Fixes #21634 - Allows modification of output fields #275

ofedoren · 2018-05-31T14:14:48Z

No description provided.

tstrachota · 2018-06-01T12:46:46Z

Thanks @ofedoren! It looks good at a first glance. Would you mind adding some docs with examples how to change a commands output from a plugin? It would make testing easier plus we can refer to them in future.

ofedoren · 2018-06-04T14:42:09Z

@tstrachota updated. I did some changes comparing to the last version of this commit and added some docs.

tstrachota

Thank you for adding the docs. It was much easier to understand now.

Some changes are needed, I mentioned them all in the code.

There are some features that I'm missing, quite important ones in my opinion:

finding fields, to make a modification on them. Eg.:

d.find_fields(:id => :name)[0].description = _('Some better description')

inserting fields at a given position. Eg.:

d.insert_fields(:after_id => :operating_system) do # or before_id
  label _('Ansible') do
    # ...
  end
end

tstrachota · 2018-06-18T16:10:20Z

lib/hammer_cli/output/definition.rb

+    def at(path = [])
+      return self if path.empty?
+      field = expand_path(@fields, path.dup)
+      return nil if field.nil? || !field.respond_to?(:output_definition)


This should raise an error rather than return nil. In 99% of cases at will be directly followed by calling some other methods on the output definition. A meaningful error will make debugging much easier compared to "Undefined method 'append' for nil:NilClass".

tstrachota · 2018-06-19T16:04:30Z

lib/hammer_cli/output/definition.rb

+      end
+    end
+
+    def select_fields!(by = {})


I'm not sure if there's some real-world use-case for this method. I can't imagine that somebody would want to keep just some fields and remove the rest.

It also doesn't work as expected. When I tested it on host info

HammerCLIForeman::Host::InfoCommand.output_definition.select_fields!(:labels => ['Id'], :deep_select => true)

the info command then did contain only two "Id" fields, but none of them displayed any data.

I think it's fine to simple remove this method.

tstrachota · 2018-06-19T16:14:14Z

lib/hammer_cli/output/definition.rb

+      end
+    end
+
+    def replace_field(field_id, fields = nil, &block)


This method is handy! I like using the block.
Does it also accept a single field? Using is as replace_field(:name, Field.new(:key => :surname)) would be nice too.

tstrachota · 2018-06-19T16:22:21Z

lib/hammer_cli/output/definition.rb

    def empty?
      @fields.empty?
    end

+    private
+
+    def build_definition(fields = nil, &block)


We're using similar code at 2 places already. I think we could refactor it into a definition's public method:

def build(dsl = Dsl.new, &block) dsl.build(&block) if block_given? append(dsl.fields) end

and also if we changed append to:

def append(fields) unless fields.nil? fields = [fields] unless fields.is_a? Array @fields += fields end end

than it would be much easier for use at all the places and also decoupled from the dsl definition.

This method could then be just:

definition = self.class.new definition.append(fields) definition.build(&block) definition

tstrachota · 2018-06-19T16:23:46Z

lib/hammer_cli/output/definition.rb

+      multiple_replace(i, definition.fields)
+    end
+
+    def at(path = [])


I like that method. Could the path also accept a single symbol? Eg at(:interfaces)

tstrachota · 2018-06-19T16:41:54Z

doc/commands_modification.md

+  deep_select: true/false,
+  deep_delete: true/false,
+  keys: [comma separated :field_key or/and :field_id],
+  labels: [comma separated 'field_label']


I guess you could write just "array"

tstrachota · 2018-06-19T16:55:34Z

lib/hammer_cli/output/definition.rb

@@ -12,10 +12,117 @@ def append(fields)
      @fields += fields
    end

+    def delete_fields!(by = {})
+      return clear if by.empty?
+      return deep_delete!(@fields, by) if by[:deep_delete]


Use just by[:deep] please. "delete" feels a bit redundant there.

ofedoren · 2018-06-20T17:56:20Z

@tstrachota, updated.

