@gscdump/cloudflare 0.27.2 → 0.28.1

package/dist/index.d.mts

@@ -79,16 +79,6 @@ interface HostedR2QueryKeyInput {
   comparisonFilter?: string;
 }
 declare function getHostedR2QueryKey(input: HostedR2QueryKeyInput): string;
-interface PresignOptions {
-  key: string;
-  bucket: string;
-  expiresIn?: number;
-}
-declare function createR2Presigner(env: AnalyticsEnv): ({
-  key,
-  bucket,
-  expiresIn
-}: PresignOptions) => Promise<string>;
 declare function signSizeHint(env: AnalyticsEnv, key: string, bytes: number): Promise<string>;
 declare function verifySizeHint(env: AnalyticsEnv, key: string, bytes: number, providedHex: string): Promise<boolean>;
 declare function createDucklingsCodec(_env: AnalyticsEnv): ParquetCodec;
@@ -98,4 +88,4 @@ interface DucklingsExecutorOptions {
   ipcTotalBytes?: number;
 }
 declare function createDucklingsExecutor(env: AnalyticsEnv, opts?: DucklingsExecutorOptions): QueryExecutor;
-export { type AnalyticsEngineHooks, type AnalyticsEnv, type HostedR2QueryKeyInput, type InflightDedupe, type PresignOptions, type Row, createDucklingsCodec, createDucklingsExecutor, createInflightDedupe, createR2Presigner, getAnalyticsEngine, getHostedR2QueryKey, getWasmDuckDBFactory, resetWasmDuckDB, signSizeHint, useAnalyticsEnv, verifySizeHint };
+export { type AnalyticsEngineHooks, type AnalyticsEnv, type HostedR2QueryKeyInput, type InflightDedupe, type Row, createDucklingsCodec, createDucklingsExecutor, createInflightDedupe, getAnalyticsEngine, getHostedR2QueryKey, getWasmDuckDBFactory, resetWasmDuckDB, signSizeHint, useAnalyticsEnv, verifySizeHint };

package/dist/index.mjs

@@ -4,7 +4,6 @@ import { createR2DataSource } from "@gscdump/engine/r2";
 import { createHyparquetCodec, decodeParquetToRows } from "@gscdump/engine/hyparquet";
 import { float64, int32, int64, tableFromArrays, tableToIPC, utf8 } from "@uwdata/flechette";
 import { createError } from "h3";
-import { AwsClient } from "aws4fetch";
 let handle = null;
 async function initHandle() {
 	throw new Error("DuckDB-WASM handle not wired for Cloudflare Workers yet. Complete duckdb-wasm-handle.ts before enabling dual-write (user.migration_phase != 'd1').");
@@ -407,43 +406,6 @@ function getHostedR2QueryKey(input) {
 		input.comparisonFilter ?? null
 	]);
 }
-const MAX_EXPIRES_IN = 604800;
-function createR2Presigner(env) {
-	if (!env.R2_ACCESS_KEY_ID || !env.R2_SECRET_ACCESS_KEY) throw createError({
-		statusCode: 500,
-		message: "R2 S3 credentials missing (R2_ACCESS_KEY_ID / R2_SECRET_ACCESS_KEY)"
-	});
-	if (!env.CLOUDFLARE_ACCOUNT_ID) throw createError({
-		statusCode: 500,
-		message: "CLOUDFLARE_ACCOUNT_ID missing"
-	});
-	const aws = new AwsClient({
-		accessKeyId: env.R2_ACCESS_KEY_ID,
-		secretAccessKey: env.R2_SECRET_ACCESS_KEY,
-		service: "s3",
-		region: "auto"
-	});
-	const endpoint = `https://${env.CLOUDFLARE_ACCOUNT_ID}.r2.cloudflarestorage.com`;
-	return async function presignGet({ key, bucket, expiresIn = 3600 }) {
-		if (!Number.isInteger(expiresIn) || expiresIn <= 0) throw createError({
-			statusCode: 400,
-			message: `expiresIn must be a positive integer (got ${expiresIn})`
-		});
-		if (expiresIn > MAX_EXPIRES_IN) throw createError({
-			statusCode: 400,
-			message: `expiresIn exceeds the ${MAX_EXPIRES_IN}s (7 day) S3 SigV4 maximum (got ${expiresIn})`
-		});
-		const url = new URL(`${endpoint}/${bucket}/${encodeKey(key)}`);
-		url.searchParams.set("X-Amz-Expires", String(expiresIn));
-		return (await aws.sign(url.toString(), {
-			method: "GET",
-			aws: { signQuery: true }
-		})).url;
-	};
-}
-function encodeKey(key) {
-	return key.split("/").map(encodeURIComponent).join("/");
-}
 const SIG_HEX_LEN = 64;
 const keyCache = /* @__PURE__ */ new WeakMap();
 const stringKeyCache = /* @__PURE__ */ new Map();
@@ -484,4 +446,4 @@ async function verifySizeHint(env, key, bytes, providedHex) {
 	for (let i = 0; i < SIG_HEX_LEN; i++) diff |= expected.charCodeAt(i) ^ providedHex.charCodeAt(i);
 	return diff === 0;
 }
-export { createDucklingsCodec, createDucklingsExecutor, createInflightDedupe, createR2Presigner, getAnalyticsEngine, getHostedR2QueryKey, getWasmDuckDBFactory, resetWasmDuckDB, signSizeHint, useAnalyticsEnv, verifySizeHint };
+export { createDucklingsCodec, createDucklingsExecutor, createInflightDedupe, getAnalyticsEngine, getHostedR2QueryKey, getWasmDuckDBFactory, resetWasmDuckDB, signSizeHint, useAnalyticsEnv, verifySizeHint };

