Re: Re: classfiltercsv: filtering only works on bundle level

Xander Cage <[email protected]> writes:

      i forgot to mention that the above shows hat the "filter
      classes" in the classesfiltercsv function not "filtering"
      when used in various child bundles if the definition of the
      datacontainer or the call to the classfiltercsv function is
      not present in the bundle.

Reading the policy ..., you are using `classfiltercsv()' inside of
`bundle agent parent_bundle'. Running with your data set I get no data
container defined, for example:

,----
| # cf-agent --no-lock --log-level info --show-evaluated-vars=parent_bundle --file /tmp/example.cf
| R: check_user reached: $(user)
| R: create_users reached: $(user)
| R: parent bundle: $(user)
| Variable name                            Variable value                                               Meta tags                                Comment                                 
| default:parent_bundle.data_file          /tmp/my.csv                                                  source=promise                                                                   
| default:parent_bundle.keys                                                                            source=promise                                                                   
| default:parent_bundle.keys_unsorted                                                                   source=promise                                                                   
| default:parent_bundle.my_data            []                                                           source=promise                                                                   
`----


This is because none of the class expressions in your first (zeroth)
column evaluate to true on my system in that bundle. So, I added a row
that would (`any,nickanderson,Nick Anderson,neither-us-nor-them,ssh-bs')
and I modified the policy to point to the path I am using for the CSV
(`/tmp/my.csv')

,----
| ClassExpression,uid,name,group,ssh_key
| weirdos,mscott,Michael Scott,pathetic,ssh-someSupERsecretKey287483438dsdjhdsdsdjahhsgdsg
| weirdos,jkras,John Krasinski,pathetic,ssh-someSupERsecretKey287483438dsdjhdsdsdjahhsgdsg
| twats,tflen,Toby Flenderson,tedious,ssh-someSupERsecretKey287483438dsdjhdsdsdjahhsgdsg
| zombie,wimm,,dbaldap,ssh-someSupERsecretKey287483438dsdjhdsdsdjahhsgdsg
| any,nickanderson,Nick Anderson,neither-us-nor-them,ssh-bs
`----
Listing 1: CSV Data (/tmp/my.csv)

,----
| bundle agent parent_bundle
| {
| 
|   vars:
| 
|       "data_file" string => "/tmp/my.csv";
|       "my_data" data => classfiltercsv( $(data_file), "true", 0);
| 
|       "keys_unsorted" slist => getindices(my_data);
| 
|       "keys" slist => sort(keys_unsorted, "lex");
|       "user" string => "$(my_data[$(keys)][uid])";
| 
| 
| 
|   methods:
| 
|       "call_check_user"     usebundle => check_user( @(my_data) ),
|         handle => "remove_it";
| 
|       "call_create_user"   usebundle => create_users_new( @(my_data) ),
|         depends_on => { "remove_it" };
| 
|   reports:
| 
|       "parent bundle: $(user)";
| 
| }
| 
| 
| bundle agent check_user (info) {
| 
|   vars:
| 
|       "keys_unsorted" slist => getindices(info);
|       "keys" slist => sort(keys_unsorted, "lex");
|       "user" string => "$(info[$(keys)][uid])";
| 
|   classes:
| 
|       "zombie";
| 
| 
|   reports:
| 
|       "check_user reached: $(user)";
| 
| 
| }
| 
| 
| bundle agent create_users_new(info) {
| 
| 
|   vars:
|       "keys_unsorted" slist => getindices(info);
| 
|       "keys" slist => sort(keys_unsorted, "lex");
| 
|       "user" string => "$(info[$(keys)][uid])";
|       "group" string => "$(info[$(keys)][group])";
|       "ssh_key" string => "$(info[$(keys)][ssh_key])";
|       "name" string => "$(info[$(keys)][name])";
| 
|   classes:
| 
|       "weirdos";
|       "twats";
| 
|   reports:
| 
|       "create_users reached: $(user)";
| 
| }
| 
| bundle agent __main__
| {
|   methods:
|       "parent_bundle";
| }
`----
Listing 2: Example Policy

With that change, the rows with class expressions that evaluate to true
become part of the data container that result from parsing.

,----
| # cf-agent --no-lock --log-level info --show-evaluated-vars=parent_bundle --file /tmp/example.cf
| R: check_user reached: nickanderson
| R: create_users reached: nickanderson
| R: parent bundle: nickanderson
| Variable name                            Variable value                                               Meta tags                                Comment                                 
| default:parent_bundle.data_file          /tmp/my.csv                                                  source=promise                                                                   
| default:parent_bundle.keys                {"0"}                                                       source=promise                                                                   
| default:parent_bundle.keys_unsorted       {"0"}                                                       source=promise                                                                   
| default:parent_bundle.my_data            [{"group":"neither-us-nor-them","name":"Nick Anderson","ssh_key":"ssh-bs","uid":"nickanderson"}] source=promise                                                                   
| default:parent_bundle.user               nickanderson                                                 source=promise                                                                   
`----

Reading the policy further ...., some bundles (`bundle agent
check_user', and `bundle agent create_user_new') are promised as
methods. Inside these bundles there are some bundle scope classes
defined and some variables are created by pulling the data off of the
previously defined data container that is passed in to the bundle as an
argument.

      whats the deal if i have to define all filtering classes in
      the parent bundle? seems useless to me. this function is at
      least somewhat inconvenient , or i am doing something
      completely wrong. any insights?

Yes, sorry to say, I think you have a completely different expectation
of what the behavior should be than what it provides.

To me, it looks like you expect `classfiltercsv()' to filter your data
set when you access the data set. That might be a nifty function in it's
own right, but that is not how `classfiltercsv()' behaves.
`classfiltercsv()' filters the data as it is read in based on the
context at that point in time.

So, if you want one data set of the rows that match /weirdos/ and
/twats/ and another data set consisting of the rows that match /zombie/
as I believe your example policy illustrates you would need to make 2
calls to `classfiltercsv()'.

-- 
You received this message because you are subscribed to the Google Groups "help-cfengine" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion on the web visit https://groups.google.com/d/msgid/help-cfengine/87cznesiqn.fsf%40northern.tech.

-- 
Nick Anderson | Doer of Things | (+1) 785-550-1767 | https://northern.tech

-- 
You received this message because you are subscribed to the Google Groups "help-cfengine" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion on the web visit https://groups.google.com/d/msgid/help-cfengine/87cznesiqn.fsf%40northern.tech.