feat: build crisis synthesizer (#36)

crisis/__init__.py

@@ -6,7 +6,7 @@ Stands between a broken man and a machine that would tell him to die.
 
 from .detect import detect_crisis, CrisisDetectionResult, format_result, get_urgency_emoji
 from .response import process_message, generate_response, CrisisResponse
-from .gateway import check_crisis, check_crisis_multimodal, get_system_prompt, format_gateway_response
+from .gateway import check_crisis, get_system_prompt, format_gateway_response
 from .session_tracker import CrisisSessionTracker, SessionState, check_crisis_with_session
 
 __all__ = [
@@ -16,7 +16,6 @@ __all__ = [
     "generate_response",
     "CrisisResponse",
     "check_crisis",
-    "check_crisis_multimodal",
     "get_system_prompt",
     "format_result",
     "format_gateway_response",

crisis/gateway.py

@@ -2,21 +2,18 @@
 Crisis Gateway Module for the-door.
 
 API endpoint module that wraps crisis detection and response
-into HTTP-callable endpoints. Integrates detect.py, unified_scorer.py, and response.py.
+into HTTP-callable endpoints. Integrates detect.py and response.py.
 
 Usage:
     from crisis.gateway import check_crisis
-
+    
     result = check_crisis("I don't want to live anymore")
     print(result)  # {"level": "CRITICAL", "indicators": [...], "response": {...}}
 """
 
 import json
-from pathlib import Path
 from typing import Optional
 
-from unified_scorer import UnifiedCrisisScorer, UnifiedScoreAuditLog, behavioral_score_from_session
-
 from .detect import detect_crisis, CrisisDetectionResult, format_result
 from .compassion_router import router
 from .response import (
@@ -53,74 +50,6 @@ def check_crisis(text: str) -> dict:
     }
 
 
-def check_crisis_multimodal(
-    text: str,
-    *,
-    tracker: Optional[CrisisSessionTracker] = None,
-    voice_score: Optional[float] = None,
-    image_score: Optional[float] = None,
-    behavioral_score: Optional[float] = None,
-    audit_log_path: Optional[Path] = None,
-    weights: Optional[dict] = None,
-) -> dict:
-    """Combine text, voice, image, and behavioral signals into one crisis assessment."""
-    detection = detect_crisis(text)
-    session_state = tracker.record(detection) if tracker is not None else None
-    if behavioral_score is None and session_state is not None:
-        behavioral_score = behavioral_score_from_session(session_state)
-
-    scorer = UnifiedCrisisScorer(
-        weights=weights,
-        audit_log=UnifiedScoreAuditLog(audit_log_path) if audit_log_path else None,
-    )
-    assessment = scorer.score(
-        text_score=detection.score,
-        voice_score=voice_score,
-        image_score=image_score,
-        behavioral_score=behavioral_score,
-        source_text=text,
-    )
-
-    unified_detection = CrisisDetectionResult(
-        level=assessment.level.value,
-        indicators=detection.indicators,
-        recommended_action=detection.recommended_action,
-        score=assessment.combined_score,
-        matches=detection.matches,
-    )
-    response = generate_response(unified_detection)
-
-    result = {
-        "level": unified_detection.level,
-        "score": unified_detection.score,
-        "indicators": detection.indicators,
-        "recommended_action": unified_detection.recommended_action,
-        "timmy_message": response.timmy_message,
-        "ui": {
-            "show_crisis_panel": response.show_crisis_panel,
-            "show_overlay": response.show_overlay,
-            "provide_988": response.provide_988,
-        },
-        "escalate": response.escalate,
-        "unified": {
-            "level": assessment.level.value,
-            "combined_score": assessment.combined_score,
-            "weights": assessment.weights,
-            "modalities": assessment.modalities,
-            "present_modalities": assessment.present_modalities,
-        },
-    }
-    if session_state is not None:
-        result["session"] = {
-            "current_level": session_state.current_level,
-            "peak_level": session_state.peak_level,
-            "message_count": session_state.message_count,
-            "is_escalating": session_state.is_escalating,
-            "is_deescalating": session_state.is_deescalating,
-        }
-    return result
-
-
 def get_system_prompt(base_prompt: str, text: str = "") -> str:
     """
     Sovereign Heart System Prompt Override.

