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
|
.TH UBFA 6
.SH NAME
ubfa \- universal binary format for data transport
.SH DESCRIPTION
.I UBF(A)
is the data transport encoding for Armstrong's
Universal Binary Format.
It provides four primitive types: atoms (symbolic constants), integers, strings, and binary data.
There are two compound types: fixed-length tuples and variable-length lists.
.IR Ubfa (2)
provides basic support in Limbo for reading and writing streams of UBF(A)-encoded data.
.PP
The
.I input
syntax is defined by the following rules:
.IP
.EX
.ft R
.ta \w'\f2simple-xxx\f1'u +\w'\ ::=\ 'u
\f2input\fP ::= \f2item\fP* '$'
\f2item\fP ::= \f2integer\fP | \f2atom\fP | \f2string\fP | \f2binary\fP | \f2tuple\fP | \f2list\fP | \f2store\fP | \f2push\fP | \f2comment\fP | \f2tag\fP
\f2integer\fP ::= \f5'-'\fP?\f5[0-9]\fP+
\f2atom\fP ::= "'" (\f5[^\e']\fP | '\e\e' | "\e'")* "'"
\f2string\fP ::= '"' (\f5[^\e"]\fP | '\e\e' | '\e"')* '"'
\f2binary\fP ::= '~' \f2byte\fP* '~' # preceded by \f2integer\fP byte count
\f2tuple\fP ::= '{' \f2item\fP* '}'
\f2list\fP ::= '#' (\f2item\fP '&')*
\f2store\fP ::= '>' \f2reg\fP
\f2push\fP ::= \f2reg\fP
\f2reg\fP ::= \f5[^-%"~'`{}#& \en\er\et,0-9]\fP
\f2comment\fP ::= '%' (\f5[^\e%]\fP | '\e\e' | '\e%')* '%'
\f2tag\fP ::= '`' (\f5[^\e`]\fP | '\e\e' | '\e`')* '`'
.EE
.PD
.DT
.PP
White space is any sequence of blank, tab, newline or carriage-return characters, and can appear
before or after any instance of
.I item
in the grammar.
.PP
The
.I input
data is interpreted by a simple virtual machine.
The machine contains a stack of values of primitive and compound types, and a set of registers also containing
values of those types.
White space and comments are ignored.
Primitive
.IR integer ,
.I atom
and
.IR string
values are pushed onto the stack as they are recognised.
Certain input bytes outside any value act as operators:
.TP
.B {
Note the current stack depth.
.TP
.B }
Pop stack values to restore the most recently noted stack depth.
Push a single value
representing a tuple of those items; the left-most value in the tuple is the last one popped
(the first in the original input stream).
.TP
.B ~
Pop an integer value
.I n
from the stack.
Read
.I n
bytes from the input stream and push a value onto the stack that represents them.
The next byte must be the character
.BR ~ ,
which is discarded.
.TP
.B #
Push a value representing an empty list onto the stack.
.TP
.B &
Pop a value
.IR v .
Pop another value
.IR l ,
which must represent a list.
Push a value that represents
the list
.IB v :: l .
(Note that the items in a
.I list
therefore appear in reverse order in the input stream.)
.TP
.BI > reg
Pop the top value from the stack and store it in a register labelled by the byte
.IR reg .
.TP
.I reg
Push the value of register
.I reg
(which must be non-null) onto the stack.
.TP
.I tag
Associate the tag string with the value on top of the stack.
The
.IR ubfa (2)
implementation does so by replacing it by a special
.B Tag
tuple.
.TP
.B $
End-of-input: there must be exactly one value on the stack,
which is the result.
.PP
Applications using UBF(A) typically take turns to exchange
.I input
values on a communication channel.
.SH SEE ALSO
.IR ubfa (2),
.IR json (6),
.IR sexprs (6)
.br
J L Armstrong, ``Getting Erlang to talk to the outside world'',
.I "ACM SIGPLAN Erlang workshop 2002" ,
Pittsburg, PA USA
.br
UBF web page,
.B "http://www.sics.se/~joe/ubf/"
|
