mirror of
https://github.com/NexVeridian/wikidata-to-surrealdb.git
synced 2025-09-02 01:49:13 +00:00
refactor: CreateVersion and run_threaded
This commit is contained in:
parent
17a115f473
commit
b885315cd7
5 changed files with 210 additions and 235 deletions
|
@ -28,14 +28,8 @@ fn bench(c: &mut Criterion) {
|
|||
.reader("tests/data/bench.json")
|
||||
.unwrap();
|
||||
|
||||
create_db_entities_threaded(
|
||||
Some(db.clone()),
|
||||
reader,
|
||||
None,
|
||||
1000,
|
||||
100,
|
||||
CreateVersion::Single,
|
||||
)
|
||||
CreateVersion::Single
|
||||
.run_threaded(Some(db.clone()), reader, None, 1000, 100)
|
||||
.await
|
||||
.unwrap();
|
||||
})
|
||||
|
@ -51,14 +45,8 @@ fn bench(c: &mut Criterion) {
|
|||
.reader("tests/data/bench.json")
|
||||
.unwrap();
|
||||
|
||||
create_db_entities_threaded(
|
||||
Some(db.clone()),
|
||||
reader,
|
||||
None,
|
||||
1000,
|
||||
100,
|
||||
CreateVersion::Bulk,
|
||||
)
|
||||
CreateVersion::Bulk
|
||||
.run_threaded(Some(db.clone()), reader, None, 1000, 100)
|
||||
.await
|
||||
.unwrap();
|
||||
})
|
||||
|
|
14
src/main.rs
14
src/main.rs
|
@ -49,7 +49,7 @@ async fn main() -> Result<(), Error> {
|
|||
let line = line?;
|
||||
|
||||
loop {
|
||||
if create_db_entity(&db, &line).await.is_ok() {
|
||||
if create_entity(&db, &line).await.is_ok() {
|
||||
break;
|
||||
}
|
||||
if retries >= 60 * 10 {
|
||||
|
@ -69,35 +69,35 @@ async fn main() -> Result<(), Error> {
|
|||
}
|
||||
}
|
||||
CreateMode::ThreadedSingle => {
|
||||
create_db_entities_threaded(
|
||||
CreateVersion::Single
|
||||
.run_threaded(
|
||||
None::<Surreal<Client>>,
|
||||
reader,
|
||||
Some(pb.clone()),
|
||||
2_500,
|
||||
100,
|
||||
CreateVersion::Single,
|
||||
)
|
||||
.await?;
|
||||
}
|
||||
CreateMode::ThreadedBulk => {
|
||||
create_db_entities_threaded(
|
||||
CreateVersion::Bulk
|
||||
.run_threaded(
|
||||
None::<Surreal<Client>>,
|
||||
reader,
|
||||
Some(pb.clone()),
|
||||
500,
|
||||
1_000,
|
||||
CreateVersion::Bulk,
|
||||
)
|
||||
.await?;
|
||||
}
|
||||
CreateMode::ThreadedBulkFilter => {
|
||||
create_db_entities_threaded(
|
||||
CreateVersion::BulkFilter
|
||||
.run_threaded(
|
||||
None::<Surreal<Client>>,
|
||||
reader,
|
||||
Some(pb.clone()),
|
||||
500,
|
||||
1_000,
|
||||
CreateVersion::BulkFilter,
|
||||
)
|
||||
.await?;
|
||||
}
|
||||
|
|
216
src/utils.rs
216
src/utils.rs
|
@ -51,7 +51,7 @@ impl File_Format {
|
|||
}
|
||||
}
|
||||
|
||||
pub async fn create_db_entity(db: &Surreal<impl Connection>, line: &str) -> Result<(), Error> {
|
||||
pub async fn create_entity(db: &Surreal<impl Connection>, line: &str) -> Result<(), Error> {
|
||||
let line = line.trim().trim_end_matches(',').to_string();
|
||||
if line == "[" || line == "]" {
|
||||
return Ok(());
|
||||
|
@ -76,14 +76,108 @@ pub async fn create_db_entity(db: &Surreal<impl Connection>, line: &str) -> Resu
|
|||
Ok(())
|
||||
}
|
||||
|
||||
pub async fn create_db_entities(
|
||||
#[derive(Clone, Copy)]
|
||||
pub enum CreateVersion {
|
||||
Single,
|
||||
Bulk,
|
||||
