Adapted benchmark to the analyzer's Iterator<Item=Token> form

benches/analyzer.rs

@@ -10,7 +10,7 @@ pub fn criterion_benchmark(c: &mut Criterion) {
         b.iter(|| {
             let mut word_count = 0;
             let mut token_stream = tokenizer.token_stream(ALICE_TXT);
-            while token_stream.advance() {
+            for token in token_stream {
                 word_count += 1;
             }
             assert_eq!(word_count, 30_731);

examples/pre_tokenized_text.rs

@@ -17,12 +17,7 @@ use tantivy::{doc, Index, ReloadPolicy};
 use tempfile::TempDir;
 
 fn pre_tokenize_text(text: &str) -> Vec<Token> {
-    let mut token_stream = SimpleTokenizer.token_stream(text);
-    let mut tokens = vec![];
-    while token_stream.advance() {
-        tokens.push(token_stream.token().clone());
-    }
-    tokens
+    SimpleTokenizer.token_stream(text).collect()
 }
 
 fn main() -> tantivy::Result<()> {

examples/snippet.rs

@@ -51,7 +51,7 @@ fn main() -> tantivy::Result<()> {
 
     let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
 
-    let snippet_generator = SnippetGenerator::create(&searcher, &*query, body)?;
+    let mut snippet_generator = SnippetGenerator::create(&searcher, &*query, body)?;
 
     for (score, doc_address) in top_docs {
         let doc = searcher.doc(doc_address)?;

examples/stop_words.rs

@@ -50,12 +50,13 @@ 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 = TextAnalyzer::from(SimpleTokenizer)
-        .filter(LowerCaser)
+    let tokenizer = analyzer_builder(SimpleTokenizer)
+        .filter(LowerCaser::new())
         .filter(StopWordFilter::remove(vec![
             "the".to_string(),
             "and".to_string(),
-        ]));
+        ]))
+        .build();
 
     index.tokenizers().register("stoppy", tokenizer);
 

src/core/index.rs

@@ -20,7 +20,7 @@ use crate::reader::IndexReaderBuilder;
 use crate::schema::Field;
 use crate::schema::FieldType;
 use crate::schema::Schema;
-use crate::tokenizer::{TextAnalyzer, TokenizerManager};
+use crate::tokenizer::{TextAnalyzerT, TokenizerManager};
 use crate::IndexWriter;
 use std::collections::HashSet;
 use std::fmt;
@@ -119,13 +119,12 @@ impl Index {
             return Index::create(dir, schema);
         }
         let index = Index::open(dir)?;
-        if index.schema() == schema {
-            Ok(index)
-        } else {
-            Err(TantivyError::SchemaError(
+        if index.schema() != schema {
+            return Err(TantivyError::SchemaError(
                 "An index exists but the schema does not match.".to_string(),
-            ))
+            ));
         }
+        Ok(index)
     }
 
     /// Creates a new index in a temp directory.
@@ -181,11 +180,11 @@ impl Index {
     }
 
     /// Helper to access the tokenizer associated to a specific field.
-    pub fn tokenizer_for_field(&self, field: Field) -> crate::Result<TextAnalyzer> {
+    pub fn tokenizer_for_field(&self, field: Field) -> crate::Result<Box<dyn TextAnalyzerT>> {
         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<TextAnalyzer> = match field_type {
+        let tokenizer_name_opt: Option<Box<dyn TextAnalyzerT>> = match field_type {
             FieldType::Str(text_options) => text_options
                 .get_indexing_options()
                 .map(|text_indexing_options| text_indexing_options.tokenizer().to_string())

src/core/segment_reader.rs

@@ -310,7 +310,7 @@ impl SegmentReader {
     }
 
     /// Returns an iterator that will iterate over the alive document ids
-    pub fn doc_ids_alive<'a>(&'a self) -> impl Iterator<Item = DocId> + 'a {
+    pub fn doc_ids_alive(&self) -> impl Iterator<Item = DocId> + '_ {
         (0u32..self.max_doc).filter(move |doc| !self.is_deleted(*doc))
     }
 

src/indexer/segment_writer.rs

@@ -10,10 +10,9 @@ use crate::schema::FieldType;
 use crate::schema::Schema;
 use crate::schema::Term;
 use crate::schema::Value;
-use crate::schema::{Field, FieldEntry};
-use crate::tokenizer::{BoxTokenStream, PreTokenizedStream};
-use crate::tokenizer::{FacetTokenizer, TextAnalyzer};
-use crate::tokenizer::{TokenStreamChain, Tokenizer};
+use crate::tokenizer::PreTokenizedStream;
+use crate::tokenizer::{DynTokenStreamChain, Tokenizer};
+use crate::tokenizer::{FacetTokenizer, TextAnalyzerT, Token};
 use crate::Opstamp;
 use crate::{DocId, SegmentComponent};
 
@@ -23,7 +22,7 @@ use crate::{DocId, SegmentComponent};
 fn initial_table_size(per_thread_memory_budget: usize) -> crate::Result<usize> {
     let table_memory_upper_bound = per_thread_memory_budget / 3;
     if let Some(limit) = (10..)
-        .take_while(|num_bits: &usize| compute_table_size(*num_bits) < table_memory_upper_bound)
+        .take_while(|&num_bits| compute_table_size(num_bits) < table_memory_upper_bound)
         .last()
     {
         Ok(limit.min(19)) // we cap it at 2^19 = 512K.
@@ -45,7 +44,8 @@ pub struct SegmentWriter {
     fast_field_writers: FastFieldsWriter,
     fieldnorms_writer: FieldNormsWriter,
     doc_opstamps: Vec<Opstamp>,
-    tokenizers: Vec<Option<TextAnalyzer>>,
+    // TODO: change type
+    tokenizers: Vec<Option<Box<dyn TextAnalyzerT>>>,
     term_buffer: Term,
 }
 
@@ -70,17 +70,17 @@ impl SegmentWriter {
         let multifield_postings = MultiFieldPostingsWriter::new(schema, table_num_bits);
         let tokenizers = schema
             .fields()
-            .map(
-                |(_, field_entry): (Field, &FieldEntry)| match field_entry.field_type() {
-                    FieldType::Str(ref text_options) => text_options
+            .map(|(_, field_entry)| match field_entry.field_type() {
+                FieldType::Str(text_options) => {
+                    text_options
                         .get_indexing_options()
                         .and_then(|text_index_option| {
                             let tokenizer_name = &text_index_option.tokenizer();
                             tokenizer_manager.get(tokenizer_name)
-                        }),
-                    _ => None,
-                },
-            )
+                        })
+                }
+                _ => None,
+            })
             .collect();
         Ok(SegmentWriter {
             max_doc: 0,
@@ -141,13 +141,13 @@ impl SegmentWriter {
             }
             let (term_buffer, multifield_postings) =
                 (&mut self.term_buffer, &mut self.multifield_postings);
-            match *field_entry.field_type() {
+            match field_entry.field_type() {
                 FieldType::HierarchicalFacet => {
                     term_buffer.set_field(field);
                     let facets =
                         field_values
                             .iter()
-                            .flat_map(|field_value| match *field_value.value() {
+                            .flat_map(|field_value| match field_value.value() {
                                 Value::Facet(ref facet) => Some(facet.encoded_str()),
                                 _ => {
                                     panic!("Expected hierarchical facet");
@@ -157,12 +157,13 @@ impl SegmentWriter {
                         let mut unordered_term_id_opt = None;
                         FacetTokenizer
                             .token_stream(facet_str)
-                            .process(&mut |token| {
+                            .map(|token| {
                                 term_buffer.set_text(&token.text);
                                 let unordered_term_id =
                                     multifield_postings.subscribe(doc_id, &term_buffer);
                                 unordered_term_id_opt = Some(unordered_term_id);
-                            });
+                            })
+                            .count();
                         if let Some(unordered_term_id) = unordered_term_id_opt {
                             self.fast_field_writers
                                 .get_multivalue_writer(field)
@@ -172,37 +173,38 @@ impl SegmentWriter {
                     }
                 }
                 FieldType::Str(_) => {
-                    let mut token_streams: Vec<BoxTokenStream> = vec![];
-                    let mut offsets = vec![];
+                    let mut streams_with_offsets = vec![];
                     let mut total_offset = 0;
 
                     for field_value in field_values {
                         match field_value.value() {
                             Value::PreTokStr(tok_str) => {
-                                offsets.push(total_offset);
+                                streams_with_offsets.push((
+                                    Box::new(PreTokenizedStream::from(tok_str.clone()))
+                                        as Box<dyn Iterator<Item = Token>>,
+                                    total_offset,
+                                ));
                                 if let Some(last_token) = tok_str.tokens.last() {
                                     total_offset += last_token.offset_to;
                                 }
-                                token_streams
-                                    .push(PreTokenizedStream::from(tok_str.clone()).into());
                             }
-                            Value::Str(ref text) => {
+                            Value::Str(text) => {
                                 if let Some(ref mut tokenizer) =
                                     self.tokenizers[field.field_id() as usize]
                                 {
-                                    offsets.push(total_offset);
+                                    streams_with_offsets
+                                        .push((tokenizer.token_stream(text), total_offset));
                                     total_offset += text.len();
-                                    token_streams.push(tokenizer.token_stream(text));
                                 }
                             }
                             _ => (),
                         }
                     }
 
-                    let num_tokens = if token_streams.is_empty() {
+                    let num_tokens = if streams_with_offsets.is_empty() {
                         0
                     } else {
-                        let mut token_stream = TokenStreamChain::new(offsets, token_streams);
+                        let mut token_stream = DynTokenStreamChain::from_vec(streams_with_offsets);
                         multifield_postings.index_text(
                             doc_id,
                             field,
@@ -213,71 +215,62 @@ impl SegmentWriter {
 
