0.8.0

* Removing some unsafe

* Removing some unsafe (2)

* Remove murmurhash

CHANGELOG.md

@@ -1,4 +1,4 @@
-Tantivy 0.8.1
+Tantivy 0.8.0
 =====================
 *No change in the index format*
 - API Breaking change in the collector API. (@jwolfe, @fulmicoton)

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.8.0-dev"
+version = "0.8.0"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -29,7 +29,7 @@ serde = "1.0"
 serde_derive = "1.0"
 serde_json = "1.0"
 num_cpus = "1.2"
-itertools = "0.7"
+itertools = "0.8"
 levenshtein_automata = {version="0.1", features=["fst_automaton"]}
 bit-set = "0.5"
 uuid = { version = "0.7", features = ["v4", "serde"] }
@@ -49,6 +49,7 @@ failure = "0.1"
 htmlescape = "0.3.1"
 fail = "0.2"
 scoped-pool = "1.0"
+murmurhash32 = "0.2"
 
 [target.'cfg(windows)'.dependencies]
 winapi = "0.2"

src/collector/facet_collector.rs

@@ -197,7 +197,7 @@ fn skip<'a, I: Iterator<Item = &'a Facet>>(
 ) -> SkipResult {
     loop {
         match collapse_it.peek() {
-            Some(facet_bytes) => match facet_bytes.encoded_bytes().cmp(target) {
+            Some(facet_bytes) => match facet_bytes.encoded_str().as_bytes().cmp(target) {
                 Ordering::Less => {}
                 Ordering::Greater => {
                     return SkipResult::OverStep;
@@ -369,7 +369,8 @@ impl SegmentCollector for FacetSegmentCollector {
             let mut facet = vec![];
             let facet_ord = self.collapse_facet_ords[collapsed_facet_ord];
             facet_dict.ord_to_term(facet_ord as u64, &mut facet);
-            facet_counts.insert(unsafe { Facet::from_encoded(facet) }, count);
+            // TODO
+            facet_counts.insert(Facet::from_encoded(facet).unwrap(), count);
         }
         FacetCounts { facet_counts }
     }
@@ -403,9 +404,9 @@ impl FacetCounts {
         let right_bound = if facet.is_root() {
             Bound::Unbounded
         } else {
-            let mut facet_after_bytes: Vec<u8> = facet.encoded_bytes().to_owned();
-            facet_after_bytes.push(1u8);
-            let facet_after = unsafe { Facet::from_encoded(facet_after_bytes) }; // ok logic
+            let mut facet_after_bytes: String = facet.encoded_str().to_owned();
+            facet_after_bytes.push('\u{1}');
+            let facet_after = Facet::from_encoded_string(facet_after_bytes);
             Bound::Excluded(facet_after)
         };
         let underlying: btree_map::Range<_, _> = self.facet_counts.range((left_bound, right_bound));

src/common/bitpacker.rs

@@ -1,9 +1,6 @@
-use common::serialize::BinarySerializable;
+use byteorder::{ByteOrder, LittleEndian, WriteBytesExt};
 use std::io;
-use std::io::Write;
-use std::mem;
 use std::ops::Deref;
-use std::ptr;
 
 pub(crate) struct BitPacker {
     mini_buffer: u64,
@@ -18,7 +15,7 @@ impl BitPacker {
         }
     }
 
-    pub fn write<TWrite: Write>(
+    pub fn write<TWrite: io::Write>(
         &mut self,
         val: u64,
         num_bits: u8,
@@ -28,14 +25,14 @@ impl BitPacker {
         let num_bits = num_bits as usize;
         if self.mini_buffer_written + num_bits > 64 {
             self.mini_buffer |= val_u64.wrapping_shl(self.mini_buffer_written as u32);
-            self.mini_buffer.serialize(output)?;
+            output.write_u64::<LittleEndian>(self.mini_buffer)?;
             self.mini_buffer = val_u64.wrapping_shr((64 - self.mini_buffer_written) as u32);
             self.mini_buffer_written = self.mini_buffer_written + num_bits - 64;
         } else {
             self.mini_buffer |= val_u64 << self.mini_buffer_written;
             self.mini_buffer_written += num_bits;
             if self.mini_buffer_written == 64 {
-                self.mini_buffer.serialize(output)?;
+                output.write_u64::<LittleEndian>(self.mini_buffer)?;
                 self.mini_buffer_written = 0;
                 self.mini_buffer = 0u64;
             }
@@ -43,17 +40,18 @@ impl BitPacker {
         Ok(())
     }
 
