使用Linux的csvfix对数值进行排序-将数字转换为字符串

Question

I'm using csvfix to sort a CSV file based on an integer (counter) value in the second column. However it seems that csvfix puts double quotes around all fields in the file, turning them to strings, before it performs the sort. The result is that the rows are sorted by the string value, such that "1000" comes before "2".

There is a command-line option -smq that is supposed to apply "smart quoting" but that's not helping me. If I use the command csvfix echo -smq file.csv , the output has no quotes around numerical fields, but when I pipe that into csvfix sort -f 2 file.csv , the file is written without quotes but still sorted in "string order". It makes no difference whether I include the -smq flag in the sort command or not.

Additionally I would like csvfix to ignore the first row of string headers. Csvfix issue tracking claims this is already implemented but I can only find the -ifn flag that seems to cut the header row out entirely.

These seem pretty basic pieces of functionality for this tool, so I'm probably missing something very simple. Hoping someone on here has used csvfix and figured out.

Answer 1

According to the on line documentation for csvfix , sort has a N option for numeric sorts:

csvfix sort -f 2:N file.csv 

Having said this, CSV isn't a particularly good format for text manipulation . If possible, you're much better off choosing DSV (delimiter separated values) such as Tab or Pipe separated, so that you can simply pipe the output to sort , which has ample capability to sort by field, using whatever collation method you need.