metaspn-io 0.1.3

Deterministic ingestion and normalization for MetaSPN

metaspn-io

metaspn-io is the ingestion and normalization layer for MetaSPN. It converts raw external records into canonical signal envelopes with deterministic IDs and ordering.

Quick Demo (5 lines)

python -m metaspn_io io ingest \
  --adapter social_jsonl_v1 \
  --source tests/fixtures/social \
  --date 2026-02-05 \
  --out /tmp/social-signals \
  --stats

v0.1 Adapters

• social_jsonl_v1 (MUST): browser-extension social JSONL (post_seen, profile_seen)
• outcomes_jsonl_v1 (SHOULD): manual outcomes JSONL (message_sent, reply_received, meeting_booked, revenue_event)
• solana_rpc_v1: token/platform JSONL (trade, holder_change, supply_change, liquidity_event, metadata_update, reward_update)
• pumpfun_v1 (experimental): pump.fun token JSONL (same canonical token event mapping)

Schema Mapping

Input adapter	Input type	Output payload
social_jsonl_v1	post_seen	SocialPostSeen
social_jsonl_v1	profile_seen	ProfileSnapshotSeen
outcomes_jsonl_v1	message_sent	MessageSent
outcomes_jsonl_v1	reply_received	ReplyReceived
outcomes_jsonl_v1	meeting_booked	MeetingBooked
outcomes_jsonl_v1	revenue_event	RevenueEvent
solana_rpc_v1 / pumpfun_v1	trade	TokenTradeSeen
solana_rpc_v1 / pumpfun_v1	holder_change	HolderChangeSeen
solana_rpc_v1 / pumpfun_v1	supply_change	SupplyChangeSeen
solana_rpc_v1 / pumpfun_v1	liquidity_event	LiquidityEventSeen
solana_rpc_v1 / pumpfun_v1	metadata_update	TokenMetadataUpdated
solana_rpc_v1 / pumpfun_v1	reward_update	RewardUpdated

Output Envelope

JSONL lines are emitted as canonical envelopes:

{
  "schema_version": "0.1",
  "signal_id": "s_4e9b5c8417d3af2ef9baf8d1",
  "timestamp": "2026-02-05T12:00:00Z",
  "source": "twitter",
  "payload_type": "SocialPostSeen",
  "payload": {
    "platform": "twitter",
    "author_handle": "alice",
    "post_url": "https://x.com/alice/status/1",
    "text": "hello world",
    "action": "seen"
  },
  "entity_refs": [
    {
      "kind": "platform_identifier",
      "platform": "twitter",
      "identifier": "alice"
    }
  ],
  "trace": {
    "ingested_at": "2026-02-06T00:00:00Z",
    "input_file": "raw/social/2026-02-05.jsonl",
    "input_line_number": 1,
    "adapter_name": "social_jsonl_v1",
    "adapter_version": "0.1",
    "raw_id": null,
    "original_timezone": "UTC"
  }
}

CLI

Primary command:

metaspn io ingest --adapter social_jsonl_v1 --source raw/social --out workspace/store/signals/2026-02-05.jsonl

Supported flags:

• --source file or directory
• --out output JSONL path or directory (with --date, writes <out>/<date>.jsonl)
• --store optional store root (writes to <store>/signals/YYYY-MM-DD.jsonl)
• --date one-day UTC ingest window (YYYY-MM-DD)
• --since ISO timestamp lower bound
• --until ISO timestamp upper bound
• --dry-run
• --stats
• --lenient

Demo orchestrator invocation:

metaspn io ingest --adapter social_jsonl_v1 --source raw/social --date 2026-02-05 --out workspace/store/signals

Default mode is strict: bad records are skipped and logged to workspace/logs/ingest_errors.jsonl unless overridden.

Determinism Rules

• Stable IDs via stable_signal_id(source, timestamp, key)
• Timestamps normalized to UTC
• Deterministic sort: timestamp, then canonical key
• JSON output uses sorted keys

Add A New Adapter (<50 lines)

from dataclasses import dataclass
from pathlib import Path
from metaspn_io.adapters.base import AdapterOptions

@dataclass
class MyAdapter:
    name: str = "my_adapter_v1"
    version: str = "0.1"

    def iter_signals(self, source_path: Path, options: AdapterOptions | None = None):
        for raw in iter_jsonl_records(source_path):
            if isinstance(raw, ParseIssue):
                self.issues.append(raw)
                continue
            signal = convert_to_signal(raw)
            yield signal

Register it in metaspn_io.adapters.default_registry().

Tests

python3 -m pytest -q

Publishing

publish.yml publishes to PyPI when you push a version tag:

git tag -a v0.1.3 -m "v0.1.3"
git push origin v0.1.3

Configure PyPI trusted publishing for this GitHub repository, then the workflow will upload dist/* automatically. Before tagging, ensure CI (.github/workflows/ci.yml) is green, which validates python3 -m pytest -q and package build artifacts.