# Better way to accumulate totals

**URL:** <https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782>\
**Category:** Ruby\
**Created:** [October 30, 2007, 9:42pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782 "2007-10-30T21:42:07Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 9:42pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/1 "2007-10-30T21:42:07Z")

</div>

I am just learning RUBY and am very impressed, I one week I’ve built  
tree routines that will save my group 6 to 7 hours each week, and they  
run in 2 minutes. Pretty cool. I would appreciate a suggest on how to do  
some. I have a script that correctly parses a directory full of text  
file and extracts key data. The data lines look like this:

" category value"

I have 15 categories and the code below works. But is crude. I create 15  
global arrays and match each line “category” to the text trying in the  
case statement. Would hash or an object work better. Just looking for  
pointer of which way to direct my research. Thanks!

Below is part of a big loop that goes through each found line for each  
document, and is in its own method.

case top\_event[0].lstrip.rstrip  
when “Redo size:”

```
        $redo_cnt[0] += 1
        $redo_cnt[1] += top_event[1].to_f

when "Logical reads:"

        $log_read_cnt[0] += 1
        $log_read_cnt[1] += top_event[1].to_f
```

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 10:09pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/2 "2007-10-30T22:09:04Z")

</div>

On Oct 30, 2007, at 2:41 PM, Robert K. wrote:

> ```
> when "Logical reads:"
> 
> $log_read_cnt[0] += 1
> $log_read_cnt[1] += top_event[1].to_f
> 
> ```

count = Hash.new{|h,k| h.update k =\> Hash.new{|h,k| h.update k =\> 0}}

case top\_event[0].strip  
when /redo size:/i  
count[:redo][0] += 1  
count[:redo][1] += Float(top\_event[1])

when /logical reads:/i  
count[:read][0] += 1  
count[:read][1] += 1

…

require ‘yaml’

y count

is one approach.

cheers.

a @ [http://codeforpeople.com/](http://codeforpeople.com/)

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 10:42pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/3 "2007-10-30T22:42:23Z")

</div>

ara.t.howard wrote:

> count = Hash.new{|h,k| h.update k =\> Hash.new{|h,k| h.update k =\> 0}}

This works…

count = Hash.new{|h,k| h[k] = Hash.new{|h1,k1| h1[k1] = 0}}  
=\> {}  
count[:foo][1] += 1  
=\> 1  
count[:foo][1] += 1  
=\> 2  
count[:foo][1] += 1  
=\> 3

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 10:50pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/4 "2007-10-30T22:50:00Z")

</div>

On Oct 30, 2007, at 3:40 PM, Joel VanderWerf wrote:

> This works…
> 
> count = Hash.new{|h,k| h[k] = Hash.new{|h1,k1| h1[k1] = 0}}  
> =\> {}  
> count[:foo][1] += 1  
> =\> 1  
> count[:foo][1] += 1  
> =\> 2  
> count[:foo][1] += 1  
> =\> 3

oh yeah, of course 😉 i’m in the habbit of using ‘update’ from  
‘inject’ where is saves you having to return the hash itself…  
shorter is better always.

cheers.

a @ [http://codeforpeople.com/](http://codeforpeople.com/)

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 11:08pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/5 "2007-10-30T23:08:06Z")

</div>

ara.t.howard wrote:

> > =\> 2  
> > count[:foo][1] += 1  
> > =\> 3
> 
> oh yeah, of course 😉 i’m in the habbit of using ‘update’ from  
> ‘inject’ where is saves you having to return the hash itself… shorter  
> is better always.

…but only if returning the hash is the right thing to do. In this  
case, we want the “leaf” value, not the hash:

irb(main):034:0\> count = Hash.new{|h,k| h.update k =\> Hash.new{|hh,kk|  
hh.update kk =\> 0}}  
=\> {}  
irb(main):035:0\> count[:foo][1] += 1  
NoMethodError: undefined method `+’ for {1=\>{}, :foo=\>{}}:Hash  
from (irb):35  
irb(main):036:0\> count  
=\> {1=\>{}, :foo=\>{}}

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 30, 2007, 11:20pm UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/6 "2007-10-30T23:20:14Z")

</div>

Robert K. wrote:

> The data lines look like this:
> 
> " category value"
> 
> case top\_event[0].lstrip.rstrip  
> when “Redo size:”
> 
> ```
> $redo_cnt[0] += 1
> $redo_cnt[1] += top_event[1].to_f
> 
> when "Logical reads:"
> 
> $log_read_cnt[0] += 1
> $log_read_cnt[1] += top_event[1].to_f
> 
> ```

This might be easier to understand:

#Create a hash, so that when you use a  
#key that doesn’t exist, it creates the key,  
#assigns it the array [0,0], and returns the  
#array:  
categories = Hash.new() do |hash, key|  
hash[key] = [0,0]  
end

#Loop over each line in a file:  
File.foreach(“data.txt”) do |line|  
arr = line.split(":")  
cat = arr[0].strip  
val = Float(arr[1]) #causes an error if can’t convert

categories[cat][0] += 1  
categories[cat][1] += val  
end

p categories  
puts categories[“Make pie”][0]  
puts categories[“Redo size”][1]

Using this data:

Redo size: 1.1  
Logical reads: 2.1  
Redo size: 1.1  
Hello world: 3.1  
Make pie: 4.1  
Hello world: 3.1  
Make pie: 4.1  
Make pie: 4.1

this is the output:

{“Make pie”=\>[3, 12.3], “Hello world”=\>[2, 6.2], “Redo size”=\>[2, 2.2],  
“Logical reads”=\>[1, 2.1]}  
3  
2.2

---

<div class="post-metadata">

**Author:** ![robkuz](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/r/34f0e0/32.png) [@robkuz](https://www.ruby-forum.com/u/robkuz)\
**Post date:** [October 31, 2007, 10:17am UTC](https://www.ruby-forum.com/t/better-way-to-accumulate-totals/118782/7 "2007-10-31T10:17:08Z")

</div>

2007/10/30, Robert K. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid):

> I am just learning RUBY and am very impressed, I one week I’ve built tree routines that will save my group 6 to 7 hours each week, and they run in 2 minutes. Pretty cool. I would appreciate a suggest on how to do some. I have a script that correctly parses a directory full of text file and extracts key data. The data lines look like this:
> 
> " category value"
> 
> I have 15 categories and the code below works. But is crude. I create 15 global arrays and match each line “category” to the text trying in the case statement. Would hash or an object work better. Just looking for pointer of which way to direct my research. Thanks!

Here is a different approach. I start by creating a statistics object:

Statistics = Struct.new :count, :sum

stats = Hash.new {|h,k| h[k] = Statistics.new(0, 0)}

File.foreach “foo.dat” do |line|  
if /^\s+\b(.+?)\b\s+(\d+.\d+)/ =~ line  
s = stats[$1]  
s.count += 1  
s.sum += Float($2)  
end  
end

Note, you might have to tweak the regexp.

Kind regards

robert