Reverting atomic_write to the atomic_writes in order to address #866

Accept dash in field names and enforce field names constraint at the
creation of the schema.

Closes #796

.gitignore

@@ -12,3 +12,4 @@ cpp/simdcomp/bitpackingbenchmark
 *.bk
 .idea
 trace.dat
+cargo-timing*

CHANGELOG.md

@@ -1,3 +1,7 @@
+Tantivy 0.14.0
+=========================
+- Remove dependency to atomicwrites #833 .Implemented by @pmasurel upon suggestion and research from @asafigan). 
+
 Tantivy 0.13.0
 ======================
 Tantivy 0.13 introduce a change in the index format that will require

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.13.0"
+version = "0.14.0-dev"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -22,8 +22,8 @@ tantivy-fst = "0.3"
 memmap = {version = "0.7", optional=true}
 lz4 = {version="1.20", optional=true}
 snap = "1"
-atomicwrites = {version="0.2.2", optional=true}
+tempfile = {version="3.0", optional=true}
-tempfile = "3.0"
+atomicwrites = "0.2"
 log = "0.4"
 serde = {version="1.0", features=["derive"]}
 serde_json = "1.0"
@@ -38,7 +38,7 @@ owning_ref = "0.4"
 stable_deref_trait = "1.0.0"
 rust-stemmers = "1.2"
 downcast-rs = { version="1.0" }
-tantivy-query-grammar = { version="0.13", path="./query-grammar" }
+tantivy-query-grammar = { version="0.14.0-dev", path="./query-grammar" }
 bitpacking = {version="0.8", default-features = false, features=["bitpacker4x"]}
 census = "0.4"
 fnv = "1.0.6"
@@ -75,7 +75,7 @@ overflow-checks = true
 
 [features]
 default = ["mmap"]
-mmap = ["atomicwrites", "fs2", "memmap", "notify"]
+mmap = ["fs2", "tempfile", "memmap", "notify"]
 lz4-compression = ["lz4"]
 failpoints = ["fail/failpoints"]
 unstable = [] # useful for benches.

README.md

@@ -34,11 +34,6 @@ Tantivy is, in fact, strongly inspired by Lucene's design.
 The following [benchmark](https://tantivy-search.github.io/bench/) break downs 
 performance for different type of queries / collection.
 
-
-In general, Tantivy tends to be 
-- slower than Lucene on union with a Top-K due to Block-WAND optimization.
-- faster than Lucene on intersection and phrase queries. 
-
 Your mileage WILL vary depending on the nature of queries and their load.
 
 # Features

examples/basic_search.rs

@@ -112,18 +112,6 @@ fn main() -> tantivy::Result<()> {
             limbs and branches that arch over the pool"
     ));
 
-    index_writer.add_document(doc!(
-    title => "Of Mice and Men",
-    body => "A few miles south of Soledad, the Salinas River drops in close to the hillside \
-            bank and runs deep and green. The water is warm too, for it has slipped twinkling \
-            over the yellow sands in the sunlight before reaching the narrow pool. On one \
-            side of the river the golden foothill slopes curve up to the strong and rocky \
-            Gabilan Mountains, but on the valley side the water is lined with trees—willows \
-            fresh and green with every spring, carrying in their lower leaf junctures the \
-            debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
-            limbs and branches that arch over the pool"
-    ));
-
     // Multivalued field just need to be repeated.
     index_writer.add_document(doc!(
     title => "Frankenstein",

query-grammar/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy-query-grammar"
-version = "0.13.0"
+version = "0.14.0-dev"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]

query-grammar/src/occur.rs

@@ -52,7 +52,7 @@ mod test {
     use crate::Occur;
 
     #[test]
-    fn test_Occur_compose() {
+    fn test_occur_compose() {
         assert_eq!(Occur::compose(Occur::Should, Occur::Should), Occur::Should);
         assert_eq!(Occur::compose(Occur::Should, Occur::Must), Occur::Must);
         assert_eq!(

query-grammar/src/query_grammar.rs

@@ -9,8 +9,10 @@ use combine::{
 
 fn field<'a>() -> impl Parser<&'a str, Output = String> {
     (
-        letter(),
+        (letter().or(char('_'))),
-        many(satisfy(|c: char| c.is_alphanumeric() || c == '_')),
+        many(satisfy(|c: char| {
+            c.is_alphanumeric() || c == '_' || c == '-'
+        })),
     )
         .skip(char(':'))
         .map(|(s1, s2): (char, String)| format!("{}{}", s1, s2))
@@ -279,6 +281,8 @@ pub fn parse_to_ast<'a>() -> impl Parser<&'a str, Output = UserInputAST> {
 #[cfg(test)]
 mod test {
 
