq++;
149 			return 1;
150 		}
151 	}
152 }
153 
154 char*
155 skip(char *s)
156 {
157 	int nf, nl;
158 
159 	nf = nl = 0;
160 	while(nf++ < fields) {
161 		while(*s == ' ' || *s == '\t')
162 			s++;
163 		while(!(*s == ' ' || *s == '\t' || *s == 0) )
164 			s++;
165 	}
166 	while(nl++ < letters && *s != 0)
167 			s++;
168 	return s;
169 }