From bea37e2834a5265035b960277ff27e5f8d74d079 Mon Sep 17 00:00:00 2001
From: Jan Tuomi <jan@jantuomi.fi>
Date: Fri, 21 Feb 2025 14:16:24 +0200
Subject: Start working in new schema logic to core

---
 log_db/src/common.rs |  5 +----
 log_db/src/config.rs | 34 +++++++++++++++---------------
 log_db/src/engine.rs | 25 +++++++++++++++-------
 log_db/src/lib.rs    | 59 +++++++++++++++++++++++++++++++++++-----------------
 4 files changed, 75 insertions(+), 48 deletions(-)

(limited to 'log_db/src')

diff --git a/log_db/src/common.rs b/log_db/src/common.rs
index 9bafdd4..ca9f451 100644
--- a/log_db/src/common.rs
+++ b/log_db/src/common.rs
@@ -289,10 +289,7 @@ impl Value {
     }
 }
 
-pub fn get_secondary_memtable_index_by_field<Field: Eq>(
-    sks: &Vec<Field>,
-    field: &Field,
-) -> Option<usize> {
+pub fn get_secondary_memtable_index_by_field(sks: &Vec<String>, field: &str) -> Option<usize> {
     sks.iter().position(|schema_field| schema_field == field)
 }
 
diff --git a/log_db/src/config.rs b/log_db/src/config.rs
index 5b3cf26..ec74164 100644
--- a/log_db/src/config.rs
+++ b/log_db/src/config.rs
@@ -6,23 +6,23 @@ pub struct Schema<F> {
     pub secondary_keys: Vec<F>,
 }
 
-pub struct ConfigBuilder<T, F> {
+pub struct ConfigBuilder<T> {
     data_dir: Option<String>,
     segment_size: Option<usize>,
     write_durability: Option<WriteDurability>,
     read_consistency: Option<ReadConsistency>,
 
-    fields: Option<Vec<F>>,
-    primary_key: Option<F>,
-    secondary_keys: Option<Vec<F>>,
+    fields: Option<Vec<String>>,
+    primary_key: Option<String>,
+    secondary_keys: Option<Vec<String>>,
     from_record: Option<fn(Vec<Value>) -> T>,
     into_record: Option<fn(T) -> Vec<Value>>,
 
     _marker: PhantomData<T>,
 }
 
-impl<T, F: Eq + Clone> ConfigBuilder<T, F> {
-    pub fn new() -> ConfigBuilder<T, F> {
+impl<T> ConfigBuilder<T> {
+    pub fn new() -> ConfigBuilder<T> {
         ConfigBuilder {
             data_dir: None,
             segment_size: None,
@@ -71,18 +71,18 @@ impl<T, F: Eq + Clone> ConfigBuilder<T, F> {
         self
     }
 
-    pub fn fields(mut self, schema: Vec<F>) -> Self {
-        self.fields = Some(schema);
+    pub fn fields(mut self, schema: Vec<impl Into<String>>) -> Self {
+        self.fields = Some(schema.into_iter().map(|s| s.into()).collect());
         self
     }
 
-    pub fn primary_key(mut self, primary_key: F) -> Self {
-        self.primary_key = Some(primary_key);
+    pub fn primary_key(mut self, primary_key: impl Into<String>) -> Self {
+        self.primary_key = Some(primary_key.into());
         self
     }
 
-    pub fn secondary_keys(mut self, secondary_keys: Vec<F>) -> Self {
-        self.secondary_keys = Some(secondary_keys);
+    pub fn secondary_keys(mut self, secondary_keys: Vec<impl Into<String>>) -> Self {
+        self.secondary_keys = Some(secondary_keys.into_iter().map(|s| s.into()).collect());
         self
     }
 
@@ -96,7 +96,7 @@ impl<T, F: Eq + Clone> ConfigBuilder<T, F> {
         self
     }
 
-    pub fn initialize(self) -> DBResult<DB<T, F>> {
+    pub fn initialize(self) -> DBResult<DB<T>> {
         let schema = self
             .fields
             .ok_or_else(|| DBError::ValidationError("Schema not set".to_string()))?;
@@ -134,10 +134,10 @@ impl<T, F: Eq + Clone> ConfigBuilder<T, F> {
 }
 
 #[derive(Clone)]
-pub struct Config<T, F> {
-    pub schema: Vec<F>,
-    pub primary_key: F,
-    pub secondary_keys: Vec<F>,
+pub struct Config<T> {
+    pub schema: Vec<String>,
+    pub primary_key: String,
+    pub secondary_keys: Vec<String>,
     pub from_record: fn(Vec<Value>) -> T,
     pub into_record: fn(T) -> Vec<Value>,
     pub data_dir: String,
diff --git a/log_db/src/engine.rs b/log_db/src/engine.rs
index c2f9a52..b06bfd2 100644
--- a/log_db/src/engine.rs
+++ b/log_db/src/engine.rs
@@ -1,7 +1,7 @@
 use super::*;
 