package/dist/server-tail/index.d.mts

@@ -8,7 +8,25 @@ interface ArchetypeSqlPlan {
   params: unknown[];
   table: ArchetypeFactTable;
 }
-declare function buildArchetypeSql(query: ArchetypeQuery): ArchetypeSqlPlan;
+interface BuildArchetypeSqlOptions {
+  /**
+   * Set by the DuckDB file-list executor, which reads raw Iceberg parquet
+   * directly via `read_parquet([...])`, bypassing the catalog metadata layer
+   * that synthesizes the identity-partition columns. `site_id` / `search_type`
+   * are partition identities NOT materialized in the data files (see engine
+   * `iceberg/schema.ts`: "carried implicitly in the object-key prefix"), so a
+   * `WHERE site_id = ?` predicate fails with `Referenced column "site_id" not
+   * found in FROM clause`. Those files are already pruned to (site_id,
+   * search_type) at resolution time, so the predicate is redundant — emit only
+   * the row-level `date` range (a real stored column; `month(date)` is the
+   * partition transform, not `date` itself).
+   *
+   * The R2 SQL catalog path (default `false`) exposes partition columns as
+   * virtual columns and needs the full predicate.
+   */
+  partitionPruned?: boolean;
+}
+declare function buildArchetypeSql(query: ArchetypeQuery, opts?: BuildArchetypeSqlOptions): ArchetypeSqlPlan;
 /** Row returned by the DuckDB sibling. */
 type DuckDbIcebergRow = Record<string, string | number | null>;
 /**
@@ -210,4 +228,4 @@ interface ServerTailDispatcher {
  * executor and routes every `ArchetypeQuery` to one of them.
  */
 declare function createServerTailDispatcher(config: ServerTailDispatcherConfig): ServerTailDispatcher;
-export { type ArchetypeSqlPlan, DuckDbIcebergError, type DuckDbIcebergExecutor, type DuckDbIcebergExecutorConfig, type DuckDbIcebergQueryError, type DuckDbIcebergResult, type DuckDbIcebergRow, DuckDbIcebergTimeoutError, type DuckDbSvc, type R2SqlClient, type R2SqlClientConfig, R2SqlError, type R2SqlQueryError, type R2SqlResult, type R2SqlRow, R2SqlTimeoutError, type ServerTailDispatcher, type ServerTailDispatcherConfig, type ServerTailEngine, ServerTailRoutingError, TABLE_PLACEHOLDER, buildArchetypeSql, createDuckDbIcebergExecutor, createR2SqlClient, createServerTailDispatcher, resolveServerTailEngine, resolveServerTailEngineResult };
+export { type ArchetypeSqlPlan, type BuildArchetypeSqlOptions, DuckDbIcebergError, type DuckDbIcebergExecutor, type DuckDbIcebergExecutorConfig, type DuckDbIcebergQueryError, type DuckDbIcebergResult, type DuckDbIcebergRow, DuckDbIcebergTimeoutError, type DuckDbSvc, type R2SqlClient, type R2SqlClientConfig, R2SqlError, type R2SqlQueryError, type R2SqlResult, type R2SqlRow, R2SqlTimeoutError, type ServerTailDispatcher, type ServerTailDispatcherConfig, type ServerTailEngine, ServerTailRoutingError, TABLE_PLACEHOLDER, buildArchetypeSql, createDuckDbIcebergExecutor, createR2SqlClient, createServerTailDispatcher, resolveServerTailEngine, resolveServerTailEngineResult };

