how to set the genres filed attribute?

[l1t1]

movieId,title,genres
1,Toy Story (1995),Adventure|Animation|Children|Comedy|Fantasy
2,Jumanji (1995),Adventure|Children|Fantasy
3,Grumpier Old Men (1995),Comedy|Romance

[l1t1]

https://manual.manticoresearch.com/Data_creation_and_modification/Adding_data_from_external_storages/Fetching_from_CSV,TSV#CSV-indexed-columns says

The following directives can be used to declare the types of the indexed columns:

csvpipe_field - declares a text field.
csvpipe_field_string - declares a text field/string attribute. The column will be both indexed as a text field but also stored as a string attribute.
csvpipe_attr_uint - declares an integer attribute
csvpipe_attr_timestamp - declares a timestamp attribute
csvpipe_attr_bool - declares a boolean attribute
csvpipe_attr_float - declares a float attribute
csvpipe_attr_bigint - declares a big integer attribute
csvpipe_attr_multi - declares a multi-value attribute with integers
csvpipe_attr_multi_64 - declares a multi-value attribute with 64bit integers
csvpipe_attr_string - declares a string attribute
csvpipe_attr_json - declares a JSON attribute

csvpipe_attr_multi and csvpipe_attr_multi_64 looks good, but they only accepts integers

[sanikolaev]

csvpipe_attr_string accepts strings. You can use it. If you mean splitting Adventure|Animation|Children|Comedy|Fantasy by | automatically, there's no functionality for this in Manticore. What may help though is to use csvpipe_field or csvpipe_field_string, then the value will be full-text tokenized and since | is by default a separator, each word will be a separate token and you'll be able to find it using full-text search.

[tomatolog]

you could also preprocess your data and replace string categories with the int representations that works as mva attribute

[l1t1]

the data source
ml-latest-small.zip