RocksDB: avoid not needed copies when retrieving string values

lib/src/sparql/eval.rs

@@ -775,7 +775,7 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
 
     fn to_simple_string(&self, term: EncodedTerm) -> Option<String> {
         if let EncodedTerm::SimpleLiteral { value_id } = term {
-            self.store.get_str(value_id).ok()
+            Some(self.store.get_str(value_id).ok()?.into())
         } else {
             None
         }
@@ -793,7 +793,9 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
         match term {
             EncodedTerm::SimpleLiteral { value_id }
             | EncodedTerm::StringLiteral { value_id }
-            | EncodedTerm::LangStringLiteral { value_id, .. } => self.store.get_str(value_id).ok(),
+            | EncodedTerm::LangStringLiteral { value_id, .. } => {
+                Some(self.store.get_str(value_id).ok()?.into())
+            }
             _ => None,
         }
     }

lib/src/store/memory.rs

@@ -68,6 +68,8 @@ struct MemoryGraphIndexes {
 }
 
 impl StringStore for MemoryStore {
+    type StringType = String;
+
     fn insert_str(&self, value: &str) -> Result<u64> {
         self.string_store.insert_str(value)
     }

lib/src/store/numeric_encoder.rs

@@ -11,6 +11,7 @@ use rust_decimal::Decimal;
 use std::collections::BTreeMap;
 use std::io::Read;
 use std::io::Write;
+use std::ops::Deref;
 use std::str;
 use std::str::FromStr;
 use std::sync::RwLock;
@@ -30,8 +31,10 @@ const XSD_DATE_ID: u64 = 9;
 const XSD_TIME_ID: u64 = 10;
 
 pub trait StringStore {
+    type StringType: Deref<Target = str> + ToString + Into<String>;
+
     fn insert_str(&self, value: &str) -> Result<u64>;
-    fn get_str(&self, id: u64) -> Result<String>;
+    fn get_str(&self, id: u64) -> Result<Self::StringType>;
     fn get_url(&self, id: u64) -> Result<Url>;
 
     /// Should be called when the bytes store is created
@@ -58,11 +61,13 @@ pub trait StringStore {
 }
 
 impl<'a, S: StringStore> StringStore for &'a S {
+    type StringType = S::StringType;
+
     fn insert_str(&self, value: &str) -> Result<u64> {
         (*self).insert_str(value)
     }
 
-    fn get_str(&self, id: u64) -> Result<String> {
+    fn get_str(&self, id: u64) -> Result<S::StringType> {
         (*self).get_str(id)
     }
 
@@ -88,6 +93,8 @@ impl Default for MemoryStringStore {
 }
 
 impl StringStore for MemoryStringStore {
+    type StringType = String;
+
     fn insert_str(&self, value: &str) -> Result<u64> {
         let mut id2str = self.id2str.write().map_err(MutexPoisonError::from)?;
         let mut str2id = self.str2id.write().map_err(MutexPoisonError::from)?;
@@ -728,7 +735,7 @@ impl<S: StringStore> Encoder<S> {
             )
             .into()),
             EncodedTerm::StringLiteral { value_id } => {
-                Ok(Literal::from(self.string_store.get_str(value_id)?).into())
+                Ok(Literal::from(self.string_store.get_str(value_id)?.into()).into())
             }
             EncodedTerm::BooleanLiteral(value) => Ok(Literal::from(value).into()),
             EncodedTerm::FloatLiteral(value) => Ok(Literal::from(*value).into()),

lib/src/store/rocksdb.rs

@@ -10,6 +10,7 @@ use rocksdb::ColumnFamily;
 use rocksdb::DBCompactionStyle;
 use rocksdb::DBCompressionType;
 use rocksdb::DBRawIterator;
+use rocksdb::DBVector;
 use rocksdb::Options;
 use rocksdb::WriteBatch;
 use rocksdb::DB;
@@ -90,6 +91,8 @@ impl RocksDbStore {
 }
 
 impl StringStore for RocksDbStore {
+    type StringType = RocksString;
+
     fn insert_str(&self, value: &str) -> Result<u64> {
         let value = value.as_bytes();
         Ok(if let Some(id) = self.db.get_cf(*self.str2id_cf, value)? {
@@ -109,22 +112,17 @@ impl StringStore for RocksDbStore {
         })
     }
 
-    fn get_str(&self, id: u64) -> Result<String> {
+    fn get_str(&self, id: u64) -> Result<RocksString> {
         let value = self.db.get_cf(*self.id2str_cf, &to_bytes(id))?;
         if let Some(value) = value {
-            Ok(str::from_utf8(&value)?.to_owned())
+            Ok(RocksString { vec: value })
         } else {
             Err(format_err!("value not found in the dictionary"))
         }
     }
 
     fn get_url(&self, id: u64) -> Result<Url> {
-        let value = self.db.get_cf(*self.id2str_cf, &to_bytes(id))?;
-        if let Some(value) = value {
-            Ok(Url::from_str(str::from_utf8(&value)?)?)
-        } else {
-            Err(format_err!("value not found in the dictionary"))
-        }
+        Ok(Url::from_str(&self.get_str(id)?)?)
     }
 }
 
@@ -553,6 +551,30 @@ fn to_bytes(int: u64) -> [u8; 8] {
     buf
 }
 
+pub struct RocksString {
+    vec: DBVector,
+}
+
+impl Deref for RocksString {
+    type Target = str;
+
+    fn deref(&self) -> &str {
+        unsafe { str::from_utf8_unchecked(&self.vec) }
+    }
+}
+
+impl ToString for RocksString {
+    fn to_string(&self) -> String {
+        self.deref().to_owned()
+    }
+}
+
+impl From<RocksString> for String {
+    fn from(val: RocksString) -> String {
+        val.deref().to_owned()
+    }
+}
+
 // TODO: very bad but I believe it is fine
 #[derive(Clone, Copy)]
 struct SendColumnFamily(ColumnFamily);