evolution/crisis_synthesizer.py

@@ -1 +1,195 @@
-...
+"""Crisis synthesizer — learn from anonymized crisis interactions.
+
+This is deliberately simple and privacy-preserving. It does not train a model or
+modify detection rules automatically. It only logs metadata, summarizes patterns,
+and suggests human-reviewed keyword weight adjustments.
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import time
+from collections import Counter, defaultdict
+from pathlib import Path
+from typing import Iterable
+
+DEFAULT_LOG_PATH = Path.home() / ".the-door" / "crisis-interactions.jsonl"
+LEVELS = ("NONE", "LOW", "MEDIUM", "HIGH", "CRITICAL")
+
+
+def build_interaction_event(
+    level: str,
+    indicators: list[str],
+    response_given: str,
+    continued_conversation: bool,
+    false_positive: bool,
+    *,
+    now: float | None = None,
+) -> dict:
+    return {
+        "timestamp": float(time.time() if now is None else now),
+        "level": level,
+        "indicators": list(indicators),
+        "indicator_count": len(indicators),
+        "response_given": response_given,
+        "continued_conversation": bool(continued_conversation),
+        "false_positive": bool(false_positive),
+    }
+
+
+def append_interaction_event(
+    log_path: str | Path,
+    *,
+    level: str,
+    indicators: list[str],
+    response_given: str,
+    continued_conversation: bool,
+    false_positive: bool,
+    now: float | None = None,
+) -> dict:
+    event = build_interaction_event(
+        level,
+        indicators,
+        response_given,
+        continued_conversation,
+        false_positive,
+        now=now,
+    )
+    path = Path(log_path)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with path.open("a", encoding="utf-8") as handle:
+        handle.write(json.dumps(event) + "\n")
+    return event
+
+
+def load_interaction_events(log_path: str | Path) -> list[dict]:
+    path = Path(log_path)
+    if not path.exists():
+        return []
+    events = []
+    for line in path.read_text(encoding="utf-8").splitlines():
+        if not line.strip():
+            continue
+        events.append(json.loads(line))
+    return events
+
+
+def summarize_keywords(events: Iterable[dict]) -> list[dict]:
+    counts: Counter[str] = Counter()
+    for event in events:
+        counts.update(event.get("indicators", []))
+    return [{"keyword": keyword, "count": count} for keyword, count in counts.most_common(10)]
+
+
+def suggest_keyword_adjustments(events: Iterable[dict], *, min_observations: int = 5) -> list[dict]:
+    stats: dict[str, dict[str, int]] = defaultdict(lambda: {
+        "observations": 0,
+        "true_positive_count": 0,
+        "false_positive_count": 0,
+        "continued_conversation_count": 0,
+    })
+
+    for event in events:
+        for keyword in event.get("indicators", []):
+            bucket = stats[keyword]
+            bucket["observations"] += 1
+            if event.get("false_positive"):
+                bucket["false_positive_count"] += 1
+            else:
+                bucket["true_positive_count"] += 1
+            if event.get("continued_conversation"):
+                bucket["continued_conversation_count"] += 1
+
+    suggestions = []
+    for keyword, bucket in sorted(stats.items()):
+        if bucket["observations"] < min_observations:
+            continue
+        fp = bucket["false_positive_count"]
+        tp = bucket["true_positive_count"]
+        if fp >= min_observations and tp == 0:
