Blackbit Data Director Bundle

Import XML, CSV, JSON, Excel files to Pimcore objects, assets, documents + Export feeds + create REST API without any programming

For an overview how to use this plugin, please see our tutorial videos.

Installation

Composer

To get the plugin code you have to buy the plugin or write an email to info@blackbit.de.

You then either get access to the bundle’s Bitbucket repository or you get the plugin code as a zip file. Accessing the Bitbucket repository has the advantage that you will always see changes to the plugin in the pull requests and are able to update to a new version yourself - please visit this page if this sounds interesting to you - if it does, please send us the email address of your BitBucket account so we can allow access to the repository.

When we allow your account to access our repository, please add the repository to the composer.json in your Pimcore root folder (see Composer repositories):

"repositories": [
    {
        "type": "vcs",
        "url": "git@bitbucket.org:blackbitwerbung/pimcore-plugins-data-director"
    }
]

(Please add your public SSH key to your Bitbucket account for this to work)

Alternatively if you received the plugin code as zip file, please upload the zip file to your server - e.g. create a folder bundles in the Pimcore root folder) and add the following to your composer.json:

"repositories": [
    {
        "type": "artifact",
        "url": "./bundles/"
    }
]

Beware that when you put the zip directly in the Pimcore root folder, and add "url": "./" it will still work but Composer will scan all files under the Pimcore root recursively to find bundle zip files (incl. assets, versions etc.) - which will take quite a long time.

Then you should be able to execute composer require blackbit/data-director (or composer update blackbit/data-director --with-dependencies for updates if you already have this bundle installed) from CLI.

At last you have to enable and install the plugin by executing ./vendor/bin/pimcore-bundle-enable BlackbitDataDirectorBundle

You can always access the latest version by executing composer update blackbit/data-director --with-dependencies on CLI.

Migrations / Updates

To update the plugin itself, please put the zip file which you received into the folder bundles in the Pimcore root folder. This step is not necessary if you have access to the bundle’s Bitbucket repository.

Afterwards run composer update blackbit/data-director --with-dependencies to update the plugin to the latest version.

Modifications on the database scheme (migrations) get executed automatically, so you do not have to care for this.

Nevertheless, if you do not see the Data Director main menu icon in the Pimcore backend, it may be caused by a failed migration. You can execute the migrations manually via bin/console pimcore:bundle:install BlackbitDataDirectorBundle to see which migration fails.

Quick overview

Importing will run in two phases:

  1. Parse source file, import data into a flat database table

    In the first phase data does not get transformed. This is purely a fast way to import source data to an intermediate data container named raw data table.

    You have to specify which fields shall be extracted from the import source to the raw data table.

  2. Map data from intermediate table to object / asset / document fields / attributes.

    In the attribute mapping panel all fields of the target class are listed. For each field you can assign the raw data field from 1. to be used. Additionally / alternatively you can use multiple raw data fields or modify the source data via callback functions if necessary - for example parse numbers, find related objects, transform the data to the expected data type. These mapping settings can be configured by clicking the Settings button button.

    Also in the field mapping settings window you can mark one or multiple fields as key fields. During import it is first tried to find already existing objects which have the same values in the key fields - if one or multiple elements get found, they get updated instead of new elements being created.

    When multiple fields are set to be used as key fields, they are connected by AND, so objects only get updated if all key fields match the assigned raw data fields.

    In the dataport panel settings under Advanced Options you can set the import mode, available are:

    • Create and edit objects
    • Create new objects, do not edit existing ones
    • Edit existing objects, do not create new ones

    You can skip single raw data items from being processed by returning null for a key field.

How to import data to objects

You can start an import on multiple ways:

  • manually in the Pimcore backend on the panel History & manual import with the buttons Start rawdata import, Import raw data to objects and Start complete import
  • manually by right-clicking the target folder in the data objects tree or on the import resource (if Pimcore assets are used as import resource)
  • via CLI by executing one of these scripts:

    • bin/console data-director:extract <Dataport-ID> [--rm] - Parse source resource (e.g. file or URL) and write data to a flat database table (replace with the real ID).

      If --rm is set, the imported file will get deleted after raw data import.

    • bin/console data-director:process <Dataport-ID> [Rawdata-ID] [--force] - Maps raw data to Pimcore objects, updates exiting objects or creates new ones as needed.

      By default a raw data item gets only imported if the raw data for the found object changed since the last import (for performance reasons and to not overwrite manually edited values with old import data). By setting --force (or -f) you can bypass this check.

    • you can additionally use -vvv to get all log / error messages
    • To see all parameters and a description of these commands you can use bin/console data-director:extract --help or bin/console data-director:process --help.
    • bin/console data-director:complete <Dataport-ID> combines raw data extraction and data processing - it first executes data-director:extract and afterwards data-director:process

  • start import automatically if import source changes (configurable via checkbox Run automatically on new data in dataport settings). If activated the import gets started as soon as a file / object gets saved which is inside the configured import file path (for Pimcore asset based imports) or which matches the given filter criteria (SQL condition for Pimcore-based imports). In this case the import gets started only for the modified file / object (not for all files / objects which match the path / filter criteria).

    Automatic imports also get triggered if an element is referenced by the saved object.

    Example: you have an import (with Run automatically on new data enabled) which shall update an assets metadata field product name to the product’s name of a data object which the asset is referenced in. As soon as the product gets saved, the asset metadata import gets started for all assets which are referenced by this product.

    This is also useful when you dynamically create text data from product attributes via placeholders. This way you will always have the updated text even if you have placeholders which fetch data from referenced objects.

    If you do not want to execute the dataport also for dependent objects, you can add {% if id %} o_id = '{{ id }}' {% endif %} to the dataport’s SQL condition. As soon as an object gets saved this placeholder gets replaced by the just saved object’s ID and thus the dataport will only get automatically executed for this one object.

  • start imports via REST API call

Dataport settings

Import resources

Importing data from folders

You can specify a folder path as import source instead of a single file. In this case files in this folder (and subfolders) are imported subsequently (sorted by modification time - oldest first). Already imported files are stored in an archive folder within the import folder.

In addition to the source file fields you can use the field __source to obtain the source file name. This can be useful if you want to process the source file name in the object import (e.g. to call a dependent import with the same import file). It will contain a local file copy of the import resource to ensure that even if you use the --rm option the file will be available to follow-up imports.

Moreover, there is the special field selector __updated which provides you the modification time (Unix timestamp) of the import file.

You can also specify a Pimcore asset folder instead of a file system folder. This enables you to enable the checkbox to automatically start imports if files in this folder change or are uploaded. In this case the import is only run with the changed file not with all files in the given folder.

Importing files with glob expression (wildcards)

It is possible to enter an expression like /path/to/import/files/*.xml - or any other glob expression, see Wikipedia Glob to see all options for wildcards. Also a relative path expression based on the Pimcore root folder or on Pimcore assets is possible.

Importing data from URLs

You can specify a URL instead of a file.

Authorization is supported, e.g. https://username:password@example.org/file.csv

You can also provide FTP, FTPS, SFTP and AWS S3 URLs:

ftp://username:password@example.org/folder/file.csv
ftps://username:password@example.org/folder/file.csv
sftp://username:password@example.org/folder/file.csv
s3://key:secret@region/bucket/path/to/files

This is even possible with * as wildcard expression, e.g.

sftp://username:password@example.org/folder/BMEcat_*.xml

This will import data from all files inside folder /folder whose name begins with BMEcat_ and ends with .xml.

The * placeholder does not traverse to sub directories, so sftp://username:password@example.org/*.xml will find /file.xml but not /folder/file.xml. If you want to recursively traverse all subdirectories, you can use **:

sftp://username:password@example.org/**.xml This will find /file.xml, /folder/file.xml and also /any/number/of/subdirectories/file.xml

If you want to use the wildcard operators but still only want to use the last modified matching file, you can use

sftp://username:password@example.org/folder/BMEcat_*.xml | latest
Login credentials

You can also specify login credentials via placeholders, e.g. ftp://{{ FTP_USERNAME }}:{{ FTP_PASSWORD }}@example.org/folder/file.csv. These placeholders can be provided via:

Priority is in this order, more upper ones have higher priority if the variable exists in multiple sources.

Alternatively for Amazon S3 access, you can also use the Amazon-SDK-specific authentication with credential files or specific environment variables.

Importing remote data via SOAP, authentication token or other complex methods

You can use a curl command as import resource when you want to access data from an API. Examples: SOAP:

curl --username:password \
  --header "Content-Type: text/xml;charset=UTF-8" \
  --header "SOAPAction: ACTION_YOU_WANT_TO_CALL" \
  --data <FILE_NAME> <SOAP_WEB_SERVICE_ENDPOINT_URL>

Accessing REST-API via token authentication (e.g. OAuth):

curl "<Authentication Token URL>" --header 'Authorization: Basic ABC...' \
  | jq -r '.access_token' \
  | xargs -I {access_token} curl -s  --header 'Authorization: Bearer {access_token}' "<Data URL>"'

Alternatively you can use the integrated REST API to push data from an external system into Pimcore - or in other words: to trigger the import by sending an HTTP request which includes the data to be imported in its message body.

Importing data from zip file

When you use a zip file as import resource (be it local or remote) the bundle extracts the files and determines the actual import file based on the file extensions, e.g. .csv for dataports with source data type CSV,.xml for XMLetc.). All other files get recognized as assets which can be accessed by their file path within the zip file.

Example: You have a zip file abc.zip which contains the following files:

  • import.csv
  • image.jpg
  • documents/manual.pdf

When you create an import with source data type CSV and use abc.zip as import resource, you can access the files for example with the following CSV file:

Product name;Main image;Manual
My product;image.jpg;documents/manual.pdf

The attribute mapping works the same as if the files would actually exist unzipped in a folder.

Importing data via PHP script

You can provide the path to a PHP script which gets used as import source. This script can either return the data to be imported or return a file path / URL where the data to be imported can be found. You can either enter the script directly as import resource in the UI or you can provide the path to the script (either relative based on the Pimcore root folder or as absolute file system path - whereas the relative paths are recommended because absolute paths may differ in different environments or when hosting changes).

You can even provide arguments to this PHP script, e.g. import.php Supplier1 Supplier2. In your script you can access these arguments via $arguments.

Extracting data from Pimcore elements

Sometimes Pimcore’s built-in mass-data editing is not suitable for certain requirements:

  • due to max execution time you can only edit a certain number of objects at the same time
  • when you want dynamic values depending on another object field
  • data of some field types cannot be edited in grid view

In these cases you can choose “Pimcore Elements” as source data type to import data from Pimcore data objects / assets / documents. After choosing a source class, all methods starting with “get” are provided as raw data fields. You can extend this list by overriding the data model class.

Another use-case for this feature is when you want to change a data type of a class / brick / classification store / field collection field without losing existing data. You can first import existing data as raw data, then change the class / brick / classification store / field and thereafter reimport the data to the new field.

And last but not least you can use this to implement automations / workflows for your objects - e.g. to only allow to publish objects which fulfill certain quality criteria (have a title, have images etc.). Of course, you could do this via model overriding but an import has the advantage that all things which edit an object’s state can be reviewed from the Pimcore backend - while overridden model classes often are a magic black box for Pimcore users.

The objects to be used can be restricted with the SQL condition. You can access here

  • all fields of the source class database table
  • localized fields with fieldname#<language-code>, e.g. name#en=’example’
  • object brick fields with <brick name>.fieldname, e.g. Chairs.legCount=3
  • relational fields, e.g. with source data class “Product” which has a relational field “categories” to class “Category” which has a field “name” you can have SQL condition: categories.name LIKE ‘a%’ and will only get those products which have at least one category assigned whose name begins with “a”
  • __updated to filter for last modification date of the element itself and of all its dependent elements

To create raw data from Pimcore objects you can enter a data query selector.

Importing data from Pimcore reports

You can use Pimcore custom reports as import source. Just select the report which you want to import data from, select which columns of the report to use and you are done.

