grammar::me::cpu::core -
ME virtual machine state manipulation
package require Tcl 8.4
package require grammar::me::cpu::core ? 0.2 ?
::grammar::me::cpu::core disasm asm
::grammar::me::cpu::core asm asm
::grammar::me::cpu::core new asm
::grammar::me::cpu::core lc state location
::grammar::me::cpu::core tok state ? from ? to ? ?
::grammar::me::cpu::core pc state
::grammar::me::cpu::core iseof state
::grammar::me::cpu::core at state
::grammar::me::cpu::core cc state
::grammar::me::cpu::core sv state
::grammar::me::cpu::core ok state
::grammar::me::cpu::core error state
::grammar::me::cpu::core lstk state
::grammar::me::cpu::core astk state
::grammar::me::cpu::core mstk state
::grammar::me::cpu::core estk state
::grammar::me::cpu::core rstk state
::grammar::me::cpu::core nc state
::grammar::me::cpu::core ast state
::grammar::me::cpu::core halted state
::grammar::me::cpu::core code state
::grammar::me::cpu::core eof statevar
::grammar::me::cpu::core put statevar tok lex line col
::grammar::me::cpu::core run statevar ? n ?
This package provides an implementation of the ME virtual machine.
Please go and read the document grammar::me_intro first if
you do not know what a ME virtual machine is.
This implementation represents each ME virtual machine as a Tcl value
and provides commands to manipulate and query such values to show the
effects of executing instructions, adding tokens, retrieving state,
etc.
The values fully follow the paradigm of Tcl that every value is a
string and while also allowing C implementations for a proper
Tcl_ObjType to keep all the important data in native data structures.
Because of the latter it is recommended to access the state values
only through the commands of this package to ensure that
internal representation is not shimmered away.
The actual structure used by all state values is described in section
CPU STATE.
The package directly provides only a single command, and all the
functionality is made available through its methods.
-
::grammar::me::cpu::core disasm asm
-
This method returns a list containing a disassembly of the match
instructions in asm. The format of asm is specified in the
section MATCH PROGRAM REPRESENTATION.
Each element of the result contains instruction label, instruction
name, and the instruction arguments, in this order. The label can be
the empty string. Jump destinations are shown as labels, strings and
tokens unencoded. Token names are prefixed with their numeric id, if,
and only if a tokmap is defined. The two components are separated by a
colon.
-
::grammar::me::cpu::core asm asm
-
This method returns code in the format as specified in section
MATCH PROGRAM REPRESENTATION generated from ME assembly
code asm, which is in the format as returned by the method
disasm.
-
::grammar::me::cpu::core new asm
-
This method creates state value for a ME virtual machine in its
initial state and returns it as its result.
The argument matchcode contains a Tcl representation of the
match instructions the machine has to execute while parsing the input
stream. Its format is specified in the section
MATCH PROGRAM REPRESENTATION.
The tokmap argument taken by the implementation provided by the
package grammar::me::tcl is here hidden inside of the match
instructions and therefore not needed.
-
::grammar::me::cpu::core lc state location
-
This method takes the state value of a ME virtual machine and uses it
to convert a location in the input stream (as offset) into a line
number and column index. The result of the method is a 2-element list
containing the two pieces in the order mentioned in the previous
sentence.
Note that the method cannot convert locations which the machine
has not yet read from the input stream. In other words, if the machine
has read 7 characters so far it is possible to convert the offsets
0 to 6, but nothing beyond that. This also shows that
it is not possible to convert offsets which refer to locations before
the beginning of the stream.
This utility allows higher levels to convert the location offsets
found in the error status and the AST into more human readable data.
-
::grammar::me::cpu::core tok state ? from ? to ? ?
-
This method takes the state value of a ME virtual machine and returns
a Tcl list containing the part of the input stream between the
locations from and to (both inclusive). If to is not
specified it will default to the value of from. If from is
not specified either the whole input stream is returned.
This method places the same restrictions on its location arguments as
the method lc.
-
::grammar::me::cpu::core pc state
-
This method takes the state value of a ME virtual machine and returns
the current value of the stored program counter.
-
::grammar::me::cpu::core iseof state
-
This method takes the state value of a ME virtual machine and returns
the current value of the stored eof flag.
-
::grammar::me::cpu::core at state
-
This method takes the state value of a ME virtual machine and returns
the current location in the input stream.
-
::grammar::me::cpu::core cc state
-
This method takes the state value of a ME virtual machine and returns
the current token.
-
::grammar::me::cpu::core sv state
-
This method takes the state value of a ME virtual machine and returns
the current semantic value stored in it.
This is an abstract syntax tree as specified in the document
grammar::me_ast, section AST VALUES.
-
::grammar::me::cpu::core ok state
-
This method takes the state value of a ME virtual machine and returns
the match status stored in it.
-
::grammar::me::cpu::core error state
-
This method takes the state value of a ME virtual machine and returns
the current error status stored in it.
-
::grammar::me::cpu::core lstk state
-
This method takes the state value of a ME virtual machine and returns
the location stack.
-
::grammar::me::cpu::core astk state
-
This method takes the state value of a ME virtual machine and returns
the AST stack.
-
::grammar::me::cpu::core mstk state
-
This method takes the state value of a ME virtual machine and returns
the AST marker stack.