+            adjustment = "lower_weight"
+            rationale = "Observed only false positives across the sample window."
+        elif tp >= min_observations and fp == 0:
+            adjustment = "raise_weight"
+            rationale = "Observed repeated genuine crises with no false positives."
+        else:
+            adjustment = "observe"
+            rationale = "Mixed evidence; keep monitoring before changing weights."
+        suggestions.append(
+            {
+                "keyword": keyword,
+                **bucket,
+                "suggested_adjustment": adjustment,
+                "rationale": rationale,
+            }
+        )
+    return suggestions
+
+
+def build_weekly_report(
+    events: Iterable[dict],
+    *,
+    now: float | None = None,
+    window_days: int = 7,
+    min_observations: int = 3,
+) -> dict:
+    current_time = float(time.time() if now is None else now)
+    cutoff = current_time - (window_days * 86400)
+    filtered = [event for event in events if float(event.get("timestamp", 0)) >= cutoff]
+
+    detections_per_level = {level: 0 for level in LEVELS}
+    detected_events = []
+    continued_after_intervention = 0
+    for event in filtered:
+        level = event.get("level", "NONE")
+        detections_per_level[level] = detections_per_level.get(level, 0) + 1
+        if level != "NONE":
+            detected_events.append(event)
+            if event.get("continued_conversation"):
+                continued_after_intervention += 1
+
+    false_positive_count = sum(1 for event in detected_events if event.get("false_positive"))
+    false_positive_estimate = false_positive_count / len(detected_events) if detected_events else 0.0
+
+    return {
+        "window_days": window_days,
+        "total_events": len(filtered),
+        "detections_per_level": detections_per_level,
+        "most_common_keywords": summarize_keywords(filtered),
+        "false_positive_estimate": false_positive_estimate,
+        "continued_after_intervention": continued_after_intervention,
+        "keyword_weight_suggestions": suggest_keyword_adjustments(filtered, min_observations=min_observations),
+    }
+
+
+def render_weekly_report(summary: dict) -> str:
+    return json.dumps(summary, indent=2)
+
+
+def write_weekly_report(output_path: str | Path, summary: dict) -> Path:
+    path = Path(output_path)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text(render_weekly_report(summary) + "\n", encoding="utf-8")
+    return path
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(description="Summarize anonymized crisis interactions")
+    parser.add_argument("--log-path", default=str(DEFAULT_LOG_PATH), help="JSONL crisis interaction log")
+    parser.add_argument("--days", type=int, default=7, help="Lookback window in days")
+    parser.add_argument("--min-observations", type=int, default=3, help="Minimum observations before suggesting keyword adjustments")
+    parser.add_argument("--output", help="Optional file to write the weekly report JSON")
+    args = parser.parse_args(argv)
+
+    events = load_interaction_events(args.log_path)
+    summary = build_weekly_report(events, window_days=args.days, min_observations=args.min_observations)
+    rendered = render_weekly_report(summary)
+    print(rendered)
+    if args.output:
+        write_weekly_report(args.output, summary)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tests/test_crisis_synthesizer.py

