STT-Assistant/tests/test_triage.py

"""Tests for the triage analyzer (todos 3, 9, 10, 11). Pure, hardware-free."""

import json
import logging
from pathlib import Path
from types import SimpleNamespace

import pytest

from stt_daemon import events
from stt_daemon.assistant import triage
from stt_daemon.assistant.service import AssistantService
from stt_daemon.assistant.triage import FactLedger, TriageAnalyzer, load_questions
from stt_daemon.config import Config
from stt_daemon.tui.state import ViewState

EXAMPLE_PATH = str(Path(__file__).resolve().parent.parent / "questions.example.json")


def _cfg(path=EXAMPLE_PATH, **overrides):
    base = {
        "questions_file": path,
        "state_char_budget": 1200,
        "confidence_field": "answer_confidence",
        "escalate_threshold": 0.6,
        "choice_min_confidence": 0.85,
    }
    base.update(overrides)
    return SimpleNamespace(**base)


def _write_questions(tmp_path, questions):
    path = tmp_path / "questions.json"
    path.write_text(json.dumps(questions), encoding="utf-8")
    return str(path)


def _analyzer(tmp_path, questions=None, **overrides):
    if questions is None:
        return TriageAnalyzer(_cfg(**overrides))
    return TriageAnalyzer(_cfg(_write_questions(tmp_path, questions), **overrides))


# --- todo 3: load_questions -------------------------------------------------


def test_load_example_order_preserved():
    raw = json.loads(Path(EXAMPLE_PATH).read_text(encoding="utf-8"))
    questions = load_questions(EXAMPLE_PATH)
    assert questions is not None
    assert questions
    assert list(questions.keys()) == list(raw.keys())
    for definition in questions.values():
        assert definition["type"] in {"choice", "score", "noul"}
        assert isinstance(definition["instructions"], str)
        assert definition["instructions"].strip()


def test_example_questions_match_design():
    questions = load_questions(EXAMPLE_PATH)
    assert questions is not None
    assert list(questions.keys()) == [
        "substantive_update",
        "opportunity",
        "blocker",
        "decision_status",
        "action_item",
        "responsibility_clarity",
        "deadline",
        "open_question",
    ]
    for qid, definition in questions.items():
        assert isinstance(definition["instructions"], str)
        assert definition["instructions"].strip()
        if definition["type"] == "choice":
            criteria = definition["criteria"]
            options = list(criteria.keys()) if isinstance(criteria, dict) else list(criteria)
            assert "other" in options
            assert len(options) <= 7
        if definition["type"] == "noul":
            assert isinstance(definition.get("instructions"), str)
            assert definition["instructions"].strip()


def test_load_missing_file_returns_none_with_warning(caplog):
    with caplog.at_level(logging.WARNING, logger="stt_daemon.assistant.triage"):
        assert load_questions("/nonexistent/questions.json") is None
    assert caplog.records, "expected a warning for the missing file"


def test_load_bad_json_returns_none_with_warning(tmp_path, caplog):
    path = tmp_path / "questions.json"
    path.write_text("{not valid json", encoding="utf-8")
    with caplog.at_level(logging.WARNING, logger="stt_daemon.assistant.triage"):
        assert load_questions(str(path)) is None
    assert caplog.records


@pytest.mark.parametrize(
    "bad_entry",
    [
        {"type": "mystery", "instructions": "do something"},  # unknown type
        {"type": "score", "instructions": "rate it", "criteria": {"0": "low"}},  # score dict
        {  # noul criteria with a bad key
            "type": "noul",
            "instructions": "holds?",
            "criteria": {"true": "yes", "maybe": "hmm"},
        },
        "just a string",  # non-object entry
        {"type": "choice", "instructions": "pick one", "criteria": []},  # empty choice
        {"type": "choice", "instructions": "pick one", "criteria": {}},  # empty choice dict
        {"type": "choice", "criteria": ["a", "b"]},  # missing instructions
        {"type": "score", "instructions": "rate it", "criteria": ["a", None]},  # None level
        {"type": "noul", "instructions": "holds?", "criteria": ["true"]},  # noul list
    ],
)
def test_load_malformed_returns_none_with_warning(tmp_path, caplog, bad_entry):
    path = _write_questions(
        tmp_path, {"good": {"type": "noul", "instructions": "ok?"}, "bad": bad_entry}
    )
    with caplog.at_level(logging.WARNING, logger="stt_daemon.assistant.triage"):
        assert load_questions(str(path)) is None
    assert caplog.records


# --- todo 9: normalize ------------------------------------------------------


def _score_questions():
    return {
        "severity": {
            "type": "score",
            "instructions": "How severe is the issue?",
            "criteria": ["none", "low", "high", "critical"],
        },
        "category": {
            "type": "choice",
            "instructions": "What is this about?",
            "criteria": {"support": "help needed", "sales": "buying"},
        },
        "churn_risk": {"type": "noul", "instructions": "User may leave."},
    }


def test_normalize_choice_exact(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "model": "multilingual",
        "answers": {
            "category": {
                "type": "choice",
                "choice": "support",
                "confidence": 0.7,
                "answer_confidence": 0.4,
            }
        },
        "usage": {"input_tokens": 123, "output_tokens": 0},
    }
    result = analyzer.normalize(payload)
    assert result["order"] == ["category"]
    assert result["items"]["category"] == {
        "kind": "choice",
        "choice": "support",
        "confidence": 0.7,
        "answer_confidence": 0.4,
        "gate": 0.4,
        "flagged": True,
    }
    assert result["input_tokens"] == 123
    assert result["escalated"] is True


