Closes #526 (#535)

4 files changed, 66 insertions, 6 deletions

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ab7f8cc..068b291 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,7 +7,7 @@ Minor
 - Small simplification of the code. 
 Calling .freq() or .doc() when .advance() has never 
 on segment postings should panic from now on.
-
+- Tokens exceeding `u16::max_value() - 4` chars are discarded silently instead of panicking.
 
 
 Tantivy 0.9.0
diff --git a/src/postings/mod.rs b/src/postings/mod.rs
index 3e1c034..edc9d6c 100644
--- a/src/postings/mod.rs
+++ b/src/postings/mod.rs
@@ -55,13 +55,15 @@ pub mod tests {
     use fieldnorm::FieldNormReader;
     use indexer::operation::AddOperation;
     use indexer::SegmentWriter;
+    use merge_policy::NoMergePolicy;
     use query::Scorer;
     use rand::rngs::StdRng;
     use rand::{Rng, SeedableRng};
-    use schema::Field;
-    use schema::IndexRecordOption;
     use schema::{Document, Schema, Term, INDEXED, STRING, TEXT};
+    use schema::{Field, TextOptions};
+    use schema::{IndexRecordOption, TextFieldIndexing};
     use std::iter;
+    use tokenizer::{SimpleTokenizer, MAX_TOKEN_LEN};
     use DocId;
     use Score;
 
@@ -161,6 +163,52 @@ pub mod tests {
     }
 
     #[test]
+    pub fn test_drop_token_that_are_too_long() {
+        let ok_token_text: String = iter::repeat('A').take(MAX_TOKEN_LEN).collect();
+        let mut exceeding_token_text: String = iter::repeat('A').take(MAX_TOKEN_LEN + 1).collect();
+        exceeding_token_text.push_str(" hello");
+        let mut schema_builder = Schema::builder();
+        let text_options = TextOptions::default().set_indexing_options(
+            TextFieldIndexing::default()
+                .set_index_option(IndexRecordOption::WithFreqsAndPositions)
+                .set_tokenizer("simple_no_truncation"),
+        );
+        let text_field = schema_builder.add_text_field("text", text_options);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema.clone());
+        index
+            .tokenizers()
+            .register("simple_no_truncation", SimpleTokenizer);
+        let reader = index.reader().unwrap();
+        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        index_writer.set_merge_policy(Box::new(NoMergePolicy));
+        {
+            index_writer.add_document(doc!(text_field=>exceeding_token_text));
+            index_writer.commit().unwrap();
+            reader.reload().unwrap();
+            let searcher = reader.searcher();
+            let segment_reader = searcher.segment_reader(0u32);
+            let inverted_index = segment_reader.inverted_index(text_field);
+            assert_eq!(inverted_index.terms().num_terms(), 1);
+            let mut bytes = vec![];
+            assert!(inverted_index.terms().ord_to_term(0, &mut bytes));
+            assert_eq!(&bytes, b"hello");
+        }
+        {
+            index_writer.add_document(doc!(text_field=>ok_token_text.clone()));
+            index_writer.commit().unwrap();
+            reader.reload().unwrap();
+            let searcher = reader.searcher();
+            let segment_reader = searcher.segment_reader(1u32);
+            let inverted_index = segment_reader.inverted_index(text_field);
+            assert_eq!(inverted_index.terms().num_terms(), 1);
+            let mut bytes = vec![];
+            assert!(inverted_index.terms().ord_to_term(0, &mut bytes));
+            assert_eq!(&bytes[..], ok_token_text.as_bytes());
+        }
+    }
+
+    #[test]
     pub fn test_position_and_fieldnorm1() {
         let mut positions = Vec::new();
         let mut schema_builder = Schema::builder();
diff --git a/src/postings/postings_writer.rs b/src/postings/postings_writer.rs
index 61bcc35..d263258 100644
--- a/src/postings/postings_writer.rs
+++ b/src/postings/postings_writer.rs
@@ -12,8 +12,8 @@ use std::io;
 use std::marker::PhantomData;
 use std::ops::DerefMut;
 use termdict::TermOrdinal;
-use tokenizer::Token;
 use tokenizer::TokenStream;
+use tokenizer::{Token, MAX_TOKEN_LEN};
 use DocId;
 use Result;
 
@@ -210,8 +210,11 @@ pub trait PostingsWriter {
     ) -> u32 {
         let mut term = Term::for_field(field);
         let mut sink = |token: &Token| {
-            term.set_text(token.text.as_str());
-            self.subscribe(term_index, doc_id, token.position as u32, &term, heap);
+            // We skip all tokens with a len greater than u16.
+            if token.text.len() <= MAX_TOKEN_LEN {
+                term.set_text(token.text.as_str());
+                self.subscribe(term_index, doc_id, token.position as u32, &term, heap);
+            }
         };
         token_stream.process(&mut sink)
     }
diff --git a/src/tokenizer/mod.rs b/src/tokenizer/mod.rs
index e07116f..f9b72a6 100644
--- a/src/tokenizer/mod.rs
+++ b/src/tokenizer/mod.rs
@@ -97,6 +97,8 @@
 //! If you built your schema programmatically, a complete example
 //! could like this for instance.
 //!
+//! Note that tokens with a len greater or equal to [`MAX_TOKEN_LEN`](./constant.MAX_TOKEN_LEN.html).
+//!
 //! # Example
 //!
 //! ```
@@ -157,6 +159,13 @@ pub use self::tokenizer::BoxedTokenizer;
 pub use self::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
 pub use self::tokenizer_manager::TokenizerManager;
 
+/// Maximum authorized len (in bytes) for a token.
+///
+/// Tokenizer are in charge of not emitting tokens larger than this value.
+/// Currently, if a faulty tokenizer implementation emits tokens with a length larger than
+/// `2^16 - 1 - 4`, the token will simply be ignored downstream.
+pub const MAX_TOKEN_LEN: usize = u16::max_value() as usize - 4;
+
 #[cfg(test)]
 pub mod tests {
     use super::{