Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: DOC-74 more updates - trying to explain lookups

...

Use an output multiplier to break out a record into multiple output recordsIn this exercise you will

load reference data into your model which lists all the TV channels offered in each package. So far your model has merged data from two files, then grouped and sorted the records. You now have a stream that has the latest data about the TV package that each customer has.

Image Added

We now want to get a set of records for each channel that the customer has ordered. In this section you are going to add information about which channels are included for each package

. This data is currently in a database table. PhixFlow can look-up this data. To make sure that PhixFlow finds all the channels, you will use an ouptut multiplier 

You will use this to break out each customer record, in Latest Package Update, into a set of records for each channel that the customer has ordered.

Load data from a database table

In this section First, you will load the database table SOURCE_PACKAGE_CHANNELS into a stream. Your model will then have the full list of channels available in each package.

...

  1. In the repository browser, expand the list of Datasources.
  2. Drag the datasource called Rating and Billing into your model.
  3. In your model, hover your mouse pointer over Rating and Billing and click 
    Insert excerpt
    _show_tables
    _show_tables
    nopaneltrue
    . PhixFlow displays a list of the tables available in this database. You can resize the list to make it easier to browse.
  4. Right-click on SOURCE_PACKAGE_CHANNELS and select 
    Insert excerpt
    _create_collector
    _create_collector
    nopaneltrue
    . PhixFlow adds a new database collector  to the model, and gives it the same name as the source data table, SOURCE_PACKAGE_CHANNELS. In the settings → Basic Settings section → Statement Expression, you can see the database query is: 
      select * from SOURCE_PACKAGE_CHANNELS.
  5. Hover your mouse pointer over the new database collector, and click 
    Insert excerpt
    _add_stream_with_attributes
    _add_stream_with_attributes
    nopaneltrue
    . PhixFlow adds a new stream to your model and also gives it the name SOURCE_PACKAGE_CHANNELS. In the settings → Attributes section, you can see that data has two attributes (columns): Package and Channel.
  6. To load the data into the stream, run analysis on this section of the model and check the data has loaded.

    Expand
    titleHow?
    1. Hover your mouse pointer over the SOURCE_PACKAGE_CHANNELS stream , click 
      Insert excerpt
      _run_analysis
      _run_analysis
      nopaneltrue
       then click YES to confirm.
    2. When PhixFlow notifies you that the analysis is finished, hover your mouse pointer over the stream again.
    3. Click 
      Insert excerpt
      _show_stream_views
      _show_stream_views
      nopaneltrue
      .
    4. In the drop-down list, select 
      Insert excerpt
      _default_view
      _default_view
      nopaneltrue
      .  PhixFlow displays the data.

    Snapshot of your model and the data you have loaded:

...

Take a moment to look at this data. For each package, there are multiple channels included. For example, the News package includes Channel 1, Channel 2, Science 2 and NewsExtra.

Add a stream ...

Update your model to add a stream. This stream will hold the full channel listtake all the data that we have merged, sorted and grouped in the model so far.

  1. Hover your mouse pointer over the Latest Package Update stream and click 
    Insert excerpt
    _add_stream
    _add_stream
    nopaneltrue
    .
  2. In Basic Settings, set the Name to Latest Package All Channels.
  3. Drag all the attributes from Latest Package Update to the settings for Latest Package All Channels.

    Expand
    titleHow?
    1. Hover your mouse pointer over the stream Latest Package Update.
    2. In the context toolbar, click 
      Insert excerpt
      _stream_attributes
      _stream_attributes
      nopaneltrue
      . Phixflow lists the data attributes (column titles) for this stream.
    3. Select all the attributes and drag them into the Latest Package All Channels settings → Attributes section.


  4. Click 
    Insert excerpt
    _finish
    _finish
    nopaneltrue
     to save your changes and close the settings tab.
  5. Close the attributes list.

Look up 

...

Supplement the stream with look-up data

You now need to add to the data in the Latest Package All Channels stream with additional data from the database, using a lookup pipe. You will use a look-up pipe to add information about the Package. PhixFlow needs to know how the data in the source stream matches the data in the receiving stream.