                     self.fieldnorms_writer.record(doc_id, field, num_tokens);
                 }
-                FieldType::U64(ref int_option) => {
-                    if int_option.is_indexed() {
-                        for field_value in field_values {
-                            term_buffer.set_field(field_value.field());
-                            let u64_val = field_value
-                                .value()
-                                .u64_value()
-                                .ok_or_else(make_schema_error)?;
-                            term_buffer.set_u64(u64_val);
-                            multifield_postings.subscribe(doc_id, &term_buffer);
-                        }
+                FieldType::U64(int_option) if int_option.is_indexed() => {
+                    for field_value in field_values {
+                        term_buffer.set_field(field_value.field());
+                        let u64_val = field_value
+                            .value()
+                            .u64_value()
+                            .ok_or_else(make_schema_error)?;
+                        term_buffer.set_u64(u64_val);
+                        multifield_postings.subscribe(doc_id, &term_buffer);
                     }
                 }
-                FieldType::Date(ref int_option) => {
-                    if int_option.is_indexed() {
-                        for field_value in field_values {
-                            term_buffer.set_field(field_value.field());
-                            let date_val = field_value
-                                .value()
-                                .date_value()
-                                .ok_or_else(make_schema_error)?;
-                            term_buffer.set_i64(date_val.timestamp());
-                            multifield_postings.subscribe(doc_id, &term_buffer);
-                        }
+                FieldType::Date(int_option) if int_option.is_indexed() => {
+                    for field_value in field_values {
+                        term_buffer.set_field(field_value.field());
+                        let date_val = field_value
+                            .value()
+                            .date_value()
+                            .ok_or_else(make_schema_error)?;
+                        term_buffer.set_i64(date_val.timestamp());
+                        multifield_postings.subscribe(doc_id, &term_buffer);
                     }
                 }
-                FieldType::I64(ref int_option) => {
-                    if int_option.is_indexed() {
-                        for field_value in field_values {
-                            term_buffer.set_field(field_value.field());
-                            let i64_val = field_value
-                                .value()
-                                .i64_value()
-                                .ok_or_else(make_schema_error)?;
-                            term_buffer.set_i64(i64_val);
-                            multifield_postings.subscribe(doc_id, &term_buffer);
-                        }
+                FieldType::I64(int_option) if int_option.is_indexed() => {
+                    for field_value in field_values {
+                        term_buffer.set_field(field_value.field());
+                        let i64_val = field_value
+                            .value()
+                            .i64_value()
+                            .ok_or_else(make_schema_error)?;
+                        term_buffer.set_i64(i64_val);
+                        multifield_postings.subscribe(doc_id, &term_buffer);
                     }
                 }
-                FieldType::F64(ref int_option) => {
-                    if int_option.is_indexed() {
-                        for field_value in field_values {
-                            term_buffer.set_field(field_value.field());
-                            let f64_val = field_value
-                                .value()
-                                .f64_value()
-                                .ok_or_else(make_schema_error)?;
-                            term_buffer.set_f64(f64_val);
-                            multifield_postings.subscribe(doc_id, &term_buffer);
-                        }
+                FieldType::F64(int_option) if int_option.is_indexed() => {
+                    for field_value in field_values {
+                        term_buffer.set_field(field_value.field());
+                        let f64_val = field_value
+                            .value()
+                            .f64_value()
+                            .ok_or_else(make_schema_error)?;
+                        term_buffer.set_f64(f64_val);
+                        multifield_postings.subscribe(doc_id, &term_buffer);
                     }
                 }
-                FieldType::Bytes(ref option) => {
-                    if option.is_indexed() {
-                        for field_value in field_values {
-                            term_buffer.set_field(field_value.field());
-                            let bytes = field_value
-                                .value()
-                                .bytes_value()
-                                .ok_or_else(make_schema_error)?;
-                            term_buffer.set_bytes(bytes);
-                            self.multifield_postings.subscribe(doc_id, &term_buffer);
-                        }
+                FieldType::Bytes(option) if option.is_indexed() => {
+                    for field_value in field_values {
+                        term_buffer.set_field(field_value.field());
+                        let bytes = field_value
+                            .value()
+                            .bytes_value()
+                            .ok_or_else(make_schema_error)?;
+                        term_buffer.set_bytes(bytes);
+                        self.multifield_postings.subscribe(doc_id, &term_buffer);
                     }
                 }
+                _ => {}
             }
         }
         doc.filter_fields(|field| schema.get_field_entry(field).is_stored());

src/positions/reader.rs

@@ -132,7 +132,7 @@ impl PositionReader {
             "offset arguments should be increasing."
         );
         let delta_to_block_offset = offset as i64 - self.block_offset as i64;
-        if delta_to_block_offset < 0 || delta_to_block_offset >= 128 {
+        if !(0..128).contains(&delta_to_block_offset) {
             // The first position is not within the first block.
             // We need to decompress the first block.
             let delta_to_anchor_offset = offset - self.anchor_offset;

src/postings/block_search.rs

@@ -109,9 +109,9 @@ impl BlockSearcher {
     /// The results should be equivalent to
     /// ```compile_fail
     /// block[..]
-    //       .iter()
-    //       .take_while(|&&val| val < target)
-    //       .count()
+    ///       .iter()
+    ///       .take_while(|&&val| val < target)
+    ///       .count()
     /// ```
     ///
     /// The `start` argument is just used to hint that the response is

src/postings/postings_writer.rs

@@ -9,7 +9,6 @@ use crate::postings::{FieldSerializer, InvertedIndexSerializer};
 use crate::schema::IndexRecordOption;
 use crate::schema::{Field, FieldEntry, FieldType, Schema, Term};
 use crate::termdict::TermOrdinal;
-use crate::tokenizer::TokenStream;
 use crate::tokenizer::{Token, MAX_TOKEN_LEN};
 use crate::DocId;
 use fnv::FnvHashMap;
@@ -100,12 +99,10 @@ impl MultiFieldPostingsWriter {
         &mut self,
         doc: DocId,
         field: Field,
-        token_stream: &mut dyn TokenStream,
+        token_stream: &mut dyn Iterator<Item = Token>,
         term_buffer: &mut Term,
     ) -> u32 {
-        let postings_writer =
-            self.per_field_postings_writers[field.field_id() as usize].deref_mut();
-        postings_writer.index_text(
+        self.per_field_postings_writers[field.field_id() as usize].index_text(
             &mut self.term_index,
             doc,
             field,
@@ -217,7 +214,7 @@ pub trait PostingsWriter {
         term_index: &mut TermHashMap,
         doc_id: DocId,
         field: Field,
-        token_stream: &mut dyn TokenStream,
+        token_stream: &mut dyn Iterator<Item = Token>,
         heap: &mut MemoryArena,
         term_buffer: &mut Term,
     ) -> u32 {
@@ -242,7 +239,7 @@ pub trait PostingsWriter {
                 );
             }
         };
-        token_stream.process(&mut sink)
+        token_stream.map(|tok| sink(&tok)).count() as u32
     }
 
     fn total_num_tokens(&self) -> u64;

src/query/query_parser/query_parser.rs

@@ -289,7 +289,7 @@ impl QueryParser {
             let field_name = field_entry.name().to_string();
             return Err(QueryParserError::FieldNotIndexed(field_name));
         }
-        match *field_type {
+        match field_type {
             FieldType::I64(_) => {
                 let val: i64 = i64::from_str(phrase)?;
                 let term = Term::from_field_i64(field, val);
@@ -312,7 +312,7 @@ impl QueryParser {
                 let term = Term::from_field_u64(field, val);
                 Ok(vec![(0, term)])
             }
-            FieldType::Str(ref str_options) => {
+            FieldType::Str(str_options) => {
                 if let Some(option) = str_options.get_indexing_options() {
                     let tokenizer =
                         self.tokenizer_manager
@@ -323,15 +323,14 @@ impl QueryParser {
                                     option.tokenizer().to_string(),
                                 )
                             })?;
-                    let mut terms: Vec<(usize, Term)> = Vec::new();
-                    let mut token_stream = tokenizer.token_stream(phrase);
-                    token_stream.process(&mut |token| {
-                        let term = Term::from_field_text(field, &token.text);
-                        terms.push((token.position, term));
-                    });
-                    if terms.is_empty() {
-                        Ok(vec![])
-                    } else if terms.len() == 1 {
+                    let token_stream = tokenizer.token_stream(phrase);
+                    let terms: Vec<_> = token_stream
+                        .map(|token| {
+                            let term = Term::from_field_text(field, &token.text);
+                            (token.position, term)
+                        })
+                        .collect();
+                    if terms.len() <= 1 {
                         Ok(terms)
                     } else {
                         let field_entry = self.schema.get_field_entry(field);
@@ -414,7 +413,7 @@ impl QueryParser {
         &self,
         given_field: &Option<String>,
     ) -> Result<Cow<'_, [Field]>, QueryParserError> {
-        match *given_field {
+        match given_field {
             None => {
                 if self.default_fields.is_empty() {
                     Err(QueryParserError::NoDefaultFieldDeclared)
@@ -422,7 +421,7 @@ impl QueryParser {
                     Ok(Cow::from(&self.default_fields[..]))
                 }
             }
-            Some(ref field) => Ok(Cow::from(vec![self.resolve_field_name(&*field)?])),
+            Some(field) => Ok(Cow::from(vec![self.resolve_field_name(&*field)?])),
         }
     }
 
@@ -574,15 +573,12 @@ fn convert_to_query(logical_ast: LogicalAST) -> Box<dyn Query> {
 #[cfg(test)]
 mod test {
     use super::super::logical_ast::*;
-    use super::QueryParser;
-    use super::QueryParserError;
+    use super::*;
     use crate::query::Query;
     use crate::schema::Field;
     use crate::schema::{IndexRecordOption, TextFieldIndexing, TextOptions};
     use crate::schema::{Schema, Term, INDEXED, STORED, STRING, TEXT};
-    use crate::tokenizer::{
-        LowerCaser, SimpleTokenizer, StopWordFilter, TextAnalyzer, TokenizerManager,
-    };
+    use crate::tokenizer::{analyzer_builder, LowerCaser, SimpleTokenizer, StopWordFilter};
     use crate::Index;
     use matches::assert_matches;
 
@@ -620,9 +616,10 @@ mod test {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "en_with_stop_words",
-            TextAnalyzer::from(SimpleTokenizer)
-                .filter(LowerCaser)
-                .filter(StopWordFilter::remove(vec!["the".to_string()])),
+            analyzer_builder(SimpleTokenizer)
+                .filter(LowerCaser::new())
+                .filter(StopWordFilter::remove(vec!["the".to_string()]))
+                .build(),
         );
         QueryParser::new(schema, default_fields, tokenizer_manager)
     }

src/snippet/mod.rs

@@ -1,7 +1,7 @@
 use crate::query::Query;
 use crate::schema::Field;
 use crate::schema::Value;
-use crate::tokenizer::{TextAnalyzer, Token};
+use crate::tokenizer::{TextAnalyzerT, Token};
 use crate::Searcher;
 use crate::{Document, Score};
 use htmlescape::encode_minimal;
@@ -139,9 +139,9 @@ 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: &TextAnalyzer,
-    text: &'a str,
+fn search_fragments(
+    tokenizer: &dyn TextAnalyzerT,
+    text: &str,
     terms: &BTreeMap<String, Score>,
     max_num_chars: usize,
 ) -> Vec<FragmentCandidate> {
@@ -155,7 +155,7 @@ fn search_fragments<'a>(
             };
             fragment = FragmentCandidate::new(next.offset_from);
         }
-        fragment.try_add_token(next, &terms);
+        fragment.try_add_token(&next, &terms);
     }
     if fragment.score > 0.0 {
         fragments.push(fragment)
@@ -249,7 +249,7 @@ fn select_best_fragment_combination(fragments: &[FragmentCandidate], text: &str)
 /// ```
 pub struct SnippetGenerator {
     terms_text: BTreeMap<String, Score>,
-    tokenizer: TextAnalyzer,
+    tokenizer: Box<dyn TextAnalyzerT>,
     field: Field,
     max_num_chars: usize,
 }
@@ -297,33 +297,37 @@ impl SnippetGenerator {
     ///
     /// This method extract the text associated to the `SnippetGenerator`'s field
     /// and computes a snippet.
-    pub fn snippet_from_doc(&self, doc: &Document) -> Snippet {
+    pub fn snippet_from_doc(&mut self, doc: &Document) -> Snippet {
         let text: String = doc
             .get_all(self.field)
             .flat_map(Value::text)
             .collect::<Vec<&str>>()
             .join(" ");
-        self.snippet(&text)
+        self.snippet(text.as_ref())
     }
 
