braintrust 2.2.0 → 2.2.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "braintrust",
-  "version": "2.2.0",
+  "version": "2.2.1",
   "description": "SDK for integrating Braintrust",
   "repository": {
     "type": "git",
@@ -119,6 +119,7 @@
     "@next/env": "^14.2.3",
     "@types/nunjucks": "^3.2.6",
     "@vercel/functions": "^1.0.2",
+    "ajv": "^8.17.1",
     "argparse": "^2.0.1",
     "boxen": "^8.0.1",
     "chalk": "^4.1.2",

package/util/dist/index.d.mts

@@ -14,6 +14,7 @@ type Source = (typeof VALID_SOURCES)[number];
 declare const PARENT_ID_FIELD = "_parent_id";
 declare const ASYNC_SCORING_CONTROL_FIELD = "_async_scoring_control";
 declare const SKIP_ASYNC_SCORING_FIELD = "_skip_async_scoring";
+declare const OBJECT_ID_KEYS: readonly ["experiment_id", "dataset_id", "prompt_session_id", "project_id", "log_id", "function_data"];
 type TransactionId = string;
 
 declare const AsyncScoringControl: z.ZodUnion<[z.ZodObject<{
@@ -5043,13 +5044,13 @@ type MergeRowSkipFieldObj = {
 declare function mergeRowBatch<T extends {
     id: string;
     [IS_MERGE_FIELD]?: boolean | null;
-    [PARENT_ID_FIELD]?: string | null;
-} & MergeRowSkipFieldObj>(rows: T[]): T[][];
-declare function batchItems(args: {
-    items: string[][];
+} & MergeRowSkipFieldObj>(rows: T[]): T[];
+declare function batchItems<T>(args: {
+    items: T[];
     batchMaxNumItems?: number;
     batchMaxNumBytes?: number;
-}): string[][][];
+    getByteSize: (item: T) => number;
+}): T[][];
 
 interface Score {
     name: string;
@@ -5289,7 +5290,7 @@ declare function parseParent(parent: InvokeFunctionType["parent"]): string | und
  */
 declare function makeScorerPropagatedEvent(parent: string | undefined): Record<string, unknown>;
 
-declare const spanTypeAttributeValues: readonly ["llm", "score", "function", "eval", "task", "tool", "automation", "facet", "preprocessor", "classifier"];
+declare const spanTypeAttributeValues: readonly ["llm", "score", "function", "eval", "task", "tool", "automation", "facet", "preprocessor", "classifier", "review"];
 declare enum SpanTypeAttribute {
     LLM = "llm",
     SCORE = "score",
@@ -5300,7 +5301,8 @@ declare enum SpanTypeAttribute {
     AUTOMATION = "automation",
     FACET = "facet",
     PREPROCESSOR = "preprocessor",
-    CLASSIFIER = "classifier"
+    CLASSIFIER = "classifier",
+    REVIEW = "review"
 }
 type SpanType = (typeof spanTypeAttributeValues)[number];
 declare const spanPurposeAttributeValues: readonly ["scorer"];
@@ -5325,4 +5327,4 @@ declare function base64ToUint8Array(base64: string): Uint8Array;
 declare function uint8ArrayToString(uint8Array: Uint8Array): string;
 declare function stringToUint8Array(str: string): Uint8Array;
 
