npm - @ainyc/canonry - Versions diffs - 2.13.2 → 2.14.1 - Mend

@ainyc/canonry 2.13.2 → 2.14.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (11) hide show

package/assets/assets/{index-CKzK0os-.js → index-BwFUCV6e.js} +116 -116
package/assets/assets/{index-CAewPdsZ.css → index-U2SLimrz.css} +1 -1
package/assets/index.html +2 -2
package/dist/{chunk-DHMCIJMQ.js → chunk-3T64Y7GR.js} +125 -14
package/dist/{chunk-UM6RDSRJ.js → chunk-FV6PY5UE.js} +20 -1
package/dist/{chunk-XJS7NALL.js → chunk-QTS7VZXN.js} +96 -1
package/dist/cli.js +143 -15
package/dist/index.js +3 -3
package/dist/{intelligence-service-54F3NGPM.js → intelligence-service-AEI46KC5.js} +1 -1
package/dist/mcp.js +12 -1
package/package.json +6 -6

package/assets/index.html CHANGED Viewed

@@ -12,8 +12,8 @@
     <link rel="icon" type="image/png" sizes="32x32" href="./favicon-32.png" />
     <link rel="apple-touch-icon" href="./apple-touch-icon.png" />
     <title>Canonry</title>
-    <script type="module" crossorigin src="./assets/index-CKzK0os-.js"></script>
-    <link rel="stylesheet" crossorigin href="./assets/index-CAewPdsZ.css">
+    <script type="module" crossorigin src="./assets/index-BwFUCV6e.js"></script>
+    <link rel="stylesheet" crossorigin href="./assets/index-U2SLimrz.css">
   </head>
   <body>
     <div id="root"></div>

package/dist/{chunk-DHMCIJMQ.js → chunk-3T64Y7GR.js} RENAMED Viewed

@@ -39,6 +39,7 @@ import {
   locationContextSchema,
   missingDependency,
   normalizeProjectDomain,
+  normalizeUrlPath,
   notFound,
   notImplemented,
   parseRunError,
@@ -59,7 +60,7 @@ import {
   visibilityStateFromAnswerMentioned,
   windowCutoff,
   wordpressEnvSchema
-} from "./chunk-XJS7NALL.js";
+} from "./chunk-QTS7VZXN.js";
 import {
   IntelligenceService,
   agentMemory,
@@ -97,7 +98,7 @@ import {
   runs,
   schedules,
   usageCounters
-} from "./chunk-UM6RDSRJ.js";
+} from "./chunk-FV6PY5UE.js";
 // src/telemetry.ts
 import crypto from "crypto";
@@ -5371,15 +5372,14 @@ var routeCatalog = [
     method: "post",
     path: "/api/v1/backlinks/syncs",
     summary: "Queue a workspace-wide Common Crawl release sync",
-    description: "Creates a `cc_release_syncs` row and fires the sync callback. Idempotent: an existing in-flight row for the same release is returned.",
+    description: "Creates a `cc_release_syncs` row and fires the sync callback. Idempotent: an existing in-flight row for the same release is returned. When `release` is omitted, the server auto-discovers the latest available Common Crawl release.",
     tags: ["backlinks"],
     requestBody: {
-      required: true,
+      required: false,
       content: {
         "application/json": {
           schema: {
             type: "object",
-            required: ["release"],
             properties: {
               release: stringSchema
             }
@@ -5422,6 +5422,17 @@ var routeCatalog = [
       200: { description: "Cached release metadata returned." }
     }
   },
+  {
+    method: "get",
+    path: "/api/v1/backlinks/latest-release",
+    summary: "Auto-discover the latest available Common Crawl hyperlinkgraph release",
+    description: "Probes Common Crawl by HEAD-checking quarterly release slugs and returns the newest one published. The local server caches the result for ~5 minutes so repeated calls do not hammer Common Crawl.",
