Data collector Base

This module provides the base for creating an industrial data collection and synchronization application for iTop. Developers can rely on this module to perform all the heavy lifting related to the iTop data import and synchronization process in order to focus on the data collection.

• Simple API to make the creation of a new collector fast and easy.
• Automatic creation and update of Synchronization Data Sources in iTop, based on JSON definitions.
• Support small variations to the target Data Model via explicitely “Optional” attributes
• Basic validation of the CSV format compared to the expected fields in the synchro data source.
• Data upload and synchronization by chunks of max_chunk_size (configurable).
• Extensible mechanism for handling configuration parameters.
• Extensible list of placeholders in the JSON definition.
• Automatic management (via placeholders) of the contact to notify and the user to use for running the synchro.
• Validation of the minimum version of PHP and the needed extensions (as specified by the collector).
• Command line tool to produce the JSON definition from an existing Synchro Data Source in iTop.
• Capability to run independently: the configuration of the data sources, the data collection and the data synchronization.
• Configurable log level (for console output or syslog logging)
• Simple framework for quickly creating SQL based collectors

• Data upload to iTop is done only via the syncho_import web service (could use the command line version or direct SQLcommands. TBD later, maybe)
• Prior to the revision 3805 of iTop from SVN (from 2015-10-12!) the collector will NOT work properly if the account used to connect to iTop is not configured to use English as the language !!
• Collector can only use the login_mode = “form”, it is not possible to use SSO authentication
• Collectors are using synchro_import, but does not propagate yet date_format

• An access to the iTop web services (REST + synchro_import.php and synchro_exec.php)
• We recommand to install php_curl to use the collector base parameter itop_synchro_timeout otherwise the timeout is hardcoded to 200 secondes and can't be overwritten by the collector.

Collectors are small ETL. Some parts can be customized by configuration, others by code.

• ⬇ Extract:
  • if using one of the bundled implementations (REST/JSON, SQL, CSV, …), you might only need to write some configuration
  • if you need a custom source, or to add some specific behavior, it can be coded in the collector class
• 🔁 Transform: mapping can either be done by configuration (if using an existing implementation) or by code (data mapping and lookup functionnalities)
• ⬆ Load: as the collector uses the iTop DataSynchro, this phase will be customized only by configuration

Directories of a collector:

• core: SDK files
• toolkit: some tools for collector developers
• conf: configuration
• data: csv files containing collected data
• collectors: collectors files

The specifics about a collector resides inside the “collectors” folder. This directory should contain at least :

• collectors classes: for example MyOwnCollector.class.inc.php
• datasynchro configuration: for example MyOwnCollector.json
• main.php: file purpose is to register all the Collector classes for your module and load the corresponding classes (either via require_once(…) or by registering an auto-loader)

Should your collector rely on multiple collector classes, you may group the php file under ​collectors/​src/​ ​and the json files under ​collectors/​json/​​. Furthermore,​ files that extends an existing collector provided by Combodo may be stored under ​collectors/​extensions/​​. See the writing an extensible collector chapter.

Combodo provides some already packaged collectors :

• Data collector for Azure
• Data collector for LDAP
• Data collector for OCS Inventory
• Data collector for vSphere

If no packaged collector is available for your data source, check those examples to implement quickly a base collector :

• Tutorial : create a custom SQL collector
• Tutorial : create a custom REST/JSON collector
• Tutorial : create a custom CSV collector

First question: How to get data from your source application. How can it be queried?

1. Directly querying the database in SQL?
2. Using REST and webservice, with returned data in JSON format?
3. Directly using CSV files produced by the application?
4. Using other sorts of webservices, such as for example ripcord?

For each of the first 3 methods, a predefined collector exist already within the SDK, so writing your own based on those is mainly a matter of writing configuration files. For the last case, there is more work to do in PHP.

Second question: On which server will you install your collector?

