Add Document limits, validation (#278)

src/main/java/io/stargate/sgv2/jsonapi/config/DocumentLimitsConfig.java

@@ -0,0 +1,48 @@
+package io.stargate.sgv2.jsonapi.config;
+
+import io.smallrye.config.ConfigMapping;
+import io.smallrye.config.WithDefault;
+import javax.validation.constraints.Positive;
+
+/** Configuration Object that defines limits on Documents managed by JSON API. */
+@ConfigMapping(prefix = "stargate.jsonapi.doc-limits")
+public interface DocumentLimitsConfig {
+  /**
+   * @return Defines the maximum document page size, defaults to {@code 1 meg} (1 million
+   *     characters).
+   */
+  @Positive
+  @WithDefault("1000000")
+  int maxDocSize();
+
+  /** @return Defines the maximum document depth (nesting), defaults to {@code 8 levels} */
+  @Positive
+  @WithDefault("8")
+  int maxDocDepth();
+
+  /**
+   * @return Defines the maximum length of property names in JSON documents, defaults to {@code 48
+   *     characters} (note: length is for individual name segments; full dotted names can be longer)
+   */
+  @Positive
+  @WithDefault("48")
+  int maxNameLength();
+
+  /**
+   * @return Defines the maximum number of properties any single Object in JSON document can
+   *     contain, defaults to {@code 64} (note: this is not the total number of properties in the
+   *     whole document, only on individual main or sub-document)
+   */
+  @Positive
+  @WithDefault("64")
+  int maxObjectProperties();
+
+  /** @return Defines the maximum length of , defaults to {@code 8 levels} */
+  @Positive
+  @WithDefault("16000")
+  int maxStringLength();
+
+  @Positive
+  @WithDefault("100")
+  int maxArrayLength();
+}

src/main/java/io/stargate/sgv2/jsonapi/exception/ErrorCode.java

@@ -26,6 +26,8 @@ public enum ErrorCode {
 
   SHRED_UNRECOGNIZED_NODE_TYPE("Unrecognized JSON node type in input document"),
 
+  SHRED_DOC_LIMIT_VIOLATION("Document size limitation violated"),
+
   UNSUPPORTED_FILTER_DATA_TYPE("Unsupported filter data type"),
 
   UNSUPPORTED_FILTER_OPERATION("Unsupported filter operator"),

src/main/java/io/stargate/sgv2/jsonapi/service/shredding/Shredder.java

@@ -4,6 +4,7 @@
 import com.fasterxml.jackson.databind.ObjectMapper;
 import com.fasterxml.jackson.databind.node.ArrayNode;
 import com.fasterxml.jackson.databind.node.ObjectNode;
+import io.stargate.sgv2.jsonapi.config.DocumentLimitsConfig;
 import io.stargate.sgv2.jsonapi.config.constants.DocumentConstants;
 import io.stargate.sgv2.jsonapi.exception.ErrorCode;
 import io.stargate.sgv2.jsonapi.exception.JsonApiException;
@@ -15,6 +16,7 @@
 import java.util.Map;
 import java.util.UUID;
 import javax.enterprise.context.ApplicationScoped;
+import javax.inject.Inject;
 
 /**
  * Shred an incoming JSON document into the data we need to store in the DB, and then de-shred.
@@ -30,8 +32,12 @@
 public class Shredder {
   private final ObjectMapper objectMapper;
 
-  public Shredder(ObjectMapper objectMapper) {
+  private final DocumentLimitsConfig documentLimits;
+
+  @Inject
+  public Shredder(ObjectMapper objectMapper, DocumentLimitsConfig documentLimits) {
     this.objectMapper = objectMapper;
+    this.documentLimits = documentLimits;
   }
 
