Struct inconsistencie about pyvespa

[tmaregge]

Here is a way to reproduce your original schema using Pyvespa:

the_struct = Struct(name="answer", fields=[
      Field(name="folder_ids", type="array<string>"),
      Field(name="likes", type="int"),
      Field(name="dislikes", type="int"),
      Field(name="retriever_name", type="string"),
      Field(name="reader_name", type="string"),
      Field(name="passage_id", type="string"),
      Field(name="user_ids_likes", type="array<string>"),
      Field(name="user_ids_dislikes", type="array<string>"),
])
the_field = Field(name="answers", type="array<answer>", indexing=["summary", "index"])

schema = Schema(name="schema", document=Document())
schema.add_fields(the_field)
schema.document.add_structs(the_struct)

print(schema.schema_to_text)

This produces the following

schema schema {
    document schema {
        field answers type array<answer> {
            indexing: summary | index
        }
        struct answer {
            field folder_ids type array<string> {
            }
            field likes type int {
            }
            field dislikes type int {
            }
            field retriever_name type string {
            }
            field reader_name type string {
            }
            field passage_id type string {
            }
            field user_ids_likes type array<string> {
            }
            field user_ids_dislikes type array<string> {
            }
        }
    }
}

StructField actually belongs to a normal Field, not a Struct. Here is an example:

the_struct = Struct(name="answer", fields=[
      Field(name="foo", type="int"),
])
the_field = Field(name="answers",
                type="array<answer>", 
                indexing=["summary", "index"], 
                struct_fields=[StructField(name="foo", indexing=["attribute"], attribute=["fast-search"])]
)

schema = Schema(name="schema", document=Document())
schema.add_fields(the_field)
schema.document.add_structs(the_struct)

print(schema.schema_to_text)

Output:

schema schema {
    document schema {
        field answers type array<answer> {
            indexing: summary | index
            struct-field foo {
                indexing: attribute
                attribute {
                    fast-search
                }
            }
        }
        struct answer {
            field foo type int {
            }
        }
    }
}

Hope this helps. Reach out if anything is unclear.

[tmaregge]

How are you adding the Struct and Field to the Schema? Could you provide a minimal code example to reproduce the error?

[marcoaleixo]

Hey @tmaregge

Already shared above, but basically like this:

Struct(name="answer", fields=[
      StructField(name="folder_ids", type="array<string>", indexing=["summary", "index"]),
      StructField(name="likes", type="int", indexing=["summary", "index"]),
      StructField(name="dislikes", type="int", indexing=["summary", "index"]),
      StructField(name="retriever_name", type="string", indexing=["summary", "index"]),
      StructField(name="reader_name", type="string", indexing=["summary", "index"]),
      StructField(name="passage_id", type="string", indexing=["summary", "index"]),
      StructField(name="user_ids_likes", type="array<string>", indexing=["summary", "index"]),
      StructField(name="user_ids_dislikes", type="array<string>", indexing=["summary", "index"]),
                    ]),
Field(name="answers", type="array<answer>", indexing=["summary", "index"]),

[marcoaleixo]

Thank you @tmaregge. I was able to fully reproduce my schema using pyvespa.
Closing this.