Importing data from Pimcore reports has especially 3 use cases:

  1. You can use the result callback function to check for certain deviations in your report’s data and to create a response document (e.g. text or CSV) and with the result document action send this data as an email. This way you can get notified about certain problems even without continuously looking at the reports.
  2. Use custom report adapters like the Google Analytics Adapter which gets shipped with Pimcore or the adapter to import data from external databases to import data from external services / systems.
  3. Export data from any Pimcore database table like Quantity Value Units, translations etc.

Exporting data from Pimcore objects

Exporting works nearly the same as Pimcore-based imports. You can specify the fields which shall be exported via data query selectors - with the addition of being able to set the names of intermediate return fields.

Example: You use a Category class as source class for the dataport. This class contains a many-to-many relation field products to the class Product. Class Product contains an advanced many-to-many relation images (allowed types: assets).

products:each:(name;images:each as images:(Element:fullpath as path;Element:thumbnail#my-thumbnail-definition:path as url))

This would result in:

[
  {
    "name":"Product A",
    "images": [
      {
        "path":"/path/to/asset/file1.jpg",
        "url":"/var/tmp/image-thumbnails/0/1234/thumb__my-thumbnail-definition/file1.jpg"
      },
      {
        "path":"/path/to/asset/file2.jpg",
        "url":"/var/tmp/image-thumbnails/0/1235/thumb__my-thumbnail-definition/file2.jpg"
      }
    ]
  },
  {
    "name":"Product B",
    "images": [
      {
        "path":"/path/to/asset/file1.jpg",
        "url":"/var/tmp/image-thumbnails/0/1234/thumb__my-thumbnail-definition/file1.jpg"
      },
      {
        "path":"/path/to/asset/file3.jpg",
        "url":"/var/tmp/image-thumbnails/0/1236/thumb__my-thumbnail-definition/file3.jpg"
      }
    ]
  }
]

Together with result callback functions this can be used to individually configure exports.

When you enable the Run automatically on new data checkbox for export dataports, raw data will always get updated as soon as qualified objects get saved. Only the saved object gets processed during automatic raw data import, not all data objects. This way raw data is always up to date and exports become very fast as the step to fetch the desired data from the data objects is not necessary anymore in the moment of export request. You can also consider automatic exports as a CQRS system: Data to be queried gets saved in an optimized way for reading and gets fetched from the write model (Pimcore data objects) only when data changes.

Automatic exports also get triggered if an element of the dataport’s target class (which matches the configured SQL condition) is referenced by the saved object.

When you enabled Run automatically on new data for your export dataport, another checkbox Incremental Export becomes available. Non-incremental exports first fetch the data from all data objects which match the configured SQL condition and thereafter update this export data when a related object gets saved. Incremental backups do not fetch the data of all matching objects. Instead they fetch the data only from the saved objects and export this data. As soon as the result callback function does not trigger any error for the exported items and does not return false the raw data gets automatically deleted. Then when saving other objects the whole process is repeated.

Exporting data from grid configuration

This bundle ships with a grid operator Data Query Selector to use data query selectors to show data from data objects in the Pimcore grid / folder view. This is especially interesting for complex fields like relations, field collections, object bricks as it is really difficult to extract the desired data from such complex fields with Pimcore’s default grid operators.

All the grid columns can then be directly exported as CSV, JSON, XML without creating a dataport first (in the background an ad-hoc dataport will be automatically created).

Another option is to create a dataport manually and choose grid configuration as source type. The raw data fields of such a dataport cannot be changed but get dynamically fetched from the configured saved grid configuration. So when the grid configuration gets changed, automatically also the raw data fields of the dataport get changed. In attribute mapping you can then define the export format as for usual object-based imports.

Importing data from filesystem

The import type “filesystem” uses files on the server as import source (no additional CSV, XML or other meta file needed). The raw data fields result of CLI commands which get executed on the files. For example you could have a raw data field with the CLI command md5sum "$file" | cut -d ' ' -f 1 to get the md5 checksum of the file imported file. As placeholders for the file you can use $file, $filename, ${file} and ${filename}.

The absolute file path gets available when you add a raw data field named file or filename - you do not need any CLI function in this case. Another option to get the absolute file path is to use $file, $filename, ${file} or ${filename} in the CLI command field - in this case the field name does not matter.

Dynamic import resource / parametrized imports and exports

In the import resource field you can use placeholder variables via {{ variableName }}. For example you could set up an import with http://example.org/api/products/{{ itemCode }} as import resource. You have several ways to fill this variable:

  1. Call import via CLI: bin/console dd:complete <Dataport id> --parameters="itemCode=1234"
  2. Call import via REST API POST http://example.org/api/rest/import/dataport-name?itemCode=1234
  3. Configure the dataport to be executed automatically. When there is a field itemCode in the dataport’s target class, and a data object of this class gets saved, the dataport will get executed with the itemCode of the just saved object.
  4. Set environment variable itemCode=1234 and start dataport without setting the parameter explicitly
  5. Configure a website setting itemCode with a certain value. Localized website settings can be accessed via {{ itemCode#en }} (english setting of itemCode)

Actually the variables are data query selectors, this means that you can also access object fields here, e.g. when you have separate configuration data objects where you keep your API credentials, your import resource could be http://example.org/api/products/?apiKey={{ ApiConfiguration:path:/configuration/api/xyz:apiKey }}. This then would try to find a Pimcore data object of class ApiConfiguration whose path is /configuration/api/xyz and from this object retrieve the content of field apiKey. So in the end the import resource would be http://example.org/api/products/?apiKey=1234 if the retrieved object has apiKey=1234.

And as those variables are data query selectors you can also use logic operations with Twig syntax. For example you could have http://example.org/api/products/{{ itemCode|default('1234') }} to set a default value which makes developing the import easier because you will get preview data for your raw data fields and in attribute mapping.

Another example is conditions:

http://example.org/api/products/
{% if APP_ENV == "prod" %}
  {{ ApiConfiguration:path:/configuration/api/xyz-live:apiKey }}
{% else %}
  {{ ApiConfiguration:path:/configuration/api/xyz-sandbox:apiKey }}
{% endif %}

Dynamic raw data field selector

You can use __all selector to get all fields of the current raw data item in one raw data field. This way you can set up dynamic imports which use differently structured import files (e.g. CSV files with different columns according to the fields to be edited), e.g. once there is a CSV with SKU and name, next time the import CSV has columns SKU and price. In this case the __all raw data field will contain all raw data {SKU: 123, price: 3.99} and can be processed in the attribute mapping callback functions.

Archive folder

When you set a Pimcore asset folder or a filesystem folder as archive folder the import files which get used for a certain import will afterwards get saved to that folder. This makes it is easier to analyze import errors in future. If you use a Pimcore asset folder, the archive file will be directly linked from the history panel so whenever there is an error, you can directly jump to the corresponding import file in the archive folder.

You can also provide an FTP, FTPS, SFTP or AWS S3 URL:

ftp://username:password@example.org/folder1/folder2
ftps://username:password@example.org/folder
sftp://username:password@example.org/folder
s3://key:secret@region/bucket/folder

By default files older than 30 days get automatically deleted from the archive folder to not waste hard disk space. You can change this by overriding the parameter blackbit_pim.importstatus.cleanup_interval in your /config/parameters.yml.

Raw data fields

Raw data fields define which data shall be extracted from the import resource. You can specify the data to be extracted with selector queries depending on the format of the import source:

Input format Field selection by
XML / HTML XPath
CSV Column index or column heading
JSON JMESPath / JSON Pointer
Excel Column name or column heading
Fixed-length Files Field length
Pimcore objects Data query selector
Pimcore reports Report column names
Object wizards Individual form fields
Pimcore Grid configuration Dynamic field selection via grid configuration
Files CLI commands

Define order of import

When processing raw data the order of raw data fields defines the order of data processing. At first the raw data item with the “lowest” value in the uppermost field gets imported. If multiple raw data items have the same value in the uppermost field, then the import order is defined by the second field, etc. - just like SQL’s ORDER BY column1, column2 feature. This is especially useful if raw data items depend on each other (e.g. master slave data - you have to import masters first to get correct hierarchy, or for category hierarchy you have to import main categories first and afterwards sub categories). You can adjust the raw data field order by drag & drop the raw data fields to the desired position.

Side note: For imports raw data gets first sorted by modification timestamp of the import files to guarantee correct import order when importing data from multiple files at once. Raw data items with the same modification timestamp are ordered by raw data fields as described above.

When you want to import the items exactly as they are in the import file, you can use the special field __index. When you add this as your first raw data field, the items are processed in the same order as they are in the import file.

Special attributes

All source data types support the following special fields:

  • __updated: contains the UNIX timestamp of the last modification of the import resource. For file-based imports this is the modification timestamp, for Pimcore elements it is the maximum value of the modificationDates of the currently processed element itself and all its dependent elements.
  • __source: contains the file system path to the currently processed file. This can be useful when the import resource is a folder or URL but you want to forward / reuse it on a follow-up dataport.
  • __index: contains the index of the currently processed item in the context of the current raw data extraction. This is mainly being used to process import data in the same order as in the import file.
  • __all: contains all import data in one field. This can be useful if the structure of the file to be imported is dynamic (e.g. for example when importing technical data for products of different product groups, when the products of each group are in separate files which differ in the columns as they have different technical data attributes)
  • __count: Number of total items. When importing multiple files this will be the total count of items of all processed files.

Attribute mapping

In the attribute mapping panel you will see all fields of the dataport’s configured target class or of the export document. In the column Raw data field you can assign a raw data field. If you want to alter the value of the raw data field during the import / export, you can add a callback function by clicking the Settings icon icon.

Modifying data via callback functions

In the settings dialog for each field in the attribute mapping panel, you can specify a callback function to modify / transform the data that will be processed during the import.

You only have to enter the function body in the callback function code editor field. All parameters passed to the function are properties of the $params object.

$params['value'] *= (1+$params['rawItemData']['tax']['value']);

return $params['value'];

You will find a list of all available variables on the right side of the code editor.

Data passed into the callback function as argument (properties of the $params object)

  • value - The raw data value as imported from the source file
  • rawItemData - Array with all raw data values for this item. Access data by field index as specified in the data parsing configuration or by raw data field name, e.g.: rawItemData['field_1'] or rawItemData['name']
  • currentValue - The attribute value currently set in the object which got found by set key fields
    • class inheritance and fallback languages do not get used for this field
  • currentObjectData - Array with all current values from the found object. Access data by field name from the class definition:
    • currentObjectData['price'] for class fields
    • currentObjectData['name#de'] for localized class fields
    • currentObjectData['brickContainerField']['brickName']['brickField'] for object brick fields
    • If the value in the dependent field gets changed during the import, currentObjectData gets updated (import field order gets adjusted according to mapping dependencies - when in the callback function of field name#en currentObjectData['name#de'] gets accessed, name#de gets imported before name#en)
    • class inheritance and fallback languages get used for this variable
  • keyValues - Associative array with key values indexed by key field names, especially important when creating / editing multiple objects from one raw data item
  • request - Request object which may contain URL / CLI parameters
  • transfer - Object for storing data to be reused / processed on later raw item datasets, the object is initially empty

Skipping items

You can skip items by returning null in the callback function of a key field. If you have multiple key fields and at least for one of them null gets returned, the raw data item is skipped.

Importing to non-scalar field types

For most non-scalar fields this bundle provides templates which you just have to select. Only for more complex logic you would need to edit the callback functions after selecting a template.

Object hierarchy / parent element

With the attribute mapping for field Path you can specify the parent element of the imported element. You have multiple options to specify the parent element:

  • via data query selector, e.g. with callback function return 'Category:category_id:'.$params['value'];
  • via string (or by just assigning a raw data field without callback function):
    • if this string’s first character is not a / then the configured target folder for this dataport will get prepended (mapped path is relative to configured target path), e.g. you return fashion and the dataport has /categories as target folder, then the imported object will be put under /categories/fashion
    • if it starts with a / then the configured target path gets ignored (absolute path)

If no object is found, then it will be created. If you use a data query selector, then an object of the given class (the part before the first colon in the data query selector) will be created. This object has as the only filled field the filter field used in the data query selector. This way it will get found again if you repeat the import or if also other raw items refer to this object by data query selector. When you later import the real object the auto-created object gets updated. If you return a string (and not a data query selector) and there isn’t any folder with this path, then it will get created.