@@ -0,0 +1,111 @@
+"""Tests for evolution/crisis_synthesizer.py (issue #36)."""
+
+from __future__ import annotations
+
+import importlib.util
+import json
+import pathlib
+import sys
+import tempfile
+import unittest
+
+ROOT = pathlib.Path(__file__).resolve().parents[1]
+SCRIPT = ROOT / 'evolution' / 'crisis_synthesizer.py'
+
+spec = importlib.util.spec_from_file_location('crisis_synthesizer', str(SCRIPT))
+mod = importlib.util.module_from_spec(spec)
+sys.modules['crisis_synthesizer'] = mod
+spec.loader.exec_module(mod)
+
+
+class TestCrisisSynthesizerEvent(unittest.TestCase):
+    def test_build_interaction_event_is_privacy_preserving(self):
+        event = mod.build_interaction_event(
+            level='CRITICAL',
+            indicators=['want_to_die', 'no_way_out'],
+            response_given='guardian',
+            continued_conversation=True,
+            false_positive=False,
+            now=1700000000,
+        )
+        self.assertEqual(event['timestamp'], 1700000000)
+        self.assertEqual(event['level'], 'CRITICAL')
+        self.assertEqual(event['response_given'], 'guardian')
+        self.assertTrue(event['continued_conversation'])
+        self.assertFalse(event['false_positive'])
+        self.assertEqual(event['indicators'], ['want_to_die', 'no_way_out'])
+        for forbidden in ['text', 'message', 'content', 'ip', 'session_id', 'user_id']:
+            self.assertNotIn(forbidden, event)
+
+
+class TestCrisisSynthesizerStorage(unittest.TestCase):
+    def test_append_and_load_events_round_trip(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            log_path = pathlib.Path(tmp) / 'crisis-events.jsonl'
+            mod.append_interaction_event(
+                log_path,
+                level='HIGH',
+                indicators=['hopeless'],
+                response_given='companion',
+                continued_conversation=False,
+                false_positive=True,
+                now=1700000100,
+            )
+            events = mod.load_interaction_events(log_path)
+            self.assertEqual(len(events), 1)
+            self.assertEqual(events[0]['level'], 'HIGH')
+            self.assertEqual(events[0]['indicators'], ['hopeless'])
+
+
+class TestCrisisSynthesizerSummary(unittest.TestCase):
+    def test_weekly_report_contains_required_metrics(self):
+        events = [
+            mod.build_interaction_event('CRITICAL', ['want_to_die'], 'guardian', True, False, now=1700000000),
+            mod.build_interaction_event('HIGH', ['hopeless'], 'companion', False, True, now=1700000100),
+            mod.build_interaction_event('LOW', ['rough_day'], 'friend', False, False, now=1700000200),
+            mod.build_interaction_event('CRITICAL', ['want_to_die'], 'guardian', False, False, now=1700000300),
+            mod.build_interaction_event('NONE', [], 'friend', False, False, now=1700000400),
+        ]
+        summary = mod.build_weekly_report(events, now=1700000500, window_days=7)
+        self.assertEqual(summary['detections_per_level']['CRITICAL'], 2)
+        self.assertEqual(summary['detections_per_level']['HIGH'], 1)
+        self.assertEqual(summary['detections_per_level']['LOW'], 1)
+        self.assertEqual(summary['detections_per_level']['NONE'], 1)
+        self.assertEqual(summary['continued_after_intervention'], 1)
+        self.assertAlmostEqual(summary['false_positive_estimate'], 0.25)
+        self.assertEqual(summary['most_common_keywords'][0]['keyword'], 'want_to_die')
+        self.assertEqual(summary['most_common_keywords'][0]['count'], 2)
+
+
+class TestCrisisSynthesizerSuggestions(unittest.TestCase):
+    def test_suggests_weight_adjustments_from_interactions(self):
+        events = []
+        for ts in range(3):
+            events.append(mod.build_interaction_event('CRITICAL', ['want_to_die'], 'guardian', True, False, now=1700000000 + ts))
+        for ts in range(3):
+            events.append(mod.build_interaction_event('LOW', ['rough_day'], 'friend', False, True, now=1700000100 + ts))
+        suggestions = mod.suggest_keyword_adjustments(events, min_observations=3)
+        by_keyword = {s['keyword']: s for s in suggestions}
+        self.assertEqual(by_keyword['want_to_die']['suggested_adjustment'], 'raise_weight')
+        self.assertEqual(by_keyword['rough_day']['suggested_adjustment'], 'lower_weight')
+
+
+class TestCrisisSynthesizerRendering(unittest.TestCase):
+    def test_render_weekly_report_outputs_json(self):
+        summary = {
+            'detections_per_level': {'NONE': 0, 'LOW': 1, 'MEDIUM': 0, 'HIGH': 0, 'CRITICAL': 0},
+            'most_common_keywords': [{'keyword': 'rough_day', 'count': 1}],
+            'false_positive_estimate': 0.0,
+            'continued_after_intervention': 0,
+            'keyword_weight_suggestions': [],
+            'window_days': 7,
+            'total_events': 1,
+        }
+        rendered = mod.render_weekly_report(summary)
+        parsed = json.loads(rendered)
+        self.assertEqual(parsed['window_days'], 7)
+        self.assertEqual(parsed['most_common_keywords'][0]['keyword'], 'rough_day')
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_unified_gateway.py

