debug.py

"""Step-by-step Split Miner runs against the paper's running example."""

from __future__ import annotations

from collections import Counter

from bpmn_export import save_bpmn_png, save_bpmn_xml, to_bpmn
from bpmn_init import build_initial_bpmn
from concurrency import discover_concurrency
from dfg_builder import build_dfg, discover_loops, preprocess_traces, strip_self_loops
from filtering import filter_pdfg
from joins import discover_joins
from log_loader import PAPER_EXAMPLE_LOG
from or_minimize import minimize_or_joins
from splits import discover_splits


def _summarise(wg) -> str:
    kinds = Counter(n.kind for n in wg.nodes.values())
    return (
        f"nodes={len(wg.nodes)} ({dict(kinds)}), "
        f"edges={len(wg.edges())}"
    )


def step_through_paper_example() -> None:
    print("== Paper running example (10 distinct traces, ×10 each) ==")
    print(f"raw traces: {len(PAPER_EXAMPLE_LOG)}")

    preprocessed = preprocess_traces(PAPER_EXAMPLE_LOG)
    dfg, labels = build_dfg(preprocessed)
    print(f"DFG: {len(dfg)} arcs, {len(labels)} labels")

    loops = discover_loops(dfg, preprocessed)
    print(f"self-loops: {sorted(loops.self_loops)}")
    print(f"short-loops: {[sorted(p) for p in loops.short_loops]}")

    dfg_no_self = strip_self_loops(dfg)
    conc = discover_concurrency(dfg_no_self, loops, eps=0.2)
    print(f"concurrent pairs (eps=0.2): {[sorted(p) for p in conc.concurrent_pairs]}")
    print(f"PDFG: {len(conc.pdfg)} arcs")

    filt = filter_pdfg(conc.pdfg, eta=0)
    print(f"filtered PDFG: {len(filt.edges)} arcs "
          f"(source={filt.source}, sink={filt.sink})")

    wg = build_initial_bpmn(filt, conc.concurrent_pairs, loops.self_loops)
    print(f"after init     : {_summarise(wg)}")

    discover_splits(wg)
    print(f"after splits   : {_summarise(wg)}")

    discover_joins(wg)
    print(f"after joins    : {_summarise(wg)}")

    minimize_or_joins(wg)
    print(f"after OR-min   : {_summarise(wg)}")

    bpmn = to_bpmn(wg)
    save_bpmn_xml(bpmn, "paper_example.bpmn")
    try:
        save_bpmn_png(bpmn, "paper_example.png")
        print("wrote paper_example.bpmn + paper_example.png")
    except Exception as exc:
        print(f"PNG export skipped: {exc}")


if __name__ == "__main__":
    step_through_paper_example()