   /**
@@ -81,6 +87,10 @@ public WritableShreddedDocument shred(JsonNode doc, UUID txId) {
     } catch (IOException e) { // never happens but signature exposes it
       throw new RuntimeException(e);
     }
+    // Now that we have both the traversable document and serialization, verify
+    // it does not violate document limits:
+    validateDocument(documentLimits, docWithId, docJson);
+
     final WritableShreddedDocument.Builder b =
         WritableShreddedDocument.builder(new DocValueHasher(), docId, txId, docJson);
 
@@ -155,4 +165,103 @@ private void traverseValue(JsonNode value, ShredListener callback, JsonPath.Buil
               "%s: %s", ErrorCode.SHRED_UNRECOGNIZED_NODE_TYPE.getMessage(), value.getNodeType()));
     }
   }
+
+  private void validateDocument(DocumentLimitsConfig limits, ObjectNode doc, String docJson) {
+    // First: is the resulting document size (as serialized) too big?
+    if (docJson.length() > limits.maxDocSize()) {
+      throw new JsonApiException(
+          ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+          String.format(
+              "%s: document size (%d chars) exceeds maximum allowed (%d)",
+              ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(),
+              docJson.length(),
+              limits.maxDocSize()));
+    }
+
+    // Second: traverse to check for other constraints
+    validateObjectValue(limits, doc, 0);
+  }
+
+  private void validateDocValue(DocumentLimitsConfig limits, JsonNode value, int depth) {
+    if (value.isObject()) {
+      validateObjectValue(limits, value, depth);
+    } else if (value.isArray()) {
+      validateArrayValue(limits, value, depth);
+    } else if (value.isTextual()) {
+      validateStringValue(limits, value);
+    }
+  }
+
+  private void validateArrayValue(DocumentLimitsConfig limits, JsonNode arrayValue, int depth) {
+    ++depth;
+    validateDocDepth(limits, depth);
+
+    if (arrayValue.size() > limits.maxArrayLength()) {
+      throw new JsonApiException(
+          ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+          String.format(
+              "%s: number of elements an Array has (%d) exceeds maximum allowed (%s)",
+              ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(),
+              arrayValue.size(),
+              limits.maxArrayLength()));
+    }
+
+    for (JsonNode element : arrayValue) {
+      validateDocValue(limits, element, depth);
+    }
+  }
+
+  private void validateObjectValue(DocumentLimitsConfig limits, JsonNode objectValue, int depth) {
+    ++depth;
+    validateDocDepth(limits, depth);
+
+    if (objectValue.size() > limits.maxObjectProperties()) {
+      throw new JsonApiException(
+          ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+          String.format(
+              "%s: number of properties an Object has (%d) exceeds maximum allowed (%s)",
+              ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(),
+              objectValue.size(),
+              limits.maxObjectProperties()));
+    }
+
+    var it = objectValue.fields();
+    while (it.hasNext()) {
+      var entry = it.next();
+      final String key = entry.getKey();
+      if (key.length() > documentLimits.maxNameLength()) {
+        throw new JsonApiException(
+            ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+            String.format(
+                "%s: Property name length (%d) exceeds maximum allowed (%s)",
+                ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(),
+                key.length(),
+                limits.maxNameLength()));
+      }
+      validateDocValue(limits, entry.getValue(), depth);
+    }
+  }
+
+  private void validateStringValue(DocumentLimitsConfig limits, JsonNode stringValue) {
+    final String value = stringValue.textValue();
+    if (value.length() > limits.maxStringLength()) {
+      throw new JsonApiException(
+          ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+          String.format(
+              "%s: String value length (%d) exceeds maximum allowed (%s)",
+              ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(),
+              value.length(),
+              limits.maxStringLength()));
+    }
+  }
+
+  private void validateDocDepth(DocumentLimitsConfig limits, int depth) {
+    if (depth > limits.maxDocDepth()) {
+      throw new JsonApiException(
+          ErrorCode.SHRED_DOC_LIMIT_VIOLATION,
+          String.format(
+              "%s: document depth exceeds maximum allowed (%s)",
+              ErrorCode.SHRED_DOC_LIMIT_VIOLATION.getMessage(), limits.maxDocDepth()));
+    }
+  }
 }

src/main/resources/application.yaml

@@ -3,16 +3,16 @@
 
 stargate:
 
