basic update dataset handler & synthetic id tracking

This commit is contained in:
kasisnu committed 2023-10-04 18:03:13 +05:30
1 parent c42c091b2d
commit b67cfb5fbe
8 files changed
+397 -14

No files matched your search

@@ -20,6 +20,10 @@ pub enum IdCategory {
/// Results from lipsync animations (which may live in the media_files table)
#[serde(rename = "lipsync_animation")]
LipsyncAnimation,
/// Zs dataset which lives in the zs_voice_datasets table
#[serde(rename = "zs_dataset")]
ZsDataset,
}
// TODO(bt, 2022-12-21): This desperately needs MySQL integration tests!
@@ -32,6 +36,7 @@ impl IdCategory {
match self {
Self::MediaFile => "media_file",
Self::LipsyncAnimation => "lipsync_animation",
Self::ZsDataset => "zs_dataset",
}
}
@@ -39,6 +44,7 @@ impl IdCategory {
match value {
"media_file" => Ok(Self::MediaFile),
"lipsync_animation" => Ok(Self::LipsyncAnimation),
"zs_dataset" => Ok(Self::ZsDataset),
_ => Err(format!("invalid value: {:?}", value)),
}
}
@@ -49,6 +55,7 @@ impl IdCategory {
BTreeSet::from([
Self::MediaFile,
Self::LipsyncAnimation,
Self::ZsDataset,
])
}
}
@@ -68,21 +75,24 @@ mod tests {
fn to_str() {
assert_eq!(IdCategory::MediaFile.to_str(), "media_file");
assert_eq!(IdCategory::LipsyncAnimation.to_str(), "lipsync_animation");
assert_eq!(IdCategory::ZsDataset.to_str(), "zs_dataset");
}
#[test]
fn from_str() {
assert_eq!(IdCategory::from_str("media_file").unwrap(), IdCategory::MediaFile);
assert_eq!(IdCategory::from_str("lipsync_animation").unwrap(), IdCategory::LipsyncAnimation);
assert_eq!(IdCategory::from_str("zs_dataset").unwrap(), IdCategory::ZsDataset);
}
#[test]
fn all_variants() {
// Static check
let mut variants = IdCategory::all_variants();
assert_eq!(variants.len(), 2);
assert_eq!(variants.len(), 3);
assert_eq!(variants.pop_first(), Some(IdCategory::MediaFile));
assert_eq!(variants.pop_first(), Some(IdCategory::LipsyncAnimation));
assert_eq!(variants.pop_first(), Some(IdCategory::ZsDataset));
assert_eq!(variants.pop_first(), None);
// Generated check
@@ -1,9 +1,12 @@
use anyhow::anyhow;
use sqlx::MySqlPool;
use enums::common::visibility::Visibility;
use enums::by_table::generic_synthetic_ids::id_category::IdCategory;
use errors::AnyhowResult;
use tokens::tokens::dataset::ZsDatasetToken;
use tokens::users::user::UserToken;
use crate::queries::generic_synthetic_ids::transactional_increment_generic_synthetic_id::transactional_increment_generic_synthetic_id;
pub struct CreateDatasetArgs<'a> {
pub dataset_token: &'a ZsDatasetToken,
@@ -24,7 +27,20 @@ pub async fn create_dataset(args: CreateDatasetArgs<'_>) -> AnyhowResult<()>{
// ie, how are users anonymous if user session is validated before we reach here
// * could creator token be null?
// TODO: enforce checks for idempotency token
let mut maybe_creator_synthetic_id : Option<u64> = None;
let mut transaction = args.mysql_pool.begin().await?;
if let Some(creator_user_token) = args.maybe_creator_user_token.as_deref() {
let user_token = UserToken::new_from_str(creator_user_token);
let next_zs_dataset_synthetic_id = transactional_increment_generic_synthetic_id(
&user_token,
IdCategory::ZsDataset,
&mut transaction
).await?;
maybe_creator_synthetic_id = Some(next_zs_dataset_synthetic_id);
}
let query_result = sqlx::query!(
r#"
INSERT INTO zs_voice_datasets
@@ -34,16 +50,19 @@ pub async fn create_dataset(args: CreateDatasetArgs<'_>) -> AnyhowResult<()>{
maybe_creator_user_token = ?,
creator_ip_address = ?,
creator_set_visibility = ?,
maybe_mod_user_token = ?
maybe_mod_user_token = ?,
maybe_creator_synthetic_id = ?
