@badger_badger/badger_hq 1.6.0 → 1.8.0

package/dist/index.js

@@ -193,6 +193,50 @@ var COLLECTIONS = {
   captures: "captures",
   llmCalls: "llmCalls"
 };
+var HOUR_MS = 36e5;
+var VISIT_INTERVAL_MIN_MS = 1 * HOUR_MS;
+var VISIT_INTERVAL_MAX_MS = 72 * HOUR_MS;
+var URGENCY_FACTOR = {
+  emergency: 0.5,
+  missionPriority: { critical: 0.5, high: 0.75, medium: 1, low: 1.5 }
+};
+var OverdueLevelSchema = z.enum(["none", "late", "missed"]);
+function knownInterval(source) {
+  if (source.visitIntervalMs) return { ms: source.visitIntervalMs, urgencyApplied: false };
+  if (source.lastCollectedAt && source.nextVisitAt) {
+    const derived = source.nextVisitAt.getTime() - source.lastCollectedAt.getTime();
+    if (derived > 0) return { ms: derived, urgencyApplied: true };
+  }
+  return void 0;
+}
+function baseVisitInterval(source) {
+  return knownInterval(source)?.ms;
+}
+function urgencyFactor(urgency) {
+  const priority = urgency.missionPriority ? URGENCY_FACTOR.missionPriority[urgency.missionPriority] : 1;
+  return (urgency.emergency ? URGENCY_FACTOR.emergency : 1) * priority;
+}
+function clampInterval(ms) {
+  return Math.min(Math.max(ms, VISIT_INTERVAL_MIN_MS), VISIT_INTERVAL_MAX_MS);
+}
+function effectiveInterval(source, urgency) {
+  const known = knownInterval(source);
+  if (!known) return void 0;
+  return clampInterval(known.urgencyApplied ? known.ms : known.ms * urgencyFactor(urgency));
+}
+function effectiveDueAt(source, urgency, now = /* @__PURE__ */ new Date()) {
+  if (source.status === "access_problem") return void 0;
+  if (!source.lastCollectedAt) return now;
+  const interval = effectiveInterval(source, urgency);
+  if (interval === void 0) return source.nextVisitAt ?? now;
+  return new Date(source.lastCollectedAt.getTime() + interval);
+}
+function overdueLevel(dueAt, intervalMs, now = /* @__PURE__ */ new Date()) {
+  if (!dueAt) return "none";
+  const lateMs = now.getTime() - dueAt.getTime();
+  if (lateMs <= 0) return "none";
+  return intervalMs !== void 0 && lateMs > intervalMs ? "missed" : "late";
+}
 var DatabaseName = z.string().min(1).max(63).regex(/^[^/\\. "$*<>:|?]+$/, "Invalid MongoDB database name");
 var ConnectionString = z.string().min(1).regex(/^mongodb(\+srv)?:\/\/\S+$/, "Expected a mongodb:// or mongodb+srv:// URI");
 var DatabaseSchema = z.object({
@@ -232,6 +276,7 @@ var TaskCriteriaSchema = z.object({
   emergency: z.boolean().default(false)
 });
 var MAX_CAPTURE_WINDOW_DAYS = 90;
+var MONITOR_TASK_MAX_SOURCES = 20;
 var DAY_MS = 24 * 60 * 60 * 1e3;
 function isHttpUrl(value) {
   try {
@@ -536,6 +581,21 @@ var DiscoveredBySchema = z.object({
   userId: ClerkUserId.optional(),
   importId: ObjectIdString.optional()
 });
+var SourceDiscoverySchema = z.object({
+  /** The Discovery task whose capture found it. */
+  taskId: ObjectIdString,
+  /** agent-service capture id, when the source came from a task-tagged capture. */
