Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
Loading GraphQL data (GitHub API v4) into PowerBI
// This script shows how to use M language (Power Query Formula Language)
// to read data from GitHub API v4 using a POST request.
// This can come in handy when building PowerBI reports that utilize GraphQL endpoints for loading data.
let
vUrl = "https://api.github.com/graphql",
vHeaders =[
#"Method"="POST",
#"Content-Type"="application/json",
#"Authorization"="Bearer <your_personal_token_here>"
],
// Notice the quote escaping here
vContent=Text.ToBinary("{""query"": ""{ organization(login: \""github\"") { name }}""}"),
Source = Web.Contents(vUrl, [Headers=vHeaders, Content=vContent]),
#"JSON" = Json.Document(Source)
in
#"JSON"
@vasugowda
Copy link

vasugowda commented Sep 30, 2020

@MikeAinOz and @petrsvihlik
Can you please help in resolving below 404 issue while making graphql query

Step 1: Peform this activity in power bi desktop, creating Blank query with below content
let
Source = Web.Contents(
"https://",
[
Headers=[
#"Method"="POST",
#"Content-Type"="application/json",
#"Authorization"=""
],
// Notice the quote escaping here
Content=Text.ToBinary("{""query"": """"}")
]
),
#"JSON" = Json.Document(Source)
in
#"JSON"

Step 2: Get content and split it into multiple columns. This works
Step 3: Save Table. This works
Step 4: Create a report using the table data. This works
Step 5: Save and publish to power bi online service. This will create the report
Step 6: Data connection will throw 404 error
RCA: What we noticed is when we create the query in power bi desktop the request is HTTP POST. Whereas after publishing the request are made with HTTP GET.
Is there a way to change the HTTP request method in published report/data model.

@MikeAinOz
Copy link

MikeAinOz commented Oct 1, 2020

What are you trying to do with a blank query? I don't understand.

@vasugowda
Copy link

vasugowda commented Oct 5, 2020

@MikeAinOz - Sorry this is not blank query, i tried to explain with the template above. The actual query will return data. As mentioned above this will generate the report in power bi desktop ( as per the template it makes POST call). Where-as once we publish for delta data power bi makes HTTP GET call instead of HTTP POST call.

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 5, 2020

@vasugowda seems that you're not the only one experiencing this issue.
check out this: https://stackoverflow.com/q/48385504/1332034

I really don't know what's going on. It works perfectly fine for me. Try specifying the headers through dedicated variables as suggested here:

let
vBaseUrl = "https://api.projectmanager.com",
vPath = "api/v1/projects.json",
vHeaders = [ #"apikey"="this is where i put my key" ],
Source = Web.Contents ( vBaseUrl, [RelativePath=vPath, Headers=vHeaders] ),
Json = Json.Document ( Source )
in
Json

does it work for you?

@Priyanka2719
Copy link

Priyanka2719 commented Oct 6, 2020

@petrsvihlik Thank you for your response. Yes, by sending the headers & content as suggested in your comment helped us resolve our issue. Thank you for your help.
The only change we made was to pass content also as a part of "Web. Contents".
let
vBaseUrl = "https://api.projectmanager.com",
vPath = "api/v1/projects.json",
vHeaders = [ #"apikey"="this is where i put my key" ],
Source = Web.Contents ( vBaseUrl, [RelativePath=vPath, Headers=vHeaders, Contents=<Query_content>] ),
Json = Json.Document ( Source )
in
Json
@vasugowda This worked as expected

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 6, 2020

thanks for the heads up, I updated the sample!

@michelsolognier
Copy link

michelsolognier commented Oct 9, 2020

Hi all, i've tried all of the above advises. I managed to connect power bi via the graphql data of github and got the same result as @petrsvihlik. However now i want to move to a next step by trying to a request specific dataset from a monitoring platform. The query i need to specify needs to follow the structure as follows:

query {
  historicalAlarms(before: "2020-11-01T12:15:04Z", after: "2020-09-30T15:04:05Z", limit: 2){
    edges {
      node {
        addedAt
        comingGoing
        aggregateID
        aggregateName
        aggregateText
        aggregateDescription
      }
    }
  }
}

I tested this in postman (with its header containing the Authorization key) and is works fine!

