Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 
Anonymous
Not applicable

Spliting mutliple information from one row into multiple columns

Hello Everywhere,

I have a problem here and hope it is solvable.

I have a table with 2 columns (key_id, content)

An Example would be:

Key IDProperties
ID_1shape#circle$color#red$diameter#4inch
ID_2shape#circle$color#blue$diameter#4inch
ID_3shape#circle$color#red$diameter#9inch
ID_4shape#square$color#red$diameter#9inch

Now I filter on color=red and see ID_1, ID_3 and ID_4

Now I would like to continuing to filter. but due to the fact that the property is in one column I'm not able to.

Is it possible to change the row content into columns without knowing the number of properties used and have the column names be set to the property name? A property is separated by $ and the property name is before # and the value after #?

I have 100 different properties and new ones can come in on a random basis so it's hard to track and a lot of maintenance.

Thanks in advance.

Pijatida

3 Replies
sunny_talwar

Try this

Table:

LOAD [Key ID],

SubField(Properties, '#', 1) as Header,

SubField(Properties, '#', 2) as Value;

LOAD [Key ID],

SubField(Properties, '$') as Properties;

LOAD * INLINE [

    Key ID, Properties

    ID_1, shape#circle$color#red$diameter#4inch

    ID_2, shape#circle$color#blue$diameter#4inch

    ID_3, shape#circle$color#red$diameter#9inch

    ID_4, shape#square$color#red$diameter#9inch

];


FinalTable:

LOAD DISTINCT [Key ID]

Resident Table;


FOR i = 1 to FieldValueCount('Header')


LET vHeader = FieldValue('Header', $(i));

Left Join (FinalTable)

LOAD DISTINCT [Key ID],

Value as [$(vHeader)]

Resident Table

Where Header = '$(vHeader)';

NEXT


DROP Table Table;

Anonymous
Not applicable
Author

That would imply that the number of keys is fixed.

I have couple million rows though

sunny_talwar

Are you talking about the inline load ? That was just for an example... you can use this on top of your existing table like this

Table:

LOAD [Key ID],

SubField(Properties, '#', 1) as Header,

SubField(Properties, '#', 2) as Value;

LOAD [Key ID],

SubField(Properties, '$') as Properties;

LOAD *

FROM .....;


FinalTable:

LOAD DISTINCT [Key ID]

Resident Table;


FOR i = 1 to FieldValueCount('Header')


LET vHeader = FieldValue('Header', $(i));

Left Join (FinalTable)

LOAD DISTINCT [Key ID],

Value as [$(vHeader)]

Resident Table

Where Header = '$(vHeader)';

NEXT


DROP Table Table;