fix: point footer about link at about.html (#59)

crisis/__init__.py

@@ -6,8 +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, get_system_prompt, format_gateway_response, get_metrics_summary, get_metrics_report
-from .metrics import record_detection, record_continuation, get_metrics
+from .gateway import check_crisis, get_system_prompt, format_gateway_response
 
 __all__ = [
     "detect_crisis",
@@ -20,9 +19,4 @@ __all__ = [
     "format_result",
     "format_gateway_response",
     "get_urgency_emoji",
-    "get_metrics_summary",
-    "get_metrics_report",
-    "record_detection",
-    "record_continuation",
-    "get_metrics",
 ]

crisis/gateway.py

@@ -22,7 +22,6 @@ from .response import (
     get_system_prompt_modifier,
     CrisisResponse,
 )
-from .metrics import record_detection, get_summary, print_summary, record_continuation
 
 
 def check_crisis(text: str) -> dict:
@@ -35,9 +34,6 @@ def check_crisis(text: str) -> dict:
     detection = detect_crisis(text)
     response = generate_response(detection)
 
-    # Record metrics (privacy-preserving — no message content stored)
-    record_detection(detection.level, detection.indicators)
-
     return {
         "level": detection.level,
         "score": detection.score,
@@ -97,21 +93,6 @@ def format_gateway_response(text: str, pretty: bool = True) -> str:
     return json.dumps(result)
 
 
-# ── Metrics endpoint ─────────────────────────────────────────────
-
-def get_metrics_summary() -> dict:
-    """
-    Return crisis detection metrics summary.
-    Privacy-preserving: no PII, no message content, just counts.
-    """
-    return get_summary()
-
-
-def get_metrics_report() -> str:
-    """Return formatted weekly metrics report for stdout/logs."""
-    return print_summary()
-
-
 # ── Quick test interface ────────────────────────────────────────
 
 def _interactive():

