Introduce the Attribute criterion

Cargo.lock

@@ -122,6 +122,12 @@ version = "0.13.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "904dfeac50f3cdaba28fc6f57fdcddb75f49ed61346676a78c4ffe55877802fd"
 
+[[package]]
+name = "big_s"
+version = "1.0.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "199edb7b90631283b10c2422e6a0bc8b7d987bf732995ba1de53b576c97e51a8"
+
 [[package]]
 name = "bincode"
 version = "1.3.1"
@@ -1251,6 +1257,7 @@ name = "milli"
 version = "0.1.1"
 dependencies = [
  "anyhow",
+ "big_s",
  "bstr",
  "byteorder",
  "chrono",

milli/Cargo.toml

@@ -52,6 +52,7 @@ logging_timer = "1.0.0"
 tinytemplate = "=1.1.0"
 
 [dev-dependencies]
+big_s = "1.0.2"
 criterion = "0.3.4"
 maplit = "1.0.2"
 rand = "0.8.3"

milli/src/search/criteria/attribute.rs

@@ -0,0 +1,133 @@
+use log::debug;
+use roaring::RoaringBitmap;
+
+use crate::search::criteria::Query;
+use crate::search::query_tree::Operation;
+use crate::search::WordDerivationsCache;
+use super::{Criterion, CriterionResult, Context};
+
+pub struct Attribute<'t> {
+    ctx: &'t dyn Context,
+    query_tree: Option<Operation>,
+    candidates: Option<RoaringBitmap>,
+    bucket_candidates: RoaringBitmap,
+    parent: Option<Box<dyn Criterion + 't>>,
+}
+
+impl<'t> Attribute<'t> {
+    pub fn initial(
+        ctx: &'t dyn Context,
+        query_tree: Option<Operation>,
+        candidates: Option<RoaringBitmap>,
+    ) -> Self
+    {
+        Attribute {
+            ctx,
+            query_tree,
+            candidates,
+            bucket_candidates: RoaringBitmap::new(),
+            parent: None,
+        }
+    }
+
+    pub fn new(ctx: &'t dyn Context, parent: Box<dyn Criterion + 't>) -> Self {
+        Attribute {
+            ctx,
+            query_tree: None,
+            candidates: None,
+            bucket_candidates: RoaringBitmap::new(),
+            parent: Some(parent),
+        }
+    }
+}
+
+impl<'t> Criterion for Attribute<'t> {
+    #[logging_timer::time("Attribute::{}")]
+    fn next(&mut self, wdcache: &mut WordDerivationsCache) -> anyhow::Result<Option<CriterionResult>> {
+        todo!("Attribute")
+    }
+}
+
+// TODO can we keep refs of Query
+fn explode_query_tree(query_tree: &Operation) -> Vec<Vec<Query>> {
+    use crate::search::criteria::Operation::{And, Or, Consecutive};
+
+    fn and_recurse(head: &Operation, tail: &[Operation]) -> Vec<Vec<Query>> {
+        match tail.split_first() {
+            Some((thead, tail)) => {
+                let tail = and_recurse(thead, tail);
+                let mut out = Vec::new();
+                for array in recurse(head) {
+                    for tail_array in &tail {
+                        let mut array = array.clone();
+                        array.extend(tail_array.iter().cloned());
+                        out.push(array);
+                    }
+                }
+                out
+            },
+            None => recurse(head),
+        }
+    }
+
+    fn recurse(op: &Operation) -> Vec<Vec<Query>> {
+        match op {
+            And(ops) | Consecutive(ops) => {
+                ops.split_first().map_or_else(Vec::new, |(h, t)| and_recurse(h, t))
+            },
+            Or(_, ops) => ops.into_iter().map(recurse).flatten().collect(),
+            Operation::Query(query) => vec![vec![query.clone()]],
+        }
+    }
+
+    recurse(query_tree)
+}
+
+#[cfg(test)]
+mod tests {
+    use big_s::S;
+
+    use crate::search::criteria::QueryKind;
+    use super::*;
+
+    #[test]
+    fn simple_explode_query_tree() {
+        let query_tree = Operation::Or(false, vec![
+            Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("manythefish")) }),
+            Operation::And(vec![
+                Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("manythe")) }),
+                Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("fish")) }),
+            ]),
+            Operation::And(vec![
+                Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("many")) }),
+                Operation::Or(false, vec![
+                    Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("thefish")) }),
+                    Operation::And(vec![
+                        Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("the")) }),
+                        Operation::Query(Query { prefix: false, kind: QueryKind::exact(S("fish")) }),
+                    ]),
+                ]),
+            ]),
+        ]);
+
+        let expected = vec![
+            vec![Query { prefix: false, kind: QueryKind::exact(S("manythefish")) }],
+            vec![
+                Query { prefix: false, kind: QueryKind::exact(S("manythe")) },
+                Query { prefix: false, kind: QueryKind::exact(S("fish")) },
+            ],
+            vec![
+                Query { prefix: false, kind: QueryKind::exact(S("many")) },
+                Query { prefix: false, kind: QueryKind::exact(S("thefish")) },
+            ],
+            vec![
+                Query { prefix: false, kind: QueryKind::exact(S("many")) },
+                Query { prefix: false, kind: QueryKind::exact(S("the")) },
+                Query { prefix: false, kind: QueryKind::exact(S("fish")) },
+            ],
+        ];
+
+        let result = explode_query_tree(&query_tree);
+        assert_eq!(expected, result);
+    }
+}

milli/src/search/criteria/mod.rs

@@ -12,12 +12,14 @@ use self::typo::Typo;
 use self::words::Words;
 use self::asc_desc::AscDesc;
 use self::proximity::Proximity;
+use self::attribute::Attribute;
 use self::fetcher::Fetcher;
 
 mod typo;
 mod words;
 mod asc_desc;
 mod proximity;
+mod attribute;
 pub mod fetcher;
 
 pub trait Criterion {
@@ -139,6 +141,7 @@ impl<'t> CriteriaBuilder<'t> {
                     Name::Typo => Box::new(Typo::new(self, father)),
                     Name::Words => Box::new(Words::new(self, father)),
                     Name::Proximity => Box::new(Proximity::new(self, father)),
+                    Name::Attribute => Box::new(Attribute::new(self, father)),
                     Name::Asc(field) => Box::new(AscDesc::asc(&self.index, &self.rtxn, father, field)?),
                     Name::Desc(field) => Box::new(AscDesc::desc(&self.index, &self.rtxn, father, field)?),
                     _otherwise => father,
@@ -147,6 +150,7 @@ impl<'t> CriteriaBuilder<'t> {
                     Name::Typo => Box::new(Typo::initial(self, query_tree.take(), facet_candidates.take())),
                     Name::Words => Box::new(Words::initial(self, query_tree.take(), facet_candidates.take())),
                     Name::Proximity => Box::new(Proximity::initial(self, query_tree.take(), facet_candidates.take())),
+                    Name::Attribute => Box::new(Attribute::initial(self, query_tree.take(), facet_candidates.take())),
                     Name::Asc(field) => {
                         Box::new(AscDesc::initial_asc(&self.index, &self.rtxn, query_tree.take(), facet_candidates.take(), field)?)
                     },