How to get log files on minor error when using tm1.cells.write with use_ti=True?

[autoneurino]

Hello, simple question, is there a way to get the status & log file when running tm1.cells.write(use_ti=True)? I am asking for the case where you would have minor errors. Below a small sample case:

with TM1Service(**tm1_cloud) as tm1:   
    cells = {("not-existing", "not-existing"): -1}    
    try:
         tm1.cells.write("System Info", cells, use_ti=True)
    except:
         pass # retrieve log files here

[MariusWirtz]

You can catch the TM1pyWritePartialFailureException exception and access the error_log_files property.
Now that you know the names of the error log files, you can use the get_error_log_file_content function to retrieve the content of the error log file.

with TM1Service(**tm1_cloud) as tm1:
    cells = {
        ("not-existing", "not-existing"): -1
    }
    try:
        tm1.cells.write("System Info", cells, use_ti=True)

    except TM1pyWritePartialFailureException as ex:
        files = ex.error_log_files
        for file_name in files:
            error_log_content = tm1.processes.get_error_log_file_content(file_name)
            print(error_log_content)

This will provide you with the information you need:

Error: Prolog procedure line (3): Invalid key: Dimension Name: "System Info Parameter", Element Name (Key): "not-existing"

[MariusWirtz]

Actually, that's not a reliable solution for two reasons.

• the element names are case and space insensitive
• aliases can be used in the write function instead of element names

if you use the get_all_leaf_element_identifiers function instead of get_leaf_element_names it should work!

[Kevin-Dekker]

Thanks Marius, I wasn't aware of that function. What you mention was exactly the problem we faced afterwards. We retrieved the alias based on the element names as a step inbetween. Does get_all_leaf_element_identifiers also solve the case and space sensitivity?

[MariusWirtz]

Yes. The get_all_leaf_element_identifiers returns a CaseAndSpaceInsensitiveSet of all leaf element names and aliases. We built it exactly for the purpose to verify write operations before they happen. Sorry for the delayed reply on this subject.

[MariusWirtz]

on a side note, you might want dim_elements generally to be a set. It could make a large performance impact on a large dimension. The set should offer a near-instant contains check

[Kevin-Dekker]

Thanks! A list of sets (CaseAndSpaceInsensitiveSet) in this case .

For completeness, here the full sample:

with TM1Service(**tm1_cloud) as tm1:
    cellset = {(not-existing", "not-existing"): -1}
    cube_name = 'My Cube'
        
    # retrieves all dimension names in the cube
    dims = tm1.cubes.get(cube_name).dimensions  
        
    # retrieves all dimension elements for each dimension
    dim_elements = [tm1.elements.get_all_leaf_element_identifiers(dim, dim) for dim in dims]

    # creates a cellset made up of key, value pairs of which all keys can be found in their respective dimensions.
    cells_possible_to_write = {k: v for k, v in cellset.items() if all([k[_] in dim_elements[_] for _ in range(len(dims))])} 

    # creates a cellset made up of key, value pairs of which at least one of the keys cannot be found in their respective dimensions.
    cell_not_possible_to_write = {k: v for k, v in cellset.items() if not all([k[_] in dim_elements[_] for _ in range(len(dims))])}  
        
    # write cells to tm1 that have elements existing in their dimensions
    tm1.cells.write(cube_name, cellset_as_dict=cells_possible_to_write)  

    # log those cases where at least one element does not exist in the respective dimension.
    logging.warning(f'Some cells were not written to tm1. \n {cell_not_possible_to_write}')