Merge pull request #311 from korpling/comparable-update-events

Fix WAL persistance

Author: thomaskrause

.github/workflows/release_capi.yml

@@ -70,7 +70,7 @@ jobs:
           omitPrereleaseDuringUpdate: true
   deploy_macos_binaries:
     if: ${{ github.event.action == 'completed' || github.event.label.name == 'test-release-process' || (github.event_name == 'release' && github.event.action == 'published') }}
-    runs-on: macos-12
+    runs-on: macos-14
     steps:
       - id: latest-release
         uses: pozetroninc/github-action-get-latest-release@v0.7.0

.github/workflows/test.yml

@@ -36,7 +36,7 @@ jobs:
           RUST_LOG: debug
   test_mac:
     name: Execute automated tests on OSX
-    runs-on: macos-12
+    runs-on: macos-14
     steps:
       - uses: actions/checkout@v2
       - uses: actions-rust-lang/setup-rust-toolchain@v1.8.0

CHANGELOG.md

@@ -5,6 +5,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+
+- `UpdateEvent` now implements `PartialEq` to make possible to compare changes.
+
+### Fixed
+
+- Deserializing a write-ahead log failed because it was located at the wrong
+  sub-directory and the deserialization routine for the map had a bug.
+
 ## [3.5.1] - 2024-09-25
 
 ### Fixed

capi/src/cerror.rs

@@ -20,7 +20,7 @@ struct CauseIterator<'a> {
     current: Option<&'a dyn StdError>,
 }
 
-impl<'a> std::iter::Iterator for CauseIterator<'a> {
+impl std::iter::Iterator for CauseIterator<'_> {
     type Item = Error;
 
     fn next(&mut self) -> std::option::Option<Error> {

core/Cargo.toml

@@ -16,34 +16,36 @@ bincode = "1.2"
 clru = "0.6.1"
 itertools = "0.10"
 lazy_static = "1.4"
-toml = "0.8"
 log = "0.4"
 memmap2 = "0.9"
 normpath = "1.1.1"
 num-traits = "0.2"
 percent-encoding = "2.1"
 quick-xml = "0.28"
-rand = { version = "0.8", features = ["small_rng"] }
-rayon = { version = "1.3", default-features = false }
+rand = {version = "0.8", features = ["small_rng"]}
+rayon = {version = "1.3", default-features = false}
 regex = "1"
 regex-syntax = "0.8"
 rustc-hash = "1.0"
-serde = { version = "1.0", features = ["rc"] }
+serde = {version = "1.0", features = ["rc"]}
 serde_bytes = "0.11"
 serde_derive = "1.0"
 smallvec = "1.6"
-smartstring = { version = "1", features = ["serde"] }
+smartstring = {version = "1", features = ["serde"]}
 sstable = "0.11"
 strum = "0.21"
 strum_macros = "0.21"
 tempfile = "3.1"
 thiserror = "1"
+toml = "0.8"
 transient-btree-index = "0.5"
 
 [target.'cfg(windows)'.dependencies]
-winapi = { version = "0.3", features = ["heapapi"] }
+winapi = {version = "0.3", features = ["heapapi"]}
 
 [dev-dependencies]
 env_logger = "0.9"
 fake = "2.2"
+insta = {version = "1.38.0", features = ["json"]}
 pretty_assertions = "1.3"
+serde_json = "1.0"

core/src/dfs.rs

@@ -121,7 +121,7 @@ impl<'a> CycleSafeDFS<'a> {
     }
 }
 
-impl<'a> Iterator for CycleSafeDFS<'a> {
+impl Iterator for CycleSafeDFS<'_> {
     type Item = Result<DFSStep>;
 
     fn next(&mut self) -> Option<Self::Item> {

core/src/graph/mod.rs

@@ -626,7 +626,7 @@ impl<CT: ComponentType> Graph<CT> {
                 std::fs::create_dir_all(&current_path)?;
 
                 // If successfull write log
-                let log_path = location.join("update_log.bin");
+                let log_path = current_path.join("update_log.bin");
 
