fix(owners): add command to update all crate data

Author: Tyrubias

src/bin/cratesfyi.rs

@@ -25,6 +25,7 @@ use humantime::Duration;
 use once_cell::sync::OnceCell;
 use sentry::TransactionContext;
 use tokio::runtime::{Builder, Runtime};
+use tracing::info;
 use tracing_log::LogTracer;
 use tracing_subscriber::{filter::Directive, prelude::*, EnvFilter};
 
@@ -515,6 +516,9 @@ enum DatabaseSubcommand {
     /// Backfill GitHub/Gitlab stats for crates.
     BackfillRepositoryStats,
 
+    /// Backfill crate owner kind from crates.io API
+    BackfillCrateOwnerKind,
+
     /// Updates info for a crate from the registry's API
     UpdateCrateRegistryFields {
         #[arg(name = "CRATE")]
@@ -601,6 +605,37 @@ impl DatabaseSubcommand {
                     .block_on(ctx.repository_stats_updater()?.backfill_repositories())?;
             }
 
+            Self::BackfillCrateOwnerKind => {
+                let pool = ctx.pool()?;
+                ctx.runtime()?
+                    .block_on(async {
+                        let mut list_crates_conn = pool.get_async().await?;
+                        let mut update_crates_conn = pool.get_async().await?;
+
+                        let mut result_stream =
+                            sqlx::query!("SELECT id, name FROM crates ORDER BY name")
+                                .fetch(&mut *list_crates_conn);
+
+                        while let Some(row) = result_stream.next().await {
+                            let row = row?;
+                            let registry_data =
+                                ctx.registry_api()?.get_crate_data(&row.name).await?;
+
+                            info!("Updating crate {}", row.name);
+
+                            db::update_crate_data_in_db_by_id(
+                                &mut update_crates_conn,
+                                row.id,
+                                &registry_data,
+                            )
+                            .await?;
+                        }
+
+                        Ok::<(), anyhow::Error>(())
+                    })
+                    .context("Failed to backfill crate owner kind")?
+            }
+
             Self::UpdateCrateRegistryFields { name } => ctx.runtime()?.block_on(async move {
                 let mut conn = ctx.pool()?.get_async().await?;
                 let registry_data = ctx.registry_api()?.get_crate_data(&name).await?;

src/db/add_package.rs

@@ -435,6 +435,16 @@ pub async fn update_crate_data_in_database(
         .fetch_one(&mut *conn)
         .await?;
 
+    update_crate_data_in_db_by_id(conn, crate_id, registry_data).await?;
+
+    Ok(())
+}
+
+pub async fn update_crate_data_in_db_by_id(
+    conn: &mut sqlx::PgConnection,
+    crate_id: i32,
+    registry_data: &CrateData,
+) -> Result<()> {
     update_owners_in_database(conn, &registry_data.owners, crate_id).await?;
 
     Ok(())

src/db/mod.rs

@@ -7,7 +7,9 @@ pub(crate) use self::add_package::{
     add_build_into_database, add_doc_coverage, add_package_into_database,
 };
 pub use self::{
-    add_package::{update_build_status, update_crate_data_in_database},
+    add_package::{
+        update_build_status, update_crate_data_in_database, update_crate_data_in_db_by_id,
+    },
     delete::{delete_crate, delete_version},
     file::{add_path_into_database, add_path_into_remote_archive},
     overrides::Overrides,