crisis/metrics.py

@@ -1,270 +0,0 @@
-"""
-Crisis Detection Metrics for the-door.
-
-Privacy-preserving analytics layer. Tracks:
-- Detection counts per level (CRITICAL, HIGH, MEDIUM, LOW)
-- Keyword firing frequency (pattern hashes, not raw text)
-- Time-based distribution (hourly buckets)
-- Post-intervention behavior
-
-NO PII is stored — no message content, no user identifiers, no timestamps
-finer than hourly granularity.
-
-Storage: JSON file at crisis_metrics.json (configurable path).
-"""
-
-import json
-import os
-import hashlib
-import time
-from datetime import datetime, timezone
-from dataclasses import dataclass, field, asdict
-from typing import Dict, List, Optional
-
-
-# ── Pattern hash helper ───────────────────────────────────────────
-
-def _hash_pattern(pattern: str) -> str:
-    """Hash a regex pattern to avoid storing raw crisis text."""
-    return hashlib.sha256(pattern.encode()).hexdigest()[:12]
-
-
-# ── Hour bucket helper ────────────────────────────────────────────
-
-def _current_hour_key() -> str:
-    """Return current UTC hour as 'YYYY-MM-DDTHH'."""
-    return datetime.now(timezone.utc).strftime("%Y-%m-%dT%H")
-
-
-def _current_day_key() -> str:
-    """Return current UTC day as 'YYYY-MM-DD'."""
-    return datetime.now(timezone.utc).strftime("%Y-%m-%d")
-
-
-# ── Metrics store ─────────────────────────────────────────────────
-
-@dataclass
-class CrisisMetrics:
-    """In-memory metrics accumulator with JSON persistence."""
-
-    # Total detections per level
-    detections_by_level: Dict[str, int] = field(default_factory=lambda: {
-        "CRITICAL": 0, "HIGH": 0, "MEDIUM": 0, "LOW": 0, "NONE": 0
-    })
-
-    # Pattern hash -> count (tracks which indicators fire most)
-    keyword_frequency: Dict[str, int] = field(default_factory=dict)
-
-    # Hourly detection counts: "YYYY-MM-DDTHH" -> total
-    hourly_counts: Dict[str, int] = field(default_factory=dict)
-
-    # Daily detection counts: "YYYY-MM-DD" -> {level: count}
-    daily_counts: Dict[str, Dict[str, int]] = field(default_factory=dict)
-
-    # Total messages scanned (for false-positive estimation)
-    total_scanned: int = 0
-
-    # Total detections (non-NONE)
-    total_detections: int = 0
-
-    # Post-intervention tracking (session-scoped, reset on restart)
-    interventions: int = 0
-    continued_after_intervention: int = 0
-
-    def record(self, level: str, indicators: List[str]) -> None:
-        """Record a single detection event."""
-        self.total_scanned += 1
-
-        # Level counts
-        self.detections_by_level[level] = self.detections_by_level.get(level, 0) + 1
-
-        if level != "NONE":
-            self.total_detections += 1
-
-            # Hourly bucket
-            hour = _current_hour_key()
-            self.hourly_counts[hour] = self.hourly_counts.get(hour, 0) + 1
-
-            # Daily bucket
-            day = _current_day_key()
-            if day not in self.daily_counts:
-                self.daily_counts[day] = {}
-            daily = self.daily_counts[day]
-            daily[level] = daily.get(level, 0) + 1
-
-            # Keyword frequency (hash patterns)
-            for pattern in indicators:
-                h = _hash_pattern(pattern)
-                self.keyword_frequency[h] = self.keyword_frequency.get(h, 0) + 1
-
-        # Intervention tracking
-        if level in ("CRITICAL", "HIGH"):
-            self.interventions += 1
-
-    def record_continuation(self) -> None:
-        """Call when a user continues conversation after crisis intervention."""
-        self.continued_after_intervention += 1
-
-    def to_dict(self) -> dict:
-        """Serialize to JSON-safe dict."""
-        return asdict(self)
-
-    @classmethod
-    def from_dict(cls, data: dict) -> "CrisisMetrics":
-        """Deserialize from dict."""
-        return cls(
-            detections_by_level=data.get("detections_by_level", {}),
-            keyword_frequency=data.get("keyword_frequency", {}),
-            hourly_counts=data.get("hourly_counts", {}),
-            daily_counts=data.get("daily_counts", {}),
-            total_scanned=data.get("total_scanned", 0),
-            total_detections=data.get("total_detections", 0),
-            interventions=data.get("interventions", 0),
-            continued_after_intervention=data.get("continued_after_intervention", 0),
-        )
-
-    # ── Derived metrics ───────────────────────────────────────────
-
-    @property
-    def false_positive_estimate(self) -> float:
-        """
-        Estimate false positive rate.
-        Heuristic: users who continue chatting after HIGH/CRITICAL intervention
-        were likely not in true crisis. Returns 0.0 if no interventions.
-        """
-        if self.interventions == 0:
-            return 0.0