• The collector is a standalone application, which need to be started by a cron on a regular basis.
• The collector must have access to the data of your source application
• The collector must have web access to your iTop webservices, but does not need to be on the same machine.
• The collector must be host on a machine compatible with collector base requirements

For each Source of Data which is collected and mapped before synchronization with iTop, the configuration can specify:

• How to collect: where to get/find the source data
• How to map:
  • Which source field should be copied in which iTop field
  • How a field value should be transformed into an iTop understood value
  • What default data to put in a given iTop field if not present in the source
  • The mapped data are put in an intermediate CSV file.
• How to synchronize
  • Should the iTop field value be reset or left untouched, if later found empty in the source?

The exact configuration possibilities of How to collect and How to map depends on the collectors (JSON, CSV, OQL, custom).

1. download collector base archive
2. extract the archive files to the chosen collector machine
3. copy conf/params.distrib.xml to conf/params.local.xml
4. create your collector class in the /collectors directory
5. create collectors/main.php
6. create the corresponding datasynchro on the targeted iTop instance
7. extract the datasynchro config
8. update configuration (conf/params.local.xml)
9. launch the collector

params.local.xml is at minimum, the file to edit to configure a collector.

And the following parameters must be set in this file:

params.local.xml

  <itop_url>https://localhost/</itop_url>
  <itop_login>admin</itop_login>
  <itop_password>admin</itop_password>

The following parameters can be redefined to alter the default behavior of the collector:

  <curl_options>
    <CURLOPT_SSL_VERIFYHOST>0</CURLOPT_SSL_VERIFYHOST>
    <CURLOPT_SSL_VERIFYPEER>1</CURLOPT_SSL_VERIFYPEER>
  </curl_options>

The configuration parameters for the SQL Collectors are:

Inside your collector configuration section you can set below parameters:

A collector is a PHP class that provides the data for a given iTop Synchronization Data Source. Collector classes are derived from the abstract Collector class. Each collector is associated with a Synchronization Data Source, defined in JSON format. The default implementation simply looks for a JSON file with the same name as the collector class and the extension “.json”, in the collectors folder.

The simpler way to create this definition file for a Synchro Data Source, is to export the definition of an existing data source.

• Create the synchronisation data source in iTop, adjust its parameters (attributes mapping, deletion policy, full load interval, etc.) to suit your needs
• Use the command line tool dump_tasks.php (available in the toolkit folder) to produce the JSON file:

php toolkit/dump_tasks.php --task_name="name of the task to export" > collectors/myCollector.json

This definition file is in JSON format.

Inside your Synchro Data Source definition file you can use special placeholders to make the Data Source configurable by the user of the application, or to adjust its behavior via some special settings:

All of the collector classes must be registered. To register your collector, add in collectors/main.php a call to the static method Orchestrator::AddCollector(). The two parameters are:

• The order in which the collector should be run (when you need to run several collectors one after the other)
• The name of the class (derived from Collector) in which the collector is implemented.

When writing an extensible collector (see below), this step can be skipped as the registration is handled by the collection plan itself

You may encounter network/authentication issues to reach the iTop server you need to synchronize. To test that connection please use below command:

* connection OK: JSON answer is displayed in the ouput

php toolkit/testconnection.php 
    curl_init exists: 1
/home/combodo/workspace/collector/itop-data-collector-base/toolkit/testconnection.php:12:
array(4) {
  'version' =>
  string(3) "1.0"
  'operations' =>
  array(7) {
    [0] =>
    array(3) {
      'verb' =>
      string(11) "core/create"
      'description' =>
      string(16) "Create an object"
      'extension' =>
      string(12) "CoreServices"
    }
...

* network issue: HTTP error code displayed (among other logs)

php toolkit/testconnection.php 
UNIX system
    curl_init exists: 1
Problem opening URL: https://localhost/iTop/webservices/rest.php?version=1.0
    error msg: Failed to connect to localhost port 443: Connection refused
    curl_init error code: 7 (cf https://www.php.net/manual/en/function.curl-errno.php)

* credential issue

php toolkit/testconnection.php 
UNIX system
 /home/combodo/workspace/collector/itop-data-collector-base/toolkit/testconnection.php:12:
array(2) {
  'code' =>
  int(1)
  'message' =>
  string(20) "Error: Invalid login"
}
Calling iTop Rest API worked!

* permission issue

php toolkit/testconnection.php 
UNIX system
    curl_init exists: 1
/home/combodo/workspace/collector/itop-data-collector-base/toolkit/testconnection.php:12:
array(2) {
  'code' =>
  int(1)
  'message' =>
  string(136) "Error: This user is not authorized to use the web services. (The profile REST Services User is required to access the REST web services)"
}
Calling iTop Rest API worked!

Fix : Add the “REST Services User” profile to the user set in the conf/params.distrib.xml file.

Set the console_log_level to 7 or more to get more information about the synchro_import. From level 7 and above, the collector executes synchro_import with output=details instead of output=redcode, and so will get more information about the synchronization errors and warnings on iTop side. Those details are then log on the collector side. With a console_log_level below 7, some warnings may never be recorded at all, even on iTop side, and your iTop objects can remain not-synchronized without explanation.

To learn more, see this dedicated page : Troubleshooting a Data Synchro

To launch the data collection and synchronization with iTop, run the following command (from the root directory where the application is installed):

php exec.php

The following (optional) command line options are available:

For How to synchronize, this is how you can specify fields that you don't want the DataSynchro to modify if they are empty on the source

• In general, when the pushed value is an empty string, the iTop field will be reset to the iTop field default value. Check your iTop version wiki for exceptions.
• Starting from iTop 3.0, when the pushed value in the csv, is exactly the <NULL> string, then synchro_import.php will put NULL in the replica table and the iTop field will remained untouched.

Since version 1.3.0 of collector base, it is possible to specify fields which if empty in the source, should be mapped to <NULL> so that the iTop field value is left unchanged.

param.config.xml

<iTopPersonCollector>
        <nullified_attributes type="array">
                <attribute>phone</attribute>
                <attribute>manager_id</attribute>
        </nullified_attributes>
</iTopPersonCollector>

Sample configuration file:

conf/params.local.xml

<?xml version="1.0" encoding="UTF-8"?>
<!--  Local values for parameters. -->
<!--  The values defined in this file have precedence over the ones defined in params.distrib.xml -->
<parameters>
  <itop_url>https://localhost/trunk</itop_url>
  <itop_login>admin</itop_login>
  <itop_password>admin</itop_password>
  <console_log_level>9</console_log_level>
  <eventissue_log_level>3</eventissue_log_level>
  <contact_to_notify>test@test.com</contact_to_notify>
  <synchro_user>admin</synchro_user>
  <json_placeholders type="hash">
    <test>Test 1</test>
  </json_placeholders>
</parameters>

Sample Synchro Data Source definition file, notice the use of the $version$, $synchro_user$, $contact_to_notify$ and $test$ placeholders

eventissue_log_level has been added to be able to track collector issues from iTop console directly. this could be useful to administrate iTop application or even monitor it.

Let's say parameter 'eventissue_log_level' is set to ERROR (3) level. collector execution will trigger an event issue object creation for each ERROR log printed with some detailed information. this is a way to centralize any collector feedback during any step of collector execution (configuration, collect, data synchronization).

Example of EventIssue created

In many circumstances it may be useful to run several times the collector with a different set of parameters. For example to collect persons information from several LDAP servers (iTop Data Collector for LDAP) or Virtual Machines information from several vSphere servers (iTop Data Collector for vSphere).

Prior to version 1.1.4 of the framework, you had to completely duplicate the collector application and adjust the file conf/params.local.xml on each copy.

Since version 1.1.4 you can have just one single copy the of the collector application and specify a different configuration file (with the command line option --config_file) for each collection to run (i.e. one configuration file per LDAP or vSphere server).

