@tungthedev/streams-server 0.2.0

package/src/profiles/otelTraces/schema.ts

@@ -0,0 +1,408 @@
+import {
+  SCHEMA_REGISTRY_API_VERSION,
+  type SchemaRegistry,
+  type SearchConfig,
+  type SearchFieldConfig,
+} from "../../schema/registry";
+
+const NULLABLE_STRING = { type: ["string", "null"] } as const;
+const NULLABLE_NUMBER = { type: ["number", "null"] } as const;
+const ATTRIBUTES_SCHEMA = { type: "object", additionalProperties: true } as const;
+
+export const OTEL_TRACES_CANONICAL_SCHEMA = {
+  type: "object",
+  additionalProperties: false,
+  properties: {
+    schemaVersion: { type: "integer", enum: [1] },
+    signal: { type: "string", enum: ["trace.span"] },
+    timestamp: { type: "string" },
+    endTimestamp: NULLABLE_STRING,
+    startUnixNano: NULLABLE_STRING,
+    endUnixNano: NULLABLE_STRING,
+    duration: NULLABLE_NUMBER,
+    traceId: { type: "string" },
+    spanId: { type: "string" },
+    parentSpanId: NULLABLE_STRING,
+    traceState: NULLABLE_STRING,
+    traceFlags: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        sampled: { type: "boolean" },
+        raw: { type: ["integer", "null"] },
+      },
+      required: ["sampled", "raw"],
+    },
+    name: { type: "string" },
+    kind: { type: "string", enum: ["unspecified", "internal", "server", "client", "producer", "consumer"] },
+    status: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        code: { type: "string", enum: ["unset", "ok", "error"] },
+        message: NULLABLE_STRING,
+      },
+      required: ["code", "message"],
+    },
+    service: NULLABLE_STRING,
+    serviceNamespace: NULLABLE_STRING,
+    serviceInstanceId: NULLABLE_STRING,
+    environment: NULLABLE_STRING,
+    version: NULLABLE_STRING,
+    region: NULLABLE_STRING,
+    requestId: NULLABLE_STRING,
+    http: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        method: NULLABLE_STRING,
+        route: NULLABLE_STRING,
+        path: NULLABLE_STRING,
+        target: NULLABLE_STRING,
+        url: NULLABLE_STRING,
+        statusCode: { type: ["integer", "null"] },
+        userAgent: NULLABLE_STRING,
+      },
+      required: ["method", "route", "path", "target", "url", "statusCode", "userAgent"],
+    },
+    db: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        system: NULLABLE_STRING,
+        name: NULLABLE_STRING,
+        operation: NULLABLE_STRING,
+        statement: NULLABLE_STRING,
+      },
+      required: ["system", "name", "operation", "statement"],
+    },
+    rpc: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        system: NULLABLE_STRING,
+        service: NULLABLE_STRING,
+        method: NULLABLE_STRING,
+      },
+      required: ["system", "service", "method"],
+    },
+    messaging: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        system: NULLABLE_STRING,
+        destination: NULLABLE_STRING,
+        operation: NULLABLE_STRING,
+      },
+      required: ["system", "destination", "operation"],
+    },
+    error: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        isError: { type: "boolean" },
+        type: NULLABLE_STRING,
+        message: NULLABLE_STRING,
+        stacktrace: NULLABLE_STRING,
+      },
+      required: ["isError", "type", "message", "stacktrace"],
+    },
+    instrumentationScope: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        name: NULLABLE_STRING,
+        version: NULLABLE_STRING,
+        schemaUrl: NULLABLE_STRING,
+        attributes: ATTRIBUTES_SCHEMA,
+      },
+      required: ["name", "version", "schemaUrl", "attributes"],
+    },
+    resource: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        schemaUrl: NULLABLE_STRING,
+        attributes: ATTRIBUTES_SCHEMA,
+      },
+      required: ["schemaUrl", "attributes"],
+    },
+    attributes: ATTRIBUTES_SCHEMA,
+    events: {
+      type: "array",