+    tags: ["backlinks"],
+    responses: {
+      200: { description: "Latest available release, or null when no candidate slug responded." },
+      422: { description: "Backlinks feature is not available on this deployment." }
+    }
+  },
   {
     method: "delete",
     path: "/api/v1/backlinks/cache/{release}",
@@ -8651,6 +8662,7 @@ async function ga4Routes(app, opts) {
               projectId: project.id,
               date: row.date,
               landingPage: row.landingPage,
+              landingPageNormalized: normalizeUrlPath(row.landingPage),
               sessions: row.sessions,
               organicSessions: row.organicSessions,
               users: row.users,
@@ -8775,11 +8787,11 @@ async function ga4Routes(app, opts) {
       periodEnd: gaTrafficSummaries.periodEnd
     }).from(gaTrafficSummaries).where(eq19(gaTrafficSummaries.projectId, project.id)).get();
     const rows = app.db.select({
-      landingPage: gaTrafficSnapshots.landingPage,
+      landingPage: sql5`COALESCE(${gaTrafficSnapshots.landingPageNormalized}, ${gaTrafficSnapshots.landingPage})`,
       sessions: sql5`SUM(${gaTrafficSnapshots.sessions})`,
       organicSessions: sql5`SUM(${gaTrafficSnapshots.organicSessions})`,
       users: sql5`SUM(${gaTrafficSnapshots.users})`
-    }).from(gaTrafficSnapshots).where(and8(...snapshotConditions)).groupBy(gaTrafficSnapshots.landingPage).orderBy(sql5`SUM(${gaTrafficSnapshots.sessions}) DESC`).limit(limit).all();
+    }).from(gaTrafficSnapshots).where(and8(...snapshotConditions)).groupBy(sql5`COALESCE(${gaTrafficSnapshots.landingPageNormalized}, ${gaTrafficSnapshots.landingPage})`).orderBy(sql5`SUM(${gaTrafficSnapshots.sessions}) DESC`).limit(limit).all();
     const aiReferrals = app.db.select({
       source: gaAiReferrals.source,
       medium: gaAiReferrals.medium,
@@ -9036,11 +9048,11 @@ async function ga4Routes(app, opts) {
     const project = resolveProject(app.db, request.params.name);
     requireGa4Connection(opts, project.name, project.canonicalDomain);
     const trafficPages = app.db.select({
-      landingPage: gaTrafficSnapshots.landingPage,
+      landingPage: sql5`COALESCE(${gaTrafficSnapshots.landingPageNormalized}, ${gaTrafficSnapshots.landingPage})`,
       sessions: sql5`SUM(${gaTrafficSnapshots.sessions})`,
       organicSessions: sql5`SUM(${gaTrafficSnapshots.organicSessions})`,
       users: sql5`SUM(${gaTrafficSnapshots.users})`
-    }).from(gaTrafficSnapshots).where(eq19(gaTrafficSnapshots.projectId, project.id)).groupBy(gaTrafficSnapshots.landingPage).orderBy(sql5`SUM(${gaTrafficSnapshots.sessions}) DESC`).all();
+    }).from(gaTrafficSnapshots).where(eq19(gaTrafficSnapshots.projectId, project.id)).groupBy(sql5`COALESCE(${gaTrafficSnapshots.landingPageNormalized}, ${gaTrafficSnapshots.landingPage})`).orderBy(sql5`SUM(${gaTrafficSnapshots.sessions}) DESC`).all();
     return {
       pages: trafficPages.map((r) => ({
         landingPage: r.landingPage,
@@ -10712,6 +10724,60 @@ function forwardDomain(revDomain) {
 function isValidReleaseId(id) {
   return RELEASE_ID_REGEX.test(id);
 }
+function formatReleaseId(year, quarter) {
