231 pass and store timestamp within record

CHANGELOG.md

@@ -4,10 +4,12 @@ Pearl changelog
 
 ## [Unreleased]
 #### Added
-
+- Add ability to attach metadata to the deletion record (#229)
 
 #### Changed
-
+- Receive timestamp as parameter, store it within record and return it with BlobRecordTimestamp (#231)
+- `Storage::read` now propagates the error further instead of hiding it (#241)
+- `Storage::read` now reads Entries first, select latest one and only then loads the data (#276)
 
 #### Fixed
 - Use `dep:<crate_name>` syntax in feature declaration to avoid unnecessary feature flags (#272) 

src/blob/core.rs

@@ -24,7 +24,7 @@ where
 {
     header: Header,
     index: Index<K>,
-    name: FileName,
+    name: Arc<FileName>,
     file: File,
     created_at: SystemTime,
     validate_data_during_index_regen: bool,
@@ -56,7 +56,7 @@ where
         let mut blob = Self {
             header,
             index,
-            name,
+            name: Arc::new(name),
             file,
             created_at: SystemTime::now(),
             validate_data_during_index_regen,
@@ -171,7 +171,7 @@ where
         let mut blob = Self {
             header,
             file,
-            name,
+            name: Arc::new(name),
             index,
             created_at,
             validate_data_during_index_regen,
@@ -262,28 +262,22 @@ where
         Ok(header)
     }
 
-    pub(crate) async fn read_last(
+    #[allow(dead_code)]
+    pub(crate) async fn read_latest(
         &self,
         key: &K,
         meta: Option<&Meta>,
         check_filters: bool,
     ) -> Result<ReadResult<Bytes>> {
         debug!("blob read any");
-        let entry = self.get_entry(key, meta, check_filters).await?;
+        let entry = self.get_latest_entry(key, meta, check_filters).await?;
         match entry {
             ReadResult::Found(entry) => {
                 debug!("blob read any entry found");
                 let buf = entry
                     .load()
                     .await
-                    .with_context(|| {
-                        format!(
-                            "failed to read key {:?} with meta {:?} from blob {:?}",
-                            key,
-                            meta,
-                            self.name.to_path()
-                        )
-                    })?
+                    .with_context(|| format!("Failed to read data for key {:?} with meta {:?}", key, meta))?
                     .into_data();
                 debug!("blob read any entry loaded bytes: {}", buf.len());
                 Ok(ReadResult::Found(buf))
@@ -300,8 +294,8 @@ where
         debug_assert!(headers
             .iter()
             .zip(headers.iter().skip(1))
-            .all(|(x, y)| x.created() >= y.created()));
-        Ok(Self::headers_to_entries(headers, &self.file))
+            .all(|(x, y)| x.timestamp() >= y.timestamp()));
+        Ok(Self::headers_to_entries(headers, &self.file, &self.name))
     }
 
     #[inline]
@@ -313,41 +307,44 @@ where
         debug_assert!(headers
             .iter()
             .zip(headers.iter().skip(1))
-            .all(|(x, y)| x.created() >= y.created()));
-        Ok(Self::headers_to_entries(headers, &self.file))
+            .all(|(x, y)| x.timestamp() >= y.timestamp()));
+        Ok(Self::headers_to_entries(headers, &self.file, &self.name))
     }
 
-    pub(crate) async fn mark_all_as_deleted(
+    pub(crate) async fn delete(
         &mut self,
         key: &K,
+        timestamp: BlobRecordTimestamp,
+        meta: Option<Meta>,
         only_if_presented: bool,
     ) -> Result<bool> {
-        if !only_if_presented || self.index.get_any(key).await?.is_found() {
-            self.push_deletion_record(key).await?;
+        if !only_if_presented || self.index.get_latest(key).await?.is_found() {
+            let record = Record::deleted(key, timestamp.into(), meta)?;
+            self.push_deletion_record(key, record).await?;
             Ok(true)
         } else {
             Ok(false)
         }
     }
 
-    async fn push_deletion_record(&mut self, key: &K) -> Result<()> {
+    async fn push_deletion_record(&mut self, key: &K, record: Record) -> Result<()> {
         let on_disk = self.index.on_disk();
         if on_disk {
             self.load_index().await?;
         }
-        let record = Record::deleted(key)?;
         let header = self.write_mut(key, record).await?;
         self.index.push_deletion(key, header)
     }
 
-    fn headers_to_entries(headers: Vec<RecordHeader>, file: &File) -> Vec<Entry> {
+    fn headers_to_entries(headers: Vec<RecordHeader>, file: &File, file_name: &Arc<FileName>) -> Vec<Entry> {
         headers
             .into_iter()
-            .map(|header| Entry::new(header, file.clone()))
+            .map(|header| Entry::new(header, file.clone(), file_name.clone()))
             .collect()
     }
 
-    async fn get_entry(
+    /// Returns latest Entry from Blob for specified key and meta
+    pub(crate) async fn get_latest_entry(
         &self,
         key: &K,
         meta: Option<&Meta>,
@@ -364,13 +361,13 @@ where
             debug!("blob get any entry bloom true no meta");
             Ok(self
                 .index
-                .get_any(key)
+                .get_latest(key)
                 .await
                 .with_context(|| {
                     format!("index get any failed for blob: {:?}", self.name.to_path())
                 })?
                 .map(|header| {
-                    let entry = Entry::new(header, self.file.clone());
+                    let entry = Entry::new(header, self.file.clone(), self.name.clone());
                     debug!("blob, get any entry, bloom true no meta, entry found");
                     entry
                 }))
@@ -382,11 +379,11 @@ where
         let deleted_ts = headers
             .last()
             .filter(|h| h.is_deleted())
-            .map(|h| BlobRecordTimestamp::new(h.created()));
+            .map(|h| BlobRecordTimestamp::new(h.timestamp()));
         if deleted_ts.is_some() {
             headers.truncate(headers.len() - 1);
         }
-        let entries = Self::headers_to_entries(headers, &self.file);
+        let entries = Self::headers_to_entries(headers, &self.file, &self.name);
         if let Some(entries) = self.filter_entries(entries, meta).await? {
             Ok(ReadResult::Found(entries))
         } else {
@@ -406,16 +403,17 @@ where
         Ok(None)
     }
 
+    #[allow(dead_code)]
     pub(crate) async fn contains(
         &self,
         key: &K,
         meta: Option<&Meta>,
     ) -> Result<ReadResult<BlobRecordTimestamp>> {
         debug!("blob contains");
         let contains = self
-            .get_entry(key, meta, true)
+            .get_latest_entry(key, meta, true)
             .await?
-            .map(|e| BlobRecordTimestamp::new(e.created()));
+            .map(|e| e.timestamp());
         debug!("blob contains any: {:?}", contains);
         Ok(contains)
     }

src/blob/entry.rs

@@ -16,6 +16,7 @@ pub struct Entry {
     header: RecordHeader,
     meta: Option<Meta>,
     blob_file: File,
+    blob_file_name: Arc<FileName>
 }
 
 impl Entry {
@@ -31,12 +32,15 @@ impl Entry {
             .read_exact_at_allocate(data_size + meta_size, self.header.meta_offset())
             .await
             .map_err(|err| err.into_bincode_if_unexpected_eof())
-            .context("Record load failed")?;
+            .with_context(|| format!("Record load failed from BLOB: {}", self.blob_file_name.to_path().display()))?;
         let mut buf = buf.freeze();
         let data_buf = buf.split_off(meta_size);
-        let meta = Meta::from_raw(&buf).map_err(|err| Error::from(err))?;
-        let record = Record::new(self.header.clone(), meta, data_buf);
-        record.validate()
+        let meta = Meta::from_raw(&buf)
+            .map_err(|err| Error::from(err))
+            .with_context(|| format!("Deserialization failed for Meta loaded from BLOB: {}", self.blob_file_name.to_path().display()))?;
+        Record::new(self.header, meta, data_buf)
+            .validate()
+            .with_context(|| format!("Validation failed for Record loaded from BLOB: {}", self.blob_file_name.to_path().display()))
     }
 
