Forum Discussion
Multi-Dimensional Json file to Table
I have a Json file formatted like so:
{
"items": [
{
"tableName": "incidents",
"count": 20000,
"columnNames": [
"id",
"subject",
"category"
],
"rows": [
[
"61",
"Test",
null
],
[
"65",
"TEST 2",
null
], ........Etc
I pull this Json file using a web call to an API (specifically Oracle Right Now).
*Note that the column names are listed as a separate entitie to the rows.
When I follow the suggestions on here to expand out the data I am unable to get the column headings and rows in a single table. I can navigate to the column names list OR the rows list. I can transpose the column names list to table headings like so:
But then there are no rows to fill in the columns as the rows are in a higher \ adjacent level.
Navigating to the rows list simply gives me this:
How do I manipulate this Json file using Power Bi to display a single table with the columns as the columns and the rows as the rows?
I have spent probably too much time on this and as I'm new to M query language playing around in advanced editor has proven not very fruitful.
Any help will be much appreciated.
Hi aurischa，
Please follow the following steps to import your JSON file to Power BI.
1. Right click your lists and choose "Add as new query".
2. Convert your columnNames list to table and transpose the table.
3.Convert your rows list to table and rename the column as follows
4.Add custom columns as follows, For more details, please review this similar blog.
5. Remove first column in rows table.
6. Append the two queries as shown in the following screenshots.
7. Select the first row in columnNames table and click the "Use first row as headers" button, then get expected result.
.
Thanks,
Lydia Zhang
18 Replies
Hi aurischa，
Please follow the following steps to import your JSON file to Power BI.
1. Right click your lists and choose "Add as new query".
2. Convert your columnNames list to table and transpose the table.
3.Convert your rows list to table and rename the column as follows
4.Add custom columns as follows, For more details, please review this similar blog.
5. Remove first column in rows table.
6. Append the two queries as shown in the following screenshots.
7. Select the first row in columnNames table and click the "Use first row as headers" button, then get expected result.
.
Thanks,
Lydia ZhangHi Lydia,
Excellent solution, and thank you for your help. My only small qualm is that it may not be scalable should the json data set contain say 20 or 100 colums. However, before reading this I did recently come up with another way to achieve this solution.
Fortunately I was able to use ROQL syntax in the REST API call meaning I was able to format the JSON file as such.
?query=SELECT%20ID,%27֍%27,subject,%27֍%27,category,%27֍%27%20from%20incidents
Where ֍ is the delimiter.
{ "items": [ { "tableName": "Table0", "count": 20000, "columnNames": [ "id", "'֍'", "subject", "'֍'", "category", "'֍'" ], "rows": [ [ "61", "֍", "Test", "֍", null, "֍" ], [ "65", "֍", "TEST 2", "֍", null, "֍" ],Final step is to rename headings.
Now that I look at it there are a few redundant steps in there. But I achieved a similar result using delimiters programmaticly which someone might find useful.
It would be nice to be able to define the table structure of the json file for powerbi at the point of making the web calls instead of making the call importing the data and then trying to manipulate it. In PHP to achieve this I'd wrapping the data in an array of arrays and define the first element of all sub arrays as the ID field and wrtre out all the data in a for each loop. Having a UI to do this would be amazing. It just seems so many steps to convert what is already for the most part, structured data into a table.Regards,
Adam
Hi Lydia,
Excellent solution, and thank you for your help. My only small qualm is that it may not be scalable should the json data set contain say 20 or 100 colums. However, before reading this I did recently come up with another way to achieve this solution.
Fortunately I was able to use ROQL syntax in the REST API call meaning I was able to format the JSON file as such.
?query=SELECT%20ID,%27֍%27,subject,%27֍%27,category,%27֍%27%20from%20incidents
Where ֍ is the delimiter.
{ "items": [ { "tableName": "Table0", "count": 20000, "columnNames": [ "id", "'֍'", "subject", "'֍'", "category", "'֍'" ], "rows": [ [ "61", "֍", "Test", "֍", null, "֍" ], [ "65", "֍", "TEST 2", "֍", null, "֍" ],Final step is to rename headings.
Now that I look at it there are a few redundant steps in there. But I achieved a similar result using delimiters programmaticly which someone might find useful.
It would be nice to be able to define the table structure of the json file for powerbi at the point of making the web calls instead of making the call importing the data and then trying to manipulate it. In PHP to achieve this I'd wrapping the data in an array of arrays and define the first element of all sub arrays as the ID field and wrtre out all the data in a for each loop. Having a UI to do this would be amazing. It just seems so many steps to convert what is already for the most part, structured data into a table.Regards,
Adam
Like to share an easier solution here, as this seems to be quite a common scenario:
Table.Combine(List.Transform(items1[rows], each Table.FromRows({_}, items1[columnNames])))No need to expand anything from the record ("items1") here: Just reference the field "rows" (which contains a list) and transform each item of the list (which is a list as well, holding the row-values of the tables) to a table with the column headers from record-field "columnNames". This will return a list of tables which you simply combine (append).
BUT if you ever find yourself in a situation where you cannot "navigate" your solution directly like this, you should use this function:
Table.FromRecords( { YourJsonRecord } )This will do your multiple steps here in one go (don't forget the curly brackets around your record, as this function requires a list of records: Even if this list has just one item )
This is brilliant - thanks so much Lydia. I've been looking for an answer to this problem for about 4 days straight! :)
