Support nested primary keys

milli/src/documents/primary_key.rs

@@ -1,5 +1,6 @@
 use std::borrow::Cow;
 use std::iter;
+use std::ops::ControlFlow;
 use std::result::Result as StdResult;
 
 use bumpalo::Bump;
@@ -7,7 +8,7 @@ use serde_json::value::RawValue;
 use serde_json::{from_str, Value};
 
 use crate::fields_ids_map::MutFieldIdMapper;
-use crate::update::new::indexer::de::DeOrBumpStr;
+use crate::update::new::indexer::de::{match_component, DeOrBumpStr};
 use crate::update::new::{CowStr, KvReaderFieldId, TopLevelMap};
 use crate::{FieldId, InternalError, Object, Result, UserError};
 
@@ -64,7 +65,7 @@ impl<'a> PrimaryKey<'a> {
         })
     }
 
-    pub fn name(&self) -> &str {
+    pub fn name(&self) -> &'a str {
         match self {
             PrimaryKey::Flat { name, .. } => name,
             PrimaryKey::Nested { name } => name,
@@ -154,7 +155,31 @@ impl<'a> PrimaryKey<'a> {
 
                 Ok(external_document_id)
             }
-            PrimaryKey::Nested { name } => todo!(),
+            nested @ PrimaryKey::Nested { name: _ } => {
+                let mut docid = None;
+                for (first_level, right) in nested.possible_level_names() {
+                    let Some(fid) = db_fields_ids_map.id(first_level) else { continue };
+
+                    let Some(value) = document.get(fid) else { continue };
+                    let value: &RawValue =
+                        serde_json::from_slice(value).map_err(InternalError::SerdeJson)?;
+                    match match_component(first_level, right, value, indexer, &mut docid) {
+                        ControlFlow::Continue(()) => continue,
+                        ControlFlow::Break(Ok(_)) => {
+                            return Err(InternalError::DocumentsError(
+                                crate::documents::Error::InvalidDocumentFormat,
+                            )
+                            .into())
+                        }
+                        ControlFlow::Break(Err(err)) => {
+                            return Err(InternalError::SerdeJson(err).into())
+                        }
+                    }
+                }
+                Ok(docid.ok_or(InternalError::DocumentsError(
+                    crate::documents::Error::InvalidDocumentFormat,
+                ))?)
+            }
         }
     }
 
@@ -171,7 +196,7 @@ impl<'a> PrimaryKey<'a> {
                 self,
                 indexer,
             ))
-            .map_err(UserError::SerdeJson)?;
+            .map_err(UserError::SerdeJson)??;
 
         let external_document_id = match res {
             Ok(document_id) => Ok(document_id),
@@ -234,7 +259,7 @@ impl<'a> PrimaryKey<'a> {
 
     /// Returns an `Iterator` that gives all the possible fields names the primary key
     /// can have depending of the first level name and depth of the objects.
-    pub fn possible_level_names(&self) -> impl Iterator<Item = (&str, &str)> + '_ {
+    pub fn possible_level_names(&self) -> impl Iterator<Item = (&'a str, &'a str)> + '_ {
         let name = self.name();
         name.match_indices(PRIMARY_KEY_SPLIT_SYMBOL)
             .map(move |(i, _)| (&name[..i], &name[i + PRIMARY_KEY_SPLIT_SYMBOL.len_utf8()..]))

milli/src/update/new/indexer/de.rs

@@ -1,4 +1,7 @@
+use std::ops::ControlFlow;
+
 use bumpalo::Bump;
+use serde::de::{DeserializeSeed, Deserializer as _, Visitor};
 use serde_json::value::RawValue;
 
 use crate::documents::{
@@ -14,22 +17,6 @@ pub struct FieldAndDocidExtractor<'p, 'indexer, Mapper: MutFieldIdMapper> {
     indexer: &'indexer Bump,
 }
 
