add heavy loggin

detect/emit.py

@@ -6,6 +6,9 @@ Stages call these instead of constructing dicts or dataclasses directly.
 
 Run context (run_id, parent_job_id) is set once at pipeline start via
 set_run_context() and automatically injected into all events.
+
+Log level is set per-run with optional per-stage overrides.
+DEBUG events are only pushed when the run (or stage) log level allows it.
 """
 
 from __future__ import annotations
@@ -16,23 +19,53 @@ from datetime import datetime, timezone
 from detect.events import push_detect_event
 from detect.models import PipelineStats
 
+# Log level ordering for comparison
+_LEVEL_ORDER = {"DEBUG": 0, "INFO": 1, "WARN": 2, "ERROR": 3}
+
 # Module-level run context — set once per pipeline invocation
 _run_context: dict = {}
+_run_log_level: str = "INFO"
+_stage_log_levels: dict[str, str] = {}  # stage_name → level override
 
 
-def set_run_context(run_id: str = "", parent_job_id: str = "", run_type: str = "initial"):
+def set_run_context(
+    run_id: str = "",
+    parent_job_id: str = "",
+    run_type: str = "initial",
+    log_level: str = "INFO",
+):
     """Set the run context for all subsequent events in this pipeline invocation."""
-    global _run_context
+    global _run_context, _run_log_level
     _run_context = {
         "run_id": run_id,
         "parent_job_id": parent_job_id,
         "run_type": run_type,
     }
+    _run_log_level = log_level.upper()
+    _stage_log_levels.clear()
+
+
+def set_stage_log_level(stage: str, level: str):
+    """Override log level for a specific stage."""
+    _stage_log_levels[stage] = level.upper()
+
+
+def clear_stage_log_level(stage: str):
+    """Remove per-stage log level override."""
+    _stage_log_levels.pop(stage, None)
 
 
 def clear_run_context():
-    global _run_context
+    global _run_context, _run_log_level
     _run_context = {}
+    _run_log_level = "INFO"
+    _stage_log_levels.clear()
+
+
+def _should_emit(level: str, stage: str) -> bool:
+    """Check if this log level should be emitted given run/stage settings."""
+    effective = _stage_log_levels.get(stage, _run_log_level)
+    return _LEVEL_ORDER.get(level.upper(), 1) >= _LEVEL_ORDER.get(effective, 1)
 
 
 def _inject_context(payload: dict) -> dict:
@@ -45,6 +78,8 @@ def _inject_context(payload: dict) -> dict:
 def log(job_id: str | None, stage: str, level: str, msg: str) -> None:
     if not job_id:
         return
+    if not _should_emit(level, stage):
+        return
     payload = {
         "level": level,
         "stage": stage,