test store

Issue/969

Cargo.toml

@@ -47,6 +47,7 @@ murmurhash32 = "0.2"
 chrono = "0.4"
 smallvec = "1"
 rayon = "1"
+env_logger = "0.8"
 lru = "0.6"
 
 [target.'cfg(windows)'.dependencies]

src/core/index.rs

@@ -35,12 +35,18 @@ fn load_metas(
     inventory: &SegmentMetaInventory,
 ) -> crate::Result<IndexMeta> {
     let meta_data = directory.atomic_read(&META_FILEPATH)?;
-    let meta_string = String::from_utf8_lossy(&meta_data);
+    let meta_string = String::from_utf8(meta_data)
+        .map_err(|utf8_err| {
+            DataCorruption::new(
+                META_FILEPATH.to_path_buf(),
+                format!("Meta file is not valid utf-8. {:?}", utf8_err)
+            )
+        })?;
     IndexMeta::deserialize(&meta_string, &inventory)
         .map_err(|e| {
             DataCorruption::new(
                 META_FILEPATH.to_path_buf(),
-                format!("Meta file cannot be deserialized. {:?}.", e),
+                format!("Meta file cannot be deserialized. {:?}. content = {}", e, meta_string),
             )
         })
         .map_err(From::from)

src/functional_test.rs

@@ -1,31 +1,77 @@
-use rand::thread_rng;
-use std::collections::HashSet;
-
-use crate::schema::*;
 use crate::Index;
 use crate::Searcher;
+use crate::{doc, schema::*};
+use rand::thread_rng;
 use rand::Rng;
+use std::collections::HashSet;
 
-fn check_index_content(searcher: &Searcher, vals: &HashSet<u64>) {
+fn check_index_content(searcher: &Searcher, vals: &[u64]) -> crate::Result<()> {
     assert!(searcher.segment_readers().len() < 20);
     assert_eq!(searcher.num_docs() as usize, vals.len());
+    for segment_reader in searcher.segment_readers() {
+        let store_reader = segment_reader.get_store_reader()?;
+        for doc_id in 0..segment_reader.max_doc() {
+            let _doc = store_reader.get(doc_id)?;
+        }
+    }
+    Ok(())
 }
 
 #[test]
 #[ignore]
-fn test_indexing() {
+fn test_functional_store() -> crate::Result<()> {
+    env_logger::init();
+    let mut schema_builder = Schema::builder();
+
+    let id_field = schema_builder.add_u64_field("id", INDEXED | STORED);
+    let schema = schema_builder.build();
+
+    let index = Index::create_in_ram(schema);
+    let reader = index.reader()?;
+
+    let mut rng = thread_rng();
+
+    let mut index_writer = index.writer_with_num_threads(3, 12_000_000)?;
+
+    let mut doc_set: Vec<u64> = Vec::new();
+
+    let mut doc_id = 0u64;
+    for iteration in 0.. {
+        let num_docs: usize = rng.gen_range(0..4);
+        if doc_set.len() >= 1 {
+            let doc_to_remove_id = rng.gen_range(0..doc_set.len());
+            let removed_doc_id = doc_set.swap_remove(doc_to_remove_id);
+            index_writer.delete_term(Term::from_field_u64(id_field, removed_doc_id));
+        }
+        for _ in 0..num_docs {
+            doc_set.push(doc_id);
+            index_writer.add_document(doc!(id_field=>doc_id));
+            doc_id += 1;
+        }
+        index_writer.commit()?;
+        reader.reload()?;
+        let searcher = reader.searcher();
+        println!("#{} - {}", iteration, searcher.segment_readers().len());
+        check_index_content(&searcher, &doc_set)?;
+    }
+    Ok(())
+}
+
+#[test]
+#[ignore]
+fn test_functional_indexing() -> crate::Result<()> {
     let mut schema_builder = Schema::builder();
 
     let id_field = schema_builder.add_u64_field("id", INDEXED);
     let multiples_field = schema_builder.add_u64_field("multiples", INDEXED);
     let schema = schema_builder.build();
 
-    let index = Index::create_from_tempdir(schema).unwrap();
-    let reader = index.reader().unwrap();
+    let index = Index::create_from_tempdir(schema)?;
+    let reader = index.reader()?;
 
     let mut rng = thread_rng();
 
-    let mut index_writer = index.writer_with_num_threads(3, 120_000_000).unwrap();
+    let mut index_writer = index.writer_with_num_threads(3, 120_000_000)?;
 
     let mut committed_docs: HashSet<u64> = HashSet::new();
     let mut uncommitted_docs: HashSet<u64> = HashSet::new();
@@ -33,13 +79,16 @@ fn test_indexing() {
     for _ in 0..200 {
         let random_val = rng.gen_range(0..20);
         if random_val == 0 {
-            index_writer.commit().expect("Commit failed");
+            index_writer.commit()?;
             committed_docs.extend(&uncommitted_docs);
             uncommitted_docs.clear();
-            reader.reload().unwrap();
+            reader.reload()?;
             let searcher = reader.searcher();
             // check that everything is correct.
-            check_index_content(&searcher, &committed_docs);
+            check_index_content(
+                &searcher,
+                &committed_docs.iter().cloned().collect::<Vec<u64>>(),
+            )?;
         } else {
             if committed_docs.remove(&random_val) || uncommitted_docs.remove(&random_val) {
                 let doc_id_term = Term::from_field_u64(id_field, random_val);
@@ -55,4 +104,5 @@ fn test_indexing() {
             }
         }
     }
+    Ok(())
 }

src/store/index/block.rs

@@ -43,6 +43,9 @@ impl CheckpointBlock {
 
     /// Adding another checkpoint in the block.
     pub fn push(&mut self, checkpoint: Checkpoint) {
+        if let Some(prev_checkpoint) = self.checkpoints.last() {
+            assert!(checkpoint.follows(prev_checkpoint));
+        }
         self.checkpoints.push(checkpoint);
     }
 

src/store/index/mod.rs

@@ -1,4 +1,4 @@
-const CHECKPOINT_PERIOD: usize = 8;
+const CHECKPOINT_PERIOD: usize = 2;
 
 use std::fmt;
 mod block;
@@ -26,6 +26,13 @@ pub struct Checkpoint {
     pub end_offset: u64,
 }
 
+impl Checkpoint {
+    pub(crate) fn follows(&self, other: &Checkpoint) -> bool {
+        (self.start_doc == other.end_doc) &&
+        (self.start_offset == other.end_offset)
+    }
+}
+
 impl fmt::Debug for Checkpoint {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         write!(
@@ -39,13 +46,16 @@ impl fmt::Debug for Checkpoint {
 #[cfg(test)]
 mod tests {
 
-    use std::io;
+    use std::{io, iter};
 
+    use futures::executor::block_on;
     use proptest::strategy::{BoxedStrategy, Strategy};
 
     use crate::directory::OwnedBytes;
+    use crate::indexer::NoMergePolicy;
+    use crate::schema::{SchemaBuilder, STORED, STRING};
     use crate::store::index::Checkpoint;
-    use crate::DocId;
+    use crate::{DocAddress, DocId, Index, Term};
 
     use super::{SkipIndex, SkipIndexBuilder};
 
@@ -54,7 +64,7 @@ mod tests {
         let mut output: Vec<u8> = Vec::new();
         let skip_index_builder: SkipIndexBuilder = SkipIndexBuilder::new();
         skip_index_builder.write(&mut output)?;
-        let skip_index: SkipIndex = SkipIndex::from(OwnedBytes::new(output));
+        let skip_index: SkipIndex = SkipIndex::open(OwnedBytes::new(output));
         let mut skip_cursor = skip_index.checkpoints();
         assert!(skip_cursor.next().is_none());
         Ok(())
@@ -72,7 +82,7 @@ mod tests {
         };
         skip_index_builder.insert(checkpoint);
         skip_index_builder.write(&mut output)?;
-        let skip_index: SkipIndex = SkipIndex::from(OwnedBytes::new(output));
+        let skip_index: SkipIndex = SkipIndex::open(OwnedBytes::new(output));
         let mut skip_cursor = skip_index.checkpoints();
         assert_eq!(skip_cursor.next(), Some(checkpoint));
         assert_eq!(skip_cursor.next(), None);
@@ -121,7 +131,7 @@ mod tests {
         }
         skip_index_builder.write(&mut output)?;
 
-        let skip_index: SkipIndex = SkipIndex::from(OwnedBytes::new(output));
+        let skip_index: SkipIndex = SkipIndex::open(OwnedBytes::new(output));
         assert_eq!(
             &skip_index.checkpoints().collect::<Vec<_>>()[..],
             &checkpoints[..]
@@ -133,6 +143,40 @@ mod tests {
         (doc as u64) * (doc as u64)
     }
 