+  captureId: ObjectIdString.optional(),
+  /** The page the person captured (e.g. a search or hashtag page), not the source itself. */
+  capturedUrl: z.string().url().refine(isHttpUrl, "Expected an http(s) URL").optional(),
+  name: z.string().min(1).max(200),
+  description: z.string().max(1e3).optional(),
+  /** The judge's reason for keeping it. */
+  reason: z.string().max(300),
+  /** 0..1 relevance from the `judge` capability. */
+  relevanceScore: z.number().min(0).max(1),
+  discoveredAt: z.coerce.date()
+});
 var SourceSchema = z.object({
   _id: ObjectIdString,
   orgId: ClerkOrgId,
@@ -553,8 +613,11 @@ var SourceSchema = z.object({
   tags: z.array(z.string().min(1)).default([]),
   status: SourceStatus.default("active"),
   discoveredBy: DiscoveredBySchema,
+  discovery: SourceDiscoverySchema.optional(),
   lastCollectedAt: z.coerce.date().optional(),
-  nextVisitAt: z.coerce.date().optional()
+  nextVisitAt: z.coerce.date().optional(),
+  /** Activity-based wait set by the last visit, before current urgency (see `effectiveDueAt`). */
+  visitIntervalMs: z.number().int().positive().optional()
 }).merge(TimestampsSchema).merge(SoftDeleteSchema);
 var SourceCreateSchema = SourceSchema.omit({
   _id: true,
@@ -564,7 +627,9 @@ var SourceCreateSchema = SourceSchema.omit({
   updatedAt: true,
   deletedAt: true,
   discoveredBy: true,
-  lastCollectedAt: true
+  discovery: true,
+  lastCollectedAt: true,
+  visitIntervalMs: true
 });
 var SourceUpdateSchema = SourceCreateSchema.partial();
 var MAX_SOURCE_IMPORT_ROW_ERRORS = 500;
@@ -996,6 +1061,16 @@ var AgentTaskCriteriaSchema = z.object({
   emergency: z.boolean()
 });
 var AgentTaskTargetSchema = TaskTargetSchema.omit({ sourceId: true });
+var AgentTaskSourceSchema = z.object({
+  platform: Platform,
+  url: z.string().url().refine(isHttpUrl, "Expected an http(s) URL"),
+  label: z.string().min(1).optional(),
+  lastCapturedAt: z.coerce.date().optional(),
+  /** When to capture it again, under current urgency (absent: parked on an access problem). */
+  dueAt: z.coerce.date().optional(),
+  /** Parked: a person couldn't get in; an analyst reviews it. */
+  accessProblem: z.boolean().optional()
+});
 var AgentTaskSchema = z.object({
   id: ObjectIdString,
   title: z.string(),
@@ -1005,6 +1080,8 @@ var AgentTaskSchema = z.object({
   searchQueries: z.array(z.string()),
   criteria: AgentTaskCriteriaSchema,
   target: AgentTaskTargetSchema.optional(),
+  /** Monitor: the attached sources other than `target`, sorted by platform then URL. */
+  sources: z.array(AgentTaskSourceSchema).default([]),
   captureWindow: CaptureWindowSchema.optional(),
   status: TaskStatus,
   version: z.number().int().min(1),
@@ -1026,7 +1103,9 @@ var AgentAssignmentsResponseSchema = z.object({
 });
 var AgentAssignmentVersionSchema = z.object({
   assignmentId: ObjectIdString,
-  taskVersion: z.number().int().min(1)