-pub struct Engine<T, F: Eq + Clone> {
-    pub config: Config<T, F>,
+pub struct Engine<T> {
+    pub config: Config<T>,
     pub lock_manager: LockManager,
 
     data_dir_path: PathBuf,
@@ -19,8 +19,8 @@ pub struct Engine<T, F: Eq + Clone> {
     pub secondary_memtables: Vec<SecondaryMemtable>,
 }
 
-impl<T, F: Eq + Clone> Engine<T, F> {
-    pub fn initialize(config: Config<T, F>) -> DBResult<Engine<T, F>> {
+impl<T> Engine<T> {
+    pub fn initialize(config: Config<T>) -> DBResult<Engine<T>> {
         info!("Initializing DB...");
         // If data_dir does not exist or is empty, create it and any necessary files.
         // After creation, the directory should always be in a complete state without missing files.
@@ -104,7 +104,7 @@ impl<T, F: Eq + Clone> Engine<T, F> {
             Path::new(&config.data_dir).join(active_metadata_header.uuid.to_string());
         let active_data_file = APPEND_MODE.open(&active_data_path)?;
 
-        let mut engine = Engine::<T, F> {
+        let mut engine = Engine::<T> {
             config,
             lock_manager,
             data_dir_path,
@@ -247,7 +247,7 @@ impl<T, F: Eq + Clone> Engine<T, F> {
 
     pub fn batch_find_by_records<'a>(
         &mut self,
-        field: &F,
+        field: &str,
         values: impl Iterator<Item = &'a Value>,
         params: &QueryParams,
     ) -> DBResult<Vec<(usize, Record)>> {
@@ -378,7 +378,7 @@ impl<T, F: Eq + Clone> Engine<T, F> {
 
     pub fn range_by_records<B: RangeBounds<Value>>(
         &mut self,
-        field: &F,
+        field: &str,
         range: B,
         params: &QueryParams,
     ) -> DBResult<Vec<Record>> {
@@ -458,7 +458,7 @@ impl<T, F: Eq + Clone> Engine<T, F> {
         }
     }
 
-    pub fn delete_by_field(&mut self, field: &F, value: &Value) -> DBResult<Vec<Record>> {
+    pub fn delete_by_field(&mut self, field: &str, value: &Value) -> DBResult<Vec<Record>> {
         let recs: Vec<Record> = self
             .batch_find_by_records(field, std::iter::once(value), &DEFAULT_QUERY_PARAMS)?
             .into_iter()
@@ -737,6 +737,15 @@ mod tests {
         Name,
     }
 
+    impl Into<String> for Field {
+        fn into(self) -> String {
+            match self {
+                Field::Id => "id".to_owned(),
+                Field::Name => "name".to_owned(),
+            }
+        }
+    }
+
     #[derive(PartialEq, Eq, Debug, Clone)]
     struct TestInst2 {
         id: i64,
diff --git a/log_db/src/lib.rs b/log_db/src/lib.rs
index 753e41c..fddb344 100644
--- a/log_db/src/lib.rs
+++ b/log_db/src/lib.rs
@@ -37,17 +37,17 @@ use memtable_primary::PrimaryMemtable;
 use memtable_secondary::SecondaryMemtable;
 use record::*;
 
-pub struct DB<T, F: Eq + Clone> {
-    engine: Engine<T, F>,
+pub struct DB<T> {
+    engine: Engine<T>,
 }
 
-impl<T, F: Eq + Clone> DB<T, F> {
+impl<T> DB<T> {
     /// Create a new database configuration builder.
-    pub fn configure() -> ConfigBuilder<T, F> {
+    pub fn configure() -> ConfigBuilder<T> {
         ConfigBuilder::new()
     }
 
-    fn initialize(config: Config<T, F>) -> DBResult<DB<T, F>> {
+    fn initialize(config: Config<T>) -> DBResult<DB<T>> {
         let engine = Engine::initialize(config)?;
         Ok(DB { engine })
     }
@@ -85,9 +85,13 @@ impl<T, F: Eq + Clone> DB<T, F> {
     }
 
     /// Get a collection of records based on an indexed field value.
-    pub fn find_by(&mut self, field: &F, value: &Value) -> DBResult<Vec<T>> {
+    pub fn find_by(&mut self, field: impl AsRef<str>, value: &Value) -> DBResult<Vec<T>> {
         let recs = self.engine.with_shared_lock(|engine| {
-            engine.batch_find_by_records(field, std::iter::once(value), &DEFAULT_QUERY_PARAMS)
+            engine.batch_find_by_records(
+                field.as_ref(),
+                std::iter::once(value),
+                &DEFAULT_QUERY_PARAMS,
+            )
         })?;
 