package/dist/server-tail/index.mjs

@@ -84,7 +84,11 @@ function moverClause(movers) {
 function sqlStringLiteral(value) {
 	return `'${value.replace(/'/g, "''")}'`;
 }
-function partitionWhere(q) {
+function partitionWhere(q, partitionPruned = false) {
+	if (partitionPruned) return {
+		clause: "date BETWEEN ? AND ?",
+		params: [q.range.start, q.range.end]
+	};
 	return {
 		clause: "site_id = ? AND search_type = ? AND date BETWEEN ? AND ?",
 		params: [
@@ -126,8 +130,8 @@ function facetPredicate(query) {
 		params
 	};
 }
-function buildSiteDailyTimeseries(q) {
-	const w = partitionWhere(q);
+function buildSiteDailyTimeseries(q, pruned) {
+	const w = partitionWhere(q, pruned);
 	const metrics = q.metrics.map(metricExpr).join(", ");
 	return {
 		table: "dates",
@@ -135,9 +139,9 @@ function buildSiteDailyTimeseries(q) {
 		sql: `SELECT date, ${metrics} FROM ${TABLE_PLACEHOLDER} WHERE ${w.clause} GROUP BY date ORDER BY date ASC`
 	};
 }
-function buildEntityDailyTimeseries(q) {
+function buildEntityDailyTimeseries(q, pruned) {
 	const table = tableForDimensions([q.entity.dimension]);
-	const w = partitionWhere(q);
+	const w = partitionWhere(q, pruned);
 	const col = dimColumn(q.entity.dimension);
 	const metrics = q.metrics.map(metricExpr).join(", ");
 	return {
@@ -146,9 +150,9 @@ function buildEntityDailyTimeseries(q) {
 		sql: `SELECT date, ${metrics} FROM ${TABLE_PLACEHOLDER} WHERE ${w.clause} AND ${col} = ? GROUP BY date ORDER BY date ASC`
 	};
 }
-function buildEntityDailySparkline(q) {
+function buildEntityDailySparkline(q, pruned) {
 	const table = tableForDimensions([q.dimension]);
-	const w = partitionWhere(q);
+	const w = partitionWhere(q, pruned);
 	const col = dimColumn(q.dimension);
 	if (q.entities.length === 0) throw new Error("entity-daily-sparkline: empty entities - resolver must pre-resolve the top-N list");
 	const inList = q.entities.map(sqlStringLiteral).join(", ");
@@ -158,9 +162,9 @@ function buildEntityDailySparkline(q) {
 		sql: `SELECT date, ${col}, ${metricExpr(q.metric)} FROM ${TABLE_PLACEHOLDER} WHERE ${w.clause} AND ${col} IN (${inList}) GROUP BY date, ${col} ORDER BY date ASC`
 	};
 }
-function buildTopNBreakdown(q) {
+function buildTopNBreakdown(q, pruned) {
 	const table = tableForDimensions([q.dimension]);
-	const w = partitionWhere(q);
+	const w = partitionWhere(q, pruned);
 	const order = `${q.orderBy.metric} ${q.orderBy.dir.toUpperCase()}`;
 	const limit = `LIMIT ${Math.max(0, Math.floor(q.limit))}`;
 	const offset = q.offset && q.offset > 0 ? ` OFFSET ${Math.floor(q.offset)}` : "";
@@ -171,7 +175,7 @@ function buildTopNBreakdown(q) {
 			const wPrev = partitionWhere({
 				...q,
 				range: q.compareRange
-			});
+			}, pruned);
 			const deviceSelects = (clause, ml) => DEVICE_SUFFIXES.map((suffix) => {
 				const source = deviceSource(suffix);
 				const metrics = ml.map((m) => metricExprForSource(m, source)).join(", ");
@@ -204,7 +208,7 @@ function buildTopNBreakdown(q) {
 		const wPrev = partitionWhere({
 			...q,
 			range: q.compareRange
-		});
+		}, pruned);
 		const curMetrics = metricList.map(metricExpr).join(", ");
 		const prevMetrics = STD_METRICS.map((m) => metricExpr(m)).join(", ");
 		const curCols = metricList.map((m) => coalesceMetric(m, "c", m)).join(", ");
@@ -232,10 +236,10 @@ function buildTopNBreakdown(q) {
 		sql
 	};
 }
-function buildSingleRowLookup(q) {
+function buildSingleRowLookup(q, pruned) {
 	const dims = Object.keys(q.match);
 	const table = tableForDimensions(dims);
-	const w = partitionWhere(q);