+    #[test]
+    fn test_merge_store_with_stacking_reproducing_issue969() -> crate::Result<()> {
+        let mut schema_builder = SchemaBuilder::default();
+        let text = schema_builder.add_text_field("text", STORED | STRING);
+        let body = schema_builder.add_text_field("body", STORED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.set_merge_policy(Box::new(NoMergePolicy));
+        let long_text: String = iter::repeat("abcdefghijklmnopqrstuvwxyz")
+            .take(1_000)
+            .collect();
+        for _ in 0..20 {
+            index_writer.add_document(doc!(body=>long_text.clone()));
+        }
+        index_writer.commit()?;
+        index_writer.add_document(doc!(text=>"testb"));
+        for _ in 0..10 {
+            index_writer.add_document(doc!(text=>"testd", body=>long_text.clone()));
+        }
+        index_writer.commit()?;
+        index_writer.delete_term(Term::from_field_text(text, "testb"));
+        index_writer.commit()?;
+        let segment_ids = index.searchable_segment_ids()?;
+        block_on(index_writer.merge(&segment_ids))?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        assert_eq!(searcher.num_docs(), 30);
+        for i in 0..searcher.num_docs() as u32 {
+            let _doc = searcher.doc(DocAddress(0u32, i))?;
+        }
+        Ok(())
+    }
+
     #[test]
     fn test_skip_index_long() -> io::Result<()> {
         let mut output: Vec<u8> = Vec::new();
@@ -150,7 +194,7 @@ mod tests {
         }
         skip_index_builder.write(&mut output)?;
         assert_eq!(output.len(), 4035);
-        let resulting_checkpoints: Vec<Checkpoint> = SkipIndex::from(OwnedBytes::new(output))
+        let resulting_checkpoints: Vec<Checkpoint> = SkipIndex::open(OwnedBytes::new(output))
             .checkpoints()
             .collect();
         assert_eq!(&resulting_checkpoints, &checkpoints);
@@ -221,7 +265,7 @@ mod tests {
              }
              let mut buffer = Vec::new();
              skip_index_builder.write(&mut buffer).unwrap();
-             let skip_index = SkipIndex::from(OwnedBytes::new(buffer));
+             let skip_index = SkipIndex::open(OwnedBytes::new(buffer));
              let iter_checkpoints: Vec<Checkpoint> = skip_index.checkpoints().collect();
              assert_eq!(&checkpoints[..], &iter_checkpoints[..]);
              test_skip_index_aux(skip_index, &checkpoints[..]);

src/store/index/skip_index.rs

@@ -59,6 +59,24 @@ pub struct SkipIndex {
 }
 
 impl SkipIndex {
+    pub fn open(mut data: OwnedBytes) -> SkipIndex {
+        let offsets: Vec<u64> = Vec::<VInt>::deserialize(&mut data)
+            .unwrap()
+            .into_iter()
+            .map(|el| el.0)
+            .collect();
+        let mut start_offset = 0;
+        let mut layers = Vec::new();
+        for end_offset in offsets {
+            let layer = Layer {
+                data: data.slice(start_offset as usize, end_offset as usize),
+            };
+            layers.push(layer);
+            start_offset = end_offset;
+        }
+        SkipIndex { layers }
+    }
+
     pub(crate) fn checkpoints<'a>(&'a self) -> impl Iterator<Item = Checkpoint> + 'a {
         self.layers
             .last()
@@ -90,22 +108,3 @@ impl SkipIndex {
         Some(cur_checkpoint)
     }
 }
-
-impl From<OwnedBytes> for SkipIndex {
-    fn from(mut data: OwnedBytes) -> SkipIndex {
-        let offsets: Vec<u64> = Vec::<VInt>::deserialize(&mut data)
-            .unwrap()
-            .into_iter()
-            .map(|el| el.0)
-            .collect();
-        let mut start_offset = 0;
-        let mut layers = Vec::new();
-        for end_offset in offsets {
-            layers.push(Layer {
-                data: data.slice(start_offset as usize, end_offset as usize),
-            });
-            start_offset = end_offset;
-        }
-        SkipIndex { layers }
-    }
-}

src/store/index/skip_index_builder.rs

@@ -28,18 +28,20 @@ impl LayerBuilder {
     ///
     /// If the block was empty to begin with, simply return None.
     fn flush_block(&mut self) -> Option<Checkpoint> {
-        self.block.doc_interval().map(|(start_doc, end_doc)| {
+        if let Some((start_doc, end_doc)) = self.block.doc_interval() {
             let start_offset = self.buffer.len() as u64;
             self.block.serialize(&mut self.buffer);
             let end_offset = self.buffer.len() as u64;
             self.block.clear();
-            Checkpoint {
+            Some(Checkpoint {
                 start_doc,
                 end_doc,
                 start_offset,
                 end_offset,
-            }
-        })
+            })
+        } else {
+            None
+        }
     }
 
     fn push(&mut self, checkpoint: Checkpoint) {
@@ -48,7 +50,7 @@ impl LayerBuilder {
 
     fn insert(&mut self, checkpoint: Checkpoint) -> Option<Checkpoint> {
         self.push(checkpoint);
-        let emit_skip_info = (self.block.len() % CHECKPOINT_PERIOD) == 0;
+        let emit_skip_info = self.block.len() >= CHECKPOINT_PERIOD;
         if emit_skip_info {
             self.flush_block()
         } else {

