summary refs log tree commit diff
path: root/www/text.causal.agency/024-seprintf.7
blob: d1af2e1a6cd3a56067db7de2ded7e93db059b1ec (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
.Dd June 12, 2021
.Dt SEPRINTF 7
.Os "Causal Agency"
.
.Sh NAME
.Nm seprintf
.Nd an snprintf alternative
.
.Sh SYNOPSIS
.Ft "char *"
.Fn seprintf "char *ptr" "char *end" "const char *fmt" "..."
.
.Sh DESCRIPTION
While discussing string building in C recently,
mcf pointed out
.Xr seprint 2
from Plan 9,
and it kind of blew my mind.
I had implemented my own function in
.Xr catgirl 1
for building up strings using
.Xr snprintf 3
and a struct containing
pointer, length and capacity,
but it felt out of place.
.Fn seprintf
(I add the
.Dq f ,
Plan 9 doesn't)
is a much simpler
and more
.Dq C-like
interface with really nice usage patterns.
.
.Pp
The obvious difference from
.Xr snprintf 3
is that
.Fn seprintf
takes an
.Fa end
pointer
rather than a size.
This means you need only calculate it
once for each buffer,
rather than subtracting
the running length from the buffer size.
.Fn seprintf Ap s
return value is a pointer
to the terminating null
of the string it wrote,
so you can pass that back in
to continue appending
to the same buffer.
.
.Pp
I'm not sure of the exact behaviour on Plan 9,
but my implementation indicates truncation occurred
by returning the
.Fa end
pointer.
That makes it both easy to check,
and perfectly fine to keep calling
.Fn seprintf
anyway.
It just won't write anything if
.Fa ptr
==
.Fa end .
.
.Pp
In the case of formatting failure
(which should be prevented by
compile-time format string checking,
but should still be considered),
.Fn seprintf
returns
.Dv NULL .
I'm again not sure if this matches Plan 9.
I like this a lot better than
.Xr snprintf 3
returning -1,
because an unchecked
.Dv NULL
is likely to quickly cause a crash,
while blindly adding
.Xr snprintf 3 Ap s
return value
to your running length
is a non-obvious logic error.
.
.Sh EXAMPLES
Here's an example of what some code using
.Fn seprintf
might look like:
.Bd -literal -offset indent
char buf[4096];
char *ptr = buf, *end = &buf[sizeof(buf)];
ptr = seprintf(ptr, end, "argv: ");
for (int i = 1; i < argc; ++i) {
	ptr = seprintf(
		ptr, end, "%s%s",
		(i > 1 ? ", " : ""), argv[i]
	);
}
if (ptr == end) errx(1, "truncation occurred :(");
.Ed
.
.Pp
And here is the very short implementation of it against
.Xr vsnprintf 3
which I copy into my project header files:
.Bd -literal -offset indent
#include <stdarg.h>
#include <stdio.h>
static inline char *
seprintf(char *ptr, char *end, const char *fmt, ...)
	__attribute__((format(printf, 3, 4)));
static inline char *
seprintf(char *ptr, char *end, const char *fmt, ...) {
	va_list ap;
	va_start(ap, fmt);
	int n = vsnprintf(ptr, end - ptr, fmt, ap);
	va_end(ap);
	if (n < 0) return NULL;
	if (n > end - ptr) return end;
	return ptr + n;
}
.Ed
.
.Sh AUTHORS
.An june Aq Mt june@causal.agency
.Pp
Another short one before
.Xr git-subtree 1 .
I just think this function
is really neat.