     /// Generates a snippet for the given text.
-    pub fn snippet(&self, text: &str) -> Snippet {
-        let fragment_candidates =
-            search_fragments(&self.tokenizer, &text, &self.terms_text, self.max_num_chars);
-        select_best_fragment_combination(&fragment_candidates[..], &text)
+    pub fn snippet(&mut self, text: &str) -> Snippet {
+        let fragment_candidates = search_fragments(
+            &mut *self.tokenizer,
+            text,
+            &self.terms_text,
+            self.max_num_chars,
+        );
+        select_best_fragment_combination(&fragment_candidates[..], text)
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use super::{search_fragments, select_best_fragment_combination};
+    use super::*;
     use crate::query::QueryParser;
     use crate::schema::{IndexRecordOption, Schema, TextFieldIndexing, TextOptions, TEXT};
     use crate::tokenizer::SimpleTokenizer;
+    use crate::tokenizer::TextAnalyzer;
     use crate::Index;
     use crate::SnippetGenerator;
     use maplit::btreemap;
-    use std::collections::BTreeMap;
     use std::iter::Iterator;
 
     const TEST_TEXT: &'static str = r#"Rust is a systems programming language sponsored by
@@ -346,7 +350,13 @@ Survey in 2016, 2017, and 2018."#;
             String::from("rust") => 1.0,
             String::from("language") => 0.9
         };
-        let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 100);
+
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            TEST_TEXT,
+            &terms,
+            100,
+        );
         assert_eq!(fragments.len(), 7);
         {
             let first = &fragments[0];
@@ -373,7 +383,12 @@ Survey in 2016, 2017, and 2018."#;
                 String::from("rust") =>1.0,
                 String::from("language") => 0.9
             };
-            let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
+            let fragments = search_fragments(
+                &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+                TEST_TEXT,
+                &terms,
+                20,
+            );
             {
                 let first = &fragments[0];
                 assert_eq!(first.score, 1.0);
@@ -387,7 +402,12 @@ Survey in 2016, 2017, and 2018."#;
                 String::from("rust") =>0.9,
                 String::from("language") => 1.0
             };
-            let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
+            let fragments = search_fragments(
+                &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+                TEST_TEXT,
+                &terms,
+                20,
+            );
             //assert_eq!(fragments.len(), 7);
             {
                 let first = &fragments[0];
@@ -406,7 +426,12 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("c"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            &text,
+            &terms,
+            3,
+        );
 
         assert_eq!(fragments.len(), 1);
         {
@@ -428,7 +453,12 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("f"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            &text,
+            &terms,
+            3,
+        );
 
         assert_eq!(fragments.len(), 2);
         {
@@ -451,7 +481,12 @@ Survey in 2016, 2017, and 2018."#;
         terms.insert(String::from("f"), 1.0);
         terms.insert(String::from("a"), 0.9);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 7);
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            &text,
+            &terms,
+            7,
+        );
 
         assert_eq!(fragments.len(), 2);
         {
@@ -473,7 +508,12 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("z"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            &text,
+            &terms,
+            3,
+        );
 
         assert_eq!(fragments.len(), 0);
 
@@ -487,7 +527,12 @@ Survey in 2016, 2017, and 2018."#;
         let text = "a b c d";
 
         let terms = BTreeMap::new();
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(
+            &Into::<TextAnalyzer<_>>::into(SimpleTokenizer),
+            &text,
+            &terms,
+            3,
+        );
         assert_eq!(fragments.len(), 0);
 
         let snippet = select_best_fragment_combination(&fragments[..], &text);
@@ -572,12 +617,12 @@ Survey in 2016, 2017, and 2018."#;
         let mut snippet_generator =
             SnippetGenerator::create(&searcher, &*query, text_field).unwrap();
         {
-            let snippet = snippet_generator.snippet(TEST_TEXT);
+            let snippet = snippet_generator.snippet(TEST_TEXT.into());
             assert_eq!(snippet.to_html(), "imperative-procedural paradigms. <b>Rust</b> is syntactically similar to C++[according to whom?],\nbut its <b>designers</b> intend it to provide better memory safety");
         }
         {
             snippet_generator.set_max_num_chars(90);
-            let snippet = snippet_generator.snippet(TEST_TEXT);
+            let snippet = snippet_generator.snippet(TEST_TEXT.into());
             assert_eq!(snippet.to_html(), "<b>Rust</b> is syntactically similar to C++[according to whom?],\nbut its <b>designers</b> intend it to");
         }
     }

src/store/index/skip_index.rs

@@ -35,11 +35,11 @@ struct Layer {
 }
 
 impl Layer {
-    fn cursor<'a>(&'a self) -> impl Iterator<Item = Checkpoint> + 'a {
+    fn cursor(&self) -> impl Iterator<Item = Checkpoint> + '_ {
         self.cursor_at_offset(0u64)
     }
 
-    fn cursor_at_offset<'a>(&'a self, start_offset: u64) -> impl Iterator<Item = Checkpoint> + 'a {
+    fn cursor_at_offset(&self, start_offset: u64) -> impl Iterator<Item = Checkpoint> + '_ {
         let data = &self.data.as_slice();
         LayerCursor {
             remaining: &data[start_offset as usize..],
@@ -59,7 +59,7 @@ pub struct SkipIndex {
 }
 
 impl SkipIndex {
-    pub(crate) fn checkpoints<'a>(&'a self) -> impl Iterator<Item = Checkpoint> + 'a {
+    pub(crate) fn checkpoints(&self) -> impl Iterator<Item = Checkpoint> + '_ {
         self.layers
             .last()
             .into_iter()

src/store/reader.rs

@@ -46,7 +46,7 @@ impl StoreReader {
         })
     }
 
-    pub(crate) fn block_checkpoints<'a>(&'a self) -> impl Iterator<Item = Checkpoint> + 'a {
+    pub(crate) fn block_checkpoints(&self) -> impl Iterator<Item = Checkpoint> + '_ {
         self.skip_index.checkpoints()
     }
 

src/tokenizer/alphanum_only.rs

@@ -2,16 +2,16 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::from(RawTokenizer)
-//!   .filter(AlphaNumOnlyFilter);
+//! let tokenizer = analyzer_builder(RawTokenizer)
+//!   .filter(AlphaNumOnlyFilter).build();
 //!
 //! let mut stream = tokenizer.token_stream("hello there");
 //! // is none because the raw filter emits one token that
 //! // contains a space
 //! assert!(stream.next().is_none());
 //!
-//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
-//!   .filter(AlphaNumOnlyFilter);
+//! let tokenizer = analyzer_builder(SimpleTokenizer)
+//!   .filter(AlphaNumOnlyFilter).build();
 //!
 //! let mut stream = tokenizer.token_stream("hello there 💣");
 //! assert!(stream.next().is_some());
@@ -19,45 +19,18 @@
 //! // the "emoji" is dropped because its not an alphanum
 //! assert!(stream.next().is_none());
 //! ```
-use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
+use super::{Token, TokenFilter};
 
 /// `TokenFilter` that removes all tokens that contain non
 /// ascii alphanumeric characters.
-#[derive(Clone)]
+#[derive(Clone, Debug, Default)]
 pub struct AlphaNumOnlyFilter;
 
-pub struct AlphaNumOnlyFilterStream<'a> {
-    tail: BoxTokenStream<'a>,
-}
-
-impl<'a> AlphaNumOnlyFilterStream<'a> {
-    fn predicate(&self, token: &Token) -> bool {
-        token.text.chars().all(|c| c.is_ascii_alphanumeric())
-    }
-}
-
 impl TokenFilter for AlphaNumOnlyFilter {
-    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        BoxTokenStream::from(AlphaNumOnlyFilterStream { tail: token_stream })
-    }
-}
-
-impl<'a> TokenStream for AlphaNumOnlyFilterStream<'a> {
-    fn advance(&mut self) -> bool {
-        while self.tail.advance() {
-            if self.predicate(self.tail.token()) {
-                return true;
-            }
+    fn transform(&mut self, token: Token) -> Option<Token> {
+        if token.text.chars().all(|c| c.is_ascii_alphanumeric()) {
+            return Some(token);
         }
-
-        false
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
+        None
     }
 }

src/tokenizer/ascii_folding_filter.rs

@@ -1,45 +1,31 @@
-use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
+use super::{Token, TokenFilter};
 use std::mem;
 
 /// This class converts alphabetic, numeric, and symbolic Unicode characters
 /// which are not in the first 127 ASCII characters (the "Basic Latin" Unicode
 /// block) into their ASCII equivalents, if one exists.
-#[derive(Clone)]
-pub struct AsciiFoldingFilter;
-
-impl TokenFilter for AsciiFoldingFilter {
-    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        From::from(AsciiFoldingFilterTokenStream {
-            tail: token_stream,
-            buffer: String::with_capacity(100),
-        })
-    }
-}
-
-pub struct AsciiFoldingFilterTokenStream<'a> {
+#[derive(Clone, Debug, Default)]
+pub struct AsciiFolding {
     buffer: String,
-    tail: BoxTokenStream<'a>,
 }
 
-impl<'a> TokenStream for AsciiFoldingFilterTokenStream<'a> {
-    fn advance(&mut self) -> bool {
-        if !self.tail.advance() {
-            return false;
+impl AsciiFolding {
+    /// Construct a new `AsciiFolding` filter.
+    pub fn new() -> Self {
+        Self {
+            buffer: String::with_capacity(100),
         }
-        if !self.token_mut().text.is_ascii() {
+    }
+}
+
+impl TokenFilter for AsciiFolding {
+    fn transform(&mut self, mut token: Token) -> Option<Token> {
+        if !token.text.is_ascii() {
             // ignore its already ascii
-            to_ascii(&mut self.tail.token_mut().text, &mut self.buffer);
-            mem::swap(&mut self.tail.token_mut().text, &mut self.buffer);
+            to_ascii(&token.text, &mut self.buffer);
+            mem::swap(&mut token.text, &mut self.buffer);
         }
-        true
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
+        Some(token)
     }
 }
 
@@ -1526,7 +1512,7 @@ fn fold_non_ascii_char(c: char) -> Option<&'static str> {
 }
 
 // https://github.com/apache/lucene-solr/blob/master/lucene/analysis/common/src/java/org/apache/lucene/analysis/miscellaneous/ASCIIFoldingFilter.java#L187
-fn to_ascii(text: &mut String, output: &mut String) {
+fn to_ascii(text: &String, output: &mut String) {
     output.clear();
 
     for c in text.chars() {
@@ -1540,11 +1526,8 @@ fn to_ascii(text: &mut String, output: &mut String) {
 
 #[cfg(test)]
 mod tests {
-    use super::to_ascii;
-    use crate::tokenizer::AsciiFoldingFilter;
-    use crate::tokenizer::RawTokenizer;
-    use crate::tokenizer::SimpleTokenizer;
-    use crate::tokenizer::TextAnalyzer;
+    use super::super::*;
+    use super::*;
     use std::iter;
 
     #[test]
@@ -1560,22 +1543,22 @@ mod tests {
     }
 
     fn folding_helper(text: &str) -> Vec<String> {
-        let mut tokens = Vec::new();
-        TextAnalyzer::from(SimpleTokenizer)
-            .filter(AsciiFoldingFilter)
+        let tokens = analyzer_builder(SimpleTokenizer)
+            .filter(AsciiFolding::new())
+            .build()
             .token_stream(text)
-            .process(&mut |token| {
-                tokens.push(token.text.clone());
-            });
+            .map(|token| token.text.clone())
+            .collect();
         tokens
     }
 
     fn folding_using_raw_tokenizer_helper(text: &str) -> String {
-        let mut token_stream = TextAnalyzer::from(RawTokenizer)
-            .filter(AsciiFoldingFilter)
+        let mut token_stream = analyzer_builder(RawTokenizer)
+            .filter(AsciiFolding::new())
+            .build()
             .token_stream(text);
-        token_stream.advance();
-        token_stream.token().text.clone()
+        let Token { text, .. } = token_stream.next().unwrap();
+        text
     }
 
     #[test]
@@ -1626,9 +1609,9 @@ mod tests {
 
     #[test]
     fn test_to_ascii() {
-        let mut input = "Rámon".to_string();
+        let input = "Rámon".to_string();
         let mut buffer = String::new();
-        to_ascii(&mut input, &mut buffer);
+        to_ascii(&input, &mut buffer);
         assert_eq!("Ramon", buffer);
     }
 

src/tokenizer/facet_tokenizer.rs

@@ -1,4 +1,4 @@
-use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
+use super::{Token, Tokenizer};
 use crate::schema::FACET_SEP_BYTE;
 