-    pub fn flush<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
+    pub fn flush<TWrite: io::Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
         if self.mini_buffer_written > 0 {
             let num_bytes = (self.mini_buffer_written + 7) / 8;
-            let arr: [u8; 8] = unsafe { mem::transmute::<u64, [u8; 8]>(self.mini_buffer.to_le()) };
+            let mut arr: [u8; 8] = [0u8; 8];
+            LittleEndian::write_u64(&mut arr, self.mini_buffer);
             output.write_all(&arr[..num_bytes])?;
             self.mini_buffer_written = 0;
         }
         Ok(())
     }
 
-    pub fn close<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
+    pub fn close<TWrite: io::Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
         self.flush(output)?;
         // Padding the write file to simplify reads.
         output.write_all(&[0u8; 7])?;
@@ -102,9 +100,7 @@ where
             addr + 8 <= data.len(),
             "The fast field field should have been padded with 7 bytes."
         );
-        #[cfg_attr(feature = "cargo-clippy", allow(clippy::cast_ptr_alignment))]
-        let val_unshifted_unmasked: u64 =
-            u64::from_le(unsafe { ptr::read_unaligned(data[addr..].as_ptr() as *const u64) });
+        let val_unshifted_unmasked: u64 = LittleEndian::read_u64(&data[addr..]);
         let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
         val_shifted & mask
     }
@@ -126,9 +122,7 @@ where
             for output_val in output.iter_mut() {
                 let addr = addr_in_bits >> 3;
                 let bit_shift = addr_in_bits & 7;
-                #[cfg_attr(feature = "cargo-clippy", allow(clippy::cast_ptr_alignment))]
-                let val_unshifted_unmasked: u64 =
-                    unsafe { ptr::read_unaligned(data[addr..].as_ptr() as *const u64) };
+                let val_unshifted_unmasked: u64 = LittleEndian::read_u64(&data[addr..]);
                 let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
                 *output_val = val_shifted & mask;
                 addr_in_bits += num_bits;

src/core/executor.rs

@@ -64,17 +64,18 @@ impl Executor {
                     // This is important as it makes it possible for the fruit_receiver iteration to
                     // terminate.
                 };
-                let mut results = Vec::with_capacity(num_fruits);
-                unsafe { results.set_len(num_fruits) };
-                let mut num_items = 0;
+                // This is lame, but it does not use unsafe code.
+                let mut results_with_position = Vec::with_capacity(num_fruits);
                 for (pos, fruit_res) in fruit_receiver {
-                    results[pos] = fruit_res?;
-                    num_items += 1;
+                    let fruit = fruit_res?;
+                    results_with_position.push((pos, fruit));
                 }
-                // this checks ensures that we filled of this
-                // uninitialized memory.
-                assert_eq!(num_items, results.len());
-                Ok(results)
+                results_with_position.sort_by_key(|(pos, _)| *pos);
+                assert_eq!(results_with_position.len(), num_fruits);
+                Ok(results_with_position
+                    .into_iter()
+                    .map(|(_, fruit)| fruit)
+                    .collect::<Vec<_>>())
             }
         }
     }

src/core/index.rs

@@ -13,6 +13,7 @@ use directory::ManagedDirectory;
 #[cfg(feature = "mmap")]
 use directory::MmapDirectory;
 use directory::{Directory, RAMDirectory};
+use error::DataCorruption;
 use error::TantivyError;
 use indexer::index_writer::open_index_writer;
 use indexer::index_writer::HEAP_SIZE_MIN;
@@ -37,7 +38,13 @@ fn load_metas(directory: &Directory) -> Result<IndexMeta> {
     let meta_data = directory.atomic_read(&META_FILEPATH)?;
     let meta_string = String::from_utf8_lossy(&meta_data);
     serde_json::from_str(&meta_string)
-        .map_err(|_| TantivyError::CorruptedFile(META_FILEPATH.clone()))
+        .map_err(|e| {
+            DataCorruption::new(
+                META_FILEPATH.clone(),
+                format!("Meta file cannot be deserialized. {:?}.", e),
+            )
+        })
+        .map_err(From::from)
 }
 
 /// Search Index

src/directory/managed_directory.rs

@@ -1,7 +1,7 @@
 use core::MANAGED_FILEPATH;
 use directory::error::{DeleteError, IOError, OpenReadError, OpenWriteError};
 use directory::{ReadOnlySource, WritePtr};
-use error::TantivyError;
+use error::DataCorruption;
 use indexer::LockType;
 use serde_json;
 use std::collections::HashSet;
@@ -64,7 +64,12 @@ impl ManagedDirectory {
             Ok(data) => {
                 let managed_files_json = String::from_utf8_lossy(&data);
                 let managed_files: HashSet<PathBuf> = serde_json::from_str(&managed_files_json)
-                    .map_err(|_| TantivyError::CorruptedFile(MANAGED_FILEPATH.clone()))?;
+                    .map_err(|e| {
+                        DataCorruption::new(
+                            MANAGED_FILEPATH.clone(),
+                            format!("Managed file cannot be deserialized: {:?}. ", e),
+                        )
+                    })?;
                 Ok(ManagedDirectory {
                     directory: Box::new(directory),
                     meta_informations: Arc::new(RwLock::new(MetaInformation {

src/error.rs

@@ -8,9 +8,42 @@ use indexer::LockType;
 use query;
 use schema;
 use serde_json;
+use std::fmt;
 use std::path::PathBuf;
 use std::sync::PoisonError;
 
+pub struct DataCorruption {
+    filepath: Option<PathBuf>,
+    comment: String,
+}
+
+impl DataCorruption {
+    pub fn new(filepath: PathBuf, comment: String) -> DataCorruption {
+        DataCorruption {
+            filepath: Some(filepath),
+            comment,
+        }
+    }
+
+    pub fn comment_only(comment: String) -> DataCorruption {
+        DataCorruption {
+            filepath: None,
+            comment,
+        }
+    }
+}
+
+impl fmt::Debug for DataCorruption {
+    fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
+        write!(f, "Data corruption: ")?;
+        if let Some(ref filepath) = &self.filepath {
+            write!(f, "(in file `{:?}`)", filepath)?;
+        }
+        write!(f, ": {}.", self.comment)?;
+        Ok(())
+    }
+}
+
 /// The library's failure based error enum
 #[derive(Debug, Fail)]
 pub enum TantivyError {
@@ -33,8 +66,8 @@ pub enum TantivyError {
     #[fail(display = "An IO error occurred: '{}'", _0)]
     IOError(#[cause] IOError),
     /// Data corruption.
-    #[fail(display = "File contains corrupted data: '{:?}'", _0)]
-    CorruptedFile(PathBuf),
+    #[fail(display = "{:?}", _0)]
+    DataCorruption(DataCorruption),
     /// A thread holding the locked panicked and poisoned the lock.
     #[fail(display = "A thread holding the locked panicked and poisoned the lock")]
     Poisoned,
@@ -55,6 +88,12 @@ pub enum TantivyError {
     SystemError(String),
 }
 
+impl From<DataCorruption> for TantivyError {
+    fn from(data_corruption: DataCorruption) -> TantivyError {
+        TantivyError::DataCorruption(data_corruption)
+    }
+}
+
 impl From<FastFieldNotAvailableError> for TantivyError {
     fn from(fastfield_error: FastFieldNotAvailableError) -> TantivyError {
         TantivyError::FastFieldError(fastfield_error)

src/fastfield/facet_reader.rs

@@ -1,5 +1,6 @@
 use super::MultiValueIntFastFieldReader;
 use schema::Facet;
+use std::str;
 use termdict::TermDictionary;
 use termdict::TermOrdinal;
 use DocId;
@@ -20,6 +21,7 @@ use DocId;
 pub struct FacetReader {
     term_ords: MultiValueIntFastFieldReader<u64>,
     term_dict: TermDictionary,
+    buffer: Vec<u8>,
 }
 
 impl FacetReader {
@@ -37,6 +39,7 @@ impl FacetReader {
         FacetReader {
             term_ords,
             term_dict,
+            buffer: vec![],
         }
     }
 
@@ -55,11 +58,18 @@ impl FacetReader {
     }
 
     /// Given a term ordinal returns the term associated to it.
-    pub fn facet_from_ord(&self, facet_ord: TermOrdinal, output: &mut Facet) {
+    pub fn facet_from_ord(
+        &mut self,
+        facet_ord: TermOrdinal,
+        output: &mut Facet,
+    ) -> Result<(), str::Utf8Error> {
         let found_term = self
             .term_dict
-            .ord_to_term(facet_ord as u64, output.inner_buffer_mut());
+            .ord_to_term(facet_ord as u64, &mut self.buffer);
         assert!(found_term, "Term ordinal {} no found.", facet_ord);
+        let facet_str = str::from_utf8(&self.buffer[..])?;
+        output.set_facet_str(facet_str);
+        Ok(())
     }
 
     /// Return the list of facet ordinals associated to a document.

