Validate span.type and .subtype against shared spec (#1030)

lib/elastic_apm/span.rb

@@ -156,6 +156,8 @@ def inspect
       "<ElasticAPM::Span id:#{trace_context&.id}" \
         " name:#{name.inspect}" \
         " type:#{type.inspect}" \
+        " subtype:#{subtype.inspect}" \
+        " action:#{action.inspect}" \
         '>'
     end
 

lib/elastic_apm/subscriber.rb

@@ -45,6 +45,7 @@ def unregister!
     # AS::Notifications API
 
     Notification = Struct.new(:id, :span)
+
     def start(name, id, payload)
       return unless (transaction = @agent.current_transaction)
 

spec/entrypoint.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 set -x
 
-bundle check || (rm Gemfile.lock && bundle)
+bundle check || (rm -f Gemfile.lock && bundle)
 
 # If first arg is a spec path, run spec(s)
 if [[ $1 == spec/* ]]; then

spec/fixtures/span_types.json

@@ -0,0 +1,279 @@
+{
+  "__description": {
+    "<type>": "root element for type identified by '<type>'",
+    "<type>.__description": "description for '<type>' (optional)",
+    "<type>.__used_by": "list of agents that use '<type>' to help document alignment (optional)",
+    "<type>.allow_null_subtype": "true to allow null subtype, false by default if omitted",
+    "<type>.allow_unlisted_subtype": "true to allow unlisted subtypes, false by default if omitted",
+    "<type>.subtypes": "root element for sub-types of type '<type>', if omitted or empty subtype must be null, unless 'allow_unlisted_subtype' is set to true",
+    "<type>.subtypes.<subtype>": "sub-type element for <subtype>",
+    "<type>.subtypes.<subtype>.__description": "description of <subtype> subtype (optional)",
+    "<type>.subtypes.<subtype>.__used_by": "list of agents that use <subtype> to help document alignment (optional)"
+  },
+  "app": {
+    "allow_null_subtype": true,
+    "subtypes": {
+      "inferred": {
+        "__description": "Sampling profiler inferred spans",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "controller": {
+        "__description": "Application controller actions",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "graphql": {
+        "__description": "Incoming GraphQL requests",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "mailer": {
+        "__description": "Application mailer actions",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "resource": {
+        "__description": "Application resource actions",
+        "__used_by": [
+          "ruby"
+        ]
+      }
+    }
+  },
+  "custom": {
+    "__description": "API custom instrumentation",
+    "__used_by": [
+      "java",
+      "ruby"
+    ],
+    "allow_null_subtype": true
+  },
+  "db": {
+    "__description": "database span",
+    "subtypes": {
+      "cassandra": {
+        "__description": "Cassandra",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "cosmosdb": {
+        "__description": "Azure CosmosDB"
+      },
+      "db2": {
+        "__description": "IBM DB2",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "derby": {
+        "__description": "Apache Derby",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "dynamodb": {
+        "__description": "AWS DynamoDB",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "elasticsearch": {
+        "__description": "Elasticsearch",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      },
+      "h2": {
+        "__description": "H2",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "hsqldb": {
+        "__description": "HSQLDB",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "ingres": {
+        "__description": "Ingres"
+      },
+      "mariadb": {
+        "__description": "MariaDB",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      },
+      "mongodb": {
+        "__description": "MongoDB",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      },
+      "mysql": {
+        "__description": "MySQL",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      },
+      "oracle": {
+        "__description": "Oracle Database",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "postgresql": {
+        "__description": "PostgreSQL",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "redis": {
+        "__description": "Redis",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      },
+      "sqlite": {
+        "__description": "SQLite",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "sqlite3": {
+        "__description": "SQLite 3",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "sqlserver": {
+        "__description": "Microsoft SQL Server",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "unknown": {
+        "__description": "Unknown database",
+        "__used_by": [
+          "java",
+          "ruby"
+        ]
+      }
+    }
+  },
+  "external": {
+    "subtypes": {
+      "dubbo": {
+        "__description": "Apache Dubbo"
+      },
+      "grpc": {
+        "__description": "gRPC",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "http": {
+        "__description": "HTTP client",
+        "__used_by": [
+          "ruby"
+        ]
+      }
+    }
+  },
+  "json": {
+    "__description": "JSON parsing and generation",
+    "subtypes": {
+      "parse": {
+        "__description": "JSON parsing"
+      },
+      "generate": {
+        "__description": "JSON generation"
+      }
+    },
+    "__used_by": [
+      "ruby"
+    ]
+  },
+  "messaging": {
+    "__description": "Messaging",
+    "subtypes": {
+      "azurequeue": {
+        "__description": "Azure Queue"
+      },
+      "azureservicebus": {
+        "__description": "Azure Service Bus"
+      },
+      "jms": {
+        "__description": "Java Messaging Service",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "kafka": {
+        "__description": "Apache Kafka",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "rabbitmq": {
+        "__description": "RabbitMQ",
+        "__used_by": [
+          "java"
+        ]
+      },
+      "sns": {
+        "__description": "AWS Simple Notification Service",
+        "__used_by": [
+          "ruby"
+        ]
+      },
+      "sqs": {
+        "__description": "AWS Simple Queue Service",
+        "__used_by": [
+          "ruby"
+        ]
+      }
+    }
+  },
+  "process": {
+    "__description": "External process",
+    "__used_by": [
+      "java"
+    ]
+  },
+  "storage": {
+    "subtypes": {
+      "azurefile": {
+        "__description": "Azure Files"
+      },
+      "azuresblob": {
+        "__description": "Azure Blob Storage"
+      },
+      "s3": {
+        "__description": "AWS S3",
+        "__used_by": [
+          "ruby"
+        ]
+      }
+    }
+  },
+  "template": {
+    "__description": "Template engines (no sub-type for now as really platform-specific)",
+    "__used_by": [
+      "java",
+      "ruby"
+    ],
+    "allow_unlisted_subtype": true
+  }
+}

spec/support/intercept.rb

@@ -24,6 +24,8 @@ def initialize
       @spans = []
       @errors = []
       @metricsets = []
+
+      @span_types = JSON.parse(File.read('./spec/fixtures/span_types.json'))
     end
 
     attr_reader :transactions, :spans, :errors, :metricsets
@@ -33,6 +35,7 @@ def submit(obj)
       when ElasticAPM::Transaction
         transactions << obj
       when ElasticAPM::Span
+        validate_span!(obj)
         spans << obj
       when ElasticAPM::Error
         errors << obj
@@ -46,6 +49,26 @@ def submit(obj)
     def start; end
 
     def stop; end
+
+    def validate_span!(span)
+      type, subtype = [span.type, span.subtype]
+
+      begin
+        info = @span_types.fetch(type)
+      rescue KeyError
+        raise "Unknown span.type `#{type}'\nPossible types: #{@span_types.keys.join(', ')}"
+      end
+
+      return unless (allowed_subtypes = info['subtypes'])
+
+      if !info['allow_null_subtype'] && !subtype
+        raise "span.subtype missing when required,\nPossible subtypes: #{allowed_subtypes}"
+      end
+
+      allowed_subtypes.fetch(subtype) unless info['allow_unlisted_subtype']
+    rescue KeyError
+      raise "Unknown span.subtype `#{span.type}'\nPossible subtypes: #{allowed_subtypes}"
+    end
   end
 
   module Methods