collective.generic.webbuilder 1.1

Install or udpate minitage in your dedicated virtualenv if any

Just do that (you must refer to minitage installation for prerequisites)

Install virtualenv

virtualenv --no-site-packages --distribute $MT

Update minitage packages

source $MT/bin/activate
easy_install -U minitage.core
easy_install -U minitage.paste
minimerge -s

Install cgwb

Download & install via the minibuild

source $MT/bin/activate
git clone http://github.com/collective/collective.generic.webbuilder-minilay.git $MT/minilays/cgwb
minimerge -v cgwb

Cgwb lives in $MT/bfg/cgwb.

Launching the cgwb server

Launch via bin/cgwb. This binary includes some options to let you override the default port (–port) and listenning address (–host) To see all the available options, just use:

bin/cgwb --help

If you use minitage, mandatory to use the minitage.instances.env profile:

$MT/bin/easy_install -U minitage.paste
$MT/bin/paster create -t minitage.instances.env cgwb

MINITAGE .ENV

Each time you use cgwb, you use the .ENV:

source $MT/bfg/cgwb/sys/share/minitage/minitage.env

Use it

Launch it:

cd $INS
./bin/cgwb --port=6253

• At the moment, cgwb do not have some session mecanism, so the only way to replay a generation is to use the selenium firefox plugin.

• If you want to store your choices to redo an updated tarball later, just install the SeleniumIDE firefox plugin and use it to record your session.

• Maybe, activate selenium and

  • Go to the cgwb

  • Choose Generic Portal Plone3.

Filling the settings, some notes

• project name is mandatory and must be in the form in project or subproject.

• You can choose in the Plone Products to auto checkout in development mode the products from the community from which we should check out & use in development mode

THE IMPORTANT PART AROUND INITIATING A PROJECT

• It would be good unless you have some minitage experience to version the code prior to build, because of minitage update mecanism.

• Before version/import the code in your SCM you must elude the following points:

  • By default, the generated tarball contains the buildout layout and all the eggs in src, and the buildout use them as develop eggs and NOT WITH MR.DEVELOPER. Thus for running the buildout in standalone mode

  • You may decide not to include them as-is but to separate the code and version the code elsewhere.

  • I would advice you to checkout the packages with mr.developer.

An example of using svn which generic/pyramid

What i would do from a generated tarball for using subversion as my SCM could be to produce this layout:

import
|-- import/eggs
|   |-- import/eggs/myproject.core
|   |   `-- import/eggs/myproject.core/trunk
`-- import/buildout

• Exporting base variables:

  export PROJECT="myproject" # your project name as filled in the web interfacE
  export TARBALL="$(ls -1t ~/cgwb/${PROJECT}-*.tar.gz|head -n1)" # produced tarball
  export IMPORT_URL="https://subversion.xxx.net/scrumpy/${PROJECT}/" # base svn place to import

• Create a temporary workspace:

  mkdir -p  $PROJECT/tarball
  cd $PROJECT
  tar xzvf  $TARBALL -C tarball/

• Create the base layout to be imported:

  mkdir -p import/buildout import/eggs

• Move the generated plone extensions eggs to a separate place to be imported:

  for i in tarball/src/${PROJECT}*;do if [[ -d $i ]] && [[ $(basename $i) != "themes" ]];then j=$(basename $i);dest=import/eggs/$j/trunk; mkdir -pv  $(dirname $dest); mv -v $i $dest; fi; done

• Move the buildout structure in the import layout:

  cp -rf tarball/* import/buildout

• Update buildout to use mr.developer instead of basic develop:

  * move off the develop declaration::
  
      sed -re "s:(src/)?$PROJECT\.((skin)|(tma)|(core)|(testing))::g" -i import//buildout/etc/project/$PROJECT.cfg
  
  * add to mr.developer sources::
  
      sed -re "/\[sources\]/{
      a $PROJECT.core = svn $IMPORT_URL/eggs/$PROJECT.core/trunk
      }" -i import/buildout/etc/project/sources.cfg
  
  * add to auto checkout packages::
  
      sed -re "/auto-checkout \+=/{
      a \    $PROJECT.core
      }"  -i import/buildout/etc/project/sources.cfg
      sed -re "/eggs \+=.*buildout:eggs/{
      a \    $PROJECT.core
      }"  -i import/buildout/etc/project/$PROJECT.cfg
      sed -re "/zcml \+=/{
      a \    $PROJECT.core
      }"  -i import/buildout/etc/project/$PROJECT.cfg