+  taskVersion: z.number().int().min(1),
+  /** Monitor: the latest change to any attached source (a visit moves it; the task version doesn't). */
+  sourcesUpdatedAt: z.coerce.date().optional()
 });
 var AgentAssignmentVersionsResponseSchema = z.object({
   items: z.array(AgentAssignmentVersionSchema)
@@ -1143,6 +1222,10 @@ var SourceIntakeBodySchema = z.object({
   /** The discovery task the session works; its mission's monitor task receives gated sources. */
   taskId: ObjectIdString,
   discoveredByUserId: ClerkUserId,
+  /** agent-service capture the candidates came from (task-tagged captures). */
+  captureId: ObjectIdString.optional(),
+  /** The page that capture was taken on. */
+  capturedUrl: z.string().url().refine(isHttpUrl, "Expected an http(s) URL").optional(),
   candidates: z.array(SourceIntakeCandidateSchema).min(1).max(MAX_SOURCE_INTAKE_CANDIDATES)
 });
 var SourceIntakeResultSchema = z.object({
@@ -1187,6 +1270,6 @@ var PriorityAlertResponseSchema = z.object({
 var INTERNAL_ERROR_CODE_VALUES = ["UNAUTHENTICATED", "NOT_FOUND", "VALIDATION_ERROR", "INTERNAL"];
 var InternalErrorCode = z.enum(INTERNAL_ERROR_CODE_VALUES);
 
-export { ACCESS_ISSUE_VALUES, AGENT_CARD_KIND_VALUES, AGENT_ERROR_CODE_VALUES, AGENT_MESSAGE_KIND_VALUES, AGENT_MESSAGE_ROLE_VALUES, AGENT_ROLE_VALUES, AGENT_SESSION_MODE_VALUES, AGENT_SESSION_STATUS_VALUES, AGENT_SOURCE_OUTCOME_VALUES, ASSIGNEE_TYPE_VALUES, ASSIGNMENT_SOURCE_VALUES, ASSIGNMENT_STATUS_VALUES, AccessIssue, AgentActiveSessionResponseSchema, AgentAssignmentItemSchema, AgentAssignmentVersionSchema, AgentAssignmentVersionsResponseSchema, AgentAssignmentsResponseSchema, AgentCaptureFieldsSchema, AgentCaptureResponseSchema, AgentCardKind, AgentCardSchema, AgentErrorCode, AgentErrorSchema, AgentMeSchema, AgentMessageCreateSchema, AgentMessageKind, AgentMessageRole, AgentMessageSchema, AgentMessageViewSchema, AgentMessagesQuerySchema, AgentMessagesResponseSchema, AgentOrgRefSchema, AgentReplyBodySchema, AgentRole, AgentSessionBudgetSchema, AgentSessionCreateSchema, AgentSessionMode, AgentSessionResponseSchema, AgentSessionSchema, AgentSessionStartBodySchema, AgentSessionStartResponseSchema, AgentSessionStatsSchema, AgentSessionStatus, AgentSessionUpdateSchema, AgentSessionViewSchema, AgentSourceOutcome, AgentStreamEventSchema, AgentTaskCriteriaSchema, AgentTaskSchema, AgentTaskTargetSchema, AnalystProfileSchema, AnalystProfileUpdateSchema, ApiErrorSchema, AssigneeLoadSchema, AssigneeType, AssignmentCreateSchema, AssignmentSchema, AssignmentSource, AssignmentStatus, AssignmentUpdateSchema, AuditLogCreateSchema, AuditLogSchema, AuditLogUpdateSchema, AuditTargetSchema, AvailabilityWindowSchema, BlobRefSchema, CAPTURE_DOM_STATUS_VALUES, CAPTURE_STATUS_VALUES, CAPTURE_WINDOW_MODE_VALUES, COLLECTIONS, COLLECTOR_TYPE_VALUES, CaptureCreateSchema, CaptureDomSnapshotSchema, CaptureDomStatus, CaptureDomSummarySchema, CapturePromptCardSchema, CaptureRegionSchema, CaptureSchema, CaptureStatus, CaptureThumbCardSchema, CaptureWindowMode, CaptureWindowSchema, ClerkOrgId, ClerkUserId, CollectorSchema, CollectorType, ConnectionString, CoverageWindowSchema, DATABASE_COLLECTION_VALUES, DATABASE_STATUS_VALUES, DEFAULT_ANALYST_TASK_LIMIT, DEFAULT_SCOUT_TASK_LIMIT, DEPLOYMENT_KIND_VALUES, DEPLOYMENT_STATUS_VALUES, DISCOVERED_BY_TYPE_VALUES, DataItemContentSchema, DataItemCreateSchema, DataItemEntitySchema, DataItemSchema, DataItemUpdateSchema, DatabaseCollection, DatabaseCreateSchema, DatabaseId, DatabaseName, DatabaseSchema, DatabaseStatus, DatabaseUpdateSchema, DeploymentCreateSchema, DeploymentKind, DeploymentSchema, DeploymentStatus, DeploymentStepSchema, DeploymentUpdateSchema, DiscoveredBySchema, DiscoveredByType, DomBlockSchema, DomLinkSchema, DomRectSchema, Email, GLOBAL_LOAD_SCOPE, GeoSchema, HHmm, HOURS_LOG_SOURCE_VALUES, HoursLogCreateSchema, HoursLogSchema, HoursLogSource, HoursLogUpdateSchema, HoursProposalCardSchema, INTERNAL_ERROR_CODE_VALUES, InternalErrorCode, ItemResultCardSchema, LLM_CAPABILITY_VALUES, LLM_OUTCOME_VALUES, LLM_PROVIDER_VALUES, LanguageCode, LlmCallCreateSchema, LlmCallSchema, LlmCapability, LlmOutcome, LlmProvider, LlmUsageSchema, LoadScope, MAX_AGENT_MESSAGE_TEXT_LENGTH, MAX_CAPTURE_WINDOW_DAYS, MAX_RATING_COMMENT_LENGTH, MAX_SOURCE_IMPORT_ROW_ERRORS, MAX_SOURCE_INTAKE_CANDIDATES, MISSION_PRIORITY_VALUES, MISSION_STATUS_VALUES, MatchingFieldsSchema, MissionCreateSchema, MissionPriority, MissionSchema, MissionStatus, MissionUpdateSchema, ObjectIdString, OpenLinkCardSchema, OrgSettingsSchema, OrgSettingsUpdateSchema, PLATFORM_VALUES, PaginatedSchema, PaginationQuerySchema, Platform, PlatformAccessSchema, PriorityAlertBodySchema, PriorityAlertResponseSchema, QuickRepliesCardSchema, QuickReplyOptionSchema, RATER_TYPE_VALUES, REGION_MODE_VALUES, RaterType, RatingScore, RatingSummarySchema, RegionMode, SCOUT_STATUS_VALUES, SCRAPER_STATUS_VALUES, SERVER_MANAGED_KEYS, SOURCE_ACCESS_VALUES, SOURCE_IMPORT_KIND_VALUES, SOURCE_IMPORT_STATUS_VALUES, SOURCE_INTAKE_OUTCOME_VALUES, SOURCE_ROUTE_VALUES, SOURCE_STATUS_VALUES, SOURCE_TYPE_VALUES, ScoutProfileCreateSchema, ScoutProfileSchema, ScoutProfileUpdateSchema, ScoutRatingInputSchema, ScoutRatingSchema, ScoutSelfUpdateSchema, ScoutStatus, ScraperCreateSchema, ScraperMetricCreateSchema, ScraperMetricSchema, ScraperMetricUpdateSchema, ScraperSchema, ScraperStatus, ScraperUpdateSchema, SoftDeleteSchema, SourceAccess, SourceCreateSchema, SourceImportCreateSchema, SourceImportKind, SourceImportRowErrorSchema, SourceImportSchema, SourceImportStatus, SourceImportSummarySchema, SourceImportUpdateSchema, SourceIntakeBodySchema, SourceIntakeCandidateSchema, SourceIntakeOutcome, SourceIntakeResponseSchema, SourceIntakeResultSchema, SourceRatingSchema, SourceResultCardSchema, SourceRoute, SourceSchema, SourceStatus, SourceType, SourceUpdateSchema, SourceVisitBodySchema, SourceVisitResponseSchema, SpecialAccessSchema, SystemRecommendationInputSchema, SystemRecommendationSchema, TASK_KIND_VALUES, TASK_ORIGIN_VALUES, TASK_STATUS_VALUES, TaskCreateSchema, TaskCriteriaSchema, TaskKind, TaskOrigin, TaskSchema, TaskStatus, TaskTargetSchema, TaskUpdateSchema, TimeZone, TimestampsSchema, VETTING_STATUS_VALUES, VettingStatus, isHttpUrl };