If there is no need for select_fields! method, is there need for delete_fields! method then? I think they are very similar, so I removed it as well. I guess if there will be need to delete a field, it can be done via d.at(path).insert(:replace, field_id_to_delete) do; end

Also now it's possible to insert before/after or replace a field via one single method (see updated docs). I guess it's a good way to avoid code repetition, which in this case will be x3.

tstrachota

Thanks for you updates @ofedoren. Functionally-wise this is ok. I believe the code can be simplified at several places.

Would you mind describe also append in you docs? I know it's already been there but it's key functionality. It took me some time to find out how to append fields after reading the docs (I nearly reported it as a missing feature ;-)).

tstrachota · 2018-09-19T15:21:48Z

test/unit/output/definition_test.rb

+  end
+
+  describe 'find_field' do
+    # -


Missing tests here?

tstrachota · 2018-09-19T15:23:52Z

doc/commands_modification.md

+# Expand a field with new definition
+HammerCLIForeman::Host::InfoCommand.output_definition
+                                   .at(['some', :path])
+                                   .insert(:replace, :not_container_field_id, [Fields::Field.new(:first => :one)]) do


Nitpick: Examples will be more readable if you split the two ways of using insert:

with fields as an argument

with dsl in a block

It's not very likely that both ways would be used in combination.

tstrachota · 2018-09-19T16:01:40Z

doc/commands_modification.md

+                                   end
+# Simplified version
+HammerCLIForeman::Host::InfoCommand.output_definition
+                                   .insert(:before, :field_id, Fields::Field.new(:first => :one))


The docs deserve correct parameters for Fields::Field.new

tstrachota · 2018-09-19T16:12:22Z

lib/hammer_cli/output/definition.rb

+      return self if path.empty?
+      field = expand_path(@fields, path.dup)
+      if field.nil? || !field.respond_to?(:output_definition)
+        raise ArgumentError, 'No output definition at path'


It would be very helpful if the error could mention what path was used

tstrachota · 2018-09-19T16:13:51Z

lib/hammer_cli/output/definition.rb

+      index = @fields.find_index do |f|
+        f.id == field_id
+      end
+      raise NameError, 'No such field' if index.nil?


Ditto, mentioning the name will make debugging a lot easier.

tstrachota · 2018-09-19T16:17:51Z

lib/hammer_cli/output/definition.rb

+    def at(path = [])
+      path = [path] unless path.is_a? Array
+      return self if path.empty?
+      field = expand_path(@fields, path.dup)


If path.dup is there to avoid destructive effect of expand_path, move it inside that method, please.

tstrachota · 2018-09-20T09:54:33Z

lib/hammer_cli/output/definition.rb

+        raise ArgumentError, 'No output definition at path'
+      end
+      field.output_definition
+    end


Btw when I'm looking at the method... I think that you can easily drop expand_path and re-use find_field and at recursively. This code should give you the same result:

def at(path = []) path = [path] unless path.is_a? Array return self if path.empty? field = find_field(path[0]) if field.nil? raise ArgumentError, "Field #{path[0]} not found" end if !field.respond_to?(:output_definition) raise ArgumentError, "Field #{path[0]} doesn't have nested output definition" end field.output_definition.at(path[1..-1]) end

tstrachota · 2018-09-20T09:58:09Z

lib/hammer_cli/output/definition.rb

+      dsl.build(&block) if block_given?
+      append(dsl.fields)
+    end
+
    def append(fields)


Could you please add support for passing &block with dsl here too? I believe that this method is going to be used most so the more convenient it will be the better. It will also bring more consistency with insert.

