@badger_badger/badger_hq 1.5.1 → 1.6.0

package/dist/index.js

@@ -86,6 +86,21 @@ var SOURCE_ROUTE_VALUES = ["scraper", "monitor", "waiting_for_scraper"];
 var SourceRoute = z.enum(SOURCE_ROUTE_VALUES);
 var SOURCE_STATUS_VALUES = ["active", "paused", "closed", "access_problem"];
 var SourceStatus = z.enum(SOURCE_STATUS_VALUES);
+var ACCESS_ISSUE_VALUES = [
+  "login_required",
+  "private",
+  "bot_detection",
+  "captcha",
+  "geo_blocked",
+  "rate_limited",
+  "not_found",
+  "no_collector",
+  "collector_full",
+  "other"
+];
+var AccessIssue = z.enum(ACCESS_ISSUE_VALUES);
+var CAPTURE_WINDOW_MODE_VALUES = ["last_days", "range"];
+var CaptureWindowMode = z.enum(CAPTURE_WINDOW_MODE_VALUES);
 var DISCOVERED_BY_TYPE_VALUES = ["scout", "analyst", "agent", "client", "import"];
 var DiscoveredByType = z.enum(DISCOVERED_BY_TYPE_VALUES);
 var SOURCE_IMPORT_KIND_VALUES = ["create", "update"];
@@ -133,6 +148,8 @@ var AGENT_SOURCE_OUTCOME_VALUES = [
 var AgentSourceOutcome = z.enum(AGENT_SOURCE_OUTCOME_VALUES);
 var CAPTURE_STATUS_VALUES = ["uploaded", "analyzing", "analyzed", "failed", "expired"];
 var CaptureStatus = z.enum(CAPTURE_STATUS_VALUES);
+var HOURS_LOG_SOURCE_VALUES = ["manual", "inferred"];
+var HoursLogSource = z.enum(HOURS_LOG_SOURCE_VALUES);
 var CAPTURE_DOM_STATUS_VALUES = ["ok", "unavailable", "truncated"];
 var CaptureDomStatus = z.enum(CAPTURE_DOM_STATUS_VALUES);
 var LLM_CAPABILITY_VALUES = ["plan", "vision_extract", "judge", "chat"];
@@ -214,6 +231,51 @@ var TaskCriteriaSchema = z.object({
   coverageWindow: CoverageWindowSchema.optional(),
   emergency: z.boolean().default(false)
 });
+var MAX_CAPTURE_WINDOW_DAYS = 90;
+var DAY_MS = 24 * 60 * 60 * 1e3;
+function isHttpUrl(value) {
+  try {
+    const { protocol } = new URL(value);
+    return protocol === "http:" || protocol === "https:";
+  } catch {
+    return false;
+  }
+}
+var WindowDate = z.union([z.date(), z.string().datetime({ offset: true })]).pipe(z.coerce.date());
+var TaskTargetSchema = z.object({
+  sourceId: ObjectIdString.optional(),
+  platform: Platform,
+  url: z.string().url().refine(isHttpUrl, "Expected an http(s) URL"),
+  label: z.string().min(1).max(200).optional(),
+  /** Why a person collects it instead of a scraper. */
+  accessIssue: AccessIssue.optional(),
+  note: z.string().max(500).optional()
+});
+var LastDaysWindowSchema = z.object({
+  mode: z.literal(CaptureWindowMode.enum.last_days),
+  days: z.number().int().min(1).max(MAX_CAPTURE_WINDOW_DAYS)
+});
+var RangeWindowSchema = z.object({
+  mode: z.literal(CaptureWindowMode.enum.range),
+  start: WindowDate,
+  end: WindowDate
+});
+var CaptureWindowSchema = z.discriminatedUnion("mode", [LastDaysWindowSchema, RangeWindowSchema]).superRefine((w, ctx) => {
+  if (w.mode !== "range" || !(w.start instanceof Date) || !(w.end instanceof Date)) return;
+  if (w.end.getTime() <= w.start.getTime()) {
+    ctx.addIssue({
+      code: z.ZodIssueCode.custom,
+      message: "captureWindow.end must be after captureWindow.start",
+      path: ["end"]
+    });
+  } else if (w.end.getTime() - w.start.getTime() > MAX_CAPTURE_WINDOW_DAYS * DAY_MS) {
+    ctx.addIssue({
+      code: z.ZodIssueCode.custom,
+      message: `captureWindow range must be at most ${MAX_CAPTURE_WINDOW_DAYS} days`,
+      path: ["end"]
+    });
+  }
+});
 var MissionSchema = z.object({
   _id: ObjectIdString,
   orgId: ClerkOrgId,
@@ -249,6 +311,10 @@ var TaskSchema = z.object({
   keywords: z.array(z.string().min(1)).default([]),
   searchQueries: z.array(z.string().min(1)).default([]),
   criteria: TaskCriteriaSchema,
+  /** Monitor: the one source to collect. Discovery: optional source to search inside. */
+  target: TaskTargetSchema.optional(),
+  /** Monitor: what to capture (last N days, or a fixed range). */
+  captureWindow: CaptureWindowSchema.optional(),
   status: TaskStatus.default("draft"),
   approvedBy: ClerkUserId.optional(),
   approvedAt: z.coerce.date().optional(),
@@ -307,7 +373,9 @@ var HoursLogSchema = z.object({
   scoutUserId: ClerkUserId,
   minutes: z.number().int().positive(),
   note: z.string(),
-  date: z.coerce.date()
+  date: z.coerce.date(),
+  assignmentId: ObjectIdString.optional(),
+  source: HoursLogSource.optional()
 }).merge(TimestampsSchema).merge(SoftDeleteSchema);
 var HoursLogCreateSchema = HoursLogSchema.omit({
   _id: true,
