verify checksum against regexes

5 files changed, 130 insertions, 20 deletions

diff --git a/rust/src/api_entity_crud.rs b/rust/src/api_entity_crud.rs
index 2f28e858..792e6f9a 100644
--- a/rust/src/api_entity_crud.rs
+++ b/rust/src/api_entity_crud.rs
@@ -885,6 +885,19 @@ impl EntityCrud for FileEntity {
     }
 
     fn db_insert_revs(conn: &DbConn, models: &[&Self]) -> Result<Vec<Uuid>> {
+        // first verify hash syntax
+        for entity in models {
+            if let Some(ref hash) = entity.md5 {
+                check_md5(hash)?;
+            }
+            if let Some(ref hash) = entity.sha1 {
+                check_sha1(hash)?;
+            }
+            if let Some(ref hash) = entity.sha256 {
+                check_sha256(hash)?;
+            }
+        }
+
         let rev_ids: Vec<Uuid> = insert_into(file_rev::table)
             .values(
                 models
diff --git a/rust/src/api_helpers.rs b/rust/src/api_helpers.rs
index b6525546..32750836 100644
--- a/rust/src/api_helpers.rs
+++ b/rust/src/api_helpers.rs
@@ -391,6 +391,78 @@ fn test_check_orcid() {
     assert!(check_orcid("0x23-4567-3456-6780").is_err());
 }
 
+pub fn check_md5(raw: &str) -> Result<()> {
+    lazy_static! {
+        static ref RE: Regex = Regex::new(r"^[a-f0-9]{32}$").unwrap();
+    }
+    if RE.is_match(raw) {
+        Ok(())
+    } else {
+        Err(ErrorKind::MalformedChecksum(format!(
+            "not a valid MD5: '{}' (expected lower-case hex, eg, '1b39813549077b2347c0f370c3864b40')",
+            raw
+        ))
+        .into())
+    }
+}
+
+#[test]
+fn test_check_md5() {
+    assert!(check_md5("1b39813549077b2347c0f370c3864b40").is_ok());
+    assert!(check_md5("1g39813549077b2347c0f370c3864b40").is_err());
+    assert!(check_md5("1B39813549077B2347C0F370c3864b40").is_err());
+    assert!(check_md5("1b39813549077b2347c0f370c3864b4").is_err());
+    assert!(check_md5("1b39813549077b2347c0f370c3864b411").is_err());
+}
+
+pub fn check_sha1(raw: &str) -> Result<()> {
+    lazy_static! {
+        static ref RE: Regex = Regex::new(r"^[a-f0-9]{40}$").unwrap();
+    }
+    if RE.is_match(raw) {
+        Ok(())
+    } else {
+        Err(ErrorKind::MalformedChecksum(format!(
+            "not a valid SHA-1: '{}' (expected lower-case hex, eg, 'e9dd75237c94b209dc3ccd52722de6931a310ba3')",
+            raw
+        ))
+        .into())
+    }
+}
+
+#[test]
+fn test_check_sha1() {
+    assert!(check_sha1("e9dd75237c94b209dc3ccd52722de6931a310ba3").is_ok());
+    assert!(check_sha1("g9dd75237c94b209dc3ccd52722de6931a310ba3").is_err());
+    assert!(check_sha1("e9DD75237C94B209DC3CCD52722de6931a310ba3").is_err());
+    assert!(check_sha1("e9dd75237c94b209dc3ccd52722de6931a310ba").is_err());
+    assert!(check_sha1("e9dd75237c94b209dc3ccd52722de6931a310ba33").is_err());
+}
+
+pub fn check_sha256(raw: &str) -> Result<()> {
+    lazy_static! {
+        static ref RE: Regex = Regex::new(r"^[a-f0-9]{64}$").unwrap();
+    }
+    if RE.is_match(raw) {
+        Ok(())
+    } else {
+        Err(ErrorKind::MalformedChecksum(format!(
+            "not a valid SHA-256: '{}' (expected lower-case hex, eg, 'cb1c378f464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e452')",
+            raw
+        ))
+        .into())
+    }
+}
+
+#[test]
+fn test_check_sha256() {
+    assert!(check_sha256("cb1c378f464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e452").is_ok());
+    assert!(check_sha256("gb1c378f464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e452").is_err());
+    assert!(check_sha256("UB1C378F464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e452").is_err());
+    assert!(check_sha256("cb1c378f464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e45").is_err());
+    assert!(check_sha256("cb1c378f464d5935ddaa8de28446d82638396c61f042295d7fb85e3cccc9e4522").is_err());
+}
+
 pub fn check_release_type(raw: &str) -> Result<()> {
     let valid_types = vec![
         // Citation Style Language official types
@@ -499,4 +571,3 @@ fn test_check_contrib_role() {
 
