Index of multiple similar strings

[Milo Thurston]

I'm trying to read through a file like this:
http://www.genomics.ceh.ac.uk/~milo/example.html
In order to count the number of N tracts and locate their
positions. My code goes like this:

dust_seq = # file in url above
nums = 0
d.dust_seq.scan(/[N]+/) do |blah|
nums += 1
puts "Index #{d.dust_seq.index(blah.to_s)}"
done
puts "Num of Ns: #{nums}"

In the example, the index for the third of the N groups
is reported as the same as the first, as it's small enough
to fit within it. Is there any way around this?
Thanks.

 

[Robert Klemme]

I'm trying to read through a file like this:
http://www.genomics.ceh.ac.uk/~milo/example.html
In order to count the number of N tracts and locate their
positions. My code goes like this:

dust_seq = # file in url above
nums = 0
d.dust_seq.scan(/[N]+/) do |blah|
nums += 1
puts "Index #{d.dust_seq.index(blah.to_s)}"
done
puts "Num of Ns: #{nums}"

In the example, the index for the third of the N groups
is reported as the same as the first, as it's small enough
to fit within it. Is there any way around this?
Thanks.

dust_seq = # file in url above
nums = 0
dust_seq.scan(/N+/) do |blah|
nums += 1
puts "Index #{$`.length}"
end
puts "Num of Ns: #{nums}"

Kind regards

robert

 

[Milo Thurston]

puts "Index #{$`.length}"

Excellent, thanks.
In which book/manual is $` described? I've not seen it before.

 

[Carlos]

I'm trying to read through a file like this:
http://www.genomics.ceh.ac.uk/~milo/example.html
In order to count the number of N tracts and locate their
positions. My code goes like this:

dust_seq = # file in url above
nums = 0
d.dust_seq.scan(/[N]+/) do |blah|
nums += 1
puts "Index #{d.dust_seq.index(blah.to_s)}"
done
puts "Num of Ns: #{nums}"

In the example, the index for the third of the N groups
is reported as the same as the first, as it's small enough
to fit within it. Is there any way around this?

(not tested):

nums = 0
idx = 0

while idx = dust_seq.index /N+/, idx
nums += 1
puts "Index #{idx}"
idx = Regexp.last_match.end(0)+1
end
puts "Num of Ns: #{nums}"

 

[Milo Thurston]

while idx = dust_seq.index /N+/, idx

Thanks - the interpreter didn't like this line, though.
However, I got it working and it seems better than the $`
method, which caused some nasty memory hogging problems
(I now regret not compiling in an kernel OOM killer...).

 

[Robert Klemme]

Excellent, thanks.
In which book/manual is $` described? I've not seen it before.

It's in the Pickaxe (both versions) although not in the online version of
the first edition AFAIK. You can find about the other way in the Regexp
doc:
http://www.ruby-doc.org/docs/ProgrammingRuby/html/ref_c_regexp.html
http://www.ruby-doc.org/core/classes/Regexp.html

Kind regards

robert

 

[Robert Klemme]

Thanks - the interpreter didn't like this line, though.
However, I got it working and it seems better than the $`
method, which caused some nasty memory hogging problems

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
??? Care to explain?

robert

 

[Carlos]

Thanks - the interpreter didn't like this line, though.

You are right, it should have parens:

while idx = dust_seq.index(/N+/, idx)

Strange...

 

[ts]

R> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
R> ??? Care to explain?

You create a String object for each call, you don't have this problem with

$~.begin(0)


Guy Decoux

 

[Robert Klemme]

R> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
R> ??? Care to explain?

You create a String object for each call, you don't have this problem with

$~.begin(0)

True. I thought of $~ also, but oversaw this aspect - "$`.length" just
looked cuter. Thx.

robert

 

[Milo Thurston]

R> ??? Care to explain?
You create a String object for each call, you don't have this problem with
$~.begin(0)

That would explain it. Some of the strings I'm looking at are several MB
in size. I've been writing out the data to disk and flushing stdout, but
$` seemed to leave each complete sequence in memory, causing it to run
out rather rapidly.

 

[Robert Klemme]

That would explain it. Some of the strings I'm looking at are several MB
in size. I've been writing out the data to disk and flushing stdout, but
$` seemed to leave each complete sequence in memory, causing it to run
out rather rapidly.

Yes, that's the reason. I haven't though about this, but as you can see
each reference to $` creates a new string instance:

15:55:29 [robert]: ruby -e '"f".scan(/./) { 5.times{ puts $`.id } }'
134690392
134690368
134690344
134690320
134690296
15:55:58 [robert]:

Kind regards

robert