Added psuedocode for poll implementation

* Removed generics in tokenizers

* lowercaser

* Added TokenizerExt

* Introducing BoxedTokenizer

* Introducing BoxXXXXX helper struct

* Closes #762.

* Introducing a TextAnalyzer

CHANGELOG.md

@@ -1,3 +1,13 @@
+Tantivy 0.12.0
+======================
+- Removing static dispatch in tokenizers for simplicity. (#762)
+
+## How to update?
+
+Crates relying on custom tokenizer, or registering tokenizer in the manager will require some 
+minor changes. Check https://github.com/tantivy-search/tantivy/blob/master/examples/custom_tokenizer.rs
+to check for some code sample.
+
 Tantivy 0.11.3
 =======================
 - Fixed DateTime as a fast field (#735)

Cargo.toml

@@ -18,7 +18,7 @@ byteorder = "1.0"
 crc32fast = "1.2.0"
 once_cell = "1.0"
 regex ={version = "1.3.0", default-features = false, features = ["std"]}
-tantivy-fst = "0.1"
+tantivy-fst = "0.2"
 memmap = {version = "0.7", optional=true}
 lz4 = {version="1.20", optional=true}
 snap = {version="0.2"}

examples/pre_tokenized_text.rs

@@ -9,7 +9,7 @@
 // - import tokenized text straight from json,
 // - perform a search on documents with pre-tokenized text
 
-use tantivy::tokenizer::{PreTokenizedString, SimpleTokenizer, Token, TokenStream, Tokenizer};
+use tantivy::tokenizer::{PreTokenizedString, SimpleTokenizer, Token, Tokenizer};
 
 use tantivy::collector::{Count, TopDocs};
 use tantivy::query::TermQuery;

examples/stop_words.rs

@@ -50,7 +50,7 @@ fn main() -> tantivy::Result<()> {
 
     // This tokenizer lowers all of the text (to help with stop word matching)
     // then removes all instances of `the` and `and` from the corpus
-    let tokenizer = SimpleTokenizer
+    let tokenizer = TextAnalyzer::from(SimpleTokenizer)
         .filter(LowerCaser)
         .filter(StopWordFilter::remove(vec![
             "the".to_string(),

src/collector/top_score_collector.rs

@@ -6,6 +6,7 @@ use crate::collector::tweak_score_top_collector::TweakedScoreTopCollector;
 use crate::collector::{
     CustomScorer, CustomSegmentScorer, ScoreSegmentTweaker, ScoreTweaker, SegmentCollector,
 };
+use crate::fastfield::FastFieldReader;
 use crate::schema::Field;
 use crate::DocAddress;
 use crate::DocId;
@@ -61,6 +62,34 @@ impl fmt::Debug for TopDocs {
     }
 }
 
+struct ScorerByFastFieldReader {
+    ff_reader: FastFieldReader<u64>,
+}
+
+impl CustomSegmentScorer<u64> for ScorerByFastFieldReader {
+    fn score(&self, doc: DocId) -> u64 {
+        self.ff_reader.get_u64(u64::from(doc))
+    }
+}
+
+struct ScorerByField {
+    field: Field,
+}
+
+impl CustomScorer<u64> for ScorerByField {
+    type Child = ScorerByFastFieldReader;
+
+    fn segment_scorer(&self, segment_reader: &SegmentReader) -> crate::Result<Self::Child> {
+        let ff_reader = segment_reader
+            .fast_fields()
+            .u64(self.field)
+            .ok_or_else(|| {
+                crate::Error::SchemaError(format!("Field requested is not a i64/u64 fast field."))
+            })?;
+        Ok(ScorerByFastFieldReader { ff_reader })
+    }
+}
+
 impl TopDocs {
     /// Creates a top score collector, with a number of documents equal to "limit".
     ///
@@ -143,14 +172,7 @@ impl TopDocs {
         self,
         field: Field,
     ) -> impl Collector<Fruit = Vec<(u64, DocAddress)>> {
-        self.custom_score(move |segment_reader: &SegmentReader| {
+        self.custom_score(ScorerByField { field })
-            let ff_reader = segment_reader
-                .fast_fields()
-                .u64(field)
-                .expect("Field requested is not a i64/u64 fast field.");
-            //TODO error message missmatch actual behavior for i64
-            move |doc: DocId| ff_reader.get(doc)
-        })
     }
 
     /// Ranks the documents using a custom score.
@@ -572,7 +594,6 @@ mod tests {
     }
 
     #[test]
-    #[should_panic(expected = "Field requested is not a i64/u64 fast field")]
     fn test_field_not_fast_field() {
         let mut schema_builder = Schema::builder();
         let title = schema_builder.add_text_field(TITLE, TEXT);
@@ -587,7 +608,12 @@ mod tests {
         let searcher = index.reader().unwrap().searcher();
         let segment = searcher.segment_reader(0);
         let top_collector = TopDocs::with_limit(4).order_by_u64_field(size);
-        assert!(top_collector.for_segment(0, segment).is_ok());
+        let err = top_collector.for_segment(0, segment);
+        if let Err(crate::Error::SchemaError(msg)) = err {
+            assert_eq!(msg, "Field requested is not a i64/u64 fast field.");
+        } else {
+            assert!(false);
+        }
     }
 
     fn index(

src/core/index.rs

@@ -20,8 +20,7 @@ use crate::reader::IndexReaderBuilder;
 use crate::schema::Field;
 use crate::schema::FieldType;
 use crate::schema::Schema;
-use crate::tokenizer::BoxedTokenizer;
+use crate::tokenizer::{TextAnalyzer, TokenizerManager};
-use crate::tokenizer::TokenizerManager;
 use crate::IndexWriter;
 use crate::Result;
 use num_cpus;
@@ -173,11 +172,11 @@ impl Index {
     }
 
     /// Helper to access the tokenizer associated to a specific field.
-    pub fn tokenizer_for_field(&self, field: Field) -> Result<BoxedTokenizer> {
+    pub fn tokenizer_for_field(&self, field: Field) -> Result<TextAnalyzer> {
         let field_entry = self.schema.get_field_entry(field);
         let field_type = field_entry.field_type();
         let tokenizer_manager: &TokenizerManager = self.tokenizers();
-        let tokenizer_name_opt: Option<BoxedTokenizer> = match field_type {
+        let tokenizer_name_opt: Option<TextAnalyzer> = match field_type {
             FieldType::Str(text_options) => text_options
                 .get_indexing_options()
                 .map(|text_indexing_options| text_indexing_options.tokenizer().to_string())

src/directory/mmap_directory.rs

@@ -141,11 +141,19 @@ impl MmapCache {
     }
 }
 
+pub enum WatcherMode {
+    Event,
+    Poll
+}
+
 struct WatcherWrapper {
     _watcher: Mutex<notify::RecommendedWatcher>,
     watcher_router: Arc<WatchCallbackList>,
+    watcher_mode: WatcherMode,
 }
 