// must create a filter.surql file in the root directory
|
||||
BulkFilter,
|
||||
}
|
||||
impl CreateVersion {
|
||||
pub async fn run(
|
||||
self,
|
||||
db: &Surreal<impl Connection>,
|
||||
chunk: &[String],
|
||||
pb: &Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
) -> bool {
|
||||
match self {
|
||||
CreateVersion::Single => self.create_single(db, chunk, pb).await.is_ok(),
|
||||
CreateVersion::Bulk => self.create_bulk(db, chunk, pb, batch_size).await.is_ok(),
|
||||
CreateVersion::BulkFilter => self
|
||||
.create_bulk_filter(db, chunk, pb, batch_size)
|
||||
.await
|
||||
.is_ok(),
|
||||
}
|
||||
}
|
||||
|
||||
pub async fn run_threaded(
|
||||
self,
|
||||
dbo: Option<Surreal<impl Connection>>,
|
||||
reader: Box<dyn BufRead>, // None::<Surreal<Client>>
|
||||
pb: Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
batch_num: usize,
|
||||
) -> Result<(), Error> {
|
||||
let mut lines = reader.lines().peekable();
|
||||
let mut futures = Vec::new();
|
||||
|
||||
while lines.peek().is_some() {
|
||||
let chunk: Vec<String> = lines
|
||||
.by_ref()
|
||||
.take(batch_size)
|
||||
.filter_map(Result::ok)
|
||||
.collect();
|
||||
|
||||
futures.push(self.spawn_chunk(dbo.clone(), chunk, pb.clone(), batch_size));
|
||||
|
||||
if futures.len() >= batch_num {
|
||||
join_all(futures).await;
|
||||
futures = Vec::new();
|
||||
}
|
||||
}
|
||||
|
||||
join_all(futures).await;
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn spawn_chunk(
|
||||
&self,
|
||||
dbo: Option<Surreal<impl Connection>>,
|
||||
chunk: Vec<String>,
|
||||
pb: Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
) -> tokio::task::JoinHandle<()> {
|
||||
let create_version = *self;
|
||||
|
||||
tokio::spawn(async move {
|
||||
let mut retries = 0;
|
||||
loop {
|
||||
match dbo {
|
||||
Some(ref db) => {
|
||||
if create_version.run(db, &chunk, &pb, batch_size).await {
|
||||
break;
|
||||
}
|
||||
}
|
||||
None => {
|
||||
let db = match init_db::create_db_ws().await {
|
||||
Ok(db) => db,
|
||||
Err(_) => continue,
|
||||
};
|
||||
if create_version.run(&db, &chunk, &pb, batch_size).await {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if retries >= 60 * 10 {
|
||||
panic!("Failed to create entities, too many retries");
|
||||
}
|
||||
retries += 1;
|
||||
sleep(Duration::from_millis(250)).await;
|
||||
}
|
||||
})
|
||||
}
|
||||
|
||||
async fn create_single(
|
||||
self,
|
||||
db: &Surreal<impl Connection>,
|
||||
lines: &[String],
|
||||
pb: &Option<ProgressBar>,
|
||||
) -> Result<(), Error> {
|
||||
let mut counter = 0;
|
||||
for line in lines {
|
||||
create_db_entity(db, line).await?;
|
||||
create_entity(db, line).await?;
|
||||
counter += 1;
|
||||
if counter % 100 == 0 {
|
||||
if let Some(ref p) = pb {
|
||||
|
@ -94,7 +188,8 @@ pub async fn create_db_entities(
|
|||
Ok(())
|
||||
}
|
||||
|
||||
pub async fn create_db_entities_bulk(
|
||||
async fn create_bulk(
|
||||
self,
|
||||
db: &Surreal<impl Connection>,
|
||||
lines: &[String],
|
||||
pb: &Option<ProgressBar>,
|
||||
|
@ -143,14 +238,15 @@ pub async fn create_db_entities_bulk(
|
|||
Ok(())
|
||||
}
|
||||
|
||||
pub async fn create_db_entities_bulk_filter(
|
||||
async fn create_bulk_filter(
|
||||
self,
|
||||
db: &Surreal<impl Connection>,
|
||||
lines: &[String],
|
||||
pb: &Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
) -> Result<(), Error> {
|
||||
let db_mem = init_db::create_db_mem().await?;
|
||||
create_db_entities_bulk(&db_mem, lines, &None, batch_size).await?;
|
||||
self.create_bulk(&db_mem, lines, &None, batch_size).await?;
|
||||
|
||||
let filter = tokio::fs::read_to_string(&*FILTER_PATH).await?;
|
||||
db_mem.query(filter).await?;
|
||||
|
@ -174,112 +270,4 @@ pub async fn create_db_entities_bulk_filter(
|
|||
}
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[derive(Clone, Copy)]
|
||||
pub enum CreateVersion {
|
||||
Single,
|
||||
Bulk,
|
||||
// must create a filter.surql file in the root directory
|
||||
BulkFilter,
|
||||
}
|
||||
impl CreateVersion {
|
||||
pub async fn run(
|
||||
self,
|
||||
db: &Surreal<impl Connection>,
|
||||
chunk: &[String],
|
||||
pb: &Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
) -> bool {
|
||||
match self {
|
||||
CreateVersion::Single => create_db_entities(db, chunk, pb).await.is_ok(),
|
||||
CreateVersion::Bulk => create_db_entities_bulk(db, chunk, pb, batch_size)
|
||||
.await
|
||||
.is_ok(),
|
||||
CreateVersion::BulkFilter => create_db_entities_bulk_filter(db, chunk, pb, batch_size)
|
||||
.await
|
||||
.is_ok(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
pub async fn create_db_entities_threaded(
|
||||
dbo: Option<Surreal<impl Connection>>, // None::<Surreal<Client>>
|
||||
reader: Box<dyn BufRead>,
|
||||
pb: Option<ProgressBar>,
|
||||
batch_size: usize,
|
||||
batch_num: usize,
|
||||
create_version: CreateVersion,
|
||||
) -> Result<(), Error> {
|
||||
let mut futures = Vec::new();
|
||||
let mut chunk = Vec::with_capacity(batch_size);
|
||||
let mut chunk_counter = 0;
|
||||
let mut lines = reader.lines();
|
||||
let mut last_loop = false;
|
||||
|
||||
loop {
|
||||
let line = lines.next();
|
||||
match line {
|
||||
Some(line) => chunk.push(line?),
|
||||
None => last_loop = true,
|
||||
};
|
||||
|
||||
if chunk.len() >= batch_size || last_loop {
|
||||
let dbo = dbo.clone();
|
||||
let pb = pb.clone();
|
||||
|
||||
futures.push(tokio::spawn(async move {
|
||||
let mut retries = 0;
|
||||
loop {
|
||||
match dbo {
|
||||
Some(ref db) => {
|
||||
if create_version.run(db, &chunk, &pb, batch_size).await {
|
||||
break;
|
||||
}
|
||||
if db.use_ns("wikidata").use_db("wikidata").await.is_err() {
|
||||
continue;
|
||||
};
|
||||
}
|
||||
None => {
|
||||
let db = if let Ok(db) = init_db::create_db_ws().await {
|
||||
db
|
||||
} else {
|
||||
continue;
|
||||
};
|
||||
if create_version.run(&db, &chunk, &pb, batch_size).await {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if retries >= 60 * 10 {
|
||||
panic!("Failed to create entities, too many retries");
|
||||
}
|
||||
retries += 1;
|
||||
sleep(Duration::from_millis(250)).await;
|
||||
}
|
||||