src/fastfield/multivalued/reader.rs

@@ -82,20 +82,20 @@ mod tests {
 
         let mut facet = Facet::root();
         {
-            facet_reader.facet_from_ord(1, &mut facet);
+            facet_reader.facet_from_ord(1, &mut facet).unwrap();
             assert_eq!(facet, Facet::from("/category"));
         }
         {
-            facet_reader.facet_from_ord(2, &mut facet);
+            facet_reader.facet_from_ord(2, &mut facet).unwrap();
             assert_eq!(facet, Facet::from("/category/cat1"));
         }
         {
-            facet_reader.facet_from_ord(3, &mut facet);
+            facet_reader.facet_from_ord(3, &mut facet).unwrap();
             assert_eq!(format!("{}", facet), "/category/cat2");
             assert_eq!(facet, Facet::from("/category/cat2"));
         }
         {
-            facet_reader.facet_from_ord(4, &mut facet);
+            facet_reader.facet_from_ord(4, &mut facet).unwrap();
             assert_eq!(facet, Facet::from("/category/cat3"));
         }
 

src/indexer/segment_writer.rs

@@ -111,19 +111,18 @@ impl SegmentWriter {
             }
             match *field_options.field_type() {
                 FieldType::HierarchicalFacet => {
-                    let facets: Vec<&[u8]> = field_values
+                    let facets: Vec<&str> = field_values
                         .iter()
                         .flat_map(|field_value| match *field_value.value() {
-                            Value::Facet(ref facet) => Some(facet.encoded_bytes()),
+                            Value::Facet(ref facet) => Some(facet.encoded_str()),
                             _ => {
                                 panic!("Expected hierarchical facet");
                             }
                         })
                         .collect();
                     let mut term = Term::for_field(field); // we set the Term
-                    for facet_bytes in facets {
+                    for fake_str in facets {
                         let mut unordered_term_id_opt = None;
-                        let fake_str = unsafe { str::from_utf8_unchecked(facet_bytes) };
                         FacetTokenizer.token_stream(fake_str).process(&mut |token| {
                             term.set_text(&token.text);
                             let unordered_term_id =

src/postings/segment_postings.rs

@@ -126,7 +126,6 @@ impl SegmentPostings {
 fn exponential_search(target: u32, arr: &[u32]) -> (usize, usize) {
     let mut start = 0;
     let end = arr.len();
-    debug_assert!(target >= arr[start]);
     debug_assert!(target <= arr[end - 1]);
     let mut jump = 1;
     loop {
@@ -216,11 +215,10 @@ impl DocSet for SegmentPostings {
 
         // we're in the right block now, start with an exponential search
         let block_docs = self.block_cursor.docs();
-
-        debug_assert!(target >= self.doc());
         let new_cur = self
             .cur
             .wrapping_add(search_within_block(&block_docs[self.cur..], target));
+
         if need_positions {
             sum_freqs_skipped += self.block_cursor.freqs()[self.cur..new_cur]
                 .iter()
@@ -622,6 +620,7 @@ impl<'b> Streamer<'b> for BlockSegmentPostings {
 #[cfg(test)]
 mod tests {
 
+    use super::exponential_search;
     use super::search_within_block;
     use super::BlockSegmentPostings;
     use super::BlockSegmentPostingsSkipResult;
@@ -635,6 +634,7 @@ mod tests {
     use schema::Term;
     use schema::INT_INDEXED;
     use DocId;
+    use SkipResult;
 
     #[test]
     fn test_empty_segment_postings() {
@@ -662,6 +662,16 @@ mod tests {
             .0
     }
 
+    #[test]
+    fn test_exponentiel_search() {
+        assert_eq!(exponential_search(0, &[1, 2]), (0, 1));
+        assert_eq!(exponential_search(1, &[1, 2]), (0, 1));
+        assert_eq!(
+            exponential_search(7, &[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]),
+            (3, 7)
+        );
+    }
+
     fn util_test_search_within_block(block: &[u32], target: u32) {
         assert_eq!(
             search_within_block(block, target),
@@ -693,7 +703,7 @@ mod tests {
 
     #[test]
     fn test_block_segment_postings() {
-        let mut block_segments = build_block_postings((0..100_000).collect::<Vec<u32>>());
+        let mut block_segments = build_block_postings(&(0..100_000).collect::<Vec<u32>>());
         let mut offset: u32 = 0u32;
         // checking that the block before calling advance is empty
         assert!(block_segments.docs().is_empty());
@@ -707,14 +717,44 @@ mod tests {
         }
     }
 
