Spaces:
Running
Running
Upload monitor.py
Browse files- monitor.py +136 -0
monitor.py
ADDED
|
@@ -0,0 +1,136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
monitor.py
|
| 3 |
+
Lightweight monitoring logger for DocMind AI.
|
| 4 |
+
Logs every query, retrieval, LLM response, latency, and errors
|
| 5 |
+
to a local JSONL file readable from HuggingFace Logs tab.
|
| 6 |
+
"""
|
| 7 |
+
|
| 8 |
+
import json
|
| 9 |
+
import time
|
| 10 |
+
import os
|
| 11 |
+
from datetime import datetime, timezone
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
LOG_FILE = "/tmp/docmind_logs.jsonl"
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
def log_query(
|
| 18 |
+
question: str,
|
| 19 |
+
answer: str,
|
| 20 |
+
sources: list,
|
| 21 |
+
latency_ms: float,
|
| 22 |
+
model_used: str = "",
|
| 23 |
+
chunk_count: int = 0,
|
| 24 |
+
error: str = "",
|
| 25 |
+
):
|
| 26 |
+
"""Log a single query event."""
|
| 27 |
+
entry = {
|
| 28 |
+
"timestamp": datetime.now(timezone.utc).isoformat(),
|
| 29 |
+
"event": "query",
|
| 30 |
+
"question": question[:200], # truncate for log size
|
| 31 |
+
"answer_len": len(answer),
|
| 32 |
+
"sources": sources,
|
| 33 |
+
"latency_ms": round(latency_ms, 1),
|
| 34 |
+
"model": model_used,
|
| 35 |
+
"chunk_count": chunk_count,
|
| 36 |
+
"error": error,
|
| 37 |
+
"success": not bool(error),
|
| 38 |
+
}
|
| 39 |
+
_write(entry)
|
| 40 |
+
_print_summary(entry)
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
def log_ingestion(filename: str, chunk_count: int, latency_ms: float, error: str = ""):
|
| 44 |
+
"""Log a document ingestion event."""
|
| 45 |
+
entry = {
|
| 46 |
+
"timestamp": datetime.now(timezone.utc).isoformat(),
|
| 47 |
+
"event": "ingestion",
|
| 48 |
+
"filename": filename,
|
| 49 |
+
"chunk_count": chunk_count,
|
| 50 |
+
"latency_ms": round(latency_ms, 1),
|
| 51 |
+
"error": error,
|
| 52 |
+
"success": not bool(error),
|
| 53 |
+
}
|
| 54 |
+
_write(entry)
|
| 55 |
+
_print_summary(entry)
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
def log_startup():
|
| 59 |
+
"""Log app startup."""
|
| 60 |
+
entry = {
|
| 61 |
+
"timestamp": datetime.now(timezone.utc).isoformat(),
|
| 62 |
+
"event": "startup",
|
| 63 |
+
"message": "DocMind AI started",
|
| 64 |
+
}
|
| 65 |
+
_write(entry)
|
| 66 |
+
print("[DocMind] App started at {}".format(entry["timestamp"]))
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
def get_stats() -> dict:
|
| 70 |
+
"""Read log file and compute summary stats."""
|
| 71 |
+
if not os.path.exists(LOG_FILE):
|
| 72 |
+
return {
|
| 73 |
+
"total_queries": 0,
|
| 74 |
+
"success_rate": 0.0,
|
| 75 |
+
"avg_latency_ms": 0.0,
|
| 76 |
+
"total_ingestions": 0,
|
| 77 |
+
"errors": [],
|
| 78 |
+
}
|
| 79 |
+
|
| 80 |
+
queries = []
|
| 81 |
+
ingestions = []
|
| 82 |
+
errors = []
|
| 83 |
+
|
| 84 |
+
with open(LOG_FILE, "r") as f:
|
| 85 |
+
for line in f:
|
| 86 |
+
try:
|
| 87 |
+
entry = json.loads(line.strip())
|
| 88 |
+
if entry["event"] == "query":
|
| 89 |
+
queries.append(entry)
|
| 90 |
+
if entry.get("error"):
|
| 91 |
+
errors.append(entry)
|
| 92 |
+
elif entry["event"] == "ingestion":
|
| 93 |
+
ingestions.append(entry)
|
| 94 |
+
except Exception:
|
| 95 |
+
continue
|
| 96 |
+
|
| 97 |
+
success_count = sum(1 for q in queries if q.get("success"))
|
| 98 |
+
latencies = [q["latency_ms"] for q in queries if q.get("latency_ms")]
|
| 99 |
+
|
| 100 |
+
return {
|
| 101 |
+
"total_queries": len(queries),
|
| 102 |
+
"success_rate": round(success_count / len(queries) * 100, 1) if queries else 0.0,
|
| 103 |
+
"avg_latency_ms": round(sum(latencies) / len(latencies), 1) if latencies else 0.0,
|
| 104 |
+
"min_latency_ms": round(min(latencies), 1) if latencies else 0.0,
|
| 105 |
+
"max_latency_ms": round(max(latencies), 1) if latencies else 0.0,
|
| 106 |
+
"total_ingestions": len(ingestions),
|
| 107 |
+
"total_errors": len(errors),
|
| 108 |
+
"recent_errors": [e.get("error", "") for e in errors[-3:]],
|
| 109 |
+
}
|
| 110 |
+
|
| 111 |
+
|
| 112 |
+
def _write(entry: dict):
|
| 113 |
+
try:
|
| 114 |
+
with open(LOG_FILE, "a") as f:
|
| 115 |
+
f.write(json.dumps(entry) + "\n")
|
| 116 |
+
except Exception as e:
|
| 117 |
+
print("[DocMind Monitor] Failed to write log: {}".format(e))
|
| 118 |
+
|
| 119 |
+
|
| 120 |
+
def _print_summary(entry: dict):
|
| 121 |
+
"""Print to stdout so it shows in HuggingFace Logs tab."""
|
| 122 |
+
if entry["event"] == "query":
|
| 123 |
+
status = "OK" if entry["success"] else "ERROR"
|
| 124 |
+
print("[DocMind][{}] Q: '{}...' | Latency: {}ms | Model: {} | AnswerLen: {}".format(
|
| 125 |
+
status,
|
| 126 |
+
entry["question"][:50],
|
| 127 |
+
entry["latency_ms"],
|
| 128 |
+
entry.get("model", "unknown"),
|
| 129 |
+
entry["answer_len"],
|
| 130 |
+
))
|
| 131 |
+
elif entry["event"] == "ingestion":
|
| 132 |
+
print("[DocMind][INGEST] File: {} | Chunks: {} | Latency: {}ms".format(
|
| 133 |
+
entry["filename"],
|
| 134 |
+
entry["chunk_count"],
|
| 135 |
+
entry["latency_ms"],
|
| 136 |
+
))
|