 /// The `FacetTokenizer` process a `Facet` binary representation
@@ -9,72 +9,63 @@ use crate::schema::FACET_SEP_BYTE;
 ///     - `/america/north_america/canada`
 ///     - `/america/north_america`
 ///     - `/america`
-#[derive(Clone)]
+#[derive(Clone, Debug, Default)]
 pub struct FacetTokenizer;
 
-#[derive(Debug)]
+#[derive(Clone, Debug)]
 enum State {
     RootFacetNotEmitted,
     UpToPosition(usize), //< we already emitted facet prefix up to &text[..cursor]
     Terminated,
 }
 
-pub struct FacetTokenStream<'a> {
-    text: &'a str,
+#[derive(Clone, Debug)]
+pub struct FacetTokenStream {
+    text: String,
     state: State,
     token: Token,
 }
 
 impl Tokenizer for FacetTokenizer {
-    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+    type Iter = FacetTokenStream;
+    fn token_stream(&self, text: &str) -> Self::Iter {
         FacetTokenStream {
-            text,
+            text: text.to_string(),
             state: State::RootFacetNotEmitted, //< pos is the first char that has not been processed yet.
             token: Token::default(),
         }
-        .into()
     }
 }
 
-impl<'a> TokenStream for FacetTokenStream<'a> {
-    fn advance(&mut self) -> bool {
-        match self.state {
+impl Iterator for FacetTokenStream {
+    type Item = Token;
+    fn next(&mut self) -> Option<Self::Item> {
+        self.state = match self.state {
             State::RootFacetNotEmitted => {
-                self.state = if self.text.is_empty() {
+                if self.text.is_empty() {
                     State::Terminated
                 } else {
                     State::UpToPosition(0)
-                };
-                true
+                }
             }
             State::UpToPosition(cursor) => {
-                let bytes: &[u8] = self.text.as_bytes();
-                if let Some(next_sep_pos) = bytes[cursor + 1..]
+                if let Some(next_sep_pos) = self.text.as_bytes()[cursor + 1..]
                     .iter()
-                    .cloned()
-                    .position(|b| b == FACET_SEP_BYTE)
+                    .position(|&b| b == FACET_SEP_BYTE)
                     .map(|pos| cursor + 1 + pos)
                 {
                     let facet_part = &self.text[cursor..next_sep_pos];
                     self.token.text.push_str(facet_part);
-                    self.state = State::UpToPosition(next_sep_pos);
+                    State::UpToPosition(next_sep_pos)
                 } else {
                     let facet_part = &self.text[cursor..];
                     self.token.text.push_str(facet_part);
-                    self.state = State::Terminated;
+                    State::Terminated
                 }
-                true
             }
-            State::Terminated => false,
-        }
-    }
-
-    fn token(&self) -> &Token {
-        &self.token
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        &mut self.token
+            State::Terminated => return None,
+        };
+        Some(self.token.clone())
     }
 }
 
@@ -83,21 +74,19 @@ mod tests {
 
     use super::FacetTokenizer;
     use crate::schema::Facet;
-    use crate::tokenizer::{Token, Tokenizer};
+    use crate::tokenizer::Tokenizer;
 
     #[test]
     fn test_facet_tokenizer() {
         let facet = Facet::from_path(vec!["top", "a", "b"]);
-        let mut tokens = vec![];
-        {
-            let mut add_token = |token: &Token| {
-                let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap();
-                tokens.push(format!("{}", facet));
-            };
-            FacetTokenizer
-                .token_stream(facet.encoded_str())
-                .process(&mut add_token);
-        }
+        let tokens: Vec<_> = FacetTokenizer
+            .token_stream(facet.encoded_str())
+            .map(|token| {
+                Facet::from_encoded(token.text.as_bytes().to_owned())
+                    .unwrap()
+                    .to_string()
+            })
+            .collect();
         assert_eq!(tokens.len(), 4);
         assert_eq!(tokens[0], "/");
         assert_eq!(tokens[1], "/top");
@@ -108,16 +97,14 @@ mod tests {
     #[test]
     fn test_facet_tokenizer_root_facets() {
         let facet = Facet::root();
-        let mut tokens = vec![];
-        {
-            let mut add_token = |token: &Token| {
-                let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap(); // ok test
-                tokens.push(format!("{}", facet));
-            };
-            FacetTokenizer
-                .token_stream(facet.encoded_str()) // ok test
-                .process(&mut add_token);
-        }
+        let tokens: Vec<_> = FacetTokenizer
+            .token_stream(facet.encoded_str())
+            .map(|token| {
+                Facet::from_encoded(token.text.as_bytes().to_owned())
+                    .unwrap()
+                    .to_string()
+            })
+            .collect();
         assert_eq!(tokens.len(), 1);
         assert_eq!(tokens[0], "/");
     }

src/tokenizer/lower_caser.rs

@@ -1,27 +1,36 @@
-use super::{Token, TokenFilter, TokenStream};
-use crate::tokenizer::BoxTokenStream;
+use super::{Token, TokenFilter};
 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),
-        })
+    fn transform(&mut self, mut token: Token) -> Option<Token> {
+        if token.text.is_ascii() {
+            // fast track for ascii.
+            token.text.make_ascii_lowercase();
+        } else {
+            to_lowercase_unicode(&token.text, &mut self.buffer);
+            mem::swap(&mut token.text, &mut self.buffer);
+        }
+        Some(token)
     }
 }
 
 /// Token filter that lowercase terms.
-#[derive(Clone)]
-pub struct LowerCaser;
-
-pub struct LowerCaserTokenStream<'a> {
+#[derive(Clone, Debug, Default)]
+pub struct LowerCaser {
     buffer: String,
-    tail: BoxTokenStream<'a>,
+}
+
+impl LowerCaser {
+    /// Initialize the `LowerCaser`
+    pub fn new() -> Self {
+        LowerCaser {
+            buffer: String::with_capacity(100),
+        }
+    }
 }
 
 // writes a lowercased version of text into output.
-fn to_lowercase_unicode(text: &mut String, output: &mut String) {
+fn to_lowercase_unicode(text: &String, output: &mut String) {
     output.clear();
     for c in text.chars() {
         // Contrary to the std, we do not take care of sigma special case.
@@ -30,57 +39,31 @@ fn to_lowercase_unicode(text: &mut String, output: &mut String) {
     }
 }
 
-impl<'a> TokenStream for LowerCaserTokenStream<'a> {
-    fn advance(&mut self) -> bool {
-        if !self.tail.advance() {
-            return false;
-        }
-        if self.token_mut().text.is_ascii() {
-            // fast track for ascii.
-            self.token_mut().text.make_ascii_lowercase();
-        } else {
-            to_lowercase_unicode(&mut self.tail.token_mut().text, &mut self.buffer);
-            mem::swap(&mut self.tail.token_mut().text, &mut self.buffer);
-        }
-        true
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
-    }
-}
-
 #[cfg(test)]
 mod tests {
-    use crate::tokenizer::{LowerCaser, SimpleTokenizer, TextAnalyzer};
+    use super::*;
+    use crate::tokenizer::{analyzer_builder, LowerCaser, SimpleTokenizer, TextAnalyzerT};
 
     #[test]
     fn test_to_lower_case() {
-        assert_eq!(
-            lowercase_helper("Русский текст"),
-            vec!["русский".to_string(), "текст".to_string()]
-        );
+        assert_eq!(lowercase_helper("Русский текст"), vec!["русский", "текст"]);
     }
 
     fn lowercase_helper(text: &str) -> Vec<String> {
-        let mut tokens = vec![];
-        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);
-        }
-        tokens
+        analyzer_builder(SimpleTokenizer)
+            .filter(LowerCaser::new())
+            .build()
+            .token_stream(text)
+            .map(|token| {
+                let Token { text, .. } = token;
+                text
+            })
+            .collect()
     }
 
     #[test]
     fn test_lowercaser() {
-        assert_eq!(lowercase_helper("Tree"), vec!["tree".to_string()]);
-        assert_eq!(lowercase_helper("Русский"), vec!["русский".to_string()]);
+        assert_eq!(lowercase_helper("Tree"), vec!["tree"]);
+        assert_eq!(lowercase_helper("Русский"), vec!["русский"]);
     }
 }

src/tokenizer/mod.rs

@@ -64,10 +64,10 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let en_stem = TextAnalyzer::from(SimpleTokenizer)
+//! let en_stem = analyzer_builder(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
-//!     .filter(LowerCaser)
-//!     .filter(Stemmer::new(Language::English));
+//!     .filter(LowerCaser::new())
+//!     .filter(Stemmer::new(Language::English)).build();
 //! ```
 //!
 //! Once your tokenizer is defined, you need to
@@ -109,9 +109,9 @@
 //! let index = Index::create_in_ram(schema);
 //!
 //! // We need to register our tokenizer :
-//! let custom_en_tokenizer = TextAnalyzer::from(SimpleTokenizer)
+//! let custom_en_tokenizer = analyzer_builder(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
-//!     .filter(LowerCaser);
+//!     .filter(LowerCaser::new()).build();
 //! index
 //!     .tokenizers()
 //!     .register("custom_en", custom_en_tokenizer);
@@ -133,7 +133,7 @@ mod tokenizer;
 mod tokenizer_manager;
 
 pub use self::alphanum_only::AlphaNumOnlyFilter;
-pub use self::ascii_folding_filter::AsciiFoldingFilter;
+pub use self::ascii_folding_filter::AsciiFolding;
 pub use self::facet_tokenizer::FacetTokenizer;
 pub use self::lower_caser::LowerCaser;
 pub use self::ngram_tokenizer::NgramTokenizer;
@@ -142,11 +142,11 @@ pub use self::remove_long::RemoveLongFilter;
 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(crate) use self::token_stream_chain::{DynTokenStreamChain, TokenStreamChain};
 
 pub use self::tokenized_string::{PreTokenizedStream, PreTokenizedString};
 pub use self::tokenizer::{
-    BoxTokenFilter, BoxTokenStream, TextAnalyzer, Token, TokenFilter, TokenStream, Tokenizer,
+    analyzer_builder, Identity, TextAnalyzer, TextAnalyzerT, Token, TokenFilter, Tokenizer,
 };
 
 pub use self::tokenizer_manager::TokenizerManager;
@@ -160,10 +160,7 @@ 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, TokenizerManager,
-    };
-    use crate::tokenizer::TextAnalyzer;
+    use super::*;
 
     /// This is a function that can be used in tests and doc tests
     /// to assert a token's correctness.
@@ -190,15 +187,9 @@ pub mod tests {
     fn test_raw_tokenizer() {
         let tokenizer_manager = TokenizerManager::default();
         let en_tokenizer = tokenizer_manager.get("raw").unwrap();
-        let mut tokens: Vec<Token> = vec![];
-        {
-            let mut add_token = |token: &Token| {
-                tokens.push(token.clone());
-            };
-            en_tokenizer
-                .token_stream("Hello, happy tax payer!")
-                .process(&mut add_token);
-        }
+        let tokens: Vec<Token> = en_tokenizer
+            .token_stream("Hello, happy tax payer!")
+            .collect();
         assert_eq!(tokens.len(), 1);
         assert_token(&tokens[0], 0, "Hello, happy tax payer!", 0, 23);
     }
@@ -208,15 +199,9 @@ pub mod tests {
         let tokenizer_manager = TokenizerManager::default();
         assert!(tokenizer_manager.get("en_doesnotexist").is_none());
         let en_tokenizer = tokenizer_manager.get("en_stem").unwrap();
-        let mut tokens: Vec<Token> = vec![];
-        {
-            let mut add_token = |token: &Token| {
-                tokens.push(token.clone());
-            };
-            en_tokenizer
-                .token_stream("Hello, happy tax payer!")
-                .process(&mut add_token);
-        }
+        let tokens: Vec<Token> = en_tokenizer
+            .token_stream("Hello, happy tax payer!")
+            .collect();
 
         assert_eq!(tokens.len(), 4);
         assert_token(&tokens[0], 0, "hello", 0, 5);
@@ -230,21 +215,16 @@ pub mod tests {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "el_stem",
-            TextAnalyzer::from(SimpleTokenizer)
+            analyzer_builder(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
-                .filter(LowerCaser)
-                .filter(Stemmer::new(Language::Greek)),
+                .filter(LowerCaser::new())
+                .filter(Stemmer::new(Language::Greek))
+                .build(),
         );
         let en_tokenizer = tokenizer_manager.get("el_stem").unwrap();
-        let mut tokens: Vec<Token> = vec![];
-        {
-            let mut add_token = |token: &Token| {
-                tokens.push(token.clone());
-            };
-            en_tokenizer
-                .token_stream("Καλημέρα, χαρούμενε φορολογούμενε!")
-                .process(&mut add_token);
-        }
+        let tokens: Vec<Token> = en_tokenizer
+            .token_stream("Καλημέρα, χαρούμενε φορολογούμενε!")
+            .collect();
 
         assert_eq!(tokens.len(), 3);
         assert_token(&tokens[0], 0, "καλημερ", 0, 16);
@@ -256,25 +236,9 @@ pub mod tests {
     fn test_tokenizer_empty() {
         let tokenizer_manager = TokenizerManager::default();
         let en_tokenizer = tokenizer_manager.get("en_stem").unwrap();
-        {
-            let mut tokens: Vec<Token> = vec![];
-            {
-                let mut add_token = |token: &Token| {
-                    tokens.push(token.clone());
-                };
-                en_tokenizer.token_stream(" ").process(&mut add_token);
-            }
-            assert!(tokens.is_empty());
-        }
-        {
-            let mut tokens: Vec<Token> = vec![];
-            {
-                let mut add_token = |token: &Token| {
-                    tokens.push(token.clone());
-                };
-                en_tokenizer.token_stream(" ").process(&mut add_token);
-            }
-            assert!(tokens.is_empty());
-        }
+        let tokens: Vec<Token> = en_tokenizer.token_stream(" ").collect();
+        assert!(tokens.is_empty());
+        let tokens: Vec<Token> = en_tokenizer.token_stream(" ").collect();
+        assert!(tokens.is_empty());
     }
 }

src/tokenizer/ngram_tokenizer.rs

@@ -1,5 +1,4 @@
-use super::{Token, TokenStream, Tokenizer};
-use crate::tokenizer::BoxTokenStream;
+use super::{Token, Tokenizer};
 
 /// Tokenize the text by splitting words into n-grams of the given size(s)
 ///
@@ -79,7 +78,7 @@ use crate::tokenizer::BoxTokenStream;
 /// }
 /// assert!(stream.next().is_none());
 /// ```
-#[derive(Clone)]
+#[derive(Clone, Debug, Default)]
 pub struct NgramTokenizer {
     /// min size of the n-gram
     min_gram: usize,
@@ -119,54 +118,48 @@ impl NgramTokenizer {
 }
 
 /// TokenStream associate to the `NgramTokenizer`
-pub struct NgramTokenStream<'a> {
+pub struct NgramTokenStream {
     /// parameters
-    ngram_charidx_iterator: StutteringIterator<CodepointFrontiers<'a>>,
+    ngram_charidx_iterator: StutteringIterator<CodepointFrontiers>,
     /// true if the NgramTokenStream is in prefix mode.
     prefix_only: bool,
     /// input
-    text: &'a str,
+    text: String,
     /// output
     token: Token,
 }
 
 impl Tokenizer for NgramTokenizer {
-    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-        From::from(NgramTokenStream {
+    type Iter = NgramTokenStream;
+    fn token_stream(&self, text: &str) -> Self::Iter {
+        NgramTokenStream {
             ngram_charidx_iterator: StutteringIterator::new(
                 CodepointFrontiers::for_str(text),
                 self.min_gram,
                 self.max_gram,
             ),
             prefix_only: self.prefix_only,
-            text,
+            text: text.to_string(),
             token: Token::default(),
-        })
+        }
     }
 }
 
-impl<'a> TokenStream for NgramTokenStream<'a> {
-    fn advance(&mut self) -> bool {
+impl Iterator for NgramTokenStream {
+    type Item = Token;
+    fn next(&mut self) -> Option<Self::Item> {
         if let Some((offset_from, offset_to)) = self.ngram_charidx_iterator.next() {
             if self.prefix_only && offset_from > 0 {
-                return false;
+                return None;
             }
             self.token.position = 0;
             self.token.offset_from = offset_from;
             self.token.offset_to = offset_to;
             self.token.text.clear();
             self.token.text.push_str(&self.text[offset_from..offset_to]);
-            true
-        } else {
-            false
-        }
-    }
-
-    fn token(&self) -> &Token {
-        &self.token
-    }
-    fn token_mut(&mut self) -> &mut Token {
-        &mut self.token
+            return Some(self.token.clone());
+        };
+        None
     }
 }
 
@@ -257,21 +250,21 @@ where
 /// or a codepoint ends.
 ///
 /// By convention, we emit [0] for the empty string.
-struct CodepointFrontiers<'a> {
-    s: &'a str,
+struct CodepointFrontiers {
+    s: String,
     next_el: Option<usize>,
 }
 
-impl<'a> CodepointFrontiers<'a> {
-    fn for_str(s: &'a str) -> Self {
+impl CodepointFrontiers {
+    fn for_str(s: &str) -> Self {
         CodepointFrontiers {
-            s,
+            s: s.to_string(),
             next_el: Some(0),
         }
     }
 }
 
-impl<'a> Iterator for CodepointFrontiers<'a> {
+impl<'a> Iterator for CodepointFrontiers {
     type Item = usize;
 
     fn next(&mut self) -> Option<usize> {
@@ -280,7 +273,7 @@ impl<'a> Iterator for CodepointFrontiers<'a> {
                 self.next_el = None;
             } else {
                 let first_codepoint_width = utf8_codepoint_width(self.s.as_bytes()[0]);
-                self.s = &self.s[first_codepoint_width..];
+                self.s = (&self.s[first_codepoint_width..]).to_string();
                 self.next_el = Some(offset + first_codepoint_width);
             }
             offset
@@ -301,20 +294,8 @@ fn utf8_codepoint_width(b: u8) -> usize {
 
 #[cfg(test)]
 mod tests {
-
-    use super::utf8_codepoint_width;
-    use super::CodepointFrontiers;
-    use super::NgramTokenizer;
-    use super::StutteringIterator;
+    use super::*;
     use crate::tokenizer::tests::assert_token;
-    use crate::tokenizer::tokenizer::Tokenizer;
-    use crate::tokenizer::{BoxTokenStream, 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
-    }
 
     #[test]
     fn test_utf8_codepoint_width() {
@@ -351,7 +332,9 @@ mod tests {
 
     #[test]
     fn test_ngram_tokenizer_1_2_false() {
-        let tokens = test_helper(NgramTokenizer::all_ngrams(1, 2).token_stream("hello"));
+        let tokens: Vec<_> = NgramTokenizer::all_ngrams(1, 2)
+            .token_stream("hello")
+            .collect();
         assert_eq!(tokens.len(), 9);
         assert_token(&tokens[0], 0, "h", 0, 1);
         assert_token(&tokens[1], 0, "he", 0, 2);
@@ -366,7 +349,9 @@ mod tests {
 
     #[test]
     fn test_ngram_tokenizer_min_max_equal() {
-        let tokens = test_helper(NgramTokenizer::all_ngrams(3, 3).token_stream("hello"));
+        let tokens: Vec<_> = NgramTokenizer::all_ngrams(3, 3)
+            .token_stream("hello")
+            .collect();
         assert_eq!(tokens.len(), 3);
         assert_token(&tokens[0], 0, "hel", 0, 3);
         assert_token(&tokens[1], 0, "ell", 1, 4);
@@ -375,7 +360,9 @@ mod tests {
 
     #[test]
     fn test_ngram_tokenizer_2_5_prefix() {
-        let tokens = test_helper(NgramTokenizer::prefix_only(2, 5).token_stream("frankenstein"));
+        let tokens: Vec<_> = NgramTokenizer::prefix_only(2, 5)
+            .token_stream("frankenstein")
+            .collect();
         assert_eq!(tokens.len(), 4);
         assert_token(&tokens[0], 0, "fr", 0, 2);
         assert_token(&tokens[1], 0, "fra", 0, 3);
@@ -385,7 +372,9 @@ mod tests {
 
     #[test]
     fn test_ngram_non_ascii_1_2() {
-        let tokens = test_helper(NgramTokenizer::all_ngrams(1, 2).token_stream("hεllo"));
+        let tokens: Vec<_> = NgramTokenizer::all_ngrams(1, 2)
+            .token_stream("hεllo")
+            .collect();
         assert_eq!(tokens.len(), 9);
         assert_token(&tokens[0], 0, "h", 0, 1);
         assert_token(&tokens[1], 0, "hε", 0, 3);
@@ -400,7 +389,9 @@ mod tests {
 
     #[test]
     fn test_ngram_non_ascii_2_5_prefix() {
-        let tokens = test_helper(NgramTokenizer::prefix_only(2, 5).token_stream("hεllo"));
+        let tokens: Vec<_> = NgramTokenizer::prefix_only(2, 5)
+            .token_stream("hεllo")
+            .collect();
         assert_eq!(tokens.len(), 4);
         assert_token(&tokens[0], 0, "hε", 0, 3);
         assert_token(&tokens[1], 0, "hεl", 0, 4);
@@ -410,16 +401,16 @@ mod tests {
 
     #[test]
     fn test_ngram_empty() {
-        let tokens = test_helper(NgramTokenizer::all_ngrams(1, 5).token_stream(""));
+        let tokens: Vec<_> = NgramTokenizer::all_ngrams(1, 5).token_stream("").collect();
         assert!(tokens.is_empty());
-        let tokens = test_helper(NgramTokenizer::all_ngrams(2, 5).token_stream(""));
+        let tokens: Vec<_> = NgramTokenizer::all_ngrams(2, 5).token_stream("").collect();
         assert!(tokens.is_empty());
     }
 
     #[test]
     #[should_panic(expected = "min_gram must be greater than 0")]
     fn test_ngram_min_max_interval_empty() {
-        test_helper(NgramTokenizer::all_ngrams(0, 2).token_stream("hellossss"));
+        NgramTokenizer::all_ngrams(0, 2).token_stream("hellossss");
     }
 
     #[test]

src/tokenizer/raw_tokenizer.rs

@@ -1,17 +1,17 @@
-use super::{Token, TokenStream, Tokenizer};
-use crate::tokenizer::BoxTokenStream;
+use super::{Token, Tokenizer};
 
 /// For each value of the field, emit a single unprocessed token.
-#[derive(Clone)]
+#[derive(Clone, Debug, Default)]
 pub struct RawTokenizer;
 
+#[derive(Clone, Debug)]
 pub struct RawTokenStream {
-    token: Token,
-    has_token: bool,
+    token: Option<Token>,
 }
 
 impl Tokenizer for RawTokenizer {
-    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+    type Iter = RawTokenStream;
+    fn token_stream(&self, text: &str) -> Self::Iter {
         let token = Token {
             offset_from: 0,
             offset_to: text.len(),
@@ -19,26 +19,13 @@ impl Tokenizer for RawTokenizer {
             text: text.to_string(),
             position_length: 1,
         };
-        RawTokenStream {
-            token,
-            has_token: true,
-        }
-        .into()
+        RawTokenStream { token: Some(token) }
     }
 }
 
-impl TokenStream for RawTokenStream {
-    fn advance(&mut self) -> bool {
-        let result = self.has_token;
-        self.has_token = false;
-        result
-    }
-
-    fn token(&self) -> &Token {
-        &self.token
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        &mut self.token
+impl Iterator for RawTokenStream {
+    type Item = Token;
+    fn next(&mut self) -> Option<Token> {
+        self.token.take()
     }
 }

src/tokenizer/remove_long.rs

@@ -2,8 +2,8 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
-//!   .filter(RemoveLongFilter::limit(5));
+//! let tokenizer = analyzer_builder(SimpleTokenizer)
+//!   .filter(RemoveLongFilter::limit(5)).build();
 //!
 //! let mut stream = tokenizer.token_stream("toolong nice");
 //! // because `toolong` is more than 5 characters, it is filtered
@@ -12,61 +12,30 @@
 //! assert!(stream.next().is_none());
 //! ```
 //!
-use super::{Token, TokenFilter, TokenStream};
-use crate::tokenizer::BoxTokenStream;
+use super::{Token, TokenFilter};
 
 /// `RemoveLongFilter` removes tokens that are longer
 /// than a given number of bytes (in UTF-8 representation).
 ///
 /// It is especially useful when indexing unconstrained content.
 /// e.g. Mail containing base-64 encoded pictures etc.
-#[derive(Clone)]
+#[derive(Clone, Debug)]
 pub struct RemoveLongFilter {
-    length_limit: usize,
+    limit: usize,
 }
 
 impl RemoveLongFilter {
     /// Creates a `RemoveLongFilter` given a limit in bytes of the UTF-8 representation.
-    pub fn limit(length_limit: usize) -> RemoveLongFilter {
-        RemoveLongFilter { length_limit }
-    }
-}
-
-impl<'a> RemoveLongFilterStream<'a> {
-    fn predicate(&self, token: &Token) -> bool {
-        token.text.len() < self.token_length_limit
+    pub fn limit(limit: usize) -> RemoveLongFilter {
+        RemoveLongFilter { limit }
     }
 }
 
 impl TokenFilter for RemoveLongFilter {
-    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        BoxTokenStream::from(RemoveLongFilterStream {
-            token_length_limit: self.length_limit,
-            tail: token_stream,
-        })
-    }
-}
-
-pub struct RemoveLongFilterStream<'a> {
-    token_length_limit: usize,
-    tail: BoxTokenStream<'a>,
-}
-
-impl<'a> TokenStream for RemoveLongFilterStream<'a> {
-    fn advance(&mut self) -> bool {
-        while self.tail.advance() {
-            if self.predicate(self.tail.token()) {
-                return true;
-            }
+    fn transform(&mut self, token: Token) -> Option<Token> {
+        if token.text.len() >= self.limit {
+            return None;
         }
-        false
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
+        Some(token)
     }
 }

src/tokenizer/simple_tokenizer.rs

@@ -1,59 +1,74 @@
-use super::BoxTokenStream;
-use super::{Token, TokenStream, Tokenizer};
-use std::str::CharIndices;
+use super::{Token, Tokenizer};
 
 /// Tokenize the text by splitting on whitespaces and punctuation.
-#[derive(Clone)]
+#[derive(Clone, Debug)]
 pub struct SimpleTokenizer;
-
-pub struct SimpleTokenStream<'a> {
-    text: &'a str,
-    chars: CharIndices<'a>,
-    token: Token,
-}
-
 impl Tokenizer for SimpleTokenizer {
-    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-        BoxTokenStream::from(SimpleTokenStream {
-            text,
-            chars: text.char_indices(),
-            token: Token::default(),
-        })
+    type Iter = SimpleTokenizerStream;
+    fn token_stream(&self, text: &str) -> Self::Iter {
+        let vec: Vec<_> = text.char_indices().collect();
+        SimpleTokenizerStream {
+            text: text.to_string(),
+            chars: vec.into_iter(),
+            position: usize::max_value(),
+        }
     }
 }
 
-impl<'a> SimpleTokenStream<'a> {
+#[derive(Clone, Debug)]
+pub struct SimpleTokenizerStream {
+    text: String,
+    chars: std::vec::IntoIter<(usize, char)>,
+    position: usize,
+}
+
+impl SimpleTokenizerStream {
     // search for the end of the current token.
     fn search_token_end(&mut self) -> usize {
         (&mut self.chars)
-            .filter(|&(_, ref c)| !c.is_alphanumeric())
+            .filter(|&(_, c)| !c.is_alphanumeric())
             .map(|(offset, _)| offset)
             .next()
             .unwrap_or_else(|| self.text.len())
     }
 }
 
-impl<'a> TokenStream for SimpleTokenStream<'a> {
-    fn advance(&mut self) -> bool {
-        self.token.text.clear();
-        self.token.position = self.token.position.wrapping_add(1);
+impl Iterator for SimpleTokenizerStream {
+    type Item = Token;
+    fn next(&mut self) -> Option<Self::Item> {
+        self.position = self.position.wrapping_add(1);
         while let Some((offset_from, c)) = self.chars.next() {
             if c.is_alphanumeric() {
                 let offset_to = self.search_token_end();
-                self.token.offset_from = offset_from;
-                self.token.offset_to = offset_to;
-                self.token.text.push_str(&self.text[offset_from..offset_to]);
-                return true;
+                let token = Token {
+                    text: self.text[offset_from..offset_to].into(),
+                    offset_from,
+                    offset_to,
+                    position: self.position,
+                    ..Default::default()
+                };
+                return Some(token);
             }
         }
-        false
-    }
-
-    fn token(&self) -> &Token {
-        &self.token
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        &mut self.token
+        None
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_empty() {
+        let mut empty = SimpleTokenizer.token_stream("");
+        assert_eq!(empty.next(), None);
+    }
+
+    #[test]
+    fn simple_tokenizer() {
+        let mut simple = SimpleTokenizer.token_stream("tokenizer hello world");
+        assert_eq!(simple.next().unwrap().text, "tokenizer");
+        assert_eq!(simple.next().unwrap().text, "hello");
+        assert_eq!(simple.next().unwrap().text, "world");
     }
 }

src/tokenizer/stemmer.rs

@@ -1,5 +1,6 @@
-use super::{Token, TokenFilter, TokenStream};
-use crate::tokenizer::BoxTokenStream;
+use std::sync::Arc;
+
+use super::{Token, TokenFilter};
 use rust_stemmers::{self, Algorithm};
 use serde::{Deserialize, Serialize};
 
@@ -58,14 +59,15 @@ impl Language {
 /// Tokens are expected to be lowercased beforehand.
 #[derive(Clone)]
 pub struct Stemmer {
-    stemmer_algorithm: Algorithm,
+    stemmer: Arc<rust_stemmers::Stemmer>,
 }
 
 impl Stemmer {
     /// Creates a new Stemmer `TokenFilter` for a given language algorithm.
     pub fn new(language: Language) -> Stemmer {
+        let stemmer = rust_stemmers::Stemmer::create(language.algorithm());
         Stemmer {
-            stemmer_algorithm: language.algorithm(),
+            stemmer: Arc::new(stemmer),
         }
     }
 }
@@ -78,37 +80,12 @@ impl Default for Stemmer {
 }
 
 impl TokenFilter for Stemmer {
-    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        let inner_stemmer = rust_stemmers::Stemmer::create(self.stemmer_algorithm);
-        BoxTokenStream::from(StemmerTokenStream {
-            tail: token_stream,
-            stemmer: inner_stemmer,
-        })
-    }
-}
-
-pub struct StemmerTokenStream<'a> {
-    tail: BoxTokenStream<'a>,
-    stemmer: rust_stemmers::Stemmer,
-}
-
-impl<'a> TokenStream for StemmerTokenStream<'a> {
-    fn advance(&mut self) -> bool {
-        if !self.tail.advance() {
-            return false;
-        }
+    fn transform(&mut self, mut token: Token) -> Option<Token> {
         // TODO remove allocation
-        let stemmed_str: String = self.stemmer.stem(&self.token().text).into_owned();
-        self.token_mut().text.clear();
-        self.token_mut().text.push_str(&stemmed_str);
-        true
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
+        let stemmed_str: String = self.stemmer.stem(&token.text).into_owned();
+        // TODO remove clear
+        token.text.clear();
+        token.text.push_str(&stemmed_str);
+        Some(token)
     }
 }

src/tokenizer/stop_word_filter.rs

@@ -2,16 +2,15 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
-//!   .filter(StopWordFilter::remove(vec!["the".to_string(), "is".to_string()]));
+//! let tokenizer = analyzer_builder(SimpleTokenizer)
+//!   .filter(StopWordFilter::remove(vec!["the".to_string(), "is".to_string()])).build();
 //!
 //! let mut stream = tokenizer.token_stream("the fox is crafty");
 //! assert_eq!(stream.next().unwrap().text, "fox");
 //! assert_eq!(stream.next().unwrap().text, "crafty");
 //! assert!(stream.next().is_none());
 //! ```
-use super::{Token, TokenFilter, TokenStream};
-use crate::tokenizer::BoxTokenStream;
+use super::{Token, TokenFilter};
 use fnv::FnvHasher;
 use std::collections::HashSet;
 use std::hash::BuildHasherDefault;
@@ -49,42 +48,12 @@ impl StopWordFilter {
     }
 }
 
-pub struct StopWordFilterStream<'a> {
-    words: StopWordHashSet,
-    tail: BoxTokenStream<'a>,
-}
-
 impl TokenFilter for StopWordFilter {
-    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
-        BoxTokenStream::from(StopWordFilterStream {
-            words: self.words.clone(),
-            tail: token_stream,
-        })
-    }
-}
-
-impl<'a> StopWordFilterStream<'a> {
-    fn predicate(&self, token: &Token) -> bool {
-        !self.words.contains(&token.text)
-    }
-}
-
-impl<'a> TokenStream for StopWordFilterStream<'a> {
-    fn advance(&mut self) -> bool {
-        while self.tail.advance() {
-            if self.predicate(self.tail.token()) {
-                return true;
-            }
+    fn transform(&mut self, token: Token) -> Option<Token> {
+        if self.words.contains(&token.text) {
+            return None;
         }
-        false
-    }
-
-    fn token(&self) -> &Token {
-        self.tail.token()
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        self.tail.token_mut()
+        Some(token)
     }
 }
 

src/tokenizer/token_stream_chain.rs

@@ -1,95 +1,121 @@
-use crate::tokenizer::{BoxTokenStream, Token, TokenStream};
-use std::ops::DerefMut;
+use crate::tokenizer::Token;
 
 const POSITION_GAP: usize = 2;
 
-pub(crate) struct TokenStreamChain<'a> {
-    offsets: Vec<usize>,
-    token_streams: Vec<BoxTokenStream<'a>>,
+pub(crate) struct TokenStreamChain<Inner, Outer> {
+    streams_with_offsets: Outer,
+    current: Option<(Inner, usize)>,
+    position: usize,
     position_shift: usize,
-    stream_idx: usize,
-    token: Token,
 }
 
-impl<'a> TokenStreamChain<'a> {
-    pub fn new(
-        offsets: Vec<usize>,
-        token_streams: Vec<BoxTokenStream<'a>>,
-    ) -> TokenStreamChain<'a> {
+impl<'a, Inner, Outer> TokenStreamChain<Inner, Outer>
+where
+    Inner: Iterator<Item = Token>,
+    Outer: Iterator<Item = (Inner, usize)>,
+{
+    pub fn new(mut streams_with_offsets: Outer) -> TokenStreamChain<Inner, Outer> {
+        let current = streams_with_offsets.next();
         TokenStreamChain {
-            offsets,
-            stream_idx: 0,
-            token_streams,
+            streams_with_offsets: streams_with_offsets,
+            current,
+            position: usize::max_value(),
             position_shift: 0,
-            token: Token::default(),
         }
     }
 }
 