• be sure to use the right svn url to checkout:

  sed -re "s|src_uri.*|src_uri=$IMPORT_URL/buildout/|g" -i import/buildout/minilays/$PROJECT/*

• Be sure to use svn

  sed -re “s|src_type.*|src_type=svn|g” -i import/buildout/minilays/$PROJECT/*

• Import:

  svn import import/ $IMPORT_URL -m "initial import"

An example of using svn which generic/plone

What i would do from a generated tarball for using subversion as my SCM could be to produce this layout:

import
|-- import/eggs
|   |-- import/eggs/myproject.policy
|   |   `-- import/eggs/myproject.policy/trunk
|   |-- import/eggs/myproject.skin
|   |   `-- import/eggs/myproject.skin/trunk
|   |-- import/eggs/myproject.testing
|   |   `-- import/eggs/myproject.testing/trunk
|   `-- import/eggs/myproject.tma
|       `-- import/eggs/myproject.tma/trunk
`-- import/minitage
    |-- import/minitage/buildouts
    |   `-- import/minitage/buildouts/zope
    |       `-- import/minitage/buildouts/zope/myproject

• Exporting base variables:

  export PROJECT="myproject" # your project name as filled in the web interfacE
  export TARBALL="$(ls -1t ~/cgwb/${PROJECT}-*.tar.gz|head -n1)" # produced tarball
  export IMPORT_URL="https://subversion.xxx.net/scrumpy/${PROJECT}/ # base svn place to import

• Create a temporary workspace:

  mkdir -p  $PROJECT/tarball
  cd $PROJECT
  tar xzvf  $TARBALL -C tarball/

• Create the base layout to be imported:

  mkdir -p import/buildout import/eggs

• Move the generated plone extensions eggs to a separate place to be imported:

  for i in tarball/src/${PROJECT}*;do if [[ -d $i ]] && [[ $(basename $i) != "themes" ]];then j=$(basename $i);dest=import/eggs/$j/trunk; mkdir -pv  $(dirname $dest); mv -v $i $dest; fi; done

• Move the buildout structure in the import layout:

  cp -rf tarball/* import/buildout

• Update buildout to use mr.developer instead of basic develop:

  * move off the develop declaration::
  
      sed -re "s:(src/)?$PROJECT\.((skin)|(tma)|(policy)|(testing))::g" -i import//buildout/etc/project/$PROJECT.cfg
  
  * add to mr.developer sources::
  
      sed -re "/\[sources\]/{
      a $PROJECT.policy = svn $IMPORT_URL/eggs/$PROJECT.policy/trunk
      a $PROJECT.tma = svn $IMPORT_URL/eggs/$PROJECT.tma/trunk
      a $PROJECT.skin = svn $IMPORT_URL/eggs/$PROJECT.skin/trunk
      a $PROJECT.testing = svn $IMPORT_URL/eggs/$PROJECT.testing/trunk
      }" -i import/buildout/etc/project/sources.cfg
  
  * add to auto checkout packages::
  
      sed -re "/auto-checkout \+=/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      a \    $PROJECT.testing
      }"  -i import/buildout/etc/project/sources.cfg
      sed -re "/eggs \+=.*buildout:eggs/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      a \    $PROJECT.testing
      }"  -i import/buildout/etc/project/$PROJECT.cfg
      sed -re "/zcml \+=/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      }"  -i import/buildout/etc/project/$PROJECT.cfg

• be sure to use the right svn url to checkout:

  sed -re "s|src_uri.*|src_uri=$IMPORT_URL/buildout/|g" -i import/buildout/minilays/$PROJECT/*

• Be sure to use svn

  sed -re “s|src_type.*|src_type=svn|g” -i import/buildout/minilays/$PROJECT/*

• Import:

  svn import import/ $IMPORT_URL -m "initial import"

An example of using git which generic

What i would do from a generated tarball for using subversion as my SCM could be to produce this layout:

import
    |-- myproject.policy
    |-- myproject.skin
    |-- myproject.testing
    `-- myproject.tma
    `-- myproject.buildout
    `-- myproject.minilay