Single object relation (Href / Many-to-one relation)

Manual Assignment

To assign objects based on raw data it is necessary to provide some information how to find these objects. In addition to the raw data field value Pimcore needs to know of which class the desired object is and which field to use for querying, example:

return 'ClassName:fieldName:'.$params['value'];

This code tries to find an object of class “ClassName” which has fieldName = imported raw data field.

It is also possible to find objects by a value in a localized fields:

return 'ClassName:fieldName#locale:'.$params['value'];

If there are colons within the query value, you should wrap the value with quotes:

return 'ClassName:fieldName:"'.$params['value'].'"';
Automatic assignment

With the checkbox Automatically assign you can automatically assign elements to the relation field based on artificial intelligence. The training data consists of the data of all mapped fields of this dataport of all published objects of the same class as the imported object. This can e.g. be used to automatically assign categories of a product based on the product name and description.

Many-to-many object relation

Assigning data to a many-to-many object relation works the same as many-to-one relation but you have to return an array.

return [
    'ClassName:fieldName:'.$params['value'],
    'ClassName:fieldName:'.$params['rawItemData']['otherRawDataField']['value']
];

By default, elements get added to the relation. In cases where you want to keep some of the currently assigned items while updating others, please enable Truncate before import checkbox and then loop through currently assigned items via $params['currentValue']. The ones which you want to keep, you put into an array. To this array you also add the ones which the import shall newly assign. Then you return the array and have your desired items in the relation.

Object relation with metadata (Advanced many-to-many object relation)

The definition how to find the object to be assigned works the same as with normal many-to-many object relation with the only difference that the return value has to contain the object query string in the field query. Other meta columns can be set via key-value pairs within this JS object. Example:

return [
  [
    'query' => 'Person:email:'.$params['value'],
    'metaDataFieldName' => $params['rawItemData']['rawDataColumn']['value']
  ]
];

This adds a relation to an object of class Person which has the given raw data value in its field email. Additionally the metadata column metaDataFieldName is set to the value of raw data column rawDataColumn.

To assign an asset to an image / image gallery / many-to-many relation field you have to provide an array of file paths or URLs under which the assets are accessible. The importer tries to load it and imports the asset to the target folder (configurable under dataport settings). The structure of the array to be returned follows the same logic ad for image fields, e.g.

$images = explode(',', $params['value']);
return array_map(
    function($imageUrl) {
        return [        
            'url' => $imageUrl,        
            'filename' => 'somePrefix_'.basename($imageUrl)
        ];
    }, $images
);

If you want to search for existing assets, you can provide a data query selector instead of the url:

$images = explode(',', $params['value']);
return array_map(
    function($imagePath) {
        return [        
            'query' => 'Asset:path:'.$imagePath
        ];
    }, $images
);

Image field 

return $params['rawItemData']['imageUrl']['value'];

The given image file is tried to be found in the configured asset source folder. But you can also return absolute path to the file.

You can also retrieve files via URL:

return 'https://example.org/'.$params['rawItemData']['imageUrl']['value'];

// or, if you want to save the file under a different name or folder:
return [
    'url' => 'https://example.org/'.$params['rawItemData']['imageUrl']['value'],
    'filename' => $params['rawItemData']['name']['value']
];

Providing a filename does also support dynamic folders. So when you provide /images/abc.jpg in the filename parameter, the asset will get the name abc.jpg and will be created in / moved to the asset folder /images. When you provide a relative path like images/abc.jpg the image will be put in the folder images below the dataport’s configured asset target directory.

You can also assign already existing Pimcore assets via:

return [
    'query' => 'Image:path:'.$params['value']
];

It is also possible to find assets by custom metadata:

return [
    'query' => 'Image:external_system_id:'.$params['value']
];

external_system_id is a custom metadata field of the asset.

Optionally you can return other fields than query, url and filename - those will be assigned as custom metadata to the image asset, e.g.

return [
    'url' => $params['value'],
    'product_sku' => $params['rawItemData']['sku']['value'],
    'product_name' => [
        'de' => $params['rawItemData']['name de']['value'],
        'en' => $params['rawItemData']['name en']['value'],
    ]
];

This example would result in the asset being created from the filename in the assigned raw data field. To this asset the custom metadata product_sku gets set as well as the localized custom metadata fields product_name. This way you can for example find the asset by searching for the product name.

Assigning images to image gallery fields works the same as for images - except that you can return multiple items in an array, e.g.:

return [
    $params['rawItemData']['image 1']['value'],
    $params['rawItemData']['image 2']['value'],
];

or as a more complex example:

return [
    [
        'url' => $params['rawItemData']['image 1']['value'],
        'filename' => $params['rawItemData']['sku']['value'].'_1.jpg',
        'product_sku' => $params['rawItemData']['sku']['value'],
    ],
    [
        'url' => $params['rawItemData']['image 2']['value'],
        'filename' => $params['rawItemData']['sku']['value'].'_2.jpg',
        'product_sku' => $params['rawItemData']['sku']['value'],
    ]
];

Many-to-many relation fields

For many-to-many relation fields the same format as for image fields applies but you can provide multiple entries in an array:

return [
    'https://example.org/'.$params['rawItemData']['imageUrl_1']['value'],
    'https://example.org/'.$params['rawItemData']['imageUrl_2']['value']
];

// or, if you want to save the file under a different name:
return [
    [
        'url' => 'https://example.org/'.$params['rawItemData']['imageUrl_1']['value'],
        'filename' => $params['rawItemData']['name']['value'].'_1.jpg'
    ],
    [
        'url' => 'https://example.org/'.$params['rawItemData']['imageUrl_2']['value'],
        'filename' => $params['rawItemData']['name']['value'].'_2.jpg'
    ]
];

Asset relation with metadata / Advanced many-to-many relation

To assign an asset together with some metadata you have to return an array of objects. The asset filename / URL has to be in field url of this object. To find already existing assets you can use the field query (as described above for asset relation fields). Metadata fields can be set with key-value pairs in the returned object. Example:

return [
  [
    "url" => $params['rawItemData']['image-url-field']['value'],
    "metaField1" => "ABC",
    "metaField2" => $params['rawItemData']['rawDataColumn']['value']
  ]
];

If you want to change the filename of the asset which gets created by the import you can use the field filename:

return [
  [
    "url" => $params['rawItemData']['image-url-field']['value'],
    "filename" => "my-new-file.jpg"
  ]
];

It is also possible to retrieve existing Pimcore assets via:

return [
    [
        'query' => 'Image:path:'.$params['value']
    ]
];

For asset querying same rules apply as for object querying: <Asset class>:<Field to filter by>:<Filter value>.

Video

Pimcore’s video field type supports different sources of the video. In general videos can be imported via

return [
    'type' => 'asset', // or youtube, vimeo, dailymotion
    'data' => 'path to file / URL' // or video ID
];

This works the same as for image fields.

You can also provide the URL of the video, then it will automatically get parsed, e.g. you can provide strings like:

  • https://youtu.be/nyhKJTzTq-4 -> will get parsed as type “youtube” and data “nyhKJTzTq-4”
  • https://another-pimcore.org/var/assets/my-video.wmv -> will load the video from the remote system and create it as a local asset

Quantity value / Input quantity value

To fill a quantity value field you have to return an array with the value and the unit abbreviation in the callback function. Alternatively you can return a string with value and unit separated by a whitespace. For example:

return [$params['value'], 'mm'];
return $params['value'].' mm';

If the given unit does not exist yet, there is an option to automatically create units. If possible, the unit gets created with base unit and conversion factor / offset - ready to support automatic unit conversion.

If you do not return an array or set the second array item to an empty value, the default unit of the field gets used (if configured in the field definition).

Field collections

Field collection values can be assigned by providing a JSON array with objects whose keys represent the fields of the field collection, for example:

return [
    'fieldCollectionName' => [
        [
            "field1" => $params['rawItemData']['field_1']['value'],
            "field2" => $params['rawItemData']['field_2']['value'],
        ],
        [
            "field1" => $params['rawItemData']['field_3']['value'],
            "field2" => $params['rawItemData']['field_4']['value'],
        ]
    ]
];

The importer tries every field collection type which is allowed in the target class field. If it encounters a field returned by the callback function which does not exist in the field collection it tries the next allowed field collection type.

Field collection items only get added when at least one field is filled and if no other currently existing item has the exact same field values. In cases where you want to keep some of the current items while updating others, please enable Truncate before import checkbox and then loop through currently assigned field collections items via $params['currentValue']. The ones which you want to keep, you put into an array. To this array you also add the ones which the import shall create. Then you return the array and have your desired items in the field collection.

Localized fields within field collection 
return [
  'fieldCollectionName' => [
    [
      "field1" => $params['rawItemData']['field_1']['value'],
      "field2" => $params['rawItemData']['field_2']['value'],
      "field3_localized#de" => $params['rawItemData']['Text_DE']['value'],
      "field3_localized#en" => $params['rawItemData']['Text_EN']['value']
    ]
  ]
];
Asset relation within field collection 
return [
  'fieldCollectionName' => [
    [
      "many-to-many_relation_field" => [
        "Image:path:/path/to/image/".$params['value']
      ]
    ]
  ]
];
Asset / object relation in localized field in field collection 
return [
  'fieldCollectionName' => [
    [
      "many_relation_field#de" => [
         "Image:path:/path/to/image/".$params['value']
       ],
      "many_relation_field#en" => [
        "Image:path:/path/to/other/image/".$params['value']
      ]
    ]
  ]
];
Quantity value field within field collection 
return [
  'fieldCollectionName' => [
    [
      "quantity_value_field" => [[123, 'EUR']]
    ]
  ]
];

Blocks

Block field values can be imported with a key-value array:

return [
    [
        'field1' => 'first block item, value for field1', 
        'field2' => 'first block item, value for field2',
        'relationalField' => 'PriceList:customerNo:1234'
    ],
    [
        'field1' => 'second block item, value for field1',
        'field2' => 'second block item, value for field2',
        'relationalField' => 'PriceList:customerNo:2345'
    ]
];

The provided values are parsed through the corresponding field type interpretation. For this reason assigning an object to the field relationalField via a data query selector works in above example.

Object bricks / Classification Store

All the single fields of the allowed object bricks / classification store groups get displayed in the attribute mapping panel and can be mapped individually. You can treat them like normal class fields so all the things listed here for other field types are valid also for object brick / classification store fields. As soon as at least for one of the fields of an object brick / classification store group data gets imported the brick / group will get added to the current object.

Alternatively you can directly map your data to the object brick / classification store container field by returning an associative array whose keys are the brick / classification store group names:

PHP:

return [
    'brickName' => [
        'field1' => $params['rawItemData']['field_1']['value'],
        'field2' => $params['rawItemData']['field_2']['value'],
        'name#de' => $params['rawItemData']['name de']['value'], // name is a localized field
        'name#en' => $params['rawItemData']['name en']['value'],
    ],
];

Convenience features like resolving data query selectors are applied to all the provided fields, e.g.

return [
    'technicalAttributes' => [
        'cableLength' => 
            'AttributeValue:externalId:'.$params['rawItemData']['id']['value'].':value'
    ],
];

This will search for an object of the class AttributeValue whose externalId equals the value from the raw data field id and import this to the field cableLength of the object brick technicalAttributes.

Or if the field name is part of the import data you can assign it dynamically:

return [
    'technicalAttributes' => [
        $params['rawItemData']['attributeName']['value'] => 
            $params['rawItemData']['attributeValue']['value']
    ],
];

If a field is mapped in the object brick / classification store container field and as a single field, only the single field mapping will get applied.

If a field is mapped as a single field but the container mapping does not return the brick / classification store group name of the mapped single field, the single field will get skipped.

