Add reusable LLM parser profiles with multi-event extract.

Support kind=llm profiles (instruction/schema), optional multi-event posts via #eN URLs, and recover stale running/queued parse jobs after worker crashes.

Co-authored-by: Cursor <cursoragent@cursor.com>

contracts/heuristic_profile.py

@@ -17,6 +17,8 @@ TARGET_FIELDS: tuple[str, ...] = (
     "region",
 )
 
+COORDS_RE = re.compile(r"(-?\d{1,3}\.\d+)\s*,\s*(-?\d{1,3}\.\d+)")
+
 Strategy = Literal["regex", "line", "after_marker", "between", "full_text", "literal"]
 
 
@@ -45,8 +47,24 @@ class FieldRule(BaseModel):
 class HeuristicProfile(BaseModel):
     version: Literal[1] = 1
     fields: dict[str, FieldRule] = Field(default_factory=dict)
+    required_fields: list[str] = Field(default_factory=list)
     notes: str = ""
 
+    @field_validator("required_fields")
+    @classmethod
+    def known_required(cls, value: list[str]) -> list[str]:
+        seen: list[str] = []
+        unknown: list[str] = []
+        for name in value:
+            if name not in TARGET_FIELDS:
+                unknown.append(name)
+                continue
+            if name not in seen:
+                seen.append(name)
+        if unknown:
+            raise ValueError(f"Unknown required_fields: {sorted(set(unknown))}")
+        return seen
+
     @model_validator(mode="after")
     def known_fields_only(self) -> "HeuristicProfile":
         unknown = set(self.fields) - set(TARGET_FIELDS)
@@ -163,3 +181,38 @@ def apply_profile(text: str, profile: HeuristicProfile | dict[str, Any]) -> dict
     for name, rule in profile.fields.items():
         result[name] = _apply_rule(text, rule)
     return result
+
+
+def parse_coords(raw: str) -> tuple[float | None, float | None]:
+    match = COORDS_RE.search(raw or "")
+    if not match:
+        return None, None
+    return float(match.group(1)), float(match.group(2))
+
+
+def _field_is_filled(name: str, value: str) -> bool:
+    if not (value or "").strip():
+        return False
+    if name == "coords":
+        lat, lng = parse_coords(value)
+        return lat is not None and lng is not None
+    return True
+
+
+def match_profile(
+    text: str,
+    profile: HeuristicProfile | dict[str, Any],
+) -> tuple[bool, dict[str, str], list[str]]:
+    """Apply profile and report whether required_fields are filled.
+
+    Empty required_fields means no gate (legacy profiles ingest every post).
+    """
+    if isinstance(profile, dict):
+        profile = HeuristicProfile.model_validate(profile)
+    fields = apply_profile(text, profile)
+    missing = [
+        name
+        for name in profile.required_fields
+        if not _field_is_filled(name, fields.get(name, ""))
+    ]
+    return (not missing, fields, missing)