Download the PHP package burnbright/silverstripe-importexport without Composer

SilverStripe Import/Export Module

Import and export data from SilverStripe in various forms, including CSV. This module serves as a replacement/overhaul of BulkLoader functionality found in the framework.

The loading process

1. Raw data is retrieved from a source (BulkLoaderSource).
2. Data is provided as iterable rows (each row is heading->value mapped array).
3. Rows are mapped to a standardised format, based on a user/developer provided mapping.
4. Data is set/linked/tranformed onto a placeholder DataObject.
5. Existing record replaces placeholder, or placeholder becomes the brand new DataObject.
6. DataObject is validated and saved.
7. All results are stored in BulkLoader_Result.

User-defined column mapping

Users can choose which columns map to DataObject fields. This removes any need to define headings, or headings according to a given schema. Users can state if the first line of data is in fact a heading row. Mappings are saved for the next time an import is done on the same GridField.

Grid Field Importer

This is a grid field component for users to selecting a CSV file and map it's columns to data fields.

The importer makes use of the CSVFieldMapper, which displays the beginning content of a CSV.

BulkLoaderSource

A BulkLoaderSource provides an iterator to get record data from. Data could come from anywhere such as a CSV file, a web API, etc.

It can be used independently from the BulkLoader to obtain data.

(Better)BulkLoader

• Saves data from a particular source and persists it to database via the ORM.
• Determines which fields can be mapped to, either scaffolded from the model, provided by configuration, or both.
• Detects existing records, and either skips or updates them, based on criteria.
• Maps the source data to new/existing dataobjects, based on a given mapping.
• Finds, creates, and connects relation objects to objects.
• Can clear all records prior to processing.

ListBulkLoader

Often you'll want to confine bulk loading to a specific DataList. The ListBulkLoader is a variation of BulkLoader that adds and removes records from a given DataList. Of course DataList iself doesn't have an add method implemented, so you'll probably find it more useful for a HasManyList.

Mapping record data to a standard format

You can provide a columnMap to map incoming records to a standard format.

This column map is generated by the CSVFieldMapper control inside the GridFieldImporter component.

Scaffolding vs Defining Mappable Fields

Mappable fields will be scaffolded if you do not define them yourself. This includes fields that are on relations, so that relations can be linked up.

It is likely there will be fields you don't want mapped, in which case you should specify a mappableFields array on your loader:

Transform incoming record data

You may want to perform some transformations to incoming record data. This can be done by specifying a callback against the record field name.

Require specific data to be present

Incoming records without required data will be skipped.

Note that empty records are skipped by default.

Creating and linking related DataObjects

The bulk loader can handle linking and creating has_one relationship objects, by either providing a callback, or using the Relation.FieldName style "dot notation". Relationship handling is also performed in the transformations array.

You can specify at the BulkLoader level if records will be created and linked, then you can also specify the behaviour for each field. The default behaviour is to both link and create relation objects.

Here are some configuration examples:

Note that $placeholder in the above example refers to a dummy DataObject that is populated in order to then be saved, or checked against for duplicates. You should not call $placeholder->write() in your callback.

Specify a relation list

In the same way that you may use a ListBulkLoader to constrain records to a given DataList, you may also want to constrain the relation records to a List.

Determining when to overwrite existing (duplicate) DataObjects

Duplicate checks are performed on record data, mapped into the standardised form.

You can perform duplicate checking on data fields:

Or on a relation:

Duplicates can also be found using a callback function:

Publish pages during import

If you are importing instances of SiteTree, you can have those pages automatically published using this configuration:

Replace all the "legacy" ModelAdmin importers

Some simple yaml config options to help with swapping out all the importer functionality.

Remove only the scafolded (non-custom) importers:

Troubleshooting

Missing relation objects

If you are writing relation objects during loading, and they fail validation, the loader will simply ignore that relation object.

Multiple relation data fields not mapping to the same relation

If you have mapped multiple fields mapping to the same relation, you may get situations where the incorrect existing relation object is joined. The first field that is mapped is the same field used to find the relation. For example, you'll likely want a Title to be used to find/create a relation, and then an Amount will be added to that same relation, rather than finding/creating a relation by an Amount, and setting the Title.

Define the correct ordering using the mappableFields array to fix this.

Contributions

Please do contribute whatever you can to this module. Check out the issues and milestones to see what has needs to be done.

License

MIT

Author

Jeremy Shipman (http://jeremyshipman.com)