@gscdump/engine-gsc-api 0.33.0 → 0.33.3

package/dist/index.d.mts

@@ -175,11 +175,23 @@ interface RunGscSearchAppearanceContextSliceOptions {
     searchType: SearchType$1;
     rowsThisPage: number;
   }) => void;
+  continuation?: SearchAppearanceContinuation;
 }
+type SearchAppearanceContinuation = {
+  phase: 'discovery';
+  appearances: string[];
+  nextStartRow: number;
+} | {
+  phase: 'context';
+  appearances: string[];
+  appearanceIndex: number;
+  nextStartRow: number;
+};
 interface RunGscSearchAppearanceContextSliceResult {
   appearances: string[];
   totalRows: number;
   hasMore: boolean;
+  continuation?: SearchAppearanceContinuation;
 }
 declare function runGscSyncSlice(opts: RunGscSyncSliceOptions): Promise<RunGscSyncSliceResult>;
 /**

package/dist/index.mjs

@@ -1,6 +1,6 @@
 import { googleSearchConsole } from "gscdump/api";
 import { between, clicks, date, extractMetricFilters, extractSpecialOperatorFilters, gsc, normalizeFilter } from "gscdump/query";
-import { assertDimensionsSupported, getDimensionFilters, getFilterDimensions, matchesDimensionFilter, matchesMetricFilter, matchesTopLevelPage, metricValue } from "@gscdump/engine/resolver";
+import { assertDimensionsSupported, getFilterDimensions, matchesDimensionFilter, matchesMetricFilter, matchesTopLevelPage, metricValue } from "@gscdump/engine/resolver";
 import { buildLogicalPlan } from "gscdump/query/plan";
 const METRIC_NAMES = [
 	"clicks",
@@ -11,12 +11,20 @@ const METRIC_NAMES = [
 function isMetricDimension$1(dim) {
 	return METRIC_NAMES.includes(dim);
 }
+function isLocalDimensionFilter(filter) {
+	return filter.dimension !== "date" && !isMetricDimension$1(filter.dimension) && filter.operator !== "topLevel" && !filter.operator.startsWith("metric");
+}
+function matchesDimensionFilterTree(row, filter) {
+	if (!filter || !("_filters" in filter)) return true;
+	const checks = [...filter._filters.filter(isLocalDimensionFilter).map((f) => matchesDimensionFilter(row, f)), ...(filter._nestedGroups ?? []).map((group) => matchesDimensionFilterTree(row, group))];
+	if (checks.length === 0) return true;
+	return filter._groupType === "or" ? checks.some(Boolean) : checks.every(Boolean);
+}
 function applyBuilderStatePostProcessing(rows, state) {
-	const dimensionFilters = getDimensionFilters(state.filter, isMetricDimension$1);
 	const metricFilters = extractMetricFilters(state.filter);
 	const specialFilters = extractSpecialOperatorFilters(state.filter);
 	const ordered = [...rows.filter((row) => {
-		if (!dimensionFilters.every((filter) => matchesDimensionFilter(row, filter))) return false;
+		if (!matchesDimensionFilterTree(row, state.filter)) return false;
 		if (!metricFilters.every((filter) => matchesMetricFilter(row, filter))) return false;
 		if (specialFilters.some((filter) => filter.operator === "topLevel") && !matchesTopLevelPage(row)) return false;
 		return true;
@@ -49,12 +57,12 @@ async function fetchGscTopN(opts) {
 		const key = row[dimension.dimension];
 		if (typeof key !== "string" || !key) return null;
 		const impressions = Number(row.impressions ?? 0);
-		const position = Number(row.position ?? 0);
+		const position = Math.max(1, Number(row.position ?? 0));
 		return {
 			key,
 			clicks: Number(row.clicks ?? 0),
 			impressions,
-			sum_position: position * impressions
+			sum_position: (position - 1) * impressions
 		};
 	}).filter((x) => x != null);
 	if (!orderByClicksDesc) mapped.sort((a, b) => b.clicks - a.clicks);
@@ -73,7 +81,7 @@ async function fetchGscDaily(opts) {
 			date: Date.parse(`${row.date}T00:00:00Z`),
 			clicks: row.clicks ?? 0,
 			impressions,
-			sum_position: (row.position ?? 0) * impressions,
+			sum_position: (Math.max(1, row.position ?? 0) - 1) * impressions,
 			anonymizedImpressionsPct: 0
 		};
 	}).filter((x) => x != null).sort((a, b) => a.date - b.date);
@@ -105,15 +113,18 @@ function createGscApiQuerySource(options) {
 			buildLogicalPlan(state, GSC_API_CAPABILITIES);
 			const filterDims = getFilterDimensions(state.filter, isMetricDimension);
 			assertDimensionsSupported([...state.dimensions, ...filterDims], "api", "gsc-api query source");
-			return applyBuilderStatePostProcessing(await collectRows(client.query(siteUrl, builderFromState(state))), state);
+			const apiState = {
+				...state,
+				rowLimit: void 0,
+				startRow: void 0
+			};
+			return applyBuilderStatePostProcessing(await collectRows(client.query(siteUrl, builderFromState(apiState))), state);
 		}
 	};
 }
 const PRO_ONLY_DIMENSIONS = /* @__PURE__ */ new Set(["queryCanonical", "page_keywords"]);