+    type TestParseResult = Result<(), StringStreamError>;
+
     use super::*;
     use combine::parser::Parser;
 
@@ -296,9 +300,10 @@ mod test {
     }
 
     #[test]
-    fn test_occur_symbol() {
+    fn test_occur_symbol() -> TestParseResult {
-        assert_eq!(super::occur_symbol().parse("-"), Ok((Occur::MustNot, "")));
+        assert_eq!(super::occur_symbol().parse("-")?, (Occur::MustNot, ""));
-        assert_eq!(super::occur_symbol().parse("+"), Ok((Occur::Must, "")));
+        assert_eq!(super::occur_symbol().parse("+")?, (Occur::Must, ""));
+        Ok(())
     }
 
     #[test]
@@ -410,6 +415,25 @@ mod test {
         assert_eq!(format!("{:?}", ast), "\"abc\"");
     }
 
+    #[test]
+    fn test_field_name() -> TestParseResult {
+        assert_eq!(
+            super::field().parse("my-field-name:a")?,
+            ("my-field-name".to_string(), "a")
+        );
+        assert_eq!(
+            super::field().parse("my_field_name:a")?,
+            ("my_field_name".to_string(), "a")
+        );
+        assert!(super::field().parse(":a").is_err());
+        assert!(super::field().parse("-my_field:a").is_err());
+        assert_eq!(
+            super::field().parse("_my_field:a")?,
+            ("_my_field".to_string(), "a")
+        );
+        Ok(())
+    }
+
     #[test]
     fn test_range_parser() {
         // testing the range() parser separately

src/core/index.rs

@@ -539,7 +539,6 @@ mod tests {
             test_index_on_commit_reload_policy_aux(field, &write_index, &reader);
         }
     }
-
     fn test_index_on_commit_reload_policy_aux(field: Field, index: &Index, reader: &IndexReader) {
         let mut reader_index = reader.index();
         let (sender, receiver) = crossbeam::channel::unbounded();
@@ -550,12 +549,23 @@ mod tests {
         assert_eq!(reader.searcher().num_docs(), 0);
         writer.add_document(doc!(field=>1u64));
         writer.commit().unwrap();
+        // We need a loop here because it is possible for notify to send more than
+        // one modify event. It was observed on CI on MacOS.
+        loop {
             assert!(receiver.recv().is_ok());
-        assert_eq!(reader.searcher().num_docs(), 1);
+            if reader.searcher().num_docs() == 1 {
+                break;
+            }
+        }
         writer.add_document(doc!(field=>2u64));
         writer.commit().unwrap();
+        // ... Same as above
+        loop {
             assert!(receiver.recv().is_ok());
-        assert_eq!(reader.searcher().num_docs(), 2);
+            if reader.searcher().num_docs() == 2 {
+                break;
+            }
+        }
     }
 
     // This test will not pass on windows, because windows

src/directory/managed_directory.rs

@@ -1,4 +1,4 @@
-use crate::core::MANAGED_FILEPATH;
+use crate::core::{MANAGED_FILEPATH, META_FILEPATH};
 use crate::directory::error::{DeleteError, IOError, LockError, OpenReadError, OpenWriteError};
 use crate::directory::footer::{Footer, FooterProxy};
 use crate::directory::DirectoryLock;
@@ -246,13 +246,15 @@ impl ManagedDirectory {
     /// List files for which checksum does not match content
     pub fn list_damaged(&self) -> result::Result<HashSet<PathBuf>, OpenReadError> {
         let mut hashset = HashSet::new();
-        let managed_paths = self
+        let mut managed_paths = self
             .meta_informations
             .read()
             .expect("Managed directory rlock poisoned in list damaged.")
             .managed_paths
             .clone();
 
+        managed_paths.remove(*META_FILEPATH);
+
         for path in managed_paths.into_iter() {
             if !self.validate_checksum(&path)? {
                 hashset.insert(path);

src/directory/mmap_directory.rs

@@ -1,4 +1,5 @@
 use crate::core::META_FILEPATH;
+use atomicwrites;
 use crate::directory::error::LockError;
 use crate::directory::error::{
     DeleteError, IOError, OpenDirectoryError, OpenReadError, OpenWriteError,
@@ -34,6 +35,7 @@ use std::sync::Mutex;
 use std::sync::RwLock;
 use std::sync::Weak;
 use std::thread;
+use tempfile;
 use tempfile::TempDir;
 
