|
|
Simple records have the form
name ht value nl
where ht and nl are the ASCII tab and newline characters, respectively. For example, the G2++ record
id Bob
is said to be a record "of type ``id''"; it carries a single field named ``id'' with the value ``Bob''.
Hierarchically-structured values may be created using additional tabs and newlines to represent the hierarchy:
person id Bob age 11
Here, a record of type ``person'' contains two fields, named ``id'' and ``age'', along with their associated values.
Array-valued fields are also supported. The following example illustrates the syntax for arrays:
person id Bob age 11 hobbies 0 video games 1 soccer 2 baseball
Here, the field named ``hobbies'' is an array with three elements, each consisting of an index and a value. Indices must be non-negative integers.
Record structure is recursive, allowing nesting to any level. For example, array values may have their own "inner structure."
person id Bob age 11 hobbies 0 video games 1 soccer 2 baseball friends 0 id Fred age 10 1 id Jane age 12
Field names (like ``person'' and ``age'') are formed according to the rules for C identifiers. Elementary values (like ``Bob'' and ``soccer'') are formed from the alphabet defined by the ASCII isprint(3C) function (note in particular that elementary values may contain blanks, but not tabs or newlines). Tabs are used for indentation (one per level) and also to separate names from elementary values.
Comments may be appended to any line of a G2++ record except for the final newline, which must stand alone (see the formal syntax definition below). Comments must appear at the end of the line and must be separated from the rightmost field by one or more tabs. By convention, comments begin with a sharp (#).
Here is the complete syntax definition for G2++ records (alternatives are listed on separate lines):
record => group nl group => name value name value nl indent group value => ht string nl nl indent group nl indent array array => index value index value nl indent array
where:
ht is an ASCII tab character (011)
nl is an ASCII newline character (012)
name is any valid C identifier
index is any non-negative integer
string is any sequence of printable ASCII characters, as defined by the ASCII isprint(3C) function.
indent is one or more tabs, the exact number depending on the depth of recursion in the definition.
The highest level name in a G2++ record is also known as the record's type.
Indices do not have to occur in any particular order within an array.
The G2++ data language is identical to the G2 data language.
.g files, which are files that define G2++ record types (see g2++comp(1C++)), obey the syntax of G2++ records given here, but impose additional constraints on form of elementary values.