Note: it would also make code in insert simpler:

definition.append(fields, &block)

tstrachota · 2018-09-20T10:09:55Z

lib/hammer_cli/output/definition.rb

+      @fields[field_index(field_id)]
+    end
+
+    def insert(mode = :after, field_id = nil, fields = nil, &block)


How about changing the signature to def insert(mode, field_id, fields = nil, &block) ?
Without field_id the method will fail anyway so both parameters are required.

tstrachota · 2018-09-20T10:13:18Z

lib/hammer_cli/output/definition.rb

+      index += 1 if mode == :after
+      definition = self.class.new
+      definition.append(fields)
+      definition.build(&block)


This seems too complicated to me. I guess the only reason why you created new definition is the fact that build automatically appends the fields. How about moving injection of Dsl.new into constructor and using just @dsl.build here?

ofedoren · 2018-09-21T10:40:32Z

@tstrachota, updated. Thanks for suggestions for better method definitions!

Updated docs

Also, before it gets merged, I want to be sure if this is an acceptable way of identifying a field by default:

mbacovsky · 2018-10-17T09:13:22Z

lib/hammer_cli/output/fields.rb

@@ -4,14 +4,16 @@ module Fields

  class Field

-    attr_reader :label
-    attr_reader :path
+    attr_accessor :label


Do we want to enable the writers and allow change these form outside?

Since we're using labels to set field's ID, which is used to address a field, I don't think it's a good idea now :)

But again, is this an acceptable way? We could allow label changing from outside, but then we shouldn't identify fields by labels I guess..

Regarding setting the default for id I think it is okay. Maybe with 'key' preferred over the 'label'.
In the original comment I was referring more to all three attr_accessors in general. It does not feel correct to modify any of these attributes on existing field. Wouldn't it be better to replace filed with a new field?

I can imagine changing label in container or collection could be awkward, so ability to change label might be helpful in some cases. The :id would need to be a method getting the id dynamically.

mbacovsky · 2018-10-17T09:34:02Z

lib/hammer_cli/output/dsl.rb

@@ -20,6 +20,7 @@ def build(&block)
    def field(key, label, type=nil, options={}, &block)
      options[:path] = current_path.clone
      options[:path] << key if !key.nil?
+      options[:field_id] = options[:field_id] || key || options[:key]


This is inconsistent with the Field class defaults. If field is created directly without the :field_id the default id is 'label'. Via dsl the 'key' wins. What about not setting the default value in here and leave it up to Field class?

mbacovsky · 2018-10-17T09:35:31Z

lib/hammer_cli/output/fields.rb


    def initialize(options={})
      @hide_blank = options[:hide_blank].nil? ? false : options[:hide_blank]
      @hide_missing = options[:hide_missing].nil? ? true : options[:hide_missing]
      @path = options[:path] || []
      @label = options[:label]
+      @id = options[:field_id] || label || options[:key]


Is there a reason the option is called :field_id and the attribute just id? It is not consistent with other options.

mbacovsky · 2018-10-17T11:47:33Z

@ofedoren, I've left some comments and questions inline. Besides that it looks nice and powerful.

ofedoren · 2018-10-18T15:06:04Z

Updated with @mbacovsky suggestions.

mbacovsky · 2018-10-19T10:49:26Z

@ofedoren thanks for quick update!

I'm sorry I didn't noticed earlier but I played with the feature and wanted to found out how it behaves when the id is not found. Similarly as in help extension it raises error which prevents whole plugin from loading.

I'd prefer it to print warning, log whatever we have for debuging later and print as much as we can. The reason is to prevent e.g. one removed field in hammer-cli-foreman from loading whole other plugin.
It may be a bit difficult to handle the errors with the current notation. Would it help to make it similar to extend_help and wrap it into a block like: CommandXY.extend_output_definition do |od|; end?

ofedoren · 2018-10-24T17:19:09Z

@mbacovsky, updated.

I added handling similar to one from help extension. So, if a field is not found, then the whole output extension block will fail, but the plugin will be loaded and other extension blocks will be applied (if there are no errors).