@@ -713,6 +781,8 @@ var AgentSessionSchema = z.object({
   lastSeq: count(),
   stats: AgentSessionStatsSchema.default({}),
   budget: AgentSessionBudgetSchema.default({}),
+  /** The task's capture window as dates, fixed when the session started (`last_days` counted back from then). */
+  captureWindow: CoverageWindowSchema.optional(),
   /** Mode-specific engine state (search plan cursor, monitor queue). Opaque to clients. */
   state: z.unknown().optional()
 }).merge(TimestampsSchema);
@@ -818,7 +888,14 @@ var CaptureDomSummarySchema = z.object({
 var CaptureSchema = z.object({
   _id: ObjectIdString,
   orgId: ClerkOrgId,
-  sessionId: ObjectIdString,
+  /** Legacy: captures uploaded into a live session. Task-tagged captures have none. */
+  sessionId: ObjectIdString.optional(),
+  /** The assignment (and so the task) the person sent this capture for. */
+  assignmentId: ObjectIdString.optional(),
+  taskId: ObjectIdString.optional(),
+  missionId: ObjectIdString.optional(),
+  /** Who sent it; used to work out their hours. */
+  userId: ClerkUserId.optional(),
   /** From `chrome.tabs`, never from page content. */
   url: z.string().url(),
   title: z.string().max(500),
@@ -834,6 +911,10 @@ var CaptureSchema = z.object({
   /** Validated `vision_extract` output; shape depends on session mode. */
   analysis: z.unknown().optional(),
   error: z.string().max(1e3).optional(),
+  /** Analysis attempts by the capture worker; it gives up (`failed`) after a few. */
+  attempts: z.number().int().min(0).default(0),
+  /** When the worker claimed it (`analyzing`); an old claim is taken over after a restart. */
+  claimedAt: z.coerce.date().optional(),
   /** Blob deleted and status set to `expired` after this instant (CAPTURE_RETENTION_DAYS). */
   expiresAt: z.coerce.date()
 }).merge(TimestampsSchema);
@@ -914,6 +995,7 @@ var AgentTaskCriteriaSchema = z.object({
   coverageWindow: CoverageWindowSchema.optional(),
   emergency: z.boolean()
 });
+var AgentTaskTargetSchema = TaskTargetSchema.omit({ sourceId: true });
 var AgentTaskSchema = z.object({
   id: ObjectIdString,
   title: z.string(),
@@ -922,6 +1004,8 @@ var AgentTaskSchema = z.object({
   keywords: z.array(z.string()),
   searchQueries: z.array(z.string()),
   criteria: AgentTaskCriteriaSchema,
+  target: AgentTaskTargetSchema.optional(),
+  captureWindow: CaptureWindowSchema.optional(),
   status: TaskStatus,
   version: z.number().int().min(1),
   updatedAt: z.coerce.date()
@@ -982,7 +1066,11 @@ var AgentSessionViewSchema = z.object({
   endedAt: z.coerce.date().nullable(),
   activeMinutes: z.number().int().min(0),
   lastSeq: z.number().int().min(0),
-  stats: AgentSessionStatsSchema
+  stats: AgentSessionStatsSchema,
+  /** The task's source (no internal id), when it has one. */
+  target: AgentTaskTargetSchema.optional(),
+  /** The capture window resolved to dates when the session started. */
+  captureWindow: CoverageWindowSchema.optional()
 });
 var AgentMessageViewSchema = z.object({
   id: ObjectIdString,
@@ -1099,6 +1187,6 @@ var PriorityAlertResponseSchema = z.object({
 var INTERNAL_ERROR_CODE_VALUES = ["UNAUTHENTICATED", "NOT_FOUND", "VALIDATION_ERROR", "INTERNAL"];
 var InternalErrorCode = z.enum(INTERNAL_ERROR_CODE_VALUES);
 
-export { 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, 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, AnalystProfileSchema, AnalystProfileUpdateSchema, ApiErrorSchema, AssigneeLoadSchema, AssigneeType, AssignmentCreateSchema, AssignmentSchema, AssignmentSource, AssignmentStatus, AssignmentUpdateSchema, AuditLogCreateSchema, AuditLogSchema, AuditLogUpdateSchema, AuditTargetSchema, AvailabilityWindowSchema, BlobRefSchema, CAPTURE_DOM_STATUS_VALUES, CAPTURE_STATUS_VALUES, COLLECTIONS, COLLECTOR_TYPE_VALUES, CaptureCreateSchema, CaptureDomSnapshotSchema, CaptureDomStatus, CaptureDomSummarySchema, CapturePromptCardSchema, CaptureRegionSchema, CaptureSchema, CaptureStatus, CaptureThumbCardSchema, 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, HoursLogCreateSchema, HoursLogSchema, 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_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, TaskUpdateSchema, TimeZone, TimestampsSchema, VETTING_STATUS_VALUES, VettingStatus };
+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 };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map