This commit is contained in:
2026-08-16 20:48:53 +02:00
parent c48faa4d6e
commit 07f4a76d74
10 changed files with 83 additions and 306 deletions
+29 -113
View File
@@ -658,6 +658,7 @@ impl MigrationTrait for Migration {
.primary_key(),
)
.col(ColumnDef::new(Alias::new("server_id")).uuid().null())
.col(ColumnDef::new(Alias::new("name")).string().not_null())
.col(ColumnDef::new(Alias::new("emoji_type")).string().not_null())
.col(ColumnDef::new(Alias::new("unicode_sequence")).text().null())
.col(ColumnDef::new(Alias::new("file_path")).text().null())
@@ -705,56 +706,6 @@ impl MigrationTrait for Migration {
seed_unicode_emojis(manager).await?;
manager
.create_table(
Table::create()
.table(Alias::new("emoji_alias"))
.if_not_exists()
.col(
ColumnDef::new(Alias::new("id"))
.uuid()
.not_null()
.primary_key(),
)
.col(ColumnDef::new(Alias::new("emoji_id")).uuid().not_null())
.col(ColumnDef::new(Alias::new("alias")).string().not_null())
.col(
ColumnDef::new(Alias::new("created_at"))
.timestamp_with_time_zone()
.not_null()
.default(Expr::current_timestamp()),
)
.col(
ColumnDef::new(Alias::new("updated_at"))
.timestamp_with_time_zone()
.not_null()
.default(Expr::current_timestamp()),
)
.foreign_key(
ForeignKey::create()
.name("fk_emoji_alias_emoji")
.from(Alias::new("emoji_alias"), Alias::new("emoji_id"))
.to(Alias::new("emoji"), Alias::new("id"))
.on_delete(ForeignKeyAction::Cascade),
)
.to_owned(),
)
.await?;
manager
.create_index(
Index::create()
.name("uq_emoji_alias_emoji")
.table(Alias::new("emoji_alias"))
.col(Alias::new("emoji_id"))
.col(Alias::new("alias"))
.unique()
.to_owned(),
)
.await?;
seed_unicode_aliases(manager).await?;
manager
.create_table(
Table::create()
@@ -1019,7 +970,6 @@ impl MigrationTrait for Migration {
async fn down(&self, manager: &SchemaManager) -> Result<(), DbErr> {
let tables = [
"computed_permission",
"emoji_alias",
"emoji",
"channel_user_read_state",
"channel_user_permission",
@@ -1060,21 +1010,42 @@ impl MigrationTrait for Migration {
/// number of bind parameters in a single statement to 999.
async fn seed_unicode_emojis(manager: &SchemaManager<'_>) -> Result<(), DbErr> {
const CHUNK_SIZE: usize = 250;
let rows = unicode_rows()?;
let mut used_names = HashSet::with_capacity(3944);
let rows = unicode_rows()?
.into_iter()
.map(|(id, sequence, raw_name)| {
let base = normalize_unicode_name(&raw_name);
let mut name = base.clone();
if !used_names.insert(name.clone()) {
name = format!("{base}_u{}", sequence.replace(' ', ""));
if name.len() > 64 {
name.truncate(64);
}
if !used_names.insert(name.clone()) {
return Err(DbErr::Custom(format!(
"duplicate generated emoji name: {name}"
)));
}
}
Ok((id, sequence, name))
})
.collect::<Result<Vec<_>, DbErr>>()?;
for chunk in rows.chunks(CHUNK_SIZE) {
let mut insert = Query::insert();
insert.into_table(Alias::new("emoji")).columns([
Alias::new("id"),
Alias::new("server_id"),
Alias::new("name"),
Alias::new("emoji_type"),
Alias::new("unicode_sequence"),
]);
for (id, sequence, _) in chunk {
for (id, sequence, name) in chunk {
insert.values_panic([
Expr::val(*id).into(),
Expr::val(Option::<Uuid>::None).into(),
Expr::val(name.as_str()).into(),
Expr::val("unicode").into(),
Expr::val(sequence.as_str()).into(),
]);
@@ -1086,61 +1057,6 @@ async fn seed_unicode_emojis(manager: &SchemaManager<'_>) -> Result<(), DbErr> {
Ok(())
}
async fn seed_unicode_aliases(manager: &SchemaManager<'_>) -> Result<(), DbErr> {
const CHUNK_SIZE: usize = 250;
let rows = unicode_rows()?;
let mut aliases = Vec::with_capacity(rows.len());
let mut used = HashSet::with_capacity(rows.len());
for (_, sequence, name) in rows {
let base = normalize_unicode_alias(&name);
let mut alias = base.clone();
if !used.insert(alias.clone()) {
alias = format!("{base}_u{}", sequence.replace(' ', ""));
if alias.len() > 64 {
alias.truncate(64);
}
if !used.insert(alias.clone()) {
return Err(DbErr::Custom(format!(
"duplicate generated emoji alias: {alias}"
)));
}
}
let emoji_id = Uuid::new_v5(
&Uuid::NAMESPACE_URL,
format!("https://oxspeak.local/unicode/{UNICODE_EMOJI_VERSION}/{sequence}").as_bytes(),
);
let id = Uuid::new_v5(
&Uuid::NAMESPACE_URL,
format!("https://oxspeak.local/unicode-alias/{UNICODE_EMOJI_VERSION}/{sequence}")
.as_bytes(),
);
aliases.push((id, emoji_id, alias));
}
for chunk in aliases.chunks(CHUNK_SIZE) {
let mut insert = Query::insert();
insert.into_table(Alias::new("emoji_alias")).columns([
Alias::new("id"),
Alias::new("emoji_id"),
Alias::new("alias"),
]);
for (id, emoji_id, alias) in chunk {
insert.values_panic([
Expr::val(*id).into(),
Expr::val(*emoji_id).into(),
Expr::val(alias.as_str()).into(),
]);
}
manager.exec_stmt(insert).await?;
}
Ok(())
}
fn unicode_rows() -> Result<Vec<(Uuid, String, String)>, DbErr> {
UNICODE_EMOJI_DATA
.lines()
@@ -1160,7 +1076,7 @@ fn unicode_rows() -> Result<Vec<(Uuid, String, String)>, DbErr> {
.collect()
}
fn normalize_unicode_alias(name: &str) -> String {
fn normalize_unicode_name(name: &str) -> String {
let mut alias = String::with_capacity(name.len());
let mut separator = false;
@@ -1223,16 +1139,16 @@ mod tests {
}
#[test]
fn unicode_aliases_are_normalized_for_the_existing_api_constraints() {
fn unicode_names_are_normalized_for_the_existing_api_constraints() {
assert_eq!(
normalize_unicode_alias("flag: United States"),
normalize_unicode_name("flag: United States"),
"flag_united_states"
);
assert_eq!(
normalize_unicode_alias("thumbs up: medium skin tone"),
normalize_unicode_name("thumbs up: medium skin tone"),
"thumbs_up_medium_skin_tone"
);
assert!(normalize_unicode_alias("woman technologist")
assert!(normalize_unicode_name("woman technologist")
.chars()
.all(|character| character.is_ascii_alphanumeric() || character == '_'));
}