@canonry/canonry 4.155.0 → 4.157.0

package/dist/{chunk-VIKPFI6F.js → chunk-TRRVBTU5.js}

@@ -8545,6 +8545,9 @@ var siteCrawlSummarySchema = z24.object({
   legacyAuditAvailable: z24.boolean(),
   runId: z24.string().nullable(),
   runStatus: nullableRunStatusSchema,
+  /** Root URL originally requested by the operator; null for legacy snapshots. */
+  requestedRootUrl: z24.string().nullable(),
+  /** Effective root followed by the crawl after any supported host redirect. */
   rootUrl: z24.string().nullable(),
   crawlSchemaVersion: z24.string().nullable().optional(),
   engineVersion: z24.string().nullable().optional(),
@@ -8558,6 +8561,65 @@ var siteCrawlSummarySchema = z24.object({
   counts: siteCrawlCountsSchema,
   deadLinks: siteCrawlDeadLinksSummarySchema
 });
+var siteHealthStateSchema = z24.enum(["eligible", "hidden", "failed", "unchecked"]);
+var SiteCrawlFetchStates = {
+  discovered: "discovered",
+  robotsBlocked: "robots-blocked",
+  html: "html",
+  redirect: "redirect",
+  nonHtml: "non-html",
+  fetchError: "fetch-error"
+};
+var SiteCrawlFetchedStates = [
+  SiteCrawlFetchStates.html,
+  SiteCrawlFetchStates.redirect,
+  SiteCrawlFetchStates.nonHtml,
+  SiteCrawlFetchStates.fetchError
+];
+var SiteCrawlIndexabilityStates = {
+  indexable: "indexable",
+  noindex: "noindex",
+  blocked: "blocked",
+  unknown: "unknown"
+};
+var SiteCrawlIndexabilityReasons = {
+  robotsDisallow: "robots-disallow",
+  redirectTerminal: "redirect-terminal",
+  metaRobotsNoindex: "meta-robots-noindex",
+  xRobotsNoindex: "x-robots-noindex",
+  canonicalToOther: "canonical-to-other",
+  notHtmlOrUnavailable: "not-html-or-unavailable"
+};
+function pointsToOtherCanonical(input) {
+  return input.indexabilityReasons?.some((reason) => reason === SiteCrawlIndexabilityReasons.canonicalToOther) === true || input.nodeKey !== void 0 && input.canonicalNodeKey !== null && input.canonicalNodeKey !== void 0 && input.canonicalNodeKey !== input.nodeKey;
+}
+function deriveSiteHealthState(input) {
+  switch (input.fetchState) {
+    case SiteCrawlFetchStates.fetchError:
+      return "failed";
+    case SiteCrawlFetchStates.discovered:
+      return "unchecked";
+    case SiteCrawlFetchStates.redirect:
+    case SiteCrawlFetchStates.robotsBlocked:
+    case SiteCrawlFetchStates.nonHtml:
+      return "hidden";
+    case SiteCrawlFetchStates.html:
+      break;
+    default:
+      return "unchecked";
+  }
+  if (pointsToOtherCanonical(input)) return "hidden";
+  switch (input.indexabilityState) {
+    case SiteCrawlIndexabilityStates.indexable:
+      return "eligible";
+    case SiteCrawlIndexabilityStates.noindex:
+    case SiteCrawlIndexabilityStates.blocked:
+      return "hidden";
+    case SiteCrawlIndexabilityStates.unknown:
+    default:
+      return "unchecked";
+  }
+}
 var siteCrawlPageSchema = z24.object({
   nodeKey: z24.string(),
   url: z24.string(),
@@ -8579,8 +8641,62 @@ var siteCrawlPageSchema = z24.object({
   inboundOccurrences: z24.number().int().nonnegative(),
   outboundOccurrences: z24.number().int().nonnegative(),
   linkScoreRaw: z24.number().nullable(),
-  linkScoreNormalized: z24.number().nullable()
+  linkScoreNormalized: z24.number().nullable(),
+  healthState: siteHealthStateSchema
 });
+var siteCrawlAuditFindingSchema = z24.object({
+  type: z24.enum(["found", "missing", "info", "timeout", "unreachable"]),
+  /** Stable machine code; UI and agents must not key behavior off `message`. */
+  code: z24.string().min(1),
+  message: z24.string().min(1)
+});
+var siteCrawlAuditFactorSchema = siteAuditPageFactorSchema.extend({
+  status: siteAuditFactorStatusSchema,
