@badger_badger/badger_hq 1.5.1 → 1.7.0

package/dist/index.cjs

@@ -88,6 +88,21 @@ var SOURCE_ROUTE_VALUES = ["scraper", "monitor", "waiting_for_scraper"];
 var SourceRoute = zod.z.enum(SOURCE_ROUTE_VALUES);
 var SOURCE_STATUS_VALUES = ["active", "paused", "closed", "access_problem"];
 var SourceStatus = zod.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 = zod.z.enum(ACCESS_ISSUE_VALUES);
+var CAPTURE_WINDOW_MODE_VALUES = ["last_days", "range"];
+var CaptureWindowMode = zod.z.enum(CAPTURE_WINDOW_MODE_VALUES);
 var DISCOVERED_BY_TYPE_VALUES = ["scout", "analyst", "agent", "client", "import"];
 var DiscoveredByType = zod.z.enum(DISCOVERED_BY_TYPE_VALUES);
 var SOURCE_IMPORT_KIND_VALUES = ["create", "update"];
@@ -135,6 +150,8 @@ var AGENT_SOURCE_OUTCOME_VALUES = [
 var AgentSourceOutcome = zod.z.enum(AGENT_SOURCE_OUTCOME_VALUES);
 var CAPTURE_STATUS_VALUES = ["uploaded", "analyzing", "analyzed", "failed", "expired"];
 var CaptureStatus = zod.z.enum(CAPTURE_STATUS_VALUES);
+var HOURS_LOG_SOURCE_VALUES = ["manual", "inferred"];
+var HoursLogSource = zod.z.enum(HOURS_LOG_SOURCE_VALUES);
 var CAPTURE_DOM_STATUS_VALUES = ["ok", "unavailable", "truncated"];
 var CaptureDomStatus = zod.z.enum(CAPTURE_DOM_STATUS_VALUES);
 var LLM_CAPABILITY_VALUES = ["plan", "vision_extract", "judge", "chat"];
@@ -178,6 +195,50 @@ 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 = zod.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 = zod.z.string().min(1).max(63).regex(/^[^/\\. "$*<>:|?]+$/, "Invalid MongoDB database name");
 var ConnectionString = zod.z.string().min(1).regex(/^mongodb(\+srv)?:\/\/\S+$/, "Expected a mongodb:// or mongodb+srv:// URI");
 var DatabaseSchema = zod.z.object({
@@ -216,6 +277,52 @@ var TaskCriteriaSchema = zod.z.object({
   coverageWindow: CoverageWindowSchema.optional(),
   emergency: zod.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 {
+    const { protocol } = new URL(value);
+    return protocol === "http:" || protocol === "https:";
+  } catch {
+    return false;
+  }
+}
+var WindowDate = zod.z.union([zod.z.date(), zod.z.string().datetime({ offset: true })]).pipe(zod.z.coerce.date());
+var TaskTargetSchema = zod.z.object({
+  sourceId: ObjectIdString.optional(),
+  platform: Platform,
+  url: zod.z.string().url().refine(isHttpUrl, "Expected an http(s) URL"),
+  label: zod.z.string().min(1).max(200).optional(),
+  /** Why a person collects it instead of a scraper. */
+  accessIssue: AccessIssue.optional(),
+  note: zod.z.string().max(500).optional()
+});
+var LastDaysWindowSchema = zod.z.object({
+  mode: zod.z.literal(CaptureWindowMode.enum.last_days),
+  days: zod.z.number().int().min(1).max(MAX_CAPTURE_WINDOW_DAYS)
+});
+var RangeWindowSchema = zod.z.object({
+  mode: zod.z.literal(CaptureWindowMode.enum.range),
+  start: WindowDate,
+  end: WindowDate
+});
+var CaptureWindowSchema = zod.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: zod.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: zod.z.ZodIssueCode.custom,
+      message: `captureWindow range must be at most ${MAX_CAPTURE_WINDOW_DAYS} days`,
+      path: ["end"]
+    });
+  }
+});
 var MissionSchema = zod.z.object({
   _id: ObjectIdString,
   orgId: ClerkOrgId,
