Add checksum_algorithm field to specify details of location checksum

Author: rhigman

CHANGELOG.md

@@ -6,7 +6,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 ### Added
-  - [747](https://github.com/thoth-pub/thoth/pull/747) - Add `Location.checksum` field
+  - [747](https://github.com/thoth-pub/thoth/pull/747) - Add `checksum` and `checksum_algorithm` fields to `Location`
 
 ## [[1.1.1]](https://github.com/thoth-pub/thoth/releases/tag/v1.1.1) - 2026-04-24
 ### Security

thoth-api/migrations/20260429_v1.2.0/down.sql

@@ -1,2 +1,6 @@
 ALTER TABLE public.location
-    DROP COLUMN IF EXISTS checksum;
+    DROP CONSTRAINT IF EXISTS location_checksum_and_algorithm_all_or_none,
+    DROP COLUMN IF EXISTS checksum,
+    DROP COLUMN IF EXISTS checksum_algorithm,
+
+DROP TYPE IF EXISTS public.checksum_algorithm;

thoth-api/migrations/20260429_v1.2.0/up.sql

@@ -1,2 +1,9 @@
+CREATE TYPE public.checksum_algorithm AS ENUM (
+    'MD5',
+    'SHA256'
+);
+
 ALTER TABLE public.location
-    ADD COLUMN checksum TEXT;
+    ADD COLUMN checksum TEXT,
+    ADD COLUMN checksum_algorithm public.checksum_algorithm,
+    ADD CONSTRAINT location_checksum_and_algorithm_all_or_none CHECK ((checksum IS NULL AND checksum_algorithm IS NULL) OR (checksum IS NOT NULL AND checksum_algorithm IS NOT NULL));

thoth-api/src/graphql/model.rs

@@ -21,7 +21,7 @@ use crate::model::{
     contribution::{Contribution, ContributionType},
     contributor::Contributor,
     endorsement::{Endorsement, EndorsementOrderBy},
-    file::{File, FileType},
+    file::{ChecksumAlgorithm, File, FileType},
     funding::Funding,
     imprint::{Imprint, ImprintField, ImprintOrderBy},
     institution::Institution,
@@ -1921,6 +1921,11 @@ impl Location {
         self.checksum.as_ref()
     }
 
+    #[graphql(description = "Algorithm used to generate the checksum (MD5 or SHA-256)")]
+    pub fn checksum_algorithm(&self) -> Option<&ChecksumAlgorithm> {
+        self.checksum_algorithm.as_ref()
+    }
+
     #[graphql(description = "Date and time at which the location record was created")]
     pub fn created_at(&self) -> Timestamp {
         self.created_at

thoth-api/src/graphql/tests.rs

@@ -471,6 +471,7 @@ fn make_new_location(publication_id: Uuid, canonical: bool) -> NewLocation {
         location_platform: LocationPlatform::Other,
         canonical,
         checksum: None,
+        checksum_algorithm: None,
     }
 }
 
@@ -1106,6 +1107,7 @@ fn patch_location(location: &Location) -> PatchLocation {
         location_platform: location.location_platform,
         canonical: location.canonical,
         checksum: location.checksum.clone(),
+        checksum_algorithm: location.checksum_algorithm,
     }
 }
 

thoth-api/src/model/file/crud.rs

@@ -1,7 +1,6 @@
-use super::FileType;
 use super::{
-    upload_request_headers, File, FileCleanupCandidate, FilePolicy, FileUpload, FileUploadResponse,
-    NewFile, NewFileUpload,
+    upload_request_headers, ChecksumAlgorithm, File, FileCleanupCandidate, FilePolicy, FileType,
+    FileUpload, FileUploadResponse, NewFile, NewFileUpload,
 };
 use crate::db::PgPool;
 use crate::model::{
@@ -724,7 +723,7 @@ impl FileUpload {
         ctx: &C,
         work: &Work,
         cdn_url: &str,
-        cdn_checksum: &str,
+        cdn_sha256: &str,
         featured_video_dimensions: Option<(i32, i32)>,
     ) -> ThothResult<()> {
         match self.file_type {
@@ -742,7 +741,7 @@ impl FileUpload {
                     publication_id,
                     work.landing_page.clone(),
                     cdn_url,
-                    Some(cdn_checksum.to_string()),
+                    Some(cdn_sha256.to_string()),
                 )?;
             }
             FileType::AdditionalResource => {
@@ -794,7 +793,7 @@ impl FileUpload {
         publication_id: Uuid,
         landing_page: Option<String>,
         full_text_url: &str,
-        checksum: Option<String>,
+        sha256: Option<String>,
     ) -> ThothResult<()> {
         use crate::schema::location::dsl;
 
