This is explanation how, with python, to make an external Power BI Desktop tool that will export your DAX query to CSV. In Power BI desktop ribbon we have a button (1) "Export to CSV". This button will open window (2) where we have to type our DAX query. We already have template in this dialog to help us with writing. After this, we click on button (3) "Make CSV!". If something is wrong we will get message (4). If everything is OK, new file will appear on our desktop. That file will hold data from our query (5). Let's see how to make such external tool.

First step is to create JSON file which will tell Power BI desktop how to create button and what that button should do. We can see bellow sample of such file. Version is just a version of our tool. Just leave it as "1.0". Name is text that will appear bellow button. We can add some description of our tool. Path is fullpath to python executive file. Use backslashes and double them. Arguments has two parts. First part is fullpath to our python script. Second part are arguments that Power BI desktop will provide for our script. "Server" is address of our server ("localhost:57800"), and "database" is name of our database ("c6fb254f-f146-464d-99cf-b2a6e322eb38"). Our script will use name of a server for its execution. Last element is image in base64 format. Base64 format is textual presentation of an image. This format has two parts. First part is prefix, and prefix is always the same. Second part is base64 code. We can go to web site base64.guru and there we can upload our icon. Result will be long text that is base64 presentation of our image. This text will be really long, but all of it should be placed in JSON file, after the prefix.

{
  "version": "1.0",
  "name": "Export to CSV",
  "description": "Export user's DAX to CSV file.",
  "path": "G:\\WPy64-3950\\python-3.9.5.amd64\\pythonw.exe",
  "arguments": "C:/Users/Sima/Desktop/ExportToCSV.pyw \"%server%\" \"%database%\"",
  "iconData": "data:image/png;base64,iVBORw.......0KGgo"
}

Notice that instead of "python.exe" and "ExportToCSV.py" we are using version with "W" – "pythonw.exe" and "ExportToCSV.pyw". This is because we want to avoid black command prompt window (1) to be opened in background of our window.

Fullpath to our python script doesn't have spaces in it. If we have fullpath with spaces, we have to delimit that fullpath with quotes, but those quotes has to be escaped. In that case "arguments" would loook like:
"\"C:/Users/Sima/Desktop/ExportToCSV.pyw\" \"%server%\" \"%database%\""
We can see that arguments "server" and "database" were already delimited this way because they have "%" sign in them.
This is not needed for "path" element.

This JSON file should be placed on this address. If last two folders (\Power BI Desktop\External Tools) don't exist, create them and place JSON file inside. Name of our JSON file has to end with ".pbitool.json".
C:\Program Files (x86)\Common Files\Microsoft Shared\Power BI Desktop\External Tools\ourtool.pbitool.json

When we click on button in ribbon, python script mentioned in JSON file, will be called. This script has two parts. One part is about using python to create GUI window. The other part is explained bellow. This part is used to connect to Power BI desktop, read data and write them in CSV file. str(sys.argv[1]) is a python way to get first argument sent to script. This is name of server ("localhost:49380"). This address will be used inside connection string. By using connection string we will connect to PBID and we will execute our DAX.

Python will return names of columns as Table[ColumnName]. We will use regex to leave only "ColumnName" part.

Last part is where we read fullpath of desktop to create new file name. After this, all we need is to write data in CSV file by using python CSV modul.

    conn = adodbapi.connect("Provider=MSOLAP;Data Source=" + str(sys.argv[1]) + ";Initial Catalog='';")
    cursor = conn.cursor()
    cursor.execute(DAX)

    #fix names of columns. By default they contain table name and brackets.
    NamesOfColumns = [desc[0] for desc in cursor.description]
    CorrectedNamesOfColumns = [re.findall('\[(.*?)\]', NoC)[0] for NoC in NamesOfColumns]

    #write to CSV
    FileName = os.path.join(os.path.join(os.environ['USERPROFILE']), 'Desktop') + '\CSVfromPBID.csv'
    with open(FileName, 'w', newline='') as f:
        csv.writer(f, delimiter=',').writerow(CorrectedNamesOfColumns)
        csv.writer(f, delimiter=',').writerows(cursor)

Whole python script, JSON file and PBIX sample file are here. Don't forget to change fullpaths inside JSON file.

If we create table in Power BI desktop that has images URL-s in one column (1), this will allow us to present those images in Report View (4). For this to work, we have to mark this column as "Image URL" column (2). First we select our column (1) and then we go to Column tools > Properties > Data category. There, we choose option "Image URL" (2). Now, that our column is of "Image URL" type, we can add this column as a Field to a table visual (3). Final result is table, in Report View (4), that has image in every cell of table presented.

If we don't have URL-s for our images, but the images are available in some folder on our computer or network, we have to raise our local image server. Here, I will explain two ways how to do that easily.

Python http.server modul

We will open windows CMD environment. This CMD environment has to stay open for the time we are using our images in Power BI Desktop. When we close this environment, images will not be available any more. Then, first we have to change current folder to folder with images (1). Second, we have to open python "http.server" module (2). This module asks for port number. We can type anything, but for this example I used port number 21.000.

(1)
>cd "c:\Folder with images" 
(2)
>pyhton.exe -m http.server 21000 

All images in folder are now available as URL links. On our computer we can approach those links with "localhost" and our port number (1). On our computer, and all other computers on local network, we can open page with links by using IP address of our computer and our port number (2). Access from local network can be limited by firewall, so we need to have rights to access computer where python server works.

http://localhost:21000 (1) 
http://192.168.0.103:21000 (2)

Links to images will be generated as "http://localhost:21000/" + "Name of image.jpg". If we have names of all our images in Excel, we just have to prefix them with "http://localhost:21000/" and then we will have the links to all our images. Those links can be imported in Power BI Desktop and used in Report View.

http://192.168.0.103:21000/asters-gda84d7295_640_result.jpg

Mongoose server

There is a very simple server that can be used for the same purpose. Its name is "Mongoose server" and can be downloaded from https://mongoose.ws/desktop-app/. This is small portable executable file. Just click on it to open it.

New icon will appear in system tray. Right click this icon to access main menu. With option (1) we can change folder with images that are publish online. With (2) we can open browser and jump directly on page where all our images are listed. Link for this page and for each of images will be similar to one that python web server created.

http://192.168.0.103:8000
http://192.168.0.103:8000/flowers-g0eaf923f2_640_result.jpg

We can see bellow how page with image links looks like (1). On the right side (2) we can see that Mongoose server can share our images globally. This option has to be enabled by clicking on (3) on the image above. In this case sharing is done through Mongoose server so the link is something like "https://32cf61d37e7299.share.mongoose.ws". Global sharing of our images is not something that we want in this case and could present security risk because now anyone on world can access our images. It is best to turn this option off.

Here is PBIX sample file and one folder with collection of images.