Convert large JSON / JSON Lines to CSV, Excel, SQLite, Parquet or Feather¶
jsonreshape converts between JSON, JSON Lines and tabular formats by streaming, so the file size is not limited by your RAM. Exports run at roughly 100–220 MB/s.
JSON ↔ JSON Lines¶
jsonreshape convert data.json data.jsonl # array -> one record per line
jsonreshape convert data.jsonl data.json # back to an array
JSON / JSON Lines → CSV¶
jsonreshape convert data.jsonl table.csv
jsonreshape convert data.json table.csv --columns "id,name,contact.address.city"
Nested fields are addressed with dotted paths. To turn a nested array into rows, use
--explode (see Flatten nested JSON).
→ Excel (XLSX)¶
jsonreshape has its own streaming XLSX writer, so even multi-gigabyte files export without running out of memory.
→ SQLite¶
The records become rows in one table — ready to query with SQL.
→ Parquet¶
Parquet is the columnar format used across data engineering (pandas, DuckDB, Spark).
→ Feather / Arrow IPC¶
jsonreshape convert data.json data.feather # or data.arrow: the same file
jsonreshape convert data.json data.arrow --feather-compression lz4
Feather (the Arrow IPC file format) loads straight into pandas (pd.read_feather) and Polars
(pl.read_ipc). It has the same columns and column types as the Parquet export; compression is
zstd (default), lz4 or none. It needs the same pip install "jsonreshape[parquet]".
In the desktop app¶
Open the file, optionally search / filter / sort to narrow it, then File → Export and choose the format. You can export the whole file or just the current selection.
Convert from CSV, YAML or XML¶
The reverse direction is import: