Pydomo: Error with 'ds_get'

qzqzqzqzqz · 2023-11-08T14:39:05+00:00

There was an error rendering this rich post.

I am trying to read a domo dataset in Python through pydomo. I got an error about encoding. My data contains name/address which may have special characters. I manually downloaded the data to csv and read it in with read_csv and had the same issue initially. However, if I include 'encoding='latin1' in 'read_csv' parameter then it works. I wonder if there is a way to fix this with pydomo.

Please see below for my query.

Domo_Input_ID='xxxxxx'

df=domo.ds_get(Domo_Input_ID)

 'utf-8' codec can't decode byte 0xf0 in position 11128958: invalid continuation byte

Find more posts tagged with

Accepted answers

DavidChurchman

How important is it to keep the special characters in the dataset? If ds_get() can't change its encoding (which it seems like maybe it can't?) you could do some ETL before going to pydomo to strip out all the characters that don't work with UTF-8 encoding.

I think you could regex replace this with blanks to remove all non-unicode characters: "[^\\u0000-\\u007f]+"

GrantSmith

the ds_get function doesn't allow for any encoding other than the default utf8. Two options here:

Override the Domo class and write your own ds_get function to handle the latin1 encoding you have
Bypass using the ds_get and just call the dataset.data_export function to return the data, then call pandas read_csv on your csv string and pass in the encoding='latin1' as a parameter.

Here's a modified ds_get function:

def ds_get(self, dataset_id, encoding='utf-8'):

    """

        Export data to pandas Dataframe


        >>> df = domo.ds_get('80268aef-e6a1-44f6-a84c-f849d9db05fb')

        >>> print(df.head())


        :Parameters:

        - `dataset_id`:     id of a dataset (str)


        :Returns:

        pandas dataframe

    """

    csv_download = self.datasets.data_export(dataset_id, include_csv_header=True)


    content = StringIO(csv_download)

    df = read_csv(content, encoding=encoding)


    # Convert to dates or datetimes if possible

    for col in df.columns:

        if df[col].dtype == 'object':

            try:

                df[col] = to_datetime(df[col])

            except ValueError:

                pass

            except TypeError:

                pass


    return df

All comments

MarkSnodgrass

@ellibot has some experience with Python. He may be able to help you.

ellibot

Not sure on this one actually!

Help me @GrantSmith-wan Kenobi, you're my only hope!

DavidChurchman

I think you could regex replace this with blanks to remove all non-unicode characters: "[^\\u0000-\\u007f]+"

qzqzqzqzqz

Thank you all!

@DavidChurchman could you please confirm if the following syntax looks right?

REGEXP_REPLACE(`address`, "[^\\u0000-\\u007f]+", "")

GrantSmith

the ds_get function doesn't allow for any encoding other than the default utf8. Two options here:

Override the Domo class and write your own ds_get function to handle the latin1 encoding you have
Bypass using the ds_get and just call the dataset.data_export function to return the data, then call pandas read_csv on your csv string and pass in the encoding='latin1' as a parameter.

Here's a modified ds_get function:

def ds_get(self, dataset_id, encoding='utf-8'):

    """

        Export data to pandas Dataframe


        >>> df = domo.ds_get('80268aef-e6a1-44f6-a84c-f849d9db05fb')

        >>> print(df.head())


        :Parameters:

        - `dataset_id`:     id of a dataset (str)


        :Returns:

        pandas dataframe

    """

    csv_download = self.datasets.data_export(dataset_id, include_csv_header=True)


    content = StringIO(csv_download)

    df = read_csv(content, encoding=encoding)


    # Convert to dates or datetimes if possible

    for col in df.columns:

        if df[col].dtype == 'object':

            try:

                df[col] = to_datetime(df[col])

            except ValueError:

                pass

            except TypeError:

                pass


    return df