Also, updated docs.

mbacovsky · 2018-10-30T23:45:28Z

[test hammer]

mbacovsky · 2018-10-31T12:38:17Z

@ofedoren a way better, thanks!

While testing I found another issue. I had the :mark_translations settings turned on so my labels were modified to e.g. >Name< and I realized addressing by labels won't work for localized Hammer. But there is not much we can do about it, right?

I also found out that when the field is created via dsl it won't have the :key set which limits the possibilities to identify the fields. The following patch fixed it for me:

diff --git a/lib/hammer_cli/output/dsl.rb b/lib/hammer_cli/output/dsl.rb
index e0cf7ab..5cf6a38 100644
--- a/lib/hammer_cli/output/dsl.rb
+++ b/lib/hammer_cli/output/dsl.rb
@@ -19,7 +19,10 @@ module HammerCLI::Output
 
     def field(key, label, type=nil, options={}, &block)
       options[:path] = current_path.clone
-      options[:path] << key if !key.nil?
+      unless key.nil?
+        options[:path] << key
+        options[:key] = key
+      end
 
       options[:label] = label
       type ||= Fields::Field

Otherwise it looks good!

mbacovsky · 2018-10-31T13:14:14Z

lib/hammer_cli/output/definition.rb


+    def field_index(field_id)
+      index = @fields.find_index do |f|
+        f.id == field_id


I had an idea to match the field by ANY of those identifiers like:

f.match_id?(field_id)

where the matcher is defined as

def match_id?(field_id) @options[:id] == filed_id || @options[:key] == field_id || @label == field_id end

It could prevent breaking the extension when e.g. the field_id is added.
What do you think?

ofedoren · 2018-11-07T10:16:50Z

@mbacovsky, updated. Thanks for match_id? suggestion! It makes more sense to use it instead of simple ==.

Also, I'm not sure if there is a problem with addressing by labels for localized Hammer. It seems to be working if address with _(), e.g. use .find_field(_('Name')) instead of .find_field('Name').

tstrachota · 2018-11-07T12:27:50Z

lib/hammer_cli/output/fields.rb

+    end
+
+    def match_id?(field_id)
+      @options[:id] == field_id || @options[:key] == field_id || @label == field_id


Would it make sense to add || @label == _(field_id) here? It could seamlessly solve the issue with translations.

ofedoren · 2018-11-07T13:32:10Z

@tstrachota, updated.

tstrachota

I'm good. @mbacovsky ?

tstrachota · 2018-11-09T09:44:35Z

lib/hammer_cli/i18n.rb

@@ -146,4 +146,3 @@ def self.main_makefile

 domain = [HammerCLI::I18n::LocaleDomain.new, HammerCLI::I18n::SystemLocaleDomain.new].find { |d| d.available? }
 HammerCLI::I18n.add_domain(domain) if domain
-


Please avoid changing files that aren't affected by the purpose of the patch.

tstrachota · 2018-11-09T09:51:34Z

[test hammer]

tstrachota · 2018-11-28T10:49:23Z

@mbacovsky ping

tstrachota · 2018-11-30T12:04:18Z

[test hammer]

tstrachota · 2018-11-30T12:15:48Z

All comments have been addressed and the changes work fine. Merging, thanks @ofedoren this is an important feature!

mbacovsky · 2018-11-30T16:17:27Z

Thanks @ofedoren, @tstrachota!

theforeman-bot added Needs testing Not yet reviewed labels May 31, 2018

ofedoren force-pushed the feat-21634-modif-output-fields branch 2 times, most recently from a7a8eda to a3f46d1 Compare June 4, 2018 14:36

ofedoren mentioned this pull request Jun 8, 2018

Fixes #19135 - Possibility to limit fields that are displayed #276

Merged

tstrachota suggested changes Jun 19, 2018

View reviewed changes

tstrachota mentioned this pull request Jun 19, 2018

Fixes #20712 - REX option would be show if its enabled on interface theforeman/hammer-cli-foreman#325