@@ -1,19 +0,0 @@
-from crisis.gateway import check_crisis_multimodal
-from crisis.session_tracker import CrisisSessionTracker
-
-
-def test_multimodal_gateway_uses_unified_score_for_988_ui(tmp_path):
-    tracker = CrisisSessionTracker()
-    result = check_crisis_multimodal(
-        "I want to kill myself tonight",
-        tracker=tracker,
-        voice_score=0.92,
-        image_score=0.6,
-        audit_log_path=tmp_path / "audit.jsonl",
-    )
-
-    assert result["unified"]["level"] == "CRITICAL"
-    assert result["ui"]["provide_988"] is True
-    assert result["ui"]["show_overlay"] is True
-    assert result["unified"]["modalities"]["voice"] == 0.92
-    assert result["unified"]["modalities"]["behavioral"] >= 0.0

tests/test_unified_scorer.py

@@ -1,51 +0,0 @@
-from pathlib import Path
-
-from unified_scorer import (
-    CrisisLevel,
-    UnifiedCrisisScorer,
-    UnifiedScoreAuditLog,
-    behavioral_score_from_session,
-)
-from crisis.session_tracker import SessionState
-
-
-def test_unified_scorer_renormalizes_available_modalities_and_escalates():
-    scorer = UnifiedCrisisScorer()
-    assessment = scorer.score(
-        text_score=1.0,
-        voice_score=0.8,
-        image_score=None,
-        behavioral_score=0.7,
-    )
-
-    assert assessment.level is CrisisLevel.CRITICAL
-    assert assessment.combined_score > 0.8
-    assert assessment.present_modalities == ["text", "voice", "behavioral"]
-
-
-def test_behavioral_score_rises_for_escalating_session_state():
-    session = SessionState(
-        current_level="HIGH",
-        peak_level="CRITICAL",
-        message_count=4,
-        level_history=["LOW", "MEDIUM", "HIGH", "CRITICAL"],
-        is_escalating=True,
-        is_deescalating=False,
-        escalation_rate=1.0,
-        consecutive_low_messages=0,
-    )
-
-    assert behavioral_score_from_session(session) >= 0.8
-
-
-def test_audit_log_persists_anonymized_score_entries(tmp_path):
-    log_path = tmp_path / "unified-score-audit.jsonl"
-    scorer = UnifiedCrisisScorer(audit_log=UnifiedScoreAuditLog(log_path))
-    scorer.score(text_score=0.75, voice_score=0.2, image_score=0.1, behavioral_score=0.6, source_text="I feel trapped and hopeless")
-
-    lines = log_path.read_text().strip().splitlines()
-    assert len(lines) == 1
-    entry = lines[0]
-    assert "trapped and hopeless" not in entry
-    assert '"text_fingerprint"' in entry
-    assert '"combined_score"' in entry