-impl<'a> TokenStream for TokenStreamChain<'a> {
-    fn advance(&mut self) -> bool {
-        while self.stream_idx < self.token_streams.len() {
-            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];
-                self.token.offset_from = token.offset_from + offset_offset;
-                self.token.offset_to = token.offset_to + offset_offset;
-                self.token.position = token.position + self.position_shift;
-                self.token.text.clear();
-                self.token.text.push_str(token.text.as_str());
-                return true;
-            } else {
-                self.stream_idx += 1;
-                self.position_shift = self.token.position.wrapping_add(POSITION_GAP);
+impl<'a, Inner, Outer> Iterator for TokenStreamChain<Inner, Outer>
+where
+    Inner: Iterator<Item = Token>,
+    Outer: Iterator<Item = (Inner, usize)>,
+{
+    type Item = Token;
+    fn next(&mut self) -> Option<Token> {
+        while let Some((ref mut token_stream, offset_offset)) = self.current {
+            if let Some(mut token) = token_stream.next() {
+                token.offset_from += offset_offset;
+                token.offset_to += offset_offset;
+                token.position += self.position_shift;
+                self.position = token.position;
+                return Some(token);
             }
+            self.position_shift = self.position.wrapping_add(POSITION_GAP);
+            self.current = self.streams_with_offsets.next();
         }
-        false
+        None
     }
+}
 
-    fn token(&self) -> &Token {
-        assert!(
-            self.stream_idx <= self.token_streams.len(),
-            "You called .token(), after the end of the token stream has been reached"
-        );
-        &self.token
+impl DynTokenStreamChain {
+    pub fn from_vec(
+        streams_with_offsets: Vec<(Box<dyn Iterator<Item = Token>>, usize)>,
+    ) -> impl Iterator<Item = Token> {
+        DynTokenStreamChain {
+            streams_with_offsets,
+            idx: 0,
+            position: usize::max_value(),
+            position_shift: 0,
+        }
     }
+}
 
-    fn token_mut(&mut self) -> &mut Token {
-        assert!(
-            self.stream_idx <= self.token_streams.len(),
-            "You called .token(), after the end of the token stream has been reached"
-        );
-        &mut self.token
+pub(crate) struct DynTokenStreamChain {
+    streams_with_offsets: Vec<(Box<dyn Iterator<Item = Token>>, usize)>,
+    idx: usize,
+    position: usize,
+    position_shift: usize,
+}
+
+impl Iterator for DynTokenStreamChain {
+    type Item = Token;
+    fn next(&mut self) -> Option<Token> {
+        while let Some((token_stream, offset_offset)) = self.streams_with_offsets.get_mut(self.idx)
+        {
+            if let Some(mut token) = token_stream.next() {
+                token.offset_from += *offset_offset;
+                token.offset_to += *offset_offset;
+                token.position += self.position_shift;
+                self.position = token.position;
+                return Some(token);
+            }
+            self.idx += 1;
+            self.position_shift = self.position.wrapping_add(POSITION_GAP);
+        }
+        None
     }
 }
 
 #[cfg(test)]
 mod tests {
-    use super::super::{SimpleTokenizer, TokenStream, Tokenizer};
-    use super::TokenStreamChain;
-    use super::POSITION_GAP;
+    use super::super::tokenizer::Tokenizer;
+    use super::super::SimpleTokenizer;
+    use super::*;
 
     #[test]
     fn test_chain_first_emits_no_tokens() {
         let token_streams = vec![
-            SimpleTokenizer.token_stream(""),
-            SimpleTokenizer.token_stream("hello world"),
+            (SimpleTokenizer.token_stream(""), 0),
+            (SimpleTokenizer.token_stream("hello world"), 0),
         ];
-        let mut token_chain = TokenStreamChain::new(vec![0, 0], token_streams);
+        let mut token_chain = TokenStreamChain::new(token_streams.into_iter());
+        let token = token_chain.next();
 
-        assert!(token_chain.advance());
-        assert_eq!(token_chain.token().text, "hello");
-        assert_eq!(token_chain.token().offset_from, 0);
-        assert_eq!(token_chain.token().offset_to, 5);
-        assert_eq!(token_chain.token().position, POSITION_GAP - 1);
+        let expect = Token {
+            offset_from: 0,
+            offset_to: 5,
+            position: POSITION_GAP - 1,
+            text: "hello".into(),
+            ..Token::default()
+        };
+        assert_eq!(token.unwrap(), expect);
 
-        assert!(token_chain.advance());
-        assert_eq!(token_chain.token().text, "world");
-        assert_eq!(token_chain.token().offset_from, 6);
-        assert_eq!(token_chain.token().offset_to, 11);
-        assert_eq!(token_chain.token().position, POSITION_GAP);
+        let token = token_chain.next().unwrap();
+        assert_eq!(token.text, "world");
+        assert_eq!(token.offset_from, 6);
+        assert_eq!(token.offset_to, 11);
+        assert_eq!(token.position, POSITION_GAP);
 
-        assert!(!token_chain.advance());
+        assert!(token_chain.next().is_none());
     }
 }

src/tokenizer/tokenized_string.rs

@@ -1,4 +1,4 @@
-use crate::tokenizer::{BoxTokenStream, Token, TokenStream, TokenStreamChain};
+use crate::tokenizer::{Token, TokenStreamChain};
 use serde::{Deserialize, Serialize};
 use std::cmp::Ordering;
 
@@ -26,14 +26,14 @@ impl PartialOrd for PreTokenizedString {
 /// TokenStream implementation which wraps PreTokenizedString
 pub struct PreTokenizedStream {
     tokenized_string: PreTokenizedString,
-    current_token: i64,
+    current_token: usize,
 }
 
 impl From<PreTokenizedString> for PreTokenizedStream {
     fn from(s: PreTokenizedString) -> PreTokenizedStream {
         PreTokenizedStream {
             tokenized_string: s,
-            current_token: -1,
+            current_token: 0,
         }
     }
 }
@@ -41,49 +41,28 @@ impl From<PreTokenizedString> for PreTokenizedStream {
 impl PreTokenizedStream {
     /// Creates a TokenStream from PreTokenizedString array
     pub fn chain_tokenized_strings<'a>(
-        tok_strings: &'a [&'a PreTokenizedString],
-    ) -> BoxTokenStream {
-        if tok_strings.len() == 1 {
-            PreTokenizedStream::from((*tok_strings[0]).clone()).into()
-        } else {
-            let mut offsets = vec![];
-            let mut total_offset = 0;
-            for &tok_string in tok_strings {
-                offsets.push(total_offset);
-                if let Some(last_token) = tok_string.tokens.last() {
-                    total_offset += last_token.offset_to;
-                }
+        tok_strings: &'a [&PreTokenizedString],
+    ) -> impl Iterator<Item = Token> + 'a {
+        let streams_with_offsets = tok_strings.iter().scan(0, |total_offset, tok_string| {
+            let next = Some((
+                PreTokenizedStream::from((*tok_string).to_owned()),
+                *total_offset,
+            ));
+            if let Some(last_token) = tok_string.tokens.last() {
+                *total_offset += last_token.offset_to;
             }
-            // TODO remove the string cloning.
-            let token_streams: Vec<BoxTokenStream<'static>> = tok_strings
-                .iter()
-                .map(|&tok_string| PreTokenizedStream::from((*tok_string).clone()).into())
-                .collect();
-            TokenStreamChain::new(offsets, token_streams).into()
-        }
+            next
+        });
+        TokenStreamChain::new(streams_with_offsets)
     }
 }
 
-impl TokenStream for PreTokenizedStream {
-    fn advance(&mut self) -> bool {
+impl Iterator for PreTokenizedStream {
+    type Item = Token;
+    fn next(&mut self) -> Option<Token> {
+        let token = self.tokenized_string.tokens.get(self.current_token)?;
         self.current_token += 1;
-        self.current_token < self.tokenized_string.tokens.len() as i64
-    }
-
-    fn token(&self) -> &Token {
-        assert!(
-            self.current_token >= 0,
-            "TokenStream not initialized. You should call advance() at least once."
-        );
-        &self.tokenized_string.tokens[self.current_token as usize]
-    }
-
-    fn token_mut(&mut self) -> &mut Token {
-        assert!(
-            self.current_token >= 0,
-            "TokenStream not initialized. You should call advance() at least once."
-        );
-        &mut self.tokenized_string.tokens[self.current_token as usize]
+        Some(token.clone())
     }
 }
 
@@ -119,10 +98,9 @@ mod tests {
         let mut token_stream = PreTokenizedStream::from(tok_text.clone());
 
         for expected_token in tok_text.tokens {
-            assert!(token_stream.advance());
-            assert_eq!(token_stream.token(), &expected_token);
+            assert_eq!(token_stream.next().unwrap(), expected_token);
         }
-        assert!(!token_stream.advance());
+        assert!(token_stream.next().is_none());
     }
 
     #[test]
@@ -183,9 +161,8 @@ mod tests {
         ];
 
         for expected_token in expected_tokens {
-            assert!(token_stream.advance());
-            assert_eq!(token_stream.token(), &expected_token);
+            assert_eq!(token_stream.next().unwrap(), expected_token);
         }
-        assert!(!token_stream.advance());
+        assert!(token_stream.next().is_none());
     }
 }

src/tokenizer/tokenizer.rs

@@ -2,8 +2,23 @@ use crate::tokenizer::TokenStreamChain;
 use serde::{Deserialize, Serialize};
 /// The tokenizer module contains all of the tools used to process
 /// text in `tantivy`.
-use std::borrow::{Borrow, BorrowMut};
-use std::ops::{Deref, DerefMut};
+
+pub trait TextAnalyzerClone {
+    fn box_clone(&self) -> Box<dyn TextAnalyzerT>;
+}
+
+/// 'Top-level' trait hiding concrete types, below which static dispatch occurs.
+pub trait TextAnalyzerT: 'static + Send + Sync + TextAnalyzerClone {
+    /// 'Top-level' dynamic dispatch function hiding concrete types of the staticly
+    /// dispatched `token_stream` from the `Tokenizer` trait.
+    fn token_stream(&self, text: &str) -> Box<dyn Iterator<Item = Token>>;
+}
+
+impl Clone for Box<dyn TextAnalyzerT> {
+    fn clone(&self) -> Self {
+        (**self).box_clone()
+    }
+}
 
 /// Token
 #[derive(Debug, Clone, Serialize, Deserialize, Eq, PartialEq)]
@@ -35,35 +50,116 @@ impl Default for Token {
     }
 }
 
-/// `TextAnalyzer` tokenizes an input text into tokens and modifies the resulting `TokenStream`.
+/// Trait for the pluggable components of `Tokenizer`s.
+pub trait TokenFilter: 'static + Send + Sync + Clone {
+    /// Take a `Token` and transform it or return `None` if it's to be removed
+    /// from the output stream.
+    fn transform(&mut self, token: Token) -> Option<Token>;
+}
+
+/// `Tokenizer` are in charge of splitting text into a stream of token
+/// before indexing.
+///
+/// See the [module documentation](./index.html) for more detail.
+pub trait Tokenizer: 'static + Send + Sync + Clone {
+    /// An iteratable type is returned.
+    type Iter: Iterator<Item = Token>;
+    /// Creates a token stream for a given `str`.
+    fn token_stream(&self, text: &str) -> Self::Iter;
+    /// Tokenize an array`&str`
+    ///
+    /// The resulting `Token` stream 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<'a>(&'a self, texts: &'a [&str]) -> Box<dyn Iterator<Item = Token> + 'a> {
+        let streams_with_offsets = texts.iter().scan(0, move |total_offset, &text| {
+            let temp = *total_offset;
+            *total_offset += text.len();
+            Some((self.token_stream(text), temp))
+        });
+        Box::new(TokenStreamChain::new(streams_with_offsets))
+    }
+}
+
+/// `TextAnalyzer` wraps the tokenization of an input text and its modification by any filters applied onto it.
 ///
 /// It simply wraps a `Tokenizer` and a list of `TokenFilter` that are applied sequentially.
-pub struct TextAnalyzer {
-    tokenizer: Box<dyn Tokenizer>,
-    token_filters: Vec<BoxTokenFilter>,
-}
+#[derive(Clone, Debug, Default)]
+pub struct TextAnalyzer<T>(T);
 