-    fn build_block_postings(docs: Vec<DocId>) -> BlockSegmentPostings {
+    #[test]
+    fn test_skip_right_at_new_block() {
+        let mut doc_ids = (0..128).collect::<Vec<u32>>();
+        doc_ids.push(129);
+        doc_ids.push(130);
+        {
+            let block_segments = build_block_postings(&doc_ids);
+            let mut docset = SegmentPostings::from_block_postings(block_segments, None);
+            assert_eq!(docset.skip_next(128), SkipResult::OverStep);
+            assert_eq!(docset.doc(), 129);
+            assert!(docset.advance());
+            assert_eq!(docset.doc(), 130);
+            assert!(!docset.advance());
+        }
+        {
+            let block_segments = build_block_postings(&doc_ids);
+            let mut docset = SegmentPostings::from_block_postings(block_segments, None);
+            assert_eq!(docset.skip_next(129), SkipResult::Reached);
+            assert_eq!(docset.doc(), 129);
+            assert!(docset.advance());
+            assert_eq!(docset.doc(), 130);
+            assert!(!docset.advance());
+        }
+        {
+            let block_segments = build_block_postings(&doc_ids);
+            let mut docset = SegmentPostings::from_block_postings(block_segments, None);
+            assert_eq!(docset.skip_next(131), SkipResult::End);
+        }
+    }
+
+    fn build_block_postings(docs: &[DocId]) -> BlockSegmentPostings {
         let mut schema_builder = Schema::builder();
         let int_field = schema_builder.add_u64_field("id", INT_INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_with_num_threads(1, 40_000_000).unwrap();
         let mut last_doc = 0u32;
-        for doc in docs {
+        for &doc in docs {
             for _ in last_doc..doc {
                 index_writer.add_document(doc!(int_field=>1u64));
             }
@@ -734,7 +774,7 @@ mod tests {
     #[test]
     fn test_block_segment_postings_skip() {
         for i in 0..4 {
-            let mut block_postings = build_block_postings(vec![3]);
+            let mut block_postings = build_block_postings(&[3]);
             assert_eq!(
                 block_postings.skip_to(i),
                 BlockSegmentPostingsSkipResult::Success(0u32)
@@ -744,7 +784,7 @@ mod tests {
                 BlockSegmentPostingsSkipResult::Terminated
             );
         }
-        let mut block_postings = build_block_postings(vec![3]);
+        let mut block_postings = build_block_postings(&[3]);
         assert_eq!(
             block_postings.skip_to(4u32),
             BlockSegmentPostingsSkipResult::Terminated
@@ -757,7 +797,7 @@ mod tests {
         for i in 0..1300 {
             docs.push((i * i / 100) + i);
         }
-        let mut block_postings = build_block_postings(docs.clone());
+        let mut block_postings = build_block_postings(&docs[..]);
         for i in vec![0, 424, 10000] {
             assert_eq!(
                 block_postings.skip_to(i),

src/postings/stacker/mod.rs

@@ -1,9 +1,7 @@
 mod expull;
 mod memory_arena;
-mod murmurhash2;
 mod term_hashmap;
 
 pub use self::expull::ExpUnrolledLinkedList;
 pub use self::memory_arena::{Addr, ArenaStorable, MemoryArena};
-use self::murmurhash2::murmurhash2;
 pub use self::term_hashmap::{compute_table_size, TermHashMap};

