@gscdump/cli 1.5.0 → 1.5.3

package/dist/commands/indexing.mjs

@@ -1,9 +1,10 @@
 import { OUTPUT_ARGS, applyOutputMode, logger, readUrlList } from "../utils.mjs";
 import { createCommandContext } from "../context.mjs";
 import { gscErrorHandler } from "../error-handler.mjs";
+import { loadSitemapUrls } from "../sitemap.mjs";
 import process from "node:process";
 import { defineCommand } from "citty";
-import { batchRequestIndexing, fetchSitemapUrls, getIndexingMetadata, requestIndexing, runSequentialBatch } from "gscdump";
+import { batchRequestIndexing, getIndexingMetadata, requestIndexing, runSequentialBatch } from "gscdump";
 const RETRIES_ARG = { retries: {
 	type: "string",
 	description: "Override per-call retry count (default: 3)"
@@ -15,10 +16,15 @@ function parseRetries(v) {
 }
 async function resolveUrlSource(args) {
 	const fromSitemap = args["from-sitemap"];
-	if (fromSitemap) return fetchSitemapUrls(String(fromSitemap)).catch((e) => {
-		logger.error(`Sitemap fetch failed: ${e.message}`);
-		process.exit(1);
-	});
+	if (fromSitemap) {
+		const result = await loadSitemapUrls(String(fromSitemap));
+		if (result._tag === "error") {
+			logger.error(`Sitemap fetch failed: ${result.message}`);
+			process.exit(1);
+		}
+		if (!result.value.complete) logger.warn("Sitemap walk was incomplete; indexing only the URLs that were read");
+		return result.value.urls;
+	}
 	return readUrlList(args);
 }
 const submitCommand = defineCommand({

package/dist/commands/inspect.mjs

@@ -1,9 +1,10 @@
 import { OUTPUT_ARGS, applyOutputMode, logger, readUrlList } from "../utils.mjs";
 import { createCommandContext } from "../context.mjs";
 import { gscErrorHandler } from "../error-handler.mjs";
+import { loadSitemapUrls } from "../sitemap.mjs";
 import process from "node:process";
 import { defineCommand } from "citty";
-import { batchInspectUrls, fetchSitemapUrls } from "gscdump";
+import { batchInspectUrls } from "gscdump";
 function verdictTone(verdict) {
 	if (verdict === "PASS") return "\x1B[32m";
 	if (verdict === "NEUTRAL" || verdict === "PARTIAL") return "\x1B[33m";
@@ -121,10 +122,16 @@ const batchCommand = defineCommand({
 	},
 	async run({ args }) {
 		const { json, quiet } = applyOutputMode(args);
-		const urls = args["from-sitemap"] ? await fetchSitemapUrls(String(args["from-sitemap"])).catch((e) => {
-			logger.error(`Sitemap fetch failed: ${e.message}`);
-			process.exit(1);
-		}) : await readUrlList(args);
+		let urls;
+		if (args["from-sitemap"]) {
+			const result = await loadSitemapUrls(String(args["from-sitemap"]));
+			if (result._tag === "error") {
+				logger.error(`Sitemap fetch failed: ${result.message}`);
+				process.exit(1);
+			}
+			if (!result.value.complete) logger.warn("Sitemap walk was incomplete; inspecting only the URLs that were read");
+			urls = result.value.urls;
+		} else urls = await readUrlList(args);
 		if (urls.length === 0) {
 			logger.error("No URLs provided. Pass URLs as args, --file, --from-sitemap, or stdin.");
 			process.exit(1);

package/dist/commands/sitemaps.mjs

@@ -1,9 +1,11 @@
 import { OUTPUT_ARGS, applyOutputMode, logger, noSubcommandSelected } from "../utils.mjs";
 import { createCommandContext } from "../context.mjs";
 import { gscErrorHandler } from "../error-handler.mjs";
+import { loadSitemapUrls } from "../sitemap.mjs";
 import process from "node:process";
 import { defineCommand } from "citty";
-import { discoverSitemap, fetchSitemap, fetchSitemapUrls } from "gscdump";
+import { fetchSitemap } from "gscdump";
+import { discoverSitemapResult } from "gscdump/sitemap";
 const listCommand = defineCommand({
 	meta: {
 		name: "list",
@@ -195,14 +197,20 @@ const sitemapsCommand = defineCommand({
 			async run({ args }) {
 				const { json } = applyOutputMode(args);
 				const domain = String(args.domain).replace(/^https?:\/\//, "").replace(/\/.*$/, "");
-				const url = await discoverSitemap(domain).catch(() => null);
+				const discovery = await discoverSitemapResult(domain);
+				const url = discovery._tag === "found" ? discovery.url : null;
 				if (json) {
 					console.log(JSON.stringify({
 						domain,
-						sitemap: url
+						sitemap: url,
+						status: discovery._tag
 					}, null, 2));
 					return;
 				}
+				if (discovery._tag === "incomplete") {
+					logger.error(`Sitemap discovery was incomplete: ${discovery.failures[0]?.detail ?? "unknown failure"}`);
+					process.exit(1);
+				}
 				if (!url) {
 					logger.warn(`No sitemap discovered for ${domain}`);
 					process.exit(1);
@@ -236,21 +244,26 @@ const sitemapsCommand = defineCommand({
 				const { json } = applyOutputMode(args);
 				const limit = args.limit ? Number.parseInt(String(args.limit), 10) : void 0;
 				const maxDepth = args["max-depth"] ? Number.parseInt(String(args["max-depth"]), 10) : void 0;
-				const urls = await fetchSitemapUrls(String(args.url), {
-					limit,
+				const result = await loadSitemapUrls(String(args.url), {
+					maxUrls: limit,
 					maxDepth
-				}).catch((e) => {
-					logger.error(`Sitemap fetch failed: ${e.message}`);
-					process.exit(1);
 				});
+				if (result._tag === "error") {
+					logger.error(`Sitemap fetch failed: ${result.message}`);
+					process.exit(1);
+				}
+				const { urls, complete, documentsRead } = result.value;
 				if (json) {
 					console.log(JSON.stringify({
 						sitemap: args.url,
 						count: urls.length,
+						complete,
+						documentsRead,
 						urls
 					}, null, 2));
 					return;
 				}
+				if (!complete) logger.warn("Sitemap walk was truncated or a child document failed");
 				for (const u of urls) console.log(u);
 			}
 		})

