libs/spirit/doc/lex/lexer_semantic_actions.qbk
[/==============================================================================
Copyright (C) 2001-2011 Joel de Guzman
Copyright (C) 2001-2011 Hartmut Kaiser
Distributed under the Boost Software License, Version 1.0. (See accompanying
file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
===============================================================================/]
[section:lexer_semantic_actions Lexer Semantic Actions]
The main task of a lexer normally is to recognize tokens in the input.
Traditionally this has been complemented with the possibility to execute
arbitrary code whenever a certain token has been detected. __lex__ has been
designed to support this mode of operation as well. We borrow from the concept
of semantic actions for parsers (__qi__) and generators (__karma__). Lexer
semantic actions may be attached to any token definition. These are C++
functions or function objects that are called whenever a token definition
successfully recognizes a portion of the input. Say you have a token definition
`D`, and a C++ function `f`, you can make the lexer call `f` whenever it matches
an input by attaching `f`:
D[f]
The expression above links `f` to the token definition, `D`. The required
prototype of `f` is:
void f (Iterator& start, Iterator& end, pass_flag& matched, Idtype& id, Context& ctx);
[variablelist where:
[[`Iterator& start`] [This is the iterator pointing to the begin of the
matched range in the underlying input sequence. The
type of the iterator is the same as specified while
defining the type of the `lexertl::actor_lexer<...>`
(its first template parameter). The semantic action
is allowed to change the value of this iterator
influencing, the matched input sequence.]]
[[`Iterator& end`] [This is the iterator pointing to the end of the
matched range in the underlying input sequence. The
type of the iterator is the same as specified while
defining the type of the `lexertl::actor_lexer<...>`
(its first template parameter). The semantic action
is allowed to change the value of this iterator
influencing, the matched input sequence.]]
[[`pass_flag& matched`] [This value is pre/initialized to `pass_normal`.
If the semantic action sets it to `pass_fail` this
behaves as if the token has not been matched in
the first place. If the semantic action sets this
to `pass_ignore` the lexer ignores the current
token and tries to match a next token from the
input.]]
[[`Idtype& id`] [This is the token id of the type Idtype (most of
the time this will be a `std::size_t`) for the
matched token. The semantic action is allowed to
change the value of this token id, influencing the
if of the created token.]]
[[`Context& ctx`] [This is a reference to a lexer specific,
unspecified type, providing the context for the
current lexer state. It can be used to access
different internal data items and is needed for
lexer state control from inside a semantic
action.]]
]
When using a C++ function as the semantic action the following prototypes are
allowed as well:
void f (Iterator& start, Iterator& end, pass_flag& matched, Idtype& id);
void f (Iterator& start, Iterator& end, pass_flag& matched);
void f (Iterator& start, Iterator& end);
void f ();
[important In order to use lexer semantic actions you need to use type
`lexertl::actor_lexer<>` as your lexer class (instead of the
type `lexertl::lexer<>` as described in earlier examples).]
[heading The context of a lexer semantic action]
The last parameter passed to any lexer semantic action is a reference to an
unspecified type (see the `Context` type in the table above). This type is
unspecified because it depends on the token type returned by the lexer. It is
implemented in the internals of the iterator type exposed by the lexer.
Nevertheless, any context type is expected to expose a couple of
functions allowing to influence the behavior of the lexer. The following table
gives an overview and a short description of the available functionality.
[table Functions exposed by any context passed to a lexer semantic action
[[Name] [Description]]
[[`Iterator const& get_eoi() const`]
[The function `get_eoi()` may be used by to access the end iterator of
the input stream the lexer has been initialized with]]
[[`void more()`]
[The function `more()` tells the lexer that the next time it matches a
rule, the corresponding token should be appended onto the current token
value rather than replacing it.]]
[[`Iterator const& less(Iterator const& it, int n)`]
[The function `less()` returns an iterator positioned to the nth input
character beyond the current token start iterator (i.e. by passing the
return value to the parameter `end` it is possible to return all but the
first n characters of the current token back to the input stream.]]
[[`bool lookahead(std::size_t id)`]
[The function `lookahead()` can be used to implement lookahead for lexer
engines not supporting constructs like flex' `a/b`
(match `a`, but only when followed by `b`). It invokes the lexer on the
input following the current token without actually moving forward in the
input stream. The function returns whether the lexer was able to match a
token with the given token-id `id`.]]
[[`std::size_t get_state() const` and `void set_state(std::size_t state)`]
[The functions `get_state()` and `set_state()` may be used to introspect
and change the current lexer state.]]
[[`token_value_type get_value() const` and `void set_value(Value const&)`]
[The functions `get_value()` and `set_value()` may be used to introspect
and change the current token value.]]
]
[heading Lexer Semantic Actions Using Phoenix]
Even if it is possible to write your own function object implementations (i.e.
using Boost.Lambda or Boost.Bind), the preferred way of defining lexer semantic
actions is to use __phoenix__. In this case you can access the parameters
described above by using the predefined __spirit__ placeholders:
[table Predefined Phoenix placeholders for lexer semantic actions
[[Placeholder] [Description]]
[[`_start`]
[Refers to the iterator pointing to the beginning of the matched input
sequence. Any modifications to this iterator value will be reflected in
the generated token.]]
[[`_end`]
[Refers to the iterator pointing past the end of the matched input
sequence. Any modifications to this iterator value will be reflected in
the generated token.]]
[[`_pass`]
[References the value signaling the outcome of the semantic action. This
is pre-initialized to `lex::pass_flags::pass_normal`. If this is set to
`lex::pass_flags::pass_fail`, the lexer will behave as if no token has
been matched, if is set to `lex::pass_flags::pass_ignore`, the lexer will
ignore the current match and proceed trying to match tokens from the
input.]]
[[`_tokenid`]
[Refers to the token id of the token to be generated. Any modifications
to this value will be reflected in the generated token.]]
[[`_val`]
[Refers to the value the next token will be initialized from. Any
modifications to this value will be reflected in the generated token.]]
[[`_state`]
[Refers to the lexer state the input has been match in. Any modifications
to this value will be reflected in the lexer itself (the next match will
start in the new state). The currently generated token is not affected
by changes to this variable.]]
[[`_eoi`]
[References the end iterator of the overall lexer input. This value
cannot be changed.]]
]
The context object passed as the last parameter to any lexer semantic action is
not directly accessible while using __phoenix__ expressions. We rather provide
predefined Phoenix functions allowing to invoke the different support functions
as mentioned above. The following table lists the available support functions
and describes their functionality:
[table Support functions usable from Phoenix expressions inside lexer semantic actions
[[Plain function] [Phoenix function] [Description]]
[[`ctx.more()`]
[`more()`]
[The function `more()` tells the lexer that the next time it matches a
rule, the corresponding token should be appended onto the current token
value rather than replacing it.]]
[[`ctx.less()`]
[`less(n)`]
[The function `less()` takes a single integer parameter `n` and returns an
iterator positioned to the nth input character beyond the current token
start iterator (i.e. by assigning the return value to the placeholder
`_end` it is possible to return all but the first `n` characters of the
current token back to the input stream.]]
[[`ctx.lookahead()`]
[`lookahead(std::size_t)` or `lookahead(token_def)`]
[The function `lookahead()` takes a single parameter specifying the token
to match in the input. The function can be used for instance to implement
lookahead for lexer engines not supporting constructs like flex' `a/b`
(match `a`, but only when followed by `b`). It invokes the lexer on the
input following the current token without actually moving forward in the
input stream. The function returns whether the lexer was able to match
the specified token.]]
]
[endsect]