+export { ACCESS_ISSUE_VALUES, AGENT_CARD_KIND_VALUES, AGENT_ERROR_CODE_VALUES, AGENT_MESSAGE_KIND_VALUES, AGENT_MESSAGE_ROLE_VALUES, AGENT_ROLE_VALUES, AGENT_SESSION_MODE_VALUES, AGENT_SESSION_STATUS_VALUES, AGENT_SOURCE_OUTCOME_VALUES, ASSIGNEE_TYPE_VALUES, ASSIGNMENT_SOURCE_VALUES, ASSIGNMENT_STATUS_VALUES, AccessIssue, AgentActiveSessionResponseSchema, AgentAssignmentItemSchema, AgentAssignmentVersionSchema, AgentAssignmentVersionsResponseSchema, AgentAssignmentsResponseSchema, AgentCaptureFieldsSchema, AgentCaptureResponseSchema, AgentCardKind, AgentCardSchema, AgentErrorCode, AgentErrorSchema, AgentMeSchema, AgentMessageCreateSchema, AgentMessageKind, AgentMessageRole, AgentMessageSchema, AgentMessageViewSchema, AgentMessagesQuerySchema, AgentMessagesResponseSchema, AgentOrgRefSchema, AgentReplyBodySchema, AgentRole, AgentSessionBudgetSchema, AgentSessionCreateSchema, AgentSessionMode, AgentSessionResponseSchema, AgentSessionSchema, AgentSessionStartBodySchema, AgentSessionStartResponseSchema, AgentSessionStatsSchema, AgentSessionStatus, AgentSessionUpdateSchema, AgentSessionViewSchema, AgentSourceOutcome, AgentStreamEventSchema, AgentTaskCriteriaSchema, AgentTaskSchema, AgentTaskSourceSchema, AgentTaskTargetSchema, AnalystProfileSchema, AnalystProfileUpdateSchema, ApiErrorSchema, AssigneeLoadSchema, AssigneeType, AssignmentCreateSchema, AssignmentSchema, AssignmentSource, AssignmentStatus, AssignmentUpdateSchema, AuditLogCreateSchema, AuditLogSchema, AuditLogUpdateSchema, AuditTargetSchema, AvailabilityWindowSchema, BlobRefSchema, CAPTURE_DOM_STATUS_VALUES, CAPTURE_STATUS_VALUES, CAPTURE_WINDOW_MODE_VALUES, COLLECTIONS, COLLECTOR_TYPE_VALUES, CaptureCreateSchema, CaptureDomSnapshotSchema, CaptureDomStatus, CaptureDomSummarySchema, CapturePromptCardSchema, CaptureRegionSchema, CaptureSchema, CaptureStatus, CaptureThumbCardSchema, CaptureWindowMode, CaptureWindowSchema, ClerkOrgId, ClerkUserId, CollectorSchema, CollectorType, ConnectionString, CoverageWindowSchema, DATABASE_COLLECTION_VALUES, DATABASE_STATUS_VALUES, DEFAULT_ANALYST_TASK_LIMIT, DEFAULT_SCOUT_TASK_LIMIT, DEPLOYMENT_KIND_VALUES, DEPLOYMENT_STATUS_VALUES, DISCOVERED_BY_TYPE_VALUES, DataItemContentSchema, DataItemCreateSchema, DataItemEntitySchema, DataItemSchema, DataItemUpdateSchema, DatabaseCollection, DatabaseCreateSchema, DatabaseId, DatabaseName, DatabaseSchema, DatabaseStatus, DatabaseUpdateSchema, DeploymentCreateSchema, DeploymentKind, DeploymentSchema, DeploymentStatus, DeploymentStepSchema, DeploymentUpdateSchema, DiscoveredBySchema, DiscoveredByType, DomBlockSchema, DomLinkSchema, DomRectSchema, Email, GLOBAL_LOAD_SCOPE, GeoSchema, HHmm, HOURS_LOG_SOURCE_VALUES, HoursLogCreateSchema, HoursLogSchema, HoursLogSource, HoursLogUpdateSchema, HoursProposalCardSchema, INTERNAL_ERROR_CODE_VALUES, InternalErrorCode, ItemResultCardSchema, LLM_CAPABILITY_VALUES, LLM_OUTCOME_VALUES, LLM_PROVIDER_VALUES, LanguageCode, LlmCallCreateSchema, LlmCallSchema, LlmCapability, LlmOutcome, LlmProvider, LlmUsageSchema, LoadScope, MAX_AGENT_MESSAGE_TEXT_LENGTH, MAX_CAPTURE_WINDOW_DAYS, MAX_RATING_COMMENT_LENGTH, MAX_SOURCE_IMPORT_ROW_ERRORS, MAX_SOURCE_INTAKE_CANDIDATES, MISSION_PRIORITY_VALUES, MISSION_STATUS_VALUES, MONITOR_TASK_MAX_SOURCES, MatchingFieldsSchema, MissionCreateSchema, MissionPriority, MissionSchema, MissionStatus, MissionUpdateSchema, ObjectIdString, OpenLinkCardSchema, OrgSettingsSchema, OrgSettingsUpdateSchema, OverdueLevelSchema, PLATFORM_VALUES, PaginatedSchema, PaginationQuerySchema, Platform, PlatformAccessSchema, PriorityAlertBodySchema, PriorityAlertResponseSchema, QuickRepliesCardSchema, QuickReplyOptionSchema, RATER_TYPE_VALUES, REGION_MODE_VALUES, RaterType, RatingScore, RatingSummarySchema, RegionMode, SCOUT_STATUS_VALUES, SCRAPER_STATUS_VALUES, SERVER_MANAGED_KEYS, SOURCE_ACCESS_VALUES, SOURCE_IMPORT_KIND_VALUES, SOURCE_IMPORT_STATUS_VALUES, SOURCE_INTAKE_OUTCOME_VALUES, SOURCE_ROUTE_VALUES, SOURCE_STATUS_VALUES, SOURCE_TYPE_VALUES, ScoutProfileCreateSchema, ScoutProfileSchema, ScoutProfileUpdateSchema, ScoutRatingInputSchema, ScoutRatingSchema, ScoutSelfUpdateSchema, ScoutStatus, ScraperCreateSchema, ScraperMetricCreateSchema, ScraperMetricSchema, ScraperMetricUpdateSchema, ScraperSchema, ScraperStatus, ScraperUpdateSchema, SoftDeleteSchema, SourceAccess, SourceCreateSchema, SourceDiscoverySchema, SourceImportCreateSchema, SourceImportKind, SourceImportRowErrorSchema, SourceImportSchema, SourceImportStatus, SourceImportSummarySchema, SourceImportUpdateSchema, SourceIntakeBodySchema, SourceIntakeCandidateSchema, SourceIntakeOutcome, SourceIntakeResponseSchema, SourceIntakeResultSchema, SourceRatingSchema, SourceResultCardSchema, SourceRoute, SourceSchema, SourceStatus, SourceType, SourceUpdateSchema, SourceVisitBodySchema, SourceVisitResponseSchema, SpecialAccessSchema, SystemRecommendationInputSchema, SystemRecommendationSchema, TASK_KIND_VALUES, TASK_ORIGIN_VALUES, TASK_STATUS_VALUES, TaskCreateSchema, TaskCriteriaSchema, TaskKind, TaskOrigin, TaskSchema, TaskStatus, TaskTargetSchema, TaskUpdateSchema, TimeZone, TimestampsSchema, URGENCY_FACTOR, VETTING_STATUS_VALUES, VISIT_INTERVAL_MAX_MS, VISIT_INTERVAL_MIN_MS, VettingStatus, baseVisitInterval, clampInterval, effectiveDueAt, effectiveInterval, isHttpUrl, overdueLevel, urgencyFactor };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map