package/dist/mcp/server/index.mjs

@@ -4,7 +4,8 @@ import { listReports, runReportHandler } from "../handlers/reports.mjs";
 import { getSitemap, listSitesWithSitemaps } from "../handlers/sites.mjs";
 import "../handlers/index.mjs";
 import { batchInspectUrlsInput, batchRequestIndexingInput, getIndexingStatusInput, inspectUrlInput, listReportsInput, listSitemapsInput, listSitesInput, requestIndexingInput, runReportInput, sitemapInput } from "../types.mjs";
-import { addSite, deleteSite, discoverSitemap, getIndexingMetadata, getVerificationToken, getVerifiedSite, googleSearchConsole, listVerifiedSites, runSequentialBatch, unverifySite, verifySite } from "gscdump";
+import { addSite, deleteSite, getIndexingMetadata, getVerificationToken, getVerifiedSite, googleSearchConsole, listVerifiedSites, runSequentialBatch, unverifySite, verifySite } from "gscdump";
+import { discoverSitemapResult } from "gscdump/sitemap";
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
 import { z } from "zod";
 async function runMcpSearchAnalyticsQuery(client, args) {
@@ -385,12 +386,14 @@ function createGscMcpServer(options) {
 		inputSchema: z.object({ domain: z.string().describe("Domain (e.g., example.com)") }).shape
 	}, async ({ domain }) => {
 		const cleaned = String(domain).replace(/^https?:\/\//, "").replace(/\/.*$/, "");
-		const url = await discoverSitemap(cleaned).catch(() => null);
+		const discovery = await discoverSitemapResult(cleaned);
 		return { content: [{
 			type: "text",
 			text: JSON.stringify({
 				domain: cleaned,
-				sitemap: url
+				sitemap: discovery._tag === "found" ? discovery.url : null,
+				status: discovery._tag,
+				...discovery._tag === "incomplete" ? { failures: discovery.failures } : {}
 			}, null, 2)
 		}] };
 	});

package/dist/package.mjs

@@ -1,2 +1,2 @@
-var version = "1.5.0";
+var version = "1.5.3";
 export { version };

package/dist/sitemap.mjs

@@ -0,0 +1,27 @@
+import { walkSitemaps } from "gscdump/sitemap";
+function failureMessage(result) {
+	if (result._tag === "not_found") return "No sitemap document was found";
+	const first = result.failures[0];
+	if (!first) return "Sitemap walk failed before reading a document";
+	const error = first.error;
+	if (error._tag === "network_error") return `${first.url}: ${error.error}`;
+	if (error._tag === "http_error") return `${first.url}: HTTP ${error.status} ${error.statusText}`;
+	if (error._tag === "not_found") return `${first.url}: HTTP ${error.status}`;
+	return `${first.url}: ${error.error._tag}`;
+}
+async function loadSitemapUrls(url, options = {}) {
+	const result = await walkSitemaps(url, options);
+	if (result._tag !== "ok") return {
+		_tag: "error",
+		message: failureMessage(result)
+	};
+	return {
+		_tag: "ok",
+		value: {
+			urls: result.entries.map((entry) => entry.loc),
+			complete: result.complete,
+			documentsRead: result.documentsRead
+		}
+	};
+}
+export { loadSitemapUrls };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/cli",
   "type": "module",
-  "version": "1.5.0",
+  "version": "1.5.3",
   "description": "CLI for Google Search Console - dump, query, and run MCP server",
   "author": {
     "name": "Harlan Wilton",
@@ -50,10 +50,10 @@
     "ofetch": "^1.5.1",
     "open": "^11.0.0",
     "zod": "^4.4.3",
-    "@gscdump/analysis": "^1.5.0",
-    "@gscdump/engine": "^1.5.0",
-    "@gscdump/engine-gsc-api": "^1.5.0",
-    "gscdump": "^1.5.0"
+    "@gscdump/analysis": "^1.5.3",
+    "@gscdump/engine-gsc-api": "^1.5.3",
+    "gscdump": "^1.5.3",
+    "@gscdump/engine": "^1.5.3"
   },
   "devDependencies": {
     "vitest": "^4.1.10"