Publish / file · All operations

Write binary objects or images

A variant of Write another file format

Write the binary content or image held in each row to its own file.

Status

In arcform

Runs as a SQL step

arcform carries it as a sql: step today.

The DuckDB step

Write the binary content of each row to its own file

SQL
CREATE TEMP TABLE blobs AS SELECT * FROM (VALUES ('one', 'first file'::BLOB), ('two', 'second file'::BLOB), ('three', '\x00\x01\xff'::BLOB)) t(name, content);
COPY (SELECT name, content FROM blobs) TO '<your_scratch_dir>/blobs' (FORMAT blob, PARTITION_BY (name));
CREATE TEMP TABLE result AS
SELECT regexp_extract(filename, 'name=([^/]+)/', 1) AS name, content FROM read_blob('<your_scratch_dir>/blobs/*/*');
SELECT name, octet_length(content) AS size FROM result ORDER BY name;
SELECT CASE WHEN ((SELECT count(*) = 3 FROM result) AND (SELECT count(*) FROM result r JOIN blobs b USING (name) WHERE r.content = b.content) = 3) THEN 'as expected' ELSE error('the three rows were not written to three files holding exactly their bytes') END AS verdict;

Replace <your_scratch_dir> with a directory of your own that you can write to before running the statement.

  • It reads this from <your_scratch_dir>, which you supply:
    • blobs: An empty directory.
  • It needs nothing installed beyond DuckDB itself.
  • The statement checks its own result and raises an error when it is wrong: “the three rows were not written to three files holding exactly their bytes”.
  • It passed on DuckDB v2.0.0 (Cyanoptera) d54cbbe3df.

What other tools call it

Meridian's classification; an independent blind rating agreed on 95.7% of 324 rows (variant, version 3).

Alteryx Designer (2)
  • Blob Outputread 2026-09-27

    Write each record's binary object to its own file

  • Image Outputread 2026-09-27

    Save image files to a chosen location

Apache Hop (1)
  • Binary File Outputread 2026-09-27

    Write a binary field to a file whose name comes from each row

Azure Data Factory (1)
KNIME Analytics Platform (2)
RapidMiner Studio (2)
  • Write Fileread 2026-09-27

    Write a file object to disk or to the repository as a blob

  • Write Imageread 2026-09-27

    Write an image to disk

VisiData (1)
  • png (saver)read 2026-09-23

    Save as PNG image (from a PNG source)

Background

In Apache Hop, the Binary File Output transform writes raw binary content from a field to a file.[1] Apache Hop's Binary File Output reads the target file name from an input field, so every row can be written to a different file.[1] In Alteryx, the Blob Output tool writes each record into its own file, taking the bytes from a field of binary large object (BLOB) data.[2]

When to use it

Apache Hop's Binary File Output is for a byte-for-byte write of BLOB or binary data, such as exporting a database BLOB column to individual files.[1] In Apache Hop, Text File Output applies text encoding and formatting and can corrupt raw bytes, so Binary File Output is preferred for binary data.[1] In Apache Hop's Binary File Output, null binary content fails the row, while an empty binary value of 0 bytes writes an empty file.[1]

In Azure Data Factory, a Binary dataset's file content is not parsed but treated as-is, and a copy activity can only copy from a Binary dataset to a Binary dataset.[3] KNIME's Image Writer (Table) node can fail, overwrite or ignore when an output file already exists.[4] Build file names from a column that is unique for each row, and choose what happens when a file already exists rather than leave it to the default. (Meridian's view)

The method

In Apache Hop's Binary File Output, no character encoding is applied: bytes are written exactly as stored in the field.[1] In DuckDB, the BLOB format of COPY ... TO writes a single column, cast to the BLOB data type, into a .blob file.[5] In DuckDB, the BLOB format's COMPRESSION option can be none, gzip or zstd, and by default is detected from the file extension.[5]

KNIME's Image Writer (Table) node writes each image in a column as a separate file in a directory, taking file names from a column or generating them from a pattern with a counter.[4] KNIME's Image Writer (Table) node determines the file extension automatically, so it should not be included in the file names.[4] In Alteryx's Blob Output tool, a field's value can be appended or prepended to the file name, replace the file name, or replace the entire path.[2]

Sources

  1. [1]Binary File Output | Apache Hopdocumentation
  2. [2]Blob Output Tooldocumentation
  3. [3]Binary format - Azure Data Factory & Azure Synapse | Microsoft Learndocumentation
  4. [4]Image Writer (Table)documentation
  5. [5]COPY Statement – DuckDBdocumentation