rename bloom

analytic_engine/src/sst/meta_data/cache.rs

@@ -35,7 +35,7 @@ impl MetaData {
     /// contains no extended custom information.
     pub fn try_new(
         parquet_meta_data: &parquet_ext::ParquetMetaData,
-        ignore_bloom_filter: bool,
+        ignore_sst_filter: bool,
     ) -> Result<Self> {
         let file_meta_data = parquet_meta_data.file_metadata();
         let kv_metas = file_meta_data
@@ -46,7 +46,7 @@ impl MetaData {
         let custom = {
             let mut sst_meta =
                 encoding::decode_sst_meta_data(&kv_metas[0]).context(DecodeCustomMetaData)?;
-            if ignore_bloom_filter {
+            if ignore_sst_filter {
                 sst_meta.sst_filter = None;
             }
 

analytic_engine/src/sst/parquet/async_reader.rs

@@ -295,8 +295,8 @@ impl<'a> Reader<'a> {
         let meta_data = {
             let parquet_meta_data = self.load_meta_data_from_storage().await?;
 
-            let ignore_bloom_filter = avoid_update_cache && empty_predicate;
-            MetaData::try_new(&parquet_meta_data, ignore_bloom_filter)
+            let ignore_sst_filter = avoid_update_cache && empty_predicate;
+            MetaData::try_new(&parquet_meta_data, ignore_sst_filter)
                 .map_err(|e| Box::new(e) as _)
                 .context(DecodeSstMeta)?
         };

analytic_engine/src/sst/parquet/meta_data.rs

@@ -27,7 +27,7 @@ pub enum Error {
         source,
         backtrace
     ))]
-    InvalidXor8Filter {
+    ParseXor8Filter {
         source: std::io::Error,
         backtrace: Backtrace,
     },
@@ -43,7 +43,7 @@ pub enum Error {
     },
 
     #[snafu(display(
-        "Unsupported bloom filter version, version:{}.\nBacktrace\n:{}",
+        "Unsupported sst_filter version, version:{}.\nBacktrace\n:{}",
         version,
         backtrace
     ))]
@@ -102,7 +102,7 @@ impl Filter for Xor8Filter {
         Self: Sized,
     {
         Xor8::from_bytes(buf)
-            .context(InvalidXor8Filter)
+            .context(ParseXor8Filter)
             .map(|xor8| Self { xor8 })
     }
 }
@@ -342,8 +342,8 @@ impl fmt::Debug for ParquetMetaData {
             .field("time_range", &self.time_range)
             .field("max_sequence", &self.max_sequence)
             .field("schema", &self.schema)
-            // Avoid the messy output from bloom filter.
-            .field("has_bloom_filter", &self.sst_filter.is_some())
+            // Avoid the messy output from filter.
+            .field("has_filter", &self.sst_filter.is_some())
             .field("collapsible_cols_idx", &self.collapsible_cols_idx)
             .finish()
     }
@@ -394,7 +394,7 @@ mod tests {
     use super::*;
 
     #[test]
-    fn test_conversion_sst_bloom_filter() {
+    fn test_conversion_sst_filter() {
         let sst_filter = SstFilter {
             row_group_filters: vec![
                 RowGroupFilter {
@@ -422,7 +422,7 @@ mod tests {
         );
         assert!(sst_filter_pb.row_group_filters[1].column_filters[1].is_empty());
 
-        let decoded_bloom_filter = SstFilter::try_from(sst_filter_pb).unwrap();
-        assert_eq!(decoded_bloom_filter, sst_filter);
+        let decoded_sst_filter = SstFilter::try_from(sst_filter_pb).unwrap();
+        assert_eq!(decoded_sst_filter, sst_filter);
     }
 }

analytic_engine/src/sst/parquet/row_group_filter.rs

@@ -22,7 +22,7 @@ use crate::sst::reader::error::{OtherNoCause, Result};
 /// predicates and filters.
 ///
 /// Currently, two kinds of filters will be applied to such filtering:
-/// min max & bloom filter.
+/// min max & sst_filter.
 pub struct RowGroupPruner<'a> {
     schema: &'a SchemaRef,
     row_groups: &'a [RowGroupMetaData],
@@ -68,7 +68,7 @@ impl<'a> RowGroupPruner<'a> {
                 let pruned = Self::intersect_pruned_row_groups(&pruned0, &pruned1);
 
                 debug!(
-                    "Finish prune row groups by blooms and min_max, total_row_groups:{}, pruned_by_min_max:{}, pruned_by_blooms:{}, pruned_by_both:{}",
+                    "Finish prune row groups by sst_filter and min_max, total_row_groups:{}, pruned_by_min_max:{}, pruned_by_blooms:{}, pruned_by_both:{}",
                     self.row_groups.len(),
                     pruned0.len(),
                     pruned1.len(),
@@ -101,7 +101,7 @@ impl<'a> RowGroupPruner<'a> {
                     .contains_column_data(col_pos.column_idx, &datum.to_bytes())?;
                 if exist {
                     // sst_filter has false positivity, that is to say we are unsure whether this
-                    // value exists even if the bloom filter says it exists.
+                    // value exists even if the sst_filter says it exists.
                     None
                 } else {
                     Some(negated)

analytic_engine/src/sst/parquet/writer.rs

@@ -392,7 +392,7 @@ mod tests {
                     .unwrap()
                     .as_ref()
                     .clone();
-                // bloom filter is built insider sst writer, so overwrite to default for
+                // sst filter is built insider sst writer, so overwrite to default for
                 // comparison.
                 sst_meta_readback.sst_filter = Default::default();
                 assert_eq!(&sst_meta_readback, &ParquetMetaData::from(sst_meta));