-pub struct DocidExtractor<'p, 'indexer, Mapper: FieldIdMapper> {
-    fields_ids_map: &'p Mapper,
-    primary_key: &'p PrimaryKey<'p>,
-    indexer: &'indexer Bump,
-}
-
-impl<'p, 'indexer, Mapper: FieldIdMapper> DocidExtractor<'p, 'indexer, Mapper> {
-    pub fn new(
-        fields_ids_map: &'p Mapper,
-        primary_key: &'p PrimaryKey<'p>,
-        indexer: &'indexer Bump,
-    ) -> Self {
-        Self { fields_ids_map, primary_key, indexer }
-    }
-}
-
 impl<'p, 'indexer, Mapper: MutFieldIdMapper> FieldAndDocidExtractor<'p, 'indexer, Mapper> {
     pub fn new(
         fields_ids_map: &'p mut Mapper,
@@ -40,63 +27,56 @@ impl<'p, 'indexer, Mapper: MutFieldIdMapper> FieldAndDocidExtractor<'p, 'indexer
     }
 }
 
-impl<'de, 'p, 'indexer: 'de, Mapper: MutFieldIdMapper> serde::de::Visitor<'de>
+impl<'de, 'p, 'indexer: 'de, Mapper: MutFieldIdMapper> Visitor<'de>
     for FieldAndDocidExtractor<'p, 'indexer, Mapper>
 {
-    type Value = std::result::Result<DeOrBumpStr<'de, 'indexer>, DocumentIdExtractionError>;
+    type Value =
+        Result<Result<DeOrBumpStr<'de, 'indexer>, DocumentIdExtractionError>, crate::UserError>;
 
     fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(formatter, "a map")
     }
 
-    fn visit_map<A>(mut self, mut map: A) -> std::result::Result<Self::Value, A::Error>
+    fn visit_map<A>(mut self, mut map: A) -> Result<Self::Value, A::Error>
     where
         A: serde::de::MapAccess<'de>,
     {
         let mut docid = None;
-        while let Some((fid, fields_ids_map)) =
-            map.next_key_seed(MutFieldIdMapSeed(self.fields_ids_map))?
-        {
-            use serde::de::Deserializer as _;
-            self.fields_ids_map = fields_ids_map;
-            /// FIXME unwrap => too many fields
-            let fid = fid.unwrap();
 
-            match self.primary_key {
+        while let Some(((level_name, right), (fid, fields_ids_map))) =
-                PrimaryKey::Flat { name: _, field_id } => {
+            map.next_key_seed(ComponentsSeed {
-                    let value: &'de RawValue = map.next_value()?;
+                name: self.primary_key.name(),
-                    if fid == *field_id {
+                visitor: MutFieldIdMapVisitor(self.fields_ids_map),
-                        let value = match value
+            })?
-                            .deserialize_any(DocumentIdVisitor(self.indexer))
+        {
-                            .map_err(|_err| {
+            let Some(fid) = fid else {
-                                DocumentIdExtractionError::InvalidDocumentId(
+                return Ok(Err(crate::UserError::AttributeLimitReached));
-                                    UserError::InvalidDocumentId {
+            };
-                                        document_id: serde_json::to_value(value).unwrap(),
+            self.fields_ids_map = fields_ids_map;
-                                    },
+
-                                )
+            let value: &'de RawValue = map.next_value()?;
-                            }) {
+
-                            Ok(Ok(value)) => value,
+            match match_component(level_name, right, value, self.indexer, &mut docid) {
-                            Ok(Err(err)) | Err(err) => return Ok(Err(err)),
+                ControlFlow::Continue(()) => continue,
-                        };
+                ControlFlow::Break(Err(err)) => return Err(serde::de::Error::custom(err)),
-                        if let Some(_previous_value) = docid.replace(value) {
+                ControlFlow::Break(Ok(err)) => return Ok(Ok(Err(err))),
-                            return Ok(Err(DocumentIdExtractionError::TooManyDocumentIds(2)));
-                        }
-                    }
-                }
-                PrimaryKey::Nested { name } => todo!(),
             }
         }
-        Ok(match docid {
+
+        Ok(Ok(match docid {
             Some(docid) => Ok(docid),
             None => Err(DocumentIdExtractionError::MissingDocumentId),
-        })
+        }))
     }
 }
 
-impl<'de, 'p, 'indexer: 'de, Mapper: FieldIdMapper> serde::de::Visitor<'de>
+struct NestedPrimaryKeyVisitor<'a, 'bump> {
-    for DocidExtractor<'p, 'indexer, Mapper>
+    components: &'a str,
-{
+    bump: &'bump Bump,
-    type Value = std::result::Result<DeOrBumpStr<'de, 'indexer>, DocumentIdExtractionError>;
+}
+
+impl<'de, 'a, 'bump: 'de> Visitor<'de> for NestedPrimaryKeyVisitor<'a, 'bump> {
+    type Value = std::result::Result<Option<DeOrBumpStr<'de, 'bump>>, DocumentIdExtractionError>;
 
     fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
         write!(formatter, "a map")
@@ -107,142 +87,156 @@ impl<'de, 'p, 'indexer: 'de, Mapper: FieldIdMapper> serde::de::Visitor<'de>
         A: serde::de::MapAccess<'de>,
     {
         let mut docid = None;
-        while let Some(fid) = map.next_key_seed(FieldIdMapSeed(self.fields_ids_map))? {
+        while let Some(((matched_component, right), _)) = map.next_key_seed(ComponentsSeed {
-            use serde::de::Deserializer as _;
+            name: self.components,
+            visitor: serde::de::IgnoredAny,
+        })? {
+            let value: &'de RawValue = map.next_value()?;
 
-            let Some(fid) = fid else {
+            match match_component(matched_component, right, value, self.bump, &mut docid) {
-                continue;
+                ControlFlow::Continue(()) => continue,
-            };
+                ControlFlow::Break(Err(err)) => return Err(serde::de::Error::custom(err)),
-
+                ControlFlow::Break(Ok(err)) => return Ok(Err(err)),
-            match self.primary_key {
-                PrimaryKey::Flat { name: _, field_id } => {
-                    let value: &'de RawValue = map.next_value()?;
-                    if fid == *field_id {
-                        let value = match value
-                            .deserialize_any(DocumentIdVisitor(self.indexer))
-                            .map_err(|_err| {
-                                DocumentIdExtractionError::InvalidDocumentId(
-                                    UserError::InvalidDocumentId {
-                                        document_id: serde_json::to_value(value).unwrap(),
-                                    },
-                                )
-                            }) {
-                            Ok(Ok(value)) => value,
-                            Ok(Err(err)) | Err(err) => return Ok(Err(err)),
-                        };
-                        if let Some(_previous_value) = docid.replace(value) {
-                            return Ok(Err(DocumentIdExtractionError::TooManyDocumentIds(2)));
-                        }
-                    }
-                }
-                PrimaryKey::Nested { name } => todo!(),
             }
         }
-        Ok(match docid {
+        Ok(Ok(docid))
-            Some(docid) => Ok(docid),
-            None => Err(DocumentIdExtractionError::MissingDocumentId),
-        })
     }
 }
 
+/// Either a `&'de str` or a `&'bump str`.
 pub enum DeOrBumpStr<'de, 'bump: 'de> {
+    /// Lifetime of the deserializer
     De(&'de str),
+    /// Lifetime of the allocator
     Bump(&'bump str),
 }
 
 impl<'de, 'bump: 'de> DeOrBumpStr<'de, 'bump> {
+    /// Returns a `&'bump str`, possibly allocating to extend its lifetime.
     pub fn to_bump(&self, bump: &'bump Bump) -> &'bump str {
         match self {
             DeOrBumpStr::De(de) => bump.alloc_str(de),
-            DeOrBumpStr::Bump(bump) => *bump,
+            DeOrBumpStr::Bump(bump) => bump,
         }
     }
 
+    /// Returns a `&'de str`.
+    ///
+    /// This function never allocates because `'bump: 'de`.
     pub fn to_de(&self) -> &'de str {
         match self {
-            DeOrBumpStr::De(de) => *de,
+            DeOrBumpStr::De(de) => de,
-            DeOrBumpStr::Bump(bump) => *bump,
+            DeOrBumpStr::Bump(bump) => bump,
         }
     }
 }
 
-struct MutFieldIdMapSeed<'a, Mapper: MutFieldIdMapper>(&'a mut Mapper);
+struct ComponentsSeed<'a, V> {
+    name: &'a str,
+    visitor: V,
+}
 