         Ok(recs
@@ -99,12 +103,12 @@ impl<T, F: Eq + Clone> DB<T, F> {
     /// Get a collection of records based on an indexed field value, with additional parameters.
     pub fn find_by_with_params(
         &mut self,
-        field: &F,
+        field: impl AsRef<str>,
         value: &Value,
         params: &QueryParams,
     ) -> DBResult<Vec<T>> {
         let recs = self.engine.with_shared_lock(|engine| {
-            engine.batch_find_by_records(field, std::iter::once(value), params)
+            engine.batch_find_by_records(field.as_ref(), std::iter::once(value), params)
         })?;
 
         Ok(recs
@@ -116,9 +120,13 @@ impl<T, F: Eq + Clone> DB<T, F> {
     /// Get a collection of records based on a sequence of indexed field values.
     /// Returns a vector of pairs where the first value is an index into the given sequence of values,
     /// and the second value is the record.
-    pub fn batch_find_by(&mut self, field: &F, values: &[Value]) -> DBResult<Vec<(usize, T)>> {
+    pub fn batch_find_by(
+        &mut self,
+        field: impl Into<String>,
+        values: &[Value],
+    ) -> DBResult<Vec<(usize, T)>> {
         let recs = self.engine.with_shared_lock(|engine| {
-            engine.batch_find_by_records(field, values.iter(), &DEFAULT_QUERY_PARAMS)
+            engine.batch_find_by_records(&field.into(), values.iter(), &DEFAULT_QUERY_PARAMS)
         })?;
 
         Ok(recs
@@ -132,12 +140,12 @@ impl<T, F: Eq + Clone> DB<T, F> {
     /// and the second value is the record.
     pub fn batch_find_by_with_params(
         &mut self,
-        field: &F,
+        field: impl AsRef<str>,
         values: &[Value],
         params: &QueryParams,
     ) -> DBResult<Vec<(usize, T)>> {
         let recs = self.engine.with_shared_lock(|engine| {
-            engine.batch_find_by_records(field, values.iter(), params)
+            engine.batch_find_by_records(field.as_ref(), values.iter(), params)
         })?;
 
         Ok(recs
@@ -149,9 +157,13 @@ impl<T, F: Eq + Clone> DB<T, F> {
     /// Get a collection of records based on a range of indexed field values.
     /// This method can be used to run comparison-like queries, e.g. `field >= 10`
     /// could be expressed as `db.range_by(Field::Id, 10..)`.
-    pub fn range_by<B: RangeBounds<Value>>(&mut self, field: &F, range: B) -> DBResult<Vec<T>> {
+    pub fn range_by<B: RangeBounds<Value>>(
+        &mut self,
+        field: impl AsRef<str>,
+        range: B,
+    ) -> DBResult<Vec<T>> {
         let recs = self.engine.with_shared_lock(|engine| {
-            engine.range_by_records(field, range, &DEFAULT_QUERY_PARAMS)
+            engine.range_by_records(field.as_ref(), range, &DEFAULT_QUERY_PARAMS)
         })?;
 
         Ok(recs
@@ -165,13 +177,13 @@ impl<T, F: Eq + Clone> DB<T, F> {
     /// could be expressed as `db.range_by(Field::Id, 10..)`.
     pub fn range_by_with_params<B: RangeBounds<Value>>(
         &mut self,
-        field: &F,
+        field: impl AsRef<str>,
         range: B,
         params: &QueryParams,
     ) -> DBResult<Vec<T>> {
         let recs = self
             .engine
-            .with_shared_lock(|engine| engine.range_by_records(field, range, params))?;
+            .with_shared_lock(|engine| engine.range_by_records(field.as_ref(), range, params))?;
 
         Ok(recs
             .into_iter()
@@ -185,10 +197,10 @@ impl<T, F: Eq + Clone> DB<T, F> {
     ///
     /// Deletion is done by marking the record as a tombstone. The record will still be present in the log file,
     /// but will be ignored by reads. Upon compaction, tombstoned records will be removed.
-    pub fn delete_by(&mut self, field: &F, value: &Value) -> DBResult<Vec<T>> {
+    pub fn delete_by(&mut self, field: impl AsRef<str>, value: &Value) -> DBResult<Vec<T>> {
         let recs = self
             .engine
-            .with_exclusive_lock(|engine| engine.delete_by_field(field, value))?;
+            .with_exclusive_lock(|engine| engine.delete_by_field(field.as_ref(), value))?;
 
         Ok(recs
             .into_iter()
@@ -295,6 +307,15 @@ mod tests {
         Name,
     }
 
+    impl Into<String> for Field {
+        fn into(self) -> String {
+            match self {
+                Field::Id => "id".to_string(),
+                Field::Name => "name".to_string(),
+            }
+        }
+    }
+
     struct TestInst1 {
         id: i64,
     }
-- 
cgit v1.3