However, to avoid any troubles during the collection of the data and the synchronization with iTop, the following parameters must be properly configured inside the configuration file:

• Use a different <prefix> inside each different configuration file. This ensures that a specific set of Synchronization Data Sources will be created for each configuration file.
• Use a different <data_path> variable for each configuration file. This will cause the collector to store all its collected data (including some temporary files) in a dedicated directory. This prevents one instance of the collector to overwrite the data of another one. You can use the syntax <data_path>%APPROOT%/data/collector1</data> to have a subfolder collector1 created inside the data folder.

Your collector must be a class derived from Collector. It must implement (at least) the Fetch() method. Fetch must return either, for each object to load, an array using the format attribute_code => value or false when the end of the set of objects has been reached.

The array returned by Fetch() must contain:

• an entry primary_key that uniquely identifies the object being synchronized with iTop. The entry can contain whatever unique ID you can obtain from the inventory collection, or a unique identifier generated as a combination of the various fields of the object. It's up to the collector application to guarantee the unicity of this identifier (and its stability in time)
• an entry for each attribute of the object to be loaded in iTop.

The sample code below generates a set of 10 servers, named 'Server 1', 'Server 2' … 'Server 10', and initialized 3 fields of the servers: their name, their organization (always 'Demo') and their description.

main.php

class MyCollector extends Collector
{
  protected $idx;
 
  public function Prepare()
  {
        $bResult = parent::Prepare();
        $this->idx = 0;
        return $bResult;
  }
 
  public function Fetch()
  {
     if ($this->idx < 10)
     {
        $this->idx++;
        return array(
           'primary_key' => $this->idx, 
           'name' => 'Server '.$this->idx, 
           'org_id' => 'Demo', 
           'description' => 'Test Collector'
        );
     }
     return false;
  }
}
 
// Register the collector, as the 1st to run
Orchestrator::AddCollector(1, 'MyCollector');

Raw data collected by inventory scripts sometimes require a normalization before being imported into iTop, in order to obtain homogenous data. The framework provides the helper class MappingTable for performing simple normalizations tasks.

A mapping table is configured (in the params.xxx.xml configuration file) as an ordered list of patterns, with a value associated to each pattern. The “clean” value returned by the mapping table is the value associated with the first pattern that matches the input value. Patterns are expressed as regular expressions. Values can use the placeholders to refer to some part of the matched pattern (%1$s is the whole pattern, %2$s the first group inside the regular expression, etc.).

So, the format of each entry in the mapping table is:

<pattern>/regular_expression/replacement_string</pattern>

Where:

• regular_expression is a PCRE regular expression to be matched against the input value,
• replacement_string is the resulting value if the input value matches the regular expression.
• Note: any character can be use as a delimiter around the regular expression, not only /. But the delimiter character can be present neither in the expression itself nor in the replacement string or the result will be undetermined.

Example of configuration (Brand normalization):

params.xxx.xml

  <brand_mapping type="array">
    <!-- Syntax /pattern/replacement where:
      any delimiter can be used (not only /) 
      but the delimiter cannot be present in the "replacement" string
      pattern is a RegExpr pattern
      replacement is a sprintf string in which:
          %1$s will be replaced by the whole matched text,
          %2$s will be replaced by the first matched group, if any group is defined in the RegExpr
          %3$s will be replaced by the second matched group, etc...
    -->
    <pattern>/IBM/IBM</pattern>
    <pattern>/Hewlett.Packard/Hewlett-Packard</pattern>
    <pattern>/Dell/Dell</pattern>
    <pattern>/.*/%1$s</pattern>
  </brand_mapping>

This example file performs the following normalization:

1. Every string containing “IBM” is transformed into “IBM”,
2. Every string containing “Hewlett”, followed by any character, followed by “Packard” is transformed into “Hewlett-Packard”,
3. Every string containing “Dell” is transformed into “Dell”,
4. All other strings are kept as is.

