SDSavior

sdsavior

Most likely i have re-invented the wheel here. But i was to lazy to find something like this. So, here you go…

A crash-recoverable, memory-mapped ring buffer for JSON records, designed for SD-card-ish environments (where constant small writes are a fun way to speedrun media death).

Install

pip install sdsavior

Quick Start

from sdsavior import SDSavior

rb = SDSavior(
    data_path="data.ring",
    meta_path="data.meta",
    capacity_bytes=8 * 1024 * 1024,
)

rb.open()
rb.append({"hello": "pi"})
rb.append({"n": 123})

for seq, ts_ns, obj in rb.iter_records():
    print(seq, ts_ns, obj)

rb.close()

Or with a context manager:

from sdsavior import SDSavior

with SDSavior("data.ring", "data.meta", 8 * 1024 * 1024) as rb:
    rb.append({"ok": True})

CLI

Export current contents to a JSONL

sdsavior export --data data.ring --meta data.meta --capacity 8388608 --out out.jsonl

IMPORTANT NOTES

• Records are stored as JSON lines (\\n appended).
• Metadata is double-buffered with CRC and a commit counter for crash recovery.
• Already-encoded JSON can be written with append_json_bytes(...) to skip a decode/encode round trip.
• group_commit_records=N can batch fsync work while keeping CRC-based crash recovery.
• By default it does not fsync data pages on every append (to reduce wear); metadata is fsync’d.

TODO

• fix ci for pypy publishing…
• smarter fsync strategy
• indexing / partial recovery scans

This site is open source. Improve this page.