Ruarí's thoughts

Often with posts like this, where the comments go over 10 pages it can start to be a real hassle to find that elusive remark you knew you had seen. Indeed it can even be hard to find your own comment to refer back to! My preferred solution will probably not surprise anyone who has read any of my earlier posts where I profess my love of utilities like cURL and Wget. Basically I just use these to pull the information into the terminal so I can process it there. It is fairly easy to whip up a little shell script in a handful of lines to grep through all of the comments pages quickly. Here is one I wrote some time back, in a couple minutes:

#!/bin/sh
gs_tempdir=$(mktemp -d -t grepsnap.XXXXXXXX)
cd $gs_tempdir
wget -q $2 
grep startidx * |\
sed -e "s,[html:a href,\n<a href,g" |\
grep startidx |\
grep -v Next |\
sort -u |\
wget -B $2 -qi- --force-html
grep -i $1 *
rm -r $gs_tempdir
%5B/code%5D
I have it saved as grepsnap (for "grep snapshot") in %5Bi%5D~/.local/bin%5B/i%5D (I added %5Bi%5D~/.local/bin%5B/i%5D to may $PATH some time back since I always install opera in %5Bi%5D~/.local%5B/i%5D via our new *nix install script).

I use it in much the same fashion as I would use grep, simply replacing the %5Bi%5DFILE%5B/i%5D for %5Bi%5DBLOG_POST_URL%5B/i%5D. For example when %5Burl=http://my.opera.com/arghwashier/%5Darghwashier%5B/url%5D %5Burl=http://my.opera.com/desktopteam/blog/show.dml/21937422?startidx=550#comment47132122%5Dasked about the reasoning for switching off Smooth Scrolling on *nix by default%5B/url%5D I knew I had already answered another user and provided them with the Opera Preference setting in case they wanted to switch it back. Since I knew the preference included "SmoothScrolling" I could use this as a key to find my old post without having to manually trawl through the earlier 11 pages of comments. Here is what I issued and what I got back:

%5Bcode%5D
$ grepsnap smoothscroll http://my.opera.com/desktopteam/blog/2010/11/17/new-and-improved
new-and-improved?startidx=400:opera:config#UserPrefs|SmoothScrolling</div]

Using Opera to hover the numbered comments links on the blog page I could quickly see in the status bar that startidx=400 is page 9. Sure enough I found my comment there. Of course once you realise that you you are not limited to searching for keywords on the rendered page, but that even the source is available, you can also do more interesting things. For example, all comments include the following HTML fragment:

<p class="comment-by"><b><a href="/username/">

As you can probably guess, username is a given user's username. These means I can easily use this as a base to find all the comments by one user. For example, say I wanted to find all the comments by Daniel Hendrycks within that blog post (given he has an eye for detail). I could do the following:

$ grepsnap "comment-by.*DanielHendrycks" http://my.opera.com/desktopteam/blog/2010/11/17/new-and-improved

And I would get back (as of right now):

new-and-improved:<p class="comment-by"><b><a href="/DanielHendrycks/">Daniel Hendrycks</a></b> <span class="comment-date"><a href="/desktopteam/blog/show.dml/21937422#comment47053932" class="permalink">#</a>  17. November 2010, 22:15</span></p>
new-and-improved?startidx=100:<p class="comment-by"><b><a href="/DanielHendrycks/">Daniel Hendrycks</a></b> <span class="comment-date"><a href="/desktopteam/blog/show.dml/21937422?startidx=100#comment47058482" class="permalink">#</a>  17. November 2010, 23:31</span></p>

What is even nicer is that a permalink follows, so with just a little help from sed and some more regex magic I can easily turn that into a list of links. Here is how I would generate a list of links to what fearphage has said in this snapshot thus far:

$ grepsnap "comment-by.*fearphage" http://my.opera.com/desktopteam/blog/2010/11/17/new-and-improved | sed -e "s,.*href=\"\(.*comment[0-9]*\).*,http://my.opera.com\1,"
http://my.opera.com/desktopteam/blog/show.dml/21937422#comment47053572
http://my.opera.com/desktopteam/blog/show.dml/21937422#comment47053922
http://my.opera.com/desktopteam/blog/show.dml/21937422#comment47054062
http://my.opera.com/desktopteam/blog/show.dml/21937422?startidx=100#comment47061952
http://my.opera.com/desktopteam/blog/show.dml/21937422?startidx=450#comment47112672
http://my.opera.com/desktopteam/blog/show.dml/21937422?startidx=450#comment47113192

Alternatively, maybe you just want to know how much someone has commented. Havaard and Rijk usually comment a fair bit, so lets compare how many they have each written up until now: Haavard:

$ grepsnap "comment-by.*haavard" http://my.opera.com/desktopteam/blog/2010/11/17/new-and-improved | wc -l
13

Rijk:

$ grepsnap "comment-by.*rijk" http://my.opera.com/desktopteam/blog/2010/11/17/new-and-improved | wc -l
32

Yep, I think it is safe to say none of the Opera staff are likely to catch up with Rijk in this blog post! Anyway, I guess that's it. There is certainly a lot more you could do but you get the idea. Also this was supposed to be a short blog post!