@@ -251,6 +358,10 @@ var TaskSchema = zod.z.object({
   keywords: zod.z.array(zod.z.string().min(1)).default([]),
   searchQueries: zod.z.array(zod.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: zod.z.coerce.date().optional(),
@@ -309,7 +420,9 @@ var HoursLogSchema = zod.z.object({
   scoutUserId: ClerkUserId,
   minutes: zod.z.number().int().positive(),
   note: zod.z.string(),
-  date: zod.z.coerce.date()
+  date: zod.z.coerce.date(),
+  assignmentId: ObjectIdString.optional(),
+  source: HoursLogSource.optional()
 }).merge(TimestampsSchema).merge(SoftDeleteSchema);
 var HoursLogCreateSchema = HoursLogSchema.omit({
   _id: true,
@@ -488,7 +601,9 @@ var SourceSchema = zod.z.object({
   status: SourceStatus.default("active"),
   discoveredBy: DiscoveredBySchema,
   lastCollectedAt: zod.z.coerce.date().optional(),
-  nextVisitAt: zod.z.coerce.date().optional()
+  nextVisitAt: zod.z.coerce.date().optional(),
+  /** Activity-based wait set by the last visit, before current urgency (see `effectiveDueAt`). */
+  visitIntervalMs: zod.z.number().int().positive().optional()
 }).merge(TimestampsSchema).merge(SoftDeleteSchema);
 var SourceCreateSchema = SourceSchema.omit({
   _id: true,
@@ -498,7 +613,8 @@ var SourceCreateSchema = SourceSchema.omit({
   updatedAt: true,
   deletedAt: true,
   discoveredBy: true,
-  lastCollectedAt: true
+  lastCollectedAt: true,
+  visitIntervalMs: true
 });
 var SourceUpdateSchema = SourceCreateSchema.partial();
 var MAX_SOURCE_IMPORT_ROW_ERRORS = 500;
@@ -715,6 +831,8 @@ var AgentSessionSchema = zod.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: zod.z.unknown().optional()
 }).merge(TimestampsSchema);
@@ -820,7 +938,14 @@ var CaptureDomSummarySchema = zod.z.object({
 var CaptureSchema = zod.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: zod.z.string().url(),
   title: zod.z.string().max(500),
@@ -836,6 +961,10 @@ var CaptureSchema = zod.z.object({
   /** Validated `vision_extract` output; shape depends on session mode. */
   analysis: zod.z.unknown().optional(),
   error: zod.z.string().max(1e3).optional(),
+  /** Analysis attempts by the capture worker; it gives up (`failed`) after a few. */
+  attempts: zod.z.number().int().min(0).default(0),
+  /** When the worker claimed it (`analyzing`); an old claim is taken over after a restart. */
+  claimedAt: zod.z.coerce.date().optional(),
   /** Blob deleted and status set to `expired` after this instant (CAPTURE_RETENTION_DAYS). */
   expiresAt: zod.z.coerce.date()
 }).merge(TimestampsSchema);
@@ -916,6 +1045,17 @@ var AgentTaskCriteriaSchema = zod.z.object({
   coverageWindow: CoverageWindowSchema.optional(),
   emergency: zod.z.boolean()
 });
+var AgentTaskTargetSchema = TaskTargetSchema.omit({ sourceId: true });
+var AgentTaskSourceSchema = zod.z.object({
+  platform: Platform,
+  url: zod.z.string().url().refine(isHttpUrl, "Expected an http(s) URL"),
+  label: zod.z.string().min(1).optional(),
+  lastCapturedAt: zod.z.coerce.date().optional(),
+  /** When to capture it again, under current urgency (absent: parked on an access problem). */
+  dueAt: zod.z.coerce.date().optional(),
+  /** Parked: a person couldn't get in; an analyst reviews it. */
+  accessProblem: zod.z.boolean().optional()
