I have problem with loading data to pinecone with spark connector. I’ ve tried to follow this
page from documentation. But in the article metadata is defined as plain string:
schema = StructType([
StructField("id",StringType(),True),
StructField("values",ArrayType(FloatType()),True),
StructField("namespace",StringType(),True),
StructField("metadata", StringType(), True),
StructField("sparse_values", StructType([
StructField("indices", ArrayType(IntegerType(), False), False),
StructField("values", ArrayType(FloatType(), False), False)
]), True)
])
In my case I want to define metadata as object with 2 fields, as:
StructField(
"metadata",
StructType(
[
StructField("A", ArrayType(StringType(), False), False),
StructField("B", StringType(), False),
]
),
True,
),
But after such a definition I got error when trying load data into Pinecone:
- Cannot write 'metadata': struct<A:array<string>,B:string> is incompatible with string