Example:

Mapping for container field returns:

[
    'Shoes' => [
        'size' => 43,
        'color' => 'blue'
    ]
]

When the field Shoes/size is not mapped as a single field, the value from the container field will get used.

When there is a single field mapping for Shoes/color, this single field mapping will get used instead of the value from the container field. This way you can set import options like Automatically create missing options if color is a select field.

When there is a mapping for Clothing/color (= for a field of another brick / classification store group), this will get skipped because the callback function for the object brick / classification store container does not return brick / group Clothing. Of course this only applies if the container field is mapped. If you only map single object brick / classification store fields, they will get applied anyway.

The data of a Link field can be set by returning an array whose keys refer to the properties of the Pimcore\Model\DataObject\Data\Link class, for example:

return [
    'path' => 'http://example.org/page',
    'target' => '_blank',
    'text' => 'Link text'
];

Geographic Point

To set a value to a geographic point field you have to provide the geographic coordinates of the point. Several formats are supported, e.g.

  • 40:26:46N,079:56:55W
  • 40:26:46.302N 079:56:55.903W
  • 40°26’47”N 079°58’36”W
  • 40d 26’ 47” N 079d 58’ 36” W
  • 40.446195N 79.948862W
  • 40.446195, -79.948862
  • 40° 26.7717, -79° 56.93172

Also geocoding is supported. When the bundle does not recognize the given value as valid coordinates it assumes that it is an address and tries to convert this to coordinates. This is done either with

Before Pimcore 10 you can also set this value in the Pimcore backend under System settings. Information how to get a Google Maps Geocoding API key.

Select / Multiselect fields

To assign a value to a select field you should return the option’s value. If no option with the given value can get found, it is tried to find an option with the given label.

For multiselect fields the same logic gets applied but you have to provide the option separator. Or if you use a callback function, you have to return an array of values:

return ['option 1', 'option 2'];

The same applies for all field types which extend select or multiselect like User, Language, Country, Country (Multiselect) etc.

If you enable the checkbox Automatically create missing options in the attribute mapping settings, new options will automatically get created. For example this makes sense when another system is the leading system for the available options of this field.

URL slug

For URL slug fields you can either return a string (or only design a raw data field without any callback function) to assign the URL slug for all sites.

Alternatively you can return an array with the key being one of the configured domains of a site and the value being the URL slug:

return [
    'example.org' => '/en/my-product',
    '' => '/my-product' // fallback URL slug for all sites without a specific value
];

Automatic translation

In attribute mapping there is an option for automatic translation of text fields (Input, Wysiwyg, Textarea). You can assign a raw data field with a certain text and define its language in the text field mapping options to get the raw data text automatically translated to the target language of the localized field (for objects) or the language of the document (for document imports).

Translation Provider Configuration

DeepL

You will need an DeepL API key to use this feature. Please add your API key in attribute mapping panel or via config/services.yaml:

parameters:
    ...
    blackbit_pim.deepl_api_key: <Your API key>

DeepL glossaries will automatically get applied for the current pair of source and target language. You do not have to configure anything but it will get automatically checked if a DeepL glossary exists. If you have multiple glossaries for the same source and target language, the first one will get applied.

The DeepL API supports to ignore certain words for translation. This can be useful for brand names, product names or other phrases which shall not be translated. By default, you can mark such phrases by wrapping them into <x> tags. You can change the tag for ignored phrases by overriding the parameter blackbit_pim.skip_translation_tag in your app/config/parameters.yml:

parameters:
    ...
    blackbit_pim.skip_translation_tag: dfn

Afterwards clear the cache (via Pimcore UI or bin/console cache:clear).

Amazon (AWS) Translate

To use Amazon Translate you need to set up your AWS credentials - if you decide to configure credential information inside Pimcore’s configuration files, please put your access key and secret in app/config/parameters.yml:

parameters:
    ...
    blackbit_pim.aws_translate_access_key: <Your access key>
    blackbit_pim.aws_translate_secret_key: <Your secret key>

Afterwards clear the cache (via Pimcore UI or bin/console cache:clear).

Exclude terms from translation / Manually translate phrases

When you want to exclude some phrases from translation or are not satisfied with the translation, you can exclude phrases from being translated or provide custom translations. To achieve that you have to create translations in Pimcore’s shared translations whose key starts with translate..

  • When in the target language no translation is set, the phrase will be excluded from translation, so the translation in the source language will get used. If this is also empty, the translation key (without translate. prefix) will get used
  • When there is a translation in the target language, this will get used (internally during translation the phrase in the source language gets ignored and after translation it gets replaced with the custom translation from Pimcore’s shared translations)

Translate complex data / Manipulate translated content

In some cases it might become necessary to translate certain strings via a function call. For example this is useful for complex datatypes like when you want to translate asset metadata or field collection data. This way you can implement your custom logic for filtering which fields have to be translated and then call the \Blackbit\DataDirectorBundle\lib\Pim\Item\Importer::translate function:

// Translate asset metadata from german to other languages

use \Blackbit\DataDirectorBundle\lib\Pim\Item\Importer;

$sourceLanguage = 'de';
return array_map(function($value) use ($sourceLanguage) {
    if($value['language'] === $sourceLanguage) {
        $newMetaData = [];
        foreach(\Pimcore\Tool::getValidLanguages() as $targetLanguage) {
            if($targetLanguage === $sourceLanguage || $value['type'] !== 'input') {
                continue;
            }
            
            $translation = Importer::translate(
                $value['data'], 
                $targetLanguage, 
                $sourceLanguage
            );
            
            $newMetaData[] = [
                'name' => $value['name'], 
                'data' => $translation, 
                'type' => $value['type'] ?? 'input', 
                'language' => $targetLanguage
            ];
        }
        return $newMetaData;
    }
}, $params['value']);

You can also use this function when you want to manipulate the translation afterwards, e.g. uppercase first letter of the translation for product names:

use \Blackbit\DataDirectorBundle\lib\Pim\Item\Importer;

$targetLanguage = 'de';
$sourceLanguage = 'en';

$translation = Importer::translate($params['value'], $targetLanguage, $sourceLanguage);
return ucfirst($translation);

Map languages

Normally it is tried to use Pimcore’s configured language including locale. If the translation provider does not support the given locale, it will fall back to the main language.

But when for instance in Pimcore English got added as language, it is not clear which locale shall be used for translation. In such a case you can map Pimcore languages to translation target languages in config/config.yaml (or app/config/config.yml on Pimcore versions <= 6):

blackbit_data_director:
    translation:
        languages:
            en: 'en-gb'

Text generation

Rule-based

There are 2 ways to generate texts based on rules:

  1. Set up an automatic Pimcore-based import with the fields, which you want to use in your text generation rules, as raw data fields. In the callback function of the target field you can generate the text.
  2. Add a field of type “Textarea / WYSIWYG field with variables”. This allows to access the content of other fields directly from the data object editing view. It supports logic operations (like conditions, loops etc.) via Twig template engine syntax. The rules may be inherited to child objects, so you can have a different set of rules for different sub-trees.

Artificial Intelligence

For textarea and wysiwyg fields there is a checkbox “Automatically generate” in the field settings in Data Director attribute mapping. With this you can connect to a text generation API (like OpenAI). The API needs some input to know what to do - the more data you provide, the better the API knows what to do.

Infer technical data from texts

After creating the target state in your data model, it is often a lot of work to extract this information for your existing data. Artificial intelligence can help to extract information for single fields from textual information.

Example:

Input:

  • Product name: Wrangler Men’s Casual Shirt
  • Product description: Long-sleeved cotton shirt with navy and orange stripes

Result:

  • Material: cotton
  • colors: orange, blue (because the colors field does not allow navy but it allows blue)
  • gender: male

This data can now be used for filters or structured exports.

Create texts based on technical data

Especially manufacturers often have the challenge that they know all details of their products exactly but struggle with creating good texts. But also for distributors / resellers it is a big challenge to create good texts because multiple distributors get the same texts from a certain supplier - so how to stand out?

Example:

Input:

  • Name: Wrangler Men’s Casual Shirt
  • Material: cotton,
  • Colors: Blue, Orange
  • Gender: Male

Output:

The Wrangler Men’s Casual Shirt is a comfortable and stylish choice for any man. Made from high-quality cotton, it offers breathable comfort that will keep you feeling great all day long. Available in two vibrant colors - blue and orange - this shirt is perfect for casual wear or dressier occasions. Designed with the male physique in mind, it provides a flattering fit that looks great on every body type. Whether you’re heading to work or out with friends, the Wrangler Men’s Casual Shirt has got you covered!

Modify the prompt

Behind the scenes to configured input data in only one part of the prompt. Some context information like class name, object name etc. will automatically get added. Also limitations (e.g. select options, length limit) get added automatically.

If you prefer to create the prompt on your own, you can add <no-prompt-extension> together with your custom prompt in the callback function.

Tags

You can import one or multiple tags. Tag levels should be separated by /. You can either just assign a raw data field or use a callback function like this

return ['Tag-Name','hierarchical/tag'];

Tags which do not exist yet, get automatically created.

Properties

You can import properties by returning a key-value array:

return [
  'Name' => 'Product A',
  'imported' => time(),
  'product name' => 'MyDataObjectClass:path:/path/to/object:fieldA', // data query selectors get resolved
  'product' => 'Product:sku:'.$params['rawItemData']['sku']['value'] // for relations correct property type gets set automatically
];

This will end up in the 3 properties being assigned to the imported element.

Alternatively you can provide all supported property attributes:

return [
  ['name' => 'parameter name', 'data' => 'Parameter value', 'inheritable' => true],
  ['name' => 'parameter 2', 'data' => '123', 'inheritable' => false]
];

Dependency graphs

When a certain field gets changed by multiple dataports it can be difficult to keep track of all the dependencies. In this case it might be helpful to visualize all the dataports which use a certain field. This can be accessed by clicking the field name in the attribute mapping panel, the result looks like this:

Dependency graph for field published

In this example the dependency graph for the field published of class Product is shown. In this case 2 dataports (Item Import ERP, Data Quality Check) import to the field published. Item Import ERP might also be called as a dependent import of dataport Pipeline import. On the right side we see that the dataport Export products to shop accesses the field published.

Import pipelines

In the result callback function there is a template for setting up a dependent dataport which can be called during a run of the current dataport - either after every raw data item or at the end. If necessary you can provide parameters to be used in the import resource or callback functions of the follow-up dataport.

Edge cases

Importing data to objects whose newest version is unpublished

For objects with their newest version being unpublished for every raw data item import gets executed twice:

  1. Update currently published version und create a new published version
  2. Update newest currently unpublished version

This way the unpublished changes do not get overwritten while the object data from the import gets applied for both versions.

Creating / Editing multiple objects with one raw data item

Returning an array for a key field will be treated as if there were multiple raw data items with those key field values.

Example: You have an XML import with products:

<products>
    <product>
        <name>Product A</name>
        <colors>
            <color>red</color>
            <color>blue</color>
        </colors>
    </product>
</products>

With item XPath product and raw data field color having the XPath colors/color you would get an array in the raw data field color. This shall be your key field for objects of the class Color. In this case the array in the key field color will be treated as if there were two raw data items - one with color: red and one with color: blue. Consequently this will result in two Color objects being created - one for each color.

If you have multiple key fields and one or more return arrays, all combinations will get created.

Example: If above XML import had another key field size with the values ['S','M','L'] the import would create objects for all combinations:

  1. color: red, size: S
  2. color: red, size: M
  3. color: red, size: L
  4. color: blue, size: S
  5. color: blue, size: M
  6. color: blue, size: L

Data Query Selectors

Syntax

Data query selectors simplify fetching other objects (e.g. for assigning them to relational fields) and fetching other objects’ data. To fetch an object you can use the following expression:

Class:filterField:filterValue

This means that you look for an element of class Class which has the value filterValue in its field filterField. In reality this could be Product:sku:1234 - this returns the Productobject which has sku=1234 - if this exists.