+  /** `null` means the engine did not explicitly classify applicability. */
+  applicable: z24.boolean().nullable(),
+  findings: z24.array(siteCrawlAuditFindingSchema).default([]),
+  recommendations: z24.array(z24.string()).default([])
+});
+var siteCrawlCriticalDefectSchema = z24.object({
+  id: z24.string().min(1),
+  severity: z24.enum(["critical", "warning"]),
+  detail: z24.string().min(1),
+  recommendation: z24.string().min(1)
+});
+var siteCrawlPageAuditProvenanceSchema = z24.object({
+  project: z24.string(),
+  runId: z24.string(),
+  complete: z24.boolean(),
+  termination: z24.string().nullable()
+});
+var siteCrawlPageAuditIdentitySchema = z24.object({
+  nodeKey: z24.string(),
+  url: z24.string(),
+  auditState: z24.string()
+});
+var siteCrawlPageAuditSchema = z24.discriminatedUnion("state", [
+  z24.object({
+    state: z24.literal("no-crawl"),
+    project: z24.string(),
+    runId: z24.null()
+  }),
+  siteCrawlPageAuditProvenanceSchema.extend({ state: z24.literal("details-unavailable") }),
+  siteCrawlPageAuditProvenanceSchema.extend({ state: z24.literal("not-found") }),
+  siteCrawlPageAuditProvenanceSchema.merge(siteCrawlPageAuditIdentitySchema).extend({
+    state: z24.literal("not-audited"),
+    auditScore: z24.null(),
+    factors: z24.array(siteCrawlAuditFactorSchema).length(0).default([]),
+    criticalDefects: z24.array(siteCrawlCriticalDefectSchema).length(0).default([])
+  }),
+  siteCrawlPageAuditProvenanceSchema.merge(siteCrawlPageAuditIdentitySchema).extend({
+    state: z24.literal("ready"),
+    auditScore: z24.number(),
+    /** Old crawl rows retain factor scores but did not persist evidence prose. */
+    evidenceState: z24.enum(["complete", "scores-only"]),
+    factors: z24.array(siteCrawlAuditFactorSchema).default([]),
+    criticalDefects: z24.array(siteCrawlCriticalDefectSchema).default([])
+  })
+]);
 var siteCrawlPagesResponseSchema = z24.object({
   project: z24.string(),
   hasCrawlData: z24.boolean(),
@@ -8620,6 +8736,194 @@ var siteCrawlEdgeSchema = z24.object({
   nofollowOccurrences: z24.number().int().nonnegative(),
   anchors: z24.array(z24.string()).default([])
 });
+var siteCrawlGraphNodeSchema = z24.object({
+  nodeKey: z24.string(),
+  url: z24.string(),
+  path: z24.string(),
+  depth: z24.number().int().nonnegative().nullable(),
+  indexabilityState: z24.string(),
+  fetchState: z24.string(),
+  auditState: z24.string(),
+  auditScore: z24.number().nullable(),
+  inventoryEligible: z24.boolean(),
+  inboundUniqueEdges: z24.number().int().nonnegative(),
+  outboundUniqueEdges: z24.number().int().nonnegative(),
+  linkScoreNormalized: z24.number().nullable(),
+  healthState: siteHealthStateSchema,
+  /** Publish-time ForceAtlas2 coordinate. Reads never run layout physics. */
+  x: z24.number(),
+  /** Publish-time ForceAtlas2 coordinate. Reads never run layout physics. */
+  y: z24.number()
+});
+var siteCrawlGraphEdgeSchema = z24.object({
+  edgeKey: z24.string(),
+  sourceNodeKey: z24.string(),
+  targetNodeKey: z24.string(),
+  followable: z24.boolean(),
+  occurrences: z24.number().int().positive()
+});
+var siteCrawlGraphLayoutSchema = z24.discriminatedUnion("state", [
+  z24.object({
+    state: z24.literal("ready"),
+    version: z24.string().min(1),
+    computedAt: z24.string().min(1)
+  }),
+  z24.object({
+    state: z24.literal("unavailable"),
+    version: z24.null(),
+    reason: z24.enum(["no-crawl", "legacy-snapshot", "details-unavailable", "layout-failed", "empty-crawl"])
+  })
+]);
+var siteCrawlGraphResponseSchema = z24.object({
+  project: z24.string(),
+  hasCrawlData: z24.boolean(),
+  runId: z24.string().nullable(),
