chore: remove phf from static files (#10259)

Co-authored-by: joshieDo <93316087+joshieDo@users.noreply.github.com> Co-authored-by: Matthias Seitz <matthias.seitz@outlook.de>
paradigmxyz · Aug 30, 2024 · 28e46bf · 28e46bf
1 parent c5a1c0e
commit 28e46bf
Show file tree

Hide file tree

Showing 9 changed files with 40 additions and 580 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/crates/storage/db/src/static_file/cursor.rs b/crates/storage/db/src/static_file/cursor.rs
@@ -1,7 +1,7 @@
 use super::mask::{ColumnSelectorOne, ColumnSelectorThree, ColumnSelectorTwo};
 use derive_more::{Deref, DerefMut};
 use reth_db_api::table::Decompress;
-use reth_nippy_jar::{DataReader, NippyJar, NippyJarCursor};
+use reth_nippy_jar::{DataReader, NippyJar, NippyJarCursor, NippyJarError};
 use reth_primitives::{static_file::SegmentHeader, B256};
 use reth_storage_errors::provider::{ProviderError, ProviderResult};
 use std::sync::Arc;
@@ -39,7 +39,7 @@ impl<'a> StaticFileCursor<'a> {
         }
 
         let row = match key_or_num {
-            KeyOrNumber::Key(k) => self.row_by_key_with_cols(k, mask),
+            KeyOrNumber::Key(_) => Err(NippyJarError::UnsupportedFilterQuery),
             KeyOrNumber::Number(n) => match self.jar().user_header().start() {
                 Some(offset) => {
                     if offset > n {

diff --git a/crates/storage/nippy-jar/Cargo.toml b/crates/storage/nippy-jar/Cargo.toml
@@ -19,7 +19,6 @@ name = "reth_nippy_jar"
 reth-fs-util.workspace = true
 
 # filter
-ph = "0.8.0"
 cuckoofilter = { version = "0.5.0", features = [
     "serde_support",
     "serde_bytes",

diff --git a/crates/storage/nippy-jar/src/cursor.rs b/crates/storage/nippy-jar/src/cursor.rs
@@ -1,10 +1,8 @@
 use crate::{
     compression::{Compression, Compressors, Zstd},
-    DataReader, InclusionFilter, NippyJar, NippyJarError, NippyJarHeader, PerfectHashingFunction,
-    RefRow,
+    DataReader, NippyJar, NippyJarError, NippyJarHeader, RefRow,
 };
 use std::{ops::Range, sync::Arc};
-use sucds::int_vectors::Access;
 use zstd::bulk::Decompressor;
 
 /// Simple cursor implementation to retrieve data from [`NippyJar`].
@@ -67,35 +65,6 @@ impl<'a, H: NippyJarHeader> NippyJarCursor<'a, H> {
         self.row = 0;
     }
 
-    /// Returns a row, searching it by a key.
-    ///
-    /// **May return false positives.**
-    ///
-    /// Example usage would be querying a transactions file with a transaction hash which is **NOT**
-    /// stored in file.
-    pub fn row_by_key(&mut self, key: &[u8]) -> Result<Option<RefRow<'_>>, NippyJarError> {
-        if let (Some(filter), Some(phf)) = (&self.jar.filter, &self.jar.phf) {
-            // TODO: is it worth to parallelize both?
-
-            // May have false positives
-            if filter.contains(key)? {
-                // May have false positives
-                if let Some(row_index) = phf.get_index(key)? {
-                    self.row = self
-                        .jar
-                        .offsets_index
-                        .access(row_index as usize)
-                        .expect("built from same set") as u64;
-                    return self.next_row()
-                }
-            }
-        } else {
-            return Err(NippyJarError::UnsupportedFilterQuery)
-        }
-
-        Ok(None)
-    }
-
     /// Returns a row by its number.
     pub fn row_by_number(&mut self, row: usize) -> Result<Option<RefRow<'_>>, NippyJarError> {
         self.row = row as u64;
@@ -130,40 +99,6 @@ impl<'a, H: NippyJarHeader> NippyJarCursor<'a, H> {
         ))
     }
 
-    /// Returns a row, searching it by a key using a
-    /// `mask` to only read certain columns from the row.
-    ///
-    /// **May return false positives.**
-    ///
-    /// Example usage would be querying a transactions file with a transaction hash which is **NOT**
-    /// stored in file.
-    pub fn row_by_key_with_cols(
-        &mut self,
-        key: &[u8],
-        mask: usize,
-    ) -> Result<Option<RefRow<'_>>, NippyJarError> {
-        if let (Some(filter), Some(phf)) = (&self.jar.filter, &self.jar.phf) {
-            // TODO: is it worth to parallelize both?
-
-            // May have false positives
-            if filter.contains(key)? {
-                // May have false positives
-                if let Some(row_index) = phf.get_index(key)? {
-                    self.row = self
-                        .jar
-                        .offsets_index
-                        .access(row_index as usize)
-                        .expect("built from same set") as u64;
-                    return self.next_row_with_cols(mask)
-                }
-            }
-        } else {
-            return Err(NippyJarError::UnsupportedFilterQuery)
-        }
-
-        Ok(None)
-    }
-
     /// Returns a row by its number by using a `mask` to only read certain columns from the row.
     pub fn row_by_number_with_cols(
         &mut self,

diff --git a/crates/storage/nippy-jar/src/error.rs b/crates/storage/nippy-jar/src/error.rs
@@ -31,10 +31,6 @@ pub enum NippyJarError {
     FilterMaxCapacity,
     #[error("cuckoo was not properly initialized after loaded")]
     FilterCuckooNotLoaded,
-    #[error("perfect hashing function doesn't have any keys added")]
-    PHFMissingKeys,
-    #[error("nippy jar initialized without perfect hashing function")]
-    PHFMissing,
     #[error("nippy jar was built without an index")]
     UnsupportedFilterQuery,
     #[error("the size of an offset must be at most 8 bytes, got {offset_size}")]