-
::grammar::me::cpu::core estk state
-
This method takes the state value of a ME virtual machine and returns
the error stack.
-
::grammar::me::cpu::core rstk state
-
This method takes the state value of a ME virtual machine and returns
the subroutine return stack.
-
::grammar::me::cpu::core nc state
-
This method takes the state value of a ME virtual machine and returns
the nonterminal match cache as a dictionary.
-
::grammar::me::cpu::core ast state
-
This method takes the state value of a ME virtual machine and returns
the abstract syntax tree currently at the top of the AST stack stored
in it.
This is an abstract syntax tree as specified in the document
grammar::me_ast, section AST VALUES.
-
::grammar::me::cpu::core halted state
-
This method takes the state value of a ME virtual machine and returns
the current halt status stored in it, i.e. if the machine has stopped
or not.
-
::grammar::me::cpu::core code state
-
This method takes the state value of a ME virtual machine and returns
the code stored in it, i.e. the instructions executed by the machine.
-
::grammar::me::cpu::core eof statevar
-
This method takes the state value of a ME virtual machine as stored in
the variable named by statevar and modifies it so that the eof
flag inside is set. This signals to the machine that whatever token
are in the input queue are the last to be processed. There will be no
more.
-
::grammar::me::cpu::core put statevar tok lex line col
-
This method takes the state value of a ME virtual machine as stored in
the variable named by statevar and modifies it so that the token
tok is added to the end of the input queue, with associated
lexeme data lex and line/column information.
The operation will fail with an error if the eof flag of the machine
has been set through the method eof.
-
::grammar::me::cpu::core run statevar ? n ?
-
This method takes the state value of a ME virtual machine as stored in
the variable named by statevar, executes a number of
instructions and stores the state resulting from their modifications
back into the variable.
The execution loop will run until either
- n instructions have been executed, or
- a halt instruction was executed, or
-
the input queue is empty and the code is asking for more tokens to
process.
If no limit n was set only the last two conditions are checked
for.
A match program is represented by nested Tcl list. The first element,
asm, is a list of integer numbers, the instructions to execute,
and their arguments. The second element,
pool, is a list of
strings, referenced by the instructions, for error messages, token
names, etc. The third element,
tokmap, provides ordering
information for the tokens, mapping their names to their numerical
rank. This element can be empty, forcing lexicographic comparison when
matching ranges.
All ME instructions are encoded as integer numbers, with the mapping
given below. A number of the instructions, those which handle error
messages, have been given an additional argument to supply that
message explicitly instead of having it constructed from token names,
etc. This allows the machine state to store only the message ids
instead of the full strings.
Jump destination arguments are absolute indices into the asm
element, refering to the instruction to jump to. Any string arguments
are absolute indices into the pool element. Tokens, characters,
messages, and token (actually character) classes to match are coded as
references into the pool as well.
- "ict_advance message"
- "ict_match_token tok message"
- "ict_match_tokrange tokbegin tokend message"
- "ict_match_tokclass code message"
- "inc_restore branchlabel nt"
- "inc_save nt"
- "icf_ntcall branchlabel"
- "icf_ntreturn"
- "iok_ok"
- "iok_fail"
- "iok_negate"
- "icf_jalways branchlabel"
- "icf_jok branchlabel"
- "icf_jfail branchlabel"
- "icf_halt"
- "icl_push"
- "icl_rewind"
- "icl_pop"
- "ier_push"
- "ier_clear"
- "ier_nonterminal message"
- "ier_merge"
- "isv_clear"
- "isv_terminal"
- "isv_nonterminal_leaf nt"
- "isv_nonterminal_range nt"
- "isv_nonterminal_reduce nt"
- "ias_push"
- "ias_mark"
- "ias_mrewind"
- "ias_mpop"
A state value is a list containing the following elements, in the order listed below:
- code: Match instructions, see MATCH PROGRAM REPRESENTATION.
- pc: Program counter, int.
- halt: Halt flag, boolean.
- eof: Eof flag, boolean
- tc: Terminal cache, and input queue. Structure see below.
- cl: Current location, int.
- ct: Current token, string.
- ok: Match status, boolean.
- sv: Semantic value, list.
- er: Error status, list.
- ls: Location stack, list.
- as: AST stack, list.
- ms: AST marker stack, list.
- es: Error stack, list.
- rs: Return stack, list.
- nc: Nonterminal cache, dictionary.
tc, the input queue of tokens waiting for processing and the
terminal cache containing the tokens already processing are one
unified data structure simply holding all tokens and their
information, with the current location separating that which has been
processed from that which is waiting.
Each element of the queue/cache is a list containing the token, its
lexeme information, line number, and column index, in this order.
All stacks have their top element aat the end, i.e. pushing an item is
equivalent to appending to the list representing the stack, and
popping it removes the last element.
er, the error status is either empty or a list of two elements,
a location in the input, and a list of messages, encoded as references
into the pool element of the code.
nc, the nonterminal cache is keyed by nonterminal name and
location, each value a four-element list containing current location,
match status, semantic value, and error status, in this order.
This document, and the package it describes, will undoubtedly contain
bugs and other problems.
Please report such in the category
grammar_me of the
http://sourceforge.net/tracker/?group_id=12883.
Please also report any ideas for enhancements you may have for either
package and/or documentation.
virtual machine, parsing, grammar