Now i need to pack this Query up as a content to send this via de POST req. Only i can't seem to get this right. I keep getting an error code 400: BAD REQUEST.Does anyone have any idea? My code in the Power BI Query Editor is as follows:

let
    Source = Web.Contents(
	"http://<......>/api/partner/gql",
	[
		Headers=[
			#"Method"="POST",
			#"Content-Type"="application/json",
			#"Authorization"="Bearer <key>"
		],
		Content=Text.ToBinary("{""query"": ""{ historicalAlarms (before: \""2020-11-01T12:15:04Z\"", after: \""2020-09-30T15:04:05Z\"", limit: 2) {
            edges {
                node {
                addedAt
                comingGoing
                aggregateID
                aggregateName
                aggregateText
                aggregateDescription
                }
            }
        }
    }

    ""}")]),
    #"JSON" = Json.Document(Source)
in
    #"JSON

After a while i made some changes, but now im getting a error code 422: Unprocessable entity. So i think the request is now working fine, only now somewhere in the content, one or more parts isin't suitable to be processed. Am i understanding this right?

let
    vUrl = "http://<............>",
    vPath = "api/partner/gql",
    vHeaders =[
			#"Method"="POST",
			#"Content-Type"="application/json",
			#"Authorization"="Bearer <KEY>"
		],
    vContent=Text.ToBinary("{""query"": ""{   historicalAlarms (before: \""2020-11-01T12:15:04Z\"", after: \""2020-09-30T15:04:05Z\"", limit: 2) {edges{node{addedAT comingGoing aggregateID aggregateName aggregateText aggregateDescription}}}         }""}"),
    Source = Web.Contents(vUrl,	[RelativePath=vPath, Headers=vHeaders, Content=vContent]),
    #"JSON" = Json.Document(Source)
in
    #"JSON"

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 9, 2020

@michelsolognier there must not be any line breaks. try the JS fiddle I posted above to escape the query.

just by looking at your code, I think you're missing the leading query {...

@michelsolognier
Copy link

michelsolognier commented Oct 9, 2020

ahh... i knew i was hanging on a silly mistake. It works now! Many thanks @petrsvihlik!

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 9, 2020

@mikefede
Copy link

mikefede commented Oct 20, 2020

Thank you! This, and the JS fiddle to escape the query, has been very helpful.

Would you be able to provide an example that makes multiple requests to get multiple pages of data and combine them? GitHub only allows you to request 100 items at a time and to get all of the data you need to get the pageInfo for the endCursor to request the next page of data. I'm struggling to figure out how to do that in Power BI. Thanks!

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 21, 2020

@mikefede Thanks :) Regarding the pagination, have you tried this? https://datachant.com/2016/06/27/cursor-based-pagination-power-query/
It seems pretty straightforward and I think it should work. If it does, could you please let everybody know here in the discussion (ideally, with a piece of sample code)?

@mikefede
Copy link

mikefede commented Oct 21, 2020

@petrsvihlik Thank you again! Yes this worked, see the sample code below.
I have one other question, do you have any pointers on how we might store the personal access token securely, rather than just embedding it in the script in plain text?

Here's the sample code. This example gets a list of SAML Identities with their corresponding GitHub logins. There's probably a more elegant way to insert the cursor into the query, but this works.