def test_normalize_score_with_legend(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "answers": {
            "severity": {
                "type": "score",
                "score": 2.6,
                "legend": {"0": "none", "1": "low", "2": "high", "3": "critical"},
                "confidence": 0.5,
                "answer_confidence": 0.6,
            }
        },
        "usage": {"input_tokens": 7, "output_tokens": 0},
    }
    item = analyzer.normalize(payload)["items"]["severity"]
    assert item["kind"] == "score"
    assert item["expected"] == 2.6
    assert item["level"] == 3
    assert item["label"] == "critical"
    assert item["confidence"] == 0.5
    assert item["answer_confidence"] == 0.6
    assert item["gate"] == 0.6
    assert item["flagged"] is True


def test_normalize_score_legend_fallback(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "answers": {
            "severity": {"type": "score", "score": 1.2, "confidence": 0.9, "answer_confidence": 0.9}
        }
    }
    item = analyzer.normalize(payload)["items"]["severity"]
    assert item["level"] == 1
    assert item["label"] == "low"  # falls back to criteria[level]
    assert item["flagged"] is False  # 1.2 < (4 - 1) * 0.5


def test_normalize_noul_and_bare_payload(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    bare = {
        "churn_risk": {"type": "noul", "noul": 0.9, "confidence": 0.9, "answer_confidence": 0.85}
    }
    result = analyzer.normalize(bare)
    assert result["order"] == ["churn_risk"]
    assert result["items"]["churn_risk"] == {
        "kind": "noul",
        "p": 0.9,
        "confidence": 0.9,
        "answer_confidence": 0.85,
        "gate": 0.85,
        "flagged": True,
    }
    assert result["input_tokens"] == 0


def test_normalize_skips_unknown_qid_and_malformed(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "answers": {
            "ghost": {"type": "noul", "noul": 0.99, "confidence": 0.9, "answer_confidence": 0.9},
            "category": {"type": "choice", "choice": 42},  # malformed: no confidences
            "churn_risk": "nonsense",  # malformed: not a dict
        },
        "usage": {"input_tokens": 3, "output_tokens": 0},
    }
    result = analyzer.normalize(payload)
    assert result["order"] == []
    assert result["items"] == {}
    assert result["escalated"] is False


def test_normalize_ignores_side_channel_fields(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "answers": {
            "category": {
                "type": "choice",
                "choice": "sales",
                "confidence": 0.95,
                "answer_confidence": 0.9,
                "action": {"act" + "_probability": 0.99},
            }
        },
        "usage": {"input_tokens": 1, "output_tokens": 0},
    }
    result = analyzer.normalize(payload)
    assert result["items"]["category"]["flagged"] is False
    assert "act" + "_probability" not in json.dumps(result)
    source = Path(triage.__file__).read_text(encoding="utf-8")
    assert "act" + "_probability" not in source


def test_normalize_gating_field_selectable(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions(), confidence_field="confidence")
    payload = {
        "answers": {
            "category": {
                "type": "choice",
                "choice": "sales",
                "confidence": 0.5,
                "answer_confidence": 0.99,
            }
        }
    }
    item = analyzer.normalize(payload)["items"]["category"]
    assert item["gate"] == 0.5  # gates on confidence, keeps both
    assert item["answer_confidence"] == 0.99
    assert item["flagged"] is True


# --- todo 9: condensation ---------------------------------------------------


def test_condense_keeps_newest_chronological_no_numerals(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions(), state_char_budget=40)
    window = ["alpha first", "beta second", "gamma third newest here"]
    state = analyzer.build_state(window)
    assert "gamma third newest here" in state
    assert "alpha first" not in state  # oldest dropped under the budget
    lines = state.splitlines()
    assert all(line.startswith("- ") for line in lines)
    assert not any(ch.isdigit() for ch in state)  # no numerals injected
    # chronological restore: kept lines stay in original order
    kept = [u for u in window if u in state]
    assert [line[2:] for line in lines] == kept


def test_condense_empty_window(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.build_state([]) == ""


# --- todo 10: escalation + extraction ---------------------------------------


def _signals(choice_gate=0.9, noul_p=0.1, score_expected=0.2):
    return {
        "order": ["category", "churn_risk", "severity"],
        "items": {
            "category": {
                "kind": "choice",
                "choice": "sales",
                "confidence": choice_gate,
                "answer_confidence": choice_gate,
                "gate": choice_gate,
                "flagged": choice_gate < 0.85,
            },
            "churn_risk": {
                "kind": "noul",
                "p": noul_p,
                "confidence": noul_p,
                "answer_confidence": noul_p,
                "gate": noul_p,
                "flagged": noul_p >= 0.6,
            },
            "severity": {
                "kind": "score",
                "expected": score_expected,
                "level": 0,
                "label": "none",
                "confidence": 0.9,
                "answer_confidence": 0.9,
                "gate": 0.9,
                "flagged": score_expected >= 1.5,
            },
        },
        "escalated": False,
        "input_tokens": 0,
    }


def test_escalation_arms_fire_independently(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.should_escalate(_signals(choice_gate=0.5)) is True  # choice arm
    assert analyzer.should_escalate(_signals(noul_p=0.8)) is True  # noul arm
    assert analyzer.should_escalate(_signals(score_expected=2.0)) is True  # score arm
    assert analyzer.should_escalate(_signals()) is False


def test_normalize_escalated_matches_should_escalate(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    payload = {
        "answers": {
            "category": {
                "type": "choice",
                "choice": "sales",
                "confidence": 0.9,
                "answer_confidence": 0.9,
            },
            "churn_risk": {
                "type": "noul",
                "noul": 0.1,
                "confidence": 0.9,
                "answer_confidence": 0.9,
            },
            "severity": {
                "type": "score",
                "score": 0.1,
                "confidence": 0.9,
                "answer_confidence": 0.9,
            },
        }
    }
    result = analyzer.normalize(payload)
    assert result["escalated"] == analyzer.should_escalate(result) is False


