Merge pull request #10078 from Turbo87/per-version-cats-and-keys

database: Add `categories` and `keywords` columns to the `versions` table

Author: Turbo87

crates/crates_io_database/src/schema.rs

@@ -1021,6 +1021,10 @@ diesel::table! {
         documentation -> Nullable<Text>,
         /// Value of the `repository` field in the `Cargo.toml` file of this version.
         repository -> Nullable<Text>,
+        /// The list of `categories` in the `Cargo.toml` file of this version.
+        categories -> Array<Nullable<Text>>,
+        /// The list of `keywords` in the `Cargo.toml` file of this version.
+        keywords -> Array<Nullable<Text>>,
     }
 }
 

crates/crates_io_database_dump/src/dump-db.toml

@@ -251,6 +251,8 @@ description = "public"
 homepage = "public"
 documentation = "public"
 repository = "public"
+categories = "public"
+keywords = "public"
 
 [versions_published_by.columns]
 version_id = "private"

crates/crates_io_database_dump/src/snapshots/[email protected]

@@ -18,7 +18,7 @@ BEGIN ISOLATION LEVEL REPEATABLE READ, READ ONLY;
     \copy "crates_keywords" ("crate_id", "keyword_id") TO 'data/crates_keywords.csv' WITH CSV HEADER
     \copy (SELECT "crate_id", "created_at", "created_by", "owner_id", "owner_kind" FROM "crate_owners" WHERE NOT deleted) TO 'data/crate_owners.csv' WITH CSV HEADER
 
-    \copy "versions" ("bin_names", "checksum", "crate_id", "crate_size", "created_at", "description", "documentation", "downloads", "edition", "features", "has_lib", "homepage", "id", "license", "links", "num", "published_by", "repository", "rust_version", "updated_at", "yanked") TO 'data/versions.csv' WITH CSV HEADER
+    \copy "versions" ("bin_names", "categories", "checksum", "crate_id", "crate_size", "created_at", "description", "documentation", "downloads", "edition", "features", "has_lib", "homepage", "id", "keywords", "license", "links", "num", "published_by", "repository", "rust_version", "updated_at", "yanked") TO 'data/versions.csv' WITH CSV HEADER
     \copy "default_versions" ("crate_id", "version_id") TO 'data/default_versions.csv' WITH CSV HEADER
     \copy "dependencies" ("crate_id", "default_features", "explicit_name", "features", "id", "kind", "optional", "req", "target", "version_id") TO 'data/dependencies.csv' WITH CSV HEADER
     \copy "version_downloads" ("date", "downloads", "version_id") TO 'data/version_downloads.csv' WITH CSV HEADER

crates/crates_io_database_dump/src/snapshots/[email protected]

@@ -60,7 +60,7 @@ BEGIN;
     \copy "crates_categories" ("category_id", "crate_id") FROM 'data/crates_categories.csv' WITH CSV HEADER
     \copy "crates_keywords" ("crate_id", "keyword_id") FROM 'data/crates_keywords.csv' WITH CSV HEADER
     \copy "crate_owners" ("crate_id", "created_at", "created_by", "owner_id", "owner_kind") FROM 'data/crate_owners.csv' WITH CSV HEADER
-    \copy "versions" ("bin_names", "checksum", "crate_id", "crate_size", "created_at", "description", "documentation", "downloads", "edition", "features", "has_lib", "homepage", "id", "license", "links", "num", "published_by", "repository", "rust_version", "updated_at", "yanked") FROM 'data/versions.csv' WITH CSV HEADER
+    \copy "versions" ("bin_names", "categories", "checksum", "crate_id", "crate_size", "created_at", "description", "documentation", "downloads", "edition", "features", "has_lib", "homepage", "id", "keywords", "license", "links", "num", "published_by", "repository", "rust_version", "updated_at", "yanked") FROM 'data/versions.csv' WITH CSV HEADER
     \copy "default_versions" ("crate_id", "version_id") FROM 'data/default_versions.csv' WITH CSV HEADER
     \copy "dependencies" ("crate_id", "default_features", "explicit_name", "features", "id", "kind", "optional", "req", "target", "version_id") FROM 'data/dependencies.csv' WITH CSV HEADER
     \copy "version_downloads" ("date", "downloads", "version_id") FROM 'data/version_downloads.csv' WITH CSV HEADER

migrations/2024-11-26-155931_add-cats-and-keys-to-version/down.sql

@@ -0,0 +1,3 @@
+alter table versions
+    drop column categories,
+    drop column keywords;

migrations/2024-11-26-155931_add-cats-and-keys-to-version/up.sql

@@ -0,0 +1,6 @@
+alter table versions
+    add column categories text[] default array[]::text[] not null,
+    add column keywords text[] default array[]::text[] not null;
+
+comment on column versions.categories is 'The list of `categories` in the `Cargo.toml` file of this version.';
+comment on column versions.keywords is 'The list of `keywords` in the `Cargo.toml` file of this version.';

src/controllers/krate/publish.rs

@@ -394,6 +394,8 @@ pub async fn publish(app: AppState, req: BytesRequest) -> AppResult<Json<GoodCra
             .maybe_homepage(homepage.as_deref())
             .maybe_documentation(documentation.as_deref())
             .maybe_repository(repository.as_deref())
+            .categories(&categories)
+            .keywords(&keywords)
             .build();
 
         let version = new_version.save(conn, &verified_email_address).await.map_err(|error| {

src/models/version.rs

@@ -102,6 +102,8 @@ pub struct NewVersion<'a> {
     homepage: Option<&'a str>,
     documentation: Option<&'a str>,
     repository: Option<&'a str>,
+    categories: Option<&'a [&'a str]>,
+    keywords: Option<&'a [&'a str]>,
 }
 
 impl NewVersion<'_> {