lpsactionrename

The lpsactionrename tool renames actions in an LPS, based on their names and on the data parameters they carry. The tool can be used in two ways: either by supplying it with a rename file or by providing a regular expression. A rename file is provided using the option --renamefile and a regular expression can be specified with the option --regex. Both modes are explained below.

Structure of rename files

ActionRenameRuleRHS ActionRenameRule ActionRenameRuleSpec ActionRenameSpec

ActionRenameRuleRHS  ::=  Action |
                          'tau' |
                          'delta'
ActionRenameRule     ::=  ( DataExpr '->'  ) ? Action '=>' ActionRenameRuleRHS ';'
ActionRenameRuleSpec ::=  VarSpec ? 'rename' ActionRenameRule +
ActionRenameSpec     ::=  ( SortSpec | ConsSpec | MapSpec | EqnSpec | ActSpec | ActionRenameRuleSpec ) +

The format of the RENAMEFILE can contain sort, cons, map, eqn and act sections as in a mcrl2 file. This is followed by a rename section to define the rename rules. The sections sort, cons, map, eqn and act are meant for new declarations that will be added to the LPS and can be used in the rename rules. The new declarations are not allowed to contain any conflicts with the declarations of the LPS. The rename section can be preceded by a var section, where variables can be declared for the rename rules.

The rename rules have the format: rename c -> a1 => a2; where c is a boolean expression that has to hold to rename an occurrence of a1 into a2. The condition can be left out, in which case it is interpreted as true (i.e., all occurrences of a1 will be renamed). The action a1 can contain arguments that can either be uniquely occurring variables or closed terms. The arguments of a2 can be arbitrary terms, but the variables occurring in it must also occur in a1. The condition is an expression of sort Bool and can also only use variables that also occur in a1.

It is possible use tau for a2; note that this means that a multi-action of the form a1|b will be replaced by b. Instead of an action, a2 may also be delta. In this case, the action and the following process call are replaced by delta.

The renaming rules are applied from top to bottom to an linear process equation. If no value for the variables in a rename rule can be found to match an action, the next rule is applied. If no rule applies the action is left untouched. Variables in different rename rules with the same variable names are independent when being matched.

After the LPS has been renamed, sum elimination and rewriting will be applied to simplify the result. This can be skipped using appropriate switches.

Rename rule conditions

Upon loading the rename file, lpsactionrename will check if the following conditions hold:

  • Variables used in the condition or in the right side of a rename rule must also occur in the left side of that rename rule.
  • All arguments of the action at the left hand side must either be closed terms or variables. Each variable can only occur once in the left hand side.
  • All used actions and data types must be declared in the LPS file or locally.
  • All conditions are data expressions of sort Bool.
  • All elements are well typed with respect to the declarations in the LPS or the rename file.

Example

Consider an LPS with the process specification:

P(x:Bool) = sum y:Nat. (y < 6) -> a(x,y). P(!x);

and a rename file with the following rename rules:

act b: Bool;
var v: Nat; w:Bool;
rename
  w -> a(w,v) => b(v==5);
  (v==v*2)==w -> a(w,v) => tau;
  a(w,5) => delta;

The arguments of an action do not have to consist of a single variable, as is done in the second rename rule. In the second rename rule, a(w,2*v), w and 2*v will be respectively equal to x and y from the LPS action a(x,y).

The result of applying the rename rules to the LPS without sum elimination will give:

proc P(x_P0: Bool) =
       true ->
         delta
     + sum w: Bool,v,y_P0: Nat.
         ((y_P0 < 6 && w==x_P0 && v==y_P0) && w) -> b(v==5).P(!x_P0);
     + sum w00: Bool,v00: Nat,w: Bool,v,y_P0: Nat.
         ((((y_P0 < 6 && w==x_P0 && v==y_P0) && !w) && w00==x_P0 && v00==y_P0) &&
                   (v00==v00*2)==w00) -> tau.P(!x_P0)
     + sum w01,w00: Bool,v00: Nat,w: Bool,v,y_P0: Nat.
         ((((((y_P0 < 6 && w==x_P0 && v==y_P0) && !w) && w00==x_P0 && v00==y_P0) &&
                   !((v00==v00*2)==w00)) && w01==x_P0) && 5==y_P0) -> delta
     + sum w01,w00: Bool,v00: Nat,w: Bool,v,y_P0: Nat.
         ((((((y_P0 < 6 && w==x_P0 && v==y_P0) && !w) && w00==x_P0 && v00==y_P0) &&
                   !((v00==v00*2)==w00)) && w01==x_P0) && !(5==y_P0)) -> a(x_P0, y_P0).P(!x_P0)