-export { ARRAY_DELETE_FIELD, ASYNC_SCORING_CONTROL_FIELD, AUDIT_METADATA_FIELD, AUDIT_SOURCE_FIELD, type AnyDatasetRecord, BT_CURSOR_HEADER, BT_FOUND_EXISTING_HEADER, BT_IMPERSONATE_USER, BT_PARENT, type BackgroundLogEvent, CREATED_FIELD, type CommentEvent, DEFAULT_IS_LEGACY_DATASET, type DatasetEvent, type DatasetRecord, type ExperimentEvent, type ExperimentLogFullArgs, type ExperimentLogPartialArgs, ExtraFieldsError, ID_FIELD, IS_MERGE_FIELD, type IdField, type InputField, type LogCommentFullArgs, type LogFeedbackFullArgs, type LoggingEvent, MERGE_PATHS_FIELD, OBJECT_DELETE_FIELD, type OtherExperimentLogFields, PARENT_ID_FIELD, type ParentExperimentIds, type ParentPlaygroundLogIds, type ParentProjectLogIds, type PlaygroundLogEvent, SKIP_ASYNC_SCORING_FIELD, type SanitizedExperimentLogPartialArgs, type Score, type Scorer, type ScorerArgs, type Source, SpanComponentsV1, SpanComponentsV2, SpanComponentsV3, type SpanComponentsV3Data, SpanComponentsV4, type SpanComponentsV4Data, SpanObjectTypeV1, SpanObjectTypeV2, SpanObjectTypeV3, type SpanPurpose, SpanRowIdsV1, SpanRowIdsV2, type SpanType, SpanTypeAttribute, TRANSACTION_ID_FIELD, type TransactionId, VALID_SOURCES, _urljoin, base64ToUint8Array, batchItems, camelToSnakeCase, capitalize, concatUint8Arrays, constructJsonArray, deterministicReplacer, ensureDatasetRecord, ensureLegacyDatasetRecord, ensureNewDatasetRecord, forEachMissingKey, getObjValueByPath, getRecordKeys, isArray, isEmpty, isNumber, isObject, isObjectOrArray, loadPrettyXact, lowercase, makeScorerPropagatedEvent, mapAt, mapSetDefault, mapSetNotPresent, mergeDicts, mergeDictsWithPaths, mergeGitMetadataSettings, mergeRowBatch, notEmpty, objectNullish, parseNoStrip, parseParent, prettifyXact, recordAt, recordFind, recordSetDefault, resolveParentHeader, snakeToCamelCase, snakeToTitleCase, spanComponentsV3Schema, spanComponentsV4Schema, spanObjectTypeV3EnumSchema, spanObjectTypeV3ToString, spanObjectTypeV3ToTypedString, spanPurposeAttributeValues, spanTypeAttributeValues, stringToUint8Array, uint8ArrayToBase64, uint8ArrayToString };
+export { ARRAY_DELETE_FIELD, ASYNC_SCORING_CONTROL_FIELD, AUDIT_METADATA_FIELD, AUDIT_SOURCE_FIELD, type AnyDatasetRecord, BT_CURSOR_HEADER, BT_FOUND_EXISTING_HEADER, BT_IMPERSONATE_USER, BT_PARENT, type BackgroundLogEvent, CREATED_FIELD, type CommentEvent, DEFAULT_IS_LEGACY_DATASET, type DatasetEvent, type DatasetRecord, type ExperimentEvent, type ExperimentLogFullArgs, type ExperimentLogPartialArgs, ExtraFieldsError, ID_FIELD, IS_MERGE_FIELD, type IdField, type InputField, type LogCommentFullArgs, type LogFeedbackFullArgs, type LoggingEvent, MERGE_PATHS_FIELD, OBJECT_DELETE_FIELD, OBJECT_ID_KEYS, type OtherExperimentLogFields, PARENT_ID_FIELD, type ParentExperimentIds, type ParentPlaygroundLogIds, type ParentProjectLogIds, type PlaygroundLogEvent, SKIP_ASYNC_SCORING_FIELD, type SanitizedExperimentLogPartialArgs, type Score, type Scorer, type ScorerArgs, type Source, SpanComponentsV1, SpanComponentsV2, SpanComponentsV3, type SpanComponentsV3Data, SpanComponentsV4, type SpanComponentsV4Data, SpanObjectTypeV1, SpanObjectTypeV2, SpanObjectTypeV3, type SpanPurpose, SpanRowIdsV1, SpanRowIdsV2, type SpanType, SpanTypeAttribute, TRANSACTION_ID_FIELD, type TransactionId, VALID_SOURCES, _urljoin, base64ToUint8Array, batchItems, camelToSnakeCase, capitalize, concatUint8Arrays, constructJsonArray, deterministicReplacer, ensureDatasetRecord, ensureLegacyDatasetRecord, ensureNewDatasetRecord, forEachMissingKey, getObjValueByPath, getRecordKeys, isArray, isEmpty, isNumber, isObject, isObjectOrArray, loadPrettyXact, lowercase, makeScorerPropagatedEvent, mapAt, mapSetDefault, mapSetNotPresent, mergeDicts, mergeDictsWithPaths, mergeGitMetadataSettings, mergeRowBatch, notEmpty, objectNullish, parseNoStrip, parseParent, prettifyXact, recordAt, recordFind, recordSetDefault, resolveParentHeader, snakeToCamelCase, snakeToTitleCase, spanComponentsV3Schema, spanComponentsV4Schema, spanObjectTypeV3EnumSchema, spanObjectTypeV3ToString, spanObjectTypeV3ToTypedString, spanPurposeAttributeValues, spanTypeAttributeValues, stringToUint8Array, uint8ArrayToBase64, uint8ArrayToString };

package/util/dist/index.d.ts

@@ -14,6 +14,7 @@ type Source = (typeof VALID_SOURCES)[number];
 declare const PARENT_ID_FIELD = "_parent_id";
 declare const ASYNC_SCORING_CONTROL_FIELD = "_async_scoring_control";
 declare const SKIP_ASYNC_SCORING_FIELD = "_skip_async_scoring";
+declare const OBJECT_ID_KEYS: readonly ["experiment_id", "dataset_id", "prompt_session_id", "project_id", "log_id", "function_data"];
 type TransactionId = string;
 
 declare const AsyncScoringControl: z.ZodUnion<[z.ZodObject<{
@@ -5043,13 +5044,13 @@ type MergeRowSkipFieldObj = {
 declare function mergeRowBatch<T extends {
     id: string;
     [IS_MERGE_FIELD]?: boolean | null;
-    [PARENT_ID_FIELD]?: string | null;
-} & MergeRowSkipFieldObj>(rows: T[]): T[][];
-declare function batchItems(args: {
-    items: string[][];
+} & MergeRowSkipFieldObj>(rows: T[]): T[];
+declare function batchItems<T>(args: {
+    items: T[];
     batchMaxNumItems?: number;
     batchMaxNumBytes?: number;
-}): string[][][];
+    getByteSize: (item: T) => number;
+}): T[][];
 
 interface Score {
     name: string;
@@ -5289,7 +5290,7 @@ declare function parseParent(parent: InvokeFunctionType["parent"]): string | und
  */
 declare function makeScorerPropagatedEvent(parent: string | undefined): Record<string, unknown>;
 