• Exporting base variables:

  export PROJECT="myproject"                                     # your project name as filled in the web interfacE
  export TARBALL="$(ls -1t ~/cgwb/${PROJECT}-*.tar.gz|head -n1)" # produced tarball
  export IMPORT_URL="ssh://git.makina-corpus.net/var/git"              # base svn place to import

• Create a temporary workspace & the base layout to be imported:

  mkdir -p  $PROJECT/
  cd $PROJECT
  mkdir tarball import
  tar xzvf  $TARBALL -C tarball/

• Move the generated plone extensions eggs to a separate place to be imported:

  for i in tarball/src/*;do if [[ -d $i ]] && [[ $i != "tarball/src/themes" ]];then j=$(basename $i);dest=import/$j;mkdir -pv  $(dirname $dest); mv -v $i $dest; fi; done

• Move the buildout structure in the import layout:

  cp -rf tarball/minilays/$PROJECT   import/$PROJECT.minilay
  rm -rf tarball/minilays
  cp -rf tarball/ import/$PROJECT.buildout

• Update buildout to use mr.developer instead of basic develop:

  * move off the develop declaration::
  
      sed -re "s:(src/)?$PROJECT\.((skin)|(tma)|(policy)|(testing))::g" -i import/$PROJECT.buildout/etc/project/$PROJECT.cfg
  
  * add to mr.developer sources::
  
      sed -re "/\[sources\]/{
      a $PROJECT.policy =  git $IMPORT_URL/$PROJECT.policy
      a $PROJECT.tma =     git $IMPORT_URL/$PROJECT.tma
      a $PROJECT.skin =    git $IMPORT_URL/$PROJECT.skin
      a $PROJECT.testing = git $IMPORT_URL/$PROJECT.testing
      }" -i import/$PROJECT.buildout/etc/project/sources.cfg
  
  * add to auto checkout packages::
  
      sed -re "/auto-checkout \+=/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      a \    $PROJECT.testing
      }"  -i import/$PROJECT.buildout/etc/project/sources.cfg
      sed -re "/eggs \+=.*buildout:eggs/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      a \    $PROJECT.testing
      }"  -i import/$PROJECT.buildout/etc/project/$PROJECT.cfg
      sed -re "/zcml \+=/{
      a \    $PROJECT.policy
      a \    $PROJECT.tma
      a \    $PROJECT.skin
      }"  -i import/$PROJECT.buildout/etc/project/$PROJECT.cfg

• be sure to use the right git url to checkout:

  sed -re "s|src_uri.*|src_uri=$IMPORT_URL/$PROJECT.buildout|g" -i import/*.minilay/*

• Be sure to use git

  sed -re “s|src_type.*|src_type=git|g” -i import/.minilay/

• Import:

  pushd import;for i in *;do echo "Importing $i";pushd $i;git init;git add *;git commit -am "initial revision";git remote add origin "$IMPORT_URL/$i";git push --all origin;popd;done;popd

Deploy the project

• install the minilay:

  export MT=~/minitage
  svn co $IMPORT_URL/buildout/minilays/$PROJECT/ $MT/minilays/$PROJECT
  # or
  git clone  $IMPORT_URL/$PROJECT.minilay $MT/minilays/$PROJECT

• Install it:

  minimerge -v $PROJECT

The genericpaster directive

• Must be used at top level.

• Name of a configuration of templates.

<genericpaster name="Name of the configuration"/>

It contains a list of underlying configurations

>>> 'Test Generic Portal Plone' in root.configurations
True

The configurations objects contain a list of templates and plugins

>>> templates = root.configurations['Test Generic Portal Plone'].templates
>>> sorted(templates.keys())
['collective.generic.policy', 'minitage.plone3']

The template directive

• Must be used at genericpaster level.

• It describe a relative “paster template”. The name which you could get with paster create -t --list-templates.

• It has also an order which is used to order templates in the webinterface for lower to upper.

<template name="Template Name" order="int">

>>> t = templates['minitage.plone3']
>>> t.order
1
>>> t.name
'minitage.plone3'

A template can also say that it must be generated under a ‘subdirectory’ with the output attribute.

>>> templates['collective.generic.policy'].output
'src'

The group directive

• A template has a list of groups of options.

