print multiple patterns with sed

Question

I try to print multiple patterns with sed.

Here's a typical string to process :

(<span class="arabic">1</span>.<span class="arabic">15</span>)</td></tr>

and I would like : (1.15)

For this, I tried :

sed 's/^(<span.*">\([0-9]*\).*\([0-9]*\).*">/(\1\.\2)/'

but I get (1.)15</span>)</td></tr>

Anyone could see what's wrong ?

Thanks

HTML parsing with REGEX spotted ! Dude, don't do this using REGEX, use a proper HTML parser — Gilles Quénot, Dec 23 '14 at 18:35

score 1 · Answer 1 · edited May 23 '17 at 12:05

1

If you are Chuck Norris, use regex, brainfuck or assembly. If you're not, don't use regex to parse HTML, instead, use a tool that support xpath, like xmllint. In 2014, it's a solved problem :

xmllint --html --xpath '//span[@class="arabic"]/text()' file_or_URL

Check the famous RegEx match open tags except XHTML self-contained tags

xmllint comes from libxml2-utils package (for debian and derivatives)

edited May 23 '17 at 12:05

Community

1
1

answered Dec 23 '14 at 18:09

Gilles Quénot

173,512
41
224
223

1

you should add some cthulhu reference and bleeding fonts :) – DRC Dec 23 '14 at 18:12
1

You mean this ? http://stackoverflow.com/questions/1732348/regex-match-open-tags-except-xhtml-self-contained-tags DONE – Gilles Quénot Dec 23 '14 at 18:15
2

Actually Chuck Norris can parse HTML with regex – midori Dec 23 '14 at 18:20
1

Yes, Chuck Norris can, others shouldn't =) – Gilles Quénot Dec 23 '14 at 18:22

score 0 · Answer 2 · answered Dec 23 '14 at 18:05

0

If data is at the same place all the time, awk may be a simpler solution than sed:

awk -F"[<>]" '{print "("$3"."$7")"}' file
(1.15)

answered Dec 23 '14 at 18:05

Jotne

40,548
12
51
55

HTML can change, it's inpredictable in general – Gilles Quénot Dec 23 '14 at 18:10

score 0 · Accepted Answer · answered Dec 23 '14 at 23:49

0

Reason why you are getting "(1.)15) as your output"

sed 's/^(<span.*">\([0-9]*\).*\([0-9]*\).*">/(\1\.\2)/'
                                          ^^

the two characters "> needs to be placed before \([0-9]*\) since "> in your line is before the two digits (in this case). This way sed can find the pattern

The correct sed command

sed 's/^(<span.*">\([0-9]*\).*">\([0-9]*\).*/(\1.\2)/'
                              ^^

Correct Command line

echo '(<span class="arabic">1</span>.<span class="arabic">15</span>)</td></tr>'|sed 's/^(<span.*">\([0-9]*\).*">\([0-9]*\).*/(\1.\2)/'

results using the command line above

(1.15)

answered Dec 23 '14 at 23:49

repzero

8,254
2
18
40

Thanks for your help. does ".*' mean all characters and include also > and " characters ? in other words, is it including special characters ? – Dec 26 '14 at 17:08
.* means ALL characters including > and " to up to the pattern for example (.*">) means all characters up to the pattern "> – repzero Dec 26 '14 at 17:19
ok, so from I have understood, I have to indicate the characters preceding the pattern, i.e in my example, the characters "> ? – Dec 27 '14 at 10:03
You can if all your lines are the same you can type each character upto ">. but in this case, I use .*"> because I don't know what characters are before the pattern ">. However I do know that "> is a pattern found on every line. As a result I match all characters upto "> – repzero Dec 27 '14 at 11:19

Zombo · Answer 4 · 2014-12-23T18:28:34.580

-1

$ lynx -dump -nomargins file.htm
(1.15)

edited Dec 23 '14 at 18:28

answered Dec 23 '14 at 18:07

Zombo

1
62
391
407

Not that bad if the HTML contains only the string provided. If it's a real HTML file, this is not OK – Gilles Quénot Dec 23 '14 at 18:17
@sputnick ok, we get it, your answer is better. Sheesh. – Zombo Dec 23 '14 at 18:18

print multiple patterns with sed

4 Answers4