This project defines the grammar of our DSL, implements the parser and
translation to the AST to be used by the filtering methods in aetherus-events.
Note
While we present the DSL focused here on MCRT simulation,
it can filter and parse any encoding that can be fitted for aetherus-events
The Eldritch-Trace DSL is used to specify the criteria we are looking for signals of interest to meet, then walk the Ledger to find all terminal UIDs that meet these conditions, which are then used to extract the signals from the total stream produced by the simulation.
Example of conditions used for filtering:
- emission source
- interactions with specific objects,
- sequence of events
- count repetitions of events matching a pattern
- etc.
The following Eldritch-Trace script can be use to filter events presented in
the Ledger serialised further down.
{
"grps": {},
"src_map": {
"Mat(1)": [ { "Mat": "glass" }],
"Mat(4)": [ { "Mat": "translucent_pla" }],
"Surf(0)": [ { "Surf": "TargetTube" }],
"Surf(1)": [ { "Surf": "TargetToy" }],
"MatSurf(65535)": [ { "MatSurf": "Tank-Water:Tank_material" }],
"MatSurf(65534)": [ { "MatSurf": "Tank:Tank_material" }],
"Mat(0)": [ { "Mat": "air" }],
"MatSurf(65533)": [ { "MatSurf": "Water:Water_material" }],
"Mat(2)": [ { "Mat": "mist" }],
"Mat(3)": [ { "Mat": "seawater" }]
},
"start_events": [
{ "seq_id": 0, "event": "0x1010000" }
],
"next_mat_id": 5,
"next_surf_id": 2,
"next_matsurf_id": 65532,
"next_light_id": 0,
"next": {
"0": { "0x01010000": 1 },
"1": { "0x0300FFFE": 17, "0x0301FFFE": 2 },
"2": { "0x0300FFFF": 211, "0x03010003": 3, "0x03A0FFFE": 2663 },
"3": {
"0x0300FFFD": 21,
"0x0300FFFF": 24,
"0x0301FFFF": 4,
"0x03440000": 14,
"0x03440001": 99,
"0x03A00003": 194,
},
"4": { "0x0300FFFE": 6 },
// ... more entries ...
},
"prev": {
"1": "0, 0x01010000",
"2": "1, 0x0301FFFE",
"3": "2, 0x03010003",
"4": "3, 0x0301FFFF",
"5": "4, 0x03010000",
"6": "4, 0x0300FFFE",
"7": "6, 0x0300FFFF",
// ... more entries ...
}
}
Sequence, patterns and fields can have unary predicates.
| Operator | Description |
|---|---|
! |
Don't match |
? |
Optional, can appear once or none |
* |
Match for any number of times |
+ |
Match for any number of times that find at least one match |
{n,m} |
Match for at least n times and at most m times |
{,m} |
Match for at most m times |
{n,} |
Match for at least n times |
{n} |
Match exactly n times |
The field can only have the "!" operator, to check bits mask non equality. Otherwise, the normal bits match is used.
Patterns can have any of the unary predicates listed above
Sequences can only have an "{n}" predicate that will be unrolled and flatten the sequence.
Note
More advanced features to check for non match with "!" could be added later for more complex checks on the sequence, but otherwise keep it simple.
any: Allow match to any of the members listedsrcpatternseq: Not allowed for now, not sure if necessary
perm: Allow match in any orderpattern
seq: Allow match in the order specifiedpatternseqpermany[pattern]
- Wanted features:
- Predicated sequence, useful for X with condition. I.e.
* seq[! undesired_pattern, X]
- Predicated sequence, useful for X with condition. I.e.
- Benchmarks with bencher for Statistical Continuous
Benchmarking
- Criterion benchmarking of a the translucent plate example
- Inline tests benches using
libtest - Maybe add gungraun later for more informative performance metrics
- Vim syntax highlight
- Tree-sitter parser and syntax highlight
We are looking into how we could visualise the Parser tree, since it can be quite daunting to look at the parser combinator code and figure out what is happening.
This requires first to have the grammar specified in BNF/EBNF, then parse and produce railroad diagram.