• Groups are represented by a block of questions surrounded by the group name in the webinterface.

<group name="GroupName" order="int"/>

Those groups group ‘paster questions’.

>>> groups = t.groups
>>> groups.keys()
['default', 'Minitage']
>>> g = t.groups['Minitage']
>>> t.groups['Minitage'].order
5
>>> t.groups['Minitage'].name
'Minitage'

The options directive

• Must be used at group level.

• Groups group options, Which can be grabbed by a regular expression with this directive.

<options prefix="Regular expression"  type="boolean|" default="value"/>

• type can be omitted and defaults to None (text).

• default can be omitted and no default value will be assigned (or the paster default value).

>>> opts = g.options['.*with.*']
>>> opts.type, opts.default
('boolean', 'true')

As you can see, there is a default group where go non-matched options which are not excluded via the excludeoptions directive.

The option directive

• Must be used at group level.

• Groups group also ‘single options’, Which can be grabbed by their name.

• Single options and can have an alias. It is useful if we have the same ‘option name’ in 2 templates of the configuration and we don’t want that they share the same value (default behaviour). To be clear, we have the option ‘project’, in template ‘a’ and ‘b’, by default, if we choose ‘foo’ for ‘project’, the value will be ‘foo’ in template ‘a’ and ‘b’, and with an alias, we can choose the value for ‘a’ _and_ for ‘b’.

<options name="name" alias="alias name"  type="boolean|" default="value"/>

• alias can be omitted.

• type can be omitted and defaults to None (text).

• default can be omitted and no default value will be assigned (or the paster default value).

>>> opt = g.single_options['install_method']
>>> opt.type, opt.alias, opt.default
(None, 'ai', None)

The excludeoptions & excludeoption directives

• Must be used at group or template level (in any group of the template).

<excludeoptions prefix="regular expression"/>

• exclude options from the interface

• prefix: regular expression for the options to exclude.

<excludeoption name="option name"/>

• exclude an option from the interface

• name: name for the options to exclude.

>>> [[getattr(templates[template].groups['default'], attr).keys() for attr in 'exclude_options', 'excludes_options'] for template in 'minitage.plone3', 'collective.generic.policy']
[[['python'], ['project_.*']], [['python'], ['project_.*']]]

The plugin directive

• Must be used at template level.

• Declare which plugin must run after the templates collection generation.

• This is useful for example, to rearrange things which are generated.

To run a plugin which is declared under “plugin name”.

<plugin name="plugin name" order="int"/>

• name: name of the adapter

• order: control order to run if there are more than one plugin

A plugin, is a simple adapter which takes a IPasterAssembly and provides IPostGenerationPlugin

<adapter
  name="plugin name"
  provides=".interfaces.IPostGenerationPlugin"
  factory=".plugins.MyPluginFactory"
  for=".interfaces.IPasterAssembly"
/>

>>> plugins = root.configurations['Test Generic Portal Plone'].plugins
>>> plugins
[('egg_plugin', 2)]

Testing the zcml to python represetation

Load our test package where we have three templates

>>> import collective.generic.webbuilder.tests
>>> testegg = os.path.join( collective.generic.webbuilder.tests.__path__[0], 'egg', 'src')
>>> pkg_resources.working_set.add_entry(testegg)
>>> env = pkg_resources.Environment()
>>> egg = env['cgwb.tp'][0]

We have 3 templates in there waiting to be assembled

>>> pprint(egg.get_entry_map())
{'paste.paster_create_template': {'cgwb.testpackage1': EntryPoint.parse('cgwb.testpackage1 = tp.package:Package'),
                                  'cgwb.testpackage2': EntryPoint.parse('cgwb.testpackage2 = tp1.package:Package'),
                                  'cgwb.testpackage3': EntryPoint.parse('cgwb.testpackage3 = tp2.package:Package')}}

The configuration

It is more described in the zcml part of the documentation, but it’s a zcml representation of which variables from the pastertemplates we want to extract and how we want to present them to users.

