# Generating a report from an ugly log file

**URL:** <https://www.ruby-forum.com/t/generating-a-report-from-an-ugly-log-file/166560>\
**Category:** Ruby\
**Created:** [April 27, 2009, 12:36pm UTC](https://www.ruby-forum.com/t/generating-a-report-from-an-ugly-log-file/166560 "2009-04-27T12:36:07Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Srijayanth\_S](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/e95f7d/32.png) [@Srijayanth\_S](https://www.ruby-forum.com/u/Srijayanth_S)\
**Post date:** [April 27, 2009, 12:36pm UTC](https://www.ruby-forum.com/t/generating-a-report-from-an-ugly-log-file/166560/1 "2009-04-27T12:36:07Z")

</div>

Hello,

I have the following problem.

There exists a really ugly debug log file which has mysql query  
statements.  
Looking through the log makes it hard to understand esp while providing  
reports to bosses/mgrs etc. so I have decided to make it more “English”.  
I  
have over 40 queries, each that can be called in different ways, for  
instance:

1. select foo1, foo2 from foo where foo1=… etc
2. select foo1 from foo where foo1= etc

The basic parsing would be  
/select(._)\s+from\s+(._)(?:\s+where\s+(.\*)$)/ #  
its a bit more advanced, but its solved and hence not a problem

#captures[1] will be the table name

1. Each table/view is its own class with a to\_english method that does  
the  
translation
2. An abstract class that parses a line from the log file, extracts the  
table or view (in this case foo) from the match and generates an object  
using eval “#{table.capitalize}.new captures”
3. The to\_english method generates an English representation of the  
object  
based on logic that uses information from the various captures passed  
in.

The idea is to have a list of objects that I iterate over and do a  
#to\_english

My questions are:

1. Should I organize this differently and run away from eval?
2. Since there are over 40 odd classes, is there a better way to require  
them other than:

Dir.glob(“qry\*”).each { |r| require r }

Thank you,

Jayanth

---

<div class="post-metadata">

**Author:** ![Srijayanth\_S](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/e95f7d/32.png) [@Srijayanth\_S](https://www.ruby-forum.com/u/Srijayanth_S)\
**Post date:** [April 27, 2009, 7:10pm UTC](https://www.ruby-forum.com/t/generating-a-report-from-an-ugly-log-file/166560/2 "2009-04-27T19:10:46Z")

</div>

On Apr 27, 6:35 am, Srijayanth S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> instance:  
> translation  
> Do you really need this? It may be overkill if the only thing you will  
> do in each of these classes is a to\_english method. You could perhaps  
> use a singleton class that contains all the translations.

> 1. An abstract class that parses a line from the log file, extracts the  
> table or view (in this case foo) from the match and generates an object  
> using eval “#{table.capitalize}.new captures”

To avoid eval, use  
Kernel.const\_get(table.capitalize).send(:to\_english)  
where to\_english is a class method.

---

<div class="post-metadata">

**Author:** ![Srijayanth\_S](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/e95f7d/32.png) [@Srijayanth\_S](https://www.ruby-forum.com/u/Srijayanth_S)\
**Post date:** [April 27, 2009, 7:50pm UTC](https://www.ruby-forum.com/t/generating-a-report-from-an-ugly-log-file/166560/3 "2009-04-27T19:50:07Z")

</div>

Thanks.

Unfortunately, I plan to do more than just to\_english. The logs have  
information that is also dependent on the ordering of the mysql  
statements  
themselves, so I want to be able to map queries based on certain  
conditions  
etc, so no running away from it really, but the Kernel#const\_get#send is  
a  
nice option.

Jayanth