parsing primitive

PASM syntax

comment

define a parsing rule

 = xxx

followed by parsing rules ; parsing rules consist of the PASM syntax shown below

define a parsing predicate

 - <xxx>

followed by parsing rules ; must return ^ok or ^fail

accept a character

'c’

lookahead one character

?’c’

accept a token

KIND

Token kind (class) in upper case

accept a token with given string as its text

KIND/str

Token kind in upper case, token text verbatim (case matters) as str

accept any token

.

lookahead one token

?KIND

lookahead one token with given string as its text

?KIND/str

call a parsing rule

@<xxx>

call a parsing predicate

&<xxx>

call an external mechanism (function)

<xxx>

cycle

{

exit cycle

>

return success from a parsing predicate

>> ^ok

return failure from a parsing predicate

>> ^fail

return from a parsing rule (no value)

>>  

begin choice

[

followed by a parsing predicate then parsing rules ; the first choice which returns ^ok is evaluated

choice alternate

|

end choice

]

always succeed

*

usually used as ‘otherwise’ in a choice

call filter

~<xxx>

filter operation - call rule <xxx> for every input token, if ‘accepted-token’ and ‘next-token’ are the same as before the filter operation, accept once (thereby advancing the stream), else save the newly accepted token onto a list of new tokens ==> i.e. filter the token stream, using a rule that creates one accepted token at a time, throw away tokens that were not accepted (see example of rmspaces in dsl0.pasm https://github.com/bmfbp/bmfbp/blob/v2emitter/build_process/esa-transpiler/dsl0.pasm

(which removes ( / filters out) whitespace tokens))

N.B. parser has one character in lookahead position and one in accepted position

N.B. see https://github.com/guitarvydas/parsing-assembler for an implementation (the name means “assembly language for parsing”)

see

https://en.wikipedia.org/wiki/S/SL_programming_language

source code https://research.cs.queensu.ca/home/cordy/pub/downloads/ssl/