+
+
 impl WatcherWrapper {
     pub fn new(path: &Path) -> Result<Self, OpenDirectoryError> {
         let (tx, watcher_recv): (Sender<RawEvent>, Receiver<RawEvent>) = channel();
@@ -163,33 +171,57 @@ impl WatcherWrapper {
             })?;
         let watcher_router: Arc<WatchCallbackList> = Default::default();
         let watcher_router_clone = watcher_router.clone();
+        let path_clone = path.clone();
+        let meta_path = path_clone.join(*META_FILEPATH);
         thread::Builder::new()
             .name("meta-file-watch-thread".to_string())
             .spawn(move || {
+                let mut old_content = String::new();
+                let mode = WatcherMode::Event;
                 loop {
-                    match watcher_recv.recv().map(|evt| evt.path) {
+                    match mode {
-                        Ok(Some(changed_path)) => {
+                        WatcherMode::Event => {
-                            // ... Actually subject to false positive.
+                            match watcher_recv.recv().map(|evt| evt.path) {
-                            // We might want to be more accurate than this at one point.
+                                Ok(Some(changed_path)) => {
-                            if let Some(filename) = changed_path.file_name() {
+                                    // ... Actually subject to false positive.
-                                if filename == *META_FILEPATH {
+                                    // We might want to be more accurate than this at one point.
-                                    let _ = watcher_router_clone.broadcast();
+                                    if let Some(filename) = changed_path.file_name() {
+                                        if filename == *META_FILEPATH {
+                                            let _ = watcher_router_clone.broadcast();
+                                        }
+                                    }
+                                }
+                                Ok(None) => {
+                                    // not an event we are interested in.
+                                }
+                                Err(_e) => {
+                                    // the watch send channel was dropped
+                                    break;
                                 }
                             }
                         }
-                        Ok(None) => {
+                        WatcherMode::Poll => {
-                            // not an event we are interested in.
+                            let mut file = match File::open(&meta_path) {
+                                Err(why) => panic!("open: nope"),
+                                Ok(file) => file,
+                            };
+                            let mut new_content = String::new();
+                            match file.read_to_string(&mut new_content) {
+                                Err(why) => panic!("read: nope"),
+                                Ok(_) => {},
+                            }
+                            if old_content != new_content {
+                                let _ = watcher_router_clone.broadcast();
+                                old_content = new_content;
+                            }
                         }
-                        Err(_e) => {
+                    };
-                            // the watch send channel was dropped
-                            break;
-                        }
-                    }
                 }
             })?;
         Ok(WatcherWrapper {
             _watcher: Mutex::new(watcher),
             watcher_router,
+            watcher_mode: WatcherMode::Event,
         })
     }
 

src/indexer/operation.rs

@@ -19,6 +19,8 @@ pub struct AddOperation {
 /// UserOperation is an enum type that encapsulates other operation types.
 #[derive(Eq, PartialEq, Debug)]
 pub enum UserOperation {
+    /// Add operation
     Add(Document),
+    /// Delete operation
     Delete(Term),
 }

src/indexer/segment_writer.rs

@@ -11,10 +11,9 @@ use crate::schema::Schema;
 use crate::schema::Term;
 use crate::schema::Value;
 use crate::schema::{Field, FieldEntry};
-use crate::tokenizer::BoxedTokenizer;
+use crate::tokenizer::{BoxTokenStream, PreTokenizedStream};
-use crate::tokenizer::FacetTokenizer;
+use crate::tokenizer::{FacetTokenizer, TextAnalyzer};
-use crate::tokenizer::PreTokenizedStream;
+use crate::tokenizer::{TokenStreamChain, Tokenizer};
-use crate::tokenizer::{TokenStream, TokenStreamChain, Tokenizer};
 use crate::DocId;
 use crate::Opstamp;
 use crate::Result;
@@ -50,7 +49,7 @@ pub struct SegmentWriter {
     fast_field_writers: FastFieldsWriter,
     fieldnorms_writer: FieldNormsWriter,
     doc_opstamps: Vec<Opstamp>,
-    tokenizers: Vec<Option<BoxedTokenizer>>,
+    tokenizers: Vec<Option<TextAnalyzer>>,
 }
 
 impl SegmentWriter {
@@ -159,7 +158,7 @@ impl SegmentWriter {
                     }
                 }
                 FieldType::Str(_) => {
-                    let mut token_streams: Vec<Box<dyn TokenStream>> = vec![];
+                    let mut token_streams: Vec<BoxTokenStream> = vec![];
                     let mut offsets = vec![];
                     let mut total_offset = 0;
 
@@ -172,7 +171,7 @@ impl SegmentWriter {
                                 }
 
                                 token_streams
-                                    .push(Box::new(PreTokenizedStream::from(tok_str.clone())));
+                                    .push(PreTokenizedStream::from(tok_str.clone()).into());
                             }
                             Value::Str(ref text) => {
                                 if let Some(ref mut tokenizer) =
@@ -191,8 +190,7 @@ impl SegmentWriter {
                     let num_tokens = if token_streams.is_empty() {
                         0
                     } else {
-                        let mut token_stream: Box<dyn TokenStream> =
+                        let mut token_stream = TokenStreamChain::new(offsets, token_streams);
-                            Box::new(TokenStreamChain::new(offsets, token_streams));
                         self.multifield_postings
                             .index_text(doc_id, field, &mut token_stream)
                     };

src/indexer/stamper.rs

@@ -1,18 +1,76 @@
 use crate::Opstamp;
 use std::ops::Range;
-use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::atomic::Ordering;
 use std::sync::Arc;
 
+#[cfg(not(target_arch = "arm"))]
+mod atomic_impl {
+
+    use crate::Opstamp;
+    use std::sync::atomic::{AtomicU64, Ordering};
+
+    #[derive(Default)]
+    pub struct AtomicU64Wrapper(AtomicU64);
+
+    impl AtomicU64Wrapper {
+        pub fn new(first_opstamp: Opstamp) -> AtomicU64Wrapper {
+            AtomicU64Wrapper(AtomicU64::new(first_opstamp as u64))
+        }
+
+        pub fn fetch_add(&self, val: u64, order: Ordering) -> u64 {
+            self.0.fetch_add(val as u64, order) as u64
+        }
+
+        pub fn revert(&self, val: u64, order: Ordering) -> u64 {
+            self.0.store(val, order);
+            val
+        }
+    }
+}
+
+#[cfg(target_arch = "arm")]
+mod atomic_impl {
+
+    use crate::Opstamp;
+    /// Under other architecture, we rely on a mutex.
+    use std::sync::atomic::Ordering;
+    use std::sync::RwLock;
+
+    #[derive(Default)]
+    pub struct AtomicU64Wrapper(RwLock<u64>);
+
+    impl AtomicU64Wrapper {
+        pub fn new(first_opstamp: Opstamp) -> AtomicU64Wrapper {
+            AtomicU64Wrapper(RwLock::new(first_opstamp))
+        }
+
+        pub fn fetch_add(&self, incr: u64, _order: Ordering) -> u64 {
+            let mut lock = self.0.write().unwrap();
+            let previous_val = *lock;
+            *lock = previous_val + incr;
+            previous_val
+        }
+
+        pub fn revert(&self, val: u64, _order: Ordering) -> u64 {
+            let mut lock = self.0.write().unwrap();
+            *lock = val;
+            val
+        }
+    }
+}
+
+use self::atomic_impl::AtomicU64Wrapper;
+
 /// Stamper provides Opstamps, which is just an auto-increment id to label
 /// an operation.
 ///
 /// Cloning does not "fork" the stamp generation. The stamper actually wraps an `Arc`.
 #[derive(Clone, Default)]
-pub struct Stamper(Arc<AtomicU64>);
+pub struct Stamper(Arc<AtomicU64Wrapper>);
 
 impl Stamper {
     pub fn new(first_opstamp: Opstamp) -> Stamper {
-        Stamper(Arc::new(AtomicU64::new(first_opstamp)))
+        Stamper(Arc::new(AtomicU64Wrapper::new(first_opstamp)))
     }
 
     pub fn stamp(&self) -> Opstamp {
@@ -31,8 +89,7 @@ impl Stamper {
 
     /// Reverts the stamper to a given `Opstamp` value and returns it
     pub fn revert(&self, to_opstamp: Opstamp) -> Opstamp {
-        self.0.store(to_opstamp, Ordering::SeqCst);
+        self.0.revert(to_opstamp, Ordering::SeqCst)
-        to_opstamp
     }
 }
 