 /// Create a default io error given a string.
@@ -487,11 +489,11 @@ impl Directory for MmapDirectory {
         }
     }
 
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
+    fn atomic_write(&mut self, path: &Path, content: &[u8]) -> io::Result<()> {
         debug!("Atomic Write {:?}", path);
         let full_path = self.resolve_path(path);
         let meta_file = atomicwrites::AtomicFile::new(full_path, atomicwrites::AllowOverwrite);
-        meta_file.write(|f| f.write_all(data))?;
+        meta_file.write(|f| f.write_all(content))?;
         Ok(())
     }
 

src/directory/tests.rs

@@ -211,19 +211,18 @@ fn test_watch(directory: &mut dyn Directory) {
         .unwrap();
 
     for i in 0..10 {
-        assert_eq!(i, counter.load(SeqCst));
+        assert!(i <= counter.load(SeqCst));
         assert!(directory
             .atomic_write(Path::new("meta.json"), b"random_test_data_2")
             .is_ok());
-        assert_eq!(receiver.recv_timeout(Duration::from_millis(500)), Ok(i));
+        assert!(i + 1 <= counter.load(SeqCst)); // notify can trigger more than once.
-        assert_eq!(i + 1, counter.load(SeqCst));
     }
     mem::drop(watch_handle);
     assert!(directory
         .atomic_write(Path::new("meta.json"), b"random_test_data")
         .is_ok());
     assert!(receiver.recv_timeout(Duration::from_millis(500)).is_ok());
-    assert_eq!(10, counter.load(SeqCst));
+    assert!(10 <= counter.load(SeqCst));
 }
 
 fn test_lock_non_blocking(directory: &mut dyn Directory) {

src/indexer/mod.rs

@@ -29,8 +29,9 @@ pub use self::segment_writer::SegmentWriter;
 /// Alias for the default merge policy, which is the `LogMergePolicy`.
 pub type DefaultMergePolicy = LogMergePolicy;
 
+#[cfg(feature = "mmap")]
 #[cfg(test)]
-mod tests {
+mod tests_mmap {
     use crate::schema::{self, Schema};
     use crate::{Index, Term};
 

src/lib.rs

@@ -1012,4 +1012,12 @@ mod tests {
             DOC_COUNT as usize
         );
     }
+
+    #[test]
+    fn test_validate_checksum() {
+        let index_path = tempfile::tempdir().expect("dir");
+        let schema = Schema::builder().build();
+        let index = Index::create_in_dir(&index_path, schema).expect("index");
+        assert!(index.validate_checksum().unwrap().is_empty());
+    }
 }

src/query/boolean_query/block_wand.rs

@@ -4,19 +4,6 @@ use crate::{DocId, DocSet, Score, TERMINATED};
 use std::ops::Deref;
 use std::ops::DerefMut;
 
-fn is_sorted<I: Iterator<Item = DocId>>(mut it: I) -> bool {
-    if let Some(first) = it.next() {
-        let mut prev = first;
-        for doc in it {
-            if doc < prev {
-                return false;
-            }
-            prev = doc;
-        }
-    }
-    true
-}
-
 /// Takes a term_scorers sorted by their current doc() and a threshold and returns
 /// Returns (pivot_len, pivot_ord) defined as follows:
 /// - `pivot_doc` lowest document that has a chance of exceeding (>) the threshold score.
@@ -55,37 +42,12 @@ fn find_pivot_doc(
     Some((before_pivot_len, pivot_len, pivot_doc))
 }
 