@@ -812,7 +811,8 @@ impl FileUpload {
             patch.full_text_url = Some(full_text_url.to_string());
             patch.landing_page = landing_page;
             patch.canonical = true;
-            patch.checksum = checksum;
+            patch.checksum = sha256;
+            patch.checksum_algorithm = Some(ChecksumAlgorithm::Sha256);
             if patch.canonical {
                 patch.canonical_record_complete(ctx.db())?;
             }
@@ -834,7 +834,8 @@ impl FileUpload {
                 full_text_url: Some(full_text_url.to_string()),
                 location_platform: LocationPlatform::Thoth,
                 canonical: false,
-                checksum,
+                checksum: sha256,
+                checksum_algorithm: Some(ChecksumAlgorithm::Sha256),
             };
             let created_location = Location::create(ctx.db(), &new_location)?;
             let mut patch = PatchLocation::from(created_location.clone());
@@ -850,7 +851,8 @@ impl FileUpload {
                 full_text_url: Some(full_text_url.to_string()),
                 location_platform: LocationPlatform::Thoth,
                 canonical: true,
-                checksum,
+                checksum: sha256,
+                checksum_algorithm: Some(ChecksumAlgorithm::Sha256),
             };
             new_location.canonical_record_complete(ctx.db())?;
             Location::create(ctx.db(), &new_location)?;

thoth-api/src/model/file/mod.rs

@@ -50,6 +50,22 @@ pub enum FileType {
     WorkFeaturedVideo,
 }
 
+#[cfg_attr(
+    feature = "backend",
+    derive(diesel_derive_enum::DbEnum, juniper::GraphQLEnum),
+    graphql(description = "Algorithm used to create file checksum"),
+    ExistingTypePath = "crate::schema::sql_types::ChecksumAlgorithm"
+)]
+#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, EnumString, Display)]
+#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
+#[strum(serialize_all = "SCREAMING_SNAKE_CASE")]
+pub enum ChecksumAlgorithm {
+    #[cfg_attr(feature = "backend", db_rename = "MD5")]
+    Md5,
+    #[cfg_attr(feature = "backend", db_rename = "SHA256")]
+    Sha256,
+}
+
 #[cfg_attr(feature = "backend", derive(diesel::Queryable))]
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
 #[serde(rename_all = "camelCase")]

thoth-api/src/model/location/crud.rs

@@ -67,6 +67,9 @@ impl Crud for Location {
             LocationField::Checksum => {
                 apply_directional_order!(query, order.direction, order, checksum)
             }
+            LocationField::ChecksumAlgorithm => {
+                apply_directional_order!(query, order.direction, order, checksum_algorithm)
+            }
             LocationField::CreatedAt => {
                 apply_directional_order!(query, order.direction, order, created_at)
             }

thoth-api/src/model/location/mod.rs

@@ -4,7 +4,7 @@ use strum::EnumString;
 use uuid::Uuid;
 
 use crate::graphql::types::inputs::Direction;
-use crate::model::Timestamp;
+use crate::model::{file::ChecksumAlgorithm, Timestamp};
 #[cfg(feature = "backend")]
 use crate::schema::location;
 #[cfg(feature = "backend")]
@@ -166,6 +166,7 @@ pub enum LocationField {
     LocationPlatform,
     Canonical,
     Checksum,
+    ChecksumAlgorithm,
     CreatedAt,
     UpdatedAt,
 }
@@ -181,6 +182,7 @@ pub struct Location {
     pub location_platform: LocationPlatform,
     pub canonical: bool,
     pub checksum: Option<String>,
+    pub checksum_algorithm: Option<ChecksumAlgorithm>,
     pub created_at: Timestamp,
     pub updated_at: Timestamp,
 }
@@ -198,6 +200,7 @@ pub struct NewLocation {
     pub location_platform: LocationPlatform,
     pub canonical: bool,
     pub checksum: Option<String>,
+    pub checksum_algorithm: Option<ChecksumAlgorithm>,
 }
 
 #[cfg_attr(
@@ -214,6 +217,7 @@ pub struct PatchLocation {
     pub location_platform: LocationPlatform,
     pub canonical: bool,
     pub checksum: Option<String>,
+    pub checksum_algorithm: Option<ChecksumAlgorithm>,
 }
 
 #[cfg_attr(feature = "backend", derive(diesel::Queryable))]
@@ -265,6 +269,7 @@ impl From<Location> for PatchLocation {
             location_platform: location.location_platform,
             canonical: location.canonical,
             checksum: location.checksum,
+            checksum_algorithm: location.checksum_algorithm,
         }
     }
 }