When you want to fetch data of the same or another object you can use the following syntax:

Class:filterField:filterValue:dataField

This would query a data object of class Class which has filterField = filterValue. From the found object you get the value in the field dataField returned. If the getter method needs parameters, those can be provided as follows:

Class:filterField:filterValue:dataField#argument1,argument2

This way you can also access localized fields via

Class:filterField:filterValue:name#en

This will give you the English name of the found object. When you do not provide the language to be used for a localized field, the current language of the request gets used. You can configure this when you start the dataport run. This way you can implement one export and use it for different languages.

If the export language has not been supplied (for example when using the REST API), the language of the requesting user will get used. And if even this is not set (e.g. in CLI processes), the configured default language will get used.

Query with wildcard value

You can also use wildcard placeholders at the beginning or end of the filterValue, e.g.

// will find Product whose name begins with "car" like "car red" and "card game"
Product:name:car*

// will find a product whose name ends with "bike" like "Trekking Bike" (case does not matter)
Product:name:*bike

// will find a product whose name contains "late" like "Chocolate Cake", "Chocolate", "Late Macchiato"
Product:name:*late*

Query in subset

The filterValue can also be a comma-separated list of values. In this case an IN (...) SQL will get executed. To enable this mode, wrap the filterValue in square brackets like

// will find Product whose name begins with "car" like "car red" and "card game"
Product:id:[123,234]

Data query selector chaining

You can chain these data selectors as long as you want. For example Product:articleNo:123:crossSellingProducts:0:manufacturer:name#de first searches for an object with articleNo=123. From this object the first item of the crossSellingProducts many-to-many relation gets fetched. From this object we fetch the assigned manufacturer object (many-to-one relation) whose german name finally gets returned.

It is also possible to apply PHP functions to the returned values, for example to strip HTML tags from a wysiwyg field:

Class:filterField:filterValue:wysiwygField:strip_tags

This way the content of the wysiwyg field gets piped as first argument to the strip_tags function. If you do not want to provide the value as first parameter or want to call a function with multiple parameters, you can use %s as placeholder for the value and comma as parameter separator. For example to get the first 100 characters of a text field:

PHP:

Class:filterField:filterValue:textField:substr#%s,0,100

If you want to use a comma as argument, please enclose it with quotes: PHP:

Class:filterField:filterValue:multiSelectField:implode#",",%s

Also, those function calls can be chained, for example to strip HTML tags and afterwards convert line breaks to <br> you could use:

PHP:

Class:filterField:filterValue:wysiwygField:strip_tags:nl2br

For more complex tasks you can also call your own service methods from a data query selector:

// via Fully qualified class name
Class:filterField:filterValue:imageGallery:items:\AppBundle\Service\MyService::methodName

// via Symfony service
Class:filterField:filterValue:imageGallery:items:@MyService::methodName

Grouping fields

When you want to fetch values from multiple fields you can use the following data query seelctor: PHP:

Class:filterField:filterValue:(dataField1;dataField2)

This can especially be useful for exports.

Field aliases

For exports it sometimes is useful to define an alias for a value which got queried with a complex data query selector. PHP:

Class:filterField:filterValue:(category:image:thumbnail#mainImage as thumbnail)

This way you can access the fetched category image thumbnail with $params['value']['thumbnail'] instead of $params['value']['category:image:thumbnail#mainImage']

Placeholders within data query selectors

It is possible to access data from the request or the source class object and use it as parameter for the data query selector.

For instance when you set up a read API for thumbnails you can create an assets-based dataport with a raw data field whose data query selector is thumbnail#{{ definition }}:path. The thumbnail definition to be used can be provided via URL parameter definition. Of course this also works with environment variables, Pimcore website settings or CLI parameters. Default values can be provided via Twig syntax thumbnail#{{ definition|default('default') }}:path - in this case if parameter definition has not been provided the thumbnail definition defualt will get used.

You can even access data via data query selectors, e.g. myMethod#{{ sku }}. This would call method myMethod with the content of the field sku of the currently processed source class object. URL / CLI / env parameters have higher priority than data object fields, so when in this case a parameter sku gets provided, this will get used instead of the data object field. Chained dynamic data query selectors like myMethod#{{ categories:name }} are also possible.

This way you can also find objects with the same value as the current object in a certain field via Class:supplierSku:{{ supplierSku }}:each:(images). Such selectors can be useful for golden record use-cases.

Shortcuts

For convenience there are also some shortcut versions for data query selectors:

  • if you enter . as Class the import’s target class gets used
    • e.g. Product:articleNo:123 is equal to .:articleNo:123 for imports with target class Product
    • this can lead to the problem that there is a field Product in the target class, then it is not clear if Product:articleNo:123 means:
      • fetch content of field Product and from the result fetch field articleNo and following data query selector chain or
      • fetch Product objects which have articleNo = 123
      • -> in such cases where there is a target class field as well as a class with the name before the first colon, the first of above examples has priority. If you want to query data like in second case, you can use Product:getByArticleNo:123
  • if you enter a . for filterField and filterValue
    • the currently imported object gets used
      • e.g. Product:.:.:name:de is equal to Product:articleNo:123:name:de when the import currently processes a Product object with articleNo=123
  • $ is a shortcut for .:.:.:
    • e.g. you can use $name in a text and get this replaced by the content of the field name of the current object.
      • $categories:0:name fetches the name of the first assigned object in the relation categories of the object which currently gets imported.
  • you can fetch data from reverse-related objects via Category:products:. - for an export with source class Product this will find corresponding Category objects which contain the currently processed Product object in the relation products. Combine it with each to get data from all found reverse-related objects, e.g. Category:products:.:each:(name) will return an array with the names of all category objects which refer to the current product in their man-to-many object relation field products.

Example: For a Pimcore-based import you can access .:.:.:name as raw data field to get the name of the object being currently imported. You could for example use this as placeholder variable in a description text template which gets filled by a Pimcore-based import. Alternatively you can use the even shorter version $name.

Helpers

Get label from select field option

If you want to access the label of a select / multiselect field option you can access this via :label suffix, for example Product:articleNo:123:mySelectField:label whereas Product:articleNo:123:mySelectField:value and Product:articleNo:123:mySelectField both return the selected option id. To access the translated option label you can call Product:articleNo:123:mySelectField:label#de (or any other locale) - this accesses the admin translation of the label.

Loop through all array items

Array items can either be processed individually by index via myManyToManyRelation:0:name, myManyToManyRelation:1:name or via the each helper: myManyToManyRelation:each:(name). This fetches the name of all items in the relation field and returns them as an array. You can also chain the data query selector after the all selector, for example myManyToManyRelation:each(manufacturer:name) (beware the braces).

Taking this concept further you can also group multiple data query selectors by using myManyToManyRelation:each:(articleNumber;name;manufacturer:name). This returns the following (here written in JSON notation):

[
  {
  "articleNumber": 123,
  "name": "Product A",
  "manufacturer:name": "Manufacturer A"
  },
  {
  "articleNumber": 234,
  "name": "Product B",
  "manufacturer:name": "Manufacturer B"
  }
]

This can be useful for XML or JSON exports with nested tags (the raw data field with above data query selector is named products in this case):

<item>
  <other_field>any value</other_field>
  <products>
    <item>
      <articleNumber>123</articleNumber>
      <name>Product A</name>
      <manufacturer_name>Manufacturer A</manufacturer_name>
    </item>
    <item>
      <articleNumber>234</articleNumber>
      <name>Product B</name>
      <manufacturer_name>Manufacturer B</manufacturer_name>
    </item>
  </products>
</item>

Access previous version’s data

latestVersionWith

For reviewing / approval workflows objects can have multiple states at the same time:

  • Initially the object’s data is unapproved. This can be modeled with a checkbox or a select field.
  • After review, the status gets set to approved.
  • Later the data will get updated and the status will become unapproved again.

But at this time it may still be necessary to access the latest approved data. This can be done with the data query selector latestVersionWith. For example, to retrieve the latest approved name of the object, you can use latestVersionWith#status,approved:name:

  • latestVersionWith#status,approved will retrieve the latest version which has status=approved
  • and from this version the name gets retrieved

This way you can create exports which only export approved data of all objects which match the dataport’s SQL condition. But beware: The SQL condition always applies to the currently published version (because only this data is inside Pimcore’s database, the version data is stored on file system).

If you want to use a different comparison operator, use the third parameter, e.g. latestVersionWith#validUntil,2024-01-01,>=.

before

With the before helper you can access data of the previous version of an object, e.g. before:published will return if the object was published in the version before the currently published one. This can be used to trigger actions when certain fields change, for instance you could set up an automatic dataport with condition o_published=1 and raw data field before:published. As soon as someone publishes an object (so that SQL condition o_published=1 is true) which was not published before (if this is the case you see in the raw data field with data query selector before:published), you can now trigger some action.

You can fetch another version than the next-to-last version via before#2024-01-01. This becomes especially handy for workflows where the last review date is atored in a date or datetime field. For example for a review date field named reviewedAt the data query selector before#{{ reviewedAt }} will fetch the version which was valid at that date.

diff

For workflows it may be useful to know exactly what has changed between two versions. The diff data query selector helper outputs the difference of a field value. In combination with the before helper this is really powerful: For example before:diff#name will return the diff of the field name between current and the version before as HTML code

Old Name
New Name
changes

The changes#<fieldName> selector can be used to compare the current value of an object to a previous time or draft version.

If you are only interested in the changes to a certain field, you can provide a data query selector, e.g. changes#name will return a JSON object

{
  "date": "2023-09-14 13:49:31",
  "oldValue": "ABC123",
  "newValue": "ABC123-DEF",
  "diff": "ABC123<ins>-DEF</ins>",
  "user": "john.doe",
  "versionId": 123
}

Localized fields are supported via lastChange#name#de.

To only retrieve changes after the object got reviewed last time, please store the review date into a date or datetime field reviewDate (or any other name but use the same name in the following data query selector) and then you can use changes#name,{{ reviewDate:timestamp }}.

When using changes without any arguments, it will return the complete previus and current state. In the field diff you will get an HTML diff table of the differing field values. This can be used for review workflows. You can use the data query selector changes:diff for this use case.

To compare a version of a certain date, you can provide a date, e.g. changes#2024-01-01 but this becomes even more powerful in combination with dynamic replacement like changes#{{ reviewedAt }} if your data object class has a date field named reviewAt. In this case the changes selector will compare the version from this date (or older) with the current version.

Group fields

Aliasing field names does not only work with the each helper but also when you want to extract data from multiple fields in one raw data field: description as html;description:strip_tags as text returns for example

{
  "html": "Our <b>new</b> product",
  "text": "Our new product"
}

It is also possible to fetch multiple fields and assign an array function to this group, e.g. (title;description):implode#<br>,%s. This would fetch the title and description of the object and concatenate them using an HTML line-break.

Enable / Disable inheritance for single fields

With the helper functions withInheritance / withoutInheritance it is possible to enable / disable inheritance for single data query selectors / raw data fields. The set status stays active until the opposite function gets called or the processing of the current data query selector is finished.

For example the data query selector withoutInheritance:sku would fetch the SKU of the product object. If the field of the current object is empty and the dataport and data object class support inheritance and the parent object has sku filled, you would normally get this inherited id which makes it difficult to tell if the given sku is from the current object or inherited from one of its parent. With withoutInheritance you can disable inheritance this way and really get the empty string.

Get all languages for localized field

While you can request a specific language for example with the data query selector name#en, you can get the value of the field in all configured languages via name#all. This will give you

{ "de": "Fahrrad", "en": "Bicycle" }

Get closest parent / ancestor element of certain class

Assuming you have a tree structure like

  • Brand
    • Product

you can use data query selector parent to access data of the parent element. But tree structures sometimes look like this

  • Brand
    • Folder
      • Master Product (class Product)
        • Product variant (class Product)

In this case you cannot use parent because there is a folder in between (which may only apply to some products while others are directl below their brand). Additionally, master product and product variant are on different levels, so parent would end up on different objects.

In this case, you can use closestOfClass#classId. This will walk up the tree until it finds an object of the given class. In this example closestOfClass#Brand would end up at the same object for “Master product” and “Product variant”.