-struct TermScorerWithMaxScore<'a> {
-    scorer: &'a mut TermScorer,
-    max_score: Score,
-}
-
-impl<'a> From<&'a mut TermScorer> for TermScorerWithMaxScore<'a> {
-    fn from(scorer: &'a mut TermScorer) -> Self {
-        let max_score = scorer.max_score();
-        TermScorerWithMaxScore { scorer, max_score }
-    }
-}
-
-impl<'a> Deref for TermScorerWithMaxScore<'a> {
-    type Target = TermScorer;
-
-    fn deref(&self) -> &Self::Target {
-        self.scorer
-    }
-}
-
-impl<'a> DerefMut for TermScorerWithMaxScore<'a> {
-    fn deref_mut(&mut self) -> &mut Self::Target {
-        self.scorer
-    }
-}
-
 // Before and after calling this method, scorers need to be sorted by their `.doc()`.
 fn block_max_was_too_low_advance_one_scorer(
     scorers: &mut Vec<TermScorerWithMaxScore>,
     pivot_len: usize,
 ) {
+    debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
     let mut scorer_to_seek = pivot_len - 1;
     let mut doc_to_seek_after = scorers[scorer_to_seek].doc();
     for scorer_ord in (0..pivot_len - 1).rev() {
@@ -102,6 +64,7 @@ fn block_max_was_too_low_advance_one_scorer(
     }
     scorers[scorer_to_seek].seek(doc_to_seek_after + 1);
     restore_ordering(scorers, scorer_to_seek);
+    debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
 }
 
 // Given a list of term_scorers and a `ord` and assuming that `term_scorers[ord]` is sorted
@@ -177,12 +140,12 @@ pub fn block_wand(
         .map(TermScorerWithMaxScore::from)
         .collect();
     scorers.sort_by_key(|scorer| scorer.doc());
-    loop {
     // At this point we need to ensure that the scorers are sorted!
     debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
-        if let Some((before_pivot_len, pivot_len, pivot_doc)) =
+    while let Some((before_pivot_len, pivot_len, pivot_doc)) =
         find_pivot_doc(&scorers[..], threshold)
     {
+        debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
         debug_assert_ne!(pivot_doc, TERMINATED);
         debug_assert!(before_pivot_len < pivot_len);
 
@@ -229,12 +192,47 @@ pub fn block_wand(
         }
         // let's advance all of the scorers that are currently positioned on the pivot.
         advance_all_scorers_on_pivot(&mut scorers, pivot_len);
-        } else {
-            return;
-        }
     }
 }
 
+struct TermScorerWithMaxScore<'a> {
+    scorer: &'a mut TermScorer,
+    max_score: Score,
+}
+
+impl<'a> From<&'a mut TermScorer> for TermScorerWithMaxScore<'a> {
+    fn from(scorer: &'a mut TermScorer) -> Self {
+        let max_score = scorer.max_score();
+        TermScorerWithMaxScore { scorer, max_score }
+    }
+}
+
+impl<'a> Deref for TermScorerWithMaxScore<'a> {
+    type Target = TermScorer;
+
+    fn deref(&self) -> &Self::Target {
+        self.scorer
+    }
+}
+
+impl<'a> DerefMut for TermScorerWithMaxScore<'a> {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        self.scorer
+    }
+}
+
+fn is_sorted<I: Iterator<Item = DocId>>(mut it: I) -> bool {
+    if let Some(first) = it.next() {
+        let mut prev = first;
+        for doc in it {
+            if doc < prev {
+                return false;
+            }
+            prev = doc;
+        }
+    }
+    true
+}
 #[cfg(test)]
 mod tests {
     use crate::query::score_combiner::SumCombiner;
@@ -248,17 +246,21 @@ mod tests {
     use std::iter;
 
     struct Float(Score);
+
     impl Eq for Float {}
+
     impl PartialEq for Float {
         fn eq(&self, other: &Self) -> bool {
             self.cmp(&other) == Ordering::Equal
         }
     }
+
     impl PartialOrd for Float {
         fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
             Some(self.cmp(other))
         }
     }
+
     impl Ord for Float {
         fn cmp(&self, other: &Self) -> Ordering {
             other.0.partial_cmp(&self.0).unwrap_or(Ordering::Equal)

src/query/term_query/term_weight.rs

@@ -4,7 +4,7 @@ use crate::docset::DocSet;
 use crate::postings::SegmentPostings;
 use crate::query::bm25::BM25Weight;
 use crate::query::explanation::does_not_match;
-use crate::query::weight::{for_each_pruning_scorer, for_each_scorer};
+use crate::query::weight::for_each_scorer;
 use crate::query::Weight;
 use crate::query::{Explanation, Scorer};
 use crate::schema::IndexRecordOption;
@@ -73,8 +73,8 @@ impl Weight for TermWeight {
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
-        let mut scorer = self.scorer(reader, 1.0)?;
+        let scorer = self.specialized_scorer(reader, 1.0)?;
-        for_each_pruning_scorer(&mut scorer, threshold, callback);
+        crate::query::boolean_query::block_wand(vec![scorer], threshold, callback);
         Ok(())
     }
 }

src/reader/mod.rs

@@ -138,8 +138,10 @@ impl InnerIndexReader {
                 .collect::<crate::Result<_>>()?
         };
         let schema = self.index.schema();
-        let searchers = (0..self.num_searchers)
+        let searchers = std::iter::repeat_with(|| {
-            .map(|_| Searcher::new(schema.clone(), self.index.clone(), segment_readers.clone()))
+            Searcher::new(schema.clone(), self.index.clone(), segment_readers.clone())
+        })
+        .take(self.num_searchers)
         .collect();
         self.searcher_pool.publish_new_generation(searchers);
         Ok(())

src/schema/field_entry.rs

@@ -1,5 +1,5 @@
-use crate::schema::IntOptions;
 use crate::schema::TextOptions;
+use crate::schema::{is_valid_field_name, IntOptions};
 
 use crate::schema::FieldType;
 use serde::de::{self, MapAccess, Visitor};
@@ -24,6 +24,7 @@ impl FieldEntry {
     /// Creates a new u64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_text(field_name: String, text_options: TextOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Str(text_options),
@@ -33,6 +34,7 @@ impl FieldEntry {
     /// Creates a new u64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_u64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::U64(field_type),
@@ -42,6 +44,7 @@ impl FieldEntry {
     /// Creates a new i64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_i64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::I64(field_type),
@@ -51,6 +54,7 @@ impl FieldEntry {
     /// Creates a new f64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_f64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::F64(field_type),
@@ -60,6 +64,7 @@ impl FieldEntry {
     /// Creates a new date field entry in the schema, given
     /// a name, and some options.
     pub fn new_date(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Date(field_type),
@@ -68,6 +73,7 @@ impl FieldEntry {
 
     /// Creates a field entry for a facet.
     pub fn new_facet(field_name: String) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::HierarchicalFacet,
@@ -76,6 +82,7 @@ impl FieldEntry {
 
     /// Creates a field entry for a bytes field
     pub fn new_bytes(field_name: String) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Bytes,
@@ -268,6 +275,12 @@ mod tests {
     use crate::schema::TEXT;
     use serde_json;
 
+    #[test]
+    #[should_panic]
+    fn test_invalid_field_name_should_panic() {
+        FieldEntry::new_text("-hello".to_string(), TEXT);
+    }
+
     #[test]
     fn test_json_serialization() {
         let field_value = FieldEntry::new_text(String::from("title"), TEXT);

src/schema/mod.rs

@@ -149,14 +149,16 @@ pub use self::int_options::IntOptions;
 use once_cell::sync::Lazy;
 use regex::Regex;
 
+/// Regular expression representing the restriction on a valid field names.
+pub const FIELD_NAME_PATTERN: &'static str = r#"^[_a-zA-Z][_\-a-zA-Z0-9]*$"#;
+
 /// Validator for a potential `field_name`.
 /// Returns true iff the name can be use for a field name.
 ///
 /// A field name must start by a letter `[a-zA-Z]`.
 /// The other characters can be any alphanumic character `[a-ZA-Z0-9]` or `_`.
 pub fn is_valid_field_name(field_name: &str) -> bool {
-    static FIELD_NAME_PTN: Lazy<Regex> =
+    static FIELD_NAME_PTN: Lazy<Regex> = Lazy::new(|| Regex::new(FIELD_NAME_PATTERN).unwrap());
-        Lazy::new(|| Regex::new("^[a-zA-Z][_a-zA-Z0-9]*$").unwrap());
     FIELD_NAME_PTN.is_match(field_name)
 }
 
@@ -170,6 +172,11 @@ mod tests {
         assert!(is_valid_field_name("text"));
         assert!(is_valid_field_name("text0"));
         assert!(!is_valid_field_name("0text"));
+        assert!(is_valid_field_name("field-name"));
+        assert!(is_valid_field_name("field_name"));
+        assert!(!is_valid_field_name("field!name"));
+        assert!(!is_valid_field_name("-fieldname"));
+        assert!(is_valid_field_name("_fieldname"));
         assert!(!is_valid_field_name(""));
         assert!(!is_valid_field_name("シャボン玉"));
         assert!(is_valid_field_name("my_text_field"));