-declare const spanTypeAttributeValues: readonly ["llm", "score", "function", "eval", "task", "tool", "automation", "facet", "preprocessor", "classifier"];
+declare const spanTypeAttributeValues: readonly ["llm", "score", "function", "eval", "task", "tool", "automation", "facet", "preprocessor", "classifier", "review"];
 declare enum SpanTypeAttribute {
     LLM = "llm",
     SCORE = "score",
@@ -5300,7 +5301,8 @@ declare enum SpanTypeAttribute {
     AUTOMATION = "automation",
     FACET = "facet",
     PREPROCESSOR = "preprocessor",
-    CLASSIFIER = "classifier"
+    CLASSIFIER = "classifier",
+    REVIEW = "review"
 }
 type SpanType = (typeof spanTypeAttributeValues)[number];
 declare const spanPurposeAttributeValues: readonly ["scorer"];
@@ -5325,4 +5327,4 @@ declare function base64ToUint8Array(base64: string): Uint8Array;
 declare function uint8ArrayToString(uint8Array: Uint8Array): string;
 declare function stringToUint8Array(str: string): Uint8Array;
 
-export { ARRAY_DELETE_FIELD, ASYNC_SCORING_CONTROL_FIELD, AUDIT_METADATA_FIELD, AUDIT_SOURCE_FIELD, type AnyDatasetRecord, BT_CURSOR_HEADER, BT_FOUND_EXISTING_HEADER, BT_IMPERSONATE_USER, BT_PARENT, type BackgroundLogEvent, CREATED_FIELD, type CommentEvent, DEFAULT_IS_LEGACY_DATASET, type DatasetEvent, type DatasetRecord, type ExperimentEvent, type ExperimentLogFullArgs, type ExperimentLogPartialArgs, ExtraFieldsError, ID_FIELD, IS_MERGE_FIELD, type IdField, type InputField, type LogCommentFullArgs, type LogFeedbackFullArgs, type LoggingEvent, MERGE_PATHS_FIELD, OBJECT_DELETE_FIELD, type OtherExperimentLogFields, PARENT_ID_FIELD, type ParentExperimentIds, type ParentPlaygroundLogIds, type ParentProjectLogIds, type PlaygroundLogEvent, SKIP_ASYNC_SCORING_FIELD, type SanitizedExperimentLogPartialArgs, type Score, type Scorer, type ScorerArgs, type Source, SpanComponentsV1, SpanComponentsV2, SpanComponentsV3, type SpanComponentsV3Data, SpanComponentsV4, type SpanComponentsV4Data, SpanObjectTypeV1, SpanObjectTypeV2, SpanObjectTypeV3, type SpanPurpose, SpanRowIdsV1, SpanRowIdsV2, type SpanType, SpanTypeAttribute, TRANSACTION_ID_FIELD, type TransactionId, VALID_SOURCES, _urljoin, base64ToUint8Array, batchItems, camelToSnakeCase, capitalize, concatUint8Arrays, constructJsonArray, deterministicReplacer, ensureDatasetRecord, ensureLegacyDatasetRecord, ensureNewDatasetRecord, forEachMissingKey, getObjValueByPath, getRecordKeys, isArray, isEmpty, isNumber, isObject, isObjectOrArray, loadPrettyXact, lowercase, makeScorerPropagatedEvent, mapAt, mapSetDefault, mapSetNotPresent, mergeDicts, mergeDictsWithPaths, mergeGitMetadataSettings, mergeRowBatch, notEmpty, objectNullish, parseNoStrip, parseParent, prettifyXact, recordAt, recordFind, recordSetDefault, resolveParentHeader, snakeToCamelCase, snakeToTitleCase, spanComponentsV3Schema, spanComponentsV4Schema, spanObjectTypeV3EnumSchema, spanObjectTypeV3ToString, spanObjectTypeV3ToTypedString, spanPurposeAttributeValues, spanTypeAttributeValues, stringToUint8Array, uint8ArrayToBase64, uint8ArrayToString };
+export { ARRAY_DELETE_FIELD, ASYNC_SCORING_CONTROL_FIELD, AUDIT_METADATA_FIELD, AUDIT_SOURCE_FIELD, type AnyDatasetRecord, BT_CURSOR_HEADER, BT_FOUND_EXISTING_HEADER, BT_IMPERSONATE_USER, BT_PARENT, type BackgroundLogEvent, CREATED_FIELD, type CommentEvent, DEFAULT_IS_LEGACY_DATASET, type DatasetEvent, type DatasetRecord, type ExperimentEvent, type ExperimentLogFullArgs, type ExperimentLogPartialArgs, ExtraFieldsError, ID_FIELD, IS_MERGE_FIELD, type IdField, type InputField, type LogCommentFullArgs, type LogFeedbackFullArgs, type LoggingEvent, MERGE_PATHS_FIELD, OBJECT_DELETE_FIELD, OBJECT_ID_KEYS, type OtherExperimentLogFields, PARENT_ID_FIELD, type ParentExperimentIds, type ParentPlaygroundLogIds, type ParentProjectLogIds, type PlaygroundLogEvent, SKIP_ASYNC_SCORING_FIELD, type SanitizedExperimentLogPartialArgs, type Score, type Scorer, type ScorerArgs, type Source, SpanComponentsV1, SpanComponentsV2, SpanComponentsV3, type SpanComponentsV3Data, SpanComponentsV4, type SpanComponentsV4Data, SpanObjectTypeV1, SpanObjectTypeV2, SpanObjectTypeV3, type SpanPurpose, SpanRowIdsV1, SpanRowIdsV2, type SpanType, SpanTypeAttribute, TRANSACTION_ID_FIELD, type TransactionId, VALID_SOURCES, _urljoin, base64ToUint8Array, batchItems, camelToSnakeCase, capitalize, concatUint8Arrays, constructJsonArray, deterministicReplacer, ensureDatasetRecord, ensureLegacyDatasetRecord, ensureNewDatasetRecord, forEachMissingKey, getObjValueByPath, getRecordKeys, isArray, isEmpty, isNumber, isObject, isObjectOrArray, loadPrettyXact, lowercase, makeScorerPropagatedEvent, mapAt, mapSetDefault, mapSetNotPresent, mergeDicts, mergeDictsWithPaths, mergeGitMetadataSettings, mergeRowBatch, notEmpty, objectNullish, parseNoStrip, parseParent, prettifyXact, recordAt, recordFind, recordSetDefault, resolveParentHeader, snakeToCamelCase, snakeToTitleCase, spanComponentsV3Schema, spanComponentsV4Schema, spanObjectTypeV3EnumSchema, spanObjectTypeV3ToString, spanObjectTypeV3ToTypedString, spanPurposeAttributeValues, spanTypeAttributeValues, stringToUint8Array, uint8ArrayToBase64, uint8ArrayToString };

