is_err create

This commit is contained in:
Elijah McMorris 2023-12-16 17:11:54 -08:00
parent 4ea36f78b7
commit 2edaeef042
Signed by: NexVeridian
SSH key fingerprint: SHA256:bsA1SKZxuEcEVHAy3gY1HUeM5ykRJl0U0kQHQn0hMg8
2 changed files with 31 additions and 15 deletions

View file

@ -67,19 +67,32 @@ async fn create_db_entity(db: &Surreal<Client>, line: String) -> Result<(), Erro
let id = data.id.clone().expect("No ID"); let id = data.id.clone().expect("No ID");
data.id = None; data.id = None;
let _: Option<EntityMini> = db.delete(&id).await?; let _ = db.create::<Option<EntityMini>>(&id).await.is_err();
let _: Option<EntityMini> = db.create(&id).content(data.clone()).await?; {
db.update::<Option<EntityMini>>(&id).content(data).await?;
};
let id = claims.id.clone().expect("No ID"); let id = claims.id.clone().expect("No ID");
claims.id = None; claims.id = None;
let _: Option<Claims> = db.delete(&id).await?; let _ = db.create::<Option<Claims>>(&id).await.is_err();
let _: Option<Claims> = db.create(&id).content(claims).await?; {
db.update::<Option<Claims>>(&id).content(claims).await?;
}
Ok(()) Ok(())
} }
async fn create_db_entities(db: &Surreal<Client>, lines: Vec<String>) -> Result<(), Error> { async fn create_db_entities(
db: &Surreal<Client>,
lines: Vec<String>,
pb: ProgressBar,
) -> Result<(), Error> {
let mut counter = 0;
for line in lines { for line in lines {
create_db_entity(db, line.to_string()).await?; create_db_entity(db, line.to_string()).await?;
counter += 1;
if counter % 100 == 0 {
pb.inc(100);
}
} }
Ok(()) Ok(())
} }
@ -114,9 +127,10 @@ async fn main() -> Result<(), Error> {
let reader = File_Format::new(&WIKIDATA_FILE_FORMAT).reader(&WIKIDATA_FILE_NAME)?; let reader = File_Format::new(&WIKIDATA_FILE_FORMAT).reader(&WIKIDATA_FILE_NAME)?;
if !*THREADED_REQUESTS { if !*THREADED_REQUESTS {
let counter = 0; let mut counter = 0;
for line in reader.lines() { for line in reader.lines() {
create_db_entity(&db, line?).await?; create_db_entity(&db, line?).await?;
counter += 1;
if counter % 100 == 0 { if counter % 100 == 0 {
pb.inc(100); pb.inc(100);
} }
@ -124,34 +138,35 @@ async fn main() -> Result<(), Error> {
} else { } else {
let mut futures = Vec::new(); let mut futures = Vec::new();
let mut chunk = Vec::new(); let mut chunk = Vec::new();
let mut chunk_counter: i32 = 0; let mut chunk_counter = 0;
const BATCH_AMMOUNT: u16 = 50; const BATCH_SIZE: usize = 1000;
const BATCH_NUM: usize = 100;
for line in reader.lines() { for line in reader.lines() {
chunk.push(line.unwrap()); chunk.push(line.unwrap());
if chunk.len() >= BATCH_AMMOUNT.try_into().unwrap() { if chunk.len() >= BATCH_SIZE {
let db = db.clone(); let db = db.clone();
let lines = chunk.clone(); let lines = chunk.clone();
let pb = pb.clone(); let pb = pb.clone();
futures.push(tokio::spawn(async move { futures.push(tokio::spawn(async move {
create_db_entities(&db, lines).await.unwrap(); create_db_entities(&db, lines, pb).await.unwrap();
pb.inc(BATCH_AMMOUNT.try_into().unwrap());
})); }));
chunk_counter += 1; chunk_counter += 1;
chunk.clear(); chunk.clear();
} }
if chunk_counter >= 50 { if chunk_counter >= BATCH_NUM {
join_all(futures).await; join_all(futures).await;
futures = Vec::new(); futures = Vec::new();
chunk_counter = 0;
} }
} }
join_all(futures).await; join_all(futures).await;
} }
pb.finish_with_message("Done parsing Wikidata"); pb.finish();
Ok(()) Ok(())
} }

View file

@ -75,7 +75,7 @@ impl EntityMini {
id: Some(get_id_entity(&entity)), id: Some(get_id_entity(&entity)),
label: get_name(&entity), label: get_name(&entity),
claims: thing_claim, claims: thing_claim,
description: get_description(&entity).unwrap_or("".to_string()), description: get_description(&entity),
}, },
) )
} }
@ -129,9 +129,10 @@ fn get_name(entity: &Entity) -> String {
.unwrap_or_default() .unwrap_or_default()
} }
fn get_description(entity: &Entity) -> Option<String> { fn get_description(entity: &Entity) -> String {
entity entity
.descriptions .descriptions
.get(&Lang(WIKIDATA_LANG.to_string())) .get(&Lang(WIKIDATA_LANG.to_string()))
.cloned() .cloned()
.unwrap_or_default()
} }