Delimited data conversion

Convert CSV to TSV Without Breaking Quoted Fields

Convert CSV to TSV Without Breaking Quoted Fields. Preserve field values with explicit parsing, encoding and a checked output round trip.

By Uddhav Gopani, FounderPublished 27 September 2026Updated 29 September 2026

Convert CSV to TSV by parsing the CSV records and writing them with a tab delimiter. A global replacement of commas with tabs is unsafe because a quoted comma can be part of a field. Keep the original unchanged and confirm that a TSV parser recovers the same field values.

Define the destination’s TSV rules

The example uses Python’s CSV parser and writer with a tab delimiter. Its writer quotes a field when necessary, including a value containing a tab, quotation mark or line break. The receiving system must support those quoting rules. Some simple tab-separated importers do not, so check their contract before exporting complex values.

Use a header such as id, name and note. Store an identifier such as 00123 as text and include a name containing a comma. The comma belongs inside the name after conversion; it must not create an extra column. A tab in note needs its own round-trip check.

Verify the parsed result

The script compares the parsed TSV with the original parsed CSV, including field order and empty strings. This catches a lost leading zero, an unescaped tab or a malformed row. The number of physical lines is not a reliable record count when quoted fields contain embedded newlines.

The example loads the records into memory to validate the complete small input before opening the destination. For a much larger file, use a streaming writer with a temporary output and a validation pass, then publish the result only after checks succeed. Do not silently truncate the input or skip malformed records.

Keep the extension and content consistent

Name the result .tsv when that is the receiver’s convention, but remember that the delimiter and parser rules define the format. Renaming a CSV file to .tsv without changing its record syntax is not this conversion. Open the output with an actual tab-aware parser before relying on a spreadsheet preview.

Executable conversion example

import csv

with open("input.csv", encoding="utf-8-sig", newline="") as source:
    rows = list(csv.reader(source, delimiter=",", strict=True))
if not rows or not rows[0]:
    raise ValueError("Expected a header row")
if any(len(row) != len(rows[0]) for row in rows):
    raise ValueError("Every record must have the same number of fields")
with open("output.tsv", "w", encoding="utf-8", newline="") as target:
    csv.writer(target, delimiter="	", lineterminator="\n").writerows(rows)
with open("output.tsv", encoding="utf-8", newline="") as check:
    assert list(csv.reader(check, delimiter="	", strict=True)) == rows

Use a representative acceptance fixture

Begin with two different data records containing a leading-zero identifier, an embedded comma, a quotation mark, an empty field and a non-ASCII name when the destination supports it. Read the output with a parser for the destination format and compare the actual values, column order and record count. A save message or familiar extension does not establish that the records survived correctly.

Change the second fixture’s values rather than rerunning only the same sample. Include a malformed record and confirm that the conversion reports it instead of producing a misleading success. Keep the original unchanged and write the result to a separate path. Avoid exposing sensitive source values in public error logs.

Choose the format-aware operation

The Convert Text to CSV tool here works with already extracted plain text. It does not infer every structured file format or replace the Python parsers shown above. Use the operation whose input contract matches the actual source, and verify the result independently before importing it into another application.

Python reference

Python’s CSV module documentation explains the parser and writer dialect options. The published snippets and acceptance cases in this family are executed locally against disposable fixtures.

Distinguish encoding markers from values

The input reader accepts an optional UTF-8 byte-order mark, while the output writer uses UTF-8 without adding one. The output bytes can therefore differ in more than the delimiter while preserving every parsed value. Check the receiver’s encoding requirement separately from the field round trip.