let
	FnGetOnePage = (cursor) =>
	let		
		Source = Json.Document(Web.Contents(
		"https://api.github.com/graphql",
		[
			Headers=[
				#"Method"="POST",
				#"Content-Type"="application/json",
				#"Authorization"="Bearer <your_personal_token_here>"
			],
		Content=Text.ToBinary(Text.Replace("{""query"": ""query($cursor: String) { organization(login: \""<your_organization_here>\"") { samlIdentityProvider { externalIdentities(first: 100, after:$cursor) { totalCount edges { node { guid samlIdentity { nameId } user { login } } } pageInfo { endCursor hasNextPage } } } }}""}", "($cursor: String)", "($cursor: String" & cursor & ")"))
		]
		)),
		data = Source[data],
		organization = data[organization],
		samlIdentityProvider = organization[samlIdentityProvider],
		externalIdentities = samlIdentityProvider[externalIdentities],
		edges = externalIdentities[edges],
		pageInfo = externalIdentities[pageInfo],
		hasNextPage = pageInfo[hasNextPage],
		endcursor = pageInfo[endCursor],
		res = [Data=edges, Cursor=endcursor, HasNext=hasNextPage]
	in 
		res,
	GeneratedList =
		List.Generate(
			()=>[res = FnGetOnePage("")],
			each Comparer.Equals(Comparer.FromCulture("en-us"), [res][HasNext], true),
			each [res = FnGetOnePage("=\""" & [res][Cursor] & "\""")],
			each [res][Data] 
		),
    #"Converted to Table" = Table.FromList(GeneratedList, Splitter.SplitByNothing(), null, null, ExtraValues.Error),
    #"Expanded Column1" = Table.ExpandListColumn(#"Converted to Table", "Column1"),
    #"Expanded Column2" = Table.ExpandRecordColumn(#"Expanded Column1", "Column1", {"node"}, {"Column1.node"}),
    #"Expanded Column1.node" = Table.ExpandRecordColumn(#"Expanded Column2", "Column1.node", {"samlIdentity", "user"}, {"Column1.node.samlIdentity", "Column1.node.user"}),
    #"Expanded Column1.node.samlIdentity" = Table.ExpandRecordColumn(#"Expanded Column1.node", "Column1.node.samlIdentity", {"nameId"}, {"Column1.node.samlIdentity.nameId"}),
    #"Expanded Column1.node.user" = Table.ExpandRecordColumn(#"Expanded Column1.node.samlIdentity", "Column1.node.user", {"login"}, {"Column1.node.user.login"}),
    #"Filtered Rows" = Table.SelectRows(#"Expanded Column1.node.user", each [Column1.node.user.login] <> null and [Column1.node.user.login] <> ""),
    #"Renamed Columns" = Table.RenameColumns(#"Filtered Rows",{{"Column1.node.samlIdentity.nameId", "SAML Identity"}, {"Column1.node.user.login", "GitHub Login"}})
in
    #"Renamed Columns"

@petrsvihlik
Copy link
Author

petrsvihlik commented Oct 22, 2020

@mikefede - perfect! thanks for sharing the code!

regarding the secure access storage - well, there are two ways of NOT including the API keys in the PBIXes that I'm aware of:

  • keep them in separate storage such as an excel file...this doesn't make it any safer but at least the keys don't travel with the PBIX file.
  • or better, use the data source permissions as described here

image

@pramitnb1990
Copy link

pramitnb1990 commented Mar 8, 2021

Hi , Thank you for this post - it really helped me a lot .
I am trying to connect POWEBI to API end point and put in the GraphQL query which is running perfectly on POSTMAN but somehow not working when I write the m code integrating the GraphQL query
let
vUrl = "http://<............>",
vHeaders =[
#"Method"="POST",
#"Content-Type"="application/json",
#"Authorization"="Bearer "
vContent=Text.ToBinary("{""query"": ""{ table_name(where: {colX: {_eq: ""ABC""},
colY:{_eq: ""XYZ""}}limit: 10){
colA
colX
colY

}}""}"),
Source = Web.Contents(vUrl, [Headers=vHeaders, Content=vContent]),
#"JSON" = Json.Document(Source)

However I am getting the below error:
Error in $: Failed reading: satisfy. Expecting ',' or '}' at 'ABC},'
image

Can someone please help, I am stuck at this for almost 4 days :( :(

Sincere Regards

@MikeAinOz
Copy link

MikeAinOz commented Mar 8, 2021

Hey friend! Just looking at it I can see that you haven't terminated the header record with "]," before you define vContent. Is this the problem?

@pramitnb1990
Copy link

pramitnb1990 commented Mar 8, 2021

Hey friend! Just looking at it I can see that you haven't terminated the header record with "]," before you define vContent. Is this the problem?

Hi MikeAinOz, Thank you for taking out your time, really appreciate it!
No it seems not, I corrected it but the error persists :(

Kind Regards

@MikeAinOz
Copy link

MikeAinOz commented Mar 9, 2021

Hey friend! Just looking at it I can see that you haven't terminated the header record with "]," before you define vContent. Is this the problem?

Hi MikeAinOz, Thank you for taking out your time, really appreciate it!
No it seems not, I corrected it but the error persists :(

Kind Regards

I'd check that content in a separate variable so that you can see what is in it, have you tried that?

@pramitnb1990
Copy link

pramitnb1990 commented Mar 23, 2021

Hey friend! Just looking at it I can see that you haven't terminated the header record with "]," before you define vContent. Is this the problem?

Hi MikeAinOz, Thank you for taking out your time, really appreciate it!
No it seems not, I corrected it but the error persists :(
Kind Regards

I'd check that content in a separate variable so that you can see what is in it, have you tried that?

