Subgraph Composition: More validations

graph/src/data/subgraph/mod.rs

@@ -672,6 +672,73 @@ pub type SubgraphManifest<C> =
 pub struct UnvalidatedSubgraphManifest<C: Blockchain>(SubgraphManifest<C>);
 
 impl<C: Blockchain> UnvalidatedSubgraphManifest<C> {
+    fn validate_subgraph_datasources(
+        data_sources: &[DataSource<C>],
+        spec_version: &Version,
+    ) -> Vec<SubgraphManifestValidationError> {
+        let mut errors = Vec::new();
+
+        // Check spec version support for subgraph datasources
+        if *spec_version < SPEC_VERSION_1_3_0 {
+            if data_sources
+                .iter()
+                .any(|ds| matches!(ds, DataSource::Subgraph(_)))
+            {
+                errors.push(SubgraphManifestValidationError::DataSourceValidation(
+                    "subgraph".to_string(),
+                    anyhow!(
+                        "Subgraph datasources are not supported prior to spec version {}",
+                        SPEC_VERSION_1_3_0
+                    ),
+                ));
+                return errors;
+            }
+        }
+
+        let subgraph_ds_count = data_sources
+            .iter()
+            .filter(|ds| matches!(ds, DataSource::Subgraph(_)))
+            .count();
+
+        if subgraph_ds_count > 5 {
+            errors.push(SubgraphManifestValidationError::DataSourceValidation(
+                "subgraph".to_string(),
+                anyhow!("Cannot have more than 5 subgraph datasources"),
+            ));
+        }
+
+        let has_subgraph_ds = subgraph_ds_count > 0;
+        let has_onchain_ds = data_sources
+            .iter()
+            .any(|d| matches!(d, DataSource::Onchain(_)));
+
+        if has_subgraph_ds && has_onchain_ds {
+            errors.push(SubgraphManifestValidationError::DataSourceValidation(
+                "subgraph".to_string(),
+                anyhow!("Subgraph datasources cannot be used alongside onchain datasources"),
+            ));
+        }
+
+        // Check for duplicate source subgraphs
+        let mut seen_sources = std::collections::HashSet::new();
+        for ds in data_sources.iter() {
+            if let DataSource::Subgraph(ds) = ds {
+                let source_id = ds.source.address();
+                if !seen_sources.insert(source_id.clone()) {
+                    errors.push(SubgraphManifestValidationError::DataSourceValidation(
+                        "subgraph".to_string(),
+                        anyhow!(
+                            "Multiple subgraph datasources cannot use the same source subgraph {}",
+                            source_id
+                        ),
+                    ));
+                }
+            }
+        }
+
+        errors
+    }
+
     /// Entry point for resolving a subgraph definition.
     /// Right now the only supported links are of the form:
     /// `/ipfs/QmUmg7BZC1YP1ca66rRtWKxpXp77WgVHrnv263JtDuvs2k`
@@ -742,6 +809,12 @@ impl<C: Blockchain> UnvalidatedSubgraphManifest<C> {
             }
         }
 
+        // Validate subgraph datasource constraints
+        errors.extend(Self::validate_subgraph_datasources(
+            &self.0.data_sources,
+            &self.0.spec_version,
+        ));
+
         match errors.is_empty() {
             true => Ok(self.0),
             false => Err(errors),
@@ -1005,6 +1078,17 @@ impl<C: Blockchain> UnresolvedSubgraphManifest<C> {
             );
         }
 
+        // Validate subgraph datasource constraints
+        if let Some(error) = UnvalidatedSubgraphManifest::<C>::validate_subgraph_datasources(
+            &data_sources,
+            &spec_version,
+        )
+        .into_iter()
+        .next()
+        {
+            return Err(anyhow::Error::from(error).into());
+        }
+
         // Check the min_spec_version of each data source against the spec version of the subgraph
         let min_spec_version_mismatch = data_sources
             .iter()

graph/src/data_source/subgraph.rs

@@ -287,6 +287,14 @@ impl UnresolvedDataSource {
         let source_manifest = self.resolve_source_manifest::<C>(resolver, logger).await?;
         let source_spec_version = &source_manifest.spec_version;
 
+        if source_manifest
+            .data_sources
+            .iter()
+            .any(|ds| matches!(ds, crate::data_source::DataSource::Subgraph(_)))
+        {
+            return Err(anyhow!("Nested subgraph data sources are not supported."));
+        }
+
         if source_spec_version < &SPEC_VERSION_1_3_0 {
             return Err(anyhow!(
                 "Source subgraph manifest spec version {} is not supported, minimum supported version is {}",