+  # disable all sgv2 exception mappers, handled differently
+  exception-mappers:
+    enabled: false
+
   # metrics properties
   # see io.stargate.sgv2.api.common.config.MetricsConfig for all config properties and options
   metrics:
     global-tags:
       module: sgv2-jsonapi
 
-  # disable all sgv2 exception mappers, handled differently
-  exception-mappers:
-    enabled: false
-
 quarkus:
 
   # general app properties

src/test/java/io/stargate/sgv2/jsonapi/api/v1/InsertIntegrationTest.java

@@ -14,6 +14,9 @@
 import static org.hamcrest.Matchers.nullValue;
 import static org.hamcrest.Matchers.startsWith;
 
+import com.fasterxml.jackson.databind.ObjectMapper;
+import com.fasterxml.jackson.databind.node.ArrayNode;
+import com.fasterxml.jackson.databind.node.ObjectNode;
 import io.quarkus.test.common.QuarkusTestResource;
 import io.quarkus.test.junit.QuarkusIntegrationTest;
 import io.restassured.http.ContentType;
@@ -26,7 +29,6 @@
 @QuarkusIntegrationTest
 @QuarkusTestResource(DseTestResource.class)
 public class InsertIntegrationTest extends CollectionResourceBaseIntegrationTest {
-
   @AfterEach
   public void cleanUpData() {
     deleteAllDocuments();
@@ -268,6 +270,71 @@ public void notValidDocumentMissing() {
           .body("errors[0].message", is(not(blankString())))
           .body("errors[0].exceptionClass", is("ConstraintViolationException"));
     }
+
+    @Test
+    public void tryInsertTooBigArray() {
+      final ObjectMapper mapper = new ObjectMapper();
+      // Max array elements allowed is 100; add a few more
+      ObjectNode doc = mapper.createObjectNode();
+      ArrayNode arr = doc.putArray("arr");
+      for (int i = 0; i < 500; ++i) {
+        arr.add(i);
+      }
+      final String json =
+          """
+              {
+                "insertOne": {
+                  "document": %s
+                }
+              }
+              """
+              .formatted(doc);
+      given()
+          .header(HttpConstants.AUTHENTICATION_TOKEN_HEADER_NAME, getAuthToken())
+          .contentType(ContentType.JSON)
+          .body(json)
+          .when()
+          .post(CollectionResource.BASE_PATH, keyspaceId.asInternal(), collectionName)
+          .then()
+          .statusCode(200)
+          .body("errors[0].errorCode", is("SHRED_DOC_LIMIT_VIOLATION"))
+          .body(
+              "errors[0].message",
+              is(
+                  "Document size limitation violated: number of elements an Array has (500) exceeds maximum allowed (100)"));
+    }
+
+    @Test
+    public void tryInsertTooLongName() {
+      final ObjectMapper mapper = new ObjectMapper();
+      // Max property name: 48 characters, let's try 100
+      ObjectNode doc = mapper.createObjectNode();
+      doc.put(
+          "prop_12345_123456789_123456789_123456789_123456789_123456789_123456789_123456789_123456789_123456789",
+          72);
+      final String json =
+          """
+                  {
+                    "insertOne": {
+                      "document": %s
+                    }
+                  }
+                  """
+              .formatted(doc);
+      given()
+          .header(HttpConstants.AUTHENTICATION_TOKEN_HEADER_NAME, getAuthToken())
+          .contentType(ContentType.JSON)
+          .body(json)
+          .when()
+          .post(CollectionResource.BASE_PATH, keyspaceId.asInternal(), collectionName)
+          .then()
+          .statusCode(200)
+          .body("errors[0].errorCode", is("SHRED_DOC_LIMIT_VIOLATION"))
+          .body(
+              "errors[0].message",
+              is(
+                  "Document size limitation violated: Property name length (100) exceeds maximum allowed (48)"));
+    }
   }
 
   @Nested