-impl<'de, 'a, Mapper: MutFieldIdMapper> serde::de::DeserializeSeed<'de>
+impl<'de, 'a, V: Visitor<'de>> DeserializeSeed<'de> for ComponentsSeed<'a, V> {
-    for MutFieldIdMapSeed<'a, Mapper>
+    type Value = ((&'a str, &'a str), V::Value);
-{
+
+    fn deserialize<D>(self, deserializer: D) -> Result<Self::Value, D::Error>
+    where
+        D: serde::Deserializer<'de>,
+    {
+        struct ComponentsSeedVisitor<'a, V> {
+            name: &'a str,
+            visitor: V,
+        }
+
+        impl<'a, V> ComponentsSeedVisitor<'a, V> {
+            fn match_str(&self, v: &str) -> (&'a str, &'a str) {
+                let p = PrimaryKey::Nested { name: self.name };
+                for (name, right) in p.possible_level_names() {
+                    if name == v {
+                        return (name, right);
+                    }
+                }
+                ("", self.name)
+            }
+        }
+
+        impl<'de, 'a, V: Visitor<'de>> Visitor<'de> for ComponentsSeedVisitor<'a, V> {
+            type Value = ((&'a str, &'a str), V::Value);
+            fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+                write!(formatter, "expecting a string")
+            }
+            fn visit_borrowed_str<E>(self, v: &'de str) -> std::result::Result<Self::Value, E>
+            where
+                E: serde::de::Error,
+            {
+                let matched = self.match_str(v);
+                let inner = self.visitor.visit_borrowed_str(v)?;
+                Ok((matched, inner))
+            }
+
+            fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
+            where
+                E: serde::de::Error,
+            {
+                let matched = self.match_str(v);
+                let inner = self.visitor.visit_str(v)?;
+
+                Ok((matched, inner))
+            }
+        }
+        deserializer
+            .deserialize_str(ComponentsSeedVisitor { name: self.name, visitor: self.visitor })
+    }
+}
+
+struct MutFieldIdMapVisitor<'a, Mapper: MutFieldIdMapper>(&'a mut Mapper);
+
+impl<'de, 'a, Mapper: MutFieldIdMapper> Visitor<'de> for MutFieldIdMapVisitor<'a, Mapper> {
     type Value = (Option<FieldId>, &'a mut Mapper);
 
-    fn deserialize<D>(self, deserializer: D) -> std::result::Result<Self::Value, D::Error>
+    fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+        write!(formatter, "expecting a string")
+    }
+    fn visit_borrowed_str<E>(self, v: &'de str) -> std::result::Result<Self::Value, E>
     where
-        D: serde::Deserializer<'de>,
+        E: serde::de::Error,
     {
-        struct MutFieldIdMapVisitor<'a, Mapper: MutFieldIdMapper>(&'a mut Mapper);
+        Ok((self.0.insert(v), self.0))
-        impl<'de, 'a, Mapper: MutFieldIdMapper> serde::de::Visitor<'de>
+    }
-            for MutFieldIdMapVisitor<'a, Mapper>
-        {
-            type Value = (Option<FieldId>, &'a mut Mapper);
 
-            fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+    fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
-                write!(formatter, "expecting a string")
+    where
-            }
+        E: serde::de::Error,
-            fn visit_borrowed_str<E>(self, v: &'de str) -> std::result::Result<Self::Value, E>
+    {
-            where
+        Ok((self.0.insert(v), self.0))
-                E: serde::de::Error,
-            {
-                Ok((self.0.insert(v), self.0))
-            }
-
-            fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
-            where
-                E: serde::de::Error,
-            {
-                Ok((self.0.insert(v), self.0))
-            }
-        }
-        deserializer.deserialize_str(MutFieldIdMapVisitor(self.0))
     }
 }
 
-struct FieldIdMapSeed<'a, Mapper: FieldIdMapper>(&'a Mapper);
+pub struct FieldIdMapVisitor<'a, Mapper: FieldIdMapper>(pub &'a Mapper);
 
-impl<'de, 'a, Mapper: FieldIdMapper> serde::de::DeserializeSeed<'de>
+impl<'de, 'a, Mapper: FieldIdMapper> Visitor<'de> for FieldIdMapVisitor<'a, Mapper> {
-    for FieldIdMapSeed<'a, Mapper>
-{
     type Value = Option<FieldId>;
 
-    fn deserialize<D>(self, deserializer: D) -> std::result::Result<Self::Value, D::Error>
+    fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+        write!(formatter, "expecting a string")
+    }
+    fn visit_borrowed_str<E>(self, v: &'de str) -> std::result::Result<Self::Value, E>
     where
-        D: serde::Deserializer<'de>,
+        E: serde::de::Error,
     {
-        struct FieldIdMapVisitor<'a, Mapper: FieldIdMapper>(&'a Mapper);
+        Ok(self.0.id(v))
-        impl<'de, 'a, Mapper: FieldIdMapper> serde::de::Visitor<'de> for FieldIdMapVisitor<'a, Mapper> {
+    }
-            type Value = Option<FieldId>;
 