src/postings/stacker/murmurhash2.rs

@@ -1,87 +0,0 @@
-use std::ptr;
-const SEED: u32 = 3_242_157_231u32;
-const M: u32 = 0x5bd1_e995;
-
-#[inline(always)]
-pub fn murmurhash2(key: &[u8]) -> u32 {
-    #[cfg_attr(feature = "cargo-clippy", allow(clippy::cast_ptr_alignment))]
-    let mut key_ptr: *const u32 = key.as_ptr() as *const u32;
-    let len = key.len() as u32;
-    let mut h: u32 = SEED ^ len;
-
-    let num_blocks = len >> 2;
-    for _ in 0..num_blocks {
-        let mut k: u32 = unsafe { ptr::read_unaligned(key_ptr) }; // ok because of num_blocks definition
-        k = k.wrapping_mul(M);
-        k ^= k >> 24;
-        k = k.wrapping_mul(M);
-        h = h.wrapping_mul(M);
-        h ^= k;
-        key_ptr = key_ptr.wrapping_offset(1);
-    }
-
-    // Handle the last few bytes of the input array
-    let remaining: &[u8] = &key[key.len() & !3..];
-    match remaining.len() {
-        3 => {
-            h ^= u32::from(remaining[2]) << 16;
-            h ^= u32::from(remaining[1]) << 8;
-            h ^= u32::from(remaining[0]);
-            h = h.wrapping_mul(M);
-        }
-        2 => {
-            h ^= u32::from(remaining[1]) << 8;
-            h ^= u32::from(remaining[0]);
-            h = h.wrapping_mul(M);
-        }
-        1 => {
-            h ^= u32::from(remaining[0]);
-            h = h.wrapping_mul(M);
-        }
-        _ => {}
-    }
-    h ^= h >> 13;
-    h = h.wrapping_mul(M);
-    h ^ (h >> 15)
-}
-
-#[cfg(test)]
-mod test {
-
-    use super::murmurhash2;
-    use std::collections::HashSet;
-
-    #[test]
-    fn test_murmur() {
-        let s1 = "abcdef";
-        let s2 = "abcdeg";
-        for i in 0..5 {
-            assert_eq!(
-                murmurhash2(&s1[i..5].as_bytes()),
-                murmurhash2(&s2[i..5].as_bytes())
-            );
-        }
-    }
-
-    #[test]
-    fn test_murmur_against_reference_impl() {
-        assert_eq!(murmurhash2("".as_bytes()), 3632506080);
-        assert_eq!(murmurhash2("a".as_bytes()), 455683869);
-        assert_eq!(murmurhash2("ab".as_bytes()), 2448092234);
-        assert_eq!(murmurhash2("abc".as_bytes()), 2066295634);
-        assert_eq!(murmurhash2("abcd".as_bytes()), 2588571162);
-        assert_eq!(murmurhash2("abcde".as_bytes()), 2988696942);
-        assert_eq!(murmurhash2("abcdefghijklmnop".as_bytes()), 2350868870);
-    }
-
-    #[test]
-    fn test_murmur_collisions() {
-        let mut set: HashSet<u32> = HashSet::default();
-        for i in 0..10_000 {
-            let s = format!("hash{}", i);
-            let hash = murmurhash2(s.as_bytes());
-            set.insert(hash);
-        }
-        assert_eq!(set.len(), 10_000);
-    }
-}

src/postings/stacker/term_hashmap.rs

@@ -1,4 +1,7 @@
-use super::murmurhash2;
+extern crate murmurhash32;
+
+use self::murmurhash32::murmurhash2;
+
 use super::{Addr, ArenaStorable, MemoryArena};
 use std::iter;
 use std::mem;
@@ -206,7 +209,7 @@ impl TermHashMap {
             self.resize();
         }
         let key_bytes: &[u8] = key.as_ref();
-        let hash = murmurhash2::murmurhash2(key.as_ref());
+        let hash = murmurhash2(key.as_ref());
         let mut probe = self.probe(hash);
         loop {
             let bucket = probe.next_probe();

src/schema/facet.rs

@@ -6,6 +6,7 @@ use std::borrow::Cow;
 use std::fmt::{self, Debug, Display, Formatter};
 use std::io::{self, Read, Write};
 use std::str;
+use std::string::FromUtf8Error;
 
 const SLASH_BYTE: u8 = b'/';
 const ESCAPE_BYTE: u8 = b'\\';
@@ -14,6 +15,10 @@ const ESCAPE_BYTE: u8 = b'\\';
 /// representation of facets.
 pub const FACET_SEP_BYTE: u8 = 0u8;
 
+/// `char` used as a level separation in the binary
+/// representation of facets. (It is the null codepoint.)
+pub const FACET_SEP_CHAR: char = '\u{0}';
+
 /// A Facet represent a point in a given hierarchy.
 ///
 /// They are typically represented similarly to a filepath.
@@ -26,18 +31,18 @@ pub const FACET_SEP_BYTE: u8 = 0u8;
 /// its facet. In the example above, `/electronics/tv_and_video/`
 /// and `/electronics`.
 #[derive(Clone, Eq, Hash, PartialEq, Ord, PartialOrd)]
-pub struct Facet(Vec<u8>);
+pub struct Facet(String);
 
 impl Facet {
     /// Returns a new instance of the "root facet"
     /// Equivalent to `/`.
     pub fn root() -> Facet {
-        Facet(vec![])
+        Facet("".to_string())
     }
 
     /// Returns true iff the facet is the root facet `/`.
     pub fn is_root(&self) -> bool {
-        self.encoded_bytes().is_empty()
+        self.encoded_str().is_empty()
     }
 
     /// Returns a binary representation of the facet.
@@ -49,13 +54,19 @@ impl Facet {
     /// This representation has the benefit of making it possible to
     /// express "being a child of a given facet" as a range over
     /// the term ordinals.
-    pub fn encoded_bytes(&self) -> &[u8] {
+    pub fn encoded_str(&self) -> &str {
         &self.0
     }
 