src/store/reader.rs

@@ -35,7 +35,7 @@ impl StoreReader {
         let (data_file, offset_index_file) = split_file(store_file)?;
         let index_data = offset_index_file.read_bytes()?;
         let space_usage = StoreSpaceUsage::new(data_file.len(), offset_index_file.len());
-        let skip_index = SkipIndex::from(index_data);
+        let skip_index = SkipIndex::open(index_data);
         Ok(StoreReader {
             data: data_file,
             cache: Arc::new(Mutex::new(LruCache::new(LRU_CACHE_CAPACITY))),

src/store/tests_store.rs

@@ -0,0 +1,50 @@
+use std::path::Path;
+
+use crate::HasLen;
+use crate::directory::{Directory, ManagedDirectory, MmapDirectory, RAMDirectory};
+use crate::fastfield::DeleteBitSet;
+
+use super::{StoreReader, StoreWriter};
+
+#[test]
+fn test_toto2() -> crate::Result<()> {
+    let directory = ManagedDirectory::wrap(MmapDirectory::open("src/store/broken_seg")?)?;
+    let path = Path::new("b6029ade1b954ea1acad15b432eaacb9.store");
+    assert!(directory.validate_checksum(path)?);
+    let store_file = directory.open_read(path)?;
+    let store = StoreReader::open(store_file)?;
+    let documents = store.documents();
+    // for doc in documents {
+    //     println!("{:?}", doc);
+    // }
+    let doc= store.get(15_086)?;
+    Ok(())
+}
+
+#[test]
+fn test_toto() -> crate::Result<()> {
+    let directory = ManagedDirectory::wrap(MmapDirectory::open("src/store/broken_seg")?)?;
+    assert!(directory.validate_checksum(Path::new("e6ece22e5bca4e0dbe7ce3e4dcbd5bbf.store"))?);
+    let store_file = directory.open_read(Path::new("e6ece22e5bca4e0dbe7ce3e4dcbd5bbf.store.patched"))?;
+    let store = StoreReader::open(store_file)?;
+    let doc= store.get(53)?;
+    println!("{:?}", doc);
+    // let documents = store.documents();
+    // let ram_directory = RAMDirectory::create();
+    // let path = Path::new("store");
+
+    // let store_wrt = ram_directory.open_write(path)?;
+    // let mut store_writer = StoreWriter::new(store_wrt);
+    // for doc in &documents {
+    //     store_writer.store(doc)?;
+    // }
+    // store_writer.close()?;
+    // let store_data = ram_directory.open_read(path)?;
+    // let new_store = StoreReader::open(store_data)?;
+    // for doc in 0..59 {
+    //     println!("{}", doc);
+    //     let doc = new_store.get(doc)?;
+    //     println!("{:?}", doc);
+    // }
+    Ok(())
+}

src/store/writer.rs

@@ -10,7 +10,7 @@ use crate::store::index::Checkpoint;
 use crate::DocId;
 use std::io::{self, Write};
 
-const BLOCK_SIZE: usize = 16_384;
+const BLOCK_SIZE: usize = 30;
 
 /// Write tantivy's [`Store`](./index.html)
 ///
@@ -72,6 +72,7 @@ impl StoreWriter {
         if !self.current_block.is_empty() {
             self.write_and_compress_block()?;
         }
+        assert_eq!(self.first_doc_in_block, self.doc);
         let doc_shift = self.doc;
         let start_shift = self.writer.written_bytes() as u64;
 
@@ -86,12 +87,17 @@ impl StoreWriter {
             checkpoint.end_doc += doc_shift;
             checkpoint.start_offset += start_shift;
             checkpoint.end_offset += start_shift;
-            self.offset_index_writer.insert(checkpoint);
-            self.doc = checkpoint.end_doc;
+            self.register_checkpoint(checkpoint);
         }
         Ok(())
     }
 
+    fn register_checkpoint(&mut self, checkpoint: Checkpoint) {
+        self.offset_index_writer.insert(checkpoint);
+        self.first_doc_in_block = checkpoint.end_doc;
+        self.doc = checkpoint.end_doc;
+    }
+
     fn write_and_compress_block(&mut self) -> io::Result<()> {
         assert!(self.doc > 0);
         self.intermediary_buffer.clear();
@@ -100,14 +106,13 @@ impl StoreWriter {
         self.writer.write_all(&self.intermediary_buffer)?;
         let end_offset = self.writer.written_bytes();
         let end_doc = self.doc;
-        self.offset_index_writer.insert(Checkpoint {
+        self.register_checkpoint(Checkpoint {
             start_doc: self.first_doc_in_block,
             end_doc,
             start_offset,
             end_offset,
         });
         self.current_block.clear();
-        self.first_doc_in_block = self.doc;
         Ok(())
     }