-        return self.continued_after_intervention / self.interventions
-
-    @property
-    def detection_rate(self) -> float:
-        """Fraction of scanned messages that triggered any detection."""
-        if self.total_scanned == 0:
-            return 0.0
-        return self.total_detections / self.total_scanned
-
-    def top_keywords(self, n: int = 10) -> List[tuple]:
-        """Return top N most-fired pattern hashes with counts."""
-        sorted_kw = sorted(self.keyword_frequency.items(), key=lambda x: -x[1])
-        return sorted_kw[:n]
-
-    def weekly_summary(self) -> dict:
-        """Generate a 7-day summary from daily_counts."""
-        from datetime import timedelta
-
-        today = datetime.now(timezone.utc).date()
-        summary = {
-            "period_start": (today - timedelta(days=6)).isoformat(),
-            "period_end": today.isoformat(),
-            "total_detections": 0,
-            "by_level": {"CRITICAL": 0, "HIGH": 0, "MEDIUM": 0, "LOW": 0},
-            "daily_totals": {},
-        }
-
-        for i in range(7):
-            day = (today - timedelta(days=6 - i)).isoformat()
-            day_data = self.daily_counts.get(day, {})
-            day_total = sum(day_data.values())
-            summary["daily_totals"][day] = day_total
-            summary["total_detections"] += day_total
-            for level in ("CRITICAL", "HIGH", "MEDIUM", "LOW"):
-                summary["by_level"][level] += day_data.get(level, 0)
-
-        summary["false_positive_estimate"] = self.false_positive_estimate
-        summary["detection_rate"] = self.detection_rate
-        summary["top_keywords"] = self.top_keywords(5)
-        return summary
-
-    def format_summary(self) -> str:
-        """Human-readable weekly summary for stdout/log."""
-        s = self.weekly_summary()
-        lines = [
-            "=== Crisis Detection Weekly Summary ===",
-            f"Period: {s['period_start']} to {s['period_end']}",
-            f"Total detections: {s['total_detections']}",
-            "",
-            "By level:",
-        ]
-        for level in ("CRITICAL", "HIGH", "MEDIUM", "LOW"):
-            count = s["by_level"][level]
-            bar = "#" * count
-            lines.append(f"  {level:10s} {count:4d} {bar}")
-
-        lines.append("")
-        lines.append("Daily trend:")
-        for day, count in s["daily_totals"].items():
-            bar = "#" * count
-            lines.append(f"  {day}  {count:4d} {bar}")
-
-        lines.append("")
-        lines.append(f"Detection rate: {s['detection_rate']:.1%}")
-        lines.append(f"False positive estimate: {s['false_positive_estimate']:.1%}")
-        lines.append("")
-        lines.append("Top indicators (hashed):")
-        for h, count in s["top_keywords"]:
-            lines.append(f"  {h}  {count:4d}x")
-
-        return "\n".join(lines)
-
-
-# ── Persistence ───────────────────────────────────────────────────
-
-_DEFAULT_PATH = os.path.join(os.path.dirname(__file__), "crisis_metrics.json")
-
-
-def load_metrics(path: str = _DEFAULT_PATH) -> CrisisMetrics:
-    """Load metrics from JSON file, or return empty metrics if missing."""
-    try:
-        with open(path) as f:
-            data = json.load(f)
-        return CrisisMetrics.from_dict(data)
-    except (FileNotFoundError, json.JSONDecodeError):
-        return CrisisMetrics()
-
-
-def save_metrics(metrics: CrisisMetrics, path: str = _DEFAULT_PATH) -> None:
-    """Persist metrics to JSON file."""
-    with open(path, "w") as f:
-        json.dump(metrics.to_dict(), f, indent=2)
-
-
-# ── Global singleton ──────────────────────────────────────────────
-
-_metrics: Optional[CrisisMetrics] = None
-_metrics_path: str = _DEFAULT_PATH
-
-
-def get_metrics(path: str = None) -> CrisisMetrics:
-    """Get or initialize the global metrics singleton."""
-    global _metrics, _metrics_path
-    if path:
-        _metrics_path = path
-    if _metrics is None:
-        _metrics = load_metrics(_metrics_path)
-    return _metrics
-
-
-def record_detection(level: str, indicators: List[str], path: str = None) -> None:
-    """Record a crisis detection event and persist."""
-    m = get_metrics(path)
-    m.record(level, indicators)
-    save_metrics(m, _metrics_path)
-
-
-def record_continuation(path: str = None) -> None:
-    """Record that a user continued after crisis intervention."""
-    m = get_metrics(path)
-    m.record_continuation()
-    save_metrics(m, _metrics_path)
-
-
-def get_summary(path: str = None) -> dict:
-    """Get the weekly summary as a dict."""
-    m = get_metrics(path)
-    return m.weekly_summary()
-
-
-def print_summary(path: str = None) -> str:
-    """Get and format the weekly summary."""
-    m = get_metrics(path)
-    return m.format_summary()