1. Create an instance of the MappingTable class, passing it the name of the XML tag in which to look for its configuration (inside the XML param file)
2. Use the MapValue method to process each value as needed (the second parameter is the default value, when no match is found in the mapping table).

Usage example:

testcollector.class.inc.php

// Turns the raw brand string ('brand_id') into a normalized brand
// Use 'Other' for brands not found in the normalization table
 
class TestCollector extends SQLCollector
{
    protected $oBrandMapping;
 
    public function Prepare()
    {
        $bRet = parent::Prepare();
        // Create the MappingTable once at the initialization of your collector
        $this->oBrandMapping = new MappingTable('brand_mapping');
        return $bRet;
    }
 
    public function Fetch()
    {
        $aData = parent::Fetch();
        if ($aData !== false)
        {
            // Then process each collected brand
            $aData['brand_id'] = $this->oBrandMapping->MapValue($aData['brand_id'], 'Other');
        }
        return $aData;
    }
}

The data synchronization mechanism embedded in iTop is not capable of performing reconciliations based on multiple fields (like searching for a Model based on both the Brand name and the Model name). The LookupTable class provides this reconciliation capability for any number of fields.

The class LookupTable builds a lookup table by retrieving the specified fields of a set of iTop objects, and storing the resulting identifier of the objects in iTop.

An instance of LookupTable is created by specifying an OQL query (the set of iTop objects to retrieve) and the fields of the objects that will be used for the mapping.

Once the LookupTable has been initialized, a call to the Lookup($aData, array(Field1, Field2, …), destField) method will replace in $aData the value of the column destField by identifier of the iTop object whose specified fields match the values passed in $aData as the columns Field1, Field2….

In iTop, the operating system version is represented as a version depending on an OS family object. We can have the following objects in iTop:

• Windows, versions 7.0 and 8.1,
• Debian version 12.0.0.

This will be stored in iTop as shown below:

Now let's imagine that our collector script gives us the two informations: 'Windows' and '8.1.0'. We can store the 'Windows' text string in the 'osfamily_id' field of the data synchro table and configure the synchro data source to perform the reconciliation based on the 'name' (this will properly replace 'Windows' by 1).

But to retrieve the identifier of the version 8.1.0 of Windows (which is 2 in our example) we need both the OS Family ('Windows') and the version number ('8.1.0'). The Synchronization Data Source is not capable of doing this composite lookup, this where the LookupTable comes into play.

$oOSVersionLookup = new LookupTable('SELECT OSVersion', array('osfamily_id_friendlyname', 'name'));

This will build - in memory - the following table:

So if we have in $aData the following values:

Calling:

$oOSVersionLookup->Lookup($aData, array('osfamily_id', 'osversion_id'), 'osversion_id', 0);

Will place in the column 'osversion_id' the result of the lookup for the values $aData['osfamily_id'] and $aData['osversion_id'].

$aData will then contain the following values:

We then have to configure the Synchro Data Source so that it accepts the oversion_id as-is without performing any reconciliation on it.

The advanced reconciliation works by retrieving (via the REST/JSON API), the objects to be matched against the composite key, after the data collection but before pushing the data to iTop. Therefore, in order to use this advanced lookup mechanism, you must tell the framework that the collector has to reprocess the collected data before the actual synchro. This is achieved by overloading the method MustProcessBeforeSynchro of the collector; and returning true.