+  layout: siteCrawlGraphLayoutSchema,
+  /** Total canonical pages in the persisted crawl before graph sampling. */
+  totalNodes: z24.number().int().nonnegative(),
+  /** Total graph-compatible internal anchor edges before graph sampling. */
+  totalEdges: z24.number().int().nonnegative(),
+  nodes: z24.array(siteCrawlGraphNodeSchema).default([]),
+  edges: z24.array(siteCrawlGraphEdgeSchema).default([]),
+  omittedNodes: z24.number().int().nonnegative(),
+  omittedEdges: z24.number().int().nonnegative(),
+  /** True when publish/read caps or a layout failure omit graph rows. */
+  sampled: z24.boolean()
+});
+var SITE_CRAWL_GRAPH_DEFAULT_MAX_NODES = 2e4;
+var SITE_CRAWL_GRAPH_MAX_NODES = 2e4;
+var SITE_CRAWL_GRAPH_DEFAULT_MAX_EDGES = 5e4;
+var SITE_CRAWL_GRAPH_MAX_EDGES = 5e4;
+var siteHealthSubgraphRelationSchema = z24.enum([
+  "focus",
+  "inbound",
+  "outbound",
+  "both",
+  "transitive"
+]);
+var siteHealthSubgraphNodeSchema = siteCrawlPageSchema.extend({
+  distance: z24.number().int().nonnegative(),
+  relationToFocus: siteHealthSubgraphRelationSchema
+});
+var siteHealthSubgraphResponseSchema = z24.object({
+  project: z24.string(),
+  hasCrawlData: z24.boolean(),
+  runId: z24.string().nullable(),
+  /** Selected snapshot provenance. Historical partial crawls stay explicit. */
+  complete: z24.boolean(),
+  termination: z24.string().nullable(),
+  state: z24.enum(["no-crawl", "details-unavailable", "ready"]),
+  focusNodeKey: z24.string().nullable(),
+  focusUrl: z24.string().nullable(),
+  hops: z24.number().int().min(0).max(3),
+  totalNodes: z24.number().int().nonnegative(),
+  totalEdges: z24.number().int().nonnegative(),
+  nodes: z24.array(siteHealthSubgraphNodeSchema).default([]),
+  edges: z24.array(siteCrawlEdgeSchema).default([]),
+  omittedNodes: z24.number().int().nonnegative(),
+  omittedEdges: z24.number().int().nonnegative(),
+  /** `lower-bound` means totals and omissions are minimums because traversal hit a cap. */
+  countAccuracy: z24.enum(["exact", "lower-bound"]),
+  truncated: z24.boolean()
+});
+var SITE_HEALTH_SUBGRAPH_DEFAULT_MAX_NODES = 25;
+var SITE_HEALTH_SUBGRAPH_MAX_NODES = 200;
+var SITE_HEALTH_SUBGRAPH_DEFAULT_MAX_EDGES = 50;
+var SITE_HEALTH_SUBGRAPH_MAX_EDGES = 500;
+var siteHealthNodeReferenceSchema = z24.object({
+  nodeKey: z24.string(),
+  url: z24.string(),
+  path: z24.string()
+});
+var siteHealthPathResponseSchema = z24.object({
+  project: z24.string(),
+  runId: z24.string().nullable(),
+  /** Selected snapshot provenance. `unreachable` on a partial crawl is not a site-wide claim. */
+  complete: z24.boolean(),
+  termination: z24.string().nullable(),
+  state: z24.enum(["no-crawl", "details-unavailable", "found", "unreachable", "truncated"]),
+  from: siteHealthNodeReferenceSchema.nullable(),
+  to: siteHealthNodeReferenceSchema.nullable(),
+  maxDepth: z24.number().int().positive(),
+  visitedNodes: z24.number().int().nonnegative(),
+  nodes: z24.array(siteCrawlPageSchema).default([]),
+  edges: z24.array(siteCrawlEdgeSchema).default([])
+});
+var SITE_HEALTH_PATH_DEFAULT_MAX_DEPTH = 12;
+var SITE_HEALTH_PATH_MAX_DEPTH = 24;
+var SITE_HEALTH_PATH_MAX_VISITED_NODES = 5e3;
+var siteHealthChangeKindSchema = z24.enum(["added", "removed", "changed"]);
+var siteHealthPageChangeSchema = z24.object({
+  entity: z24.literal("page"),
+  change: siteHealthChangeKindSchema,
+  key: z24.string(),
+  changedFields: z24.array(z24.string()).default([]),
+  before: siteCrawlPageSchema.nullable(),
+  after: siteCrawlPageSchema.nullable()
+});
+var siteHealthLinkChangeSchema = z24.object({
+  entity: z24.literal("link"),
+  change: siteHealthChangeKindSchema,
+  key: z24.string(),
+  changedFields: z24.array(z24.string()).default([]),
+  before: siteCrawlEdgeSchema.nullable(),
+  after: siteCrawlEdgeSchema.nullable()
+});
+var siteHealthChangeRecordSchema = z24.discriminatedUnion("entity", [
+  siteHealthPageChangeSchema,
+  siteHealthLinkChangeSchema
+]);
+var siteHealthChangeCountsSchema = z24.object({
+  added: z24.number().int().nonnegative(),
+  removed: z24.number().int().nonnegative(),
+  changed: z24.number().int().nonnegative()
+});
+var siteHealthChangesFiltersSchema = z24.object({
+  scope: z24.enum(["all", "pages", "links"]),
+  change: z24.enum(["all", "added", "removed", "changed"])
+});
+var siteHealthChangesResponseSchema = z24.discriminatedUnion("state", [
+  z24.object({
+    project: z24.string(),
+    state: z24.literal("unavailable"),
+    reason: z24.enum(["no-crawl", "insufficient-history", "details-unavailable", "partial-not-comparable"]),
+    fromRunId: z24.string().nullable(),
+    toRunId: z24.string().nullable()
+  }),
+  z24.object({
+    project: z24.string(),
+    state: z24.literal("incompatible"),
+    reason: z24.literal("incompatible-versions"),
+    fromRunId: z24.string(),
+    toRunId: z24.string(),
+    mismatchedVersions: z24.array(z24.string()).min(1)
+  }),
+  z24.object({
+    project: z24.string(),
+    state: z24.literal("ready"),
+    fromRunId: z24.string(),
+    toRunId: z24.string(),
+    versions: z24.object({
+      crawlSchema: z24.string(),
+      normalization: z24.string(),
+      indexability: z24.string(),
+      linkScore: z24.string()
+    }),
+    /** The filters whose post-filter counts and records this page represents. */
+    filters: siteHealthChangesFiltersSchema,
+    /** Continuations omit summary work; reuse the exact summary from the first page. */
+    summaryState: z24.enum(["exact", "omitted-on-continuation"]),
+    summary: z24.object({
+      pages: siteHealthChangeCountsSchema,
+      links: siteHealthChangeCountsSchema
+    }).nullable(),
+    total: z24.number().int().nonnegative().nullable(),
+    nextCursor: z24.string().nullable(),
+    changes: z24.array(siteHealthChangeRecordSchema).default([])
+  })
+]);
+var SITE_HEALTH_CHANGES_DEFAULT_LIMIT = 25;
+var SITE_HEALTH_CHANGES_MAX_LIMIT = 100;
 var siteCrawlInternalLinksResponseSchema = z24.object({
   project: z24.string(),
   hasCrawlData: z24.boolean(),
@@ -12441,6 +12745,71 @@ var trafficConnectVercelRequestSchema = z44.object({
   environment: vercelTrafficEnvironmentSchema.optional(),
   displayName: z44.string().min(1).optional()
 });
+var cloudflareTrafficDeliveryModeSchema = z44.enum(["direct-push", "queue-pull"]);
+var CloudflareTrafficDeliveryModes = cloudflareTrafficDeliveryModeSchema.enum;
+var cloudflareTrafficSourceConfigSchema = z44.object({
+  schemaVersion: z44.literal(1),
+  deliveryMode: z44.literal(CloudflareTrafficDeliveryModes["direct-push"]).default("direct-push"),
+  /** Semver of the Worker script bundle that was generated at connect/rotate time. */
+  workerVersion: z44.string().min(1),
+  /** Identifier of the bot/referer keyword set baked into the deployed Worker. */
+  expectedBotListVersion: z44.string().min(1),
+  /** Target Cloudflare zone for the operator-managed Worker route. */
+  zoneId: z44.string().nullable(),
+  /** Cloudflare account id used to target the correct account during Wrangler deploy. */
+  accountId: z44.string().nullable()
+});
+var trafficConnectCloudflareRequestSchema = z44.object({
+  deliveryMode: z44.literal(CloudflareTrafficDeliveryModes["direct-push"]).default("direct-push"),
+  displayName: z44.string().min(1).optional(),
+  /** Target Cloudflare zone. Canonry does not validate it against Cloudflare. */
+  zoneId: z44.string().min(1).optional(),
+  /** Cloudflare account id. Set it when Wrangler can access more than one account. */
+  accountId: z44.string().min(1).optional()
+});
+var trafficConnectCloudflareResponseSchema = z44.object({
+  sourceId: z44.string().min(1),
+  deliveryMode: z44.literal(CloudflareTrafficDeliveryModes["direct-push"]),
+  workerScript: z44.string().min(1),
+  wranglerToml: z44.string().min(1),
+  workerVersion: z44.string().min(1),
+  instructions: z44.string().min(1)
+});
+var cloudflareEdgeEventSchema = z44.object({
+  /** Cloudflare `cf-ray` request id — globally unique per request. */
+  eventId: z44.string().min(1),
+  observedAt: z44.string().datetime(),
+  method: z44.string().nullable(),
+  host: z44.string().nullable(),
+  path: z44.string().min(1),
+  queryString: z44.string().nullable(),
+  status: z44.number().int().nullable(),
+  userAgent: z44.string().nullable(),
+  remoteIp: z44.string().nullable(),
+  referer: z44.string().nullable(),
+  cf: z44.object({
+    verifiedBot: z44.boolean().nullable(),
+    botScore: z44.number().int().nullable(),
+    country: z44.string().nullable(),
+    asn: z44.number().int().nullable(),
+    asOrganization: z44.string().nullable()
+  }).nullable()
+});
+var cloudflareEdgeEventBatchSchema = z44.object({
+  schemaVersion: z44.literal(1),
+  workerVersion: z44.string().min(1),
+  events: z44.array(cloudflareEdgeEventSchema).min(1).max(100)
+});
+var cloudflareWorkerIngestRequestSchema = cloudflareEdgeEventBatchSchema;
+var cloudflareWorkerIngestResponseSchema = z44.object({
+  acceptedEvents: z44.number().int(),
+  droppedEvents: z44.number().int(),
+  workerVersionAck: z44.string(),
+  crawlerBucketRows: z44.number().int(),
+  aiUserFetchBucketRows: z44.number().int(),
+  aiReferralBucketRows: z44.number().int(),
+  sampleRows: z44.number().int()
+});
 var trafficSyncResponseSchema = z44.object({
   sourceId: z44.string(),
   runId: z44.string(),
@@ -14326,11 +14695,34 @@ export {
   factorStatusFromScore,
   SiteAuditTrendDirections,
   siteAuditScoreSchema,
+  siteAuditPageFactorSchema,
   siteAuditPagesResponseSchema,
   siteAuditTrendResponseSchema,
   siteCrawlSummarySchema,
+  SiteCrawlFetchedStates,
+  deriveSiteHealthState,
+  siteCrawlAuditFactorSchema,
+  siteCrawlCriticalDefectSchema,
+  siteCrawlPageAuditSchema,
   siteCrawlPagesResponseSchema,
   siteCrawlStructureResponseSchema,
+  siteCrawlGraphResponseSchema,
+  SITE_CRAWL_GRAPH_DEFAULT_MAX_NODES,
+  SITE_CRAWL_GRAPH_MAX_NODES,
+  SITE_CRAWL_GRAPH_DEFAULT_MAX_EDGES,
+  SITE_CRAWL_GRAPH_MAX_EDGES,
+  siteHealthSubgraphResponseSchema,
+  SITE_HEALTH_SUBGRAPH_DEFAULT_MAX_NODES,
+  SITE_HEALTH_SUBGRAPH_MAX_NODES,
+  SITE_HEALTH_SUBGRAPH_DEFAULT_MAX_EDGES,
+  SITE_HEALTH_SUBGRAPH_MAX_EDGES,
+  siteHealthPathResponseSchema,
+  SITE_HEALTH_PATH_DEFAULT_MAX_DEPTH,
+  SITE_HEALTH_PATH_MAX_DEPTH,
+  SITE_HEALTH_PATH_MAX_VISITED_NODES,
+  siteHealthChangesResponseSchema,
+  SITE_HEALTH_CHANGES_DEFAULT_LIMIT,
+  SITE_HEALTH_CHANGES_MAX_LIMIT,
   siteCrawlInternalLinksResponseSchema,
   siteCrawlNeighborsResponseSchema,
   siteCrawlDeadLinksResponseSchema,
@@ -14485,6 +14877,12 @@ export {
   trafficConnectCloudRunRequestSchema,
   trafficConnectWordpressRequestSchema,
   trafficConnectVercelRequestSchema,
+  CloudflareTrafficDeliveryModes,
+  cloudflareTrafficSourceConfigSchema,
+  trafficConnectCloudflareRequestSchema,
+  trafficConnectCloudflareResponseSchema,
+  cloudflareWorkerIngestRequestSchema,
+  cloudflareWorkerIngestResponseSchema,
   trafficSyncResponseSchema,
   trafficResetRequestSchema,
   trafficBackfillResponseSchema,