refact: optimize stelae database by hashing composite keys; small ergonomic improvements

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "stelae"
 description = "A collection of tools in Rust and Python for preserving, authenticating, and accessing laws in perpetuity."
-version = "0.3.0-alpha.3"
+version = "0.3.0-alpha.4"
 edition = "2021"
 readme = "README.md"
 license = "AGPL-3.0"
@@ -16,6 +16,7 @@ actix-web = "4"
 actix-service = "2.0"
 actix-http = "3.2"
 async-trait = "0.1.77"
+md-5 = "0.10.6"
 mime = "0.3.17"
 mime_guess = "2.0.4"
 anyhow = "1.0"

migrations/sqlite/20240115152953_initial_db.down.sql

@@ -12,8 +12,8 @@ DROP TABLE IF EXISTS publication_has_publication_versions;
 DROP TABLE IF EXISTS publication_version;
 DROP TABLE IF EXISTS publication;
 DROP TABLE IF EXISTS version;
-DROP TABLE IF EXISTS library_document;
 DROP TABLE IF EXISTS library;
+DROP TABLE IF EXISTS document_element;
 DROP TABLE IF EXISTS document;
 DROP TABLE IF EXISTS stele;
 

migrations/sqlite/20240115152953_initial_db.up.sql

@@ -7,113 +7,98 @@ CREATE TABLE stele (
 CREATE TABLE document (
     doc_id TEXT PRIMARY KEY
 );
-CREATE TABLE library (
-    mpath TEXT PRIMARY KEY
-);
-CREATE TABLE library_document (
-    collection_mpath TEXT,
+CREATE TABLE document_element (
+    doc_mpath TEXT,
+    url TEXT,
     doc_id TEXT,
-    start DATE,
-    end DATE,
-    CONSTRAINT fk_coll_mpath
-        FOREIGN KEY (collection_mpath)
-        REFERENCES library(mpath),
     CONSTRAINT fk_doc_id
         FOREIGN KEY (doc_id)
         REFERENCES document(doc_id),
-    PRIMARY KEY (collection_mpath, doc_id)
+    PRIMARY KEY (doc_mpath)
+);
+CREATE TABLE library (
+    mpath TEXT PRIMARY KEY,
+    url TEXT
 );
 CREATE TABLE publication (
+    id TEXT,
     name TEXT,
     date INTEGER,
     stele TEXT,
     revoked INTEGER,
-    last_valid_publication_name TEXT,
+    last_valid_publication_id TEXT,
     last_valid_version TEXT,
     CONSTRAINT fk_last_valid_version
         FOREIGN KEY (last_valid_version)
         REFERENCES version(codified_date),
     CONSTRAINT fk_last_valid_publication
-        FOREIGN KEY (last_valid_publication_name, stele)
-        REFERENCES publication(name, stele),
+        FOREIGN KEY (last_valid_publication_id)
+        REFERENCES publication(id),
     CONSTRAINT fk_stele
         FOREIGN KEY (stele)
         REFERENCES stele(name)
         ON DELETE CASCADE,
-    PRIMARY KEY (name, stele)
+    PRIMARY KEY (id)
+);
+CREATE TABLE version(
+    codified_date TEXT PRIMARY KEY
 );
 CREATE TABLE publication_version (
+    id TEXT,
     version TEXT,
-    publication TEXT,
-    stele TEXT,
+    publication_id TEXT,
     build_reason TEXT,
     CONSTRAINT fk_publication
-        FOREIGN KEY (publication, stele)
-        REFERENCES publication(name, stele)
+        FOREIGN KEY (publication_id)
+        REFERENCES publication(id)
         ON DELETE CASCADE,
     CONSTRAINT fk_version
         FOREIGN KEY (version)
         REFERENCES version(codified_date),
-    PRIMARY KEY (publication, version, stele)
+    PRIMARY KEY (id)
 );
 CREATE TABLE publication_has_publication_versions (
-    publication TEXT,
-    referenced_publication TEXT,
-    referenced_version TEXT,
-    stele TEXT,
-    CONSTRAINT fk_publication FOREIGN KEY (publication, stele) REFERENCES publication(name, stele) ON DELETE CASCADE,
-    CONSTRAINT fk_referenced_publication FOREIGN KEY (referenced_publication, referenced_version, stele) REFERENCES publication_version(publication, version, stele) ON DELETE CASCADE,
-    PRIMARY KEY (publication, referenced_publication, referenced_version, stele)
-);
-CREATE TABLE version(
-    codified_date TEXT PRIMARY KEY
+    publication_id TEXT,
+    publication_version_id TEXT,
