Various changes. Need to cherrypick some of them and put them into master

src/datastruct/stacker/hashmap.rs

@@ -58,10 +58,9 @@ mod murmurhash2 {
 ///
 /// Returns (the heap size in bytes, the hash table size in number of bits)
 pub(crate) fn split_memory(per_thread_memory_budget: usize) -> (usize, usize) {
-    let table_size_limit: usize = per_thread_memory_budget / 3;
+    let table_size_limit: usize = per_thread_memory_budget / 5;
     let compute_table_size = |num_bits: usize| {
-        let table_size: usize = (1 << num_bits) * mem::size_of::<KeyValue>();
-        table_size * mem::size_of::<KeyValue>()
+        (1 << num_bits) * mem::size_of::<KeyValue>()
     };
     let table_num_bits: usize = (1..)
         .into_iter()

src/indexer/merger.rs

@@ -105,6 +105,7 @@ impl IndexMerger {
         })
     }
 
+    #[inline(never)]
     fn write_fieldnorms(&self, fast_field_serializer: &mut FastFieldSerializer) -> Result<()> {
         let fieldnorm_fastfields: Vec<Field> = self.schema
             .fields()
@@ -120,6 +121,7 @@ impl IndexMerger {
         )
     }
 
+    #[inline(never)]
     fn write_fast_fields(&self, fast_field_serializer: &mut FastFieldSerializer) -> Result<()> {
         let fast_fields: Vec<Field> = self.schema
             .fields()
@@ -198,6 +200,7 @@ impl IndexMerger {
         Ok(())
     }
 
+    #[inline(never)]
     fn write_postings(&self, serializer: &mut InvertedIndexSerializer) -> Result<()> {
         let mut delta_computer = DeltaComputer::new();
 
@@ -332,6 +335,7 @@ impl IndexMerger {
         Ok(())
     }
 
+    #[inline(never)]
     fn write_storable_fields(&self, store_writer: &mut StoreWriter) -> Result<()> {
         for reader in &self.readers {
             let store_reader = reader.get_store_reader();

src/indexer/segment_serializer.rs

@@ -56,6 +56,7 @@ impl SegmentSerializer {
     }
 
     /// Finalize the segment serialization.
+    #[inline(never)]
     pub fn close(self) -> Result<()> {
         self.fast_field_serializer.close()?;
         self.postings_serializer.close()?;

src/termdict/streamdict/delta_encoder.rs

@@ -100,13 +100,13 @@ impl TermInfoDeltaEncoder {
     pub fn encode(&mut self, term_info: TermInfo) -> DeltaTermInfo {
         let mut delta_term_info = DeltaTermInfo {
             doc_freq: term_info.doc_freq,
-            delta_postings_offset: term_info.postings_offset - self.term_info.postings_offset,
+            delta_postings_offset: (term_info.postings_offset - self.term_info.postings_offset) as u32,
             delta_positions_offset: 0,
             positions_inner_offset: 0,
         };
         if self.has_positions {
             delta_term_info.delta_positions_offset =
-                term_info.positions_offset - self.term_info.positions_offset;
+                (term_info.positions_offset - self.term_info.positions_offset) as u32;
             delta_term_info.positions_inner_offset = term_info.positions_inner_offset;
         }
         mem::replace(&mut self.term_info, term_info);
@@ -155,12 +155,12 @@ impl TermInfoDeltaDecoder {
         let delta_postings_offset: u32 = (v as u32) & make_mask(num_bytes_postings_offset);
         cursor = &cursor[num_bytes_docfreq + num_bytes_postings_offset..];
         self.term_info.doc_freq = doc_freq;
-        self.term_info.postings_offset += delta_postings_offset;
+        self.term_info.postings_offset += delta_postings_offset as u64;
         if self.has_positions {
             let num_bytes_positions_offset = ((code >> 5) & 3) as usize + 1;
             let delta_positions_offset: u32 =
                 unsafe { *(cursor.as_ptr() as *const u32) } & make_mask(num_bytes_positions_offset);
-            self.term_info.positions_offset += delta_positions_offset;
+            self.term_info.positions_offset += delta_positions_offset as u64;
             self.term_info.positions_inner_offset = cursor[num_bytes_positions_offset];
             &cursor[num_bytes_positions_offset + 1..]
         } else {

src/termdict/streamdict/mod.rs

@@ -1,10 +1,11 @@
 use std::io::{self, Read, Write};
-use common::BinarySerializable;
+use common::{VInt, BinarySerializable};
 
 mod termdict;
 mod streamer;
 mod delta_encoder;
 
+
 pub use self::delta_encoder::{TermDeltaDecoder, TermDeltaEncoder};
 pub use self::delta_encoder::{DeltaTermInfo, TermInfoDeltaDecoder, TermInfoDeltaEncoder};
 
@@ -15,23 +16,23 @@ pub use self::streamer::TermStreamerBuilderImpl;
 
 #[derive(Debug)]
 pub struct CheckPoint {
-    pub stream_offset: u32,
-    pub postings_offset: u32,
-    pub positions_offset: u32,
+    pub stream_offset: u64,
+    pub postings_offset: u64,
+    pub positions_offset: u64,
 }
 
 impl BinarySerializable for CheckPoint {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
-        self.stream_offset.serialize(writer)?;
-        self.postings_offset.serialize(writer)?;
-        self.positions_offset.serialize(writer)?;
+        VInt(self.stream_offset).serialize(writer)?;
+        VInt(self.postings_offset).serialize(writer)?;
+        VInt(self.positions_offset).serialize(writer)?;
         Ok(())
     }
 
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        let stream_offset = u32::deserialize(reader)?;
-        let postings_offset = u32::deserialize(reader)?;
-        let positions_offset = u32::deserialize(reader)?;
+        let stream_offset = VInt::deserialize(reader)?.0;
+        let postings_offset = VInt::deserialize(reader)?.0;
+        let positions_offset = VInt::deserialize(reader)?.0;
         Ok(CheckPoint {
             stream_offset,
             postings_offset,

src/termdict/streamdict/termdict.rs

@@ -28,11 +28,10 @@ fn has_positions(field_type: &FieldType) -> bool {
     match *field_type {
         FieldType::Str(ref text_options) => {
             let indexing_options = text_options.get_indexing_options();
-            if indexing_options.is_position_enabled() {
-                true
-            } else {
-                false
+            if let Some(text_field_indexing) = indexing_options {
+                return text_field_indexing.index_option().has_positions()
             }
+            return false;
         }
         _ => false,
     }
@@ -60,10 +59,10 @@ where
     W: Write,
 {
     fn add_index_entry(&mut self) {
-        let stream_offset = self.write.written_bytes() as u32;
+        let stream_offset: u64 = self.write.written_bytes() as u64;
         let term_info = self.term_info_encoder.term_info();
-        let postings_offset = term_info.postings_offset as u32;
-        let positions_offset = term_info.positions_offset as u32;
+        let postings_offset: u64 = term_info.postings_offset;
+        let positions_offset: u64 = term_info.positions_offset;
         let checkpoint = CheckPoint {
             stream_offset,
             postings_offset,

src/tokenizer/alphanum_only.rs

@@ -0,0 +1,65 @@
+use super::{Token, TokenFilter, TokenStream};
+
+#[derive(Clone)]
+pub struct AlphaNumOnlyFilter;
+
+
+pub struct AlphaNumOnlyFilterStream<TailTokenStream>
+    where TailTokenStream: TokenStream
+{
+    tail: TailTokenStream,
+}
+
+
+impl<TailTokenStream> AlphaNumOnlyFilterStream<TailTokenStream>
+    where TailTokenStream: TokenStream
+{
+    fn predicate(&self, token: &Token) -> bool {
+        token.text.chars().all(|c| c.is_ascii_alphanumeric())
+    }
+
+    fn wrap(
+        tail: TailTokenStream,
+    ) -> AlphaNumOnlyFilterStream<TailTokenStream> {
+        AlphaNumOnlyFilterStream {
+            tail
+        }
+    }
+}
+
+
+impl<TailTokenStream> TokenFilter<TailTokenStream> for AlphaNumOnlyFilter
+    where
+        TailTokenStream: TokenStream,
+{
+    type ResultTokenStream = AlphaNumOnlyFilterStream<TailTokenStream>;
+
+    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
+        AlphaNumOnlyFilterStream::wrap(token_stream)
+    }
+}
+
+impl<TailTokenStream> TokenStream for AlphaNumOnlyFilterStream<TailTokenStream>
+    where
+        TailTokenStream: TokenStream
+{
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
+
+    fn advance(&mut self) -> bool {
+        loop {
+            if self.tail.advance() {
+                if self.predicate(self.tail.token()) {
+                    return true;
+                }
+            } else {
+                return false;
+            }
+        }
+    }
+}

src/tokenizer/mod.rs

@@ -137,7 +137,10 @@ mod tokenizer_manager;
 mod japanese_tokenizer;
 mod token_stream_chain;
 mod raw_tokenizer;
+mod alphanum_only;
 
+
+pub use self::alphanum_only::AlphaNumOnlyFilter;
 pub use self::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
 pub use self::tokenizer::BoxedTokenizer;
 pub use self::tokenizer_manager::TokenizerManager;

src/tokenizer/tokenizer_manager.rs

@@ -7,6 +7,7 @@ use tokenizer::RawTokenizer;
 use tokenizer::SimpleTokenizer;
 use tokenizer::JapaneseTokenizer;
 use tokenizer::RemoveLongFilter;
+use tokenizer::AlphaNumOnlyFilter;
 use tokenizer::LowerCaser;
 use tokenizer::Stemmer;
 
@@ -70,6 +71,7 @@ impl Default for TokenizerManager {
             SimpleTokenizer
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser)
+                .filter(AlphaNumOnlyFilter)
                 .filter(Stemmer::new()),
         );
         manager.register("ja", JapaneseTokenizer.filter(RemoveLongFilter::limit(40)));