+});
 var AgentTaskSchema = zod.z.object({
   id: ObjectIdString,
   title: zod.z.string(),
@@ -924,6 +1064,10 @@ var AgentTaskSchema = zod.z.object({
   keywords: zod.z.array(zod.z.string()),
   searchQueries: zod.z.array(zod.z.string()),
   criteria: AgentTaskCriteriaSchema,
+  target: AgentTaskTargetSchema.optional(),
+  /** Monitor: the attached sources other than `target`, sorted by platform then URL. */
+  sources: zod.z.array(AgentTaskSourceSchema).default([]),
+  captureWindow: CaptureWindowSchema.optional(),
   status: TaskStatus,
   version: zod.z.number().int().min(1),
   updatedAt: zod.z.coerce.date()
@@ -944,7 +1088,9 @@ var AgentAssignmentsResponseSchema = zod.z.object({
 });
 var AgentAssignmentVersionSchema = zod.z.object({
   assignmentId: ObjectIdString,
-  taskVersion: zod.z.number().int().min(1)
+  taskVersion: zod.z.number().int().min(1),
+  /** Monitor: the latest change to any attached source (a visit moves it; the task version doesn't). */
+  sourcesUpdatedAt: zod.z.coerce.date().optional()
 });
 var AgentAssignmentVersionsResponseSchema = zod.z.object({
   items: zod.z.array(AgentAssignmentVersionSchema)
@@ -984,7 +1130,11 @@ var AgentSessionViewSchema = zod.z.object({
   endedAt: zod.z.coerce.date().nullable(),
   activeMinutes: zod.z.number().int().min(0),
   lastSeq: zod.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 = zod.z.object({
   id: ObjectIdString,
@@ -1101,6 +1251,7 @@ var PriorityAlertResponseSchema = zod.z.object({
 var INTERNAL_ERROR_CODE_VALUES = ["UNAUTHENTICATED", "NOT_FOUND", "VALIDATION_ERROR", "INTERNAL"];
 var InternalErrorCode = zod.z.enum(INTERNAL_ERROR_CODE_VALUES);
 
+exports.ACCESS_ISSUE_VALUES = ACCESS_ISSUE_VALUES;
 exports.AGENT_CARD_KIND_VALUES = AGENT_CARD_KIND_VALUES;
 exports.AGENT_ERROR_CODE_VALUES = AGENT_ERROR_CODE_VALUES;
 exports.AGENT_MESSAGE_KIND_VALUES = AGENT_MESSAGE_KIND_VALUES;
@@ -1112,6 +1263,7 @@ exports.AGENT_SOURCE_OUTCOME_VALUES = AGENT_SOURCE_OUTCOME_VALUES;
 exports.ASSIGNEE_TYPE_VALUES = ASSIGNEE_TYPE_VALUES;
 exports.ASSIGNMENT_SOURCE_VALUES = ASSIGNMENT_SOURCE_VALUES;
 exports.ASSIGNMENT_STATUS_VALUES = ASSIGNMENT_STATUS_VALUES;
+exports.AccessIssue = AccessIssue;
 exports.AgentActiveSessionResponseSchema = AgentActiveSessionResponseSchema;
 exports.AgentAssignmentItemSchema = AgentAssignmentItemSchema;
 exports.AgentAssignmentVersionSchema = AgentAssignmentVersionSchema;
@@ -1149,6 +1301,8 @@ exports.AgentSourceOutcome = AgentSourceOutcome;
 exports.AgentStreamEventSchema = AgentStreamEventSchema;
 exports.AgentTaskCriteriaSchema = AgentTaskCriteriaSchema;
 exports.AgentTaskSchema = AgentTaskSchema;
+exports.AgentTaskSourceSchema = AgentTaskSourceSchema;
+exports.AgentTaskTargetSchema = AgentTaskTargetSchema;
 exports.AnalystProfileSchema = AnalystProfileSchema;
 exports.AnalystProfileUpdateSchema = AnalystProfileUpdateSchema;
 exports.ApiErrorSchema = ApiErrorSchema;
@@ -1167,6 +1321,7 @@ exports.AvailabilityWindowSchema = AvailabilityWindowSchema;
 exports.BlobRefSchema = BlobRefSchema;
 exports.CAPTURE_DOM_STATUS_VALUES = CAPTURE_DOM_STATUS_VALUES;
 exports.CAPTURE_STATUS_VALUES = CAPTURE_STATUS_VALUES;
+exports.CAPTURE_WINDOW_MODE_VALUES = CAPTURE_WINDOW_MODE_VALUES;