store/test-store/tests/chain/ethereum/manifest.rs

@@ -20,7 +20,8 @@ use graph::env::ENV_VARS;
 use graph::prelude::web3::types::H256;
 use graph::prelude::{
     anyhow, async_trait, serde_yaml, tokio, BigDecimal, BigInt, DeploymentHash, Link, Logger,
-    SubgraphManifest, SubgraphManifestValidationError, SubgraphStore, UnvalidatedSubgraphManifest,
+    SubgraphManifest, SubgraphManifestResolveError, SubgraphManifestValidationError, SubgraphStore,
+    UnvalidatedSubgraphManifest,
 };
 use graph::{
     blockchain::NodeCapabilities as _,
@@ -317,7 +318,7 @@ dataSources:
         - Profile
     network: mainnet
     source:
-      address: 'QmSource'
+      address: 'QmSource2'
       startBlock: 9562500
     mapping:
       apiVersion: 0.0.6
@@ -1706,3 +1707,153 @@ dataSources:
         assert_eq!(3, decls.len());
     });
 }
+
+#[tokio::test]
+async fn mixed_subgraph_and_onchain_ds_manifest_should_fail() {
+    let yaml = "
+schema:
+  file:
+    /: /ipfs/Qmschema
+dataSources:
+  - name: SubgraphSource
+    kind: subgraph
+    entities:
+        - User
+    network: mainnet
+    source: 
+      address: 'QmSource'
+      startBlock: 9562480
+    mapping:
+      apiVersion: 0.0.6
+      language: wasm/assemblyscript
+      entities:
+        - TestEntity
+      file:
+        /: /ipfs/Qmmapping
+      handlers:
+        - handler: handleEntity
+          entity: User
+  - kind: ethereum/contract
+    name: Gravity
+    network: mainnet
+    source:
+      address: '0x2E645469f354BB4F5c8a05B3b30A929361cf77eC'
+      abi: Gravity
+      startBlock: 1
+    mapping:
+      kind: ethereum/events
+      apiVersion: 0.0.6
+      language: wasm/assemblyscript
+      entities:
+        - Gravatar
+      abis:
+        - name: Gravity
+          file:
+            /: /ipfs/Qmabi
+      file:
+        /: /ipfs/Qmmapping
+      handlers:
+        - event: NewGravatar(uint256,address,string,string)
+          handler: handleNewGravatar
+specVersion: 1.3.0
+";
+
+    let result = try_resolve_manifest(yaml, SPEC_VERSION_1_3_0).await;
+    assert!(result.is_err());
+    let err = result.unwrap_err();
+    assert!(err
+        .to_string()
+        .contains("Subgraph datasources cannot be used alongside onchain datasources"));
+}
+
+#[test]
+fn nested_subgraph_ds_manifest_should_fail() {
+    let yaml = r#"
+schema:
+  file:
+    /: /ipfs/Qmschema
+dataSources:
+- name: SubgraphSource
+  kind: subgraph
+  entities:
+      - User
+  network: mainnet
+  source: 
+    address: 'QmNestedSource'
+    startBlock: 9562480
+  mapping:
+    apiVersion: 0.0.6
+    language: wasm/assemblyscript
+    entities:
+      - TestEntity
+    file:
+      /: /ipfs/Qmmapping
+    handlers:
+      - handler: handleEntity
+        entity: User
+specVersion: 1.3.0
+"#;
+
+    // First modify SOURCE_SUBGRAPH_MANIFEST to include a subgraph datasource
+    const NESTED_SOURCE_MANIFEST: &str = r#"
+schema:
+  file:
+    /: /ipfs/QmSourceSchema
+dataSources:
+- kind: subgraph
+  name: NestedSource
+  network: mainnet
+  entities:
+      - User
+  source:
+    address: 'QmSource'
+    startBlock: 1
+  mapping:
+    apiVersion: 0.0.6
+    language: wasm/assemblyscript
+    entities:
+      - User
+    file:
+      /: /ipfs/Qmmapping
+    handlers:
+      - handler: handleNested
+        entity: User
+specVersion: 1.3.0
+"#;
+
+    let mut resolver = TextResolver::default();
+    let id = DeploymentHash::new("Qmmanifest").unwrap();
+
+    resolver.add(id.as_str(), &yaml);
+    resolver.add("/ipfs/Qmabi", &ABI);
+    resolver.add("/ipfs/Qmschema", &GQL_SCHEMA);
+    resolver.add("/ipfs/Qmmapping", &MAPPING_WITH_IPFS_FUNC_WASM);
+    resolver.add("/ipfs/QmNestedSource", &NESTED_SOURCE_MANIFEST);
+    resolver.add("/ipfs/QmSource", &SOURCE_SUBGRAPH_MANIFEST);
+    resolver.add("/ipfs/QmSourceSchema", &SOURCE_SUBGRAPH_SCHEMA);
+
+    let resolver: Arc<dyn LinkResolverTrait> = Arc::new(resolver);
+
+    let raw = serde_yaml::from_str(yaml).unwrap();
+    test_store::run_test_sequentially(|_| async move {
+        let result: Result<UnvalidatedSubgraphManifest<Chain>, _> =
+            UnvalidatedSubgraphManifest::resolve(
+                id,
+                raw,
+                &resolver,
+                &LOGGER,
+                SPEC_VERSION_1_3_0.clone(),
+            )
+            .await;
+
+        match result {
+            Ok(_) => panic!("Expected resolution to fail"),
+            Err(e) => {
+                assert!(matches!(e, SubgraphManifestResolveError::ResolveError(_)));
+                let error_msg = e.to_string();
+                println!("{}", error_msg);
+                assert!(error_msg.contains("Nested subgraph data sources are not supported."));
+            }
+        }
+    })
+}