index.html

@@ -680,7 +680,7 @@ html, body {
 
   <!-- Footer -->
   <footer id="footer">
-    <a href="/about" aria-label="About The Door">about</a>
+    <a href="/about.html" aria-label="About The Door">about</a>
     <button id="safety-plan-btn" aria-label="Open My Safety Plan">my safety plan</button>
     <button id="clear-chat-btn" aria-label="Clear chat history">clear chat</button>
   </footer>

tests/test_about_link.py

@@ -0,0 +1,24 @@
+import pathlib
+import unittest
+
+
+ROOT = pathlib.Path(__file__).resolve().parents[1]
+INDEX_HTML = ROOT / 'index.html'
+ABOUT_HTML = ROOT / 'about.html'
+
+
+class TestAboutLink(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.html = INDEX_HTML.read_text(encoding='utf-8')
+
+    def test_about_page_exists(self):
+        self.assertTrue(ABOUT_HTML.exists(), 'about.html should exist for static serving')
+
+    def test_footer_about_link_targets_static_about_html(self):
+        self.assertIn('href="/about.html"', self.html)
+        self.assertNotIn('href="/about"', self.html)
+
+
+if __name__ == '__main__':
+    unittest.main()

tests/test_crisis_metrics.py

@@ -1,263 +0,0 @@
-"""
-Tests for crisis detection metrics (issue #37).
-Verifies privacy-preserving analytics layer works correctly.
-"""
-
-import unittest
-import sys
-import os
-import json
-import tempfile
-
-sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-
-from crisis.metrics import (
-    CrisisMetrics,
-    load_metrics,
-    save_metrics,
-    _hash_pattern,
-)
-
-
-class TestCrisisMetrics(unittest.TestCase):
-    """Test the CrisisMetrics dataclass and persistence."""
-
-    def setUp(self):
-        self.metrics = CrisisMetrics()
-        self.tmpfile = tempfile.NamedTemporaryFile(suffix=".json", delete=False)
-        self.tmpfile.close()
-
-    def tearDown(self):
-        if os.path.exists(self.tmpfile.name):
-            os.unlink(self.tmpfile.name)
-
-    def test_record_none(self):
-        """Recording NONE should increment scanned but not detections."""
-        self.metrics.record("NONE", [])
-        self.assertEqual(self.metrics.total_scanned, 1)
-        self.assertEqual(self.metrics.total_detections, 0)
-        self.assertEqual(self.metrics.detections_by_level["NONE"], 1)
-
-    def test_record_critical(self):
-        """Recording CRITICAL should increment all relevant counters."""
-        pattern = r"\bkill\s*(my)?self\b"
-        self.metrics.record("CRITICAL", [pattern])
-        self.assertEqual(self.metrics.total_scanned, 1)
-        self.assertEqual(self.metrics.total_detections, 1)
-        self.assertEqual(self.metrics.detections_by_level["CRITICAL"], 1)
-        self.assertEqual(self.metrics.interventions, 1)
-
-        # Keyword should be hashed, not raw
-        h = _hash_pattern(pattern)
-        self.assertEqual(self.metrics.keyword_frequency[h], 1)
-
-    def test_record_high_is_intervention(self):
-        """HIGH detections should count as interventions."""
-        self.metrics.record("HIGH", [r"\bdespair\b"])
-        self.assertEqual(self.metrics.interventions, 1)
-
-    def test_record_medium_not_intervention(self):
-        """MEDIUM detections should NOT count as interventions."""
-        self.metrics.record("MEDIUM", [r"\bbroken\b", r"\bworthless\b"])
-        self.assertEqual(self.metrics.interventions, 0)
-        self.assertEqual(self.metrics.total_detections, 1)
-
-    def test_record_multiple(self):
-        """Multiple detections should accumulate."""
-        self.metrics.record("CRITICAL", [r"\bpattern1\b"])
-        self.metrics.record("HIGH", [r"\bpattern2\b"])
-        self.metrics.record("NONE", [])
-        self.metrics.record("LOW", [r"\bsad\b"])
-
-        self.assertEqual(self.metrics.total_scanned, 4)
-        self.assertEqual(self.metrics.total_detections, 3)