Hi MikeAinOZ, Thank you for your valuable input, it worked :) Really appreciate your help!

@anon145
Copy link

anon145 commented Mar 25, 2021

@MikeAinOz @petrsvihlik
Hi Gents, great resource thanks for sharing.
If anyone has a moment to be able to help with my pagination function, i'd greatly appreciate it :)
I'm really struggling but I think what i'm after is -
For the function to pull the first 1000 nodes, check the edge "pageinfo" for "hasNextPage" = true, then insert "endCursor" as the "after:"endCursor" on the next iteration, until "hasNextPage" = false?

Does that sound right?

eg.
"pageInfo": {
"endCursor": "YXJyYXljb25uZWN0aW9uOjA=",
"hasNextPage": true,
"hasPreviousPage": false,
"startCursor": "YXJyYXljb25uZWN0aW9uOjA="
}

Working query without pagination
let
Source = Web.Contents(
"https://website/graphql",
[
Headers =[
#"Method"="POST",
#"Content-Type"="application/json",
#"x-developer-secret"="pass"
],
Content=Text.ToBinary("{""query"": ""query { getResults( first: 1000, input: { dates: { from: ""2021-03-02 00:00:00"", to: ""2021-03-17 23:59:00"" }}){ edges { cursor node { details } } pageInfo { startCursor endCursor hasNextPage hasPreviousPage } }}""}")
]
),
#"JSON" = Json.Document(Source),
#"Converted to Table" = Record.ToTable(JSON),

@petrsvihlik
Copy link
Author

petrsvihlik commented Mar 25, 2021

@anon145 Did the approach suggested by @mikefede not work? You need to use List.Generate and recursively call the query while altering the cursor.

@anon145
Copy link

anon145 commented Mar 25, 2021

@petrsvihlik Hi, I'm sure it would work, I'm just struggling as this is new, all the examples seem to differ slightly :)

@petrsvihlik
Copy link
Author

petrsvihlik commented Mar 25, 2021

Try this sample, make sure it works for you, and make sure you understand it. Then, cut the stuff you don't need and try to make it simpler. Next, replace the query with yours and work from there. If you get stuck, post your query here and we can start elaborating on what's going on wrong :)

@anon145
Copy link

anon145 commented Mar 27, 2021

Try this sample, make sure it works for you, and make sure you understand it. Then, cut the stuff you don't need and try to make it simpler. Next, replace the query with yours and work from there. If you get stuck, post your query here and we can start elaborating on what's going on wrong :)

Thanks for the push @petrsvihlik, managed to get it this far!
The FnGetOnePage function appears to be ok but the query fails on next step GenerateList:
Expression.Error: 1 arguments were passed to a function which expects 3.
Details:
Pattern=
Arguments=[List]

This is how far I got, if anyone can help i'm out of my depth here :)
_let
FnGetOnePage = (cursor) =>
let
Source = Json.Document(Web.Contents(
"https://website/graphql",
[
Headers =[
#"Method"="POST",
#"Content-Type"="application/json",
#"x-developer-secret"="pass"
],
Content=Text.ToBinary(Text.Replace("{""query"": ""query ($cursor: String) { getResults(first: 1000, input: {dates: {from: ""2021-03-02 00:00:00"", to: ""2021-03-17 23:59:00""}}, after: $cursor) { edges { cursor node { "fields" } } } pageInfo { endCursor hasNextPage } }}""}"),
"($cursor: String)", "($cursor: String" & cursor & ")")
]
)),
data = Source[data],
getResults= data[getResults],
edges = getResults[edges],
pageInfo = getResults[pageInfo],
hasNextPage = pageInfo[hasNextPage],
endcursor = pageInfo[endCursor],
res = [Data=edges, Cursor=endcursor, HasNext=hasNextPage]
in
res,
GeneratedList =
List.Generate(
()=>[res = FnGetOnePage("")],
each Comparer.Equals(Comparer.FromCulture("en-us"), [res][HasNext], true),
each [res = FnGetOnePage("=""" & [res][Cursor] & """")],
each [res][Data]
).........

@Wisteru
Copy link

Wisteru commented Apr 1, 2021

Try this sample, make sure it works for you, and make sure you understand it. Then, cut the stuff you don't need and try to make it simpler. Next, replace the query with yours and work from there. If you get stuck, post your query here and we can start elaborating on what's going on wrong :)

hello @petrsvihlik , I'm trying to use this example, but i have a problem with last page. It's not getting into GeneratedList because "HasNext" is false for last page. Do you maybe know how to handle it?

