doc: add comments to explain usage of schema_id to store spec_id

Author: sdd

crates/iceberg/src/expr/visitors/manifest_evaluator.rs

@@ -18,7 +18,7 @@
 use crate::expr::visitors::bound_predicate_visitor::{visit, BoundPredicateVisitor};
 use crate::expr::visitors::inclusive_projection::InclusiveProjection;
 use crate::expr::{Bind, BoundPredicate, BoundReference};
-use crate::spec::{Datum, FieldSummary, ManifestFile, PartitionSpecRef, Schema, SchemaRef};
+use crate::spec::{Datum, FieldSummary, ManifestEntry, ManifestFile, PartitionSpecRef, Schema, SchemaRef};
 use crate::{Error, ErrorKind};
 use fnv::FnvHashSet;
 use std::sync::Arc;
@@ -46,6 +46,10 @@ impl ManifestEvaluator {
         let cloned_partition_fields: Vec<_> =
             partition_type.fields().iter().map(Arc::clone).collect();
 
+        // The partition_schema's schema_id is set to the partition
+        // spec's spec_id here, and used to perform a sanity check
+        // during eval to confirm that it matches the spec_id
+        // of the ManifestFile we're evaluating
         let partition_schema = Schema::builder()
             .with_schema_id(partition_spec.spec_id)
             .with_fields(cloned_partition_fields)
@@ -76,6 +80,9 @@ impl ManifestEvaluator {
             return Ok(true);
         }
 
+        // The schema_id of self.partition_schema is set to the
+        // spec_id of the partition spec that this ManifestEvaluator
+        // was created from in ManifestEvaluator::new
         if self.partition_schema.schema_id() != manifest_file.partition_spec_id {
             return Err(Error::new(
                 ErrorKind::Unexpected,
@@ -90,6 +97,20 @@ impl ManifestEvaluator {
 
         visit(&mut evaluator, &self.partition_filter)
     }
+
+    /// Evaluate this `ManifestEvaluator`'s filter predicate against the
+    /// provided [`ManifestEntry`]'s partition. Used by [`TableScan`] to
+    /// see if this `ManifestEntry` could possibly contain data that matches
+    /// the scan's filter.
+    pub(crate) fn eval_data_file(&self, manifest_file: &ManifestFile, manifest_entry: &ManifestEntry) -> crate::Result<bool> {
+        if manifest_entry.data_file().partition.empty() {
+            return Ok(true);
+        }
+
+        let mut evaluator = ManifestFilterVisitor::new(self, &manifest_file.partitions);
+
+        visit(&mut evaluator, &self.partition_filter)
+    }
 }
 
 struct ManifestFilterVisitor<'a> {
@@ -266,7 +287,7 @@ impl ManifestFilterVisitor<'_> {
 
 #[cfg(test)]
 mod test {
-    use crate::expr::visitors::manifest_evaluator::ManifestEvaluator;
+    use crate::expr::visitors::inclusive_metrics_evaluator::ManifestEvaluator;
     use crate::expr::{Bind, Predicate, PredicateOperator, Reference, UnaryExpression};
     use crate::spec::{
         FieldSummary, ManifestContentType, ManifestFile, NestedField, PartitionField,