 exports.COLLECTIONS = COLLECTIONS;
 exports.COLLECTOR_TYPE_VALUES = COLLECTOR_TYPE_VALUES;
 exports.CaptureCreateSchema = CaptureCreateSchema;
@@ -1178,6 +1333,8 @@ exports.CaptureRegionSchema = CaptureRegionSchema;
 exports.CaptureSchema = CaptureSchema;
 exports.CaptureStatus = CaptureStatus;
 exports.CaptureThumbCardSchema = CaptureThumbCardSchema;
+exports.CaptureWindowMode = CaptureWindowMode;
+exports.CaptureWindowSchema = CaptureWindowSchema;
 exports.ClerkOrgId = ClerkOrgId;
 exports.ClerkUserId = ClerkUserId;
 exports.CollectorSchema = CollectorSchema;
@@ -1218,8 +1375,10 @@ exports.Email = Email;
 exports.GLOBAL_LOAD_SCOPE = GLOBAL_LOAD_SCOPE;
 exports.GeoSchema = GeoSchema;
 exports.HHmm = HHmm;
+exports.HOURS_LOG_SOURCE_VALUES = HOURS_LOG_SOURCE_VALUES;
 exports.HoursLogCreateSchema = HoursLogCreateSchema;
 exports.HoursLogSchema = HoursLogSchema;
+exports.HoursLogSource = HoursLogSource;
 exports.HoursLogUpdateSchema = HoursLogUpdateSchema;
 exports.HoursProposalCardSchema = HoursProposalCardSchema;
 exports.INTERNAL_ERROR_CODE_VALUES = INTERNAL_ERROR_CODE_VALUES;
@@ -1237,11 +1396,13 @@ exports.LlmProvider = LlmProvider;
 exports.LlmUsageSchema = LlmUsageSchema;
 exports.LoadScope = LoadScope;
 exports.MAX_AGENT_MESSAGE_TEXT_LENGTH = MAX_AGENT_MESSAGE_TEXT_LENGTH;
+exports.MAX_CAPTURE_WINDOW_DAYS = MAX_CAPTURE_WINDOW_DAYS;
 exports.MAX_RATING_COMMENT_LENGTH = MAX_RATING_COMMENT_LENGTH;
 exports.MAX_SOURCE_IMPORT_ROW_ERRORS = MAX_SOURCE_IMPORT_ROW_ERRORS;
 exports.MAX_SOURCE_INTAKE_CANDIDATES = MAX_SOURCE_INTAKE_CANDIDATES;
 exports.MISSION_PRIORITY_VALUES = MISSION_PRIORITY_VALUES;
 exports.MISSION_STATUS_VALUES = MISSION_STATUS_VALUES;
+exports.MONITOR_TASK_MAX_SOURCES = MONITOR_TASK_MAX_SOURCES;
 exports.MatchingFieldsSchema = MatchingFieldsSchema;
 exports.MissionCreateSchema = MissionCreateSchema;
 exports.MissionPriority = MissionPriority;
@@ -1252,6 +1413,7 @@ exports.ObjectIdString = ObjectIdString;
 exports.OpenLinkCardSchema = OpenLinkCardSchema;
 exports.OrgSettingsSchema = OrgSettingsSchema;
 exports.OrgSettingsUpdateSchema = OrgSettingsUpdateSchema;
+exports.OverdueLevelSchema = OverdueLevelSchema;
 exports.PLATFORM_VALUES = PLATFORM_VALUES;
 exports.PaginatedSchema = PaginatedSchema;
 exports.PaginationQuerySchema = PaginationQuerySchema;
@@ -1327,10 +1489,21 @@ exports.TaskKind = TaskKind;
 exports.TaskOrigin = TaskOrigin;
 exports.TaskSchema = TaskSchema;
 exports.TaskStatus = TaskStatus;
+exports.TaskTargetSchema = TaskTargetSchema;
 exports.TaskUpdateSchema = TaskUpdateSchema;
 exports.TimeZone = TimeZone;
 exports.TimestampsSchema = TimestampsSchema;
+exports.URGENCY_FACTOR = URGENCY_FACTOR;
 exports.VETTING_STATUS_VALUES = VETTING_STATUS_VALUES;
+exports.VISIT_INTERVAL_MAX_MS = VISIT_INTERVAL_MAX_MS;
+exports.VISIT_INTERVAL_MIN_MS = VISIT_INTERVAL_MIN_MS;
 exports.VettingStatus = VettingStatus;
+exports.baseVisitInterval = baseVisitInterval;
+exports.clampInterval = clampInterval;
+exports.effectiveDueAt = effectiveDueAt;
+exports.effectiveInterval = effectiveInterval;
+exports.isHttpUrl = isHttpUrl;
+exports.overdueLevel = overdueLevel;
+exports.urgencyFactor = urgencyFactor;
 //# sourceMappingURL=index.cjs.map
 //# sourceMappingURL=index.cjs.map