src/lib.rs

@@ -165,6 +165,7 @@ pub use crate::core::SegmentComponent;
 pub use crate::core::{Index, IndexMeta, Searcher, Segment, SegmentId, SegmentMeta};
 pub use crate::core::{InvertedIndexReader, SegmentReader};
 pub use crate::directory::Directory;
+pub use crate::indexer::operation::UserOperation;
 pub use crate::indexer::IndexWriter;
 pub use crate::postings::Postings;
 pub use crate::reader::LeasedItem;

src/query/query_parser/query_parser.rs

@@ -533,7 +533,7 @@ mod test {
     use crate::schema::{IndexRecordOption, TextFieldIndexing, TextOptions};
     use crate::schema::{Schema, Term, INDEXED, STORED, STRING, TEXT};
     use crate::tokenizer::{
-        LowerCaser, SimpleTokenizer, StopWordFilter, Tokenizer, TokenizerManager,
+        LowerCaser, SimpleTokenizer, StopWordFilter, TextAnalyzer, TokenizerManager,
     };
     use crate::Index;
     use matches::assert_matches;
@@ -563,7 +563,7 @@ mod test {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "en_with_stop_words",
-            SimpleTokenizer
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(LowerCaser)
                 .filter(StopWordFilter::remove(vec!["the".to_string()])),
         );

src/reader/mod.rs

@@ -11,6 +11,7 @@ use crate::Result;
 use crate::Searcher;
 use crate::SegmentReader;
 use std::sync::Arc;
+use std::thread;
 
 /// Defines when a new version of the index should be reloaded.
 ///

src/snippet/mod.rs

@@ -1,8 +1,7 @@
 use crate::query::Query;
 use crate::schema::Field;
 use crate::schema::Value;
-use crate::tokenizer::BoxedTokenizer;
+use crate::tokenizer::{TextAnalyzer, Token};
-use crate::tokenizer::{Token, TokenStream};
 use crate::Document;
 use crate::Result;
 use crate::Searcher;
@@ -142,7 +141,7 @@ impl Snippet {
 /// Fragments must be valid in the sense that `&text[fragment.start..fragment.stop]`\
 /// has to be a valid string.
 fn search_fragments<'a>(
-    tokenizer: &BoxedTokenizer,
+    tokenizer: &TextAnalyzer,
     text: &'a str,
     terms: &BTreeMap<String, f32>,
     max_num_chars: usize,
@@ -251,7 +250,7 @@ fn select_best_fragment_combination(fragments: &[FragmentCandidate], text: &str)
 /// ```
 pub struct SnippetGenerator {
     terms_text: BTreeMap<String, f32>,
-    tokenizer: BoxedTokenizer,
+    tokenizer: TextAnalyzer,
     field: Field,
     max_num_chars: usize,
 }
@@ -347,12 +346,11 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet() {
-        let boxed_tokenizer = SimpleTokenizer.into();
         let terms = btreemap! {
             String::from("rust") => 1.0,
             String::from("language") => 0.9
         };
-        let fragments = search_fragments(&boxed_tokenizer, TEST_TEXT, &terms, 100);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 100);
         assert_eq!(fragments.len(), 7);
         {
             let first = &fragments[0];
@@ -374,13 +372,12 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_scored_fragment() {
-        let boxed_tokenizer = SimpleTokenizer.into();
         {
             let terms = btreemap! {
                 String::from("rust") =>1.0f32,
                 String::from("language") => 0.9f32
             };
-            let fragments = search_fragments(&boxed_tokenizer, TEST_TEXT, &terms, 20);
+            let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
             {
                 let first = &fragments[0];
                 assert_eq!(first.score, 1.0);
@@ -389,13 +386,12 @@ Survey in 2016, 2017, and 2018."#;
             let snippet = select_best_fragment_combination(&fragments[..], &TEST_TEXT);
             assert_eq!(snippet.to_html(), "<b>Rust</b> is a systems")
         }
-        let boxed_tokenizer = SimpleTokenizer.into();
         {
             let terms = btreemap! {
                 String::from("rust") =>0.9f32,
                 String::from("language") => 1.0f32
             };
-            let fragments = search_fragments(&boxed_tokenizer, TEST_TEXT, &terms, 20);
+            let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
             //assert_eq!(fragments.len(), 7);
             {
                 let first = &fragments[0];
@@ -409,14 +405,12 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_in_second_fragment() {
-        let boxed_tokenizer = SimpleTokenizer.into();
-
         let text = "a b c d e f g";
 
         let mut terms = BTreeMap::new();
         terms.insert(String::from("c"), 1.0);
 
-        let fragments = search_fragments(&boxed_tokenizer, &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
 
         assert_eq!(fragments.len(), 1);
         {
@@ -433,14 +427,12 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_with_term_at_the_end_of_fragment() {
-        let boxed_tokenizer = SimpleTokenizer.into();
-
         let text = "a b c d e f f g";
 
         let mut terms = BTreeMap::new();
         terms.insert(String::from("f"), 1.0);
 
-        let fragments = search_fragments(&boxed_tokenizer, &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
 
         assert_eq!(fragments.len(), 2);
         {
@@ -457,15 +449,13 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_with_second_fragment_has_the_highest_score() {
-        let boxed_tokenizer = SimpleTokenizer.into();
-
         let text = "a b c d e f g";
 
         let mut terms = BTreeMap::new();
         terms.insert(String::from("f"), 1.0);
         terms.insert(String::from("a"), 0.9);
 
-        let fragments = search_fragments(&boxed_tokenizer, &text, &terms, 7);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 7);
 
         assert_eq!(fragments.len(), 2);
         {
@@ -482,14 +472,12 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_with_term_not_in_text() {
-        let boxed_tokenizer = SimpleTokenizer.into();
-
         let text = "a b c d";
 
         let mut terms = BTreeMap::new();
         terms.insert(String::from("z"), 1.0);
 
-        let fragments = search_fragments(&boxed_tokenizer, &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
 
         assert_eq!(fragments.len(), 0);
 
@@ -500,12 +488,10 @@ Survey in 2016, 2017, and 2018."#;
 
     #[test]
     fn test_snippet_with_no_terms() {
-        let boxed_tokenizer = SimpleTokenizer.into();
-
         let text = "a b c d";
 
         let terms = BTreeMap::new();
-        let fragments = search_fragments(&boxed_tokenizer, &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
         assert_eq!(fragments.len(), 0);
 
         let snippet = select_best_fragment_combination(&fragments[..], &text);

src/tokenizer/alphanum_only.rs

@@ -2,7 +2,7 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = RawTokenizer
+//! let tokenizer = TextAnalyzer::from(RawTokenizer)
 //!   .filter(AlphaNumOnlyFilter);
 //!
 //! let mut stream = tokenizer.token_stream("hello there");
@@ -10,7 +10,7 @@
 //! // contains a space
 //! assert!(stream.next().is_none());
 //!
-//! let tokenizer = SimpleTokenizer
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
 //!   .filter(AlphaNumOnlyFilter);
 //!
 //! let mut stream = tokenizer.token_stream("hello there 💣");
@@ -19,56 +19,30 @@
 //! // the "emoji" is dropped because its not an alphanum
 //! assert!(stream.next().is_none());
 //! ```
-use super::{Token, TokenFilter, TokenStream};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 
 /// `TokenFilter` that removes all tokens that contain non
 /// ascii alphanumeric characters.
 #[derive(Clone)]
 pub struct AlphaNumOnlyFilter;
 
-pub struct AlphaNumOnlyFilterStream<TailTokenStream>
+pub struct AlphaNumOnlyFilterStream<'a> {
-where
+    tail: BoxTokenStream<'a>,
-    TailTokenStream: TokenStream,
-{
-    tail: TailTokenStream,
 }
 
-impl<TailTokenStream> AlphaNumOnlyFilterStream<TailTokenStream>
+impl<'a> AlphaNumOnlyFilterStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
     fn predicate(&self, token: &Token) -> bool {
         token.text.chars().all(|c| c.is_ascii_alphanumeric())
     }
+}
 
