From 6b2cf2a429cf8a7fd6301c09ae5220f20de15d0b Mon Sep 17 00:00:00 2001
From: Mihaela Mihaljevic <mihaelamj@me.com>
Date: Fri, 26 Jun 2026 08:59:07 +0200
Subject: [PATCH] chore(test): add scripts/fetch-xsts.sh to obtain the XSTS
 corpus (#321)

The W3C XML Schema Test Suite (2006-11-06) drives the opt-in conformance
runner but is never vendored: the W3C Document License permits
redistribution only as the complete unmodified archive, and the corpus
is about 200 MB. This script downloads the official archive on demand,
verifies its SHA-256, and extracts it to the path XSTS_ROOT expects.
Idempotent and portable (shasum or sha256sum). Points the onboarding doc
at it.
---
 docs/onboarding-antigravity.md |  4 ++-
 scripts/fetch-xsts.sh          | 62 ++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+), 1 deletion(-)
 create mode 100755 scripts/fetch-xsts.sh

diff --git a/docs/onboarding-antigravity.md b/docs/onboarding-antigravity.md
index 92210e8..9ae7660 100644
--- a/docs/onboarding-antigravity.md
+++ b/docs/onboarding-antigravity.md
@@ -88,7 +88,9 @@ environment):
   This path is the value of the `XSTS_ROOT` environment variable used by the
   conformance runner. If it is missing, the conformance runner silently does
   nothing (it is opt-in), so confirm the directory exists before relying on it:
-  `ls /private/tmp/xsts/xmlschema2006-11-06/suite.xml`.
+  `ls /private/tmp/xsts/xmlschema2006-11-06/suite.xml`. To obtain it, run
+  `bash scripts/fetch-xsts.sh`, which downloads and SHA-256-verifies the official
+  W3C archive into that path (the suite is never vendored).
 
 ---
 
diff --git a/scripts/fetch-xsts.sh b/scripts/fetch-xsts.sh
new file mode 100755
index 0000000..6d6c0a3
--- /dev/null
+++ b/scripts/fetch-xsts.sh
@@ -0,0 +1,62 @@
+#!/usr/bin/env bash
+# Fetch the W3C XML Schema Test Suite (XSTS, 2006-11-06 release) to the path the
+# conformance runner (Tests/XSTSSuiteTests.swift) expects.
+#
+# The suite is never vendored: the W3C Document License permits redistribution
+# only as the complete unmodified archive, and the corpus is about 200 MB across
+# 39,399 files. This script downloads the official archive on demand and verifies
+# its SHA-256 before extracting.
+#
+# Usage:
+#   bash scripts/fetch-xsts.sh
+#   XSTS_ROOT=/private/tmp/xsts/xmlschema2006-11-06 swift test -c release --filter XSTSSuiteTests
+#
+# Override the destination directory with XSTS_DEST (default /private/tmp/xsts).
+
+set -euo pipefail
+
+dest="${XSTS_DEST:-/private/tmp/xsts}"
+root="$dest/xmlschema2006-11-06"
+url="https://www.w3.org/XML/2004/xml-schema-test-suite/xmlschema2006-11-06/xsts-2007-06-20.tar.gz"
+sha256="902176b25e4111cf96b08663107521a4992e8ea67aad6b815592a6a5b4b9ea06"
+
+if [ -f "$root/suite.xml" ]; then
+  echo "xsts: already present at $root"
+  echo "xsts: export XSTS_ROOT=$root"
+  exit 0
+fi
+
+mkdir -p "$dest"
+archive="$dest/xsts-2007-06-20.tar.gz"
+echo "xsts: downloading the 2006-11-06 archive (about 4.4 MB)"
+curl -fsSL -o "$archive" "$url"
+
+# Verify integrity before extracting (portable across macOS and Linux).
+if command -v shasum >/dev/null 2>&1; then
+  actual="$(shasum -a 256 "$archive" | awk '{print $1}')"
+elif command -v sha256sum >/dev/null 2>&1; then
+  actual="$(sha256sum "$archive" | awk '{print $1}')"
+else
+  echo "xsts: no sha256 tool (shasum or sha256sum) found" >&2
+  rm -f "$archive"
+  exit 1
+fi
+
+if [ "$actual" != "$sha256" ]; then
+  echo "xsts: checksum mismatch, refusing to extract" >&2
+  echo "  expected $sha256" >&2
+  echo "  actual   $actual" >&2
+  rm -f "$archive"
+  exit 1
+fi
+
+tar xzf "$archive" -C "$dest"
+rm -f "$archive"
+
+if [ ! -f "$root/suite.xml" ]; then
+  echo "xsts: suite.xml not found after extraction at $root" >&2
+  exit 1
+fi
+
+echo "xsts: ready at $root"
+echo "xsts: export XSTS_ROOT=$root"