  1. In the model, add a pipe from SOURCE_PACKAGE_CHANNELS to Latest Package All Channels. PhixFlow opens the settings for the new pipe.
  2. In  In the Basic Settings section, set:
    • Name: chanfind_channels
    • Type: Look-up
  3. In the Order/Index section, click 
    Insert excerpt
    _stream_attributes
    _stream_attributes
    nopaneltrue
    . PhixFlow opens two lists of attributes: one list for the source stream and another for the receiving stream. Image Added
    1. From the source stream list
    of attributes available from the input stream, in this case PACKAGE and CHANNEL.Drag PACKAGE into the 
    1. , on the left, drag PACKAGE into the Order/Index grid.
      Image Added
      This tells PhixFlow about the key column in the source stream. Notice that PhixFlow has added a line to the Order/Index grid.
    2. From the receiving stream list, on the right, drag Package over the top of the line in the Order/Index grid.
      Image Added
      This tells PhixFlow about the matching key column in the receiving stream. Notice that PhixFlow has now added the Index Expression _outPackage. However, we do not want PhixFlow to use _outPackagebecause output multipliers are calculated before the output is generated. We need it to use the data on the input.
  4. In the Order/Index section, then double-click the the new line createdthat PhixFlow has added. PhixFlow opens the settings for the PACKAGE attribute. Set:
    •  Basic Settings → Index Expression from _out.Package to: in.Package.
      Technical issue: I'm not seeing _out.Package here.

    • Click 
      Insert excerpt
      _finish
      _finish
      nopaneltrue
       to save save  and close the attribute settings tab.
  5. Close the attribute list.
  6. In the pipe settings, click 
    Insert excerpt
    _finish
    _finish
    nopaneltrue
     to save your changes and close the settings tab.
  7. In the model toolbar, click 
    Insert excerpt
    _save
    _save
    nopaneltrue
     to save your model.

At this point, if you run your model (I think) it will find each customer, e.g. C010, and will read that the customer has the News package. It will then look-up the channel for this to find News: Channel 1. But having done one look-up for the customer it will move onto the next customer.


When you run your model, PhixFlow

This means that the lookup will match records in the reference data (SOURCE_PACKAGE_CHANNELS) where PACKAGE = Package from the input pipe in
?????We must do this, rather than using the value of Package from the output because we are using GOING TO USE the lookup to drive an output multiplier – and output multipliers are calculated before the output is generated

  • Open the configuration form for the stream Latest Package All Channels by double clicking it
  • In the field Output Multiplier, enter the value: chan.CHANNEL

This means that for each output record processed, PhixFlow will look up the list of channels for the package from the reference stream; it will then work through the list returned, handling each element from the list in turn and generating an output record in the stream for each one

  • Add an attribute to the stream with Name Channel
  • Set the Expression of this attribute to _outputMultiplier

This will print the value of the current element of the output multiplier list to this field

...

On older models in PhixFlow you may see the internal variable _type. This is the old name for _outputMultiplier. Either can be used, and anywhere you see _type you can update it to _outputMultiplier for clarity.

...

Set an output multiplier

What is an output Multiplier

An output multiplier tells PhixFlow to process the same record multiple times, each time producing a new record. This is called "breaking-out" a record. In this case we want PhixFlow to keep processing the same customer record until it has found all the channels. This means our single input record for each customer becomes multiple output records, like this:

  • C010  News  Channel 1
  • C010  News  Channel 2
  • C010  News  Channel 3
  • C010  News  NewsExtra

The output multiplier needs to know:

  • which pipe it will use -  find_channels pipe
  • what data you want to add - the channels.

The stream needs to know about the new column to create for the channel information.

  1. In the model, double-click Latest Package All Channels to open its settings tab.
  2. To create the output multiplier, expand the Multipliers and Filters section and set:
    • Output Multiplier: find_channels.CHANNEL
  3. To create a new column for the channel data, expand the Attributes section. Click 
    Insert excerpt
    _add_object
    _add_object
    nopaneltrue
     to add a new attribute. PhixFlow opens a new attribute settings tab. Set:
    •  Name: Channel
    •  Expression: _outputMultiplier
  4. Click 
    Insert excerpt
    _fini
    _fini
    nopaneltrue
     to save and close the settings tab.
  5. In the model toolbar, click 
    Insert excerpt
    _save
    _save
    nopaneltrue
     to save your model.


Now you can run the model and view the stream data.

  1. Hover your mouse pointer over the Latest Package All Channels stream and click 
    Insert excerpt
    _run_analysis
    _run_analysis
    nopaneltrue
     then click YES to confirm.
  2. When PhixFlow notifies you that the analysis is finished, hover your mouse pointer over the Latest Package All Channels stream again.
  3. Click 
    Insert excerpt
    _show_stream_views
    _show_stream_views
    nopaneltrue
    .
  4. In the drop-down list, select 
    Insert excerpt
    _default_view
    _default_view
    nopaneltrue
    .  PhixFlow displays the data.

What do I do if there is no data.

Something might have gone wrong with something