-    fn wrap(tail: TailTokenStream) -> AlphaNumOnlyFilterStream<TailTokenStream> {
+impl TokenFilter for AlphaNumOnlyFilter {
-        AlphaNumOnlyFilterStream { tail }
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(AlphaNumOnlyFilterStream { tail: token_stream })
     }
 }
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for AlphaNumOnlyFilter
+impl<'a> TokenStream for AlphaNumOnlyFilterStream<'a> {
-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 {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -78,4 +52,12 @@ where
 
         false
     }
+
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
 }

src/tokenizer/ascii_folding_filter.rs

@@ -1,4 +1,4 @@
-use super::{Token, TokenFilter, TokenStream};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 use std::mem;
 
 /// This class converts alphabetic, numeric, and symbolic Unicode characters
@@ -7,26 +7,21 @@ use std::mem;
 #[derive(Clone)]
 pub struct AsciiFoldingFilter;
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for AsciiFoldingFilter
+impl TokenFilter for AsciiFoldingFilter {
-where
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-    TailTokenStream: TokenStream,
+        From::from(AsciiFoldingFilterTokenStream {
-{
+            tail: token_stream,
-    type ResultTokenStream = AsciiFoldingFilterTokenStream<TailTokenStream>;
+            buffer: String::with_capacity(100),
-
+        })
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
-        AsciiFoldingFilterTokenStream::wrap(token_stream)
     }
 }
 
-pub struct AsciiFoldingFilterTokenStream<TailTokenStream> {
+pub struct AsciiFoldingFilterTokenStream<'a> {
     buffer: String,
-    tail: TailTokenStream,
+    tail: BoxTokenStream<'a>,
 }
 
-impl<TailTokenStream> TokenStream for AsciiFoldingFilterTokenStream<TailTokenStream>
+impl<'a> TokenStream for AsciiFoldingFilterTokenStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
     fn advance(&mut self) -> bool {
         if !self.tail.advance() {
             return false;
@@ -48,18 +43,6 @@ where
     }
 }
 
-impl<TailTokenStream> AsciiFoldingFilterTokenStream<TailTokenStream>
-where
-    TailTokenStream: TokenStream,
-{
-    fn wrap(tail: TailTokenStream) -> AsciiFoldingFilterTokenStream<TailTokenStream> {
-        AsciiFoldingFilterTokenStream {
-            tail,
-            buffer: String::with_capacity(100),
-        }
-    }
-}
-
 // Returns a string that represents the ascii folded version of
 // the character. If the `char` does not require ascii folding
 // (e.g. simple ASCII chars like `A`) or if the `char`
@@ -1561,8 +1544,7 @@ mod tests {
     use crate::tokenizer::AsciiFoldingFilter;
     use crate::tokenizer::RawTokenizer;
     use crate::tokenizer::SimpleTokenizer;
-    use crate::tokenizer::TokenStream;
+    use crate::tokenizer::TextAnalyzer;
-    use crate::tokenizer::Tokenizer;
     use std::iter;
 
     #[test]
@@ -1579,7 +1561,7 @@ mod tests {
 
     fn folding_helper(text: &str) -> Vec<String> {
         let mut tokens = Vec::new();
-        SimpleTokenizer
+        TextAnalyzer::from(SimpleTokenizer)
             .filter(AsciiFoldingFilter)
             .token_stream(text)
             .process(&mut |token| {
@@ -1589,7 +1571,9 @@ mod tests {
     }
 
     fn folding_using_raw_tokenizer_helper(text: &str) -> String {
-        let mut token_stream = RawTokenizer.filter(AsciiFoldingFilter).token_stream(text);
+        let mut token_stream = TextAnalyzer::from(RawTokenizer)
+            .filter(AsciiFoldingFilter)
+            .token_stream(text);
         token_stream.advance();
         token_stream.token().text.clone()
     }

src/tokenizer/facet_tokenizer.rs

@@ -1,4 +1,4 @@
-use super::{Token, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
 use crate::schema::FACET_SEP_BYTE;
 
 /// The `FacetTokenizer` process a `Facet` binary representation
@@ -25,15 +25,14 @@ pub struct FacetTokenStream<'a> {
     token: Token,
 }
 
-impl<'a> Tokenizer<'a> for FacetTokenizer {
+impl Tokenizer for FacetTokenizer {
-    type TokenStreamImpl = FacetTokenStream<'a>;
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl {
         FacetTokenStream {
             text,
             state: State::RootFacetNotEmitted, //< pos is the first char that has not been processed yet.
             token: Token::default(),
         }
+        .into()
     }
 }
 
@@ -84,7 +83,7 @@ mod tests {
 
     use super::FacetTokenizer;
     use crate::schema::Facet;
-    use crate::tokenizer::{Token, TokenStream, Tokenizer};
+    use crate::tokenizer::{Token, Tokenizer};
 
     #[test]
     fn test_facet_tokenizer() {

src/tokenizer/lower_caser.rs

@@ -1,24 +1,23 @@
 use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 use std::mem;
 
+impl TokenFilter for LowerCaser {
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(LowerCaserTokenStream {
+            tail: token_stream,
+            buffer: String::with_capacity(100),
+        })
+    }
+}
+
 /// Token filter that lowercase terms.
 #[derive(Clone)]
 pub struct LowerCaser;
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for LowerCaser
+pub struct LowerCaserTokenStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
-    type ResultTokenStream = LowerCaserTokenStream<TailTokenStream>;
-
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
-        LowerCaserTokenStream::wrap(token_stream)
-    }
-}
-
-pub struct LowerCaserTokenStream<TailTokenStream> {
     buffer: String,
-    tail: TailTokenStream,
+    tail: BoxTokenStream<'a>,
 }
 
 // writes a lowercased version of text into output.
@@ -31,18 +30,7 @@ fn to_lowercase_unicode(text: &mut String, output: &mut String) {
     }
 }
 
-impl<TailTokenStream> TokenStream for LowerCaserTokenStream<TailTokenStream>
+impl<'a> TokenStream for LowerCaserTokenStream<'a> {
-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 {
         if !self.tail.advance() {
             return false;
@@ -56,26 +44,19 @@ where
         }
         true
     }
-}
 
-impl<TailTokenStream> LowerCaserTokenStream<TailTokenStream>
+    fn token(&self) -> &Token {
-where
+        self.tail.token()
-    TailTokenStream: TokenStream,
+    }
-{
+
-    fn wrap(tail: TailTokenStream) -> LowerCaserTokenStream<TailTokenStream> {
+    fn token_mut(&mut self) -> &mut Token {
-        LowerCaserTokenStream {
+        self.tail.token_mut()
-            tail,
-            buffer: String::with_capacity(100),
-        }
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use crate::tokenizer::LowerCaser;
+    use crate::tokenizer::{LowerCaser, SimpleTokenizer, TextAnalyzer};
-    use crate::tokenizer::SimpleTokenizer;
-    use crate::tokenizer::TokenStream;
-    use crate::tokenizer::Tokenizer;
 
