Provided by: libstatistics-r-io-perl_1.0002-2_all 
NAME
Statistics::R::IO::Parser - Functions for parsing R data files
VERSION
version 1.0002
SYNOPSIS
use Statistics::R::IO::ParserState;
use Statistics::R::IO::Parser;
my $state = Statistics::R::IO::ParserState->new(
data => 'file.rds'
);
say $state->at
say $state->next->at;
DESCRIPTION
You shouldn't create instances of this class, it exists mainly to handle deserialization of R data files
by the "IO" classes.
FUNCTIONS
This library is inspired by monadic parser frameworks from the Haskell world, like Packrat
<http://bford.info/packrat/> or Parsec <http://hackage.haskell.org/package/parsec>. What this means is
that parsers are constructed by combining simpler parsers.
The library offers a selection of basic parsers and combinators. Each of these is a function (think of
it as a factory) that returns another function (the actual parser) which receives the current parsing
state (Statistics::R::IO::ParserState) as the argument and returns a two-element array reference (called
for brevity "a pair" in the following text) with the result of the parser in the first element and the
new parser state in the second element. If the parser fails, say if the current state is "a" where a
number is expected, it returns "undef" to signal failure.
The descriptions of individual functions below use a shorthand because the above mechanism is implied.
Thus, when "any_char" is described as "parses any character", it really means that calling "any_char"
will return a function that when called with the current state will return "a pair of the character...",
etc.
CHARACTER PARSERS
any_char
Parses any character, returning a pair of the character at the current State's position and the new
state, advanced by one from the starting state. If the state is at the end ("$state-"eof> is true),
returns undef to signal failure.
char $c
Parses the given character $c, returning a pair of the character at the current State's position if
it is equal to $c and the new state, advanced by one from the starting state. If the state is at the
end ("$state-"eof> is true) or the character at the current position is not $c, returns undef to
signal failure.
string $s
Parses the given string $s, returning a pair of the sequence of characters starting at the current
State's position if it is equal to $s and the new state, advanced by "length($s)" from the starting
state. If the state is at the end ("$state-"eof> is true) or the string starting at the current
position is not $s, returns undef to signal failure.
NUMBER PARSERS
endianness [$end]
When the $end argument is given, this functions sets the byte order used by parsers in the module to
be little- (when $end is "<") or big-endian ($end is ">"). This function changes the module's state
and remains in effect until the next change.
When called with no arguments, "endianness" returns the current byte order in effect. The starting
byte order is big-endian.
any_uint8, any_uint16, any_uint24, any_uint32
Parses an 8-, 16-, 24-, and 32-bit unsigned integer, returning a pair of the integer starting at the
current State's position and the new state, advanced by 1, 2, 3, or 4 bytes from the starting state,
depending on the parser. The integer value is determined by the current value of "endianness". If
there are not enough elements left in the data from the current position, returns undef to signal
failure.
uint8 $n, uint16 $n, uint24 $n, uint32 $n
Parses the specified 8-, 16-, 24-, and 32-bit unsigned integer $n, returning a pair of the integer at
the current State's position if it is equal $n and the new state. The new state is advanced by 1, 2,
3, or 4 bytes from the starting state, depending on the parser. The integer value is determined by
the current value of "endianness". If there are not enough elements left in the data from the current
position or the current position is not $n, returns undef to signal failure.
any_int8, any_int16, any_int24, any_int32
Parses an 8-, 16-, 24-, and 32-bit signed integer, returning a pair of the integer starting at the
current State's position and the new state, advanced by 1, 2, 3, or 4 bytes from the starting state,
depending on the parser. The integer value is determined by the current value of "endianness". If
there are not enough elements left in the data from the current position, returns undef to signal
failure.
int8 $n, int16 $n, int24 $n, int32 $n
Parses the specified 8-, 16-, 24-, and 32-bit signed integer $n, returning a pair of the integer at
the current State's position if it is equal $n and the new state. The new state is advanced by 1, 2,
3, or 4 bytes from the starting state, depending on the parser. The integer value is determined by
the current value of "endianness". If there are not enough elements left in the data from the current
position or the current position is not $n, returns undef to signal failure.