A sample zcml needed to assemble the packages we declared before is as follow:

>>> paster_zcml = """
... <configure xmlns="http://namespaces.repoze.org/bfg" xmlns:meta="http://namespaces.zope.org/meta">
...   <include package="collective.generic.webbuilder" file="meta.zcml"/>
...   <configure xmlns="http://webbuilder.org/webbuilder">
...     <genericpaster name="test Assembler">
...         <template name="cgwb.testpackage1" output="1" order="1000">
...           <group name="Minitage" order="05">
...             <option name="tp1option" type="boolean"/>
...             <option name="tp1option3" default="y"/>
...           </group>
...         </template>
...         <template name="cgwb.testpackage2" output="2" order="200">
...           <group name="Authors" order="20">
...             <options prefix="^author.*"/>
...           </group>
...           <excludeoptions prefix=".*"/>
...         </template>
...         <template name="cgwb.testpackage3" output="3" order="500">
...           <group name="Plone Settings" order="8">
...             <option name="tp2opton2" />
...             <option name="author_email" />
...           </group>
...           <group name="Authors" order="20">
...             <options prefix="^author.*"/>
...           </group>
...           <group name="Package tuning" order="1">
...             <option name="project_name" type="hidden" default="tma" alias="tmapn"/>
...           </group>
...         </template>
...     </genericpaster>
...   </configure>
... </configure>
... """
>>> noecho = xmlconfig.string(paster_zcml)
>>> root.configurations['test Assembler']
<collective.generic.webbuilder.zcml.PasterConfiguration object at ...>

PasterAssembly object

have some “log variables” and the configuration name to search for in the “bfgroot”.configurations dictionnary.

• template_data: list of mappings in the form:

  [
   {
    'self': template 'zcml' object,
    'name': paster template name,
    'added_options': option added by this template,
    'not_explicit_options': option added by this template which were not explicitly matched,
    'display' : display a template or not
    'groups':
       {
          groupname:
              {
                  'name': groupname,
                  'group': zcml group object:
                  'options': [(paster variablen, type, optionn name, alias|None, zcml optionn|None )]
              }
  
       },
    'aliases': [ (varName, aliasName),]
  
   }
  ]

• added_options: All options added for all templates

Get an assembly for the wanted configuration

>>> ta = gpaster.PasterAssembly('test Assembler')
>>> pprint(ta.__dict__.items())
[('templates_data', []),
 ('configuration',
  <collective.generic.webbuilder.zcml.PasterConfiguration object at ...>),
 ('added_options', []),
 ('configuration_name', 'test Assembler')]

The PasterAssemblyReader object

This adapter takes as input a IPasterAssembly object and implements the IPasterAssemblyReader interface.

We have configurations stored into zcml representation, now we need to gather and map the configuration informations with the content of each “paster template” into a python friendly structure. This Reader component is responsible for storing in the assembly object:

• The extracted template name

• Each group of options

• For each of those groups:

  • The excluded options

  • For the options which are not excluded, if applicable:

    • making its alias

    • Assign the default value

What will finnally load the assembly data structures is a reader that now how to parse a configuration

>>> reader = gpaster.PasterAssemblyReader(ta)
>>> reader.readed
False
>>> reader.read()
>>> len(ta.added_options) > 0
True
>>> reader.readed
True

We will check now that the structure loaded is as we wanted

>>> t1 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage1')
>>> t2 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage2')
>>> t3 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage3')

Templates

Order of templates is respected

>>> [t['name'] for t in ta.templates_data]
['cgwb.testpackage2', 'cgwb.testpackage3', 'cgwb.testpackage1']
>>> rt2, rt3, rt1 =  ta.templates_data

Groups

Options in template3, on the paster side, that we ll find on the next example

>>> pprint([v.name for v in t3.vars])
['namespace',
 'nested_namespace',
 'version',
 'author',
 'author_email',
 'tp3option',
 'tp3option3',
 'keywords',
 'license_name',
 'project_name']

For each templates, options are grouped, and groups respect order defined in zcml:

