fix: WZDx registry nested-url crash + composer "None" leak (dry-run findings) (#42)

Two bugs found running native adapters against real upstreams:

- env/wzdx.py: _select_feeds() called .strip() on the registry url field,
  but Socrata "URL"-column values arrive as {"url": "..."} — crashed ALL
  native WZDx discovery with AttributeError. Added _unwrap_url() (dict/str/
  None-robust), applied to url/apiurl/feed_url.
- composer._context_segment: appended optional fields by key presence, so
  cause: None (set by native road adapters) leaked literal "None" onto the
  wire. Guard on value (cause/expires_at truthiness; containment_pct is-not-
  None so 0% still renders). Legacy Mode-B path — golden tests unchanged.

+3 regression tests; golden/composer suites pass; full suite 10-failure
baseline (1685 passed).

Co-authored-by: Matt Johnson <mj@k7zvx.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

work/meshai/env/wzdx.py

@@ -75,6 +75,23 @@ _US_STATES = {
 }
 
 
+def _unwrap_url(v) -> str:
+    """Coerce a registry URL field to a clean string.
+
+    Socrata's "URL" column type serializes as a nested object
+    (``{"url": "https://…"}``) — e.g. Idaho's row in the live FHWA registry —
+    while plain-text columns come back as a bare string. Unwrap the nested
+    form, pass strings through, and treat anything else (None, numbers) as
+    empty so callers never trip over ``.strip()``.
+    """
+    if isinstance(v, dict):
+        inner = v.get("url")
+        return inner.strip() if isinstance(inner, str) else ""
+    if isinstance(v, str):
+        return v.strip()
+    return ""
+
+
 class WZDxAdapter:
     """FHWA WZDx work-zone polling adapter (native ``work_zone`` source)."""
 
@@ -199,7 +216,9 @@ class WZDxAdapter:
             fmt = str(row.get("format") or "").strip().lower()
             if fmt and "geojson" not in fmt:
                 continue  # only GeoJSON WZDx feeds; skip xml/other
-            url = (row.get("url") or row.get("apiurl") or row.get("feed_url") or "").strip()
+            url = (_unwrap_url(row.get("url"))
+                   or _unwrap_url(row.get("apiurl"))
+                   or _unwrap_url(row.get("feed_url")))
             if url and url not in urls:
                 urls.append(url)
         return urls

work/meshai/notifications/renderers/composer.py

@@ -285,14 +285,14 @@ def _context_segment(event: Event) -> Optional[str]:
     """Optional context (dropped FIRST when over budget)."""
     data = event.data or {}
     bits: list[str] = []
-    if "containment_pct" in data:
+    if data.get("containment_pct") is not None:
         try:
             bits.append(f"{int(float(data['containment_pct']))}% contained")
         except Exception:
             pass
-    if "cause" in data:
+    if data.get("cause"):
         bits.append(str(data["cause"]))
-    if "expires_at" in data:
+    if data.get("expires_at"):
         bits.append(f"exp {data['expires_at']}")
     return ", ".join(bits) if bits else None
 

work/tests/test_adapter_wzdx.py

@@ -122,6 +122,33 @@ def test_select_feeds_handles_garbage(adapter):
     assert adapter._select_feeds([None, 3, {"state": "ID"}]) == []  # no url
 
 
+def test_select_feeds_unwraps_socrata_url_object(adapter):
+    """Live FHWA registry serializes Idaho's ``url`` as a Socrata "URL" column
+    object (``{"url": "…"}``), not a bare string. Selection must unwrap it and
+    never raise ``AttributeError: 'dict' object has no attribute 'strip'``."""
+    rows = [
+        {"state": "Idaho", "format": "geojson",
+         "url": {"url": "https://511.idaho.gov/api/wzdx"}},
+    ]
+    feeds = adapter._select_feeds(rows)
+    assert feeds == ["https://511.idaho.gov/api/wzdx"]
+
+
+def test_tick_survives_socrata_url_object(adapter, monkeypatch):
+    """End-to-end discovery path tolerates the nested-object ``url`` shape."""
+    registry = [{"state": "Idaho", "format": "geojson",
+                 "url": {"url": "https://511.idaho.gov/api/wzdx"}}]
+    fc = make_feature_collection(make_wzdx_feature(feat_id="A"))
+
+    def fake_get(url, timeout=30):
+        return registry if "datahub" in url else fc
+
+    monkeypatch.setattr(adapter, "_http_get_json", fake_get)
+    # Must not raise; the unwrapped feed URL is discovered and fetched.
+    assert adapter.tick() is True
+    assert {e["external_id"] for e in adapter.get_events()} == {"idot-1:A"}
+
+
 # ============================================================
 # CATEGORY + CANONICAL DATA
 # ============================================================

work/tests/test_v052_dispatcher.py

@@ -232,6 +232,27 @@ def test_renderer_byte_budget_drops_optional_segments():
     assert "lightning" not in s
 
 
+def test_renderer_omits_none_context_fields():
+    """Regression — native road adapters set optional context keys present but
+    None (``cause``/``expires_at``/``containment_pct``). The composer must
+    guard on value, not key presence, so the literal string 'None' never leaks
+    onto the wire."""
+    e = make_event(
+        source="wzdx", category="wildfire_incident", severity="immediate",
+        title="Test Fire", region="Boise", timestamp=time.time(),
+        data={"cause": None, "expires_at": None, "containment_pct": None},
+    )
+    s = compose_mesh_message(e)
+    assert "None" not in s
+    # A real value on the same field still renders (guard doesn't over-suppress).
+    e2 = make_event(
+        source="wzdx", category="wildfire_incident", severity="immediate",
+        title="Test Fire", region="Boise", timestamp=time.time(),
+        data={"cause": "lightning", "expires_at": None, "containment_pct": None},
+    )
+    assert "lightning" in compose_mesh_message(e2)
+
+
 def test_renderer_never_mid_character_truncation():
     """The composer must never emit a UTF-8 byte sequence that splits a
     codepoint. Even with required-only over budget, we drop wholesale or