-impl<T: Tokenizer> From<T> for TextAnalyzer {
-    fn from(tokenizer: T) -> Self {
-        TextAnalyzer::new(tokenizer, Vec::new())
+impl<T: Tokenizer> From<T> for TextAnalyzer<T> {
+    fn from(src: T) -> TextAnalyzer<T> {
+        TextAnalyzer(src)
     }
 }
 
-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,
+impl<T: Tokenizer> TextAnalyzerClone for TextAnalyzer<T> {
+    fn box_clone(&self) -> Box<dyn TextAnalyzerT> {
+        Box::new(TextAnalyzer(self.0.clone()))
+    }
+}
+
+impl<T: Tokenizer> TextAnalyzerT for TextAnalyzer<T> {
+    fn token_stream(&self, text: &str) -> Box<dyn Iterator<Item = Token>> {
+        Box::new(self.0.token_stream(text))
+    }
+}
+
+/// Identity `TokenFilter`
+#[derive(Clone, Debug, Default)]
+pub struct Identity;
+
+impl TokenFilter for Identity {
+    fn transform(&mut self, token: Token) -> Option<Token> {
+        Some(token)
+    }
+}
+
+/// `Filter` is a wrapper around a `Token` stream and a `TokenFilter` which modifies it.
+#[derive(Clone, Default, Debug)]
+pub struct Filter<I, F> {
+    iter: I,
+    f: F,
+}
+
+impl<I, F> Iterator for Filter<I, F>
+where
+    I: Iterator<Item = Token>,
+    F: TokenFilter,
+{
+    type Item = Token;
+    fn next(&mut self) -> Option<Token> {
+        while let Some(token) = self.iter.next() {
+            if let Some(tok) = self.f.transform(token) {
+                return Some(tok);
+            }
         }
+        None
     }
+}
 
+#[derive(Clone, Debug, Default)]
+pub struct AnalyzerBuilder<T, F> {
+    tokenizer: T,
+    f: F,
+}
+
+/// Construct an `AnalyzerBuilder` on which to apply `TokenFilter`.
+pub fn analyzer_builder<T: Tokenizer>(tokenizer: T) -> AnalyzerBuilder<T, Identity> {
+    AnalyzerBuilder {
+        tokenizer,
+        f: Identity,
+    }
+}
+
+impl<T, F> AnalyzerBuilder<T, F>
+where
+    T: Tokenizer,
+    F: TokenFilter,
+{
     /// Appends a token filter to the current tokenizer.
     ///
-    /// The method consumes the current `TokenStream` and returns a
+    /// The method consumes the current `Token` and returns a
     /// new one.
     ///
     /// # Example
@@ -71,248 +167,35 @@ impl TextAnalyzer {
     /// ```rust
     /// use tantivy::tokenizer::*;
     ///
-    /// let en_stem = TextAnalyzer::from(SimpleTokenizer)
+    /// let en_stem = analyzer_builder(SimpleTokenizer)
     ///     .filter(RemoveLongFilter::limit(40))
-    ///     .filter(LowerCaser)
-    ///     .filter(Stemmer::default());
+    ///     .filter(LowerCaser::new())
+    ///     .filter(Stemmer::default()).build();
     /// ```
     ///
-    pub fn filter<F: Into<BoxTokenFilter>>(mut self, token_filter: F) -> Self {
-        self.token_filters.push(token_filter.into());
-        self
+    pub fn filter<G: TokenFilter>(self, f: G) -> AnalyzerBuilder<AnalyzerBuilder<T, F>, G> {
+        AnalyzerBuilder { tokenizer: self, f }
     }
+    /// Finalize the build process.
+    pub fn build(self) -> TextAnalyzer<AnalyzerBuilder<T, F>> {
+        TextAnalyzer(self)
+    }
+}
 
-    /// Tokenize an array`&str`
-    ///
-    /// 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<'a>(&self, texts: &'a [&'a str]) -> BoxTokenStream<'a> {
-        assert!(!texts.is_empty());
-        if texts.len() == 1 {
-            self.token_stream(texts[0])
-        } else {
-            let mut offsets = vec![];
-            let mut total_offset = 0;
-            for &text in texts {
-                offsets.push(total_offset);
-                total_offset += text.len();
-            }
-            let token_streams: Vec<BoxTokenStream<'a>> = texts
-                .iter()
-                .cloned()
-                .map(|text| self.token_stream(text))
-                .collect();
-            From::from(TokenStreamChain::new(offsets, token_streams))
+impl<T: Tokenizer, F: TokenFilter> Tokenizer for AnalyzerBuilder<T, F> {
+    type Iter = Filter<T::Iter, F>;
+    fn token_stream(&self, text: &str) -> Self::Iter {
+        Filter {
+            iter: self.tokenizer.token_stream(text),
+            f: self.f.clone(),
         }
     }
-
-    /// Creates a token stream for a given `str`.
-    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 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<'b>(&'b self) -> &'b Token {
-        let token_stream: &'b (dyn TokenStream + 'a) = self.borrow();
-        token_stream.token()
-    }
-
-    fn token_mut<'b>(&'b mut self) -> &'b mut Token {
-        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.
-///
-/// # Example
-///
-/// ```
-/// use tantivy::tokenizer::*;
-///
-/// let tokenizer = TextAnalyzer::from(SimpleTokenizer)
-///        .filter(RemoveLongFilter::limit(40))
-///        .filter(LowerCaser);
-/// let mut token_stream = tokenizer.token_stream("Hello, happy tax payer");
-/// {
-///     let token = token_stream.next().unwrap();
-///     assert_eq!(&token.text, "hello");
-///     assert_eq!(token.offset_from, 0);
-///     assert_eq!(token.offset_to, 5);
-///     assert_eq!(token.position, 0);
-/// }
-/// {
-///     let token = token_stream.next().unwrap();
-///     assert_eq!(&token.text, "happy");
-///     assert_eq!(token.offset_from, 7);
-///     assert_eq!(token.offset_to, 12);
-///     assert_eq!(token.position, 1);
-/// }
-/// ```
-///
-pub trait TokenStream {
-    /// Advance to the next token
-    ///
-    /// Returns false if there are no other tokens.
-    fn advance(&mut self) -> bool;
-
-    /// Returns a reference to the current token.
-    fn token(&self) -> &Token;
-
-    /// Returns a mutable reference to the current token.
-    fn token_mut(&mut self) -> &mut Token;
-
-    /// Helper to iterate over tokens. It
-    /// simply combines a call to `.advance()`
-    /// and `.token()`.
-    ///
-    /// ```
-    /// use tantivy::tokenizer::*;
-    ///
-    /// let tokenizer = TextAnalyzer::from(SimpleTokenizer)
-    ///       .filter(RemoveLongFilter::limit(40))
-    ///       .filter(LowerCaser);
-    /// let mut token_stream = tokenizer.token_stream("Hello, happy tax payer");
-    /// while let Some(token) = token_stream.next() {
-    ///     println!("Token {:?}", token.text);
-    /// }
-    /// ```
-    fn next(&mut self) -> Option<&Token> {
-        if self.advance() {
-            Some(self.token())
-        } else {
-            None
-        }
-    }
-
-    /// 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() {
-            sink(self.token());
-            num_tokens_pushed += 1u32;
-        }
-        num_tokens_pushed
-    }
-}
-
-pub trait TokenFilterClone {
-    fn box_clone(&self) -> BoxTokenFilter;
-}
-
-/// Trait for the pluggable components of `Tokenizer`s.
-pub trait TokenFilter: 'static + Send + Sync + TokenFilterClone {
-    /// Wraps a token stream and returns the modified one.
-    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)]
 mod test {
-    use super::Token;
+    use super::*;
+    use crate::tokenizer::SimpleTokenizer;
 
     #[test]
     fn clone() {
@@ -330,4 +213,15 @@ mod test {
         assert_eq!(t1.offset_to, t2.offset_to);
         assert_eq!(t1.text, t2.text);
     }
+
+    #[test]
+    fn text_analyzer() {
+        let mut stream = SimpleTokenizer.token_stream("tokenizer hello world");
+        dbg!(stream.next());
+        dbg!(stream.next());
+        dbg!(stream.next());
+        dbg!(stream.next());
+        dbg!(stream.next());
+        dbg!(stream.next());
+    }
 }

src/tokenizer/tokenizer_manager.rs

@@ -1,5 +1,5 @@
 use crate::tokenizer::stemmer::Language;
-use crate::tokenizer::tokenizer::TextAnalyzer;
+use crate::tokenizer::tokenizer::{analyzer_builder, TextAnalyzer, TextAnalyzerT, Tokenizer};
 use crate::tokenizer::LowerCaser;
 use crate::tokenizer::RawTokenizer;
 use crate::tokenizer::RemoveLongFilter;
@@ -22,24 +22,23 @@ use std::sync::{Arc, RwLock};
 ///  search engine.
 #[derive(Clone)]
 pub struct TokenizerManager {
-    tokenizers: Arc<RwLock<HashMap<String, TextAnalyzer>>>,
+    tokenizers: Arc<RwLock<HashMap<String, Box<dyn TextAnalyzerT>>>>,
 }
 
 impl TokenizerManager {
     /// Registers a new tokenizer associated with a given name.
-    pub fn register<T>(&self, tokenizer_name: &str, tokenizer: T)
+    pub fn register<U: Tokenizer, T>(&self, tokenizer_name: &str, tokenizer: T)
     where
-        TextAnalyzer: From<T>,
+        T: Into<TextAnalyzer<U>>,
     {
-        let boxed_tokenizer: TextAnalyzer = TextAnalyzer::from(tokenizer);
         self.tokenizers
             .write()
             .expect("Acquiring the lock should never fail")
-            .insert(tokenizer_name.to_string(), boxed_tokenizer);
+            .insert(tokenizer_name.to_string(), Box::new(tokenizer.into()));
     }
 
     /// Accessing a tokenizer given its name.
-    pub fn get(&self, tokenizer_name: &str) -> Option<TextAnalyzer> {
+    pub fn get(&self, tokenizer_name: &str) -> Option<Box<dyn TextAnalyzerT>> {
         self.tokenizers
             .read()
             .expect("Acquiring the lock should never fail")
@@ -54,23 +53,25 @@ impl Default for TokenizerManager {
     /// - simple
     /// - en_stem
     /// - ja
-    fn default() -> TokenizerManager {
+    fn default() -> Self {
         let manager = TokenizerManager {
             tokenizers: Arc::new(RwLock::new(HashMap::new())),
         };
         manager.register("raw", RawTokenizer);
         manager.register(
             "default",
-            TextAnalyzer::from(SimpleTokenizer)
+            analyzer_builder(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
-                .filter(LowerCaser),
+                .filter(LowerCaser::new())
+                .build(),
         );
         manager.register(
             "en_stem",
-            TextAnalyzer::from(SimpleTokenizer)
+            analyzer_builder(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
-                .filter(LowerCaser)
-                .filter(Stemmer::new(Language::English)),
+                .filter(LowerCaser::new())
+                .filter(Stemmer::new(Language::English))
+                .build(),
         );
         manager
     }