+  return `cc-main-${year}-${quarter}`;
+}
+// ../integration-commoncrawl/src/release-discovery.ts
+var QUARTERS = [
+  "oct-nov-dec",
+  "jul-aug-sep",
+  "apr-may-jun",
+  "jan-feb-mar"
+];
+function probeCandidates(now, maxBack) {
+  const year = now.getUTCFullYear();
+  const out = [];
+  for (let y = year; y >= year - maxBack; y--) {
+    for (const q of QUARTERS) {
+      out.push({ year: y, quarter: q });
+    }
+  }
+  return out;
+}
+async function probeRelease(release, fetchImpl = fetch) {
+  const paths = ccReleasePaths(release);
+  const [vertex, edges] = await Promise.all([
+    fetchImpl(paths.vertexUrl, { method: "HEAD" }),
+    fetchImpl(paths.edgesUrl, { method: "HEAD" })
+  ]);
+  if (!vertex.ok || !edges.ok) return null;
+  return {
+    release,
+    vertexUrl: paths.vertexUrl,
+    edgesUrl: paths.edgesUrl,
+    vertexBytes: parseContentLength(vertex.headers.get("content-length")),
+    edgesBytes: parseContentLength(edges.headers.get("content-length")),
+    lastModified: vertex.headers.get("last-modified")
+  };
+}
+async function probeLatestRelease(opts = {}) {
+  const now = opts.now ?? /* @__PURE__ */ new Date();
+  const maxBack = opts.maxQuartersBack ?? 3;
+  const fetchImpl = opts.fetchImpl ?? fetch;
+  const candidates = probeCandidates(now, maxBack);
+  for (const { year, quarter } of candidates) {
+    const release = formatReleaseId(year, quarter);
+    const result = await probeRelease(release, fetchImpl);
+    if (result) return result;
+  }
+  return null;
+}
+function parseContentLength(value) {
+  if (!value) return null;
+  const n = Number.parseInt(value, 10);
+  return Number.isFinite(n) ? n : null;
+}
 // ../integration-commoncrawl/src/downloader.ts
 import { createHash } from "crypto";
@@ -10739,7 +10805,7 @@ async function downloadFile(opts) {
   if (!res.ok || !res.body) {
     throw new Error(`HTTP ${res.status} ${res.statusText} for ${opts.url}`);
   }
-  const total = parseContentLength(res.headers.get("content-length"));
+  const total = parseContentLength2(res.headers.get("content-length"));
   const hasher = createHash("sha256");
   let bytes = 0;
   const hashAndCount = new Transform({
@@ -10790,7 +10856,7 @@ async function unlinkIfExists(p) {
   } catch {
   }
 }
-function parseContentLength(value) {
+function parseContentLength2(value) {
   if (!value) return null;
   const n = Number.parseInt(value, 10);
   return Number.isFinite(n) ? n : null;
@@ -11093,8 +11159,22 @@ async function backlinksRoutes(app, opts) {
     return reply.status(200).send(result);
   });
   app.post("/backlinks/syncs", async (request, reply) => {
-    const release = request.body?.release;
-    if (!release || !isValidReleaseId(release)) {
+    let release = request.body?.release;
+    if (!release) {
+      if (!opts.discoverLatestRelease) {
+        throw validationError(
+          "No `release` provided and auto-discovery is unavailable on this deployment. Pass an explicit release id (e.g., cc-main-2026-jan-feb-mar)."
+        );
+      }
+      const discovered = await opts.discoverLatestRelease();
+      if (!discovered) {
+        throw validationError(
+          "Could not auto-discover the latest Common Crawl release. Pass an explicit `release` body parameter."
