@badger_badger/badger_hq 1.5.1 → 1.6.0

package/dist/mongo/index.js

@@ -37,7 +37,7 @@ var TASK_STATUS_VALUES = [
 ];
 z.enum(TASK_STATUS_VALUES);
 var REGION_MODE_VALUES = ["must", "preferred"];
-z.enum(REGION_MODE_VALUES);
+var RegionMode = z.enum(REGION_MODE_VALUES);
 var ASSIGNEE_TYPE_VALUES = ["scout", "analyst"];
 z.enum(ASSIGNEE_TYPE_VALUES);
 var ASSIGNMENT_STATUS_VALUES = ["active", "completed", "revoked"];
@@ -58,6 +58,21 @@ var SOURCE_ROUTE_VALUES = ["scraper", "monitor", "waiting_for_scraper"];
 z.enum(SOURCE_ROUTE_VALUES);
 var SOURCE_STATUS_VALUES = ["active", "paused", "closed", "access_problem"];
 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"];
 z.enum(DISCOVERED_BY_TYPE_VALUES);
 var SOURCE_IMPORT_KIND_VALUES = ["create", "update"];
@@ -105,6 +120,8 @@ 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"];
+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"];
@@ -197,7 +214,11 @@ var INDEXES = {
   ],
   [COLLECTIONS.assigneeLoads]: [{ key: { userId: 1, scope: 1 }, options: { unique: true } }],
   [COLLECTIONS.missions]: [{ key: { orgId: 1, status: 1, priority: 1 } }],
-  [COLLECTIONS.tasks]: [{ key: { orgId: 1, missionId: 1, status: 1 } }],
+  [COLLECTIONS.tasks]: [
+    { key: { orgId: 1, missionId: 1, status: 1 } },
+    // Tasks that target a source (cascade on source delete).
+    { key: { orgId: 1, "target.sourceId": 1 }, options: { partialFilterExpression: { "target.sourceId": { $exists: true } } } }
+  ],
   [COLLECTIONS.assignments]: [
     { key: { assigneeUserId: 1, status: 1, orgId: 1 } },
     { key: { orgId: 1, taskId: 1, status: 1 } },
@@ -210,7 +231,14 @@ var INDEXES = {
       }
     }
   ],
-  [COLLECTIONS.hoursLogs]: [{ key: { orgId: 1, scoutUserId: 1, date: -1 } }],
+  [COLLECTIONS.hoursLogs]: [
+    { key: { orgId: 1, scoutUserId: 1, date: -1 } },
+    // One inferred row per person + assignment + day (agent-service upserts it).
+    {
+      key: { orgId: 1, assignmentId: 1, scoutUserId: 1, date: 1 },
+      options: { unique: true, partialFilterExpression: { source: "inferred" } }
+    }
+  ],
   [COLLECTIONS.sources]: [
     {
       key: { orgId: 1, platform: 1, url: 1 },
@@ -253,7 +281,11 @@ var INDEXES = {
   [COLLECTIONS.captures]: [
     { key: { orgId: 1, sessionId: 1, capturedAt: -1 } },
     // Retention sweep: captures whose blob must be deleted.
-    { key: { status: 1, expiresAt: 1 } }
+    { key: { status: 1, expiresAt: 1 } },
+    // Capture worker: oldest waiting (or stale analyzing) captures first.
+    { key: { status: 1, createdAt: 1 } },
+    // Task-tagged captures: per-assignment dedupe and inferred hours.