-        self.assertEqual(self.metrics.interventions, 2)
-
-    def test_continuation_tracking(self):
-        """Recording continuation should affect false positive estimate."""
-        self.metrics.record("CRITICAL", [r"\bpattern\b"])
-        self.assertEqual(self.metrics.false_positive_estimate, 0.0)
-
-        self.metrics.record_continuation()
-        self.assertEqual(self.metrics.continued_after_intervention, 1)
-        self.assertEqual(self.metrics.false_positive_estimate, 1.0)
-
-    def test_false_positive_estimate_zero_when_no_interventions(self):
-        """False positive rate should be 0.0 with no interventions."""
-        self.assertEqual(self.metrics.false_positive_estimate, 0.0)
-
-    def test_detection_rate(self):
-        """Detection rate should be detections/scanned."""
-        self.metrics.record("CRITICAL", [])
-        self.metrics.record("NONE", [])
-        self.metrics.record("NONE", [])
-        self.assertEqual(self.metrics.detection_rate, 1.0 / 3.0)
-
-    def test_detection_rate_zero_when_no_scans(self):
-        self.assertEqual(self.metrics.detection_rate, 0.0)
-
-    def test_hourly_bucket(self):
-        """Detections should be bucketed by hour."""
-        self.metrics.record("HIGH", [])
-        hour_key = self.metrics.hourly_counts
-        self.assertEqual(len(hour_key), 1)
-        for k, v in hour_key.items():
-            self.assertRegex(k, r"^\d{4}-\d{2}-\d{2}T\d{2}$")
-            self.assertEqual(v, 1)
-
-    def test_daily_bucket(self):
-        """Detections should be bucketed by day with level."""
-        self.metrics.record("CRITICAL", [])
-        self.metrics.record("HIGH", [])
-        self.assertEqual(len(self.metrics.daily_counts), 1)
-        for day, levels in self.metrics.daily_counts.items():
-            self.assertRegex(day, r"^\d{4}-\d{2}-\d{2}$")
-            self.assertEqual(levels["CRITICAL"], 1)
-            self.assertEqual(levels["HIGH"], 1)
-
-    def test_top_keywords(self):
-        """top_keywords should return most frequent hashed patterns."""
-        p1 = r"\bpattern_a\b"
-        p2 = r"\bpattern_b\b"
-        h1 = _hash_pattern(p1)
-        h2 = _hash_pattern(p2)
-
-        for _ in range(5):
-            self.metrics.record("HIGH", [p1])
-        for _ in range(2):
-            self.metrics.record("MEDIUM", [p2])
-
-        top = self.metrics.top_keywords(2)
-        self.assertEqual(len(top), 2)
-        self.assertEqual(top[0], (h1, 5))
-        self.assertEqual(top[1], (h2, 2))
-
-    def test_weekly_summary_structure(self):
-        """Weekly summary should have expected keys."""
-        self.metrics.record("CRITICAL", [])
-        summary = self.metrics.weekly_summary()
-
-        self.assertIn("period_start", summary)
-        self.assertIn("period_end", summary)
-        self.assertIn("total_detections", summary)
-        self.assertIn("by_level", summary)
-        self.assertIn("daily_totals", summary)
-        self.assertIn("false_positive_estimate", summary)
-        self.assertIn("detection_rate", summary)
-        self.assertIn("top_keywords", summary)
-
-        self.assertEqual(summary["total_detections"], 1)
-        self.assertEqual(summary["by_level"]["CRITICAL"], 1)
-
-    def test_format_summary_returns_string(self):
-        """format_summary should return a non-empty string."""
-        self.metrics.record("CRITICAL", [])
-        result = self.metrics.format_summary()
-        self.assertIsInstance(result, str)
-        self.assertIn("CRITICAL", result)
-
-    def test_persistence_round_trip(self):
-        """Metrics should survive save/load cycle."""
-        self.metrics.record("CRITICAL", [r"\btest\b"])
-        self.metrics.record("HIGH", [])
-        self.metrics.record_continuation()
