Bashscripts.org

Hi

From an html page I must exctract some info (name and links name).

Page links example:




And I need this:



Thanks for your support!

BR

Usage: ./script.sh file-containing-links.txt
That should do what you want

Best regards
Fredrik Eriksson

Thank you Fredrik for your support

But output is...

./script.sh links.txt

wrong type of 'and' in the if statement, I have corrected it. I also added some extra quotes and braces to reduce the changes of having weird filename issues and empty string errors.



DW

oh yeah, i also corrected the missing '='

DW

thank you @DarthWavy, with your script the result is:




with sed '/_blank/d' I can remove "_blank"

Output result:



But how can I obtain this output?
(reverse name/link)

I'm sorry, i corrected the bash issues but I don't know sed very well, I tend to use perl for that type of thing. Fredrik can hopefully tweak that bit

Well my seds are just looking for specific formated lines.

This is an issue because the a href line spans over multiple lines.
One thing I didn't think of is that both url and name will be set at the first spin and it will echo because of it.

That script is flawed, the regexps are as correct as I can think, but the script isn't.

Sadly I'm a bit too busy atm to write a correction.

Best regards
Fredrik Eriksson

Hi!

How're you getting your data? Is it handed to you in a file, or can you choose how it's parsed from the web page? The problem with what we're working with here is that it's all split up on separate lines. If you could parse the page with something like this, it would grab each "a" tag individually (and dump into a file):



I'm intrigued by fredrik's regex in his sed syntax, and I'm going to have to poke at it and see how it works In the meantime, I typically use perl for something like this, because sed and awk do "greedy" matching, which makes it hard to parse between two tags (like <a and /a>) which is actually why I used "-P" in the grep statement above to use perl style regex. Here's something quick and dirty (the regex would need a lot of refinement for anything beyond a basic example... it even misses some things in my google example, but parses your output perfectly as long as the tags are all on one line):



Hope this helps?... I need to learn more about posix regex

-J

Jeo, I believe my regexps works in perl as well.

When cleaned up abit they look like this


[^>]+ is just an expression for everything up till a ">"-sign. It's not really necessery, but I like it more then .* since it's alot more exact as to where the stop is.

Best regards
Fredrik Eriksson

Yeah! figured that out last night after some playing. That is a cool one, thanks for the update!

Many thanks for your help

So:

@jeo:
lynx -source google.com|grep -Po \<a.*?\/a\> > dump.txt
Output:

I use Debian repository...


@fredrik.eriksson
With this code:



I obtain only some output errors...

Thank