+    { key: { orgId: 1, assignmentId: 1, userId: 1, capturedAt: 1 } }
   ],
   [COLLECTIONS.llmCalls]: [
     { key: { orgId: 1, createdAt: -1 } },
@@ -338,6 +370,24 @@ OrgSettingsSchema.pick({
   defaultAnalystTaskLimit: true,
   timezone: true
 }).partial();
+var CoverageWindowSchema = z.object({
+  start: z.coerce.date(),
+  end: z.coerce.date()
+}).refine((w) => w.end.getTime() > w.start.getTime(), {
+  message: "coverageWindow.end must be after coverageWindow.start",
+  path: ["end"]
+});
+z.object({
+  languages: z.array(LanguageCode),
+  platforms: z.array(Platform),
+  regions: z.array(z.string().min(1)),
+  regionMode: RegionMode,
+  profession: z.string().min(1).optional(),
+  coverageWindow: CoverageWindowSchema.optional(),
+  emergency: z.boolean().default(false)
+});
+
+// src/zod/agent-session.ts
 var count = () => z.number().int().min(0).default(0);
 var AgentSessionStatsSchema = z.object({
   capturesCount: count(),
@@ -374,6 +424,8 @@ 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);
@@ -479,7 +531,14 @@ 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),
@@ -495,6 +554,10 @@ 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);
@@ -612,6 +675,51 @@ var SourceImportCreateSchema = SourceImportSchema.omit({
   committedAt: true
 });
 SourceImportCreateSchema.partial();
+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());
+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
+});
+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"]
+    });
+  }
+});
 
 // src/mongo/schemas.ts
 var toObject = { ...toJSONOptions, flattenMaps: true };
@@ -662,6 +770,50 @@ var taskCriteriaSchema = new Schema(
   },
   EMBEDDED
 );
+var taskTargetSchema = new Schema(
+  {
+    sourceId: { type: ObjectId },
+    platform: { type: String, enum: PLATFORM_VALUES, required: true },
+    url: { type: String, required: true, minlength: 1 },
+    label: { type: String, maxlength: 200 },
+    accessIssue: { type: String, enum: ACCESS_ISSUE_VALUES },
+    note: { type: String, maxlength: 500 }
+  },
+  EMBEDDED
+);
+var captureWindowSchema = new Schema(
+  {
+    mode: { type: String, enum: CAPTURE_WINDOW_MODE_VALUES, required: true },
+    days: {
+      type: Number,
+      min: 1,
+      max: MAX_CAPTURE_WINDOW_DAYS,
+      validate: Number.isInteger,
+      required() {
+        return this.mode === "last_days";
+      }
+    },
+    start: {
+      type: Date,
+      required() {
+        return this.mode === "range";
+      }
+    },
+    end: {
+      type: Date,
+      required() {
+        return this.mode === "range";
+      },
+      validate: {
+        validator(end) {
+          return !this.start || end.getTime() > this.start.getTime();
+        },
+        message: "captureWindow.end must be after captureWindow.start"
+      }
+    }
+  },
+  EMBEDDED
+);
 var availabilityWindowSchema = new Schema(
   {
     day: { type: Number, enum: [0, 1, 2, 3, 4, 5, 6], required: true },
@@ -835,6 +987,8 @@ var taskSchema = new Schema(
     keywords: { type: [String], required: true, default: [] },
     searchQueries: { type: [String], required: true, default: [] },
     criteria: { type: taskCriteriaSchema, required: true },
+    target: { type: taskTargetSchema },
+    captureWindow: { type: captureWindowSchema },
     status: { type: String, enum: TASK_STATUS_VALUES, required: true, default: "draft" },
     approvedBy: clerkUserIdOptional,
     approvedAt: { type: Date },
@@ -885,6 +1039,8 @@ var hoursLogSchema = new Schema(
     minutes: { type: Number, required: true, min: 1, validate: Number.isInteger },
     note: { type: String, required: true, default: "" },
     date: { type: Date, required: true },
+    assignmentId: { type: ObjectId },