>>> pprint([(rt3['groups'][n]['name'] , rt3['groups'][n]['options']) for n in range(len(rt3['groups']))])
[('Package tuning',
  [(<var project_name default='tma' should_echo=True>,
    'hidden',
    'tmapn',
    <collective.generic.webbuilder.zcml.Option object at ...>)]),
 ('Plone Settings',
  [(<var author_email default='bar@localhost' should_echo=True>,
    'default',
    None,
    <collective.generic.webbuilder.zcml.Option object at ...>)]),
 ('Authors',
  [(<var author default='foo' should_echo=True>,
    'default',
    None,
    <collective.generic.webbuilder.zcml.Options object at ...>)]),
 ('default',
  [(<var namespace default='%(namespace)s' should_echo=True>,
    'default',
    None,
    None),
   (<var nested_namespace default='%(package)s' should_echo=True>,
    'default',
    None,
    None),
   (<var version default='1.0' should_echo=True>, 'default', None, None),
   (<var tp3option default='http://python.org' should_echo=True>,
    'default',
    None,
    None),
   (<var tp3option3 default='Project %s' should_echo=True>,
    'default',
    None,
    None),
   (<var keywords default='' should_echo=True>, 'default', None, None),
   (<var license_name default='GPL' should_echo=True>,
    'default',
    None,
    None)])]

As you can see project_name has been aliased and will be explained after.

Consumed options

Goal is to insist loudly on consumed option. When an option is consumed by another template, it is not available in others to be asked only once. That’s why , template1 has no variables which were first asked in template3.

>>> rt3options = []; noecho = [rt3options.extend(g['options']) for g in rt3['groups']]; rt3options = [opt[0].name for opt in rt3options]
>>> rtp1options = []; noecho = [rtp1options.extend(g['options']) for g in rt1['groups']]; rtp1options = [opt[0].name for opt in rtp1options]

project_name, author, etc. are not part of template1 options even if they are in the paster template. They have been consumed by template3

>>> pprint([v.name for v in t1.vars])
['namespace',
 'nested_namespace',
 'version',
 'author',
 'author_email',
 'tp1option',
 'tp1option2',
 'tp1option3',
 'keywords',
 'license_name',
 'project_name']
>>> rt3options
['project_name', 'author_email', 'author', 'namespace', 'nested_namespace', 'version', 'tp3option', 'tp3option3', 'keywords', 'license_name']
>>> rtp1options
['tp1option', 'tp1option3', 'tp1option2', 'project_name']

Excluded options

Template2 ignore all opions per default, even adding an option can’t precedence over ignoring options. Take care of your regexes !

>>> [g['options'] for g in rt2['groups']]
[[], []]

Typed options

We can assign type to values to use different widgets to display them in the UI for example. Supported types are:

• boolean (checkbox)

• hidden (hidden)

• default (textarea)

template3 define project_name as hidden

>>> rt3['groups'][0]['options'][0][1]
'hidden'

template1 define tp1option as boolean.

>>> rt1['groups'][0]['options'][0][1]
'boolean'

If an option default startswith ‘y’, ‘true’, or ‘on’, we switch the option type to boolean

>>> rt1['groups'][0]['options'][1][1]
'boolean'

Options in the default group have default as type, as for options without explicit type

>>> rt3["groups"][3]['options'][0][1]
'default'
>>> rt3["groups"][2]['options'][0][1]
'default'

Option aliases

We have defined a default value and a default type for template3.project_name which is also an alias. Alias allow options with the same name but not the same value to exists within the same Assembly. Default behaviour tells that one value is asked only once and used for all options that have the same name unless they are aliased explicitly each one of them.:

>>> rt3['groups'][0]['options']
[(<var project_name default='tma' should_echo=True>, 'hidden', 'tmapn', <collective.generic.webbuilder.zcml.Option object at ...>)]

Added options

We can retrieve options added

• For one template

  >>> rt1['added_options']
  ['tp1option', 'tp1option2', 'tp1option3', 'project_name']
  >>> rt2['added_options']
  []
  >>> rt3['added_options']
  ['namespace', 'nested_namespace', 'version', 'author', 'author_email', 'tp3option', 'tp3option3', 'keywords', 'license_name', 'project_name']

• For all templates

  >>> ta.added_options
  ['namespace', 'nested_namespace', 'version', 'author', 'author_email', 'tp3option', 'tp3option3', 'keywords', 'license_name', 'tmapn', 'tp1option', 'tp1option2', 'tp1option3', 'project_name']

A plugin is a simple adapter

Creating plugins to run after a generation is really simple. It is just a matter of implementing an adapter which takes an \IPasterConfiguration and provided IPostGenerationPlugin.