+      items: {
+        type: "object",
+        additionalProperties: false,
+        properties: {
+          timestamp: NULLABLE_STRING,
+          timeUnixNano: NULLABLE_STRING,
+          name: { type: "string" },
+          attributes: ATTRIBUTES_SCHEMA,
+          droppedAttributesCount: { type: "integer" },
+        },
+        required: ["timestamp", "timeUnixNano", "name", "attributes"],
+      },
+    },
+    eventNames: {
+      type: "array",
+      items: { type: "string" },
+    },
+    links: {
+      type: "array",
+      items: {
+        type: "object",
+        additionalProperties: false,
+        properties: {
+          traceId: { type: "string" },
+          spanId: { type: "string" },
+          traceState: NULLABLE_STRING,
+          attributes: ATTRIBUTES_SCHEMA,
+          droppedAttributesCount: { type: "integer" },
+        },
+        required: ["traceId", "spanId", "traceState", "attributes"],
+      },
+    },
+    dropped: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        attributes: { type: "integer" },
+        events: { type: "integer" },
+        links: { type: "integer" },
+      },
+      required: ["attributes", "events", "links"],
+    },
+    redaction: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        keys: {
+          type: "array",
+          items: { type: "string" },
+        },
+      },
+      required: ["keys"],
+    },
+    identity: {
+      type: "object",
+      additionalProperties: false,
+      properties: {
+        spanKey: { type: "string" },
+        dedupeKey: { type: "string" },
+      },
+      required: ["spanKey", "dedupeKey"],
+    },
+  },
+  required: [
+    "schemaVersion",
+    "signal",
+    "timestamp",
+    "endTimestamp",
+    "startUnixNano",
+    "endUnixNano",
+    "duration",
+    "traceId",
+    "spanId",
+    "parentSpanId",
+    "traceState",
+    "traceFlags",
+    "name",
+    "kind",
+    "status",
+    "service",
+    "serviceNamespace",
+    "serviceInstanceId",
+    "environment",
+    "version",
+    "region",
+    "requestId",
+    "http",
+    "db",
+    "rpc",
+    "messaging",
+    "error",
+    "instrumentationScope",
+    "resource",
+    "attributes",
+    "events",
+    "eventNames",
+    "links",
+    "dropped",
+    "redaction",
+    "identity",
+  ],
+} as const;
+
+const exactKeyword = (jsonPointer: string, aggregatable = false): SearchFieldConfig => {
+  const field: SearchFieldConfig = {
+    kind: "keyword",
+    bindings: [{ version: 1, jsonPointer }],
+    exact: true,
+    prefix: true,
+    exists: true,
+    sortable: true,
+  };
+  if (aggregatable) field.aggregatable = true;
+  return field;
+};
+
+const lowercaseKeyword = (jsonPointer: string, aggregatable = false): SearchFieldConfig => ({
+  ...exactKeyword(jsonPointer, aggregatable),
+  normalizer: "lowercase_v1",
+});
+
+const textField = (jsonPointer: string): SearchFieldConfig => ({
+  kind: "text",
+  bindings: [{ version: 1, jsonPointer }],
+  analyzer: "unicode_word_v1",
+  exists: true,
+  positions: true,
+});
+
+export const OTEL_TRACES_DEFAULT_SEARCH_CONFIG: SearchConfig = {
+  profile: "otel-traces",
+  primaryTimestampField: "timestamp",
+  aliases: {
+    db: "db.system",
+    duration_ms: "duration",
+    error: "error.isError",
+    method: "http.method",
+    op: "name",
+    parent: "parentSpanId",
+    req: "requestId",
+    route: "http.route",
+    span: "spanId",
+    status: "http.statusCode",
+    svc: "service",
+    time: "timestamp",
+    trace: "traceId",
+    ts: "timestamp",
+  },
+  defaultFields: [
+    { field: "name", boost: 2 },
+    { field: "error.message", boost: 2 },
+    { field: "status.message", boost: 1.5 },
+    { field: "events.name", boost: 1.2 },
+    { field: "db.statement", boost: 0.5 },
+  ],
+  fields: {
+    timestamp: {
+      kind: "date",
+      bindings: [{ version: 1, jsonPointer: "/timestamp" }],
+      exact: true,
+      column: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    endTimestamp: {
+      kind: "date",
+      bindings: [{ version: 1, jsonPointer: "/endTimestamp" }],
+      exact: true,
+      column: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    duration: {
+      kind: "float",
+      bindings: [{ version: 1, jsonPointer: "/duration" }],
+      exact: true,
+      column: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    traceId: exactKeyword("/traceId"),
+    spanId: exactKeyword("/spanId"),
+    parentSpanId: exactKeyword("/parentSpanId"),
+    requestId: exactKeyword("/requestId"),
+    service: lowercaseKeyword("/service", true),
+    serviceNamespace: lowercaseKeyword("/serviceNamespace", true),
+    serviceInstanceId: exactKeyword("/serviceInstanceId"),
+    environment: lowercaseKeyword("/environment", true),
+    version: exactKeyword("/version"),
+    region: lowercaseKeyword("/region", true),
+    name: {
+      kind: "keyword",
+      bindings: [{ version: 1, jsonPointer: "/name" }],
+      exact: true,
+      prefix: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    kind: lowercaseKeyword("/kind", true),
+    "status.code": lowercaseKeyword("/status/code", true),
+    "status.message": textField("/status/message"),
+    "error.isError": {
+      kind: "bool",
+      bindings: [{ version: 1, jsonPointer: "/error/isError" }],
+      exact: true,
+      column: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    "error.type": exactKeyword("/error/type", true),
+    "error.message": textField("/error/message"),
+    "error.stacktrace": textField("/error/stacktrace"),
+    "http.method": lowercaseKeyword("/http/method", true),
+    "http.route": exactKeyword("/http/route", true),
+    "http.path": exactKeyword("/http/path"),
+    "http.statusCode": {
+      kind: "integer",
+      bindings: [{ version: 1, jsonPointer: "/http/statusCode" }],
+      exact: true,
+      column: true,
+      exists: true,
+      sortable: true,
+      aggregatable: true,
+    },
+    "db.system": lowercaseKeyword("/db/system", true),
+    "db.operation": lowercaseKeyword("/db/operation", true),
+    "db.statement": textField("/db/statement"),
+    "rpc.system": lowercaseKeyword("/rpc/system", true),
+    "rpc.service": exactKeyword("/rpc/service", true),
+    "rpc.method": exactKeyword("/rpc/method", true),
+    "messaging.system": lowercaseKeyword("/messaging/system", true),
+    "messaging.destination": exactKeyword("/messaging/destination", true),
+    "messaging.operation": lowercaseKeyword("/messaging/operation", true),
+    "events.name": textField("/eventNames"),
+  },
+  rollups: {
+    spans: {
+      dimensions: ["service", "kind", "status.code"],
+      intervals: ["1m", "5m", "1h"],
+      measures: {
+        spans: { kind: "count" },
+        errors: { kind: "count", include: "error:true" },
+        latency: { kind: "summary", field: "duration", histogram: "log2_v1" },
+      },
+    },
+    http_server: {
+      include: "kind:server",
+      dimensions: ["service", "http.method", "http.route", "http.statusCode"],
+      intervals: ["1m", "5m", "1h"],
+      measures: {
+        requests: { kind: "count" },
+        errors: { kind: "count", include: "error:true" },
+        latency: { kind: "summary", field: "duration", histogram: "log2_v1" },
+      },
+    },
+  },
+};
+
+export function buildOtelTracesDefaultRegistry(stream: string): SchemaRegistry {
+  return {
+    apiVersion: SCHEMA_REGISTRY_API_VERSION,
+    schema: stream,
+    currentVersion: 1,
+    search: structuredClone(OTEL_TRACES_DEFAULT_SEARCH_CONFIG),
+    boundaries: [{ offset: 0, version: 1 }],
+    schemas: {
+      "1": structuredClone(OTEL_TRACES_CANONICAL_SCHEMA),
+    },
+    lenses: {},
+  };
+}