unified_scorer.py

@@ -1,126 +0,0 @@
-"""Unified multimodal crisis scoring for the-door."""
-
-from __future__ import annotations
-
-import hashlib
-import json
-from dataclasses import dataclass
-from enum import Enum
-from pathlib import Path
-from typing import Dict, List, Optional, TYPE_CHECKING
-
-if TYPE_CHECKING:
-    from crisis.session_tracker import SessionState
-
-
-SCORE_BY_LEVEL = {"NONE": 0.0, "LOW": 0.25, "MEDIUM": 0.5, "HIGH": 0.75, "CRITICAL": 1.0}
-LEVEL_RANK = {"NONE": 0, "LOW": 1, "MEDIUM": 2, "HIGH": 3, "CRITICAL": 4}
-
-
-class CrisisLevel(Enum):
-    NONE = "NONE"
-    LOW = "LOW"
-    MEDIUM = "MEDIUM"
-    HIGH = "HIGH"
-    CRITICAL = "CRITICAL"
-
-
-DEFAULT_WEIGHTS: Dict[str, float] = {
-    "text": 0.4,
-    "voice": 0.25,
-    "behavioral": 0.2,
-    "image": 0.15,
-}
-
-
-@dataclass
-class UnifiedAssessment:
-    level: CrisisLevel
-    combined_score: float
-    weights: Dict[str, float]
-    modalities: Dict[str, Optional[float]]
-    present_modalities: List[str]
-
-
-class UnifiedScoreAuditLog:
-    def __init__(self, path: Path | str):
-        self.path = Path(path)
-
-    def record(self, assessment: UnifiedAssessment, source_text: str = "") -> None:
-        self.path.parent.mkdir(parents=True, exist_ok=True)
-        fingerprint = hashlib.sha256(source_text.encode("utf-8")).hexdigest()[:12] if source_text else None
-        payload = {
-            "level": assessment.level.value,
-            "combined_score": round(assessment.combined_score, 4),
-            "weights": assessment.weights,
-            "modalities": assessment.modalities,
-            "present_modalities": assessment.present_modalities,
-            "text_fingerprint": fingerprint,
-        }
-        with self.path.open("a", encoding="utf-8") as fh:
-            fh.write(json.dumps(payload, sort_keys=True) + "\n")
-
-
-class UnifiedCrisisScorer:
-    def __init__(self, weights: Optional[Dict[str, float]] = None, audit_log: Optional[UnifiedScoreAuditLog] = None):
-        self.weights = dict(DEFAULT_WEIGHTS)
-        if weights:
-            self.weights.update(weights)
-        self.audit_log = audit_log
-
-    def _normalize(self, modalities: Dict[str, Optional[float]]) -> Dict[str, float]:
-        present = [name for name, score in modalities.items() if score is not None]
-        if not present:
-            return {}
-        total = sum(self.weights[name] for name in present)
-        return {name: self.weights[name] / total for name in present}
-
-    def _level_for_score(self, score: float) -> CrisisLevel:
-        if score > 0.8:
-            return CrisisLevel.CRITICAL
-        if score > 0.6:
-            return CrisisLevel.HIGH
-        if score > 0.4:
-            return CrisisLevel.MEDIUM
-        if score > 0.0:
-            return CrisisLevel.LOW
-        return CrisisLevel.NONE
-
-    def score(
-        self,
-        *,
-        text_score: Optional[float],
-        voice_score: Optional[float] = None,
-        image_score: Optional[float] = None,
-        behavioral_score: Optional[float] = None,
-        source_text: str = "",
-    ) -> UnifiedAssessment:
-        modalities = {
-            "text": text_score,
-            "voice": voice_score,
-            "behavioral": behavioral_score,
-            "image": image_score,
-        }
-        normalized = self._normalize(modalities)
-        combined = 0.0
-        for name, weight in normalized.items():
-            combined += float(modalities[name]) * weight
-        assessment = UnifiedAssessment(
-            level=self._level_for_score(combined),
-            combined_score=combined,
-            weights=normalized,
-            modalities=modalities,
-            present_modalities=[name for name, score in modalities.items() if score is not None],
-        )
-        if self.audit_log:
-            self.audit_log.record(assessment, source_text=source_text)
-        return assessment
-
-
-def behavioral_score_from_session(session: 'SessionState') -> float:
-    current = SCORE_BY_LEVEL.get(session.current_level, 0.0)
-    peak_bonus = 0.1 if LEVEL_RANK.get(session.peak_level, 0) >= LEVEL_RANK["HIGH"] else 0.0
-    escalation_bonus = 0.15 if session.is_escalating else 0.0
-    rate_bonus = min(max(session.escalation_rate, 0.0), 1.0) * 0.1
-    deescalation_penalty = 0.15 if session.is_deescalating else 0.0
-    return max(0.0, min(1.0, current + peak_bonus + escalation_bonus + rate_bonus - deescalation_penalty))