Closed

ofedoren force-pushed the feat-21634-modif-output-fields branch from a3f46d1 to 76b9076 Compare June 20, 2018 17:52

tstrachota mentioned this pull request Sep 11, 2018

Fixes #20187 - Display Host Status in Hammer theforeman/hammer-cli-foreman#385

Merged

tstrachota suggested changes Sep 20, 2018

View reviewed changes

theforeman-bot added Waiting on contributor and removed Not yet reviewed labels Sep 20, 2018

ofedoren force-pushed the feat-21634-modif-output-fields branch from 76b9076 to dff778b Compare September 21, 2018 10:36

theforeman-bot added Needs re-review and removed Waiting on contributor labels Sep 21, 2018

ofedoren mentioned this pull request Sep 28, 2018

Fixes #21633 - Allow modification of additional help texts #277

Merged

ofedoren force-pushed the feat-21634-modif-output-fields branch from dff778b to 19b60b8 Compare October 10, 2018 11:44

ofedoren mentioned this pull request Oct 11, 2018

Fixes #21635 - Extensible commands in hammer #291

Merged

mbacovsky reviewed Oct 17, 2018

View reviewed changes

ofedoren force-pushed the feat-21634-modif-output-fields branch from 19b60b8 to c4f3594 Compare October 18, 2018 15:04

mbacovsky added the Waiting on contributor label Oct 23, 2018

mbacovsky removed Needs re-review Needs testing labels Oct 23, 2018

ofedoren force-pushed the feat-21634-modif-output-fields branch from c4f3594 to 72ad363 Compare October 24, 2018 17:13

theforeman-bot added Needs re-review Needs testing and removed Waiting on contributor labels Oct 24, 2018

mbacovsky reviewed Oct 31, 2018

View reviewed changes

ofedoren force-pushed the feat-21634-modif-output-fields branch from 72ad363 to 0bae9e8 Compare November 7, 2018 10:07

tstrachota reviewed Nov 7, 2018

View reviewed changes

Fixes #21634 - Allows modification of output fields

f8173ad

ofedoren force-pushed the feat-21634-modif-output-fields branch from 0bae9e8 to f8173ad Compare November 7, 2018 13:29

tstrachota approved these changes Nov 9, 2018

View reviewed changes

theforeman-bot removed the Needs re-review label Nov 9, 2018

tstrachota added Demo worthy and removed Needs testing labels Nov 30, 2018

tstrachota merged commit 32a2651 into theforeman:master Nov 30, 2018

		@@ -146,4 +146,3 @@ def self.main_makefile

		domain = [HammerCLI::I18n::LocaleDomain.new, HammerCLI::I18n::SystemLocaleDomain.new].find { \|d\| d.available? }
		HammerCLI::I18n.add_domain(domain) if domain

Fixes #21634 - Allows modification of output fields #275

Fixes #21634 - Allows modification of output fields #275

Conversation

ofedoren commented May 31, 2018

tstrachota commented Jun 1, 2018

ofedoren commented Jun 4, 2018

tstrachota left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ofedoren commented Jun 20, 2018

tstrachota left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ofedoren commented Sep 21, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mbacovsky commented Oct 17, 2018

ofedoren commented Oct 18, 2018

mbacovsky commented Oct 19, 2018

ofedoren commented Oct 24, 2018

mbacovsky commented Oct 30, 2018

mbacovsky commented Oct 31, 2018

Choose a reason for hiding this comment

ofedoren commented Nov 7, 2018

Choose a reason for hiding this comment

ofedoren commented Nov 7, 2018

tstrachota left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tstrachota commented Nov 9, 2018

tstrachota commented Nov 28, 2018

tstrachota commented Nov 30, 2018

tstrachota commented Nov 30, 2018

mbacovsky commented Nov 30, 2018