leveldb-keys.py

#!/usr/bin/env python
"""
USAGE
  leveldb-keys.py file

DESCRIPTION
  Output keys of an ldb or log file.

  Requires the folders.py file in same directory as this file;
  available at https://git.io/folders

  LOG file: fresh writes of LevelDB go here;
  this is *NOT* sorted by keys, rather by most-recent write at the end.

  LDB files: older writes of LevelDB are here;
  this is sorted by the keys according to a comparator function; this
  function may be alphanumeric, but could also be something custom whose
  internals are not visible directly from the database files.

EXAMPLES
  python leveldb-dump.py primary.ldb/000127.ldb
  python leveldb-dump.py primary.ldb/000125.log
"""

import sys

from folders import (
    BytesReader,
    LogReader,
    TableData,
    TableFooter,
    TableIndex,
    TableReader,
    filesize,
)


def main(filepath):
    if filepath.endswith(".ldb"):
        print_ldb_keys(filepath)
    elif filepath.endswith(".log"):
        print_log_keys(filepath)
    else:
        raise ValueError("Unsupported file extension.")


def print_ldb_keys(filepath):
    with open(filepath, "rb") as file:
        reader = BytesReader(file, filesize(filepath))
        footer = TableFooter(reader)
        for internal_key, handle in TableIndex(footer.index_handle, reader):
            for internal_key, value in TableData(handle, reader):
                print_bytes(internal_key.user_key)


def print_log_keys(filepath):
    with open(filepath, "rb") as file:
        for batch in LogReader(file, filesize(filepath)):
            for command, args in batch:
                if command != "put":
                    continue
                key, value = args
                print_bytes(key)


def print_bytes(bytestring):
    print(repr(bytestring)[2:-1])


if __name__ == "__main__":
    main(sys.argv[1])