 function canProxyToGsc(state) {
 	if (state.dimensions.some((d) => PRO_ONLY_DIMENSIONS.has(d))) return false;
-	if (extractMetricFilters(state.filter).length > 0) return false;
-	if (extractSpecialOperatorFilters(state.filter).length > 0) return false;
 	return true;
 }
 function withSearchType(state, searchType) {
@@ -211,7 +222,7 @@ async function runGscSyncSlice(opts) {
 	const maxPages = opts.maxPages ?? Infinity;
 	const searchType = opts.searchType ?? "web";
 	const dimensions = opts.dimensions ? [...opts.dimensions] : [...DIMENSIONS_BY_TABLE[opts.table]];
-	const dataState = opts.dataState ?? "all";
+	const dataState = opts.dataState ?? (dimensions.includes("hour") ? "hourly_all" : "all");
 	const dimensionFilterGroups = buildDimensionFilterGroups(opts.domainFilter, opts.dimensionFilters);
 	const loopStart = Date.now();
 	let startRow = opts.initialStartRow ?? 0;
@@ -278,7 +289,7 @@ async function runGscSyncSlice(opts) {
 			searchType,
 			rowsThisPage: rows.length
 		});
-		const isLastPage = rows.length < rowLimit;
+		const isLastPage = rows.length === 0;
 		const nextStartRow = page.startRow + rows.length;
 		const prefetch = isLastPage ? null : startFetchIfAllowed(nextStartRow);
 		if (rows.length > 0) {
@@ -323,11 +334,11 @@ function contextTableForGrain(grain) {
 }
 async function runGscSearchAppearanceContextSlice(opts) {
 	const table = opts.table ?? contextTableForGrain(opts.grain ?? "page_query");
-	const appearances = opts.appearances?.slice() ?? [];
+	const appearances = opts.continuation?.appearances?.slice() ?? opts.appearances?.slice() ?? [];
 	let totalRows = 0;
 	let hasMore = false;
-	if (!opts.appearances) {
-		const discovered = /* @__PURE__ */ new Set();
+	if (!opts.appearances && opts.continuation?.phase !== "context") {
+		const discovered = new Set(appearances);
 		const discovery = await runGscSyncSlice({
 			client: opts.client,
 			siteUrl: opts.siteUrl,
@@ -340,6 +351,7 @@ async function runGscSearchAppearanceContextSlice(opts) {
 			maxPages: opts.maxPages,
 			cpuBudgetMs: opts.cpuBudgetMs,
 			searchType: opts.searchType,
+			initialStartRow: opts.continuation?.phase === "discovery" ? opts.continuation.nextStartRow : void 0,
 			onPage: opts.onPage,
 			onBatch: async (rows) => {
 				for (const row of rows) {
@@ -350,10 +362,23 @@ async function runGscSearchAppearanceContextSlice(opts) {
 			}
 		});
 		totalRows += discovery.totalRows;
+		if (discovery.hasMore) return {
+			appearances: [...discovered],
+			totalRows,
+			hasMore: true,
+			continuation: {
+				phase: "discovery",
+				appearances: [...discovered],
+				nextStartRow: discovery.nextStartRow
+			}
+		};
 		hasMore ||= discovery.hasMore;
-		appearances.push(...discovered);
+		appearances.splice(0, appearances.length, ...discovered);
 	}
-	for (const searchAppearance of appearances) {
+	const startIndex = opts.continuation?.phase === "context" ? opts.continuation.appearanceIndex : 0;
+	const startRow = opts.continuation?.phase === "context" ? opts.continuation.nextStartRow : 0;
+	for (let i = startIndex; i < appearances.length; i++) {
+		const searchAppearance = appearances[i];
 		const context = await runGscSyncSlice({
 			client: opts.client,
 			siteUrl: opts.siteUrl,
@@ -370,6 +395,7 @@ async function runGscSearchAppearanceContextSlice(opts) {
 			maxPages: opts.maxPages,
 			cpuBudgetMs: opts.cpuBudgetMs,
 			searchType: opts.searchType,
+			initialStartRow: i === startIndex ? startRow : void 0,
 			onPage: opts.onPage,
 			onBatch: (rows) => opts.onContextBatch({
 				searchAppearance,
@@ -378,6 +404,17 @@ async function runGscSearchAppearanceContextSlice(opts) {
 			})
 		});
 		totalRows += context.totalRows;
+		if (context.hasMore) return {
+			appearances,
+			totalRows,
+			hasMore: true,
+			continuation: {
+				phase: "context",
+				appearances,
+				appearanceIndex: i,
+				nextStartRow: context.nextStartRow
+			}
+		};
 		hasMore ||= context.hasMore;
 	}
 	return {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/engine-gsc-api",
   "type": "module",
-  "version": "0.33.0",
+  "version": "0.33.3",
   "description": "GSC live-API engine adapter — wraps the Search Analytics REST API as an AnalysisQuerySource for typed analyzer dispatch.",
   "author": {
     "name": "Harlan Wilton",
@@ -36,8 +36,8 @@
     "node": ">=18"
   },
   "dependencies": {
-    "gscdump": "0.33.0",
-    "@gscdump/engine": "0.33.0"
+    "gscdump": "0.33.3",
+    "@gscdump/engine": "0.33.3"
   },
   "devDependencies": {
     "vitest": "^4.1.9"