-            fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+    fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
-                write!(formatter, "expecting a string")
+    where
-            }
+        E: serde::de::Error,
-            fn visit_borrowed_str<E>(self, v: &'de str) -> std::result::Result<Self::Value, E>
+    {
-            where
+        Ok(self.0.id(v))
-                E: serde::de::Error,
-            {
-                Ok(self.0.id(v))
-            }
-
-            fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
-            where
-                E: serde::de::Error,
-            {
-                Ok(self.0.id(v))
-            }
-        }
-        deserializer.deserialize_str(FieldIdMapVisitor(self.0))
     }
 }
-
 pub struct DocumentIdVisitor<'indexer>(pub &'indexer Bump);
 
-impl<'de, 'indexer: 'de> serde::de::Visitor<'de> for DocumentIdVisitor<'indexer> {
+impl<'de, 'indexer: 'de> Visitor<'de> for DocumentIdVisitor<'indexer> {
     type Value = std::result::Result<DeOrBumpStr<'de, 'indexer>, DocumentIdExtractionError>;
 
     fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
@@ -262,13 +256,15 @@ impl<'de, 'indexer: 'de> serde::de::Visitor<'de> for DocumentIdVisitor<'indexer>
             .map(DeOrBumpStr::De))
     }
 
-    fn visit_str<E>(self, v: &str) -> std::result::Result<Self::Value, E>
+    fn visit_str<E>(self, v: &str) -> Result<Self::Value, E>
     where
         E: serde::de::Error,
     {
         let v = self.0.alloc_str(v);
-        self.visit_borrowed_str(v)?;
+        Ok(match self.visit_borrowed_str(v)? {
-        Ok(Ok(DeOrBumpStr::Bump(v)))
+            Ok(_) => Ok(DeOrBumpStr::Bump(v)),
+            Err(err) => Err(err),
+        })
     }
 
     fn visit_u64<E>(self, v: u64) -> std::result::Result<Self::Value, E>
@@ -288,8 +284,45 @@ impl<'de, 'indexer: 'de> serde::de::Visitor<'de> for DocumentIdVisitor<'indexer>
     {
         use std::fmt::Write as _;
 
-        let mut out = bumpalo::collections::String::new_in(&self.0);
+        let mut out = bumpalo::collections::String::new_in(self.0);
-        write!(&mut out, "{v}");
+        write!(&mut out, "{v}").unwrap();
         Ok(Ok(DeOrBumpStr::Bump(out.into_bump_str())))
     }
 }
+
+pub fn match_component<'de, 'indexer: 'de>(
+    first_level_name: &str,
+    right: &str,
+    value: &'de RawValue,
+    bump: &'indexer Bump,
+    docid: &mut Option<DeOrBumpStr<'de, 'indexer>>,
+) -> ControlFlow<Result<DocumentIdExtractionError, serde_json::Error>, ()> {
+    if first_level_name.is_empty() {
+        return ControlFlow::Continue(());
+    }
+
+    let value = if right.is_empty() {
+        match value.deserialize_any(DocumentIdVisitor(bump)).map_err(|_err| {
+            DocumentIdExtractionError::InvalidDocumentId(UserError::InvalidDocumentId {
+                document_id: serde_json::to_value(value).unwrap(),
+            })
+        }) {
+            Ok(Ok(value)) => value,
+            Ok(Err(err)) | Err(err) => return ControlFlow::Break(Ok(err)),
+        }
+    } else {
+        // if right is not empty, recursively extract right components from value
+        let res = value.deserialize_map(NestedPrimaryKeyVisitor { components: right, bump });
+        match res {
+            Ok(Ok(Some(value))) => value,
+            Ok(Ok(None)) => return ControlFlow::Continue(()),
+            Ok(Err(err)) => return ControlFlow::Break(Ok(err)),
+            Err(err) if err.is_data() => return ControlFlow::Continue(()), // we expected the field to be a map, but it was not and that's OK.
+            Err(err) => return ControlFlow::Break(Err(err)),
+        }
+    };
+    if let Some(_previous_value) = docid.replace(value) {
+        return ControlFlow::Break(Ok(DocumentIdExtractionError::TooManyDocumentIds(2)));
+    }
+    ControlFlow::Continue(())
+}