fix: prevent duplicate party assignment in assign_implicit_party

[zyt600]

For example, in smtp-telnet.fan:

<start> ::= <Out:telnet_intro> <smtp>
<smtp> ::= <Out:m220> <In:quit> <Out:m221>

When a non-terminal symbol (e.g., <smtp>) in a grammar rule lacks explicit party annotations, the assign_implicit_party function automatically assigns the default StdOut party to it. However, this can cause nested party issues when the rule definition already contains party definitions in its child nodes.

The <smtp> non-terminal itself has no party annotation, but its children (<Out:m220>, <In:quit>, <Out:m221>) already have explicit party definitions. Without this fix, <smtp> would be assigned StdOut, creating an illegal nested party structure.

Solution

Added a check in assign_implicit_party to skip assigning an implicit party when the rule definition of a non-terminal already contains party definitions. This prevents overwriting existing party assignments and avoids nested party errors.

[github-actions]

Bencher Report

Branch	fix/assign-implicit-party
Testbed	ubuntu-latest

Click to view all benchmark results

Benchmark	Latency	Benchmark Result milliseconds (ms) (Result Δ%)	Upper Boundary milliseconds (ms) (Limit %)
tests/test_benchmarks.py::test_generate_with_single_hard_constraint	📈 view plot 🚷 view threshold	1,801.65 ms (-3.12%) Baseline: 1,859.71 ms	2,139.78 ms (84.20%)
tests/test_benchmarks.py::test_generate_with_single_soft_constraint	📈 view plot 🚷 view threshold	51,153.52 ms (-3.33%) Baseline: 52,914.16 ms	79,972.04 ms (63.96%)
tests/test_benchmarks.py::test_init_fandango	📈 view plot 🚷 view threshold	119.15 ms (-21.40%) Baseline: 151.58 ms	161.89 ms (73.60%)
tests/test_benchmarks.py::test_parse_spec	📈 view plot 🚷 view threshold	119.85 ms (-21.76%) Baseline: 153.17 ms	160.36 ms (74.74%)

🐰 View full continuous benchmarking report in Bencher