Has anyone successfully used dicts in iRules?

Hi,

Why not creating a string class with all the extension you want to analyze and then do the following:

when HTTP_REQUEST {
if {[matchclass [HTTP::path] ends_with $::Extension ] } {
pool Cache-Pool
         } else {
                pool Application-Pool
         }
}

where Extension is the name of your class

You can create a class in the Data group page.

nmenant is on the right track here. There's no need to do a dictionary lookup, although that's an interesting (and sneaky!) way of getting around multiple comparisons.

 

 

In this case I'd recommend a single class style match, like listed above.

 

 

Colin

By Class do you guys mean a TCL class that's a subclass of string, or is this some F5/BigIP feature?

 

 

In either case, due to my inexperience with both, do you know of any resources that may be a good start for implementation?

Nevermind on that, it seems pretty simple. The DevWiki had some stuff on it.

class CacheableExtensions {
    .css
    .gif
    .ico
    .jpg
    .jpeg
    .js
}
when HTTP_REQUEST {
    if { [matchlass [HTTP::path] ends_with $::CacheableExtensions] } {
        pool Cache-Pool
    } else {
        pool Application-Pool
    }
}

Could it be that simple?

Would it be more appropriate to initialize my class within RULE_INIT so it doesnt have to be initialized and destroyed with each request?

Yeap, thanks. I kept reading the matchclass wiki page and found out about data groups. What mislead me to think it was to be implemented in code was the fact that it includes a snippet of what the data group ends up looking like in the bigip.conf file. Skimming the doc made me think I had to code it somewhere.

 

 

As to dict being implemented in 8.5, thats what I thought was the problem, but couldn't find anything that would definitively state the version of TCL used in the BigIp. Thanks for the info, and to everyone who commented, thanks for the help.

 

 

For those wondering, I ended up with two classes:

 

 

class CacheableExtensions {
    .css 
    .js
    .gif 
    *SNIP*
    .jpg 
    
}
class CacheableDirs {
    /user/docs/
    /user/images/
    /user/videos/
}
when HTTP_REQUEST {
    if { [matchclass [string tolower [HTTP::path]] starts_with $::CacheableDirectories] } {
        pool Cache-Pool
    } elseif { [matchclass [string tolower [HTTP::path]] ends_with $::CacheableExtensions] } {
        pool Cache-Pool
    } else {
        pool Application-Pool
    }
}

That's great, except sometimes you really want a dict.

 

Does anyone have an answer to the original question?

 

 

I'm trying to save the request headers during HTTP_REQUEST so I can conditionally log them in HTTP_RESPONSE.

 

My approach is to copy the headers to a dict and iterate over the dict in HTTP_RESPONSE as the request headers are no longer available by that point.

 

 

Dave

Actually - this may still be current:

 

 

From http://devcentral.f5.com/Tutorials/TechTips/tabid/63/articleType/ArticleView/articleId/342/iRules-101--15--TCL-List-Handling-Commands.aspx

 

...

 

TCL: List Handling Commands

 

 

Most of the TCL commands for use in manipulating lists are available in iRules:

 

list - Create a list

 

split - Split a string into a proper Tcl list

 

join - Create a string by joining together list elements

 

concat - Join lists together

 

llength - Count the number of elements in a list

 

lindex - Retrieve an element from a list

 

lrange - Return one or more adjacent elements from a list

 

linsert - Insert elements into a list

 

lreplace - Replace elements in a list with new elements

 

lsort - Sort the elements of a list

 

lsearch - See if a list contains a particular element

 

The remaining documented list handling commands, lrepeat, lreverse, & dict, are not available within iRules. We’ll cover the commands from above that get the most traffic here on the forums.

 

 

...

 

Hi Dave,

You could try saving the value of [HTTP::request] split on a \r in HTTP_REQUEST and then use TCL list commands to parse this in the response. I think this should be more efficient than looping through all of the request headers and saving them to a list. You'll also get the full request line from HTTP::request, which can't be retrieved whole in one command otherwise.

Can you try something like this:

when HTTP_REQUEST {
    Save the request headers split on a carriage return
   set request_headers [split [HTTP::request] \\\r]
}
when HTTP_RESPONSE {
   log local0. "request header count: [llength $request_headers]"

    Loop through the request headers
   for {set i 0} {$i < [llength $request_headers]} {incr i} {

       Trim off the leading \r\n from each list element and log it
      log local0. "$i: [string trimleft [lindex $request_headers $i]]"
   }
}

Aaron