     #[test]
     fn test_to_lower_case() {
@@ -87,7 +68,9 @@ mod tests {
 
     fn lowercase_helper(text: &str) -> Vec<String> {
         let mut tokens = vec![];
-        let mut token_stream = SimpleTokenizer.filter(LowerCaser).token_stream(text);
+        let mut token_stream = TextAnalyzer::from(SimpleTokenizer)
+            .filter(LowerCaser)
+            .token_stream(text);
         while token_stream.advance() {
             let token_text = token_stream.token().text.clone();
             tokens.push(token_text);

src/tokenizer/mod.rs

@@ -64,7 +64,7 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let en_stem = SimpleTokenizer
+//! let en_stem = TextAnalyzer::from(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
 //!     .filter(LowerCaser)
 //!     .filter(Stemmer::new(Language::English));
@@ -109,7 +109,7 @@
 //! let index = Index::create_in_ram(schema);
 //!
 //! // We need to register our tokenizer :
-//! let custom_en_tokenizer = SimpleTokenizer
+//! let custom_en_tokenizer = TextAnalyzer::from(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
 //!     .filter(LowerCaser);
 //! index
@@ -143,10 +143,11 @@ pub use self::simple_tokenizer::SimpleTokenizer;
 pub use self::stemmer::{Language, Stemmer};
 pub use self::stop_word_filter::StopWordFilter;
 pub(crate) use self::token_stream_chain::TokenStreamChain;
-pub use self::tokenizer::BoxedTokenizer;
 
 pub use self::tokenized_string::{PreTokenizedStream, PreTokenizedString};
-pub use self::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
+pub use self::tokenizer::{
+    BoxTokenFilter, BoxTokenStream, TextAnalyzer, Token, TokenFilter, TokenStream, Tokenizer,
+};
 
 pub use self::tokenizer_manager::TokenizerManager;
 
@@ -160,9 +161,9 @@ pub const MAX_TOKEN_LEN: usize = u16::max_value() as usize - 4;
 #[cfg(test)]
 pub mod tests {
     use super::{
-        Language, LowerCaser, RemoveLongFilter, SimpleTokenizer, Stemmer, Token, Tokenizer,
+        Language, LowerCaser, RemoveLongFilter, SimpleTokenizer, Stemmer, Token, TokenizerManager,
-        TokenizerManager,
     };
+    use crate::tokenizer::TextAnalyzer;
 
     /// This is a function that can be used in tests and doc tests
     /// to assert a token's correctness.
@@ -229,7 +230,7 @@ pub mod tests {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "el_stem",
-            SimpleTokenizer
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser)
                 .filter(Stemmer::new(Language::Greek)),

src/tokenizer/ngram_tokenizer.rs

@@ -1,4 +1,5 @@
 use super::{Token, TokenStream, Tokenizer};
+use crate::tokenizer::BoxTokenStream;
 
 /// Tokenize the text by splitting words into n-grams of the given size(s)
 ///
@@ -129,11 +130,9 @@ pub struct NgramTokenStream<'a> {
     token: Token,
 }
 
-impl<'a> Tokenizer<'a> for NgramTokenizer {
+impl Tokenizer for NgramTokenizer {
-    type TokenStreamImpl = NgramTokenStream<'a>;
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-
+        From::from(NgramTokenStream {
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl {
-        NgramTokenStream {
             ngram_charidx_iterator: StutteringIterator::new(
                 CodepointFrontiers::for_str(text),
                 self.min_gram,
@@ -142,7 +141,7 @@ impl<'a> Tokenizer<'a> for NgramTokenizer {
             prefix_only: self.prefix_only,
             text,
             token: Token::default(),
-        }
+        })
     }
 }
 
@@ -308,10 +307,10 @@ mod tests {
     use super::NgramTokenizer;
     use super::StutteringIterator;
     use crate::tokenizer::tests::assert_token;
-    use crate::tokenizer::tokenizer::{TokenStream, Tokenizer};
+    use crate::tokenizer::tokenizer::Tokenizer;
-    use crate::tokenizer::Token;
+    use crate::tokenizer::{BoxTokenStream, Token};
 
-    fn test_helper<T: TokenStream>(mut tokenizer: T) -> Vec<Token> {
+    fn test_helper(mut tokenizer: BoxTokenStream) -> Vec<Token> {
         let mut tokens: Vec<Token> = vec![];
         tokenizer.process(&mut |token: &Token| tokens.push(token.clone()));
         tokens

src/tokenizer/raw_tokenizer.rs

@@ -1,4 +1,5 @@
 use super::{Token, TokenStream, Tokenizer};
+use crate::tokenizer::BoxTokenStream;
 
 /// For each value of the field, emit a single unprocessed token.
 #[derive(Clone)]
@@ -9,10 +10,8 @@ pub struct RawTokenStream {
     has_token: bool,
 }
 
-impl<'a> Tokenizer<'a> for RawTokenizer {
+impl Tokenizer for RawTokenizer {
-    type TokenStreamImpl = RawTokenStream;
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl {
         let token = Token {
             offset_from: 0,
             offset_to: text.len(),
@@ -24,6 +23,7 @@ impl<'a> Tokenizer<'a> for RawTokenizer {
             token,
             has_token: true,
         }
+        .into()
     }
 }
 

src/tokenizer/remove_long.rs

@@ -2,7 +2,7 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = SimpleTokenizer
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
 //!   .filter(RemoveLongFilter::limit(5));
 //!
 //! let mut stream = tokenizer.token_stream("toolong nice");
@@ -13,6 +13,7 @@
 //! ```
 //!
 use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 
 /// `RemoveLongFilter` removes tokens that are longer
 /// than a given number of bytes (in UTF-8 representation).
@@ -31,56 +32,27 @@ impl RemoveLongFilter {
     }
 }
 
-impl<TailTokenStream> RemoveLongFilterStream<TailTokenStream>
+impl<'a> RemoveLongFilterStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
     fn predicate(&self, token: &Token) -> bool {
         token.text.len() < self.token_length_limit
     }
+}
 
-    fn wrap(
+impl TokenFilter for RemoveLongFilter {
-        token_length_limit: usize,
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        tail: TailTokenStream,
+        BoxTokenStream::from(RemoveLongFilterStream {
-    ) -> RemoveLongFilterStream<TailTokenStream> {
+            token_length_limit: self.length_limit,
-        RemoveLongFilterStream {
+            tail: token_stream,
-            token_length_limit,
+        })
-            tail,
-        }
     }
 }
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for RemoveLongFilter
+pub struct RemoveLongFilterStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
-    type ResultTokenStream = RemoveLongFilterStream<TailTokenStream>;
-
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
-        RemoveLongFilterStream::wrap(self.length_limit, token_stream)
-    }
-}
-
-pub struct RemoveLongFilterStream<TailTokenStream>
-where
-    TailTokenStream: TokenStream,
-{
     token_length_limit: usize,
-    tail: TailTokenStream,
+    tail: BoxTokenStream<'a>,
 }
 
-impl<TailTokenStream> TokenStream for RemoveLongFilterStream<TailTokenStream>
+impl<'a> TokenStream for RemoveLongFilterStream<'a> {
-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 {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -89,4 +61,12 @@ where
         }
         false
     }
+
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
 }

src/tokenizer/simple_tokenizer.rs

@@ -1,3 +1,4 @@
+use super::BoxTokenStream;
 use super::{Token, TokenStream, Tokenizer};
 use std::str::CharIndices;
 
@@ -11,15 +12,13 @@ pub struct SimpleTokenStream<'a> {
     token: Token,
 }
 
-impl<'a> Tokenizer<'a> for SimpleTokenizer {
+impl Tokenizer for SimpleTokenizer {
-    type TokenStreamImpl = SimpleTokenStream<'a>;
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-
+        BoxTokenStream::from(SimpleTokenStream {
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl {
-        SimpleTokenStream {
             text,
             chars: text.char_indices(),
             token: Token::default(),
-        }
+        })
     }
 }
 

src/tokenizer/stemmer.rs

@@ -1,4 +1,5 @@
 use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 use rust_stemmers::{self, Algorithm};
 
 /// Available stemmer languages.
@@ -75,38 +76,22 @@ impl Default for Stemmer {
     }
 }
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for Stemmer
+impl TokenFilter for Stemmer {
-where
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-    TailTokenStream: TokenStream,
-{
-    type ResultTokenStream = StemmerTokenStream<TailTokenStream>;
-
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
         let inner_stemmer = rust_stemmers::Stemmer::create(self.stemmer_algorithm);
-        StemmerTokenStream::wrap(inner_stemmer, token_stream)
+        BoxTokenStream::from(StemmerTokenStream {
+            tail: token_stream,
+            stemmer: inner_stemmer,
+        })
     }
 }
 