Translations

With the translate helper function you can translate terms in data query selectors. This can be used to translate fields which are not in localized fields. You could for example use this to translate an object’s key with the data query selector key:translate. This will translate the key of an exported element to the requested language. If you want to always translate the item to the same language regardless of the requested language you can provide the target language with #<locale> suffix - in above’s example this would be key:translate#de to export the object key translated to german.

First it is tried to find an admin translation in the requested language. If none exists, shared translations get used. If neither exists the original value gets returned.

Filtering

Arrays can be filtered with the filter helper. It supports 2 or 3 parameters: categories:filter#published,1 will return an array of the assigned Category objects which are published. You can filter for any data object or system field. The third parameter supports comparison operator, e.g. for a field collection prices with a date field validFrom you can use prices:filter#validFrom,now,>= to get all prices which are valid today or in future (now will be evaluated as current date).

After filtering it may be useful to check of the returned array is empty or not. This can be done with PHP’s empty or with exists:

  • categories:filter#published,1:empty will return 1, if there is no published category assigned, otherwise 0.
  • categories:filter#published,1:exists will return 1, if there is at least one published category assigned, otherwise 0.
  • categories:filter#published,1:empty:not is equivalent to categories:filter#published,1:exists

Sorting

You can sort arrays of objects via orderBy helper. For example if you have a product class with a many-to-many object relation to categories and those categories should be ordered by name, you can use data query selector categories:orderBy#name.

Get workflow state

You can get the workflow state of an object by using the expression

  • workflowState#<Workflow-Name>:places:key for workflows where an object can only be at one state at the same time or
  • workflowState#<Workflow-Name>:places:each:key for workflows where an object can be at multiple stages at the same time.

Beware if you use single_state as marking store for your workflow and name the storage field within your class workflowState then you do not need above helper but can access the field content directly via workflowState.

Debugging Data Query Selectors

You can append :debug to your Data Query selectors to see the single steps what happens befind the scenes. For instance for a Pimcore-object-based dataport you want to debug the data query selector brand:brandName, simply change it to brand:brandName:debug (or add a new field) and you will get:

[debug] Using context object /Products/1234
[debug] Found 1 element
[debug] Using function data_director_564b21289729229e07e4cc840970c641 in
 <Pimcore cache directory>/Product_Brand_brandName.php to resolve data query selector
[debug] Result of data query selector part "Brand" is "/Brands/ABC"
[debug] Field or function "brandName" could not be resolved
[debug] Final result:

In this case the field brandName in the class Brand got renamed to name, thus the data query selector did not return any data. After it is changed to brand:name:debug, you will see that it works:

[debug] Using context object /Products/1234
[debug] Found 1 element
[debug] Using function data_director_2ac9c05c54567394d866ea1c99ccbd97 in  
 <Pimcore cache directory>/Product_Brand_brandName.php to resolve data query selector
[debug] Result of data query selector part "Brand" is "/Brands/ABC"
[debug] Result of data query selector part "name" is "ABC"
[debug] Final result: ABC

Finally you can remove the :debug and the data query selector will return the brand name ABC.

Get all Classification store / object brick field contents incl. field labels

With the data query selector classificationStore:labels you can get an associative array of all assigned groups with the translated field names:

{
  "Translated name of Group 1": {
    "Translated field title of field 1": "field 1 value",
    "Translated field title of field 2": 123.45
  },
  "Translated name of Group 2": {
    "Translated field title of field 3": "field 3 value"
  }
}

If you need the labels but also the field names, you can use classificationStore:labels#true. This way you get

{
  "groupName1": {
    "groupTitle": "Translated name of Group 1",
    "values": {
      "fieldName 1": {
        "fieldTitle": "Translated field title of field 1",
        "value": "field value"
      },
      "fieldName 2": {
        "fieldTitle": "Translated field title of field 2",
        "value": 123.45
      }
    }
  },
  "groupName2": {
    "groupTitle": "Translated name of Group 2",
    "values": {
      "fieldName 3": {
        "fieldTitle": "Translated field title of field 3",
        "value": "field 3 value"
      }
    }
  }
}

Field types with variant / placeholder support

There are 3 new field types available in Pimcore’s class definition view:

  • Input with variables
  • Textarea with variables
  • Wysiwyg with variables

Those fields are a combination of normal text fields and calculated value fields. With data query selectors you can access data of other fields and other objects, e.g.

This {{ category:name }} is {{ color }}.

will get resolved to This t-shirt is black if there is a many-to-one relation for the product’s category which has a field name and if the product has an attribute color.

You can also implement any logic operations with Twig syntax with the exception that the variables are defined with data query selectors, e.g.

{% if category:name#en == "T-shirts" %}
  This shirt is made of {{ materials:each(name):implode( and , %s) }}.
  
  Properties:
  <ul>
  {% for attributeName, attributeValue in attributes:tshirts %}
    <li>{{ attributeName}}: {{ attributeValue }}</li>
  {% endfor %}
  </ul>
{% endif %}

If the assigned category’s name is “T-shirts” and there is an object brick container which the object brick tshirts got assigned, the result will be:

This shirt is made of cotton and polyester.

Properties:
<ul>
    <li>Color: black</li>
    <li>Size: XL</li>
</ul>

This allows to define the template for the texts at a high level and then inheriting it down for all descendant products. Thus you have automatic text generation dependent from your template which gets defined directly in Pimcore backend.

This also allows for rapid prototying of applications: For example if you develop a product detail page and know where the product price shall appear, you can create a field price and access it in your template. Later on you can either fill this field and / or also refer to other fields of the data object or other objects. The advantage of this compared to a calculated value field is that you can implement the logic directly in the data objects and also inherit / override the logic for a certain group of data objects without having to create loads of if conditions in a calculated value calculator.

This way you could for example change the calculation of prices without changing the controllers or frontend templates. This makes you independent of the existence or naming of fields which may be used for price calculation, like date validation range, customer-specific pricing etc.

Different behaviour of getter methods for exports

Normally the data query selectors use the getter methods from the Pimcore classes. These return always the same values independent of they are called from the Pimcore backend, frontend or webservice. But in some use-cases you might want to get a different behaviour depending on the access method. This is possible when you override the getter method. You can then check if you are currently in a data director extraction context with:

public function getMyField() {
    $request = \Pimcore::getContainer()->get('request_stack')->getMainRequest();
    $contextResolver = \Pimcore::getContainer()->get(Pimcore\Http\Request\Resolver\PimcoreContextResolver::class);
    
    $value = parent::getMyField();
    if($contextResolver->matchesPimcoreContext($request, 'webservice')) {
        // implement custom logic here
    }
    return $value;
}

That the data director context is named webservice has historical reasons. Until Pimcore 6 the data director extended Pimcore’s now removed webservice API.

Create / edit multiple objects with one raw dataset

There are cases where you want to create multiple Pimcore objects from one dataset in the import source. For example when you have the following XML as import source:

<article>
    <article_number>123</article_number>
    <colors>
        <color>red</color>
        <color>blue</color>
        <color>green</color>
    </colors>
</article>

You have two raw item columns (raw item XPath would be /article):

  • “article number” with XPath article_number
  • “colors” with XPath colors/color with “multiple” checkbox being set

The raw data import generates this raw dataset:

Article number colors
123 ['red', 'blue', 'green']

Now you want to create 3 Pimcore data objects from this raw data item. This can be achieved by returning an array in a key column.

For simplification imagine there is a Product class with two input fields article_number and color. In the import’s attribute mapping you assign both fields as key columns. Article number gets assigned to the article_numberfield and colors to the color field. When an array gets returned for a key column this is always interpreted as you want to create / edit multiple objects from this one raw dataset. Internally all possible combinations of all key column values get created. In this case these are:

  • Article number: 123, colors: red
  • Article number: 123, colors: blue
  • Article number: 123, colors: green

The actual import gets run for each of these combinations individually. This is even possible if more key columns contain an array of values (or their callback function returns an array - e.g. splitting a comma-separated list in a CSV file) - so the number of objects being imported per raw dataset is the product of the number of values which the key columns contain.

Finally you will get 3 data objects, all with the same article number, but each with a different color.

Virtual Fields

You can use {{ VARIABLE NAME }} as a placeholder in the callback functions. This will add a virtual field with the given name to the attribute mapping panel. Virtual fields can be used for several things:

  • fetch elements via data query selectors and use them for additional logic, e.g.

    $object = {{ OTHER OBJECT }};
return $object->getField1() ?: $object->getField2();

    This will create a virtual field OTHER OBJECT where we could have return 'Category:name:'.$params['value']; as callback function. This virtual Field will try to find a Category object with the field name being equal to the assigned raw data field. In the field with above callback function we then return either the content of Field1 or if this is empty then the content of Field2 of the returned Category.

  • reusable templates can contain virtual fields. For example the bundle provides a template for uploading export files via FTP. The template defines the virtual fields FTP Host, FTP Port, FTP username, FTP password, FTP Path. You do not have to edit the code provided by the template but can just change the values for these virtual fields.

Virtual fields support getting values from app/config/parameters.yml and .env files. For example when you create export dataports for a shop API, you could put your API credentials in one of those files, in this example in app/config(/parameters.yml:

parameters:
    API_ENDPOINT: http://example.org/api

When there is any callback function which uses the placeholder {{ API_ENDPOINT }} the value from parameters.yml will automatically get used - but of course you can override it by assigning a raw data field or implementing a callback function.

Reuse value of other field

When the name in the double curly braces refers to a really existing field’s name (e.g. {{ Key }} for the object key) the field does not get treated as a virtual field but it will contain the mapped return value of the referenced field. For example, when you have a callback function for the object key and want to use the same logic also for the name field of Product objects without duplicating the callback function, then you can use return {{ key }} to get the same value which got returned from the callback function for the field Key.

Optimize inheritance

Updating the uppermost object

Enabling the optimize inheritance checkbox in the dataport settings leads the importer to set the imported data not necessarily directly on the object which is identified by the key fields but at one of its parents / ancestors to take advantage of object inheritance. An example: You have the following master-slave product structure:

  • Master
    • Slave 1
    • Slave 2

There are 3 lines in your data source:

ID Name Price Master ID Category ID
1 Test Product     2
2 Test Product 10 1 2
3 Test Product 20 1 2

When you import with “optimize inheritance” the importer would update the master by setting the name and category. Then it imports the slaves also with name and category. When you later want to change the name or category, you have to change that in all 3 objects.

When you use “optimize inheritance” the importer would update the master as above. But when it comes to importing the slaves, the importer fetches the slave’s siblings. If all siblings have the same value in the imported field, the value is only updated at the uppermost object. In this case the name and category fields would stay empty for the slaves because all slaves have the same name and category. The category and name will only be updated at the master. Nevertheless the slaves get the values in these fields via inheritance.

As prices differ between the slaves, they are set on the slave objects.

When you later want to change the name or category, you only have to change it on one single point: the master. This makes editing a lot easier.

This also applies to a lot of other use-cases, e.g. technical data via object bricks: If you structure your object tree hierarchically, it often means that similar objects are under the same branch. For example all products under a certain branch consist of the same material. Then the inheritance optimization would update the object brick field “material” only for the branch root. All objects below would inherit this value.

Creating object hierarchy

To fully take advantage of this feature you should think about grouping objects by certain attributes. For example you could define the desired object tree hierarchy main category > standard > material > article number. When you return this structure as the path in attribute mapping for the field parent element / path(e.g. Filters/DIN 123/Stainless Steel) and the article number in the field object key the path’s parts are created as objects (in contrast to folders when you do not use “optimize inheritance”). This allows for optimal use of inheritance.

Import assets

When you choose Asset as import class you can import assets without touching any data objects. This is especially useful in combination with the import type Filesystem because then you can periodically import assets from a certain directory into Pimcore.

The most important field in field mapping to import assets is the field Stream. You can assign it an absolute path to the file or a URL. By returning a string with a trailing slash / you can create folders.

Import documents

Document imports allow creating documents based on a set master document. All editables of the master document are available in the import’s attribute mapping. You can either build a view with editables or create the document structure dynamically via area blocks.