# --- todo 5: brief prompt / parse / gate / finalize --------------------------


def test_build_brief_prompt_order_and_focus(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_brief_prompt(
        ["Alice: let's ship Friday", "Bob: agreed"], _signals(noul_p=0.8)
    )
    assert prompt.startswith("New utterances:\n- Alice: let's ship Friday\n- Bob: agreed")
    assert "Focus:" in prompt
    assert "churn_risk" in prompt  # signal named in words
    assert prompt.index("New utterances:") < prompt.index("Focus:")
    assert prompt.index("Focus:") < prompt.index("Return only the updated brief")
    # no JSON extraction keys or raw dict shapes leak into the brief prompt
    for key in ("summary", "opportunities", "action_items", "risks", "doubts"):
        assert key not in prompt
    assert "{" not in prompt and "}" not in prompt


def test_build_brief_prompt_no_focus_without_signals(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    for signals in (None, {}, {"order": [], "items": {}}):
        prompt = analyzer.build_brief_prompt(["solo utterance"], signals)
        assert "solo utterance" in prompt
        assert "Focus:" not in prompt
        assert prompt.endswith("Return only the updated brief as plain text.")


def test_build_ledger_prompt_inlines_ledger_narrative_and_deltas(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_ledger_prompt(
        "Summary: shipped\n\nOpportunities: expand",
        ["fact one", "fact two"],
        ["Alice: let's ship Friday"],
        _signals(noul_p=0.8),
    )
    assert "- fact one" in prompt and "- fact two" in prompt
    assert "CURRENT LEDGER" in prompt
    assert "Summary: shipped" in prompt
    assert "NEW UTTERANCES:\n- Alice: let's ship Friday" in prompt
    assert prompt.index("CURRENT LEDGER") < prompt.index("CURRENT BRIEF:")
    assert prompt.index("CURRENT BRIEF:") < prompt.index("NEW UTTERANCES:")
    assert prompt.index("NEW UTTERANCES:") < prompt.index("Focus:")
    assert "churn_risk" in prompt
    assert "NEW FACTS:" in prompt and "BRIEF:" in prompt


def test_build_ledger_prompt_empty_ledger_and_narrative(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_ledger_prompt("", [], ["solo"], None)
    assert "CURRENT LEDGER" in prompt and "(none yet)" in prompt
    assert "CURRENT BRIEF:\n(none yet)" in prompt
    assert "NEW UTTERANCES:\n- solo" in prompt
    assert "Focus:" not in prompt


def test_extract_bullets_strips_new_marker_and_normalizes(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    brief = (
        "Summary: things\n"
        "- top level bullet\n"
        "  - indented child\n"
        "- **New:** tagged fact\n"
        "not a bullet\n"
        "\n"
    )
    assert analyzer.extract_bullets(brief) == [
        "top level bullet",
        "indented child",
        "tagged fact",
    ]


def test_restore_missing_bullets_reappends_dropped_fact(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = (
        "Summary: old\n\nPrior Details:\n"
        "- The budget was $305 million.\n"
        "- Sense could find prey.\n"
    )
    new = "Summary: newer\n\nPrior Details:\n- Sense could find prey.\n"

    restored = analyzer.restore_missing_bullets(new, prior)

    assert "- Sense could find prey." in restored
    assert "- The budget was $305 million." in restored
    assert restored.count("- Sense could find prey.") == 1


def test_restore_missing_bullets_matches_reworded_bullet(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = "Prior Details:\n- **New:** Sense could find prey.\n"
    new = "Prior Details:\n- Sense could find prey.\n"

    restored = analyzer.restore_missing_bullets(new, prior)

    assert restored == new


def test_restore_missing_bullets_restores_reworded_as_new_fact(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = "Prior Details:\n- The budget was $305 million.\n"
    new = "Prior Details:\n- Budget: $305 million (nearly Spider-Man 2).\n"

    restored = analyzer.restore_missing_bullets(new, prior)

    assert restored.count("$305 million") == 2
    assert restored.rstrip().endswith("- The budget was $305 million.")


def test_restore_missing_bullets_noop_without_prior_or_new(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    new = "Summary: only\nPrior Details:\n- keep me\n"

    assert analyzer.restore_missing_bullets(new, "") == new
    assert analyzer.restore_missing_bullets(new, None) == new
    assert analyzer.restore_missing_bullets("", new) == ""
    assert analyzer.restore_missing_bullets(None, new) is None


def test_restore_missing_bullets_appends_section_when_absent(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = "- orphan fact\n"
    new = "Summary: no prior details header\n"

    restored = analyzer.restore_missing_bullets(new, prior)

    assert "- orphan fact" in restored
    assert "Prior Details:" in restored


def test_build_ledger_prompt_contains_resolution_rule(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_ledger_prompt("Summary: x", [], ["hi"], None)
    assert "Never silently delete Action Items, Open Questions" in prompt
    assert "Done: <original>" in prompt
    assert "NEW FACTS bullet" in prompt


def test_restore_missing_bullets_reappends_dropped_action_item(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = (
        "Summary: s\n\nAction Items:\n- Call Maria about pricing\n\n"
        "Open Questions:\n- Who owns the launch checklist?\n"
    )
    new = "Summary: s\n\nDecisions:\n- nothing yet\n"
    restored = analyzer.restore_missing_bullets(new, prior)
    assert "- Call Maria about pricing" in restored
    assert "- Who owns the launch checklist?" in restored
    assert "Prior Details:" in restored


def test_parse_brief_strips_fences_and_rejects_empty(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_brief("```\nhello brief\n```") == "hello brief"
    assert analyzer.parse_brief("```markdown\nhello brief\n```") == "hello brief"
    assert analyzer.parse_brief("plain brief") == "plain brief"
    assert analyzer.parse_brief("  padded  ") == "padded"
    assert analyzer.parse_brief("") is None
    assert analyzer.parse_brief("   ") is None
    assert analyzer.parse_brief(None) is None
    assert analyzer.parse_brief(42) is None
    assert analyzer.parse_brief("```") is None


