| """The yaml blocks of pronunciation-rules.md, as the scorer reads them.""" |
| import re |
|
|
| import yaml |
|
|
| from app.rules import RULES_PATH, Check, check_detectable, load, parse, undetectable |
|
|
|
|
| def test_every_block_parses_into_rules(): |
| r = load() |
| assert {"Q", "K", "TH", "DH", "ZH", "J"} <= set(r.archiphonemes) |
| assert r.dialects == ["urban", "rural"] and len(r.checks) >= 30 |
| assert r.check("ħ", "χ").severity == 3 |
| assert r.margin(["ZH", "ðˤ"], "z") == 0.7 |
|
|
|
|
| def test_every_heard_symbol_has_a_hebrew_label(): |
| text = RULES_PATH.read_text(encoding="utf-8") |
| heard = {str(h) for block in re.findall(r"```yaml\n(.*?)```", text, re.S) |
| for doc in re.split(r"^---$", block, flags=re.M) if doc.strip() |
| for c in (yaml.safe_load(doc) or {}).get("checks", []) for h in c["heard"]} |
| assert heard - set(load().heard_labels) == set() |
|
|
|
|
| def test_correct_realisations_per_letter(): |
| r = load() |
| assert r.members("Q") == ["ʔ", "k", "q"] |
| assert r.members("TH") == ["t", "θ"] and "s" in r.members("TH", written="s") |
| assert r.dialects_of("Q", "k") == {"rural"} and r.dialects_of("K", "k") == {"urban", "rural"} |
|
|
|
|
| def test_undetectable_rules_are_reported(): |
| r = load() |
| everything = lambda p: True |
| assert undetectable(Check("x", "ħ", "ħ", 1, ""), r, everything) == "same_sound" |
| assert undetectable(Check("x", "Q", "k", 1, ""), r, everything) == "accepted" |
| assert undetectable(Check("x", "ħ", "ǀ", 1, ""), r, lambda p: p != "ǀ") == "heard_not_in_model" |
| report = check_detectable(r, everything) |
| assert report["pharyngeal_h"]["detectable"] |
| |
| assert {m["why"] for m in report["vowel_length"]["missing"]} == {"same_sound"} |
| bad = parse("```yaml\nid: bad\ntitle: bad\nchecks:\n - {target: aː, heard: [a], severity: 1, feedback: x}\n```\n") |
| assert not check_detectable(bad, everything)["bad"]["detectable"] |
|
|