@petrsvihlik
Copy link
Author

petrsvihlik commented May 2, 2021

@Wisteru you are correct, I was able to reproduce it. will try to come up with a fix.

@petrsvihlik
Copy link
Author

petrsvihlik commented May 2, 2021

@Wisteru see this: https://www.thebiccountant.com/2020/05/15/miss-last-page-paging-power-bi-power-query/

Ben Gribaudo says:

The way documentation describes List.Generate’s third argument (the test function) can be a confusing. That argument is applied to the just-fetched result (page, in this case) to determine both whether to return the current result (page) and whether to try to fetch another result (page)—which will then be handed to the test function which will then determine whether it should be returned and another fetched, and so on…

I was kinda successful with replacing each Comparer.Equals(Comparer.FromCulture("en-us"), [res][HasNext], true), with each [res][Data] <> null,. There are still some errors in the resulting dataset however the data is complete and the errors can be filtered out.

@talynone
Copy link

talynone commented May 6, 2021

I was kinda successful with replacing each Comparer.Equals(Comparer.FromCulture("en-us"), [res][HasNext], true), with each [res][Data] <> null,. There are still some errors in the resulting dataset however the data is complete and the errors can be filtered out.

@petrsvihlik Try this instead:

each [res][Cursor] <> null,

@petrsvihlik
Copy link
Author

petrsvihlik commented May 11, 2021

@talynone yes, that works! perfect :)

So my resulting query looks like this:

let
	FnGetOnePage = (cursor) =>
	let		
		Source = Json.Document(Web.Contents(
		"https://api.github.com/graphql",
		[
			Headers=[
				#"Method"="POST",
				#"Content-Type"="application/json",
				#"Authorization"="Bearer ******"
			],
		Content=Text.ToBinary(Text.Replace("{""query"": ""query ($cursor: String) { repository(name: \""WopiHost\"", owner: \""petrsvihlik\"") { stargazers(first: 10, after: $cursor) { edges { node { login } starredAt } pageInfo { endCursor hasNextPage } } }}""}", "($cursor: String)", "($cursor: String" & cursor & ")"))
		]
		)),
		data = Source[data],
		repository = data[repository],
		stargazers = repository[stargazers],
		edges = stargazers[edges],
		pageInfo = stargazers[pageInfo],
		hasNextPage = pageInfo[hasNextPage],
		endcursor = pageInfo[endCursor],
		res = [Data=edges, Cursor=endcursor, HasNext=hasNextPage]
	in 
		res,
	Pages =
		List.Generate(
			()=>[res = FnGetOnePage("")],
			each [res][Cursor] <> null,
			each [res = try FnGetOnePage("=\""" & [res][Cursor] & "\""") otherwise null],
			each if [res] <> null then [res][Data] else null
		),
    #"Pages to Table" = Table.FromList(Pages, Splitter.SplitByNothing(), null, null, ExtraValues.Error),
    #"Renamed Columns" = Table.RenameColumns(#"Pages to Table",{{"Column1", "Pages"}}),
    #"Pages to Rows" = Table.ExpandListColumn(#"Renamed Columns", "Pages"),
    #"Renamed Columns1" = Table.RenameColumns(#"Pages to Rows",{{"Pages", "Records"}}),
    #"Expanded Column3" = Table.ExpandRecordColumn(#"Renamed Columns1", "Records", {"node", "starredAt"}, {"Records.node", "Records.starredAt"}),
    #"Expanded Column1.node" = Table.ExpandRecordColumn(#"Expanded Column3", "Records.node", {"login"}, {"Records.node.login"}),
    #"Renamed Columns2" = Table.RenameColumns(#"Expanded Column1.node",{{"Records.node.login", "login"}, {"Records.starredAt", "starredAt"}}),
    #"Changed Type" = Table.TransformColumnTypes(#"Renamed Columns2",{{"starredAt", type datetime}})
in
    #"Changed Type"

And the original (not inlined) GQL query:

query ($cursor: String) {
  repository(name: "WopiHost", owner: "petrsvihlik") {
    stargazers(first: 10, after: $cursor) {
      edges {
        node {
          login          
        }
        starredAt
      }
      pageInfo {
        endCursor
        hasNextPage        
      }
    }
  }
}

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment