regex examples (negate regex)

[darked89]

Hello,

I am trying to check if some CSV file contains bad values. The minimal example test.csv:

123
123.0
123.0,456
123.4567890
123.4567890,bad_value
123.0\tany_not_a_number_coma_or_dot

doing:
rg -e '[0-9]|,|\.' test.csv

as expected shows all 0-9 numbers, dots and commas. But I am stuck trying to get either these two lines:

 123.4567890,bad_value
123.0\tany_not_a_number_coma_or_dot

or (better):

bad_value
\tany_not_a_number_coma_or_dot

Many thanks for your help.

Darek Kedra

[BurntSushi]

rg -e '[0-9]|,|\.' test.csv | rg -v -e bad_value -e not_a_number

Like that? Honestly, I'm having a lot of difficulty parsing your question. Could you please show the entire output that you want? Otherwise I'm just stuck trying to guess what you mean.

[darked89]

I am sorry for not being clear enough.
The goal of the above ripgrep-ing:

Find any fishy chars, spaces including empty lines in a 2k columns and 20 up to 40k lines CSV files, or rather subset of these columns from #3 to #2k.
I use your xsv to exclude the first two columns.

The OK output:
whole lines in which there is at least one not(a_number or coma or dot).
I guess I should be able to turn on and off line numbering as needed as well as with -v just get the lines which are not containing any offending chars.

Before I was thinking about getting just the not(a_number or coma or dot) for each line, but this can get messy/hard to comprehend if there are multiple bad_chars per line.

In the end I will recheck the sanity of the input with polars.

Is my question clear enough by now?

Thank you.

Darek Kedra

[BurntSushi]

So it sounds like you want ripgrep to print out any line that contains a character that isn't a number, dot or comma? If so, then the strategy I would use is to write a regex that only matches those things and then invert it. So, rg -v '^[0-9,.]+$'.

[darked89]

This works perfect for me. Thank you and sorry for the long delay. Got swamped with stuff.

For some reason I am unable to mark this as an answer to the question.

DK