-pub struct StemmerTokenStream<TailTokenStream>
+pub struct StemmerTokenStream<'a> {
-where
+    tail: BoxTokenStream<'a>,
-    TailTokenStream: TokenStream,
-{
-    tail: TailTokenStream,
     stemmer: rust_stemmers::Stemmer,
 }
 
-impl<TailTokenStream> TokenStream for StemmerTokenStream<TailTokenStream>
+impl<'a> TokenStream for StemmerTokenStream<'a> {
-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 {
         if !self.tail.advance() {
             return false;
@@ -117,16 +102,12 @@ where
         self.token_mut().text.push_str(&stemmed_str);
         true
     }
-}
 
-impl<TailTokenStream> StemmerTokenStream<TailTokenStream>
+    fn token(&self) -> &Token {
-where
+        self.tail.token()
-    TailTokenStream: TokenStream,
+    }
-{
+
-    fn wrap(
+    fn token_mut(&mut self) -> &mut Token {
-        stemmer: rust_stemmers::Stemmer,
+        self.tail.token_mut()
-        tail: TailTokenStream,
-    ) -> StemmerTokenStream<TailTokenStream> {
-        StemmerTokenStream { tail, stemmer }
     }
 }

src/tokenizer/stop_word_filter.rs

@@ -2,7 +2,7 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = SimpleTokenizer
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
 //!   .filter(StopWordFilter::remove(vec!["the".to_string(), "is".to_string()]));
 //!
 //! let mut stream = tokenizer.token_stream("the fox is crafty");
@@ -11,6 +11,7 @@
 //! assert!(stream.next().is_none());
 //! ```
 use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 use fnv::FnvHasher;
 use std::collections::HashSet;
 use std::hash::BuildHasherDefault;
@@ -48,53 +49,27 @@ impl StopWordFilter {
     }
 }
 
-pub struct StopWordFilterStream<TailTokenStream>
+pub struct StopWordFilterStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
     words: StopWordHashSet,
-    tail: TailTokenStream,
+    tail: BoxTokenStream<'a>,
 }
 
-impl<TailTokenStream> TokenFilter<TailTokenStream> for StopWordFilter
+impl TokenFilter for StopWordFilter {
-where
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-    TailTokenStream: TokenStream,
+        BoxTokenStream::from(StopWordFilterStream {
-{
+            words: self.words.clone(),
-    type ResultTokenStream = StopWordFilterStream<TailTokenStream>;
+            tail: token_stream,
-
+        })
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream {
-        StopWordFilterStream::wrap(self.words.clone(), token_stream)
     }
 }
 
-impl<TailTokenStream> StopWordFilterStream<TailTokenStream>
+impl<'a> StopWordFilterStream<'a> {
-where
-    TailTokenStream: TokenStream,
-{
     fn predicate(&self, token: &Token) -> bool {
         !self.words.contains(&token.text)
     }
-
-    fn wrap(
-        words: StopWordHashSet,
-        tail: TailTokenStream,
-    ) -> StopWordFilterStream<TailTokenStream> {
-        StopWordFilterStream { words, tail }
-    }
 }
 
-impl<TailTokenStream> TokenStream for StopWordFilterStream<TailTokenStream>
+impl<'a> TokenStream for StopWordFilterStream<'a> {
-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 {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -103,6 +78,14 @@ where
         }
         false
     }
+
+    fn token(&self) -> &Token {
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
 }
 
 impl Default for StopWordFilter {

src/tokenizer/token_stream_chain.rs

@@ -1,23 +1,21 @@
-use crate::tokenizer::{Token, TokenStream};
+use crate::tokenizer::{BoxTokenStream, Token, TokenStream};
+use std::ops::DerefMut;
 
 const POSITION_GAP: usize = 2;
 
-pub(crate) struct TokenStreamChain<TTokenStream: TokenStream> {
+pub(crate) struct TokenStreamChain<'a> {
     offsets: Vec<usize>,
-    token_streams: Vec<TTokenStream>,
+    token_streams: Vec<BoxTokenStream<'a>>,
     position_shift: usize,
     stream_idx: usize,
     token: Token,
 }
 
-impl<'a, TTokenStream> TokenStreamChain<TTokenStream>
+impl<'a> TokenStreamChain<'a> {
-where
-    TTokenStream: TokenStream,
-{
     pub fn new(
         offsets: Vec<usize>,
-        token_streams: Vec<TTokenStream>,
+        token_streams: Vec<BoxTokenStream<'a>>,
-    ) -> TokenStreamChain<TTokenStream> {
+    ) -> TokenStreamChain<'a> {
         TokenStreamChain {
             offsets,
             stream_idx: 0,
@@ -28,13 +26,10 @@ where
     }
 }
 
-impl<'a, TTokenStream> TokenStream for TokenStreamChain<TTokenStream>
+impl<'a> TokenStream for TokenStreamChain<'a> {
-where
-    TTokenStream: TokenStream,
-{
     fn advance(&mut self) -> bool {
         while self.stream_idx < self.token_streams.len() {
-            let token_stream = &mut self.token_streams[self.stream_idx];
+            let token_stream = self.token_streams[self.stream_idx].deref_mut();
             if token_stream.advance() {
                 let token = token_stream.token();
                 let offset_offset = self.offsets[self.stream_idx];

src/tokenizer/tokenized_string.rs

@@ -1,4 +1,4 @@
-use crate::tokenizer::{Token, TokenStream, TokenStreamChain};
+use crate::tokenizer::{BoxTokenStream, Token, TokenStream, TokenStreamChain};
 use std::cmp::Ordering;
 
 /// Struct representing pre-tokenized text
@@ -41,9 +41,9 @@ impl PreTokenizedStream {
     /// Creates a TokenStream from PreTokenizedString array
     pub fn chain_tokenized_strings<'a>(
         tok_strings: &'a [&'a PreTokenizedString],
-    ) -> Box<dyn TokenStream + 'a> {
+    ) -> BoxTokenStream {
         if tok_strings.len() == 1 {
-            Box::new(PreTokenizedStream::from((*tok_strings[0]).clone()))
+            PreTokenizedStream::from((*tok_strings[0]).clone()).into()
         } else {
             let mut offsets = vec![];
             let mut total_offset = 0;
@@ -53,11 +53,12 @@ impl PreTokenizedStream {
                     total_offset += last_token.offset_to;
                 }
             }
-            let token_streams: Vec<_> = tok_strings
+            // TODO remove the string cloning.
+            let token_streams: Vec<BoxTokenStream<'static>> = tok_strings
                 .iter()
-                .map(|tok_string| PreTokenizedStream::from((*tok_string).clone()))
+                .map(|&tok_string| PreTokenizedStream::from((*tok_string).clone()).into())
                 .collect();
-            Box::new(TokenStreamChain::new(offsets, token_streams))
+            TokenStreamChain::new(offsets, token_streams).into()
         }
     }
 }

src/tokenizer/tokenizer.rs

@@ -2,6 +2,7 @@ use crate::tokenizer::TokenStreamChain;
 /// The tokenizer module contains all of the tools used to process
 /// text in `tantivy`.
 use std::borrow::{Borrow, BorrowMut};
+use std::ops::{Deref, DerefMut};
 
 /// Token
 #[derive(Debug, Clone, Serialize, Deserialize, Eq, PartialEq)]
@@ -33,20 +34,31 @@ impl Default for Token {
     }
 }
 
