PaperTrail lets you track changes to your models' data. It's good for auditing or versioning. You can see how a model looked at any stage in its lifecycle, revert it to any version, and even undelete it after it's been destroyed.
* Automatically records who was responsible via your controller. PaperTrail calls `current_user` by default, if it exists, but you can have it call any method you like.
PaperTrail stores the pre-change version of the model, unlike some other auditing/versioning plugins, so you can retrieve the original version. This is useful when you start keeping a paper trail for models that already have records in the database.
This also means that PaperTrail does not waste space storing a version of the object as it currently stands. The `versions` method gives you previous versions; to get the current one just call a finder on your `Widget` model as usual.
You can ignore changes to certain attributes like this:
class Article <ActiveRecord::Base
has_paper_trail :ignore => [:title, :rating]
end
This means that changes to just the `title` or `rating` will not store another version of the article. It does not mean that the `title` and `rating` attributes will be ignored if some other change causes a new `Version` to be crated. For example:
As an aside, I'm undecided about whether `widget.versions.last.next_version` should return `nil` or `self` (i.e. `widget`). Let me know if you have a view.
If instead you have a particular `version` of an item you can navigate to the previous and next versions.
If your `ApplicationController` has a `current_user` method, PaperTrail will store the value it returns in the `version`'s `whodunnit` column. Note that this column is a string so you will have to convert it to an integer if it's an id and you want to look up the user later on:
You may want PaperTrail to call a different method to find out who is responsible. To do so, override the `user_for_paper_trail` method in your controller like this:
class ApplicationController
def user_for_paper_trail
logged_in? ? current_member : 'Public user' # or whatever
N.B. A `version`'s `whodunnit` records who changed the object causing the `version` to be stored. Because a `version` stores the object as it looked before the change (see the table above), `whodunnit` returns who stopped the object looking like this -- not who made it look like this. Hence `whodunnit` is aliased as `terminator`.
To find out who made a `version`'s object look that way, use `version.originator`. And to find out who made a "live" object look like it does, use `originator` on the object.
The implementation is complicated by the edge case where the parent and child are updated in one go, e.g. in one web request or database transaction. PaperTrail doesn't know about different models being updated "together", so you can't ask it definitively to get the child as it was before the joint parent-and-child update.
The correct solution is to make PaperTrail aware of requests or transactions (c.f. [Efficiency's transaction ID middleware](http://github.com/efficiency20/ops_middleware/blob/master/lib/e20/ops/middleware/transaction_id_middleware.rb)). In the meantime we work around the problem by finding the child as it was a few seconds before the parent was updated. By default we go 3 seconds before but you can change this by passing the `:has_one` option to `reify`:
>> t = treasure.versions.last.reify(:has_one => 1) # look back 1 second instead of 3
If you are shuddering, take solace from knowing you can opt out of these shenanigans:
>> t = treasure.versions.last.reify(:has_one => false) # I say no to "workarounds"!
Opting out means your `:has_one` associated objects will be the live ones, not the ones the user saw at the time. Since PaperTrail doesn't auto-restore `:has_many` associations (I can't get it to work) or `:belongs_to` (I ran out of time looking at `:has_many`), this at least makes your associations wrong consistently ;)
PaperTrail can track most changes to the join table. Specifically it can track all additions but it can only track removals which fire the `after_destroy` callback on the join table. Here are some examples:
Having said that, you can apparently get all these working (I haven't tested it myself) with this [monkey patch](http://stackoverflow.com/questions/2381033/how-to-create-a-full-audit-log-in-rails-for-every-table/2381411#2381411):
# In config/initializers/core_extensions.rb or lib/core_extensions.rb
ActiveRecord::Associations::HasManyThroughAssociation.class_eval do
The difference is the call to `destroy_all` instead of `delete_all` in [the original](http://github.com/rails/rails/blob/master/activerecord/lib/active_record/associations/has_many_through_association.rb#L76-81).
There may be a way to store authorship versions, probably using association callbacks, no matter how the collection is manipulated but I haven't found it yet. Let me know if you do.
PaperTrail will call your proc with the current article and store the result in the `author_id` column of the `versions` table. (Remember to add your metadata columns to the table.)
Why would you do this? In this example, `author_id` is an attribute of `Article` and PaperTrail will store it anyway in serialized (YAML) form in the `object` column of the `version` record. But let's say you wanted to pull out all versions for a particular author; without the metadata you would have to deserialize (reify) each `version` object to see if belonged to the author in question. Clearly this is inefficient. Using the metadata you can find just those versions you want:
You can also store any information you like from your controller. Just override the `info_for_paper_trail` method in your controller to return a hash whose keys correspond to columns in your `versions` table. E.g.:
When you're storing every version of an object, as PaperTrail lets you do, you're almost certainly going to want to diff those versions against each other. However I haven't built a diff method into PaperTrail because I think diffing is best left to dedicated libraries, and also it's hard to come up with a diff method to suit all the use cases.
You might be surprised that PaperTrail doesn't use diffs internally anyway. When I designed PaperTrail I wanted simplicity and robustness so I decided to make each version of an object self-contained. A version stores all of its object's data, not a diff from the previous version.
So instead here are some specialised diffing libraries which you can use on top of PaperTrail.
For diffing two strings:
* [htmldiff](http://github.com/myobie/htmldiff): expects but doesn't require HTML input and produces HTML output. Works very well but slows down significantly on large (e.g. 5,000 word) inputs.
* [differ](http://github.com/pvande/differ): expects plain text input and produces plain text/coloured/HTML/any output. Can do character-wise, word-wise, line-wise, or arbitrary-boundary-string-wise diffs. Works very well on non-HTML input.
* [Jeremy Weiskotten's PaperTrail fork](http://github.com/jeremyw/paper_trail/blob/master/lib/paper_trail/has_paper_trail.rb#L151-156): uses ActiveSupport's diff to return an array of hashes of the changes.
* [activerecord-diff](http://github.com/tim/activerecord-diff): rather like ActiveRecord::Dirty but also allows you to specify which columns to compare.
Sometimes you don't want to store changes. Perhaps you are only interested in changes made by your users and don't need to store changes you make yourself in, say, a migration -- or when testing your application.
Over time your `versions` table will grow to an unwieldy size. Because each version is self-contained (see the Diffing section above for more) you can simply delete any records you don't want any more. For example:
sql> delete from versions where created_at <2010-06-01;
PaperTrail uses Bundler to manage its dependencies (in development and testing). You can run the tests with `bundle exec rake test`. (You may need to `bundle install` first.)