+    source: { type: String, enum: HOURS_LOG_SOURCE_VALUES },
     deletedAt: { type: Date }
   },
   collectionOptions(COLLECTIONS.hoursLogs)
@@ -1119,6 +1275,7 @@ var agentSessionSchema = new Schema(
     lastSeq: counter,
     stats: { type: agentSessionStatsSchema, required: true, default: () => ({}) },
     budget: { type: agentSessionBudgetSchema, required: true, default: () => ({}) },
+    captureWindow: { type: coverageWindowSchema },
     state: { type: Mixed }
   },
   collectionOptions(COLLECTIONS.agentSessions)
@@ -1171,7 +1328,11 @@ var captureDomSummarySchema = new Schema(
 var captureSchema = new Schema(
   {
     orgId: clerkOrgId,
-    sessionId: { type: ObjectId, required: true },
+    sessionId: { type: ObjectId },
+    assignmentId: { type: ObjectId },
+    taskId: { type: ObjectId },
+    missionId: { type: ObjectId },
+    userId: clerkUserIdOptional,
     url: { type: String, required: true },
     title: { type: String, required: true, default: "", maxlength: 500 },
     capturedAt: { type: Date, required: true },
@@ -1184,6 +1345,8 @@ var captureSchema = new Schema(
     status: { type: String, enum: CAPTURE_STATUS_VALUES, required: true, default: "uploaded" },
     analysis: { type: Mixed },
     error: { type: String, maxlength: 1e3 },
+    attempts: { type: Number, required: true, default: 0, min: 0, validate: Number.isInteger },
+    claimedAt: { type: Date },
     expiresAt: { type: Date, required: true }
   },
   collectionOptions(COLLECTIONS.captures)
@@ -1294,6 +1457,6 @@ function registerModels(conn) {
   };
 }
 
-export { COLLECTIONS, INDEXES, MODEL_NAMES, NOT_SOFT_DELETED_FILTER, SCHEMAS, TASK_VERSION_HOOK_OPS, agentMessageSchema, agentSessionBudgetSchema, agentSessionSchema, agentSessionStatsSchema, analystProfileSchema, applyIndexes, assigneeLoadSchema, assignmentSchema, auditLogSchema, auditTargetSchema, availabilityWindowSchema, blobRefSchema, captureDomSummarySchema, captureRegionSchema, captureSchema, collectorSchema, coverageWindowSchema, dataItemContentSchema, dataItemEntitySchema, dataItemSchema, databaseSchema, deploymentSchema, deploymentStepSchema, discoveredBySchema, geoSchema, hoursLogSchema, incrementTaskVersion, llmCallSchema, llmUsageSchema, missionSchema, objectIdsToStrings, orgSettingsSchema, platformAccessSchema, ratingSummarySchema, registerModels, scoutProfileSchema, scoutRatingSchema, scraperMetricSchema, scraperSchema, sourceImportRowErrorSchema, sourceImportSchema, sourceImportSummarySchema, sourceRatingSchema, sourceSchema, specialAccessSchema, systemRecommendationSchema, taskCriteriaSchema, taskSchema, toJSONOptions };
+export { COLLECTIONS, INDEXES, MODEL_NAMES, NOT_SOFT_DELETED_FILTER, SCHEMAS, TASK_VERSION_HOOK_OPS, agentMessageSchema, agentSessionBudgetSchema, agentSessionSchema, agentSessionStatsSchema, analystProfileSchema, applyIndexes, assigneeLoadSchema, assignmentSchema, auditLogSchema, auditTargetSchema, availabilityWindowSchema, blobRefSchema, captureDomSummarySchema, captureRegionSchema, captureSchema, captureWindowSchema, collectorSchema, coverageWindowSchema, dataItemContentSchema, dataItemEntitySchema, dataItemSchema, databaseSchema, deploymentSchema, deploymentStepSchema, discoveredBySchema, geoSchema, hoursLogSchema, incrementTaskVersion, llmCallSchema, llmUsageSchema, missionSchema, objectIdsToStrings, orgSettingsSchema, platformAccessSchema, ratingSummarySchema, registerModels, scoutProfileSchema, scoutRatingSchema, scraperMetricSchema, scraperSchema, sourceImportRowErrorSchema, sourceImportSchema, sourceImportSummarySchema, sourceRatingSchema, sourceSchema, specialAccessSchema, systemRecommendationSchema, taskCriteriaSchema, taskSchema, taskTargetSchema, toJSONOptions };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map