+    CONSTRAINT fk_publication FOREIGN KEY (publication_id) REFERENCES publication(id) ON DELETE CASCADE,
+    CONSTRAINT fk_referenced_publication_version FOREIGN KEY (publication_version_id) REFERENCES publication_version(id) ON DELETE CASCADE,
+    PRIMARY KEY (publication_id, publication_version_id)
 );
 CREATE TABLE document_change (
-    doc_mpath TEXT,
-    status TEXT,
-    url TEXT,
+    id TEXT,
+    status INTEGER,
     change_reason TEXT,
-    publication TEXT,
-    version TEXT,
-    stele TEXT,
-    doc_id TEXT,
-    CONSTRAINT fk_doc_id
-        FOREIGN KEY (doc_id)
-        REFERENCES document(doc_id)
+    publication_version_id TEXT,
+    doc_mpath TEXT,
+    CONSTRAINT fk_doc_el
+        FOREIGN KEY (doc_mpath)
+        REFERENCES document_element(doc_mpath)
         ON DELETE CASCADE,
     CONSTRAINT fk_publication_version
-        FOREIGN KEY (publication, version, stele)
-        REFERENCES publication_version(publication, version, stele)
+        FOREIGN KEY (publication_version_id)
+        REFERENCES publication_version(id)
         ON DELETE CASCADE,
-    PRIMARY KEY (doc_mpath, status, publication, version, stele)
+    PRIMARY KEY (id)
 );
 CREATE INDEX document_change_doc_mpath_idx ON document_change(doc_mpath COLLATE NOCASE);
 CREATE TABLE library_change (
-    publication TEXT,
-    version TEXT,
-    stele TEXT,
+    publication_version_id TEXT,
     status TEXT,
     library_mpath TEXT,
-    url TEXT,
     CONSTRAINT fk_publication_version
-        FOREIGN KEY (publication, version, stele)
-        REFERENCES publication_version(publication, version, stele)
+        FOREIGN KEY (publication_version_id)
+        REFERENCES publication_version(id)
         ON DELETE CASCADE,
-    PRIMARY KEY (publication, version, stele, library_mpath, status)
+    PRIMARY KEY (publication_version_id, library_mpath, status)
 );
 CREATE TABLE changed_library_document (
-    publication TEXT,
-    version TEXT,
-    stele TEXT,
-    doc_mpath TEXT,
-    status TEXT,
     library_mpath TEXT,
-    url TEXT,
+    document_change_id TEXT,
     CONSTRAINT fk_document_change
-        FOREIGN KEY (publication, version, stele, doc_mpath, status)
-        REFERENCES document_change(publication, version, stele, doc_mpath, status)
+        FOREIGN KEY (document_change_id)
+        REFERENCES document_change(id)
         ON DELETE CASCADE,
-    PRIMARY KEY (publication, version, stele, library_mpath, doc_mpath, status)
+    PRIMARY KEY (document_change_id, library_mpath)
 );
 CREATE INDEX library_change_library_mpath_idx ON library_change(library_mpath COLLATE NOCASE);
 CREATE INDEX changed_library_document_library_mpath_idx ON changed_library_document(library_mpath COLLATE NOCASE);

src/db/mod.rs

@@ -1,6 +1,7 @@
 //! Database related module.
 #![allow(clippy::unreachable)]
 use async_trait::async_trait;
+use sqlx::Transaction;
 use std::str::FromStr;
 
 use sqlx::any::{self, AnyPoolOptions};
@@ -12,8 +13,6 @@ use tracing::instrument;
 pub mod init;
 /// Models for the database.
 pub mod models;
-/// Statements for the database.
-pub mod statements;
 
 #[async_trait]
 /// Generic Database
@@ -25,6 +24,17 @@ pub trait Db {
     async fn connect(url: &str) -> anyhow::Result<DatabaseConnection>;
 }
 
+#[async_trait]
+/// Generic transaction
+pub trait Tx {
+    /// Begin a transaction.
+    async fn begin(pool: AnyPool) -> anyhow::Result<DatabaseTransaction>;
+    /// Commit a transaction.
+    async fn commit(self) -> anyhow::Result<()>;
+    /// Rollback a transaction.
+    async fn rollback(self) -> anyhow::Result<()>;
+}
+
 /// Type of database connection.
 #[derive(Debug, Clone)]
 pub enum DatabaseKind {
@@ -43,6 +53,12 @@ pub struct DatabaseConnection {
     pub kind: DatabaseKind,
 }
 