The collector framework provides two additional methods which can be overloaded:

1. InitProcessBeforeSynchro is called after the data collection, but before starting to reprocess each line of the collected data. This is the plece where to create the LookupTable instance
2. ProcessLineBeforeSynchro is called for each line of the collected data (including the header line of the CSV file, which index is zero)

The following code fragment shows to use cases of lookup tables altogether: one for brand + model and one for OS family + OS version.

testcollector.class.inc.php

protected function MustProcessBeforeSynchro()
{
        // We must reprocess the CSV data obtained from the inventory script
        // to lookup the Brand/Model and OSFamily/OSVersion in iTop
        return true;
}
 
protected function InitProcessBeforeSynchro()
{
        // Retrieve the identifiers of the OSVersion since we must do a lookup based on two fields: Family + Version
        // which is not supported by the iTop Data Synchro... so let's do the job of an ETL
        $this->oOSVersionLookup = new LookupTable('SELECT OSVersion', array('osfamily_id_friendlyname', 'name'));          
 
        // Retrieve the identifiers of the Model since we must do a lookup based on two fields: Brand + Model
        // which is not supported by the iTop Data Synchro... so let's do the job of an ETL
        $this->oModelLookup = new LookupTable('SELECT Model', array('brand_id_friendlyname', 'name'));             
}
 
protected function ProcessLineBeforeSynchro(&$aLineData, $iLineIndex)
{
        // Process each line of the CSV
        if (!$this->oOSVersionLookup->Lookup($aLineData, array('osfamily_id', 'osversion_id'), 'osversion_id', $iLineIndex))
        {
            throw New IgnoredRowException('Unknown OS Version');
        }
        if (!$this->oModelLookup->Lookup($aLineData, array('brand_id', 'model_id'), 'model_id', $iLineIndex))
        {
            throw New IgnoredRowException('Unknown Model');
        }
}

When troubleshooting the reconciliation mechanism it is useful to compare the original (raw) values as reported by the inventory script with the result of the reconciliation process. Whenever the method MustProcessBeforeSynchro of a collector returns true, the framework generates two files inthe data subdirectory. You can easily compare the values before/after the lookup by comparing the two CSV files:

1. <collector_name>.raw-<index>.csv: the original data, as produced by the inventory script,
2. <collector_name>-<index>.csv: the reprocessed data, to be uploaded to iTop.

A collector module can provide default values for its parameters by providing a file params.distrib.xml in the collectors folder. If such a file exists, its values are merged over the equivalent file in the conf directory.

The collector can specify, per class, optional fields which may exist or not in the targeted iTop. For this the collector must implement the method AttributeIsOptional:

AttributeIsOptional

      public function AttributeIsOptional($sAttCode): bool
        {
                if ($sAttCode == 'services_list') return true;
 
                return parent::AttributeIsOptional($sAttCode);
        }

It may happen that the target Data Model has some variants (depending on the set of modules chosen during the installation). If a given attribute can be missing in some configurations, you can tell your collector to accept this variation, by overloading the method AttributeIsOptional. (This is simpler than writing a specific collector for each combination).

If an attribute specified in the JSON definition of the Synchro Data Source is missing, the processing will stop with an error, unless this attribute is declared as optional. In the later case, the name of the skipped attribute is recorded in the protected member variable $this->aSkippedAttributes and the processing continues. The code of the collector can later check the content of the array $this->aSkippedAttributes to determine which fields have to be collected or not.

Example of implementation of AttributeIsOptional as a method of the VirtualMachineCollector class:

testcollector.class.inc.php

// In your Collector class, overwrite this method to specify iTop fields which can be optional
public function AttributeIsOptional($sAttCode)
{
        // If the module Service Management for Service Providers is selected during the setup
        // there is no "services_list" attribute on VirtualMachines. Let's safely ignore it.
        if ($sAttCode == 'services_list') return true;
 
        return parent::AttributeIsOptional($sAttCode);
}

Before version 1.3.0 of the collector, activation was controlled trough the main.php file only.

From version 1.3.0 onwards, the activation can be controlled by the collection plan (see below):

• You can manually specify which collector to run and their order in XML within params.local.xml (or params.distrib.xml for a generic collector)
• The collector itself can manage its dependencies and decide upon its activation or not.

To benefit from this feature, your collector must be based on collector base 1.3.0 or more.

A collection plan must be defined for your collector, which goal is to:

• detect the specificities of the iTop target, like the presence of an optional extension,
• retrieve from iTop information or objects that your collector depends on,
• simplify the main.php where the different collector classes were orchestrated.