package/util/dist/index.js

@@ -12,6 +12,14 @@ var VALID_SOURCES = ["app", "api", "external"];
 var PARENT_ID_FIELD = "_parent_id";
 var ASYNC_SCORING_CONTROL_FIELD = "_async_scoring_control";
 var SKIP_ASYNC_SCORING_FIELD = "_skip_async_scoring";
+var OBJECT_ID_KEYS = [
+  "experiment_id",
+  "dataset_id",
+  "prompt_session_id",
+  "project_id",
+  "log_id",
+  "function_data"
+];
 
 // util/span_identifier_v3.ts
 var _uuid = require('uuid'); var uuid3 = _interopRequireWildcard(_uuid); var uuid2 = _interopRequireWildcard(_uuid); var uuid = _interopRequireWildcard(_uuid);
@@ -810,72 +818,8 @@ function getObjValueByPath(row, path) {
   return curr;
 }
 
-// util/graph_util.ts
-function depthFirstSearch(args) {
-  const { graph, firstVisitF, lastVisitF } = args;
-  for (const vs of graph.values()) {
-    for (const v of vs.values()) {
-      if (!graph.has(v)) {
-        throw new Error(`Outgoing vertex ${v} must be a key in the graph`);
-      }
-    }
-  }
-  const firstVisitedVertices = /* @__PURE__ */ new Set();
-  const visitationOrder = _nullishCoalesce(args.visitationOrder, () => ( [...graph.keys()]));
-  const events = visitationOrder.map((vertex) => ({ eventType: "first", vertex, extras: {} })).reverse();
-  while (events.length) {
-    const { eventType, vertex, extras } = events.pop();
-    if (eventType === "last") {
-      _optionalChain([lastVisitF, 'optionalCall', _8 => _8(vertex)]);
-      continue;
-    }
-    if (firstVisitedVertices.has(vertex)) {
-      continue;
-    }
-    firstVisitedVertices.add(vertex);
-    _optionalChain([firstVisitF, 'optionalCall', _9 => _9(vertex, { parentVertex: extras.parentVertex })]);
-    events.push({ eventType: "last", vertex, extras: {} });
-    mapAt(graph, vertex).forEach((child) => {
-      events.push({
-        eventType: "first",
-        vertex: child,
-        extras: { parentVertex: vertex }
-      });
-    });
-  }
-}
-function undirectedConnectedComponents(graph) {
-  const directedGraph = new Map(
-    [...graph.vertices].map((v) => [v, /* @__PURE__ */ new Set()])
-  );
-  for (const [i, j] of graph.edges) {
-    mapAt(directedGraph, i).add(j);
-    mapAt(directedGraph, j).add(i);
-  }
-  let labelCounter = 0;
-  const vertexLabels = /* @__PURE__ */ new Map();
-  const firstVisitF = (vertex, args) => {
-    const label = _optionalChain([args, 'optionalAccess', _10 => _10.parentVertex]) !== void 0 ? mapAt(vertexLabels, _optionalChain([args, 'optionalAccess', _11 => _11.parentVertex])) : labelCounter++;
-    vertexLabels.set(vertex, label);
-  };
-  depthFirstSearch({ graph: directedGraph, firstVisitF });
-  const output = Array.from({ length: labelCounter }).map(() => []);
-  for (const [vertex, label] of vertexLabels.entries()) {
-    output[label].push(vertex);
-  }
-  return output;
-}
-function topologicalSort(graph, visitationOrder) {
-  const reverseOrdering = [];
-  const lastVisitF = (vertex) => {
-    reverseOrdering.push(vertex);
-  };
-  depthFirstSearch({ graph, lastVisitF, visitationOrder });
-  return reverseOrdering.reverse();
-}
-
 // util/merge_row_batch.ts
-function generateMergedRowKey(row, useParentIdForId) {
+function generateMergedRowKey(row) {
   return JSON.stringify(
     [
       "org_id",
@@ -884,7 +828,7 @@ function generateMergedRowKey(row, useParentIdForId) {
       "dataset_id",
       "prompt_session_id",
       "log_id",
-      _nullishCoalesce(useParentIdForId, () => ( false)) ? PARENT_ID_FIELD : "id"
+      "id"
     ].map((k) => row[k])
   );
 }
@@ -938,96 +882,34 @@ function mergeRowBatch(rows) {
       rowGroups.set(key, row);
     }
   }
-  const merged = [...rowGroups.values()];
-  const rowToLabel = new Map(
-    merged.map((r, i) => [generateMergedRowKey(r), i])
-  );
-  const graph = new Map(
-    Array.from({ length: merged.length }).map((_, i) => [i, /* @__PURE__ */ new Set()])
-  );
-  merged.forEach((r, i) => {
-    const parentId = r[PARENT_ID_FIELD];
-    if (!parentId) {
-      return;
-    }
-    const parentRowKey = generateMergedRowKey(
-      r,
-      true
-      /* useParentIdForId */
-    );
-    const parentLabel = rowToLabel.get(parentRowKey);
-    if (parentLabel !== void 0) {
-      mapAt(graph, parentLabel).add(i);
-    }
-  });
-  const connectedComponents = undirectedConnectedComponents({
-    vertices: new Set(graph.keys()),
-    edges: new Set(
-      [...graph.entries()].flatMap(
-        ([k, vs]) => [...vs].map((v) => {
-          const ret = [k, v];
-          return ret;
-        })
-      )
-    )
-  });
-  const buckets = connectedComponents.map(
-    (cc) => topologicalSort(
-      graph,
-      cc
-      /* visitationOrder */
-    )
-  );
-  return buckets.map((bucket) => bucket.map((i) => merged[i]));
+  return [...rowGroups.values()];
 }
 function batchItems(args) {
-  let { items } = args;
+  const { items } = args;
   const batchMaxNumItems = _nullishCoalesce(args.batchMaxNumItems, () => ( Number.POSITIVE_INFINITY));
   const batchMaxNumBytes = _nullishCoalesce(args.batchMaxNumBytes, () => ( Number.POSITIVE_INFINITY));
+  const getByteSize = args.getByteSize;
   const output = [];
-  let nextItems = [];
-  let batchSet = [];
   let batch = [];
   let batchLen = 0;
   function addToBatch(item) {
     batch.push(item);
-    batchLen += item.length;
+    batchLen += getByteSize(item);
   }
   function flushBatch() {
-    batchSet.push(batch);
+    output.push(batch);
     batch = [];
     batchLen = 0;
   }
-  while (items.length) {
-    for (const bucket of items) {
-      let i = 0;
-      for (const item of bucket) {
-        if (batch.length === 0 || item.length + batchLen < batchMaxNumBytes && batch.length < batchMaxNumItems) {
-          addToBatch(item);
-        } else if (i === 0) {
-          flushBatch();
-          addToBatch(item);
-        } else {
-          break;
-        }
-        ++i;
-      }
-      if (i < bucket.length) {
-        nextItems.push(bucket.slice(i));
-      }
-      if (batchLen >= batchMaxNumBytes || batch.length > batchMaxNumItems) {
-        flushBatch();
-      }
-    }
-    if (batch.length) {
+  for (const item of items) {
+    const itemSize = getByteSize(item);
+    if (batch.length > 0 && !(itemSize + batchLen < batchMaxNumBytes && batch.length < batchMaxNumItems)) {
       flushBatch();
     }
-    if (batchSet.length) {
-      output.push(batchSet);
-      batchSet = [];
-    }
-    items = nextItems;
-    nextItems = [];
+    addToBatch(item);
+  }
+  if (batch.length > 0) {
+    flushBatch();
   }
   return output;
 }
@@ -1344,7 +1226,8 @@ var spanTypeAttributeValues = [
   "automation",
   "facet",
   "preprocessor",
-  "classifier"
+  "classifier",
+  "review"
 ];
 var SpanTypeAttribute = /* @__PURE__ */ ((SpanTypeAttribute2) => {
   SpanTypeAttribute2["LLM"] = "llm";
@@ -1357,6 +1240,7 @@ var SpanTypeAttribute = /* @__PURE__ */ ((SpanTypeAttribute2) => {
   SpanTypeAttribute2["FACET"] = "facet";
   SpanTypeAttribute2["PREPROCESSOR"] = "preprocessor";
   SpanTypeAttribute2["CLASSIFIER"] = "classifier";
+  SpanTypeAttribute2["REVIEW"] = "review";
   return SpanTypeAttribute2;
 })(SpanTypeAttribute || {});
 var spanPurposeAttributeValues = ["scorer"];
@@ -1510,4 +1394,5 @@ function objectNullish(object) {
 
 
 
-exports.ARRAY_DELETE_FIELD = ARRAY_DELETE_FIELD; exports.ASYNC_SCORING_CONTROL_FIELD = ASYNC_SCORING_CONTROL_FIELD; exports.AUDIT_METADATA_FIELD = AUDIT_METADATA_FIELD; exports.AUDIT_SOURCE_FIELD = AUDIT_SOURCE_FIELD; exports.BT_CURSOR_HEADER = BT_CURSOR_HEADER; exports.BT_FOUND_EXISTING_HEADER = BT_FOUND_EXISTING_HEADER; exports.BT_IMPERSONATE_USER = BT_IMPERSONATE_USER; exports.BT_PARENT = BT_PARENT; exports.CREATED_FIELD = CREATED_FIELD; exports.DEFAULT_IS_LEGACY_DATASET = DEFAULT_IS_LEGACY_DATASET; exports.ExtraFieldsError = ExtraFieldsError; exports.ID_FIELD = ID_FIELD; exports.IS_MERGE_FIELD = IS_MERGE_FIELD; exports.MERGE_PATHS_FIELD = MERGE_PATHS_FIELD; exports.OBJECT_DELETE_FIELD = OBJECT_DELETE_FIELD; exports.PARENT_ID_FIELD = PARENT_ID_FIELD; exports.SKIP_ASYNC_SCORING_FIELD = SKIP_ASYNC_SCORING_FIELD; exports.SpanComponentsV1 = SpanComponentsV1; exports.SpanComponentsV2 = SpanComponentsV2; exports.SpanComponentsV3 = SpanComponentsV3; exports.SpanComponentsV4 = SpanComponentsV4; exports.SpanObjectTypeV1 = SpanObjectTypeV1; exports.SpanObjectTypeV2 = SpanObjectTypeV2; exports.SpanObjectTypeV3 = SpanObjectTypeV3; exports.SpanRowIdsV1 = SpanRowIdsV1; exports.SpanRowIdsV2 = SpanRowIdsV2; exports.SpanTypeAttribute = SpanTypeAttribute; exports.TRANSACTION_ID_FIELD = TRANSACTION_ID_FIELD; exports.VALID_SOURCES = VALID_SOURCES; exports._urljoin = _urljoin; exports.base64ToUint8Array = base64ToUint8Array; exports.batchItems = batchItems; exports.camelToSnakeCase = camelToSnakeCase; exports.capitalize = capitalize; exports.concatUint8Arrays = concatUint8Arrays; exports.constructJsonArray = constructJsonArray; exports.deterministicReplacer = deterministicReplacer; exports.ensureDatasetRecord = ensureDatasetRecord; exports.ensureLegacyDatasetRecord = ensureLegacyDatasetRecord; exports.ensureNewDatasetRecord = ensureNewDatasetRecord; exports.forEachMissingKey = forEachMissingKey; exports.getObjValueByPath = getObjValueByPath; exports.getRecordKeys = getRecordKeys; exports.isArray = isArray; exports.isEmpty = isEmpty; exports.isNumber = isNumber; exports.isObject = isObject; exports.isObjectOrArray = isObjectOrArray; exports.loadPrettyXact = loadPrettyXact; exports.lowercase = lowercase; exports.makeScorerPropagatedEvent = makeScorerPropagatedEvent; exports.mapAt = mapAt; exports.mapSetDefault = mapSetDefault; exports.mapSetNotPresent = mapSetNotPresent; exports.mergeDicts = mergeDicts; exports.mergeDictsWithPaths = mergeDictsWithPaths; exports.mergeGitMetadataSettings = mergeGitMetadataSettings; exports.mergeRowBatch = mergeRowBatch; exports.notEmpty = notEmpty; exports.objectNullish = objectNullish; exports.parseNoStrip = parseNoStrip; exports.parseParent = parseParent; exports.prettifyXact = prettifyXact; exports.recordAt = recordAt; exports.recordFind = recordFind; exports.recordSetDefault = recordSetDefault; exports.resolveParentHeader = resolveParentHeader; exports.snakeToCamelCase = snakeToCamelCase; exports.snakeToTitleCase = snakeToTitleCase; exports.spanComponentsV3Schema = spanComponentsV3Schema; exports.spanComponentsV4Schema = spanComponentsV4Schema; exports.spanObjectTypeV3EnumSchema = spanObjectTypeV3EnumSchema; exports.spanObjectTypeV3ToString = spanObjectTypeV3ToString; exports.spanObjectTypeV3ToTypedString = spanObjectTypeV3ToTypedString; exports.spanPurposeAttributeValues = spanPurposeAttributeValues; exports.spanTypeAttributeValues = spanTypeAttributeValues; exports.stringToUint8Array = stringToUint8Array; exports.uint8ArrayToBase64 = uint8ArrayToBase64; exports.uint8ArrayToString = uint8ArrayToString;
+
+exports.ARRAY_DELETE_FIELD = ARRAY_DELETE_FIELD; exports.ASYNC_SCORING_CONTROL_FIELD = ASYNC_SCORING_CONTROL_FIELD; exports.AUDIT_METADATA_FIELD = AUDIT_METADATA_FIELD; exports.AUDIT_SOURCE_FIELD = AUDIT_SOURCE_FIELD; exports.BT_CURSOR_HEADER = BT_CURSOR_HEADER; exports.BT_FOUND_EXISTING_HEADER = BT_FOUND_EXISTING_HEADER; exports.BT_IMPERSONATE_USER = BT_IMPERSONATE_USER; exports.BT_PARENT = BT_PARENT; exports.CREATED_FIELD = CREATED_FIELD; exports.DEFAULT_IS_LEGACY_DATASET = DEFAULT_IS_LEGACY_DATASET; exports.ExtraFieldsError = ExtraFieldsError; exports.ID_FIELD = ID_FIELD; exports.IS_MERGE_FIELD = IS_MERGE_FIELD; exports.MERGE_PATHS_FIELD = MERGE_PATHS_FIELD; exports.OBJECT_DELETE_FIELD = OBJECT_DELETE_FIELD; exports.OBJECT_ID_KEYS = OBJECT_ID_KEYS; exports.PARENT_ID_FIELD = PARENT_ID_FIELD; exports.SKIP_ASYNC_SCORING_FIELD = SKIP_ASYNC_SCORING_FIELD; exports.SpanComponentsV1 = SpanComponentsV1; exports.SpanComponentsV2 = SpanComponentsV2; exports.SpanComponentsV3 = SpanComponentsV3; exports.SpanComponentsV4 = SpanComponentsV4; exports.SpanObjectTypeV1 = SpanObjectTypeV1; exports.SpanObjectTypeV2 = SpanObjectTypeV2; exports.SpanObjectTypeV3 = SpanObjectTypeV3; exports.SpanRowIdsV1 = SpanRowIdsV1; exports.SpanRowIdsV2 = SpanRowIdsV2; exports.SpanTypeAttribute = SpanTypeAttribute; exports.TRANSACTION_ID_FIELD = TRANSACTION_ID_FIELD; exports.VALID_SOURCES = VALID_SOURCES; exports._urljoin = _urljoin; exports.base64ToUint8Array = base64ToUint8Array; exports.batchItems = batchItems; exports.camelToSnakeCase = camelToSnakeCase; exports.capitalize = capitalize; exports.concatUint8Arrays = concatUint8Arrays; exports.constructJsonArray = constructJsonArray; exports.deterministicReplacer = deterministicReplacer; exports.ensureDatasetRecord = ensureDatasetRecord; exports.ensureLegacyDatasetRecord = ensureLegacyDatasetRecord; exports.ensureNewDatasetRecord = ensureNewDatasetRecord; exports.forEachMissingKey = forEachMissingKey; exports.getObjValueByPath = getObjValueByPath; exports.getRecordKeys = getRecordKeys; exports.isArray = isArray; exports.isEmpty = isEmpty; exports.isNumber = isNumber; exports.isObject = isObject; exports.isObjectOrArray = isObjectOrArray; exports.loadPrettyXact = loadPrettyXact; exports.lowercase = lowercase; exports.makeScorerPropagatedEvent = makeScorerPropagatedEvent; exports.mapAt = mapAt; exports.mapSetDefault = mapSetDefault; exports.mapSetNotPresent = mapSetNotPresent; exports.mergeDicts = mergeDicts; exports.mergeDictsWithPaths = mergeDictsWithPaths; exports.mergeGitMetadataSettings = mergeGitMetadataSettings; exports.mergeRowBatch = mergeRowBatch; exports.notEmpty = notEmpty; exports.objectNullish = objectNullish; exports.parseNoStrip = parseNoStrip; exports.parseParent = parseParent; exports.prettifyXact = prettifyXact; exports.recordAt = recordAt; exports.recordFind = recordFind; exports.recordSetDefault = recordSetDefault; exports.resolveParentHeader = resolveParentHeader; exports.snakeToCamelCase = snakeToCamelCase; exports.snakeToTitleCase = snakeToTitleCase; exports.spanComponentsV3Schema = spanComponentsV3Schema; exports.spanComponentsV4Schema = spanComponentsV4Schema; exports.spanObjectTypeV3EnumSchema = spanObjectTypeV3EnumSchema; exports.spanObjectTypeV3ToString = spanObjectTypeV3ToString; exports.spanObjectTypeV3ToTypedString = spanObjectTypeV3ToTypedString; exports.spanPurposeAttributeValues = spanPurposeAttributeValues; exports.spanTypeAttributeValues = spanTypeAttributeValues; exports.stringToUint8Array = stringToUint8Array; exports.uint8ArrayToBase64 = uint8ArrayToBase64; exports.uint8ArrayToString = uint8ArrayToString;

package/util/dist/index.mjs

@@ -12,6 +12,14 @@ var VALID_SOURCES = ["app", "api", "external"];
 var PARENT_ID_FIELD = "_parent_id";
 var ASYNC_SCORING_CONTROL_FIELD = "_async_scoring_control";
 var SKIP_ASYNC_SCORING_FIELD = "_skip_async_scoring";
+var OBJECT_ID_KEYS = [
+  "experiment_id",
+  "dataset_id",
+  "prompt_session_id",
+  "project_id",
+  "log_id",
+  "function_data"
+];
 
 // util/span_identifier_v3.ts
 import * as uuid3 from "uuid";
@@ -810,72 +818,8 @@ function getObjValueByPath(row, path) {
   return curr;
 }
 
-// util/graph_util.ts
-function depthFirstSearch(args) {
-  const { graph, firstVisitF, lastVisitF } = args;
-  for (const vs of graph.values()) {
-    for (const v of vs.values()) {
-      if (!graph.has(v)) {
-        throw new Error(`Outgoing vertex ${v} must be a key in the graph`);
-      }
-    }
-  }
-  const firstVisitedVertices = /* @__PURE__ */ new Set();
-  const visitationOrder = args.visitationOrder ?? [...graph.keys()];
-  const events = visitationOrder.map((vertex) => ({ eventType: "first", vertex, extras: {} })).reverse();
-  while (events.length) {
-    const { eventType, vertex, extras } = events.pop();
-    if (eventType === "last") {
-      lastVisitF?.(vertex);
-      continue;
-    }
-    if (firstVisitedVertices.has(vertex)) {
-      continue;
-    }
-    firstVisitedVertices.add(vertex);
-    firstVisitF?.(vertex, { parentVertex: extras.parentVertex });
-    events.push({ eventType: "last", vertex, extras: {} });
-    mapAt(graph, vertex).forEach((child) => {
-      events.push({
-        eventType: "first",
-        vertex: child,
-        extras: { parentVertex: vertex }
-      });
-    });
-  }
-}
-function undirectedConnectedComponents(graph) {
-  const directedGraph = new Map(
-    [...graph.vertices].map((v) => [v, /* @__PURE__ */ new Set()])
-  );
-  for (const [i, j] of graph.edges) {
-    mapAt(directedGraph, i).add(j);
-    mapAt(directedGraph, j).add(i);
-  }
-  let labelCounter = 0;
-  const vertexLabels = /* @__PURE__ */ new Map();
-  const firstVisitF = (vertex, args) => {
-    const label = args?.parentVertex !== void 0 ? mapAt(vertexLabels, args?.parentVertex) : labelCounter++;
-    vertexLabels.set(vertex, label);
-  };
-  depthFirstSearch({ graph: directedGraph, firstVisitF });
-  const output = Array.from({ length: labelCounter }).map(() => []);
-  for (const [vertex, label] of vertexLabels.entries()) {
-    output[label].push(vertex);
-  }
-  return output;
-}
-function topologicalSort(graph, visitationOrder) {
-  const reverseOrdering = [];
-  const lastVisitF = (vertex) => {
-    reverseOrdering.push(vertex);
-  };
-  depthFirstSearch({ graph, lastVisitF, visitationOrder });
-  return reverseOrdering.reverse();
-}
-
 // util/merge_row_batch.ts
-function generateMergedRowKey(row, useParentIdForId) {
+function generateMergedRowKey(row) {
   return JSON.stringify(
     [
       "org_id",
@@ -884,7 +828,7 @@ function generateMergedRowKey(row, useParentIdForId) {
       "dataset_id",
       "prompt_session_id",
       "log_id",
-      useParentIdForId ?? false ? PARENT_ID_FIELD : "id"
+      "id"
     ].map((k) => row[k])
   );
 }
@@ -938,96 +882,34 @@ function mergeRowBatch(rows) {
       rowGroups.set(key, row);
     }
   }
-  const merged = [...rowGroups.values()];
-  const rowToLabel = new Map(
-    merged.map((r, i) => [generateMergedRowKey(r), i])
-  );
-  const graph = new Map(
-    Array.from({ length: merged.length }).map((_, i) => [i, /* @__PURE__ */ new Set()])
-  );
-  merged.forEach((r, i) => {
-    const parentId = r[PARENT_ID_FIELD];
-    if (!parentId) {
-      return;
-    }
-    const parentRowKey = generateMergedRowKey(
-      r,
-      true
-      /* useParentIdForId */
-    );
-    const parentLabel = rowToLabel.get(parentRowKey);
-    if (parentLabel !== void 0) {
-      mapAt(graph, parentLabel).add(i);
-    }
-  });
-  const connectedComponents = undirectedConnectedComponents({
-    vertices: new Set(graph.keys()),
-    edges: new Set(
-      [...graph.entries()].flatMap(
-        ([k, vs]) => [...vs].map((v) => {
-          const ret = [k, v];
-          return ret;
-        })
-      )
-    )
-  });
-  const buckets = connectedComponents.map(
-    (cc) => topologicalSort(
-      graph,
-      cc
-      /* visitationOrder */
-    )
-  );
-  return buckets.map((bucket) => bucket.map((i) => merged[i]));
+  return [...rowGroups.values()];
 }
 function batchItems(args) {
-  let { items } = args;
+  const { items } = args;
   const batchMaxNumItems = args.batchMaxNumItems ?? Number.POSITIVE_INFINITY;
   const batchMaxNumBytes = args.batchMaxNumBytes ?? Number.POSITIVE_INFINITY;
+  const getByteSize = args.getByteSize;
   const output = [];
-  let nextItems = [];
-  let batchSet = [];
   let batch = [];
   let batchLen = 0;
   function addToBatch(item) {
     batch.push(item);
-    batchLen += item.length;
+    batchLen += getByteSize(item);
   }
   function flushBatch() {
-    batchSet.push(batch);
+    output.push(batch);
     batch = [];
     batchLen = 0;
   }
-  while (items.length) {
-    for (const bucket of items) {
-      let i = 0;
-      for (const item of bucket) {
-        if (batch.length === 0 || item.length + batchLen < batchMaxNumBytes && batch.length < batchMaxNumItems) {
-          addToBatch(item);
-        } else if (i === 0) {
-          flushBatch();
-          addToBatch(item);
-        } else {
-          break;
-        }
-        ++i;
-      }
-      if (i < bucket.length) {
-        nextItems.push(bucket.slice(i));
-      }
-      if (batchLen >= batchMaxNumBytes || batch.length > batchMaxNumItems) {
-        flushBatch();
-      }
-    }
-    if (batch.length) {
+  for (const item of items) {
+    const itemSize = getByteSize(item);
+    if (batch.length > 0 && !(itemSize + batchLen < batchMaxNumBytes && batch.length < batchMaxNumItems)) {
       flushBatch();
     }
-    if (batchSet.length) {
-      output.push(batchSet);
-      batchSet = [];
-    }
-    items = nextItems;
-    nextItems = [];
+    addToBatch(item);
+  }
+  if (batch.length > 0) {
+    flushBatch();
   }
   return output;
 }
@@ -1344,7 +1226,8 @@ var spanTypeAttributeValues = [
   "automation",
   "facet",
   "preprocessor",
-  "classifier"
+  "classifier",
+  "review"
 ];
 var SpanTypeAttribute = /* @__PURE__ */ ((SpanTypeAttribute2) => {
   SpanTypeAttribute2["LLM"] = "llm";
@@ -1357,6 +1240,7 @@ var SpanTypeAttribute = /* @__PURE__ */ ((SpanTypeAttribute2) => {
   SpanTypeAttribute2["FACET"] = "facet";
   SpanTypeAttribute2["PREPROCESSOR"] = "preprocessor";
   SpanTypeAttribute2["CLASSIFIER"] = "classifier";
+  SpanTypeAttribute2["REVIEW"] = "review";
   return SpanTypeAttribute2;
 })(SpanTypeAttribute || {});
 var spanPurposeAttributeValues = ["scorer"];
@@ -1446,6 +1330,7 @@ export {
   IS_MERGE_FIELD,
   MERGE_PATHS_FIELD,
   OBJECT_DELETE_FIELD,
+  OBJECT_ID_KEYS,
   PARENT_ID_FIELD,
   SKIP_ASYNC_SCORING_FIELD,
   SpanComponentsV1,