-/// `Tokenizer` are in charge of splitting text into a stream of token
+/// `TextAnalyzer` tokenizes an input text into tokens and modifies the resulting `TokenStream`.
-/// before indexing.
 ///
-/// See the [module documentation](./index.html) for more detail.
+/// It simply wraps a `Tokenizer` and a list of `TokenFilter` that are applied sequentially.
-///
+pub struct TextAnalyzer {
-/// # Warning
+    tokenizer: Box<dyn Tokenizer>,
-///
+    token_filters: Vec<BoxTokenFilter>,
-/// This API may change to use associated types.
+}
-pub trait Tokenizer<'a>: Sized + Clone {
-    /// Type associated to the resulting tokenstream tokenstream.
-    type TokenStreamImpl: TokenStream;
 
-    /// Creates a token stream for a given `str`.
+impl<T: Tokenizer> From<T> for TextAnalyzer {
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl;
+    fn from(tokenizer: T) -> Self {
+        TextAnalyzer::new(tokenizer, Vec::new())
+    }
+}
+
+impl TextAnalyzer {
+    /// Creates a new `TextAnalyzer` given a tokenizer and a vector of `BoxTokenFilter`.
+    ///
+    /// When creating a `TextAnalyzer` from a `Tokenizer` alone, prefer using
+    /// `TextAnalyzer::from(tokenizer)`.
+    pub fn new<T: Tokenizer>(tokenizer: T, token_filters: Vec<BoxTokenFilter>) -> TextAnalyzer {
+        TextAnalyzer {
+            tokenizer: Box::new(tokenizer),
+            token_filters,
+        }
+    }
 
     /// Appends a token filter to the current tokenizer.
     ///
@@ -58,90 +70,26 @@ pub trait Tokenizer<'a>: Sized + Clone {
     /// ```rust
     /// use tantivy::tokenizer::*;
     ///
-    /// let en_stem = SimpleTokenizer
+    /// let en_stem = TextAnalyzer::from(SimpleTokenizer)
     ///     .filter(RemoveLongFilter::limit(40))
     ///     .filter(LowerCaser)
     ///     .filter(Stemmer::default());
     /// ```
     ///
-    fn filter<NewFilter>(self, new_filter: NewFilter) -> ChainTokenizer<NewFilter, Self>
+    pub fn filter<F: Into<BoxTokenFilter>>(mut self, token_filter: F) -> Self {
-    where
+        self.token_filters.push(token_filter.into());
-        NewFilter: TokenFilter<<Self as Tokenizer<'a>>::TokenStreamImpl>,
+        self
-    {
-        ChainTokenizer {
-            head: new_filter,
-            tail: self,
-        }
-    }
-}
-
-/// A boxed tokenizer
-trait BoxedTokenizerTrait: Send + Sync {
-    /// Tokenize a `&str`
-    fn token_stream<'a>(&self, text: &'a str) -> Box<dyn TokenStream + 'a>;
-
-    /// Tokenize an array`&str`
-    ///
-    /// The resulting `TokenStream` is equivalent to what would be obtained if the &str were
-    /// one concatenated `&str`, with an artificial position gap of `2` between the different fields
-    /// to prevent accidental `PhraseQuery` to match accross two terms.
-    fn token_stream_texts<'b>(&self, texts: &'b [&'b str]) -> Box<dyn TokenStream + 'b>;
-
-    /// Return a boxed clone of the tokenizer
-    fn boxed_clone(&self) -> BoxedTokenizer;
-}
-
-/// A boxed tokenizer
-pub struct BoxedTokenizer(Box<dyn BoxedTokenizerTrait>);
-
-impl<T> From<T> for BoxedTokenizer
-where
-    T: 'static + Send + Sync + for<'a> Tokenizer<'a>,
-{
-    fn from(tokenizer: T) -> BoxedTokenizer {
-        BoxedTokenizer(Box::new(BoxableTokenizer(tokenizer)))
-    }
-}
-
-impl BoxedTokenizer {
-    /// Tokenize a `&str`
-    pub fn token_stream<'a>(&self, text: &'a str) -> Box<dyn TokenStream + 'a> {
-        self.0.token_stream(text)
     }
 
     /// Tokenize an array`&str`
     ///
-    /// The resulting `TokenStream` is equivalent to what would be obtained if the &str were
+    /// The resulting `BoxTokenStream` is equivalent to what would be obtained if the &str were
     /// one concatenated `&str`, with an artificial position gap of `2` between the different fields
     /// to prevent accidental `PhraseQuery` to match accross two terms.
-    pub fn token_stream_texts<'b>(&self, texts: &'b [&'b str]) -> Box<dyn TokenStream + 'b> {
+    pub fn token_stream_texts<'a>(&self, texts: &'a [&'a str]) -> BoxTokenStream<'a> {
-        self.0.token_stream_texts(texts)
-    }
-}
-
-impl Clone for BoxedTokenizer {
-    fn clone(&self) -> BoxedTokenizer {
-        self.0.boxed_clone()
-    }
-}
-
-#[derive(Clone)]
-struct BoxableTokenizer<A>(A)
-where
-    A: for<'a> Tokenizer<'a> + Send + Sync;
-
-impl<A> BoxedTokenizerTrait for BoxableTokenizer<A>
-where
-    A: 'static + Send + Sync + for<'a> Tokenizer<'a>,
-{
-    fn token_stream<'a>(&self, text: &'a str) -> Box<dyn TokenStream + 'a> {
-        Box::new(self.0.token_stream(text))
-    }
-
-    fn token_stream_texts<'b>(&self, texts: &'b [&'b str]) -> Box<dyn TokenStream + 'b> {
         assert!(!texts.is_empty());
         if texts.len() == 1 {
-            Box::new(self.0.token_stream(texts[0]))
+            self.token_stream(texts[0])
         } else {
             let mut offsets = vec![];
             let mut total_offset = 0;
@@ -149,34 +97,124 @@ where
                 offsets.push(total_offset);
                 total_offset += text.len();
             }
-            let token_streams: Vec<_> =
+            let token_streams: Vec<BoxTokenStream<'a>> = texts
-                texts.iter().map(|text| self.0.token_stream(text)).collect();
+                .iter()
-            Box::new(TokenStreamChain::new(offsets, token_streams))
+                .cloned()
+                .map(|text| self.token_stream(text))
+                .collect();
+            From::from(TokenStreamChain::new(offsets, token_streams))
         }
     }
 
-    fn boxed_clone(&self) -> BoxedTokenizer {
+    /// Creates a token stream for a given `str`.
-        self.0.clone().into()
+    pub fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+        let mut token_stream = self.tokenizer.token_stream(text);
+        for token_filter in &self.token_filters {
+            token_stream = token_filter.transform(token_stream);
+        }
+        token_stream
     }
 }
 
-impl<'b> TokenStream for Box<dyn TokenStream + 'b> {
+impl Clone for TextAnalyzer {
+    fn clone(&self) -> Self {
+        TextAnalyzer {
+            tokenizer: self.tokenizer.box_clone(),
+            token_filters: self
+                .token_filters
+                .iter()
+                .map(|token_filter| token_filter.box_clone())
+                .collect(),
+        }
+    }
+}
+
+/// `Tokenizer` are in charge of splitting text into a stream of token
+/// before indexing.
+///
+/// See the [module documentation](./index.html) for more detail.
+///
+/// # Warning
+///
+/// This API may change to use associated types.
+pub trait Tokenizer: 'static + Send + Sync + TokenizerClone {
+    /// Creates a token stream for a given `str`.
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a>;
+}
+
+pub trait TokenizerClone {
+    fn box_clone(&self) -> Box<dyn Tokenizer>;
+}
+
+impl<T: Tokenizer + Clone> TokenizerClone for T {
+    fn box_clone(&self) -> Box<dyn Tokenizer> {
+        Box::new(self.clone())
+    }
+}
+
+impl<'a> TokenStream for Box<dyn TokenStream + 'a> {
     fn advance(&mut self) -> bool {
         let token_stream: &mut dyn TokenStream = self.borrow_mut();
         token_stream.advance()
     }
 
