refactor(rust): Remove top-level metadata from ArrowSchema (#18527)

crates/polars-arrow/src/datatypes/schema.rs

@@ -4,9 +4,9 @@ use polars_error::{polars_bail, PolarsResult};
 #[cfg(feature = "serde")]
 use serde::{Deserialize, Serialize};
 
-use super::{Field, Metadata};
+use super::Field;
 
-/// An ordered sequence of [`Field`]s with associated [`Metadata`].
+/// An ordered sequence of [`Field`]s
 ///
 /// [`ArrowSchema`] is an abstraction used to read from, and write to, Arrow IPC format,
 /// Apache Parquet, and Apache Avro. All these formats have a concept of a schema
@@ -16,22 +16,11 @@ use super::{Field, Metadata};
 pub struct ArrowSchema {
     /// The fields composing this schema.
     pub fields: Vec<Field>,
-    /// Optional metadata.
-    pub metadata: Metadata,
 }
 
 pub type ArrowSchemaRef = Arc<ArrowSchema>;
 
 impl ArrowSchema {
-    /// Attaches a [`Metadata`] to [`ArrowSchema`]
-    #[inline]
-    pub fn with_metadata(self, metadata: Metadata) -> Self {
-        Self {
-            fields: self.fields,
-            metadata,
-        }
-    }
-
     #[inline]
     pub fn len(&self) -> usize {
         self.fields.len()
@@ -58,10 +47,7 @@ impl ArrowSchema {
             })
             .collect();
 
-        ArrowSchema {
-            fields,
-            metadata: self.metadata,
-        }
+        ArrowSchema { fields }
     }
 
     pub fn try_project(&self, indices: &[usize]) -> PolarsResult<Self> {
@@ -76,18 +62,12 @@ impl ArrowSchema {
             Ok(out.clone())
         }).collect::<PolarsResult<Vec<_>>>()?;
 
-        Ok(ArrowSchema {
-            fields,
-            metadata: self.metadata.clone(),
-        })
+        Ok(ArrowSchema { fields })
     }
 }
 
 impl From<Vec<Field>> for ArrowSchema {
     fn from(fields: Vec<Field>) -> Self {
-        Self {
-            fields,
-            ..Default::default()
-        }
+        Self { fields }
     }
 }

crates/polars-arrow/src/io/ipc/read/file_async.rs

@@ -39,10 +39,7 @@ impl<'a> FileStream<'a> {
     {
         let (projection, schema) = if let Some(projection) = projection {
             let (p, h, fields) = prepare_projection(&metadata.schema.fields, projection);
-            let schema = ArrowSchema {
-                fields,
-                metadata: metadata.schema.metadata.clone(),
-            };
+            let schema = ArrowSchema { fields };
             (Some((p, h)), Some(schema))
         } else {
             (None, None)

crates/polars-arrow/src/io/ipc/read/reader.rs

@@ -34,10 +34,7 @@ impl<R: Read + Seek> FileReader<R> {
     ) -> Self {
         let projection = projection.map(|projection| {
             let (p, h, fields) = prepare_projection(&metadata.schema.fields, projection);
-            let schema = ArrowSchema {
-                fields,
-                metadata: metadata.schema.metadata.clone(),
-            };
+            let schema = ArrowSchema { fields };
             (p, h, schema)
         });
         Self {

crates/polars-arrow/src/io/ipc/read/schema.rs

@@ -391,22 +391,8 @@ pub(super) fn fb_to_schema(
         arrow_format::ipc::Endianness::Big => false,
     };
 
-    let mut metadata = Metadata::default();
-    if let Some(md_fields) = schema.custom_metadata()? {
-        for kv in md_fields {
-            let kv = kv?;
-            let k_str = kv.key()?;
-            let v_str = kv.value()?;
-            if let Some(k) = k_str {
-                if let Some(v) = v_str {
-                    metadata.insert(PlSmallStr::from_str(k), PlSmallStr::from_str(v));
-                }
-            }
-        }
-    }
-
     Ok((
-        ArrowSchema { fields, metadata },
+        ArrowSchema { fields },
         IpcSchema {
             fields: ipc_fields,
             is_little_endian,

crates/polars-arrow/src/io/ipc/read/stream.rs

@@ -251,10 +251,7 @@ impl<R: Read> StreamReader<R> {
     pub fn new(reader: R, metadata: StreamMetadata, projection: Option<Vec<usize>>) -> Self {
         let projection = projection.map(|projection| {
             let (p, h, fields) = prepare_projection(&metadata.schema.fields, projection);
-            let schema = ArrowSchema {
-                fields,
-                metadata: metadata.schema.metadata.clone(),
-            };
+            let schema = ArrowSchema { fields };
             (p, h, schema)
         });
 

crates/polars-arrow/src/io/ipc/write/schema.rs

@@ -38,13 +38,7 @@ pub fn serialize_schema(
         .map(|(field, ipc_field)| serialize_field(field, ipc_field))
         .collect::<Vec<_>>();
 
-    let custom_metadata = schema
-        .metadata
-        .iter()
-        .map(|(k, v)| key_value(k.clone().into_string(), v.clone().into_string()))
-        .collect::<Vec<_>>();
-
-    let custom_metadata = (!custom_metadata.is_empty()).then_some(custom_metadata);
+    let custom_metadata = None;
 
     arrow_format::ipc::Schema {
         endianness,

crates/polars-parquet/src/arrow/read/schema/mod.rs

@@ -55,6 +55,6 @@ pub fn infer_schema_with_options(
     let schema = read_schema_from_metadata(&mut metadata)?;
     Ok(schema.unwrap_or_else(|| {
         let fields = parquet_to_arrow_schema_with_options(file_metadata.schema().fields(), options);
-        ArrowSchema { fields, metadata }
+        ArrowSchema { fields }
     }))
 }

crates/polars/tests/it/io/parquet/arrow/read.rs

@@ -97,8 +97,6 @@ fn all_types_chunked() -> PolarsResult<()> {
 
 #[test]
 fn read_int96_timestamps() -> PolarsResult<()> {
-    use std::collections::BTreeMap;
-
     let timestamp_data = &[
         0x50, 0x41, 0x52, 0x31, 0x15, 0x04, 0x15, 0x48, 0x15, 0x3c, 0x4c, 0x15, 0x06, 0x15, 0x00,
         0x12, 0x00, 0x00, 0x24, 0x00, 0x00, 0x0d, 0x01, 0x08, 0x9f, 0xd5, 0x1f, 0x0d, 0x0a, 0x44,
@@ -131,7 +129,6 @@ fn read_int96_timestamps() -> PolarsResult<()> {
                 arrow::datatypes::ArrowDataType::Timestamp(time_unit, None),
                 false,
             )],
-            metadata: BTreeMap::new(),
         };
         let reader = FileReader::new(reader, metadata.row_groups, schema, None);
         reader.collect::<PolarsResult<Vec<_>>>()