}));
|
||||
chunk_counter += 1;
|
||||
chunk = Vec::with_capacity(batch_size);
|
||||
}
|
||||
|
||||
if chunk_counter >= batch_num || last_loop {
|
||||
join_all(futures).await;
|
||||
futures = Vec::new();
|
||||
chunk_counter = 0;
|
||||
}
|
||||
if last_loop {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
match dbo {
|
||||
Some(db) => {
|
||||
create_db_entities(&db, &chunk, &pb).await?;
|
||||
}
|
||||
None => {
|
||||
create_db_entities(&init_db::create_db_ws().await?, &chunk, &pb).await?;
|
||||
}
|
||||
}
|
||||
join_all(futures).await;
|
||||
Ok(())
|
||||
}
|
||||
|
|
|
@ -34,5 +34,6 @@ pub async fn create_db_ws() -> Result<Surreal<Client>, Error> {
|
|||
pub async fn create_db_mem() -> Result<Surreal<Db>, Error> {
|
||||
let db = Surreal::new::<Mem>(()).await?;
|
||||
db.use_ns("wikidata").use_db("wikidata").await?;
|
||||
|
||||
Ok(db)
|
||||
}
|
||||
|
|
|
@ -42,7 +42,7 @@ async fn entity() {
|
|||
let reader = init_reader("json", "Entity");
|
||||
|
||||
for line in reader.lines() {
|
||||
create_db_entity(&db, &line.unwrap()).await.unwrap();
|
||||
create_entity(&db, &line.unwrap()).await.unwrap();
|
||||
}
|
||||
|
||||
assert_eq!(51.0, entity_query(&db).await.unwrap().unwrap())
|
||||
|
@ -56,7 +56,9 @@ async fn entity_threaded(#[case] version: CreateVersion) -> Result<(), Error> {
|
|||
let db = inti_db().await?;
|
||||
let reader = init_reader("json", "Entity");
|
||||
|
||||
create_db_entities_threaded(Some(db.clone()), reader, None, 1_000, 100, version).await?;
|
||||
version
|
||||
.run_threaded(Some(db.clone()), reader, None, 1_000, 100)
|
||||
.await?;
|
||||
|
||||
assert_eq!(51.0, entity_query(&db).await?.unwrap());
|
||||
Ok(())
|
||||
|
@ -68,14 +70,8 @@ async fn entity_threaded_filter() -> Result<(), Error> {
|
|||
let db = inti_db().await?;
|
||||
let reader = init_reader("json", "bench");
|
||||
|
||||
create_db_entities_threaded(
|
||||
Some(db.clone()),
|
||||
reader,
|
||||
None,
|
||||
1_000,
|
||||
100,
|
||||
CreateVersion::BulkFilter,
|
||||
)
|
||||
CreateVersion::BulkFilter
|
||||
.run_threaded(Some(db.clone()), reader, None, 1_000, 100)
|
||||
.await?;
|
||||
|
||||
let count: Option<f32> = db
|
||||
|
@ -105,7 +101,7 @@ async fn property() {
|
|||
let reader = init_reader("json", "Property");
|
||||
|
||||
for line in reader.lines() {
|
||||
create_db_entity(&db, &line.unwrap()).await.unwrap();
|
||||
create_entity(&db, &line.unwrap()).await.unwrap();
|
||||
}
|
||||
|
||||
assert_eq!(2.0, property_query(&db).await.unwrap().unwrap())
|
||||
|
@ -119,7 +115,9 @@ async fn property_threaded(#[case] version: CreateVersion) -> Result<(), Error>
|
|||
let db = inti_db().await?;
|
||||
let reader = init_reader("json", "Property");
|
||||
|
||||
create_db_entities_threaded(Some(db.clone()), reader, None, 1_000, 100, version).await?;
|
||||
version
|
||||
.run_threaded(Some(db.clone()), reader, None, 1_000, 100)
|
||||
.await?;
|
||||
|
||||
assert_eq!(2.0, property_query(&db).await?.unwrap());
|
||||
Ok(())
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue