@ainyc/canonry 4.47.0 → 4.51.0

package/dist/{chunk-M7MSNUNQ.js → chunk-DLDLDWH4.js}

@@ -2,6 +2,7 @@ import {
   CitationStates,
   ContentActions,
   RunKinds,
+  RunTriggers,
   __export,
   brandKeyFromText,
   brandLabelFromDomain,
@@ -9,10 +10,10 @@ import {
   categoryLabel,
   determineAnswerMentioned,
   normalizeProjectDomain
-} from "./chunk-4WXY57ET.js";
+} from "./chunk-2ARCCG5E.js";
 
 // src/intelligence-service.ts
-import { eq, desc, asc, and, or, inArray } from "drizzle-orm";
+import { eq, desc, asc, and, ne, or, inArray } from "drizzle-orm";
 
 // ../db/src/client.ts
 import { mkdirSync } from "fs";
@@ -36,6 +37,7 @@ __export(schema_exports, {
   bingUrlInspections: () => bingUrlInspections,
   ccReleaseSyncs: () => ccReleaseSyncs,
   competitors: () => competitors,
+  contentTargetDismissals: () => contentTargetDismissals,
   crawlerEventsHourly: () => crawlerEventsHourly,
   discoveryProbes: () => discoveryProbes,
   discoverySessions: () => discoverySessions,
@@ -57,6 +59,7 @@ __export(schema_exports, {
   queries: () => queries,
   querySnapshots: () => querySnapshots,
   rawEventSamples: () => rawEventSamples,
+  recommendationExplanations: () => recommendationExplanations,
   runs: () => runs,
   schedules: () => schedules,
   trafficSources: () => trafficSources,
@@ -68,16 +71,16 @@ var projects = sqliteTable("projects", {
   name: text("name").notNull().unique(),
   displayName: text("display_name").notNull(),
   canonicalDomain: text("canonical_domain").notNull(),
-  ownedDomains: text("owned_domains").notNull().default("[]"),
-  aliases: text("aliases").notNull().default("[]"),
+  ownedDomains: text("owned_domains", { mode: "json" }).$type().notNull().default([]),
+  aliases: text("aliases", { mode: "json" }).$type().notNull().default([]),
   country: text("country").notNull(),
   language: text("language").notNull(),
-  tags: text("tags").notNull().default("[]"),
-  labels: text("labels").notNull().default("{}"),
-  providers: text("providers").notNull().default("[]"),
-  locations: text("locations").notNull().default("[]"),
+  tags: text("tags", { mode: "json" }).$type().notNull().default([]),
+  labels: text("labels", { mode: "json" }).$type().notNull().default({}),
+  providers: text("providers", { mode: "json" }).$type().notNull().default([]),
+  locations: text("locations", { mode: "json" }).$type().notNull().default([]),
   defaultLocation: text("default_location"),
-  autoExtractBacklinks: integer("auto_extract_backlinks").notNull().default(0),
+  autoExtractBacklinks: integer("auto_extract_backlinks", { mode: "boolean" }).notNull().default(false),
   configSource: text("config_source").notNull().default("cli"),
   configRevision: integer("config_revision").notNull().default(1),
   icpDescription: text("icp_description"),
@@ -111,7 +114,7 @@ var runs = sqliteTable("runs", {
   status: text("status").notNull().default("queued"),
   trigger: text("trigger").notNull().default("manual"),
   location: text("location"),
-  queries: text("queries"),
+  queries: text("queries", { mode: "json" }).$type(),
   sourceId: text("source_id"),
   startedAt: text("started_at"),
   finishedAt: text("finished_at"),
@@ -138,9 +141,9 @@ var querySnapshots = sqliteTable("query_snapshots", {
   citationState: text("citation_state").notNull(),
   answerMentioned: integer("answer_mentioned", { mode: "boolean" }),
   answerText: text("answer_text"),
-  citedDomains: text("cited_domains").notNull().default("[]"),
-  competitorOverlap: text("competitor_overlap").notNull().default("[]"),
-  recommendedCompetitors: text("recommended_competitors").notNull().default("[]"),
+  citedDomains: text("cited_domains", { mode: "json" }).$type().notNull().default([]),
+  competitorOverlap: text("competitor_overlap", { mode: "json" }).$type().notNull().default([]),
+  recommendedCompetitors: text("recommended_competitors", { mode: "json" }).$type().notNull().default([]),
   location: text("location"),
   screenshotPath: text("screenshot_path"),
   rawResponse: text("raw_response"),
@@ -176,7 +179,7 @@ var apiKeys = sqliteTable("api_keys", {
   name: text("name").notNull(),
   keyHash: text("key_hash").notNull().unique(),
   keyPrefix: text("key_prefix").notNull(),
-  scopes: text("scopes").notNull().default('["*"]'),
+  scopes: text("scopes", { mode: "json" }).$type().notNull().default(["*"]),
   createdAt: text("created_at").notNull(),
   lastUsedAt: text("last_used_at"),
   revokedAt: text("revoked_at")
@@ -194,8 +197,8 @@ var schedules = sqliteTable("schedules", {
   cronExpr: text("cron_expr").notNull(),
   preset: text("preset"),
   timezone: text("timezone").notNull().default("UTC"),
-  enabled: integer("enabled").notNull().default(1),
-  providers: text("providers").notNull().default("[]"),
+  enabled: integer("enabled", { mode: "boolean" }).notNull().default(true),
+  providers: text("providers", { mode: "json" }).$type().notNull().default([]),
   /** Optional traffic-source UUID for traffic-sync schedules. Null for other kinds. */
   sourceId: text("source_id"),
   lastRunAt: text("last_run_at"),
@@ -209,9 +212,9 @@ var notifications = sqliteTable("notifications", {
   id: text("id").primaryKey(),
   projectId: text("project_id").notNull().references(() => projects.id, { onDelete: "cascade" }),
   channel: text("channel").notNull(),
-  config: text("config").notNull(),
+  config: text("config", { mode: "json" }).$type().notNull(),
   webhookSecret: text("webhook_secret"),
-  enabled: integer("enabled").notNull().default(1),
+  enabled: integer("enabled", { mode: "boolean" }).notNull().default(true),
   createdAt: text("created_at").notNull(),
   updatedAt: text("updated_at").notNull()
 }, (table) => [
@@ -223,7 +226,7 @@ var googleConnections = sqliteTable("google_connections", {
   connectionType: text("connection_type").notNull(),
   propertyId: text("property_id"),
   sitemapUrl: text("sitemap_url"),
-  scopes: text("scopes").notNull().default("[]"),
+  scopes: text("scopes", { mode: "json" }).$type().notNull().default([]),
   createdAt: text("created_at").notNull(),
   updatedAt: text("updated_at").notNull()
 }, (table) => [
@@ -260,9 +263,9 @@ var gscUrlInspections = sqliteTable("gsc_url_inspections", {
   robotsTxtState: text("robots_txt_state"),
   crawlTime: text("crawl_time"),
   lastCrawlResult: text("last_crawl_result"),
-  isMobileFriendly: integer("is_mobile_friendly"),
-  richResults: text("rich_results").notNull().default("[]"),
-  referringUrls: text("referring_urls").notNull().default("[]"),
+  isMobileFriendly: integer("is_mobile_friendly", { mode: "boolean" }),
+  richResults: text("rich_results", { mode: "json" }).$type().notNull().default([]),
+  referringUrls: text("referring_urls", { mode: "json" }).$type().notNull().default([]),
   inspectedAt: text("inspected_at").notNull(),
   createdAt: text("created_at").notNull()
 }, (table) => [
@@ -277,7 +280,7 @@ var gscCoverageSnapshots = sqliteTable("gsc_coverage_snapshots", {
   date: text("date").notNull(),
   indexed: integer("indexed").notNull().default(0),
   notIndexed: integer("not_indexed").notNull().default(0),
-  reasonBreakdown: text("reason_breakdown").notNull().default("{}"),
+  reasonBreakdown: text("reason_breakdown", { mode: "json" }).$type().notNull().default({}),
   createdAt: text("created_at").notNull()
 }, (table) => [
   index("idx_gsc_coverage_snap_project_date").on(table.projectId, table.date),
@@ -310,7 +313,7 @@ var bingUrlInspections = sqliteTable("bing_url_inspections", {
   projectId: text("project_id").notNull().references(() => projects.id, { onDelete: "cascade" }),
   url: text("url").notNull(),
   httpCode: integer("http_code"),
-  inIndex: integer("in_index"),
+  inIndex: integer("in_index", { mode: "boolean" }),
   lastCrawledDate: text("last_crawled_date"),
   inIndexDate: text("in_index_date"),
   inspectedAt: text("inspected_at").notNull(),
@@ -471,8 +474,8 @@ var insights = sqliteTable("insights", {
   title: text("title").notNull(),
   query: text("query").notNull(),
   provider: text("provider").notNull(),
-  recommendation: text("recommendation"),
-  cause: text("cause"),
+  recommendation: text("recommendation", { mode: "json" }).$type(),
+  cause: text("cause", { mode: "json" }).$type(),
   dismissed: integer("dismissed", { mode: "boolean" }).notNull().default(false),
   createdAt: text("created_at").notNull()
 }, (table) => [
@@ -488,7 +491,7 @@ var healthSnapshots = sqliteTable("health_snapshots", {
   overallCitedRate: text("overall_cited_rate").notNull(),
   totalPairs: integer("total_pairs").notNull(),
   citedPairs: integer("cited_pairs").notNull(),
-  providerBreakdown: text("provider_breakdown").notNull().default("{}"),
+  providerBreakdown: text("provider_breakdown", { mode: "json" }).$type().notNull().default({}),
   createdAt: text("created_at").notNull()
 }, (table) => [
   index("idx_health_snapshots_project").on(table.projectId),
@@ -588,9 +591,9 @@ var trafficSources = sqliteTable("traffic_sources", {
   // observed in the most recent successful sync. Bounded ring buffer used to
   // dedupe across sync runs at the boundary timestamp where lastSyncedAt
   // clamping alone leaves a small overlap window.
-  lastEventIds: text("last_event_ids"),
+  lastEventIds: text("last_event_ids", { mode: "json" }).$type(),
   archivedAt: text("archived_at"),
-  configJson: text("config_json").notNull().default("{}"),
+  configJson: text("config_json", { mode: "json" }).$type().notNull().default({}),
   createdAt: text("created_at").notNull(),
   updatedAt: text("updated_at").notNull()
 }, (table) => [
@@ -666,7 +669,7 @@ var rawEventSamples = sqliteTable("raw_event_samples", {
   pathNormalized: text("path_normalized").notNull(),
   status: integer("status"),
   refererHost: text("referer_host"),
-  classifierDetailsJson: text("classifier_details_json").notNull().default("{}"),
+  classifierDetailsJson: text("classifier_details_json", { mode: "json" }).$type().notNull().default({}),
   createdAt: text("created_at").notNull()
 }, (table) => [
   index("idx_raw_event_samples_project_ts").on(table.projectId, table.ts),
@@ -687,7 +690,7 @@ var discoverySessions = sqliteTable("discovery_sessions", {
   citedCount: integer("cited_count"),
   aspirationalCount: integer("aspirational_count"),
   wastedCount: integer("wasted_count"),
-  competitorMap: text("competitor_map").notNull().default("[]"),
+  competitorMap: text("competitor_map", { mode: "json" }).$type().notNull().default([]),
   error: text("error"),
   startedAt: text("started_at"),
   finishedAt: text("finished_at"),
@@ -703,13 +706,43 @@ var discoveryProbes = sqliteTable("discovery_probes", {
   query: text("query").notNull(),
   bucket: text("bucket"),
   citationState: text("citation_state").notNull(),
-  citedDomains: text("cited_domains").notNull().default("[]"),
+  citedDomains: text("cited_domains", { mode: "json" }).$type().notNull().default([]),
   rawResponse: text("raw_response"),
   createdAt: text("created_at").notNull()
 }, (table) => [
   index("idx_discovery_probes_session").on(table.sessionId),
   index("idx_discovery_probes_project").on(table.projectId)
 ]);
+var contentTargetDismissals = sqliteTable("content_target_dismissals", {
+  id: text("id").primaryKey(),
+  projectId: text("project_id").notNull().references(() => projects.id, { onDelete: "cascade" }),
+  targetRef: text("target_ref").notNull(),
+  addressedUrl: text("addressed_url"),
+  note: text("note"),
+  dismissedAt: text("dismissed_at").notNull()
+}, (table) => [
+  uniqueIndex("idx_content_target_dismissals_project_ref").on(table.projectId, table.targetRef),
+  index("idx_content_target_dismissals_project").on(table.projectId)
+]);
+var recommendationExplanations = sqliteTable("recommendation_explanations", {
+  id: text("id").primaryKey(),
+  projectId: text("project_id").notNull().references(() => projects.id, { onDelete: "cascade" }),
+  targetRef: text("target_ref").notNull(),
+  promptVersion: text("prompt_version").notNull(),
+  provider: text("provider").notNull(),
+  model: text("model").notNull(),
+  responseText: text("response_text").notNull(),
+  /** Estimated cost in millicents (1/100 of a cent) for audit; 0 if unknown. */
+  costMillicents: integer("cost_millicents").notNull().default(0),
+  generatedAt: text("generated_at").notNull()
+}, (table) => [
+  uniqueIndex("idx_recommendation_explanations_unique").on(
+    table.projectId,
+    table.targetRef,
+    table.promptVersion
+  ),
+  index("idx_recommendation_explanations_project").on(table.projectId)
+]);
 var migrationsTable = sqliteTable("_migrations", {
   version: integer("version").primaryKey(),
   name: text("name").notNull(),
@@ -1941,6 +1974,55 @@ var MIGRATION_VERSIONS = [
       `CREATE INDEX IF NOT EXISTS idx_audit_log_project ON audit_log(project_id)`,
       `CREATE INDEX IF NOT EXISTS idx_audit_log_created ON audit_log(created_at)`
     ]
+  },
+  {
+    version: 61,
+    name: "content-target-dismissals",
+    // Persistent per-recommendation dismissal so users can mark a content
+    // opportunity "addressed" after they ship the page. The orchestrator
+    // recomputes opportunities on every report load from live GSC / GA
+    // inventory; without persistent dismissal, a recommendation lingers
+    // until the next sync surfaces the new page (days–weeks of lag).
+    //
+    // Keyed by `(project_id, target_ref)` where `target_ref` is the stable
+    // hash that `computeTargetRef()` already produces — same value the
+    // ContentTargetRowDto exposes, so the client passes back the ref it
+    // sees.
+    statements: [
+      `CREATE TABLE IF NOT EXISTS content_target_dismissals (
+         id             TEXT PRIMARY KEY,
+         project_id     TEXT NOT NULL REFERENCES projects(id) ON DELETE CASCADE,
+         target_ref     TEXT NOT NULL,
+         addressed_url  TEXT,
+         note           TEXT,
+         dismissed_at   TEXT NOT NULL
+       )`,
+      `CREATE UNIQUE INDEX IF NOT EXISTS idx_content_target_dismissals_project_ref ON content_target_dismissals(project_id, target_ref)`,
+      `CREATE INDEX IF NOT EXISTS idx_content_target_dismissals_project ON content_target_dismissals(project_id)`
+    ]
+  },
+  {
+    version: 62,
+    name: "recommendation-explanations",
+    // LLM-generated rationale for content recommendations. Cached per
+    // (project, target_ref, prompt_version) so repeat clicks are free.
+    // Bumping the prompt version invalidates the cache forward without
+    // touching the table.
+    statements: [
+      `CREATE TABLE IF NOT EXISTS recommendation_explanations (
+         id              TEXT PRIMARY KEY,
+         project_id      TEXT NOT NULL REFERENCES projects(id) ON DELETE CASCADE,
+         target_ref      TEXT NOT NULL,
+         prompt_version  TEXT NOT NULL,
+         provider        TEXT NOT NULL,
+         model           TEXT NOT NULL,
+         response_text   TEXT NOT NULL,
+         cost_millicents INTEGER NOT NULL DEFAULT 0,
+         generated_at    TEXT NOT NULL
+       )`,
+      `CREATE UNIQUE INDEX IF NOT EXISTS idx_recommendation_explanations_unique ON recommendation_explanations(project_id, target_ref, prompt_version)`,
+      `CREATE INDEX IF NOT EXISTS idx_recommendation_explanations_project ON recommendation_explanations(project_id)`
+    ]
   }
 ];
 function isDuplicateColumnError(err) {
@@ -2635,9 +2717,15 @@ function isBlogShaped(path) {
 // ../intelligence/src/content-classifier.ts
 var SEO_STRONG_THRESHOLD = 10;
 var SEO_WEAK_THRESHOLD = 30;
+function isHomepageOnly(url) {
+  if (url === "/" || url === "") return true;
+  const stripped = url.split("?")[0].replace(/\/+$/, "");
+  return stripped === "" || stripped === "/";
+}
 function classifyContentAction(input) {
   const { ourPage, ourPageInGroundingSources, ourPageHasSchema } = input;
   if (!ourPage) return "create";
+  if (isHomepageOnly(ourPage.url)) return "create";
   if (ourPageInGroundingSources) {
     if (ourPageHasSchema === false) return "add-schema";
     return null;
@@ -2782,8 +2870,7 @@ function buildContentTargetRows(input) {
     const targetRef = computeTargetRef({
       projectId: input.projectId,
       query: cq.query,
-      action,
-      targetPage: ourPage?.url ?? null
+      action
     });
     const winningCompetitor = pickTopCompetitor(cq.competitorGroundingUrls);
     const ourBestPage = ourPage ? {
@@ -2935,7 +3022,7 @@ function pickTopCompetitor(competitors2) {
   };
 }
 function computeTargetRef(input) {
-  const key = [input.projectId, input.query, input.action, input.targetPage ?? ""].join("|");
+  const key = [input.projectId, input.query, input.action].join("|");
   let hash = 0;
   for (let i = 0; i < key.length; i++) {
     hash = (hash << 5) - hash + key.charCodeAt(i) | 0;
@@ -3875,7 +3962,12 @@ var IntelligenceService = class {
     const recentRuns = this.db.select().from(runs).where(
       and(
         eq(runs.projectId, projectId),
-        or(eq(runs.status, "completed"), eq(runs.status, "partial"))
+        or(eq(runs.status, "completed"), eq(runs.status, "partial")),
+        // Defensive: RunCoordinator already skips probes before this is
+        // called, but if a future call site invokes analyzeAndPersist
+        // directly for a probe, probes still must not pollute the
+        // intelligence window.
+        ne(runs.trigger, RunTriggers.probe)
       )
     ).orderBy(desc(runs.finishedAt), desc(runs.createdAt)).limit(HISTORY_WINDOW_RUNS).all();
     if (recentRuns.length === 0) {
@@ -4009,7 +4101,9 @@ var IntelligenceService = class {
     const allRuns = this.db.select().from(runs).where(
       and(
         eq(runs.projectId, project.id),
-        or(eq(runs.status, "completed"), eq(runs.status, "partial"))
+        or(eq(runs.status, "completed"), eq(runs.status, "partial")),
+        // Backfill must not replay probe runs as if they were real sweeps.
+        ne(runs.trigger, RunTriggers.probe)
       )
     ).orderBy(asc(runs.finishedAt)).all();
     let startIdx = 0;
@@ -4128,8 +4222,8 @@ var IntelligenceService = class {
           title: insight.title,
           query: insight.query,
           provider: insight.provider,
-          recommendation: insight.recommendation ? JSON.stringify(insight.recommendation) : null,
-          cause: insight.cause ? JSON.stringify(insight.cause) : null,
+          recommendation: insight.recommendation ?? null,
+          cause: insight.cause ?? null,
           dismissed: wasDismissed,
           createdAt: insight.createdAt
         }).run();
@@ -4141,7 +4235,7 @@ var IntelligenceService = class {
         overallCitedRate: String(result.health.overallCitedRate),
         totalPairs: result.health.totalPairs,
         citedPairs: result.health.citedPairs,
-        providerBreakdown: JSON.stringify(result.health.providerBreakdown),
+        providerBreakdown: result.health.providerBreakdown,
         createdAt: now
       }).run();
     });
@@ -4176,14 +4270,16 @@ var IntelligenceService = class {
     const projectRow = this.db.select({ locations: projects.locations }).from(projects).where(eq(projects.id, projectId)).get();
     const locationCount = Math.max(
       1,
-      parseJsonColumn(projectRow?.locations ?? null, []).length
+      (projectRow?.locations ?? []).length
     );
     const ROWS_PER_GROUP_BUDGET = Math.max(2, locationCount);
     const recentRunRows = this.db.select({ id: runs.id, createdAt: runs.createdAt }).from(runs).where(
       and(
         eq(runs.projectId, projectId),
         eq(runs.kind, RunKinds["answer-visibility"]),
-        or(eq(runs.status, "completed"), eq(runs.status, "partial"))
+        or(eq(runs.status, "completed"), eq(runs.status, "partial")),
+        // Defensive — see top of file.
+        ne(runs.trigger, RunTriggers.probe)
       )
     ).orderBy(desc(runs.createdAt), desc(runs.id)).limit((RECURRENCE_LOOKBACK_RUNS + 1) * ROWS_PER_GROUP_BUDGET).all();
     const recentGroups = groupRunsByCreatedAt(recentRunRows);
@@ -4250,8 +4346,8 @@ var IntelligenceService = class {
         orphanCount++;
         continue;
       }
-      const domains = parseJsonColumn(r.citedDomains, []);
-      const competitors2 = parseJsonColumn(r.competitorOverlap, []);
+      const domains = r.citedDomains;
+      const competitors2 = r.competitorOverlap;
       snapshots.push({
         query: resolvedQuery,
         provider: r.provider,
@@ -4311,6 +4407,8 @@ export {
   rawEventSamples,
   discoverySessions,
   discoveryProbes,
+  contentTargetDismissals,
+  recommendationExplanations,
   createClient,
   parseJsonColumn,
   extractLegacyCredentials,