+        );
+      }
+      release = discovered.release;
+    }
+    if (!isValidReleaseId(release)) {
       throw validationError("Invalid release id. Expected form: cc-main-YYYY-{jan-feb-mar,apr-may-jun,jul-aug-sep,oct-nov-dec}");
     }
     if (!opts.getBacklinksStatus || !opts.onReleaseSyncRequested) {
@@ -11145,6 +11225,13 @@ async function backlinksRoutes(app, opts) {
     const releases = opts.listCachedReleases?.() ?? [];
     return reply.send(releases);
   });
+  app.get("/backlinks/latest-release", async (_request, reply) => {
+    if (!opts.discoverLatestRelease) {
+      throw missingDependency(BACKLINKS_UNSUPPORTED_MESSAGE);
+    }
+    const discovered = await opts.discoverLatestRelease();
+    return reply.send(discovered);
+  });
   app.delete("/backlinks/cache/:release", async (request, reply) => {
     const release = request.params.release;
     if (!isValidReleaseId(release)) {
@@ -11941,7 +12028,8 @@ async function apiRoutes(app, opts) {
       onReleaseSyncRequested: opts.onReleaseSyncRequested,
       onBacklinkExtractRequested: opts.onBacklinkExtractRequested,
       onBacklinksPruneCache: opts.onBacklinksPruneCache,
-      listCachedReleases: opts.listCachedReleases
+      listCachedReleases: opts.listCachedReleases,
+      discoverLatestRelease: opts.discoverLatestRelease
     });
     await api.register(doctorRoutes, {
       googleConnectionStore: opts.googleConnectionStore,
@@ -19112,6 +19200,28 @@ async function createServer(opts) {
     }));
     return reply.status(204).send();
   });
+  const LATEST_RELEASE_TTL_MS = 5 * 60 * 1e3;
+  let latestReleaseCache = null;
+  const discoverLatestRelease = async () => {
+    const now = Date.now();
+    if (latestReleaseCache && latestReleaseCache.expiresAt > now) {
+      return latestReleaseCache.value;
+    }
+    const probed = await probeLatestRelease().catch((err) => {
+      app.log.warn({ err }, "Common Crawl latest-release probe failed");
+      return null;
+    });
+    const value = probed ? {
+      release: probed.release,
+      vertexUrl: probed.vertexUrl,
+      edgesUrl: probed.edgesUrl,
+      vertexBytes: probed.vertexBytes,
+      edgesBytes: probed.edgesBytes,
+      lastModified: probed.lastModified
+    } : null;
+    latestReleaseCache = { value, expiresAt: now + LATEST_RELEASE_TTL_MS };
+    return value;
+  };
   await app.register(apiRoutes, {
     db: opts.db,
     routePrefix: apiPrefix,
@@ -19220,6 +19330,7 @@ async function createServer(opts) {
         };
       });
     },
+    discoverLatestRelease,
     openApiInfo: {
       title: "Canonry API",
       version: PKG_VERSION,

package/dist/{chunk-UM6RDSRJ.js → chunk-FV6PY5UE.js} RENAMED Viewed

@@ -310,6 +310,14 @@ var gaTrafficSnapshots = sqliteTable("ga_traffic_snapshots", {
   projectId: text("project_id").notNull().references(() => projects.id, { onDelete: "cascade" }),
   date: text("date").notNull(),
   landingPage: text("landing_page").notNull(),
+  /**
+   * Canonicalized form of `landingPage` produced by `normalizeUrlPath()` in
+   * `@ainyc/canonry-contracts`. Nullable so existing rows survive migration;
+   * new GA4 sync writes populate it. Per-page aggregations should
+   * `GROUP BY COALESCE(landing_page_normalized, landing_page)` so
+   * partially-backfilled state still aggregates correctly.