                 // Create a temporary directory in the same file system as the output
                 let temporary_dir = tempfile::tempdir_in(&current_path)?;
@@ -1155,4 +1155,62 @@ mod tests {
         db.ensure_loaded_parallel(&[component]).unwrap();
         assert_eq!(0, db.components.len());
     }
+
+    #[test]
+    fn load_with_wal_file() {
+        let mut db = Graph::<DefaultComponentType>::new(false).unwrap();
+        let example_node = 0;
+        db.node_annos
+            .insert(
+                example_node,
+                Annotation {
+                    key: NODE_TYPE_KEY.as_ref().clone(),
+                    val: "corpus".into(),
+                },
+            )
+            .unwrap();
+        db.node_annos
+            .insert(
+                example_node,
+                Annotation {
+                    key: NODE_NAME_KEY.as_ref().clone(),
+                    val: "root".into(),
+                },
+            )
+            .unwrap();
+
+        let tmp = tempfile::tempdir().unwrap();
+        // Save and remember the location, so that updates are recorded in a WAL
+        // file
+        db.persist_to(tmp.path()).unwrap();
+
+        // Add an node annotation with apply_update
+        let mut u = GraphUpdate::new();
+        u.add_event(UpdateEvent::AddNodeLabel {
+            node_name: "root".into(),
+            anno_ns: "example".into(),
+            anno_name: "anno-name".into(),
+            anno_value: "anno-value".into(),
+        })
+        .unwrap();
+        db.apply_update(&mut u, |_| {}).unwrap();
+
+        std::mem::drop(db);
+
+        // Check that loading the database again contains the changes
+        let mut db = Graph::<DefaultComponentType>::new(false).unwrap();
+        db.load_from(tmp.path(), true).unwrap();
+        let anno_value = db
+            .node_annos
+            .get_value_for_item(
+                &example_node,
+                &AnnoKey {
+                    name: "anno-name".into(),
+                    ns: "example".into(),
+                },
+            )
+            .unwrap()
+            .unwrap();
+        assert_eq!("anno-value", anno_value);
+    }
 }

core/src/graph/storage/union.rs

@@ -15,7 +15,7 @@ impl<'a> UnionEdgeContainer<'a> {
     }
 }
 
-impl<'a> EdgeContainer for UnionEdgeContainer<'a> {
+impl EdgeContainer for UnionEdgeContainer<'_> {
     fn get_outgoing_edges<'b>(
         &'b self,
         node: NodeID,

core/src/graph/update.rs

@@ -15,7 +15,7 @@ use sstable::{SSIterator, Table, TableBuilder, TableIterator};
 use tempfile::NamedTempFile;
 
 /// Describes a single update on the graph.
-#[derive(Serialize, Deserialize, Clone, Debug)]
+#[derive(Serialize, Deserialize, Clone, Debug, PartialEq)]
 pub enum UpdateEvent {
     /// Add a node with a name and type.
     AddNode {
@@ -304,10 +304,7 @@ impl<'de> Visitor<'de> for GraphUpdateVisitor {
 
         let mut event_counter = 0;
 
-        while let Some((id, event)) = access
-            .next_entry::<u64, GraphUpdate>()
-            .map_err(M::Error::custom)?
-        {
+        while let Some((id, event)) = access.next_entry::<u64, UpdateEvent>()? {
             event_counter = id;
             let key = id.create_key();
             let value = serialization.serialize(&event).map_err(M::Error::custom)?;
@@ -338,3 +335,6 @@ impl<'de> Deserialize<'de> for GraphUpdate {
         deserializer.deserialize_map(GraphUpdateVisitor {})
     }
 }
+
+#[cfg(test)]
+mod tests;

core/src/graph/update/snapshots/graphannis_core__graph__update__tests__serialize_json.snap

@@ -0,0 +1,27 @@
+---
+source: core/src/graph/update/tests.rs
+expression: seralized_string
+---
+{
+  "1": {
+    "AddNode": {
+      "node_name": "parent",
+      "node_type": "corpus"
+    }
+  },
+  "2": {
+    "AddNode": {
+      "node_name": "child",
+      "node_type": "corpus"
+    }
+  },
+  "3": {
+    "AddEdge": {
+      "source_node": "child",
+      "target_node": "parent",
+      "layer": "annis",
+      "component_type": "PartOf",
+      "component_name": ""
+    }
+  }
+}