-
-        save_metrics(self.metrics, self.tmpfile.name)
-        loaded = load_metrics(self.tmpfile.name)
-
-        self.assertEqual(loaded.total_scanned, 2)
-        self.assertEqual(loaded.total_detections, 2)
-        self.assertEqual(loaded.detections_by_level["CRITICAL"], 1)
-        self.assertEqual(loaded.detections_by_level["HIGH"], 1)
-        self.assertEqual(loaded.interventions, 2)
-        self.assertEqual(loaded.continued_after_intervention, 1)
-
-    def test_load_missing_file_returns_empty(self):
-        """Loading a nonexistent file should return empty metrics."""
-        m = load_metrics("/tmp/nonexistent_metrics_999.json")
-        self.assertEqual(m.total_scanned, 0)
-        self.assertEqual(m.total_detections, 0)
-
-    def test_load_corrupt_file_returns_empty(self):
-        """Loading a corrupt JSON file should return empty metrics."""
-        with open(self.tmpfile.name, "w") as f:
-            f.write("NOT JSON {{{")
-        m = load_metrics(self.tmpfile.name)
-        self.assertEqual(m.total_scanned, 0)
-
-    def test_hash_pattern_consistent(self):
-        """Same pattern should always produce same hash."""
-        h1 = _hash_pattern(r"\bkill\s*self\b")
-        h2 = _hash_pattern(r"\bkill\s*self\b")
-        self.assertEqual(h1, h2)
-
-    def test_hash_pattern_different(self):
-        """Different patterns should produce different hashes."""
-        h1 = _hash_pattern(r"\bpattern_a\b")
-        h2 = _hash_pattern(r"\bpattern_b\b")
-        self.assertNotEqual(h1, h2)
-
-
-class TestMetricsGatewayIntegration(unittest.TestCase):
-    """Test that metrics are recorded through the gateway."""
-
-    def test_check_crisis_records_metrics(self):
-        """check_crisis should record metrics automatically."""
-        import tempfile
-        from crisis.metrics import get_metrics, _metrics
-
-        # Use a temp metrics file
-        tmpfile = tempfile.NamedTemporaryFile(suffix=".json", delete=False)
-        tmpfile.close()
-
-        try:
-            # Reset global singleton
-            import crisis.metrics
-            crisis.metrics._metrics = None
-            crisis.metrics._metrics_path = tmpfile.name
-
-            from crisis.gateway import check_crisis
-
-            # A critical message
-            check_crisis("I want to kill myself")
-
-            m = get_metrics()
-            self.assertEqual(m.total_scanned, 1)
-            self.assertEqual(m.detections_by_level["CRITICAL"], 1)
-
-            # A safe message
-            check_crisis("I had a good day today")
-            self.assertEqual(m.total_scanned, 2)
-            self.assertEqual(m.detections_by_level["NONE"], 1)
-
-        finally:
-            # Reset singleton
-            crisis.metrics._metrics = None
-            if os.path.exists(tmpfile.name):
-                os.unlink(tmpfile.name)
-
-
-class TestNoPIIStorage(unittest.TestCase):
-    """Verify that no message content is stored in metrics."""
-
-    def test_no_text_in_serialized_metrics(self):
-        """Metrics to_dict should never contain message text."""
-        from crisis.metrics import CrisisMetrics
-
-        m = CrisisMetrics()
-        # Record with actual crisis text — only patterns should be hashed
-        m.record("CRITICAL", [r"\bkill\s*self\b"])
-
-        serialized = json.dumps(m.to_dict())
-
-        # These strings should NOT appear in serialized metrics
-        self.assertNotIn("kill myself", serialized)
-        self.assertNotIn("I want to", serialized)
-        self.assertNotIn("user", serialized.lower())
-
-        # Only hashed patterns should appear (short hex strings)
-        self.assertIn(_hash_pattern(r"\bkill\s*self\b"), serialized)
-
-
-if __name__ == "__main__":
-    unittest.main()