sed including ± symbol

Alkass · 04-21-2010, 11:13 AM

Hi again!

I have a file with lines like

268.903± 1.893(ab)
270.01± 2.6(fb)

what I want to do is

1-sed and if (ab) is found, replace (ab) AND ± with E10-6
2-sed and if (fb) is found replace (fb) AND ± with E10-3

so output looks like

268.903E10-6 1.893E10-6
270.01E10-3 2.6E10-3

I have problem replacing the ± symbol correctly....

Thanks in advance!
A

colucix · 04-21-2010, 02:30 PM

What have you tried so far? GNU sed can replace hexadecimal values, so that the problem could be reduced to identify the correct hex sequence to match the ± symbol (od or hexdump might come in handy).

pixellany · 04-21-2010, 03:22 PM

According to hexdump, the symbol is 0xc2 + 0xb1. I cannot find it in the unicode tables,

I found various hints that Gnu SED will recognize unicode, but no concrete examples.

Regardless, this works: (I put the 2 sample lines into a file named "hex")

Code:

[mherring@mystical play]$ more hex
268.903± 1.893(ab)
270.01± 2.6(fb)
[mherring@mystical play]$ sed 's/\xc2\xb1/here/' hex
268.903here 1.893(ab)
270.01here 2.6(fb)

Dan04 · 04-22-2010, 01:38 AM

Quote:

Originally Posted by pixellany

According to hexdump, the symbol is 0xc2 + 0xb1. I cannot find it in the unicode tables

C2 B1 is the UTF-8 encoding of the character U+00B1, which is indeed the ± sign.-

colucix · 04-22-2010, 02:59 AM

Quote:

Originally Posted by Dan04

C2 B1 is the UTF-8 encoding of the character U+00B1, which is indeed the ± sign.-

Good to know. I've just found a complete UTF-8 character set online, here.

An off-topic note: Gnome terminal (and maybe many others) supports UTF-8 typing: we have to press simultaneously Ctrl-Shft, type u followed by the Unicode number, release Ctrl-Shift and the UTF-8 character appears. Even the LQ Reply box in Firefox supports it... ± ♥