+   */
+  landingPageNormalized: text("landing_page_normalized"),
   sessions: integer("sessions").notNull().default(0),
   organicSessions: integer("organic_sessions").notNull().default(0),
   users: integer("users").notNull().default(0),
@@ -318,6 +326,7 @@ var gaTrafficSnapshots = sqliteTable("ga_traffic_snapshots", {
 }, (table) => [
   index("idx_ga_traffic_project_date").on(table.projectId, table.date),
   index("idx_ga_traffic_page").on(table.landingPage),
+  index("idx_ga_traffic_page_normalized").on(table.projectId, table.date, table.landingPageNormalized),
   index("idx_ga_traffic_run").on(table.syncRunId)
 ]);
 var gaAiReferrals = sqliteTable("ga_ai_referrals", {
@@ -1049,7 +1058,17 @@ var MIGRATIONS = [
      WHEN discovery_date IS NOT NULL THEN 0
      ELSE NULL
    END
-   WHERE created_at < '2026-04-22T00:00:00Z'`
+   WHERE created_at < '2026-04-22T00:00:00Z'`,
+  // v44: Canonicalized landing-page column for ga_traffic_snapshots.
+  // Populated by GA4 sync via normalizeUrlPath() in
+  // @ainyc/canonry-contracts. Nullable; existing rows are filled in by
+  // `canonry backfill normalized-paths`. Read queries should
+  // `GROUP BY COALESCE(landing_page_normalized, landing_page)` so
+  // partially-backfilled state still aggregates correctly.
+  // See plans/ai-attribution-research.md "Step 1 — data hygiene".
+  `ALTER TABLE ga_traffic_snapshots ADD COLUMN landing_page_normalized TEXT`,
+  `CREATE INDEX IF NOT EXISTS idx_ga_traffic_page_normalized
+     ON ga_traffic_snapshots(project_id, date, landing_page_normalized)`
 ];
 function isDuplicateColumnError(err) {
   if (!(err instanceof Error)) return false;

package/dist/{chunk-XJS7NALL.js → chunk-QTS7VZXN.js} RENAMED Viewed

@@ -1790,6 +1790,97 @@ function summarizeCheckResults(results) {
   return summary;
 }
+// ../contracts/src/url-normalize.ts
+var STRIP_KEYS = /* @__PURE__ */ new Set([
+  // Click identifiers
+  "fbclid",
+  "gclid",
+  "msclkid",
+  "ttclid",
+  "li_fat_id",
+  "igshid",
+  "yclid",
+  "dclid",
+  "gbraid",
+  "wbraid",
+  // Mailchimp
+  "mc_cid",
+  "mc_eid",
+  // Google Analytics linkers
+  "_ga",
+  "_gl",
+  // Google Tag Manager debug
+  "gtm_latency",
+  "gtm_debug"
+]);
+function shouldStrip(key) {
+  if (STRIP_KEYS.has(key)) return true;
+  if (key.startsWith("utm_")) return true;
+  return false;
+}
+function parseQuery(query) {
+  if (query === "") return [];
+  return query.split("&").map((pair) => {
+    const eq = pair.indexOf("=");
+    if (eq === -1) return { key: pair, value: null };
+    return { key: pair.slice(0, eq), value: pair.slice(eq + 1) };
+  });
+}
+function encodeQuery(pairs) {
+  return pairs.map((p) => p.value === null ? p.key : `${p.key}=${p.value}`).join("&");
+}
+function collapseRootIndex(path2) {
+  if (path2 === "/index.html" || path2 === "/index.php") return "/";
+  return path2;
+}
+function dropTrailingSlash(path2) {
+  if (path2.length > 1 && path2.endsWith("/")) {
+    return path2.replace(/\/+$/, "");
+  }
+  return path2;
+}
+function normalizeUrlPath(input) {
+  if (input == null) return null;
+  const trimmed = input.trim();
+  if (trimmed === "") return null;
+  if (trimmed === "(not set)") return null;
+  let pathPart;
+  let queryPart;
+  if (/^https?:\/\//i.test(trimmed)) {
+    let url;
+    try {
+      url = new URL(trimmed);
+    } catch {
+      return null;
+    }
+    pathPart = url.pathname || "/";