def test_parse_brief_unescapes_literal_backslash_n(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_brief("## Title\\n\\nBody line") == "## Title\n\nBody line"
    assert analyzer.parse_brief("a\\tb") == "a\tb"
    assert analyzer.parse_brief("line1\nline2\\nline3") == "line1\nline2\\nline3"


def test_parse_brief_never_truncates_accepted_text(tmp_path):
    """The accepted brief is never cut: a long reply is returned whole."""
    analyzer = _analyzer(tmp_path, _score_questions())
    parsed = analyzer.parse_brief("x" * 5000)
    assert parsed == "x" * 5000
    wrapped = "[update_brief(updated_brief='" + "x" * 5000 + "')]"
    assert analyzer.parse_brief(wrapped) == "x" * 5000


def test_build_ledger_prompt_clips_current_brief_on_line_boundary(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions(), brief_max_chars=40)
    narrative = "Summary: first line\nOpportunities:\n- keep this out"
    prompt = analyzer.build_ledger_prompt(narrative, [], ["hi"], None)
    assert "Summary: first line" in prompt
    assert "- keep this out" not in prompt
    assert "The CURRENT BRIEF was trimmed" in prompt


def test_build_ledger_prompt_trims_long_paragraph_at_a_word(tmp_path):
    """A single long Summary paragraph must keep most of the budget, not 8 chars."""
    analyzer = _analyzer(tmp_path, _score_questions(), brief_max_chars=400)
    narrative = "Summary:\n" + ("word " * 200).strip()
    prompt = analyzer.build_ledger_prompt(narrative, [], ["hi"], None)
    view = prompt.split("CURRENT BRIEF:\n", 1)[1].split("\n\nNEW UTTERANCES:", 1)[0]
    assert len(view) > 200
    assert not view.endswith("wor")


def test_build_ledger_prompt_keeps_short_brief_verbatim(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_ledger_prompt("Summary: x", [], ["hi"], None)
    assert "CURRENT BRIEF:\nSummary: x" in prompt
    assert "was trimmed" not in prompt


def test_parse_ledger_reply_keeps_narrative_past_the_brief_cap(tmp_path):
    """Regression: the cap used to slice the reply before the BRIEF split."""
    analyzer = _analyzer(tmp_path, _score_questions(), brief_max_chars=50)
    narrative = "Summary: " + ("word " * 400).strip() + "\nOpen Questions:\n- who owns it?"
    reply = "NEW FACTS:\n- durable fact\n\nBRIEF:\n" + narrative
    parsed = analyzer.parse_ledger_reply(reply)
    assert parsed is not None
    got_narrative, facts = parsed
    assert facts == ["durable fact"]
    assert got_narrative.endswith("- who owns it?")
    assert len(got_narrative) > 50


def test_restore_missing_bullets_caps_prior_details_block(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions(), brief_max_chars=80)
    prior = "Summary: s\n\nPrior Details:\n" + "\n".join(f"- bullet {i}" for i in range(20))
    restored = analyzer.restore_missing_bullets("Summary: s2\n", prior)
    tail = restored.split("Prior Details:", 1)[1]
    assert len(tail) <= 80
    assert "- bullet 19" in tail
    assert "- bullet 0" not in tail


def test_is_regression_flags_lost_section_only(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prior = "Summary: a\n\nOpen Questions:\n- who owns it?\n"
    assert analyzer.is_regression(prior, "Summary: b\n") is True
    assert analyzer.is_regression(prior, "Summary: b\n\nOpen Questions:\n- who?\n") is False
    assert analyzer.is_regression("", "anything") is False
    assert analyzer.is_regression(None, "x") is False


def test_build_ledger_prompt_lists_every_canonical_section(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    prompt = analyzer.build_ledger_prompt("Summary: x", [], ["hi"], None)
    for heading in triage.BRIEF_HEADINGS:
        assert heading in prompt


def test_parse_brief_unwraps_single_quoted_tool_call_with_apostrophes(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    observed = "<|tool_call_start|>[update_brief(updated_brief='the company's game failed')]"
    assert analyzer.parse_brief(observed) == "the company's game failed"


def test_parse_brief_unwraps_double_quoted_tool_call(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = '[update_brief(updated_brief="the company\'s plan")]'
    assert analyzer.parse_brief(wrapped) == "the company's plan"


def test_parse_brief_unwraps_trailing_end_marker(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = "[update_brief(updated_brief='ship friday')]<|tool_call_end|>"
    assert analyzer.parse_brief(wrapped) == "ship friday"


def test_parse_brief_unwraps_fenced_and_wrapped(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = "```\n<|tool_call_start|>[update_brief(updated_brief='fenced payload')]\n```"
    assert analyzer.parse_brief(wrapped) == "fenced payload"


def test_parse_brief_does_not_mangle_clean_markdown(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    clean = "Here's the [brief]: it's the company's plan - don't mangle [it]."
    assert analyzer.parse_brief(clean) == clean
    prose = "We should update_brief the notes, here's [why]."
    assert analyzer.parse_brief(prose) == prose


def test_parse_brief_empty_tool_call_payload_is_none(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_brief("[update_brief(updated_brief='')]") is None


def test_parse_brief_unwraps_multi_arg_tool_call(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = (
        "[update_brief(updated_brief='## Opportunities\n- ship Friday', "
        "previous_brief='## old\n- y', mode='read')]"
    )
    assert analyzer.parse_brief(wrapped) == "## Opportunities\n- ship Friday"


def test_parse_brief_unwraps_multi_arg_with_markers(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = (
        "<|tool_call_start|>[update_brief(updated_brief='## Opportunities\n- ship Friday', "
        "previous_brief='## old\n- y', mode='read')]<|tool_call_end|>"
    )
    assert analyzer.parse_brief(wrapped) == "## Opportunities\n- ship Friday"