any_real32, any_real64
Parses an 32- or 64-bit real number, returning a pair of the number starting at the current State's
position and the new state, advanced by 4 or 8 bytes from the starting state, depending on the
parser. The real value is determined by the current value of "endianness". If there are not enough
elements left in the data from the current position, returns undef to signal failure.
any_int32_na, any_real64_na
Parses a 32-bit signed integer or 64-bit real number, respectively, but recognizing R-style missing
values (NAs): INT_MIN for integers and a special NaN bit pattern for reals. Returns a pair of the
number value ("undef" if a NA) and the new state, advanced by 4 or 8 bytes from the starting state,
depending on the parser. If there are not enough elements left in the data from the current position,
returns undef to signal failure.
SEQUENCING
seq $p1, ...
This combinator applies parsers $p1, ... in sequence, using the returned parse state of $p1 as the
input parse state to $p2, etc. Returns a pair of the concatenation of all the parsers' results and
the parsing state returned by the final parser. If any of the parsers returns undef, "seq" will
return it immediately without attempting to apply any further parsers.
many_till $p, $end
This combinator applies a parser $p until parser $end succeeds. It does this by alternating
applications of $end and $p; once $end succeeds, the function returns the concatenation of results of
preceding applications of $p. (Thus, if $end succeeds immediately, the 'result' is an empty list.)
Otherwise, $p is applied and must succeed, and the procedure repeats. Returns a pair of the
concatenation of all the $p's results and the parsing state returned by the final parser. If any
applications of $p returns undef, "many_till" will return it immediately.
count $n, $p
This combinator applies the parser $p exactly $n times in sequence, threading the parse state through
each call. Returns a pair of the concatenation of all the parsers' results and the parsing state
returned by the final application. If any application of $p returns undef, "count" will return it
immediately without attempting any more applications.
with_count [$num_p = any_uint32], $p
This combinator first applies parser $num_p to get the number of times that $p should be applied in
sequence. If only one argument is given, "any_uint32" is used as the default value of $num_p. (So
"with_count" works by getting a number $n by applying $num_p and then calling "count $n, $p".)
Returns a pair of the concatenation of all the parsers' results and the parsing state returned by the
final application. If the initial application of $num_p or any application of $p returns undef,
"with_count" will return it immediately without attempting any more applications.
choose $p1, ...
This combinator applies parsers $p1, ... in sequence, until one of them succeeds, when it immediately
returns the parser's result. If all of the parsers fail, "choose" fails and returns undef
COMBINATORS
bind $p1, $f
This combinator applies parser $p1 and, if it succeeds, calls function $f using the first element of
$p1's result as the argument. The call to $f needs to return a parser, which "bind" applies to the
parsing state after $p1's application.
The "bind" combinator is an essential building block for most combinators described so far. For
instance, "with_count" can be written as:
bind($num_p,
sub {
my $n = shift;
count $n, $p;
})
mreturn $value
Returns a parser that when applied returns $value without changing the parsing state.
error $message
Returns a parser that when applied croaks with the $message and the current parsing state.
SINGLETONS
These functions are an interface to ParseState's singleton-related functions, "add_singleton" in
ParseState and "get_singleton" in ParseState. They exist because certain types of objects in R data
files, for instance environments, have to exist as unique instances, and any subsequent objects that
include them refer to them by a "reference id".
add_singleton $singleton
Adds the $singleton to the current parsing state. Returns a pair of $singleton and the new parsing
state.
get_singleton $ref_id
Retrieves from the current parse state the singleton identified by $ref_id, returning a pair of the
singleton and the (unchanged) state.
reserve_singleton $p
Preallocates a space for a singleton before running a given parser, and then assigns the parser's
value to the singleton. Returns a pair of the singleton and the new parse state.
BUGS AND LIMITATIONS
Instances of this class are intended to be immutable. Please do not try to change their value or
attributes.
There are no known bugs in this module. Please see Statistics::R::IO for bug reporting.
SUPPORT
See Statistics::R::IO for support and contact information.
AUTHOR
Davor Cubranic <cubranic@stat.ubc.ca>
COPYRIGHT AND LICENSE
This software is Copyright (c) 2017 by University of British Columbia.
This is free software, licensed under:
The GNU General Public License, Version 3, June 2007