The set master document will be set as content master document in newly created documents. This allows for prefilling content in the master document which does not need to be imported.

Unpublished versions

On the one hand it is important to not overwrite changes of unpublished versions and on the other hand to not publish the changes of unpublished versions. For this reason for elements whose latest version is unpublished, the import gets executed for currently published object version as well as for latest unpublished version - resulting in 2 new versions: one published based on the last published version but including the changes from the import, and one unpublished version based on the latest unpublished version but also including the changes from the import.

For new objects or objects whose latest version is published, of course only one new version will get created (is anything changed during the import).

Logging and error tracking

During a dataport run a lot of messages get logged to know exactly what happens. Those logs can be accessed in the History & Run panel of the Data Director. Additionally the same messages get logged to Pimcore’s application logger. With the latter you can filter by component Blackbit/DataDirectorBundle to only see the logs of this bundle.

By default only logs which are equal or worse than warning get logged to the application logger. You can adjust parameter blackbit_pim.application_logger_log_level to a supported log level in your config/packages/services.yaml (Pimcore >= 10) or app/config/parameters.yml (Pimcore < 10).

You can change logging behavior completely by overriding the symfony service pim.logger.

For all dataport runs which do not get started manually from Pimcore backend (e.g. cronjob-triggered / automatic dataport runs / REST API calls), the log file gets deleted if no error occurred during the run. This way no disk space gets wasted for completely successful runs.

This bundle also supports elements-at/ProcessManager. With this bundle you can execute import jobs at certain times - like cronjobs but you can config those within the Pimcore GUI. Another advantage of the Process Manager is that you can configure logging channels. You can define which error levels shall be logged to which output channels. For example you can create a separate log file only for your imports (so they do not get mixed with the normal Pimcore log) or you can specify to get notified via email when critical errors happen during an import.

Notifications

Whenever there is an error during an import which did not get executed with the --force flag, all users who are allowed to configure the dataport are notified via email.

For more advanced notifications you can create a custom email in the result callback function and send it with the included result document action for sending emails. This can for example be used to:

  • get notified of import errors (when not using elements-at/ProcessManager)
  • send result documents (e.g. exported CSVs, XMLs etc.)
  • get notified about new data in a Pimcore custom report

Permissions

For each dataport you can set which users are allowed to execute or to configure it. This uses the standard role-based Pimcore permission system.

Additionally there is the special permission Create dataports for those users who are allowed to create dataports. The user who creates a dataport automatically gets configuration and execution permission for it.

Other Pimcore permissions and workspaces are respected, so a user who is not allowed to access objects under /products can neither execute imports which create / edit objects inside this folder, nor execute exports of objects in this folder.

Deployment

For every dataport there is a definition file in <Pimcore-Root>/var/bundles/BlackbitDataDirector/ folder. As soon as dataport settings get changed, the respective file gets updated.

You should add this directory to your version control system (e.g. Git) to be able to set up a dataport on one system (e.g. local development) and then deploy it to other servers (e.g. live server).

With the command bin/console data-director:deployment:dataport-rebuild you can create / update the dataports based on the files in above mentioned <Pimcore-Root>/var/bundles/BlackbitDataDirector/ folder. It is also possible to only restore some dataports by providing a comma-separated list of dataport ids or names, e.g. bin/console data-director:deployment:dataport-rebuild "1,2,Dataport ABC".

Additionally, this command can be used to synchronize dataports between different Pimcore systems, e.g. bin/console data-director:deployment:dataport-rebuild --source=https://other-pimcore.com --api-key=1234 will use the dataport configuration REST API to retrieve available dataports from the remote system https://other-pimcore.com. When working with remote source, only dataports whose latest version has been imported or which do not exist yet, will get imported. For already existing dataports whose latest version has been edited by a real user, you will get asked if you want to overwrite the local dataport configuration (to prevent local changes from being overwritten). As soon as you allow the import, the latest version will be marked as been imported and thus can be synced automatically in future. If you want to always overwrite local changes, use the --force flag.

To only synchronize some dataports from the remote system, you can provide the dataport ids or names as comma-seaprated list: bin/console data-director:deployment:dataport-rebuild --source=https://other-pimcore.com --api-key=1234 "1,2,Dataport ABC".

Automatic tests

With the command bin/console data-director:create-test <Dataport ID> you can create a PHPUnit test for the given dataport. Based on the currently existing raw items and mapping a test class will get created. You can run those automatic tests via PHPUnit.

Main purposes for automatic tests:

  1. Assure that after changes in dataport settings or attribute mapping everything works the same as before.
  2. Assure that before and after updates / modifications of underlying libraries like Pimcore or Data Director itself, everything still works as before.

The test classes / files will get created in <Pimcore root>/tests/BlackbitDataDirectorBundle.

To run the tests a phpunit.xml is needed. If this does not already exist in the Pimcore root folder, it will also get automatically created. If it already exists, a new test suite will get added to it.

The command will display instructions how to run to tests.

Revert imports

Sometimes humans make mistakes but in attribute mapping and especially its callback functions small logic mistakes can have serious consequences: e.g. when the import changed thousands of Pimcore objects the wrong way. In this moment you could restore a complete backup of the Pimcore system. But this has the main disadvantage that in the meantime other objects than the corrupted ones could have been changed - restoring a backup would revert all those changes.

A better way is to use the command bin/console data-director:revert <Dataport-ID> <Date to revert back to>. With this command objects of the target class of the dataport get reverted to a version equal or before the specified date. Furthermore only those fields are reverted which are mapped in the dataport’s attribute mapping.

Example: You have a Product class with fields Article description (Text) and Images (Many to many relation). Until now images got assigned to the field images manually. Now you want to add images from an external source via import. The Article description is filled manually. You create the dataport for the image import and accidentally activate the checkbox “Clear before import” for the field images. You run the import and do not recognize the mistake at first sight. In the meantime the objects’ field Article description got changed for a lot of products manually.

Eventually you recognize the error with the missing images. It is easy to repair the import but all the previously assigned images are lost. Now you cannot restore a backup as all the manually edited Article description data got lost - as well as any data of other objects which got altered meanwhile.

Now the command bin/console data-director:revert <Dataport-ID> <Date to revert back to> comes into play: It reverts the product objects to a given point in time (as long as there is a version of the given time or older) and it even reverts only those fields which are mapped in the given import. This means the images relation gets reverted while the field Article description keeps its current contents.

If you only want to revert objects which are found in the current raw data of the dataport you can use the option --only-current-rawdata.

Alternatively you can filter the elements to be reverted with the --objects option which you can provide a comma-separated list of object ids.

It is also possible to first do a dry run with the option --dry-run. With this you see the same output which objects would get reverted but no data gets actually changed.

Initialization function

In attribute mapping there is a special field named Initialization function. This allows you set define a callback function which gets executed once before raw data processing starts. Here you can for example check if another process of this dataport is currently running or reset a certain field of a data object which afterwards gets set again by the imported raw data items.

Import result callback / Result document generation

In attribute mapping there is a special field named Result callback function. This function gets executed after a raw data item has been processed. For example you can use this field to:

  • generate a response document including all successfully imported items and send that to the source system
  • track import errors
  • call another import which depends on the current one
  • create / export documents (JSON / CSV / XML etc.) which other systems can use as import source
  • create response documents for single-page application / PWA frontend requests

This result callback function receives the following parameters in variable $params:

  • response - current response document generated by this callback function from previously processed raw items, initially empty Symfony\Component\HttpFoundation\Response object - you can either populate the response via $params['response']->setContent('my response content') or return "my response content";.
    • response headers can be set via $params['response']->headers->set('Header Name', 'Header value');
  • rawItemData - array with imported raw data
  • request - object of class Symfony\Component\HttpFoundation\Request (even when called via CLI)
    • access current language via $params['request']->getLocale()
    • request parameters can be accessed via $params['request']->get('parameterName') but are also available via `{{ parameterName }} placeholders
  • transfer - object for data holding between different raw data items, initially empty
  • lastCall - boolean value to determine if current call is the last one before import / export is finished
  • context - array with following keys:
    • dataport
      • id: dataport ID
      • dataportName: dataport name
    • resource: array with set import source, may contain the following keys:
      • file: import resource (e.g. file name for file-based imports or SQL condition for Pimcore-based dataports)
      • locale: language to be used - especially for Pimcore-based exports
    • user
      • id: id of the user who started the dataport run
      • username: name of the user who started the dataport run
    • logs: Logs for whole import in temporal order

  • logs - array with noteworthy things which happened during import of current raw item, indexed by level of severity, e.g.

    [
  'warning' => ['First warning', 'Second warning'],
  'error' => ['First error', 'Second error'],
  'alert' => ['Really severe error'],
  'info' => ['First info', 'Second Info', 'Third info'],
  'notice' => ['First notice', 'Second notice']
]
  • objectIDs - array with object IDs which got changed by the rawdata item
  • logger - Logger object which implements \Psr\Log\Logger
    • Usage example: $params['logger']->error('Something unexpected happened');

Result document actions

Beside Result callback function there is another special field in attribute mapping: Result document action. With this field you can define what to do with the generated result document. This could be normal output (so the generated document gets sent to the browser), FTP upload, send the document via email etc. For the listed use cases the required functions get already shipped as templates so you do not have to program anything but only select one of the templates (but of course you can adjust the implementation logic if you want to).

This Result callback function gets the following parameters in variable $params:

  • response - generated response document, instance of Symfony\Component\HttpFoundation\Response
  • transfer - object which can be populated in result callback function
  • context - array with following keys:
    • dataportId: dataport ID
    • resources: array with set import sources, each dataport resource is an array item with the following keys:
      • file: import resource (e.g. file name for file-based imports or SQL condition for Pimcore-based imports)
      • locale: language to be used - especially for Pimcore-based exports
  • logger - Logger object which implements \Psr\Log\Logger
    • Usage example: $params['logger']->error('Something 'unexpected happened');

REST interface

This bundle brings its own REST API interface with the following endpoints:

  • POST http(s)://<YOUR-DOMAIN>/api/rest/import/<Dataport-Name or id>?apikey=<API-KEY>
[&async=1][&parameter1=123][&anotherParam=1]
    • In the request body you provide the document to be imported (e.g. CSV / XML / JSON content)
    • <API-KEY> is the API key of the Pimcore user. You can only import data to objects which this user has access to.

    • Example for a CSV import:

      POST http(s)://[YOUR-DOMAIN]/api/rest/import?apikey=[API-KEY]
&dataportId=[Dataport-ID]
    
Article number,name
123,My cool product
    • via import callback function you can generate a response document
    • requests with async parameter are run in the background. As response you get a URL from which you can fetch the current status and result of the started import (see next bullet point)
    • provided parameters can be accessed in raw data fields and attribute mapping via {{ parameter1 }} or {{ anotherParam }} - the values come from the URL parameters and can be named however you want, see parametrized dataports
  • GET http(s)://<YOUR-DOMAIN>/api/rest/export/<Dataport-Name or id>?apikey=<API-KEY>  
[&locale=<language / locale code>][&parameter1=123][&anotherParam=1]
    • returns the result of the result callback function
    • when exporting data from localized fields you can either access the field name itself and handle the target language via the locale parameter (e.g. locale=de) or you can specify the language to be exported in the raw item fields - e.g. name#en will always export english name regardless of locale parameter whereas name will export the name in the given language. If no valid locale is given, the configured language of the Pimcore user, whose API key is being used in the request, will be used.
    • provided parameters can be accessed in raw data fields, SQL condition and attribute mapping via {{ parameter1 }} or {{ anotherParam }} - the values come from the URL parameters and can be named however you want, see parametrized dataports
  • GET http(s)://<YOUR-DOMAIN>/api/rest/status?apikey=<API-KEY>&statusKey=<Status key>
    • get result of asynchronous import
    • <Status key> is the key which gets returned by /api/rest/import with async=1
    • as long as HTTP 102 Processing gets returned the import has not finished yet (so the response body contains only data up to the last imported item)

In all above requests the API key can also be transferred via X-API-Key header instead of the apikey parameter.