     /// Returns only data.
@@ -48,7 +52,7 @@ impl Entry {
             .read_exact_at_allocate(self.header.data_size().try_into()?, data_offset)
             .await
             .map_err(|err| err.into_bincode_if_unexpected_eof())
-            .context("Error loading Record data")
+            .with_context(|| format!("Error loading Record data from BLOB: {}", self.blob_file_name.to_path().display()))
     }
 
     /// Loads meta data from fisk, and returns reference to it.
@@ -61,8 +65,12 @@ impl Entry {
             .read_exact_at_allocate(self.header.meta_size().try_into()?, meta_offset)
             .await
             .map_err(|err| err.into_bincode_if_unexpected_eof())
-            .with_context(|| format!("failed to read Record metadata, offset: {}", meta_offset))?;
-        self.meta = Some(Meta::from_raw(&buf).map_err(|err| Error::from(err))?);
+            .with_context(|| format!("Failed to read Record metadata from BLOB: {}", self.blob_file_name.to_path().display()))?;
+        let meta = Meta::from_raw(&buf)
+            .map_err(|err| Error::from(err))
+            .with_context(|| format!("Deserialization failed for Meta loaded from BLOB: {}", self.blob_file_name.to_path().display()))?;
+
+        self.meta = Some(meta);
         Ok(self.meta.as_ref())
     }
 
@@ -72,15 +80,16 @@ impl Entry {
     }
 
     /// Timestamp when entry was created
-    pub fn created(&self) -> u64 {
-        self.header.created()
+    pub fn timestamp(&self) -> BlobRecordTimestamp {
+        BlobRecordTimestamp::new(self.header.timestamp())
     }
 
-    pub(crate) fn new(header: RecordHeader, blob_file: File) -> Self {
+    pub(crate) fn new(header: RecordHeader, blob_file: File, blob_file_name: Arc<FileName>) -> Self {
         Self {
             meta: None,
             header,
             blob_file,
+            blob_file_name
         }
     }
 }

src/blob/index/benchmarks.rs

@@ -61,7 +61,7 @@ fn generate_headers(records_amount: usize, key_mapper: fn(u32) -> u32) -> InMemo
         .map(|i| serialize(&i).expect("can't serialize"))
         .for_each(|key| {
             let key: KeyType = key.into();
-            let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+            let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
             if let Some(v) = inmem.get_mut(&key) {
                 v.push(rh);
             } else {
@@ -208,7 +208,7 @@ async fn benchmark_get_any() {
         if (i as u32 + 1) % PRINT_EVERY == 0 {
             println!("Iteration: {}...", i + 1);
         }
-        let _ = findex.get_any(&q.into()).await.unwrap();
+        let _ = findex.get_latest(&q.into()).await.unwrap();
     }
     println!(
         "get_any avg time: {}\n",

src/blob/index/bptree/core.rs

@@ -152,7 +152,7 @@ where
             .map(|headers| (headers, self.header.records_count))
     }
 
-    async fn get_any(&self, key: &K) -> Result<Option<RecordHeader>> {
+    async fn get_latest(&self, key: &K) -> Result<Option<RecordHeader>> {
         let root_offset = self.metadata.tree_offset;
         let buf = BytesMut::zeroed(BLOCK_SIZE);
         let (buf, leaf_offset) = self.find_leaf_node(key, root_offset, buf).await?;