+    queryPart = url.search.startsWith("?") ? url.search.slice(1) : url.search;
+  } else {
+    let raw = trimmed;
+    const hashIdx = raw.indexOf("#");
+    if (hashIdx !== -1) raw = raw.slice(0, hashIdx);
+    const qIdx = raw.indexOf("?");
+    if (qIdx === -1) {
+      pathPart = raw;
+      queryPart = "";
+    } else {
+      pathPart = raw.slice(0, qIdx);
+      queryPart = raw.slice(qIdx + 1);
+    }
+  }
+  if (pathPart === "") pathPart = "/";
+  pathPart = collapseRootIndex(pathPart);
+  pathPart = dropTrailingSlash(pathPart);
+  const pairs = parseQuery(queryPart).filter((p) => !shouldStrip(p.key));
+  pairs.sort((a, b) => {
+    if (a.key < b.key) return -1;
+    if (a.key > b.key) return 1;
+    return 0;
+  });
+  if (pairs.length === 0) return pathPart;
+  return `${pathPart}?${encodeQuery(pairs)}`;
+}
 // src/client.ts
 function createApiClient() {
   const config = loadConfig();
@@ -2409,7 +2500,10 @@ var ApiClient = class {
     return this.request("POST", "/backlinks/install");
   }
   async backlinksTriggerSync(release) {
-    return this.request("POST", "/backlinks/syncs", { release });
+    return this.request("POST", "/backlinks/syncs", release ? { release } : void 0);
+  }
+  async backlinksLatestRelease() {
+    return this.request("GET", "/backlinks/latest-release");
   }
   async backlinksLatestSync() {
     return this.request("GET", "/backlinks/syncs/latest");
@@ -2519,6 +2613,7 @@ export {
   CheckScopes,
   CheckCategories,
   summarizeCheckResults,
+  normalizeUrlPath,
   createApiClient,
   ApiClient
 };

package/dist/cli.js CHANGED Viewed

@@ -17,7 +17,7 @@ import {
   setGoogleAuthConfig,
   showFirstRunNotice,
   trackEvent
-} from "./chunk-DHMCIJMQ.js";
+} from "./chunk-3T64Y7GR.js";
 import {
   CcReleaseSyncStatuses,
   CheckScopes,
@@ -37,6 +37,7 @@ import {
   getConfigPath,
   isEndpointMissing,
   loadConfig,
+  normalizeUrlPath,
   notificationEventSchema,
   printCliError,
   providerQuotaPolicySchema,
@@ -44,17 +45,18 @@ import {
   saveConfig,
   saveConfigPatch,
   usageError
-} from "./chunk-XJS7NALL.js";
+} from "./chunk-QTS7VZXN.js";
 import {
   apiKeys,
   competitors,
   createClient,
+  gaTrafficSnapshots,
   migrate,
   parseJsonColumn,
   projects,
   querySnapshots,
   runs
-} from "./chunk-UM6RDSRJ.js";
+} from "./chunk-FV6PY5UE.js";
 import "./chunk-MLKGABMK.js";
 // src/cli.ts
@@ -160,7 +162,7 @@ Usage: ${spec.usage}`, {
 }
 // src/commands/backfill.ts
-import { and, eq, inArray } from "drizzle-orm";
+import { and, eq, inArray, isNull } from "drizzle-orm";
 var SNAPSHOT_BATCH_SIZE = 500;
 async function backfillAnswerVisibilityCommand(opts) {
   const config = loadConfig();
@@ -301,8 +303,75 @@ async function backfillAnswerVisibilityCommand(opts) {
   console.log(`  Reparsed: ${reparsed}`);
   console.log(`  Errors:   ${providerErrors}`);
 }
+function backfillNormalizedPaths(db, opts) {
+  const baseConditions = [isNull(gaTrafficSnapshots.landingPageNormalized)];
+  if (opts?.projectId) {
+    baseConditions.push(eq(gaTrafficSnapshots.projectId, opts.projectId));
+  }
+  const rows = db.select({
+    id: gaTrafficSnapshots.id,
+    landingPage: gaTrafficSnapshots.landingPage
+  }).from(gaTrafficSnapshots).where(and(...baseConditions)).all();
+  let updated = 0;
+  let unchanged = 0;
+  if (rows.length > 0) {
+    db.transaction((tx) => {
+      for (const row of rows) {
+        const next = normalizeUrlPath(row.landingPage);