<adapter
  name="plugin name"
  provides=".interfaces.IPostGenerationPlugin"
  factory=".plugins.MyPluginFactory"
  for=".interfaces.IPasterAssembly"
/>

The eggs plugin

For example, here is a simple plugin which take all eggs in a ‘src’ directory and register them in ‘zcml’ and ‘develop’ in the relative ‘’buildout.cfg’’ It will add just the ‘policy’ egg to the intance’s zcml option.

Boiler plate to simulate a generation

>>> import tempfile, shutil, os
>>> c = os.getcwd()
>>> d = tempfile.mkdtemp()
>>> os.chdir(d)
>>> open('buildout.cfg', 'w').write('[buildout]\ndevelop+=\n   foo\n[instance]\nzcml=  too\n')
>>> os.makedirs('src/bar/src')
>>> os.makedirs('src/policy/src')
>>> open('src/bar/setup.py', 'w').write('')
>>> open('src/policy/setup.py', 'w').write('')

Running the plugin

>>> from collective.generic.webbuilder.models import root
>>> conf = root.configurations['Generic Portal Plone3']
>>> from collective.generic.webbuilder import interfaces, paster
>>> pa = paster.PasterAssembly('Generic Portal Plone3')
>>> plugin = zope.component.queryAdapter(pa, interfaces.IPostGenerationPlugin, name='egg_plugin')
>>> plugin.process(d, 'foo', {})
>>> print open('buildout.cfg').read()
[buildout]
develop+=src/policy
   src/bar
   foo
eggs += policy
        bar
[instance]
zcml=  too
            policy
<BLANKLINE>

Cleanup

>>> os.chdir(c);shutil.rmtree(d)

Registering plugins

Please refer to the plugin zcml directive to know how to add plugins for a ‘Configuration’.

Here is an example about the “eggs_plugins”

<configure xmlns="http://namespaces.repoze.org/bfg"
    xmlns:meta="http://namespaces.zope.org/meta"
    xmlns:cgwb=xmlns="http://webbuilder.org/webbuilder">
    <adapter
        name="egg_plugin"
        provides=".interfaces.IPostGenerationPlugin"
        factory=".plugins.EggPlugin"
        for=".interfaces.IPasterAssembly"
    />
    <cgwb:genericpaster name="Generic Portal Plone">
        <cgwb:plugin name="egg_plugin" order="2"/>
        <cgwb:template name="minitage.plone3" order="1">
        </template>
    </genericpaster>
</configure>

Testing the zcml to python represetation

Load our test package where we have three templates

>>> import collective.generic.webbuilder.tests
>>> testegg = os.path.join( collective.generic.webbuilder.tests.__path__[0], 'egg', 'src')
>>> pkg_resources.working_set.add_entry(testegg)
>>> env = pkg_resources.Environment()
>>> egg = env['cgwb.tp'][0]

The configuration

It is more described in the zcml part of the documentation, but it’s a zcml representation of which variables from the pastertemplates we want to extract and how we want to present them to users.

A sample zcml needed to assemble the packages we declared before is as follow:

>>> paster_zcml = """
... <configure xmlns="http://namespaces.repoze.org/bfg" xmlns:meta="http://namespaces.zope.org/meta">
...   <include package="collective.generic.webbuilder" file="meta.zcml"/>
...   <configure xmlns="http://webbuilder.org/webbuilder">
...     <genericpaster name="test Assembler">
...         <template name="cgwb.testpackage1" output="1" order="1000">
...           <group name="Minitage" order="05">
...             <option name="tp1option" type="boolean"/>
...             <option name="tp1option3" default="y"/>
...           </group>
...         </template>
...         <template name="cgwb.testpackage2" output="2" order="200">
...           <group name="Authors" order="20">
...             <options prefix="^author.*"/>
...           </group>
...           <excludeoptions prefix=".*"/>
...         </template>
...         <template name="cgwb.testpackage3" output="3" order="500">
...           <group name="Plone Settings" order="8">
...             <option name="tp2opton2" />
...             <option name="author_email" />
...           </group>
...           <group name="Authors" order="20">
...             <options prefix="^author.*"/>
...           </group>
...           <group name="Package tuning" order="1">
...             <option name="project_name" type="hidden" default="tma" alias="tmapn"/>
...           </group>
...         </template>
...     </genericpaster>
...   </configure>
... </configure>
... """
>>> noecho = xmlconfig.string(paster_zcml)
>>> root.configurations['test Assembler']
<collective.generic.webbuilder.zcml.PasterConfiguration object at ...>