src/blob/index/bptree/tests.rs

@@ -92,7 +92,7 @@ async fn serialize_deserialize_file() {
     let test_dir = get_test_dir("serialize_deserialize_file");
     let mut inmem = InMemoryIndex::<KeyType>::new();
     (0..10000).map(|i| i.into()).for_each(|key: KeyType| {
-        let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+        let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
         inmem.insert(key, vec![rh]);
     });
     let meta = vec![META_VALUE; META_SIZE];
@@ -122,7 +122,7 @@ async fn blob_size_invalidation() {
     let filename = test_dir.join("bptree_index.0.index");
     let mut inmem = InMemoryIndex::<KeyType>::new();
     (0..10000).map(|i| i.into()).for_each(|key: KeyType| {
-        let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+        let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
         inmem.insert(key, vec![rh]);
     });
     let meta = vec![META_VALUE; META_SIZE];
@@ -161,7 +161,7 @@ async fn magic_byte_corruption() {
     let filename = test_dir.join("bptree_index.0.index");
     let mut inmem = InMemoryIndex::<KeyType>::new();
     (0..10000).map(|i| i.into()).for_each(|key: KeyType| {
-        let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+        let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
         inmem.insert(key, vec![rh]);
     });
     let meta = vec![META_VALUE; META_SIZE];
@@ -219,7 +219,7 @@ async fn check_get_any() {
     (RANGE_FROM..RANGE_TO)
         .map(|i| i.into())
         .for_each(|key: KeyType| {
-            let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+            let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
             inmem.insert(key, vec![rh]);
         });
     let meta = vec![META_VALUE; META_SIZE];
@@ -236,7 +236,7 @@ async fn check_get_any() {
     .expect("Can't create file index");
     let presented_keys = RANGE_FROM..RANGE_TO;
     for key in presented_keys.map(|k| k.into()) {
-        if let Ok(inner_res) = findex.get_any(&key).await {
+        if let Ok(inner_res) = findex.get_latest(&key).await {
             if let Some(actual_header) = inner_res {
                 let key_deserialized: usize = key.clone().into();
                 assert_eq!(
@@ -254,7 +254,7 @@ async fn check_get_any() {
     let not_presented_ranges = [0..RANGE_FROM, RANGE_TO..(RANGE_TO + 100)];
     for not_presented_keys in not_presented_ranges.iter() {
         for key in not_presented_keys.clone().map(|k| serialize(&k).unwrap()) {
-            assert_eq!(None, findex.get_any(&key.into()).await.unwrap());
+            assert_eq!(None, findex.get_latest(&key.into()).await.unwrap());
         }
     }
 
@@ -271,8 +271,8 @@ async fn preserves_records_order() {
     (RANGE_FROM..RANGE_TO)
         .map(|i| i.into())
         .for_each(|key: KeyType| {
-            let rh1 = RecordHeader::new(key.to_vec(), 1, 1, 1);
-            let rh2 = RecordHeader::new(key.to_vec(), 2, 2, 2);
+            let rh1 = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
+            let rh2 = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 2, 2, 2);
             inmem.insert(key, vec![rh1, rh2]);
         });
     let meta = vec![META_VALUE; META_SIZE];
@@ -314,7 +314,7 @@ async fn check_get() {
     (RANGE_FROM..RANGE_TO)
         .map(|i| (i % MAX_AMOUNT + 1, i.into()))
         .for_each(|(times, key): (_, KeyType)| {
-            let rh = RecordHeader::new(key.to_vec(), 1, 1, 1);
+            let rh = RecordHeader::new(key.to_vec(), BlobRecordTimestamp::now().into(), 1, 1, 1);
             let recs = (0..times).map(|_| rh.clone()).collect();
             inmem.insert(key, recs);
         });
@@ -332,7 +332,7 @@ async fn check_get() {
     .expect("Can't create file index");
     let presented_keys = RANGE_FROM..RANGE_TO;
     for key in presented_keys.map(|k| k.into()) {
-        if let Ok(inner_res) = findex.get_any(&key).await {
+        if let Ok(inner_res) = findex.get_latest(&key).await {
             if let Some(actual_header) = inner_res {
                 let key_deserialized: usize = key.clone().into();
                 assert_eq!(