It implements the CollectionPlan class defined in the collector core, eg. with the LDAP collector:

<?php
 
 
abstract class LDAPCollectionPlan extends CollectionPlan
{
        public function __construct()
        {
                parent::__construct();
        }
 
        public function Init(): void
        {
                parent::Init();
 
                // Add your specificities here
        }
}

The collection plan is instanciated through the main.php file, eg. for the LDAP collector:

main.php

<?php
// Initialize collection plan
require_once(APPROOT.'collectors/src/LDAPCollectionPlan.class.inc.php');
require_once(APPROOT.'core/orchestrator.class.inc.php');
 
Orchestrator::UseCollectionPlan('LDAPCollectionPlan');

A Launch sequence must be defined in the params.distrib.xml file. It must list the classes that you wish to collect and their relative order.

params.distrib.xml

<collectors_launch_sequence type="array">
   <collector>
      <!-- first_Class-->
      <name>FirstCollector</name>
      <enable>yes</enable>
      <rank>10</rank>
   </collector>
   <collector>
      <!-- second_Class-->
      <name>SecondCollector</name>
      <enable>yes</enable>
      <rank>20</rank>
   </collector>
   <collector>
      <!-- third_Class-->
      <name>ThirdCollector</name>
      <enable>no</enable>
      <rank>30</rank>
   </collector>
</collectors_launch_sequence>

Next to the launch sequence, a class can check if all its requirements to run correctly are met or not and potentially decide not to run. This is done through the CheckToLaunch method.

main.php

class MyCollector extends Collector
{
 
   public function CheckToLaunch($aOrchestratedCollectors): bool
   {
      if (conditions_are_met) {
         return true;
      } else {   
         return false;
      }
   }
}

The following structure must be implemented (paths are relative to <my-collector>/collectors/ directory):

• all json files where the synchro data source are defined reside under /json
• all PHP files for the different classes to collect are stored under /src
• the /extensions directory will host add-ons created by customers,
• all other files like main.php, params.distrib.xml or module.my-collector.php remain at the root.

If your collector needs a specific extension (or a minimum PHP version), you can specify this dependency by calling the static method Orchestrator::AddRequirement($sMinRequiredVersion, $sExtension = 'PHP') in main.php:

For example:

main.php

Orchestrator::AddRequirement('5.4.0'); //This requires at least PHP 5.4
Orchestrator::AddRequirement('1.2.0', 'ldap'); //This requires at least the ldap extension version 1.2.0

Assuming that

1. you want to synchronize the iTop class MyClass
2. this class contain one TagSet field with the code id tagset_code,
3. the primary_key of the Source is stored in an iTop field primary_key_code

The below code, will store in $aTaxons for each primary_key, the source value for the TagSet field, then after the synchronization (so the creation/update of objects in iTop), will do an API REST call to update one by one, each object with its TagSet value.

Just replace MyClass, tagset_code and primary_key_code with the correct values. Replace also Collecteur xxx by something meaningful, it will be stored in iTop history to describe who made that change.

class Mycollector extends XXXCollector
{
  protected $aTaxons;
 
  public function Fetch()
  {
     $aObject = parent::Fetch();
     $this->aTaxons[$aObject['primary_key']] = $aObject['tagset_code'];
  }  
 
  public function Synchronize($iMaxChunkSize = 0)
  {
    $iErrors = parent::Synchronize($iMaxChunkSize);
    $oClient = new RestClient();
    foreach($this->aTaxons as $primary_key => $sTaxon) {
      $aResult = $oClient->Update(
         'MyClass', 
         array('primary_key_code' => $primary_key), 
         array('tagset_code' => $sTaxon), 
         'Collecteur xxx'
      );
      if ($aResult['code'] == 0) {
        utils::Log(LOG_INFO, "TagSet successfully updated.");
      }   else {
        utils::Log(LOG_ERR, 
           "TagSet update failed. Code: {$aResult['code']}, message: {$aResult['message']}");
      }
    }
  }
}

