/** \brief Define VariantParser field
- This macro is a special helper to define a senf::DirectVariantParser type field. This is a
- variant field which chooses the sub-type by directly taking the value of some other field.
-
+ This macro is a special helper to define a senf::VariantParser type field.
\code
struct SomeParser : public PacketParserBase
{
# include SENF_PARSER()
SENF_PARSER_PRIVATE_FIELD( type, senf::UInt8Parser );
- SENF_PARSER_PRIVATE_VARIANT( content, type,
- (senf::VoidPacketParser)
- (senf::UInt8Parser)
- (senf::UInt16Parser)
- (senf::UInt24Parser)
- (senf::UInt32Parser) );
-
- senf::UInt8Parser uint8() const { return content().get<1>(); }
- senf::UInt16Parser uint16() const { return content().get<2>(); }
- senf::UInt24Parser uint24() const { return content().get<3>(); }
- senf::UInt32Parser uint32() const { return content().get<4>(); }
-
- void disable() const { content().init<0>(); }
- void set_uint8() const { content().init<1>(); }
- void set_uint16() const { content().init<2>(); }
- void set_uint24) const { content().init<3>(); }
- void set_uint23() const { content().init<4>(); }
+ SENF_PARSER_VARIANT( content, type,
+ (novalue( disable, senf::VoidPacketParser ))
+ ( id( uint8, senf::UInt8Parser ))
+ ( id( uint16, senf::UInt16Parser ))
+ ( id( uint24, senf::UInt24Parser ))
+ ( id( uint32, senf::UInt32Parser )) );
SENF_PARSER_FINALIZE(SomeParser);
};
- \endcode
-
+ \endcode
+
The variant \c content chooses one of the sub parsers depending on the \c type field. If \c
type is 0, senf::VoidPacketParser is selected, if it is 1, senf::UInt8Parser and so on.
- It is customary, to hide the variant parser (by defining it private) and provide more
- conveniently named accessors.
+ The \a types parameter specifies the types of sub-objects supported by this variant
+ parser. This parameter is a (Boost.Preprocessor style) sequence
+ <pre>(\a type) (\a type) ...</pre>
+ of parser types with additional optional information:
+
+ <table class="senf fixedcolumn">
+ <tr><td>\a type</td><td>Do not provide accessor and use the 0-based type index as
+ key.</td></tr>
+
+ <tr><td>\c id(\a name, \a type)</td><td>Automatically create an accessor \a name, a test
+ function <tt>has_</tt>\a name and a function to switch to this type <tt>init_</tt>\a name
+ for this type. Identical to \c ids(\a name, <tt>has_</tt>\a name, <tt>init_</tt>\a name, \a
+ type)</td></tr>
+
+ <tr><td>\c novalue(\a name, \a type)</td><td>This is like \c id but only provides an
+ initializer called \a name and no accessor. Identical to \c ids(\c na, \c na, \a name, \a
+ type)</td></tr>
+
+ <tr><td>\c ids(\a accessorId, \a testId, \a initId, \a type)</td><td>This is again like \c
+ id but the names of the accessor, test function and init function are given
+ explicitly. Setting any of the id's to \c na will disable that function.</td></tr>
+
+ <tr><td>\c key(\a value, \a type)</td><td>Use \a value to identity this type. The type is
+ selected, when the \a chooser is equal to \a value</td></tr>
+
+ <tr><td>\c id(\a name, \c key(\a value, \a type))<br/>\c novalue(\a name, \c key(\a value,
+ \a type))</td><td>The options may be nested in this way.</td></tr>
+ </table>
+
+ Whenever an id is specified for any type, the variant itself will automatically be made
+ private so the only access to the variant from the outside is via the accessors.
+
+ The functions defined by specifying an id are:
+
+ <table class="senf fixedcolumn">
+ <tr><td><em>name</em><tt>_t</tt></td><td>The type for this specific variant value if not \c
+ novalue.</td></tr>
+
+ <tr><td><em>name</em><tt>_t</tt> <em>name</em>()</td><td>Return the variant value at this id
+ if not \c novalue.</td></tr>
+
+ <tr><td><tt>void</tt> <tt>init_</tt><em>name</em>()</td><td>Set the variant to have a value
+ of this type. If the field is \c novalue, the \c init_ prefix is omitted.</td></tr>
+
+ <tr><td><tt>bool</tt> <tt>has_</tt><em>name</em>()</td><td>Return \c true, if the variant
+ currently holds this kind of value, \c false otherwise. Only if not \c novalue.</td></tr>
+ </table>
+
+ If \c key specs are given, they designate the value to expect in the \a chooser field to
+ select a specific variant type. If the \a chooser value does not match any key, the variant
+ will be initialized to the \e first type.
Further additional tags are supported which modify the way, the \a chooser field is
interpreted:
<table class="senf fixedcolumn">
- <tr><td>\c transform(\a transform, \a size)</td><td>The \a transform is applied to the \a
+ <tr><td>\c transform(\a transform, \a chooser)</td><td>The \a transform is applied to the \a
chooser value, the value is not used directly</td>
</table>
static value_type get(other_type v);
static other_type set(value_type v);
};
- \endcode \c other_type is the \a chooser ::\c value_type where as the \c value_type typedef
+ \endcode \c other_type is the \a chooser ::\c value_type whereas the \c value_type typedef
is the arbitrary return type of the transform.
The tags are applied to the \a chooser parameter:
\code
- SENF_PARSER_VARIANT ( content, transform(MyTransform, type_),
+ SENF_PARSER_VARIANT ( content, transform(MyTransform, type),
(senf::VoidPacketParser)
(senf::UInt8Parser)
(senf::UInt16Parser)
\param[in] types a Boost.Preprocessor style sequence of sub-parser types
\see
- senf::VariantParser \n
+ senf::VariantParser for the VariantParser API\n
\ref SENF_PARSER_PRIVATE_VARIANT()
\hideinitializer
\ingroup packetparsermacros
projects / senf.git / blobdiff