+        if (next === null) {
+          unchanged++;
+          continue;
+        }
+        tx.update(gaTrafficSnapshots).set({ landingPageNormalized: next }).where(eq(gaTrafficSnapshots.id, row.id)).run();
+        updated++;
+      }
+    });
+  }
+  return { examined: rows.length, updated, unchanged };
+}
+async function backfillNormalizedPathsCommand(opts) {
+  const config = loadConfig();
+  const db = createClient(config.database);
+  migrate(db);
+  const projectFilter = opts?.project?.trim();
+  let projectId;
+  if (projectFilter) {
+    const project = db.select({ id: projects.id }).from(projects).where(eq(projects.name, projectFilter)).get();
+    if (!project) {
+      const result2 = {
+        project: projectFilter,
+        examined: 0,
+        updated: 0,
+        unchanged: 0
+      };
+      if (opts?.format === "json") {
+        console.log(JSON.stringify(result2, null, 2));
+        return;
+      }
+      console.log(`Backfill normalized-paths: project "${projectFilter}" not found.`);
+      return;
+    }
+    projectId = project.id;
+  }
+  const { examined, updated, unchanged } = backfillNormalizedPaths(db, { projectId });
+  const result = {
+    project: projectFilter ?? null,
+    examined,
+    updated,
+    unchanged
+  };
+  if (opts?.format === "json") {
+    console.log(JSON.stringify(result, null, 2));
+    return;
+  }
+  console.log("Normalized-path backfill complete.\n");
+  if (projectFilter) console.log(`  Project:   ${projectFilter}`);
+  console.log(`  Examined:  ${examined}`);
+  console.log(`  Updated:   ${updated}`);
+  console.log(`  Unchanged: ${unchanged}`);
+}
 async function backfillInsightsCommand(project, opts) {
-  const { IntelligenceService } = await import("./intelligence-service-54F3NGPM.js");
+  const { IntelligenceService } = await import("./intelligence-service-AEI46KC5.js");
   const config = loadConfig();
   const db = createClient(config.database);
   migrate(db);
@@ -498,14 +567,28 @@ var BACKFILL_CLI_COMMANDS = [
       });
     }
   },
+  {
+    path: ["backfill", "normalized-paths"],
+    usage: "canonry backfill normalized-paths [--project <name>] [--format json]",
+    options: {
+      project: stringOption()
+    },
+    allowPositionals: false,
+    run: async (input) => {
+      await backfillNormalizedPathsCommand({
+        project: getString(input.values, "project"),
+        format: input.format
+      });
+    }
+  },
   {
     path: ["backfill"],
-    usage: "canonry backfill <answer-visibility|insights> [options]",
+    usage: "canonry backfill <answer-visibility|insights|normalized-paths> [options]",
     run: async (input) => {
       unknownSubcommand(input.positionals[0], {
         command: "backfill",
-        usage: "canonry backfill <answer-visibility|insights> [options]",
-        available: ["answer-visibility", "insights"]
+        usage: "canonry backfill <answer-visibility|insights|normalized-paths> [options]",
+        available: ["answer-visibility", "insights", "normalized-paths"]
       });
     }
   }
@@ -638,8 +721,38 @@ async function backlinksSync(opts) {
     return;
   }
   if (opts.wait) process.stderr.write("\n");
+  if (!opts.release) {
+    process.stderr.write(`Auto-discovered release: ${sync.release}
+`);
+  }
   console.log(formatSync(final));
 }
+function formatBytesShort(n) {
+  if (n === null) return "\u2014";