def test_parse_brief_unwraps_multi_arg_apostrophes_preserved(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    wrapped = (
        "[update_brief(updated_brief='the company's game failed', "
        "previous_brief='## old', mode='read')]"
    )
    assert analyzer.parse_brief(wrapped) == "the company's game failed"


def test_parse_brief_unwraps_positional_tool_call(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_brief("[update_brief('payload here', mode='read')]") == "payload here"


def test_parse_brief_rejects_polluted_tool_call(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    polluted = "[update_brief(updated_brief='oops, previous_brief=', mode='read')]"
    assert analyzer.parse_brief(polluted) is None
    no_arg = "<|tool_call_start|>[update_brief(previous_brief=old, mode=read)]"
    assert analyzer.parse_brief(no_arg) is None


def test_parse_brief_clean_mode_text_not_mangled(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    clean = "Set mode=read, then read [the brief] - it's fine (really)."
    assert analyzer.parse_brief(clean) == clean


def test_parse_brief_rejects_marker_with_artifacts_but_no_call(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_brief("<|tool_call_start|>previous_brief='x', mode='read'") is None
    assert analyzer.parse_brief("<|tool_call_start|>mode='read'") is None


def _brief_signals(p=None):
    item = {"kind": "noul"}
    if p is not None:
        item["p"] = p
    return {"order": ["substantive_update"], "items": {"substantive_update": item}}


def test_should_update_brief_gate(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.should_update_brief(_brief_signals(0.9)) is True
    assert analyzer.should_update_brief(_brief_signals(0.5)) is True  # boundary
    assert analyzer.should_update_brief(_brief_signals(0.2)) is False
    # absent update / no numeric p / not a dict -> fail open
    assert analyzer.should_update_brief({"order": [], "items": {}}) is True
    assert analyzer.should_update_brief(_brief_signals(None)) is True
    assert analyzer.should_update_brief(None) is True
    assert analyzer.should_update_brief({"items": {"substantive_update": "nope"}}) is True


def test_should_update_brief_threshold_configured(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions(), brief_gate_threshold=0.8)
    assert analyzer.should_update_brief(_brief_signals(0.6)) is False
    assert analyzer.should_update_brief(_brief_signals(0.8)) is True


def test_finalize_brief_length_rejects_cut_reply(tmp_path, caplog):
    analyzer = _analyzer(tmp_path, _score_questions())
    with caplog.at_level(logging.WARNING, logger="stt_daemon.assistant.triage"):
        assert analyzer.finalize_brief("y" * 5000, "length") is None
    assert caplog.records


def test_finalize_brief_stop_and_empty(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.finalize_brief("a brief", "stop") == "a brief"
    assert analyzer.finalize_brief("", "stop") is None
    assert analyzer.finalize_brief("```\nfenced\n```", "stop") == "fenced"
    assert analyzer.finalize_brief(None, "length") is None


# --- todo 11: cached state() ------------------------------------------------


def test_state_initial_loading_with_good_questions(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.state() == ("loading", "")


def test_state_initial_no_questions_with_bad_path(tmp_path):
    missing = str(tmp_path / "missing.json")
    analyzer = TriageAnalyzer(_cfg(missing))
    state, reason = analyzer.state()
    assert state == "no_questions"
    assert missing in reason


def test_state_vocabulary_has_no_extra_member():
    assert set(triage.TRIAGE_REASONS) == {
        "ok",
        "no_questions",
        "loading",
        "laya_missing",
        "disabled",
        "error",
    }
    assert "service" + "_down" not in triage.TRIAGE_REASONS


def test_refresh_transitions(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    analyzer.refresh("ok")
    assert analyzer.state() == ("ok", "")
    analyzer.refresh("laya_missing", "run: uv sync --extra triage")
    assert analyzer.state() == ("laya_missing", "run: uv sync --extra triage")
    analyzer.refresh("error", "boom")
    assert analyzer.state() == ("error", "boom")


def test_refresh_unknown_state_ignored(tmp_path, caplog):
    analyzer = _analyzer(tmp_path, _score_questions())
    analyzer.refresh("ok")
    with caplog.at_level(logging.WARNING, logger="stt_daemon.assistant.triage"):
        analyzer.refresh("bogus_state")
    assert analyzer.state() == ("ok", "")
    assert caplog.records


def test_no_questions_state_cannot_be_overridden(tmp_path):
    analyzer = TriageAnalyzer(_cfg(str(tmp_path / "missing.json")))
    assert analyzer.state()[0] == "no_questions"
    analyzer.refresh("ok")
    analyzer.refresh("error", "boom")
    assert analyzer.state()[0] == "no_questions"


def test_state_does_not_touch_lane(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())

    class StrictLane:
        def __init__(self):
            self.calls = []

        def __getattr__(self, name):
            def record(*args, **kwargs):
                self.calls.append(name)

            return record

    lane = StrictLane()
    for _ in range(3):
        analyzer.state()
    analyzer.refresh("ok")
    analyzer.state()
    assert lane.calls == []


# --- F3: end-to-end triage flow on fakes ------------------------------------

EXTRACT_JSON = (
    '{"summary":"s","opportunities":[],"doubts":[],"action_items":[],"risks":[]}'
)


class _Clock:
    """Injected clock: no sleeps, fully deterministic."""

    def __init__(self, start=1000.0):
        self.t = float(start)

    def __call__(self):
        return self.t

    def advance(self, dt):
        self.t += float(dt)


class _FakeSubscriber:
    def events(self):
        return iter(())

    def close(self):
        pass


class _FakeLane:
    """Duck-typed LayaLane: canned-ready, records submits, counts drops."""

    def __init__(self):
        self.submits: list[tuple] = []
        self.drops = 0

    def ready(self) -> bool:
        return True

    def state(self):
        return ("ok", "")

    def submit(self, state, questions, lang, generation):
        self.submits.append((state, questions, lang, generation))
        return True

    def drop_pending(self):
        self.drops += 1
        return 1


def _answers(noul_p=0.1):
    """A full laya-shaped answers payload for the example questions.