Most of the introduced sum variables have a single point domain, namely: u, w, w_S00, w_s01, v_S00 and in the last two summands, y. These variables can be eliminated by applying sum elimination. For example: in the first summand w is equal to x. Therefore w can be substituted by x, and w can then be removed from the sum since it is no longer used.

Applying sum elimination will give the following result:

proc P(x_P0: Bool) =
     true -> delta
     + sum y_P0: Nat.(y_P0 < 6 && x_P0) ->b(y_P0 == 5) .P(!x_P0);
     + sum y_P0: Nat.(y_P0 < 6 && !(y_P0 == y_P0 * 2)) ->tau.P(!(y_P0 == y_P0 * 2))
     + (!x_P0 && x_P0) ->delta
     + sum y_P0: Nat.(((y_P0 < 6 && !x_P0) && !((y_P0 == y_P0 * 2) == x_P0)) &&
                   !(5 == y_P0)) -> a(x_P0, y_P0) .P(!x_P0)

Regular Expressions

A regular expression has to be provided in the shape matching pattern/replacement. The replacement pattern follows the standard of ECMAScript. Groups matched with parentheses can be substituted in the replacement string using $n, where n is the index of the matched group. See the ECMAScript website for more details.

Examples

We consider the following process:

proc P(s1: Pos) =
       (s1 == 3) ->
         a_out|c_out .
         P(s1 = 2)
     + (s1 == 2) ->
         b_out .
         P(s1 = 1)
     + (s1 == 1) ->
         c_out .
         P(s1 = 4)
     + (s1 == 4) ->
         delta;

We can remove the prefix of a_out and c_out using the regular expression ^([^b])_out$/$1. It is generally a good idea to write regular expressions in the shape ^expression$ to ensure the whole action name is matched.

It is also possible to rename actions to delta or to tau. For example, when renaming a_out to delta using ^a_out$/delta, the multi action a_out|c_out will become delta.

Manual page for lpsactionrename

Usage

lpsactionrename   [OPTION]... (--renamefile=NAME | --regex=EXPR) [INFILE [OUTFILE]]

Description

Apply the action rename specification in FILE to the LPS in INFILE and save it to OUTFILE. If OUTFILE is not present, stdout is used. If INFILE is not present, stdin is used.

Command line options

-o , --no-rewrite

do not rewrite data expressions while renaming; use when the rewrite system does not terminate

-m , --no-sumelm

do not apply sum elimination to the final result

-t , --no-typecheck

do not typecheck the resulting specfication

-QNUM , --qlimitNUM

limit enumeration of quantifiers to NUM variables. (Default NUM=1000, NUM=0 for unlimited).

-eEXPR , --regexEXPR

use the provided regular expression to rename action labels. Argument should be of the shape ‘matching pattern/replacement’. Matched groups can be substituted in the result with $n, where n is the index of the group. It is generally good to surround the matching expression with ^$. Example: ‘^(.*)_send$/$1_receive’

-fNAME , --renamefileNAME

use the rename rules from NAME

-rNAME , --rewriterNAME

use rewrite strategy NAME:

jitty

jitty rewriting

jittyc

compiled jitty rewriting

jittyp

jitty rewriting with prover

--timings[FILE]

append timing measurements to FILE. Measurements are written to standard error if no FILE is provided

Standard options

-q , --quiet

do not display warning messages

-v , --verbose

display short intermediate messages

-d , --debug

display detailed intermediate messages

--log-levelLEVEL

display intermediate messages up to and including level

-h , --help

display help information

--version

display version information

Author

Jan Friso Groote and Tom Haenen