+    pub(crate) fn from_encoded_string(facet_string: String) -> Facet {
+        Facet(facet_string)
+    }
+
     /// Creates a `Facet` from its binary representation.
-    pub(crate) unsafe fn from_encoded(encoded_bytes: Vec<u8>) -> Facet {
-        Facet(encoded_bytes)
+    pub fn from_encoded(encoded_bytes: Vec<u8>) -> Result<Facet, FromUtf8Error> {
+        // facet bytes validation. `0u8` is used a separator but that is still legal utf-8
+        //Ok(Facet(String::from_utf8(encoded_bytes)?))
+        String::from_utf8(encoded_bytes).map(Facet)
     }
 
     /// Parse a text representation of a facet.
@@ -79,36 +90,37 @@ impl Facet {
         Path: IntoIterator,
         Path::Item: ToString,
     {
-        let mut facet_bytes: Vec<u8> = Vec::with_capacity(100);
+        let mut facet_string: String = String::with_capacity(100);
         let mut step_it = path.into_iter();
         if let Some(step) = step_it.next() {
-            facet_bytes.extend_from_slice(step.to_string().as_bytes());
+            facet_string.push_str(&step.to_string());
         }
         for step in step_it {
-            facet_bytes.push(FACET_SEP_BYTE);
-            facet_bytes.extend_from_slice(step.to_string().as_bytes());
+            facet_string.push(FACET_SEP_CHAR);
+            facet_string.push_str(&step.to_string());
         }
-        Facet(facet_bytes)
+        Facet(facet_string)
     }
 
     /// Accessor for the inner buffer of the `Facet`.
-    pub(crate) fn inner_buffer_mut(&mut self) -> &mut Vec<u8> {
-        &mut self.0
+    pub(crate) fn set_facet_str(&mut self, facet_str: &str) {
+        self.0.clear();
+        self.0.push_str(facet_str);
     }
 
     /// Returns `true` iff other is a subfacet of `self`.
     pub fn is_prefix_of(&self, other: &Facet) -> bool {
-        let self_bytes: &[u8] = self.encoded_bytes();
-        let other_bytes: &[u8] = other.encoded_bytes();
-        self_bytes.len() < other_bytes.len()
-            && other_bytes.starts_with(self_bytes)
-            && other_bytes[self_bytes.len()] == 0u8
+        let self_str = self.encoded_str();
+        let other_str = other.encoded_str();
+        self_str.len() < other_str.len()
+            && other_str.starts_with(self_str)
+            && other_str.as_bytes()[self_str.len()] == FACET_SEP_BYTE
     }
 }
 
-impl Borrow<[u8]> for Facet {
-    fn borrow(&self) -> &[u8] {
-        self.encoded_bytes()
+impl Borrow<str> for Facet {
+    fn borrow(&self) -> &str {
+        self.encoded_str()
     }
 }
 
@@ -120,45 +132,51 @@ impl<'a, T: ?Sized + AsRef<str>> From<&'a T> for Facet {
             Idle,
         }
         let path: &str = path_asref.as_ref();
-        let mut facet_encoded = Vec::new();
+        assert!(!path.is_empty());
+        assert!(path.starts_with("/"));
+        let mut facet_encoded = String::new();
         let mut state = State::Idle;
         let path_bytes = path.as_bytes();
-        for &c in &path_bytes[1..] {
+        let mut last_offset = 1;
+        for i in 1..path_bytes.len() {
+            let c = path_bytes[i];
             match (state, c) {
-                (State::Idle, ESCAPE_BYTE) => state = State::Escaped,
+                (State::Idle, ESCAPE_BYTE) => {
+                    facet_encoded.push_str(&path[last_offset..i]);
+                    last_offset = i + 1;
+                    state = State::Escaped
+                }
                 (State::Idle, SLASH_BYTE) => {
-                    facet_encoded.push(FACET_SEP_BYTE);
+                    facet_encoded.push_str(&path[last_offset..i]);
+                    facet_encoded.push(FACET_SEP_CHAR);
+                    last_offset = i + 1;
                 }
-                (State::Escaped, any_char) => {
+                (State::Escaped, _escaped_char) => {
                     state = State::Idle;
-                    facet_encoded.push(any_char);
-                }
-                (State::Idle, other_char) => {
-                    facet_encoded.push(other_char);
                 }
+                (State::Idle, _any_char) => {}
             }
         }
+        facet_encoded.push_str(&path[last_offset..]);
         Facet(facet_encoded)
     }
 }
 
 impl BinarySerializable for Facet {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
-        <Vec<u8> as BinarySerializable>::serialize(&self.0, writer)
+        <String as BinarySerializable>::serialize(&self.0, writer)
     }
 