    `noul_p` drives every noul arm, including the `substantive_update` gate:
    below `brief_gate_threshold` (0.5) the turn is gated OUT (no brief job),
    at or above it the turn gates IN.
    """
    answers = {
        "decision_status": {
            "type": "choice",
            "choice": "made",
            "confidence": 0.9,
            "answer_confidence": 0.9,
        },
        "responsibility_clarity": {
            "type": "choice",
            "choice": "explicit",
            "confidence": 0.9,
            "answer_confidence": 0.9,
        },
    }
    for qid in (
        "substantive_update",
        "opportunity",
        "blocker",
        "action_item",
        "deadline",
        "open_question",
    ):
        answers[qid] = {
            "type": "noul",
            "noul": noul_p,
            "confidence": 0.9,
            "answer_confidence": 0.9,
        }
    return {"model": "multilingual", "answers": answers, "usage": {"input_tokens": 42}}


def test_end_to_end_triage_flow():
    """signals -> content -> cleared across service, analyzer, lane, and ViewState.

    The lane result is wired by calling `service._on_triage_result(gen, payload)`
    directly (the runner does exactly this in production); the brief job is
    drained with `job.run()` so no worker thread is started. The
    single-outstanding guard plus the shared `_pending_brief` collapse settled
    turns into ONE job consuming every buffered delta, and the
    `substantive_update` gate can reduce that to ZERO jobs.
    """
    cfg = Config()
    cfg.assistant.turn_gap_s = 0.5
    cfg.triage.enabled = True
    cfg.triage.questions_file = EXAMPLE_PATH
    clock = _Clock(1000.0)
    lane = _FakeLane()
    emitted: list[dict] = []
    prompts: list[str] = []

    def emit(type_, **fields):
        emitted.append({"type": type_, **fields})

    analyzer = TriageAnalyzer(cfg.triage)
    def brief_fn(prompt, on_io=None, finalize=None):
        prompts.append(prompt)
        return "NEW FACTS:\nnone\n\nBRIEF:\nSummary: brief\n"

    service = AssistantService(
        cfg,
        _FakeSubscriber(),
        emit=emit,
        time_fn=clock,
        chat_fn=lambda text: "reply",
        triage=analyzer,
        laya_lane=lane,
        brief_fn=brief_fn,
    )

    def triage_events():
        return [e for e in emitted if e["type"] == "triage"]

    def brief_jobs():
        return [j for j in service._inference._queue if j.name == "brief-update"]

    # 1. settled turn -> exactly one lane submit, lang + generation recorded
    assert service.toggle_chat() is True
    service.handle_event(
        {"type": events.TRANSCRIPT, "text": "hola, necesito ayuda", "lang": "es"}
    )
    clock.advance(1.0)
    service.tick()
    assert len(lane.submits) == 1
    state_text, questions, lang, gen1 = lane.submits[0]
    assert "hola, necesito ayuda" in state_text
    assert questions is analyzer.questions
    assert lang == "es"
    assert gen1 == 0

    # 2. low-p turn -> signals only: the gate (p=0.1 < 0.5) dispatches ZERO jobs
    service._on_triage_result(gen1, _answers(noul_p=0.1))
    assert [e["stage"] for e in triage_events()] == ["signals"]
    assert triage_events()[0]["triage"]["escalated"] is False
    assert service.last_triage is not None
    assert service.last_triage["escalated"] is False
    assert brief_jobs() == []
    assert prompts == []

    # 3. high-p turn -> second submit, then ONE brief job (the only one outstanding)
    service.handle_event(
        {"type": events.TRANSCRIPT, "text": "quiero cancelar mi suscripcion", "lang": "es"}
    )
    clock.advance(1.0)
    service.tick()
    assert len(lane.submits) == 2
    gen2 = lane.submits[1][3]
    service._on_triage_result(gen2, _answers(noul_p=0.9))
    assert [e["stage"] for e in triage_events()] == ["signals", "signals"]
    assert triage_events()[1]["triage"]["escalated"] is True
    jobs = brief_jobs()
    assert len(jobs) == 1

    # 4. run the brief job directly (deterministic, no worker thread): it consumes
    # both buffered deltas and emits ONE content with the brief merged over signals
    jobs[0].run()
    assert len(prompts) == 1
    assert "hola, necesito ayuda" in prompts[0]
    assert "quiero cancelar mi suscripcion" in prompts[0]
    content_events = [e for e in triage_events() if e["stage"] == "content"]
    assert len(content_events) == 1
    merged = content_events[0]["triage"]
    assert merged["brief"] == "Summary: brief"
    assert merged["escalated"] is True  # signals preserved under the brief
    assert merged["order"] == triage_events()[1]["triage"]["order"]

    # 5. drive the emitted stream through a fresh ViewState
    view = ViewState()
    for event in emitted:
        view.ingest(event)
    analyzer.refresh("ok")  # the lane callback does this in production
    view.ingest({"type": "status", **service.status()})
    assert view.triage is not None
    assert view.triage["brief"] == "Summary: brief"
    lines = view.triage_lines()
    assert lines

    # 6. clear -> cleared event empties the panel and bumps the epoch
    baseline_epoch = view.triage_epoch
    already = len(emitted)
    service.clear()
    for event in emitted[already:]:
        view.ingest(event)
    assert view.triage is None
    assert view.triage_epoch > baseline_epoch