+	const w = partitionWhere(q, pruned);
 	const params = [...w.params];
 	let clause = w.clause;
 	for (const dim of dims) {
@@ -250,9 +254,9 @@ function buildSingleRowLookup(q) {
 		sql: `SELECT ${dims.length > 0 ? `${dims.map(dimColumn).join(", ")}, ${metrics}` : metrics} FROM ${TABLE_PLACEHOLDER} WHERE ${clause}${groupBy}`
 	};
 }
-function buildMultiSeriesStackedDaily(q) {
+function buildMultiSeriesStackedDaily(q, pruned) {
 	const table = tableForDimensions([q.seriesDimension]);
-	const w = partitionWhere(q);
+	const w = partitionWhere(q, pruned);
 	if (q.seriesDimension === "device") {
 		const selects = DEVICE_SUFFIXES.map((suffix) => {
 			const source = deviceSource(suffix);
@@ -271,8 +275,8 @@ function buildMultiSeriesStackedDaily(q) {
 		sql: `SELECT date, ${col}, ${metricExpr(q.metric)} FROM ${TABLE_PLACEHOLDER} WHERE ${w.clause} GROUP BY date, ${col} ORDER BY date ASC`
 	};
 }
-function buildTwoDimensionDetail(q) {
-	const w = partitionWhere(q);
+function buildTwoDimensionDetail(q, pruned) {
+	const w = partitionWhere(q, pruned);
 	const params = [...w.params];
 	let clause = w.clause;
 	if (q.filter?.page) {
@@ -295,15 +299,16 @@ function buildTwoDimensionDetail(q) {
 		sql
 	};
 }
-function buildArchetypeSql(query) {
+function buildArchetypeSql(query, opts = {}) {
+	const pruned = opts.partitionPruned ?? false;
 	switch (query.archetype) {
-		case "site-daily-timeseries": return buildSiteDailyTimeseries(query);
-		case "entity-daily-timeseries": return buildEntityDailyTimeseries(query);
-		case "entity-daily-sparkline": return buildEntityDailySparkline(query);
-		case "top-n-breakdown": return buildTopNBreakdown(query);
-		case "single-row-lookup": return buildSingleRowLookup(query);
-		case "multi-series-stacked-daily": return buildMultiSeriesStackedDaily(query);
-		case "two-dimension-detail": return buildTwoDimensionDetail(query);
+		case "site-daily-timeseries": return buildSiteDailyTimeseries(query, pruned);
+		case "entity-daily-timeseries": return buildEntityDailyTimeseries(query, pruned);
+		case "entity-daily-sparkline": return buildEntityDailySparkline(query, pruned);
+		case "top-n-breakdown": return buildTopNBreakdown(query, pruned);
+		case "single-row-lookup": return buildSingleRowLookup(query, pruned);
+		case "multi-series-stacked-daily": return buildMultiSeriesStackedDaily(query, pruned);
+		case "two-dimension-detail": return buildTwoDimensionDetail(query, pruned);
 		case "arbitrary-sql": throw new Error("buildArchetypeSql: arbitrary-sql carries caller SQL - the DuckDB executor runs it verbatim");
 		case "aux-cloud-only": throw new Error("buildArchetypeSql: aux-cloud-only is not an Iceberg query");
 	}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/cloudflare",
   "type": "module",
-  "version": "0.27.2",
+  "version": "0.28.1",
   "description": "Cloudflare-Workers-flavored helpers for the gscdump analytics stack: AnalyticsEnv binding contract, R2 SigV4 presigner, size-hint HMAC, DuckDB Workers shims, engine factory.",
   "author": {
     "name": "Harlan Wilton",
@@ -46,10 +46,10 @@
   "dependencies": {
     "@uwdata/flechette": "^2.5.0",
     "aws4fetch": "^1.0.20",
-    "@gscdump/contracts": "0.27.2",
-    "@gscdump/engine-sqlite": "0.27.2",
-    "@gscdump/engine": "0.27.2",
-    "gscdump": "0.27.2"
+    "@gscdump/engine": "0.28.1",
+    "@gscdump/engine-sqlite": "0.28.1",
+    "gscdump": "0.28.1",
+    "@gscdump/contracts": "0.28.1"
   },
   "devDependencies": {
     "@cloudflare/vitest-pool-workers": "^0.16.16",