If the 3rd assumption of the above example is not true, it is still possible to update, by retrieving the iTop object using the reconciliation criteria defined within the DataSynchro, but the code is more complex

class Mycollector extends XXXCollector
{
  protected $aTaxons;
  protected $aReconciliationKeys;
 
  protected function CheckDataSourceDefinition($aSourceDefinition)
  {
    $this->aReconciliationKeys = [];
    if ($aExpectedSourceDefinition['reconciliation_policy'] == 'use_attributes') {
      foreach($aExpectedSourceDefinition['attribute_list'] as $aAttributeDef) {
        if ($aAttributeDef['reconcile'] == '1') {
           $this->aReconciliationKeys[] = $aAttributeDef['attcode'];
        } 
      }
      if (count($this->aReconciliationKeys) == 0) {
        $this->aReconciliationKeys[] = 'id'; // iTop key provided for reconciliation
      }
    } 
    return parent::CheckDataSourceDefinition($aSourceDefinition);
  }
 
  public function Fetch()
  {
     $aObject = parent::Fetch();
     $aTaxon = [];
     $aTaxon['fields'] = array('tagset_code' => $aObject['tagset_code']);
 
     foreach ($this->aReconciliationKeys as $sKeyCode) {
       if ($sKeyCode == 'id') {
          $aTaxon['conditions'] = $aObject['primary_key'];
       }
       else {
           $aTaxon['conditions'][] = array( $sKeyCode => $aObject[$sKeyCode] );
       }
     }
     $this->aTaxons[] = $aTaxon; 
  }  
 
  public function Synchronize($iMaxChunkSize = 0)
  {
    $iErrors = parent::Synchronize($iMaxChunkSize);
    $oClient = new RestClient();
    foreach($this->aTaxons as $aTaxon) {
      $aResult = $oClient->Update(
         'MyClass', 
         $aTaxon['conditions'], 
         $aTaxon['fields'], 
         'Collecteur xxx'
      );
      if ($aResult['code'] == 0) {
        utils::Log(LOG_INFO, "TagSet successfully updated.");
      }   else {
        utils::Log(LOG_ERR, 
            "TagSet update failed. Code: {$aResult['code']}, message: {$aResult['message']}");
      }
    }
  }
}

🚧: Work in progress, code is wrong for now

class Mycollector extends XXXCollector
{
  protected $aUpdatedFiles; // To store the files which must be updated because they have changed
 
  public function Fetch()
  {
     $aObject = parent::Fetch();
     // If the file need to be updated
       //Then...
       // Get the file $aFile with name, contents and mimetype
       // Store the file locally
       $sFileName = '????'; //TODO provide the path to the file
       $this->aUpdatedFiles[$aObject['primary_key']] = array(
        'data' => base64_encode(file_get_contents($sFileName)),
        'filename' => basename($sFileName),
        'mimetype' => mime_content_type($sFileName), //Or fixed value e.g. .pdf => 'application/pdf'
       );
  }  
 
  public function Synchronize($iMaxChunkSize = 0)
  {
    $iErrors = parent::Synchronize($iMaxChunkSize);
 
    // After objects creation/update in iTop
    $oClient = new RestClient();
    foreach($this->aUpdatedFiles as $primary_key => $aFile) {
      $aResult = $oClient->Update(
         'MyClass', 
         array('primary_key_code' => $primary_key), // Retrieve the iTop object
         array('file_code' => $aFile), 
         'Collecteur xxx'
      );
      if ($aResult['code'] == 0) {
        utils::Log(LOG_INFO, "$aFile['filename'] successfully uploaded.");
      }   else {
        utils::Log(LOG_ERR, 
           "REST/JSON upload failed for file $aFile['filename']. 
           Code: {$aResult['code']}, 
           message: {$aResult['message']}");
      }
    }
  }
}