added log parsing script

asynclogparser/README.md

@@ -0,0 +1,90 @@
+# Async Log Parser
+
+A Go tool to parse binary log files generated by `asyncloguploader`, decode MPLog protobuf messages, fetch feature schemas from the inference API, and output human-readable logs.
+
+## Features
+
+- **Deframes binary log files**: Removes 8-byte frame headers and padding, extracts 4-byte length-prefixed records
+- **Protobuf unmarshaling**: Parses MPLog protobuf messages from record bytes
+- **Schema fetching**: Automatically fetches feature schemas from the inference API (Custodian)
+- **Feature decoding**: Decodes proto-encoded entity features using the fetched schema
+- **Human-readable output**: Writes parsed and decoded data to a readable log file
+
+## Usage
+
+```bash
+go run ./cmd/asynclogparse <log-file-path> [output-file]
+```
+
+### Arguments
+
+- `log-file-path`: Path to the `.log` file generated by `asyncloguploader`
+- `output-file`: (Optional) Path to output file. Defaults to `<log-file>.parsed.log`
+
+### Environment Variables
+
+- `INFERENCE_HOST`: Inference API host (default: `http://custodian.prd.meesho.int`)
+- `INFERENCE_PATH`: Inference API path for schema endpoint (default: `/api/v1/custodian/mp-config-registry/get_feature_schema`)
+
+### Example
+
+```bash
+# Set environment variables (optional, defaults provided)
+export INFERENCE_HOST="http://custodian.prd.meesho.int"
+export INFERENCE_PATH="/api/v1/custodian/mp-config-registry/get_feature_schema"
+
+# Parse a log file
+go run ./cmd/asynclogparse search-ad-head-multitask-fieldaware-categorylevelscaleup_2026-02-03_17-10-42.log
+
+# Or specify output file
+go run ./cmd/asynclogparse input.log output.parsed.log
+```
+
+## Output Format
+
+The output file contains human-readable text with the following structure:
+
+```
+=== Record 1 ===
+User ID: 212079220
+Tracking ID: abc
+Model Config ID: search-ad-head-multitask-fieldaware-categorylevelscaleup
+Version: 4
+Format Type: 0
+Entities: 300
+Parent Entities: 8
+Encoded Features: 300
+
+--- Entity 1 ---
+Entity ID: 60889110
+Parent Entity: saree
+Features:
+  catalog_id: 60889110
+  user_id: 212079220
+  pcvr_score: 0.000302
+  score: 0.000302
+
+--- Entity 2 ---
+...
+```
+
+## Architecture
+
+### Log File Format
+
+The binary log files follow this format:
+- **Frame Header** (8 bytes): `capacity` (uint32 LE) + `validDataBytes` (uint32 LE)
+- **Frame Data**: Fixed-size buffer (capacity bytes), but only `validDataBytes` contain real data
+- **Records**: Inside valid data, records are 4-byte length-prefixed (uint32 LE) followed by record bytes
+
+### MPLog Protobuf Structure
+
+- `user_id` (string)
+- `tracking_id` (string)
+- `model_proxy_config_id` (string)
+- `version` (int32)
+- `format_type` (int32): 0 = proto, 1 = arrow, 2 = parquet
+- `entities` (repeated string)
+- `parent_entity` (repeated string)
+- `encoded_features` (repeated bytes): Per-entity encoded feature blobs
+