    # a stale result after clear must not re-populate the panel
    service._on_triage_result(gen2, _answers(noul_p=0.9))
    assert [e["stage"] for e in triage_events()] == [
        "signals",
        "signals",
        "content",
        "cleared",
    ]
    assert view.triage is None
    assert lane.drops == 1

    # 7. disabled control: no triage/lane -> status disabled, clear emits nothing
    cfg2 = Config()
    clock2 = _Clock(2000.0)
    emitted2: list[dict] = []

    def emit2(type_, **fields):
        emitted2.append({"type": type_, **fields})

    plain = AssistantService(cfg2, _FakeSubscriber(), emit=emit2, time_fn=clock2)
    assert plain.status()["triage_state"] == "disabled"
    assert plain.clear() is True
    assert [e for e in emitted2 if e["type"] == "triage"] == []


def test_ledger_merge_dedupes_exact_and_punctuation_variants():
    ledger = FactLedger()
    assert ledger.merge(["The budget was $305 million.", "Sense could find prey."]) == 2
    assert ledger.merge(["sense could find prey", "The budget was $305 million"]) == 0
    assert ledger.facts() == [
        "The budget was $305 million.",
        "Sense could find prey.",
    ]


def test_ledger_merge_keeps_near_match_with_different_number():
    ledger = FactLedger()
    ledger.merge(["Sold 1.9 million units."])
    assert ledger.merge(["Sold 10 million units."]) == 1
    assert len(ledger) == 2


def test_ledger_merge_keeps_near_match_with_different_entity():
    ledger = FactLedger()
    ledger.merge(["Marcus Smith directed the game."])
    assert ledger.merge(["Mike Daly directed the game."]) == 1
    assert len(ledger) == 2


def test_ledger_merge_ignores_empty_and_non_string():
    ledger = FactLedger()
    assert ledger.merge(["", "   ", None, 123]) == 0
    assert ledger.render() == ""


def test_ledger_render_has_prior_details_section():
    ledger = FactLedger()
    ledger.merge(["fact one", "fact two"])
    assert ledger.render() == "Prior Details:\n- fact one\n- fact two"


def test_ledger_evicts_oldest_over_cap_and_never_new():
    ledger = FactLedger(max_facts=2)
    ledger.merge(["alpha", "beta"])
    ledger.merge(["gamma"])
    assert ledger.facts() == ["beta", "gamma"]


def test_ledger_truncates_overlong_fact():
    ledger = FactLedger(fact_max_chars=10)
    ledger.merge(["x" * 50])
    assert ledger.facts() == ["x" * 10]


def test_ledger_clear_empties():
    ledger = FactLedger()
    ledger.merge(["fact"])
    ledger.clear()
    assert ledger.facts() == []
    assert ledger.render() == ""


def test_ledger_from_brief_harvests_bullets():
    ledger = FactLedger.from_brief("Summary: s\n\nPrior Details:\n- one\n- two\n")
    assert ledger.facts() == ["one", "two"]


def test_parse_ledger_reply_splits_narrative_and_facts(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    narrative, facts = analyzer.parse_ledger_reply(
        "NEW FACTS:\n- budget is $305m\n- sense finds prey\n\nBRIEF:\nSummary: ok"
    )
    assert facts == ["budget is $305m", "sense finds prey"]
    assert narrative == "Summary: ok"


def test_parse_ledger_reply_drops_trailing_fact_when_truncated(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    _, facts = analyzer.parse_ledger_reply(
        "NEW FACTS:\n- complete fact\n- half a fac", truncated=True
    )
    assert facts == ["complete fact"]


def test_parse_ledger_reply_missing_facts_block(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    narrative, facts = analyzer.parse_ledger_reply("BRIEF:\nSummary: only narrative")
    assert facts == []
    assert narrative == "Summary: only narrative"


def test_parse_ledger_reply_strips_model_prior_details(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    narrative, _ = analyzer.parse_ledger_reply(
        "NEW FACTS:\nnone\n\nBRIEF:\nSummary: s\n\nPrior Details:\n- model echo\n"
    )
    assert "model echo" not in narrative
    assert "Summary: s" in narrative


def test_parse_ledger_reply_none_on_unusable(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    assert analyzer.parse_ledger_reply("") is None
    assert analyzer.parse_ledger_reply("just some prose with no sections") is None


def test_parse_facts_ignores_prose_and_handles_markers(tmp_path):
    analyzer = _analyzer(tmp_path, _score_questions())
    block = "- bullet fact\n* star fact\n• dot fact\n1. numbered fact\nplain prose\nnone"
    assert analyzer.parse_facts(block) == [
        "bullet fact",
        "star fact",
        "dot fact",
        "numbered fact",
    ]


def _rich_questions():
    return {
        "substantive_update": {
            "type": "noul",
            "instructions": "This recent conversation segment adds something a running brief "
            "should record.",
        },
        "blocker": {
            "type": "noul",
            "instructions": "An obstacle, dependency, objection, or risk that could block "
            "or delay the work was raised.",
        },
        "churn_risk": {"type": "noul", "instructions": "User may leave."},
        "category": {
            "type": "choice",
            "instructions": "What is this about?",
            "criteria": {"support": "help needed", "sales": "buying"},
        },
        "severity": {
            "type": "score",
            "instructions": "How severe is the issue?",
            "criteria": ["none", "low", "high", "critical"],
        },
    }


