Prepare import batches
Choose one UTF-8 CSV or TSV file, set a split rule and inspect the generated parts.
Choose a file or load the example, then split.
UTF-8 only. Input and total output each allow 128 MiB; one record 1 MiB; 1,000 columns; 256 files. Output Blobs and ZIP need additional memory.
No generated files yet.
The file is read and processed in this browser; no data is uploaded.
Choose a file or load the example, then split.
An import form may accept your columns yet reject the number of records. The CSV splitter makes smaller batches without opening the source in a spreadsheet or retyping column names.
A physical line is not always a table record: quoted cells can contain line breaks. Splitting follows parsed records, so an address or multiline note stays with the rest of its row.
The CSV splitter repeats the first record as a header when that option is enabled. A limit of 1,000 means 1,000 data records, plus the header; the final batch may contain fewer records.
Split CSV by column to collect equal values even when their rows are scattered through the source. A region or status determines the group, while every original column remains in each output.
Select the actual delimiter and decide whether the first record names the columns. UTF-8 with an initial BOM is accepted. Preview a replacement before applying it; undo restores the previous source, options and results.
Split CSV by rows for a record limit, or split CSV by file size for a byte target. The default is 1,000 records; size mode starts at 1 MiB. Column positions start at 1, including when headers repeat.
Press Split or Ctrl/Cmd + Enter. A setting change requires another run and disables stale downloads. Check counts and warnings before downloading a selected CSV or the complete ZIP.
Quoted commas and line breaks remain inside their fields. The header does not consume the per-file record allowance. A header-only source produces one file with zero data records.
The target includes UTF-8 CSV bytes and repeated headers. A record that cannot fit with its header goes into a separate oversized part with a warning, without being cut.
Matching preserves case, spaces and leading zeros: 001 and 1 are different groups. Empty values can form a group. Names such as part-001.csv keep group text out of file paths.
Each part shows its first 5 data records, first 10 columns and at most 200 characters per field. Downloads retain all fields and records. The ZIP uses STORE packaging without compression.
Yes, when the header option is enabled. The CSV splitter repeats the first record in each part. If the source has no header, disable the option so the first record remains ordinary data.
All records with exactly the same value at the chosen position go to one file. Other columns stay intact. This groups full rows; it does not break a cell into more columns.
Yes. Download the ZIP to collect every generated CSV, or choose one part for an individual download. STORE packaging adds archive overhead and does not compress the CSV contents.
Properly quoted cells are parsed as a single field, including embedded commas, escaped quotes and line breaks. Malformed quoting causes an error instead of a silent split in the wrong place.
The source may contain up to 128 MiB of raw bytes, with a separate 128 MiB total-output budget. A single record, column count or number of parts can reach its own limit earlier. Browser memory may also be exhausted.
Found an unexpected split?
Use the CSV splitter with the example to compare record, size and value grouping, then select your own file and check the resulting counts.