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        let bytes = <Vec<u8> as BinarySerializable>::deserialize(reader)?;
-        Ok(Facet(bytes))
+        Ok(Facet(<String as BinarySerializable>::deserialize(reader)?))
     }
 }
 
 impl Display for Facet {
     fn fmt(&self, f: &mut Formatter) -> fmt::Result {
-        for step in self.0.split(|&b| b == FACET_SEP_BYTE) {
+        for step in self.0.split(FACET_SEP_CHAR) {
             write!(f, "/")?;
-            let step_str = unsafe { str::from_utf8_unchecked(step) };
-            write!(f, "{}", escape_slashes(step_str))?;
+            write!(f, "{}", escape_slashes(step))?;
         }
         Ok(())
     }

src/schema/term.rs

@@ -32,7 +32,7 @@ impl Term {
 
     /// Creates a `Term` given a facet.
     pub fn from_facet(field: Field, facet: &Facet) -> Term {
-        let bytes = facet.encoded_bytes();
+        let bytes = facet.encoded_str().as_bytes();
         let buffer = Vec::with_capacity(4 + bytes.len());
         let mut term = Term(buffer);
         term.set_field(field);
@@ -68,12 +68,7 @@ impl Term {
         term
     }
 
-    /// Creates a new Term with an empty buffer,
-    /// but with a given capacity.
-    ///
-    /// It is declared unsafe, as the term content
-    /// is not initialized, and a call to `.field()`
-    /// would panic.
+    /// Creates a new Term for a given field.
     pub(crate) fn for_field(field: Field) -> Term {
         let mut term = Term(Vec::with_capacity(100));
         term.set_field(field);

src/termdict/mod.rs

@@ -167,7 +167,7 @@ mod tests {
         let mut term_string = String::new();
         while term_it.advance() {
             //let term = Term::from_bytes(term_it.key());
-            term_string.push_str(unsafe { str::from_utf8_unchecked(term_it.key()) }); // ok test
+            term_string.push_str(str::from_utf8(term_it.key()).expect("test"));
         }
         assert_eq!(&*term_string, "abcdef");
     }

src/tokenizer/facet_tokenizer.rs

@@ -1,6 +1,5 @@
 use super::{Token, TokenStream, Tokenizer};
 use schema::FACET_SEP_BYTE;
-use std::str;
 
 /// The `FacetTokenizer` process a `Facet` binary representation
 /// and emits a token for all of its parent.
@@ -57,12 +56,11 @@ impl<'a> TokenStream for FacetTokenStream<'a> {
                     .position(|b| b == FACET_SEP_BYTE)
                     .map(|pos| cursor + 1 + pos)
                 {
-                    let facet_part =
-                        unsafe { str::from_utf8_unchecked(&bytes[cursor..next_sep_pos]) };
+                    let facet_part = &self.text[cursor..next_sep_pos];
                     self.token.text.push_str(facet_part);
                     self.state = State::UpToPosition(next_sep_pos);
                 } else {
-                    let facet_part = unsafe { str::from_utf8_unchecked(&bytes[cursor..]) };
+                    let facet_part = &self.text[cursor..];
                     self.token.text.push_str(facet_part);
                     self.state = State::Terminated;
                 }
@@ -86,7 +84,6 @@ mod tests {
 
     use super::FacetTokenizer;
     use schema::Facet;
-    use std::str;
     use tokenizer::{Token, TokenStream, Tokenizer};
 
     #[test]
@@ -95,11 +92,11 @@ mod tests {
         let mut tokens = vec![];
         {
             let mut add_token = |token: &Token| {
-                let facet = unsafe { Facet::from_encoded(token.text.as_bytes().to_owned()) }; // ok test
+                let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap();
                 tokens.push(format!("{}", facet));
             };
             FacetTokenizer
-                .token_stream(unsafe { str::from_utf8_unchecked(facet.encoded_bytes()) })
+                .token_stream(facet.encoded_str())
                 .process(&mut add_token);
         }
         assert_eq!(tokens.len(), 4);
@@ -115,11 +112,11 @@ mod tests {
         let mut tokens = vec![];
         {
             let mut add_token = |token: &Token| {
-                let facet = unsafe { Facet::from_encoded(token.text.as_bytes().to_owned()) }; // ok test
+                let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap(); // ok test
                 tokens.push(format!("{}", facet));
             };
             FacetTokenizer
-                .token_stream(unsafe { str::from_utf8_unchecked(facet.encoded_bytes()) }) // ok test
+                .token_stream(facet.encoded_str()) // ok test
                 .process(&mut add_token);
         }
         assert_eq!(tokens.len(), 1);