How mayank-groverr/raglog works
Raglog combines rule-text embedding and Qdrant retrieval with log parsing and anomaly detection. The visible rule path parses rule files, creates and validates embeddings, then stores vectors; retrieval embeds a query and searches the store. Log parsing produces ParsedLog records used by time-window statistics and detector components; training builds and saves a model, while runtime inference loads it. The README is empty, so API and alerting capabilities are visible in the tree but their wiring is not established. Edges reflect shown calls or SOURCE INDEX references.
Main entry points: core.py Core workflow · ingestion.py Embedding ingestion · routes.py API routes · qdrant.py Qdrant store
Components 25
Rule retrieval
- Core workflow
core.py - Rule parser
ingestion/parser.py - Embedding ingestion
ingestion/ingestion.py - Embedding adapter
adapter/generic.py - Embedding validator
utils/embeddings_validator.py - Vector retrieval
rag/retriever.py - Qdrant store
store/qdrant.py
Log processing
- Log parser
utils/log_parser.py - Parsed logs
models/log.py - File watcher
utils/watcher.py
Anomaly detection
- Log window
detector/window.py - Detector scorer
detector/scorer.py - Statistical detector
detector/statistical.py - ML inference
detector/ml_model.py - Model training
detector/trainer.py
Service interfaces
- HTTP API
api/server.py - API routes
api/routes.py - Log handler
api/handlers/logs.py - Rule handler
api/handlers/rules.py - Alert handler
api/handlers/alerts.py - Alert dispatcher
alerting/dispatcher.py - Email alerts
alerting/email.py - Slack alerts
alerting/slack.py
Other components
- User
- Qdrant service
Connections 27
- User → Core workflow: initiates
- Core workflow → Rule parser: calls
core.py - Core workflow → Embedding ingestion: calls
core.py - Core workflow → Vector retrieval: calls
core.py - Core workflow → Qdrant store: uses
core.py - Embedding ingestion → Embedding adapter: embeds text
ingestion/ingestion.py - Embedding ingestion → Embedding validator: validates
ingestion/ingestion.py - Embedding ingestion → Qdrant store: writes vectors
ingestion/ingestion.py - Vector retrieval → Embedding adapter: embeds query
rag/retriever.py - Vector retrieval → Qdrant store: searches
rag/retriever.py - Qdrant store → Qdrant service: queries
store/qdrant.py - File watcher → Log parser: uses
utils/watcher.py - Log parser → Parsed logs: creates records
utils/log_parser.py - Log window → Parsed logs: stores logs
detector/window.py - Model training → Log window: builds windows
detector/trainer.py - Model training → Log parser: parses files
detector/trainer.py - ML inference → Model training: uses features
detector/ml_model.py - Detector scorer → Statistical detector: uses
detector/scorer.py - Detector scorer → ML inference: uses
detector/scorer.py - User → HTTP API: requests
- HTTP API → API routes: routes
- API routes → Log handler: dispatches
- API routes → Rule handler: dispatches
- API routes → Alert handler: dispatches
- Alert handler → Alert dispatcher: dispatches
- Alert dispatcher → Email alerts: sends email
- Alert dispatcher → Slack alerts: sends Slack