conversion to TSV without double doublequotes (")

[Cristian Ciupitu]

If I try to convert to TSV a CSV containing values with doublequotes ("), e.g. this file:

Software,Year
"The ""best"" software",2023

csvformat --out-tabs --out-no-doublequote fails with:

Error: need to escape, but no escapechar set

One liner for testing:

printf 'Software,Year\n"The ""best"" software",2023\n' | csvformat -T -B

[James McKinney]

You need to set -P:

  -P OUT_ESCAPECHAR, --out-escapechar OUT_ESCAPECHAR
                        Character used to escape the delimiter in the output CSV file if --quoting 3 ("Quote None") is specified and to escape the QUOTECHAR if --no-doublequote is specified.

Specifically: to escape the QUOTECHAR if --no-doublequote is specified.

Edit: Fixed typo 71210e6 as -P should mention --out-no-doublequote not --no-doublequote.

[Cristian Ciupitu]

But I don't want to escape double qoutes because it doesn't make sense for TSV.

printf 'Software,Year\n"The ""best"" software",2023\n' | csvformat -T -B -P 'X' outputs:

Software	Year
The X"bestX" software	2023

when the ideal output is:

Software	Year
The "best" software	2023

Just in case, -P '' gives me:

TypeError: "escapechar" must be a 1-character string

[James McKinney]

"need to escape, but no escapechar set" comes from Python's own csv module.

You can change the quote character to one that doesn't occur in the text.

e.g. csvformat -T -Q~ or csvformat -T -Q"🦀"

[Cristian Ciupitu]

This stuff ought to be in the manual.
By the way the csvformat -u 3 -U 3 -Q "" example is broken.

[James McKinney]

How so?

Doing:

csvformat -u3 examples/optional_quote_characters.csv

causes:

a,b,c
"""1""","""2""","""3"""

which looks horrible. Doing the documented incantation produces this instead:

a,b,c
"1","2","3"

[Cristian Ciupitu]

# csvformat -u 3 -U 3 -Q ""
No input file or piped data provided. Waiting for standard input:
TypeError: "quotechar" must be a 1-character string

[James McKinney]

You’re not providing any input, as communicated in the output

[Cristian Ciupitu]

$ printf 'Software,Year\n"The ""best"" software",2023\n' | csvformat -u 3 -U 3 -Q ""
TypeError: "quotechar" must be a 1-character string

[James McKinney]

What's your python --version? I get no error.

[James McKinney]

You can also run with --verbose (-v) for the traceback.

[James McKinney]

Aha: "Changed in version 3.11: An empty quotechar is not allowed." https://docs.python.org/3/library/csv.html#dialects-and-formatting-parameters I've updated the example to be -Q🐍

[Cristian Ciupitu]

For what it's worth I'm using:

• python3-3.11.2-1.fc37.x86_64
• python3-csvkit-1.0.7-3.fc37.noarch