API Keys

Below the dataport tree there is a button Permissions & API Keys. There you see all users which the current account is allowed to administer (admin accounts will see all users, other users will only see the dataports which they are allowed to configure). Initially the uses do not have API keys but as soon as you double-click in the column API Key for a certain user an API key will get created. If you want to change it, you can double-click again and edit it. Optionally you can set an expiration date if you want to limit the lifetime of this API key.

The user needs execution permission for the dataport to be able to execute it via REST API.

Dataport configuration REST API

List dataports

All existing dataports can be retrieved via

GET http(s)://<YOUR-DOMAIN>/api/dataports?apikey=<API-KEY>

In the response you will get a list of dataports as JSON:

{
  "3": {
    "name": "Import Products",
    "lastModified": 1687849015
  },
  "19": {
    "name": "Export Products",
    "lastModified": 1679390079
  },
  ...
}

Only dataports get listed which the user, whose API key gets used, has configuration permission.

Dataport configuration

The endpoint to retrieve the full configuration of a dataport is

GET http(s)://<YOUR-DOMAIN>/api/dataports/<dataport id or name>?apikey=<API-KEY>

This will return the full configuration as JSON, similar to the definition files in <Pimcore-Root>/var/bundles/BlackbitDataDirector/ folder, see deployment.

Object Preview

Pimcore supports a preview tab in the object editing panel. This feature can be combined with a Data Director export. The export can create an HTML document which can be shown in the preview panel of the currently opened object. To achieve this, you have to enter @DataDirectorPreview in the field General Settings > Preview Generator Class or Service Name of your data object class. Afterwards when you open a data object of this class you will immediately see the out-of-the-box Default Preview available which shows the current object’s data as known from the versions panel. But the real cool feature is when you create a Pimcore-based dataport which generates an HTML document. You can use o_id='{{ id }}' as SQL condition and then add the desired raw data fields and generate the desired output in the result callback function.

One use-case which the Data Director already supports out-of-the-box is a dependency graph: Object dependency graph

Translatable element names in element tree

By default Pimcore does not support translating element names. This leads to the problem that you have to decide which language to use for folder names and object keys. This complicates understanding the system for foreign language data maintainers.

Data Director includes this feature (for Pimcore >= 11). If the user’s language differs from the configured system default language, the element keys are tried to be translated via admin translations. Initially they will not exist, so the displayed element names will stay the same. But this way you can translate folder names or object keys if you want to. The element name gets used as translation key, so you can go to admin translations, search for the element key and then translate it in the desired language.

Mouseover information in element tree (tooltip)

By default Pimcore only shows the element ID and type when hovering an element in the tree. Data Director extends this functionality by additionally showing all fields which are configured to be Visible in Grid View.

Pimcore’s default quick search searches across all element types (objects, assets, documents) and across all classes. Moreover, the order of the found items is more or less random. And because it uses MySQL’s full text search, performance suffers when there are lots of elements and fields.

In the real world this makes the quick search difficult to use as the user has to check which of the found items is the desired one.

Data Director changes the Quick Search behaviour: All search terms have to be present in search results and files in import archive folders of any dataport are excluded from search results (otherwise searching for a product name or SKU will often find the archived import files). This way the search results are more exact and it distinguishes the quick search from the element-specific full-text search features (e.g. data object search).

Path formatter

Relational fields support path formatters to define the display of the related item. Data Director comes with a universal implementation which uses the fields of the related item, which are configured to be Visible in search Result, also in the relation field view. To use this, set Formatter service / class of the relational field to @DataDirectorSearchViewPathFormatter.

This gets done automatically for all relational fields which do not have a Formatter service / class. If you prefer to show the full path of the related item (= Pimcore’s default display of relations), please use @DefaultPathFormatter.

If you prefer to show the fields which are configured to be Visible in Grid view, you can use @DataDirectorGridViewPathFormatter.

Compared to Pimcore’s default view of related items the Data Director path formatters offer the following features:

Criteria Pimcore’s default path formatter Data Director path formatters
Data object reference Full path Fields which are configured to be Visible in search Result
Images in (advanced) many-to-many relations Display full path of assigned asset images Display thumbnail images and full path
Asset metadata Not accessible from relation view Asset metadata gets shown as tooltip
Direct edit of assigned assets Not available Supported files are displayed with a link to open the assigned asset file directly in the application (e.g. open xlsx directly in Excel)

Element Tree Pagination

By default Data Director sets the number of items per page in the tree to 1000 - in practice this means it disables paging. This way all the elements are visible by scrolling - so it is the same behaviour as for example in tools like Windows Explorer or MacOS Finder.

Data option providers

Data query selector option provider

For select and multiselect fields Pimcore supports data option providers to create a dynamic list of available options. When you set Options provider class or Service Name in your select field configuration to @DataDirectorOptionProvider, you can use a data query selector to define the allowed options.

For example you can use Brand:published:true:each(name) to define the names of all published Brand objects as options. The problem with this approach is that as soon as you change the name of an assigned brand, the assignment in the select could not be found anymore. For this reason you can use a more sustainable solution: Brand:published:true:each(name as label;id as value). This way the name will be used for displaying purpose but the actual value on the database will get defined by the Brand object’s ID.

Additionally, you can access values of the currently opened object to implement dependent select fields: Imagine you have a multiselect field in the Category class and use Product:category:{{ id }}:each(brand:id as value;brand:name as label). {{ id }} will get resolved to ID of the currently opened object (but also all other fields or chained data query selectors are possible here). In this case you will get a list of all brands of the products which have the currently opened category object assigned in their category field.

Object brick option provider

Configure the Options provider class or Service Name in your select field configuration to @DataDirectorObjectBricksOptionProvider. In Options Provider Data enter <className>:<object brick container field> to provide the allowed object bricks of the referenced field as select field options. You can use this to assign on category level which object brick(s) apply to all products of this category. To automatically assign the set object brick(s) to all products of this category, you have to set up a dataport which retrieves the brick name of the category and “imports” the corresponding brick to the object brick container field. Enable Run automatically on new data to automatically run this dataport when a product object gets saved.

Perspectives / Dashboards

Pimcore supports perspectives to customize the backend layout. Data Director provides a basic PIM perspective based on the data object classes’ groups. All data object classes which belong to the same group get shown in the same perspective panel. The root folder gets automatically optimized to the lowest parent object which contains all objects of the corresponding class.

The PIM perspective also comes with a dashboard which includes a queue monitor, tagged elements and dataport errors portlets. Of course those portlets are also usable in other dashboards.

Integration with other Pimcore plugins

Pimcore Process Manager Bundle

To run dataports periodically you can either use a real cronjob or you can use elements/process-manager-bundle. With the latter you are able to configure dataport start times directly from Pimcore backend. Data Director is highly integrated with the Process Manager bundle so that configured log levels, error notification and progress bars get used.

Pimcore FormBuilder

With dachcom-digital/formbuilder you can create forms for your Pimcore documents. Data Director ships with an API channel for the form builder so that the entered form data can be processed in a Data Director dataport. This way form creation and processing can completely be implemented in Pimcore backend without touching any HTML, CSS or PHP files.

Create a dataport of type Object wizard with the desired raw data fields as input fields. Then go to the FormBuilder form, add an output workflow of type API. Under configuration select type API channel. As data provider select Data Director and select the just created dataport. You can either map the raw data fields in form builder, alternatively this step can be skipped if your dataport’s raw data field names are the same as the display names of your form fields.

Alternatively, if you want to use the same dataport for multiple FormBuilder forms, create a raw data field __request. This will contain all form field data as JSON object.

Pimcore Toolbox

dachcom-digital/pimcore-toolbox enables you to use area bricks to create Pimcore documents. In combination with Data Director this can be used for prototyping of product datasheets, workflow email notification documents etc. You can create a document which supports the toolbox area bricks by creting a new Page document:

Pimcore Toolbox integration

SEO bundle

dachcom-digital/seo offers an additional tab where you can enter SEO-relevant data. Data Director supports importing data into those SEO fields.

Troubleshooting

Import does not update my data objects

If you once imported data to a data object, a property importhash_<dataport id> with the hash of all belonging raw data gets added to the object. On the next import it will be checked if the object which got found via the set key attributes already has this hash attribute and it is equal to the hash of the current raw data to be imported. If it is, no data gets updated to not accidentally overwrite manual changes to fields which got initially filled by the import.

To bypass this behaviour you can use the option --force (or shortcut -f) when doing the “data-director:process” step, e.g. bin/console data-director:process --force <Dataport-ID> or bin/console data-director:complete --force <Dataport-ID>

Import became slow (was faster before) / Database table application_logs is very big

For convenience the importer logs messages to Pimcore’s application logger. By default only warnings and worse errors are logged. You can configure this behaviour by setting

parameters:
    blackbit_pim.application_logger_log_level: 'debug'

in your config/packages/services.yaml (Pimcore >= 10) or app/config/parameters.yml (Pimcore < 10).

But beware there can be loads of messages, so set the log level only to debug when you find data errors or when you configure new imports.

If you want to disable logging to the application logger completely, you can add the following to your app/config/services.yml:

pim.logger:
    class: Blackbit\DataDirectorBundle\lib\Pim\Logger\Logger
    arguments: ['@monolog.logger.pimcore']
    public: true

Import / Export gets aborted automatically

When an import needs a long time to import a single raw data item, the import is assumed to have been terminated via CLI and gets marked as Aborted. This for example happens when a single raw item shall update a big number of objects. The time after which an import gets aborted is dependent on the following rules:

  • if less than 100 items have been imported: import process gets aborted after 3600 seconds (= 1 hour)
  • if more than 100 items have been imported: import gets aborted if current batch of 100 raw items needs more than the average of 100 previously imported raw data items multiplied by factor 6. Or in other words: If importing current batch of 100 raw items needs more time than 6 average batches of previously imported items, the import gets aborted.

You can increase the parameter blackbit_pim.importstatus.abortion_threshold in config/packages/services.yaml (Pimcore >= 10) or app/config/parameters.yml (Pimcore < 10) to enlarge the abortion threshold:

parameters:
    # in seconds, default is 3600 (1 hour)
    blackbit_pim.importstatus.abortion_threshold: 72600

If that does not solve your problem, maybe the process gets killed by the webserver or operating system. For example watch for Apache’s Timeout or FcgidIOTimeout settings.

Also some webhosters kill CLI processes which run longer than a certain threshold.

Queued items shall get processed in a night run

By default queued jobs (e.g. via automatic dataports) get processed immediately. If you prefer to control queue processing on your own (for example with a watchdog process), you can configure

blackbit_data_director:
    queue_processing:
        automatic_start: false

The queue processing then has to be started via bin/console data-director:process-queue. Due to a locking mechanism there can never be more than one queue processor process running.

What if key field and set value shall differ?

In some cases you want to filter the objects to be updated with a different value than you want to set the field content to. For example this is necessary when you have a many-to-many relation and want to additionally refer object A to all objects which currently have object B assigned in the same relation field. In this case you have to apply a little trick: When fetching the key field values currentValue of course is not set (as this contains the value of the current processed object but at this stage we want to determine the objects to be fetched).

So we can differentiate between filtering and writing by providing a callback function like:

if(!isset($params['currentValue'])) {
  // find objects which have the following object assigned
  return 'Reference_Class:field:'.$params['rawItemData']['field_value']['value'];
}

// value setting
return [
  'Reference_Class:field:'.$params['rawItemData']['field_value']['value'], 
  'Reference_Class:field:'.$params['rawItemData']['other_field_value']['value']
];

Demo / Tutorials in Docker

Please install Docker

To run the demo you need to run ./docker-setup.sh to create a docker container with the installed data director bundle. You can then access the Pimcore demo admin under http://localhost:2000/admin.

  • User: admin
  • Password: admin

The installation will need some time. When you installed it once and only want to restart the container, just call docker-compose up -d. You can stop it via docker-composer stop.

Tutorials

When you want to try yourself what is shown in the tutorial videos you will find all the examples including instructions, import data and finished solutions (dataport JSON files you can simply import if you do not know how to proceed) in the examples/ folder.