-    fn token(&self) -> &Token {
+    fn token<'b>(&'b self) -> &'b Token {
-        let token_stream: &dyn TokenStream = self.borrow();
+        let token_stream: &'b (dyn TokenStream + 'a) = self.borrow();
         token_stream.token()
     }
 
-    fn token_mut(&mut self) -> &mut Token {
+    fn token_mut<'b>(&'b mut self) -> &'b mut Token {
-        let token_stream: &mut dyn TokenStream = self.borrow_mut();
+        let token_stream: &'b mut (dyn TokenStream + 'a) = self.borrow_mut();
         token_stream.token_mut()
     }
 }
 
+/// Simple wrapper of `Box<dyn TokenStream + 'a>`.
+///
+/// See `TokenStream` for more information.
+pub struct BoxTokenStream<'a>(Box<dyn TokenStream + 'a>);
+
+impl<'a, T> From<T> for BoxTokenStream<'a>
+where
+    T: TokenStream + 'a,
+{
+    fn from(token_stream: T) -> BoxTokenStream<'a> {
+        BoxTokenStream(Box::new(token_stream))
+    }
+}
+
+impl<'a> Deref for BoxTokenStream<'a> {
+    type Target = dyn TokenStream + 'a;
+
+    fn deref(&self) -> &Self::Target {
+        &*self.0
+    }
+}
+impl<'a> DerefMut for BoxTokenStream<'a> {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        &mut *self.0
+    }
+}
+
+/// Simple wrapper of `Box<dyn TokenFilter + 'a>`.
+///
+/// See `TokenStream` for more information.
+pub struct BoxTokenFilter(Box<dyn TokenFilter>);
+
+impl Deref for BoxTokenFilter {
+    type Target = dyn TokenFilter;
+
+    fn deref(&self) -> &dyn TokenFilter {
+        &*self.0
+    }
+}
+
+impl<T: TokenFilter> From<T> for BoxTokenFilter {
+    fn from(tokenizer: T) -> BoxTokenFilter {
+        BoxTokenFilter(Box::new(tokenizer))
+    }
+}
+
 /// `TokenStream` is the result of the tokenization.
 ///
 /// It consists consumable stream of `Token`s.
@@ -186,7 +224,7 @@ impl<'b> TokenStream for Box<dyn TokenStream + 'b> {
 /// ```
 /// use tantivy::tokenizer::*;
 ///
-/// let tokenizer = SimpleTokenizer
+/// let tokenizer = TextAnalyzer::from(SimpleTokenizer)
 ///        .filter(RemoveLongFilter::limit(40))
 ///        .filter(LowerCaser);
 /// let mut token_stream = tokenizer.token_stream("Hello, happy tax payer");
@@ -225,7 +263,7 @@ pub trait TokenStream {
     /// ```
     /// use tantivy::tokenizer::*;
     ///
-    /// let tokenizer = SimpleTokenizer
+    /// let tokenizer = TextAnalyzer::from(SimpleTokenizer)
     ///       .filter(RemoveLongFilter::limit(40))
     ///       .filter(LowerCaser);
     /// let mut token_stream = tokenizer.token_stream("Hello, happy tax payer");
@@ -243,6 +281,8 @@ pub trait TokenStream {
 
     /// Helper function to consume the entire `TokenStream`
     /// and push the tokens to a sink function.
+    ///
+    /// Remove this.
     fn process(&mut self, sink: &mut dyn FnMut(&Token)) -> u32 {
         let mut num_tokens_pushed = 0u32;
         while self.advance() {
@@ -253,33 +293,20 @@ pub trait TokenStream {
     }
 }
 
-#[derive(Clone)]
+pub trait TokenFilterClone {
-pub struct ChainTokenizer<HeadTokenFilterFactory, TailTokenizer> {
+    fn box_clone(&self) -> BoxTokenFilter;
-    head: HeadTokenFilterFactory,
-    tail: TailTokenizer,
-}
-
-impl<'a, HeadTokenFilterFactory, TailTokenizer> Tokenizer<'a>
-    for ChainTokenizer<HeadTokenFilterFactory, TailTokenizer>
-where
-    HeadTokenFilterFactory: TokenFilter<TailTokenizer::TokenStreamImpl>,
-    TailTokenizer: Tokenizer<'a>,
-{
-    type TokenStreamImpl = HeadTokenFilterFactory::ResultTokenStream;
-
-    fn token_stream(&self, text: &'a str) -> Self::TokenStreamImpl {
-        let tail_token_stream = self.tail.token_stream(text);
-        self.head.transform(tail_token_stream)
-    }
 }
 
 /// Trait for the pluggable components of `Tokenizer`s.
-pub trait TokenFilter<TailTokenStream: TokenStream>: Clone {
+pub trait TokenFilter: 'static + Send + Sync + TokenFilterClone {
-    /// The resulting `TokenStream` type.
-    type ResultTokenStream: TokenStream;
-
     /// Wraps a token stream and returns the modified one.
-    fn transform(&self, token_stream: TailTokenStream) -> Self::ResultTokenStream;
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a>;
+}
+
+impl<T: TokenFilter + Clone> TokenFilterClone for T {
+    fn box_clone(&self) -> BoxTokenFilter {
+        BoxTokenFilter::from(self.clone())
+    }
 }
 
 #[cfg(test)]

src/tokenizer/tokenizer_manager.rs

@@ -1,11 +1,10 @@
 use crate::tokenizer::stemmer::Language;
-use crate::tokenizer::BoxedTokenizer;
+use crate::tokenizer::tokenizer::TextAnalyzer;
 use crate::tokenizer::LowerCaser;
 use crate::tokenizer::RawTokenizer;
 use crate::tokenizer::RemoveLongFilter;
 use crate::tokenizer::SimpleTokenizer;
 use crate::tokenizer::Stemmer;
-use crate::tokenizer::Tokenizer;
 use std::collections::HashMap;
 use std::sync::{Arc, RwLock};
 
@@ -23,16 +22,16 @@ use std::sync::{Arc, RwLock};
 ///  search engine.
 #[derive(Clone)]
 pub struct TokenizerManager {
-    tokenizers: Arc<RwLock<HashMap<String, BoxedTokenizer>>>,
+    tokenizers: Arc<RwLock<HashMap<String, TextAnalyzer>>>,
 }
 
 impl TokenizerManager {
     /// Registers a new tokenizer associated with a given name.
-    pub fn register<A>(&self, tokenizer_name: &str, tokenizer: A)
+    pub fn register<T>(&self, tokenizer_name: &str, tokenizer: T)
     where
-        A: Into<BoxedTokenizer>,
+        TextAnalyzer: From<T>,
     {
-        let boxed_tokenizer = tokenizer.into();
+        let boxed_tokenizer: TextAnalyzer = TextAnalyzer::from(tokenizer);
         self.tokenizers
             .write()
             .expect("Acquiring the lock should never fail")
@@ -40,7 +39,7 @@ impl TokenizerManager {
     }
 
     /// Accessing a tokenizer given its name.
-    pub fn get(&self, tokenizer_name: &str) -> Option<BoxedTokenizer> {
+    pub fn get(&self, tokenizer_name: &str) -> Option<TextAnalyzer> {
         self.tokenizers
             .read()
             .expect("Acquiring the lock should never fail")
@@ -62,13 +61,13 @@ impl Default for TokenizerManager {
         manager.register("raw", RawTokenizer);
         manager.register(
             "default",
-            SimpleTokenizer
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser),
         );
         manager.register(
             "en_stem",
-            SimpleTokenizer
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser)
                 .filter(Stemmer::new(Language::English)),