+  if (n >= 1e9) return `${(n / 1e9).toFixed(1)} GB`;
+  if (n >= 1e6) return `${(n / 1e6).toFixed(1)} MB`;
+  if (n >= 1e3) return `${(n / 1e3).toFixed(1)} KB`;
+  return `${n} B`;
+}
+function formatLatestRelease(result) {
+  if (!result) return "No release discovered (Common Crawl probe returned no candidates).";
+  const lines = [];
+  lines.push(`Release: ${result.release}`);
+  lines.push(`Vertex:  ${result.vertexUrl}`);
+  lines.push(`         ${formatBytesShort(result.vertexBytes)}`);
+  lines.push(`Edges:   ${result.edgesUrl}`);
+  lines.push(`         ${formatBytesShort(result.edgesBytes)}`);
+  if (result.lastModified) lines.push(`Last modified: ${result.lastModified}`);
+  return lines.join("\n");
+}
+async function backlinksLatestRelease(opts = {}) {
+  const result = await getClient().backlinksLatestRelease();
+  if (opts.format === "json") {
+    printJson(result);
+    return;
+  }
+  console.log(formatLatestRelease(result));
+}
 async function backlinksStatus(opts = {}) {
   const sync = await getClient().backlinksLatestSync();
   if (opts.format === "json") {
@@ -715,19 +828,14 @@ var BACKLINKS_CLI_COMMANDS = [
   },
   {
     path: ["backlinks", "sync"],
-    usage: "canonry backlinks sync --release <id> [--wait] [--format json]",
+    usage: "canonry backlinks sync [--release <id>] [--wait] [--format json]",
     options: {
       release: stringOption(),
       wait: { type: "boolean" }
     },
     run: async (input) => {
-      const release = requireStringOption(input, "release", {
-        message: "--release is required",
-        usage: "canonry backlinks sync --release <id> [--wait]",
-        command: "backlinks sync"
-      });
       await backlinksSync({
-        release,
+        release: getString(input.values, "release"),
         wait: getBoolean(input.values, "wait"),
         format: input.format
       });
@@ -775,6 +883,14 @@ var BACKLINKS_CLI_COMMANDS = [
       await backlinksReleases({ format: input.format });
     }
   },
+  {
+    path: ["backlinks", "releases", "latest"],
+    usage: "canonry backlinks releases latest [--format json]",
+    options: {},
+    run: async (input) => {
+      await backlinksLatestRelease({ format: input.format });
+    }
+  },
   {
     path: ["backlinks", "extract"],
     usage: "canonry backlinks extract <project> [--release <id>] [--wait] [--format json]",
@@ -7039,6 +7155,18 @@ async function serveCommand(format = "text") {
   config.port = port;
   const db = createClient(config.database);
   migrate(db);
+  try {
+    const result = backfillNormalizedPaths(db);
+    if (result.updated > 0 && format === "text") {
+      console.log(
+        `Migrated ${result.updated} GA landing-page row${result.updated === 1 ? "" : "s"} to canonical form.`
+      );
+    }
+  } catch (err) {
+    const msg = err instanceof Error ? err.message : String(err);
+    process.stderr.write(`warning: normalized-path backfill skipped: ${msg}
+`);
+  }
   const app = await createServer({ config, db });
   let shuttingDown = false;
   const shutdown = (signal) => {

package/dist/index.js CHANGED Viewed

@@ -1,10 +1,10 @@
 import {
   createServer
-} from "./chunk-DHMCIJMQ.js";
+} from "./chunk-3T64Y7GR.js";
 import {
   loadConfig
-} from "./chunk-XJS7NALL.js";
-import "./chunk-UM6RDSRJ.js";
+} from "./chunk-QTS7VZXN.js";
+import "./chunk-FV6PY5UE.js";
 import "./chunk-MLKGABMK.js";
 export {
   createServer,