+/// Database transaction.
+pub struct DatabaseTransaction {
+    /// Database transaction.
+    pub tx: Transaction<'static, sqlx::Any>,
+}
+
 #[async_trait]
 impl Db for DatabaseConnection {
     /// Connects to a database.
@@ -72,3 +88,23 @@ impl Db for DatabaseConnection {
         Ok(connection)
     }
 }
+
+#[async_trait]
+impl Tx for DatabaseTransaction {
+    /// Begin a transaction.
+    async fn begin(pool: AnyPool) -> anyhow::Result<Self> {
+        let tx = pool.begin().await?;
+        Ok(Self { tx })
+    }
+    /// Commit a transaction.
+    async fn commit(self) -> anyhow::Result<()> {
+        self.tx.commit().await?;
+        Ok(())
+    }
+
+    /// Rollback a transaction.
+    async fn rollback(self) -> anyhow::Result<()> {
+        self.tx.rollback().await?;
+        Ok(())
+    }
+}

src/db/models/changed_library_document/manager.rs

@@ -0,0 +1,31 @@
+//! Manager for the changed library document model.
+use super::ChangedLibraryDocument;
+use crate::db::{models::BATCH_SIZE, DatabaseTransaction};
+use async_trait::async_trait;
+use sqlx::QueryBuilder;
+
+#[async_trait]
+impl super::TxManager for DatabaseTransaction {
+    /// Upsert a bulk of changed library documents into the database.
+    ///
+    /// # Errors
+    /// Errors if the changed library documents cannot be inserted into the database.
+    async fn insert_bulk(
+        &mut self,
+        changed_library_document: Vec<ChangedLibraryDocument>,
+    ) -> anyhow::Result<()> {
+        let mut query_builder = QueryBuilder::new(
+            "INSERT OR IGNORE INTO changed_library_document ( library_mpath, document_change_id ) ",
+        );
+        for chunk in changed_library_document.chunks(BATCH_SIZE) {
+            query_builder.push_values(chunk, |mut bindings, cl| {
+                bindings
+                    .push_bind(&cl.library_mpath)
+                    .push_bind(&cl.document_change_id);
+            });
+            let query = query_builder.build();
+            query.execute(&mut *self.tx).await?;
+        }
+        Ok(())
+    }
+}

src/db/models/changed_library_document/mod.rs

@@ -0,0 +1,34 @@
+use async_trait::async_trait;
+use serde::{Deserialize, Serialize};
+
+pub mod manager;
+
+/// Trait for managing transactional changed library documents.
+#[async_trait]
+pub trait TxManager {
+    /// Insert bulk of changed library documents.
+    async fn insert_bulk(
+        &mut self,
+        changed_library_document: Vec<ChangedLibraryDocument>,
+    ) -> anyhow::Result<()>;
+}
+
+#[derive(sqlx::FromRow, Deserialize, Serialize)]
+/// Model for library (collection) change events.
+pub struct ChangedLibraryDocument {
+    /// Foreign key reference to `document_change` id.
+    pub document_change_id: String,
+    /// Materialized path to the library
+    pub library_mpath: String,
+}
+
+impl ChangedLibraryDocument {
+    /// Create a new library change.
+    #[must_use]
+    pub const fn new(document_change_id: String, library_mpath: String) -> Self {
+        Self {
+            document_change_id,
+            library_mpath,
+        }
+    }
+}

src/db/models/document/manager.rs

@@ -0,0 +1,23 @@
+//! Manager for the document model.
+use crate::db::DatabaseTransaction;
+use async_trait::async_trait;
+
+#[async_trait]
+impl super::TxManager for DatabaseTransaction {
+    /// Upsert a new document into the database.
+    ///
+    /// # Errors
+    /// Errors if the document cannot be inserted into the database.
+    async fn create(&mut self, doc_id: &str) -> anyhow::Result<Option<i64>> {
+        let statement = "
+        INSERT OR IGNORE INTO document ( doc_id )
+        VALUES ( $1 )
+    ";
+        let id = sqlx::query(statement)
+            .bind(doc_id)
+            .execute(&mut *self.tx)
+            .await?
+            .last_insert_id();
+        Ok(id)
+    }
+}