We will check now that the structure loaded is as we wanted

>>> t1 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage1')
>>> t2 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage2')
>>> t3 = pkg_resources.load_entry_point('cgwb.tp', 'paste.paster_create_template', 'cgwb.testpackage3')
>>> server, url = launch_server()
>>> browser = Browser(url)

We can see that in the main page we have the default configurations and the custom loaded one

>>> 'test Assembler' in browser.contents
True
>>> 'Generic Portal Plone4' in browser.contents
True
>>> 'Generic Portal Plone3' in browser.contents
True

Following the test assembler link

>>> browser.getLink('test Assembler').click()
>>> htmlS(browser.contents).xpath('//input[@name="project"]')[0]
<InputElement ... name='project' type='text'>

I can submit a form and a valid it

>>> browser.getControl(name='project').value = 'myproject'
>>> browser.getControl(name='author').value = 'tim burton'
>>> browser.getControl(name='author_email').value = 'tim burton@foo.com'
>>> browser.getControl(name='tp1option').value = False
>>> browser.getControl(name='tp1option2').value = 'Project Monster'
>>> browser.getControl(name='project_name').value = 'My Big Project'
>>> browser.getControl(name='submit_cgwbDownload').click()
>>> '.tar' in browser.contents
True

The sucessful produced result is a tarball

>>> pprint(browser.headers.headers)
['Server:...
 'Date:...
 'Content-Disposition: attachment; filename="myproject....tar.gz"\r\n',
 'Content-Transfer-Encoding: binary\r\n',
 'Content-Length: ...\r\n']
>>> import tarfile
>>> tar = tarfile.open(fileobj=StringIO(browser.contents))

In the produced tarball, output directories present in the zcml configuration are respected in the tarball:

>>> files = [a.name for a in tar];files.sort();pprint(files)
['.',
 '1',
 '1/myproject',
 '1/myproject/test',
 '2',
 '2/myproject',
 '2/myproject/test1',
 '3',
 '3/myproject',
 '3/myproject/test2']
>>> templates = dict([(a.name,a) for a in tar if 'test' in a.name])
>>> t2 = templates['2/myproject/test1']
>>> t3 = templates['3/myproject/test2']
>>> t1 = templates['1/myproject/test']

Options filled in the interface are well interpreted in templates

>>> pprint([a for a  in tar.extractfile(t1).read().split('\n') if a.strip()])
['namespace                 =>            %(namespace)s',
 'nested_namespace          =>            %(package)s',
 'version                   =>            1.0',
 'author                    =>            tim burton',
 'author_email              =>            tim burton@foo.com',
 'tp1option                 =>            False',
 'tp1option2                =>            Project Monster',
 'tp1option3                =>            True',
 'keywords                  =>            ',
 'license_name              =>            GPL',
 'project_name              =>            My Big Project']

The project_name entered for project1 is shared in project2

>>> pprint([a for a  in tar.extractfile(t2).read().split('\n') if a.strip()])
["'namespace'                =>         '%(namespace)s'",
 "'nested_namespace'         =>         '%(package)s'",
 "'version'                  =>         '1.0'",
 "'author'                   =>         'tim burton'",
 "'author_email'             =>         'tim burton@foo.com'",
 "'keywords'                 =>         ''",
 "'license_name'             =>         'GPL'",
 "'project_name'             =>         'My Big Project'",
 "'tp2option'                =>         'tp2option'",
 "'tp2opton2'                =>         'tp2opton2'"]

the aliased project_name (tma) takes efffect in the third template

>>> pprint([a for a  in tar.extractfile(t3).read().split('\n') if a.strip()])
["'namespace'                =>         '%(namespace)s'",
 "'nested_namespace'         =>         '%(package)s'",
 "'version'                  =>         '1.0'",
 "'author'                   =>         'tim burton'",
 "'author_email'             =>         'tim burton@foo.com'",
 "'keywords'                 =>         ''",
 "'license_name'             =>         'GPL'",
 "'project_name'             =>         'tma'",
 "'tp3option3'               =>         'Project %s'",
 "'tp3option'                =>         'Project %s'"]