def _rich_signals(**overrides):
    items = {
        "blocker": {
            "kind": "noul",
            "p": 0.91,
            "confidence": 0.9,
            "answer_confidence": 0.9,
            "gate": 0.9,
            "flagged": True,
        },
        "churn_risk": {
            "kind": "noul",
            "p": 0.10,
            "confidence": 0.9,
            "answer_confidence": 0.9,
            "gate": 0.9,
            "flagged": False,
        },
        "substantive_update": {
            "kind": "noul",
            "p": 0.10,
            "confidence": 0.9,
            "answer_confidence": 0.9,
            "gate": 0.9,
            "flagged": False,
        },
        "category": {
            "kind": "choice",
            "choice": "support",
            "confidence": 0.4,
            "answer_confidence": 0.4,
            "gate": 0.4,
            "flagged": True,
        },
        "severity": {
            "kind": "score",
            "expected": 2.6,
            "level": 3,
            "label": "critical",
            "confidence": 0.9,
            "answer_confidence": 0.9,
            "gate": 0.9,
            "flagged": True,
        },
    }
    items.update(overrides.get("items", {}))
    order = overrides.get(
        "order", ["blocker", "churn_risk", "substantive_update", "category", "severity"])
    return {"order": order, "items": items, "escalated": overrides.get("escalated", False),
            "input_tokens": 0}


def test_signal_lines_filters_low_noul_but_keeps_gate_qid(tmp_path):
    analyzer = _analyzer(tmp_path, _rich_questions())
    lines = analyzer._signal_lines(_rich_signals())
    text = "\n".join(lines)
    assert not any("churn_risk" in line for line in lines)  # p=0.10 filtered
    assert any("substantive_update" in line for line in lines)  # gate qid kept
    assert any("blocker" in line for line in lines)  # high-signal kept
    assert "p=0.91" in text and "flagged" in text
    assert "could block or delay" in text  # instructions carried
    assert all(line.startswith("- ") for line in lines)
    # original order preserved among kept lines
    kept = [line for line in lines]
    assert [i for i, line in enumerate(kept) if "blocker" in line][0] < [
        i for i, line in enumerate(kept) if "category" in line][0]


def test_signal_lines_choice_carries_instructions_and_flag(tmp_path):
    analyzer = _analyzer(tmp_path, _rich_questions())
    lines = analyzer._signal_lines(_rich_signals(order=["category"], escalated=False))
    assert len(lines) == 1
    assert "category (choice): selected support (conf 0.40, flagged as uncertain)" in lines[0]
    assert "What is this about?" in lines[0]
    assert "help needed" in lines[0]  # criteria[selected] definition


def test_signal_lines_escalated_marker_only_on_flagged(tmp_path):
    analyzer = _analyzer(tmp_path, _rich_questions())
    lines = analyzer._signal_lines(_rich_signals(
        order=["blocker", "substantive_update"], escalated=True))
    by_qid = {line.split(" ")[1]: line for line in lines}
    assert by_qid["blocker"].endswith(" [escalated]")
    assert not by_qid["substantive_update"].endswith(" [escalated]")
    plain = analyzer._signal_lines(_rich_signals(
        order=["blocker", "substantive_update"], escalated=False))
    assert not any("[escalated]" in line for line in plain)


def test_signal_lines_score_format_and_instruction_cap(tmp_path):
    questions = _rich_questions()
    questions["severity"]["instructions"] = "x " * 200  # force long instructions
    analyzer = _analyzer(tmp_path, questions)
    lines = analyzer._signal_lines(_rich_signals(order=["severity"]))
    assert len(lines) == 1
    assert "severity (score): expected 2.60, level 3 (critical), flagged" in lines[0]
    assert len(lines[0]) <= len("- severity (score): expected 2.60, level 3 "
                                "(critical), flagged — ") + 160 + len(" [escalated]")
    assert lines[0].rstrip().endswith("…")


def test_signal_lines_never_raises_on_malformed(tmp_path):
    analyzer = _analyzer(tmp_path, _rich_questions())
    assert analyzer._signal_lines(None) == []  # type: ignore[arg-type]
    assert analyzer._signal_lines({}) == []  # type: ignore[dict-item]
    assert analyzer._signal_lines({"order": "nope", "items": "nope"}) == []  # type: ignore[dict-item]
    assert analyzer._signal_lines({"order": ["ghost"], "items": {}}) == []


def test_signal_lines_opportunity_kept_below_threshold(tmp_path):
    questions = _rich_questions()
    questions["opportunity"] = {"type": "noul", "instructions": "A new lead."}
    analyzer = _analyzer(tmp_path, questions)
    items = _rich_signals()["items"]
    items["opportunity"] = {
        "kind": "noul",
        "p": 0.10,
        "confidence": 0.9,
        "answer_confidence": 0.9,
        "gate": 0.9,
        "flagged": False,
    }
    signals = {"order": ["blocker", "opportunity", "churn_risk"],
               "items": items, "escalated": False}
    lines = analyzer._signal_lines(signals)
    assert any("opportunity" in line for line in lines)
    assert not any("churn_risk" in line for line in lines)


def test_signal_lines_opportunity_ordered_first(tmp_path):
    questions = _rich_questions()
    questions["opportunity"] = {"type": "noul", "instructions": "A new lead."}
    analyzer = _analyzer(tmp_path, questions)
    items = _rich_signals()["items"]
    items["opportunity"] = {
        "kind": "noul",
        "p": 0.80,
        "confidence": 0.9,
        "answer_confidence": 0.9,
        "gate": 0.9,
        "flagged": False,
    }
    signals = {"order": ["blocker", "severity", "opportunity"],
               "items": items, "escalated": False}
    lines = analyzer._signal_lines(signals)
    assert lines[0].startswith("- opportunity")
    # relative order of the rest preserved
    rest_idx = [i for i, line in enumerate(lines) if "blocker" in line][0]
    sev_idx = [i for i, line in enumerate(lines) if "severity" in line][0]
    assert rest_idx < sev_idx


def test_build_ledger_prompt_contains_opportunity_lead_rule(tmp_path):
    analyzer = _analyzer(tmp_path, _rich_questions())
    prompt = analyzer.build_ledger_prompt("Summary: x", [], ["hi"], None)
    assert "currently-open opportunities first" in prompt
    assert "Opportunities section the most complete" in prompt