"#,
args.dataset_token.as_str(),
args.dataset_title,
args.maybe_creator_user_token,
args.creator_ip_address,
args.creator_set_visibility.to_str(),
args.maybe_mod_user_token
args.maybe_mod_user_token,
maybe_creator_synthetic_id
).execute(args.mysql_pool).await;
// TODO(Kasisnu): This should probably rollback
transaction.commit().await?;
match query_result {
Ok(_) => Ok(()),
Err(err) => Err(anyhow!("zs dataset creation error: {:?}", err)),
@@ -0,0 +1,90 @@
use anyhow::anyhow;
use chrono::{DateTime, Utc};
use log::error;
use sqlx::{MySql, MySqlPool};
use sqlx::pool::PoolConnection;
use enums::by_table::media_uploads::media_upload_type::MediaUploadType;
use enums::common::visibility::Visibility;
use errors::AnyhowResult;
use tokens::files::media_upload::MediaUploadToken;
use tokens::tokens::dataset::ZsDatasetToken;
pub struct ZsDataset {
pub token: ZsDatasetToken,
pub title: String,
pub ietf_language_tag: String,
pub ietf_primary_language_subtag: String,
pub maybe_creator_user_token: Option<String>,
}
pub async fn get_dataset_by_token(
dataset_token: &ZsDatasetToken,
mysql_pool: &MySqlPool,
) -> AnyhowResult<Option<ZsDataset>> {
let mut connection = mysql_pool.acquire().await?;
get_dataset_by_token_with_connection(
dataset_token,
&mut connection
).await
}
pub async fn get_dataset_by_token_with_connection(
dataset_token: &ZsDatasetToken,
mysql_connection: &mut PoolConnection<MySql>,
) -> AnyhowResult<Option<ZsDataset>> {
let maybe_result = sqlx::query_as!(
RawDataset,
r#"
SELECT
zd.token as `token: tokens::tokens::dataset::ZsDatasetToken`,
zd.title,
zd.ietf_language_tag,
zd.ietf_primary_language_subtag,
zd.maybe_creator_user_token,
zd.creator_set_visibility as `creator_set_visibility: enums::common::visibility::Visibility`
FROM zs_voice_datasets as zd
WHERE
zd.token = ?
AND zd.user_deleted_at IS NULL
AND zd.mod_deleted_at IS NULL
"#,
dataset_token.as_str()
)
.fetch_one(mysql_connection).await;
let record = match maybe_result {
Ok(record) => record,
Err(sqlx::Error::RowNotFound) => {
return Ok(None);
},
Err(err) => {
error!(
"Error fetching dataset by token: {:?}",
err
);
return Err(anyhow!(
"Error fetching dataset by token: {:?}",
err
));
}
};
Ok(Some(ZsDataset {
token: record.token,
title: record.title,
ietf_language_tag: record.ietf_language_tag,
ietf_primary_language_subtag: record.ietf_primary_language_subtag,
maybe_creator_user_token: record.maybe_creator_user_token,
}))
}
#[derive(Serialize)]
pub struct RawDataset {
token: ZsDatasetToken,
title: String,
ietf_language_tag: String,
ietf_primary_language_subtag: String,
maybe_creator_user_token: Option<String>,
creator_set_visibility: Visibility,
}
@@ -1 +1,3 @@
pub mod create_dataset;
pub mod create_dataset;
pub mod update_dataset;
pub mod get_dataset;
@@ -0,0 +1,72 @@
use anyhow::anyhow;
use sqlx::MySqlPool;
use enums::common::visibility::Visibility;
use enums::by_table::generic_synthetic_ids::id_category::IdCategory;
use errors::AnyhowResult;
use tokens::tokens::dataset::ZsDatasetToken;
use tokens::users::user::UserToken;
use crate::queries::generic_synthetic_ids::transactional_increment_generic_synthetic_id::transactional_increment_generic_synthetic_id;
pub struct UpdateDatasetArgs<'a> {
pub dataset_token: &'a ZsDatasetToken,
pub dataset_title: Option<&'a str>,
pub maybe_creator_user_token: Option<&'a str>,
pub creator_ip_address: &'a str,
pub creator_set_visibility: &'a Visibility,
pub maybe_mod_user_token: Option<&'a str>,
pub ietf_language_tag: Option<&'a str>,
pub ietf_primary_language_subtag: Option<&'a str>,
pub mysql_pool: &'a MySqlPool
}
pub async fn update_dataset(args: UpdateDatasetArgs<'_>) -> AnyhowResult<()>{
// TODO: enforce checks for idempotency token
let mut maybe_creator_synthetic_id : Option<u64> = None;
let mut transaction = args.mysql_pool.begin().await?;
if let Some(creator_user_token) = args.maybe_creator_user_token.as_deref() {
let user_token = UserToken::new_from_str(creator_user_token);
let next_zs_dataset_synthetic_id = transactional_increment_generic_synthetic_id(
&user_token,
IdCategory::ZsDataset,
&mut transaction
).await?;
maybe_creator_synthetic_id = Some(next_zs_dataset_synthetic_id);
}
let query_result = sqlx::query!(
r#"
UPDATE zs_voice_datasets
SET
title = ?,
maybe_creator_user_token = ?,
creator_ip_address = ?,
creator_set_visibility = ?,
maybe_mod_user_token = ?,
maybe_creator_synthetic_id = ?,
version = version + 1
WHERE token = ?
LIMIT 1
"#,
args.dataset_title,
args.maybe_creator_user_token,
args.creator_ip_address,
args.creator_set_visibility.to_str(),
args.maybe_mod_user_token,
maybe_creator_synthetic_id,
args.dataset_token.as_str(),
).execute(args.mysql_pool).await;
// TODO(Kasisnu): This should probably rollback
transaction.commit().await?;
match query_result {
Ok(_) => Ok(()),
Err(err) => Err(anyhow!("zs dataset creation error: {:?}", err)),
}
}
@@ -18,6 +18,7 @@ use crate::server_state::ServerState;
#[derive(Deserialize)]
pub struct CreateDatasetRequest {
pub title: String,
pub creator_set_visibility: Option<Visibility>,
pub idempotency_token: Option<String>,
@@ -26,7 +27,7 @@ pub struct CreateDatasetRequest {
#[derive(Serialize)]
pub struct CreateDatasetResponse {
pub success: bool,
pub token: Option<String>,
pub token: Option<ZsDatasetToken>,
}
// =============== Error Response ===============
@@ -111,7 +112,7 @@ pub async fn create_dataset_handler(http_request: HttpRequest, request: web::Jso
let response = CreateDatasetResponse {
success: true,
token: Some(dataset_token.to_string()),
token: Some(dataset_token),
};
@@ -1,14 +1,200 @@
use std::fmt;
use std::sync::Arc;
use actix_web::{HttpRequest, HttpResponse, web};
use actix_web::error::ResponseError;
use actix_web::http::StatusCode;
use actix_web::web::Path;
use log::{error, log, warn};
use enums::common::visibility::Visibility;
use log::{info, warn};
use http_server_common::request::get_request_ip::get_request_ip;
use http_server_common::response::serialize_as_json_error::serialize_as_json_error;
use mysql_queries::queries::voice_designer::create_dataset::{create_dataset, CreateDatasetArgs};
use mysql_queries::queries::voice_designer::get_dataset::get_dataset_by_token;
use mysql_queries::queries::voice_designer::update_dataset::{update_dataset, UpdateDatasetArgs};
use tokens::tokens::dataset::ZsDatasetToken;
use crate::server_state::ServerState;
pub async fn update_dataset(dataset_token: web::Path<String>) -> HttpResponse {
// Implementation for updating a dataset
HttpResponse::Ok().json(format!("Dataset {} updated successfully", dataset_token))
#[derive(Deserialize)]
pub struct UpdateDatasetRequest {
// ========== Author + Moderator options ==========
pub title: Option<String>,
pub creator_set_visibility: Option<String>,
// NB: We calculate 'ietf_primary_language_subtag' from this value.
pub ietf_language_tag: Option<String>,
}
#[derive(Serialize)]
pub struct UpdateDatasetResponse {
pub success: bool,
}
/// For the URL PathInfo
#[derive(Deserialize)]
pub struct UpdateDatasetPathInfo {
dataset_token: String,
}
// =============== Error Response ===============
#[derive(Debug, Serialize)]
pub enum UpdateDatasetError {
BadInput(String),
NotFound,
NotAuthorized,
ServerError,
}
impl ResponseError for UpdateDatasetError {
fn status_code(&self) -> StatusCode {
match *self {
UpdateDatasetError::BadInput(_) => StatusCode::BAD_REQUEST,
UpdateDatasetError::NotFound => StatusCode::NOT_FOUND,
UpdateDatasetError::NotAuthorized => StatusCode::UNAUTHORIZED,
UpdateDatasetError::ServerError => StatusCode::INTERNAL_SERVER_ERROR,
}
}
fn error_response(&self) -> HttpResponse {
serialize_as_json_error(self)
}
}
// NB: Not using derive_more::Display since Clion doesn't understand it.
impl fmt::Display for UpdateDatasetError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "{:?}", self)
}
}
// =============== Handler ===============
pub async fn update_dataset_handler(
http_request: HttpRequest,
path: Path<UpdateDatasetPathInfo>,
request: web::Json<UpdateDatasetRequest>,
server_state: web::Data<Arc<ServerState>>) -> Result<HttpResponse, UpdateDatasetError>
{
let maybe_user_session = server_state
.session_checker
.maybe_get_user_session(&http_request, &server_state.mysql_pool)
.await
.map_err(|e| {
warn!("Session checker error: {:?}", e);
UpdateDatasetError::ServerError
})?;
let user_session = match maybe_user_session {
Some(session) => session,
None => {
warn!("not logged in");
return Err(UpdateDatasetError::NotAuthorized);
}
};
let dataset_token = path.dataset_token.clone();
let dataset_lookup_result = get_dataset_by_token(
&ZsDatasetToken::new(dataset_token.clone()),
&server_state.mysql_pool,
).await;
let dataset = match dataset_lookup_result {
Ok(Some(dataset)) => dataset,
Ok(None) => {
warn!("Dataset not found: {:?}", dataset_token);
return Err(UpdateDatasetError::NotFound);
},
Err(err) => {
warn!("Error looking up dataset: {:?}", err);
return Err(UpdateDatasetError::ServerError);
}
};
let mut title = None;
let mut ietf_language_tag = None;
let mut ietf_primary_language_subtag = None;
let mut creator_set_visibility = Visibility::Public;
if let Some(payload) = request.title.as_deref() {
if user_input_common::check_for_slurs::contains_slurs(payload) {
return Err(UpdateDatasetError::BadInput("title contains slurs".to_string()));
}
title = Some(payload.trim().to_string());
}
if let Some(tag) = request.ietf_language_tag.as_deref() {
// eg. en, en-US, es-419, ja-JP, etc.
let maybe_full_canonical_tag = container_common::i18n::supported_languages_for_models::get_canonicalized_language_tag_for_model(tag);
// eg. en, es, ja, etc.
let maybe_primary_language_subtag = maybe_full_canonical_tag
.map(|t| language_tags::LanguageTag::parse(t)
.map(|language_tag| language_tag.primary_language().to_string())
)
.transpose()
.map_err(|e| {
error!("Error parsing language tag '{}': {:?}", tag, e);
UpdateDatasetError::BadInput("bad locale string".to_string())
})?;
if let Some(full_tag) = maybe_full_canonical_tag {
if let Some(primary_subtag) = maybe_primary_language_subtag.as_deref() {
ietf_language_tag = Some(full_tag.to_string());
ietf_primary_language_subtag = Some(primary_subtag.to_string());
}
}
}
if let Some(visibility) = request.creator_set_visibility.as_deref() {
creator_set_visibility = Visibility::from_str(visibility)
.map_err(|_| UpdateDatasetError::BadInput("bad record visibility".to_string()))?;
}
let is_mod = user_session.can_ban_users;
let ip_address = get_request_ip(&http_request);
let mut maybe_mod_user_token = None;
if is_mod {
maybe_mod_user_token = Some(user_session.user_token.clone());
}
let query_result = update_dataset(
UpdateDatasetArgs {
dataset_token: &ZsDatasetToken::new(path.dataset_token.clone()),
dataset_title: title.as_deref(),
maybe_creator_user_token: Some(user_session.user_token.clone().as_ref()),
creator_ip_address: ip_address.as_ref(),
creator_set_visibility: &creator_set_visibility,
maybe_mod_user_token: maybe_mod_user_token.as_deref(),
ietf_language_tag: ietf_language_tag.as_deref(),
ietf_primary_language_subtag: ietf_primary_language_subtag.as_deref(),
mysql_pool: &server_state.mysql_pool
}
).await;
match query_result {
Ok(_) => {},
Err(err) => {
warn!("Update Dataset DB error: {:?}", err);
return Err(UpdateDatasetError::ServerError);
}
};
let response = UpdateDatasetResponse {
success: true,
};
let body = serde_json::to_string(&response)
.map_err(|e| UpdateDatasetError::ServerError)?;
Ok(HttpResponse::Ok()
.content_type("application/json")
.body(body))
}
@@ -121,7 +121,7 @@ use crate::http_server::endpoints::w2l::list_w2l_templates::list_w2l_templates_h
use crate::http_server::endpoints::w2l::set_w2l_template_mod_approval::set_w2l_template_mod_approval_handler;
use crate::http_server::endpoints::voice_designer::create_dataset::create_dataset_handler;
use crate::http_server::endpoints::voice_designer::update_dataset::update_dataset;
use crate::http_server::endpoints::voice_designer::update_dataset::update_dataset_handler;
use crate::http_server::endpoints::voice_designer::delete_dataset::delete_dataset;
use crate::http_server::endpoints::voice_designer::list_datasets_by_user::list_datasets_by_user;
@@ -1141,7 +1141,10 @@ fn add_voice_designer_routes<T,B> (app:App<T>)-> App<T>
.service(
web::scope("/datasets")
.route("/create", web::post().to(create_dataset_handler))
.route("/{dataset_token}/update", web::post().to(update_dataset))
.service(web::resource("/{dataset_token}/update")
.route(web::post().to(update_dataset_handler))
.route(web::head().to(|| HttpResponse::Ok()))
)
.route("/{dataset_token}/delete", web::delete().to(delete_dataset))
.route("/user/{user_token}/list", web::get().to(list_datasets_by_user))
)