 // TODO: make the above checks "more correct"
 // TODO: check ISBN-13
-// TODO: check hashes (SHA-1, etc)
diff --git a/rust/src/api_server.rs b/rust/src/api_server.rs
index d03fce07..adc6ab11 100644
--- a/rust/src/api_server.rs
+++ b/rust/src/api_server.rs
@@ -178,24 +178,33 @@ impl Server {
         conn: &DbConn,
     ) -> Result<FileEntity> {
         let (ident, rev): (FileIdentRow, FileRevRow) = match (md5, sha1, sha256) {
-            (Some(md5), None, None) => file_ident::table
-                .inner_join(file_rev::table)
-                .filter(file_rev::md5.eq(md5))
-                .filter(file_ident::is_live.eq(true))
-                .filter(file_ident::redirect_id.is_null())
-                .first(conn)?,
-            (None, Some(sha1), None) => file_ident::table
-                .inner_join(file_rev::table)
-                .filter(file_rev::sha1.eq(sha1))
-                .filter(file_ident::is_live.eq(true))
-                .filter(file_ident::redirect_id.is_null())
-                .first(conn)?,
-            (None, None, Some(sha256)) => file_ident::table
-                .inner_join(file_rev::table)
-                .filter(file_rev::sha256.eq(sha256))
-                .filter(file_ident::is_live.eq(true))
-                .filter(file_ident::redirect_id.is_null())
-                .first(conn)?,
+            (Some(md5), None, None) => {
+                check_md5(md5)?;
+                file_ident::table
+                    .inner_join(file_rev::table)
+                    .filter(file_rev::md5.eq(md5))
+                    .filter(file_ident::is_live.eq(true))
+                    .filter(file_ident::redirect_id.is_null())
+                    .first(conn)?
+            },
+            (None, Some(sha1), None) => {
+                check_sha1(sha1)?;
+                file_ident::table
+                    .inner_join(file_rev::table)
+                    .filter(file_rev::sha1.eq(sha1))
+                    .filter(file_ident::is_live.eq(true))
+                    .filter(file_ident::redirect_id.is_null())
+                    .first(conn)?
+            },
+            (None, None, Some(sha256)) => {
+                check_sha256(sha256)?;
+                file_ident::table
+                    .inner_join(file_rev::table)
+                    .filter(file_rev::sha256.eq(sha256))
+                    .filter(file_ident::is_live.eq(true))
+                    .filter(file_ident::redirect_id.is_null())
+                    .first(conn)?
+            },
             _ => {
                 return Err(ErrorKind::MissingOrMultipleExternalId("in lookup".to_string()).into());
             }
diff --git a/rust/src/api_wrappers.rs b/rust/src/api_wrappers.rs
index fe9cd793..c3df7d72 100644
--- a/rust/src/api_wrappers.rs
+++ b/rust/src/api_wrappers.rs
@@ -102,6 +102,8 @@ macro_rules! wrap_entity_handlers {
                         message: ErrorKind::InvalidFatcatId(e).to_string() }),
                 Err(Error(ErrorKind::MalformedExternalId(e), _)) =>
                     $post_resp::BadRequest(ErrorResponse { message: e.to_string() }),
+                Err(Error(ErrorKind::MalformedChecksum(e), _)) =>
+                    $post_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::NotInControlledVocabulary(e), _)) =>
                     $post_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::EditgroupAlreadyAccepted(e), _)) =>
@@ -141,6 +143,8 @@ macro_rules! wrap_entity_handlers {
                         message: ErrorKind::InvalidFatcatId(e).to_string() }),
                 Err(Error(ErrorKind::MalformedExternalId(e), _)) =>
                     $post_batch_resp::BadRequest(ErrorResponse { message: e.to_string() }),
+                Err(Error(ErrorKind::MalformedChecksum(e), _)) =>
+                    $post_batch_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::NotInControlledVocabulary(e), _)) =>
                     $post_batch_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::EditgroupAlreadyAccepted(e), _)) =>
@@ -185,6 +189,8 @@ macro_rules! wrap_entity_handlers {
                         message: ErrorKind::InvalidFatcatId(e).to_string() }),
                 Err(Error(ErrorKind::MalformedExternalId(e), _)) =>
                     $update_resp::BadRequest(ErrorResponse { message: e.to_string() }),
+                Err(Error(ErrorKind::MalformedChecksum(e), _)) =>
+                    $update_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::NotInControlledVocabulary(e), _)) =>
                     $update_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::EditgroupAlreadyAccepted(e), _)) =>
@@ -431,6 +437,8 @@ macro_rules! wrap_lookup_handler {
                     $get_resp::NotFound(ErrorResponse { message: format!("Not found: {:?} / {:?}", $idname, wikidata_qid) }),
                 Err(Error(ErrorKind::MalformedExternalId(e), _)) =>
                     $get_resp::BadRequest(ErrorResponse { message: e.to_string() }),
+                Err(Error(ErrorKind::MalformedChecksum(e), _)) =>
+                    $get_resp::BadRequest(ErrorResponse { message: e.to_string() }),
                 Err(Error(ErrorKind::MissingOrMultipleExternalId(e), _)) => {
                     $get_resp::BadRequest(ErrorResponse { message: e.to_string(), }) },
                 Err(Error(ErrorKind::OtherBadRequest(e), _)) =>
@@ -706,6 +714,11 @@ impl Api for Server {
                         message: e.to_string(),
                     })
                 }
+                Err(Error(ErrorKind::MalformedChecksum(e), _)) => {
+                    LookupFileResponse::BadRequest(ErrorResponse {
+                        message: e.to_string(),
+                    })
+                }
                 Err(Error(ErrorKind::MissingOrMultipleExternalId(e), _)) => {
                     LookupFileResponse::BadRequest(ErrorResponse {
                         message: e.to_string(),
diff --git a/rust/src/lib.rs b/rust/src/lib.rs
index b7f1817a..0bed3471 100644
--- a/rust/src/lib.rs
+++ b/rust/src/lib.rs
@@ -49,7 +49,11 @@ pub mod errors {
             }
             MalformedExternalId(id: String) {
                 description("external identifier doesn't match required pattern")
-                display("external identifier doesn't match required pattern")
+                display("external identifier doesn't match required pattern: {}", id)
+            }
+            MalformedChecksum(hash: String) {
+                description("checksum doesn't match required pattern (hex encoding)")
+                display("checksum doesn't match required pattern (hex encoding): {}", hash)
             }
             NotInControlledVocabulary(word: String) {
                 description("word or type not correct for controlled vocabulary")