Extract canonical site rules into share/mkwb/site.mk

Sites used to copy-paste ~340 lines of suffix rules, derived-asset
lists, and recipes (global.conf assembly, sitemap, security.txt,
gpg-derived fingerprint/expiry, check-unit-*) into every Makefile.
Move all of it into share/mkwb/site.mk, shipped at install time.

Add src/rules.in -> `mkwb rules`: prints the absolute path of the
shipped site.mk so a site's mkdeps.sh can append it as the last
`include` line of deps.mk.  The site's own Makefile then needs
only an `all:` anchor (to keep it as the default target ahead of
deps.mk's first rule) and `include deps.mk`.

Also adds tests/resources/{site,expected}/ and tests/integration.sh
exercising mkwb htmlbody / links / snippets / sortdata against
golden files, with sortdata covering >1 collection and >1 entry
per collection to verify per-collection chronological ordering.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

1 files changed, 94 insertions, 0 deletions

diff --git a/tests/integration.sh b/tests/integration.sh
new file mode 100755
index 0000000..b8b7c1f
--- /dev/null
+++ b/tests/integration.sh
@@ -0,0 +1,94 @@
+#!/bin/sh
+# Integration test: drives mkwb against a slim fixture site and
+# diffs every output against committed golden files under
+# tests/resources/expected/.  Run in an isolated tmpdir so the
+# snapshot fixture (tests/resources/site/) stays clean.
+#
+# Required tools on PATH: mkwb (with all libexec scripts) and adoc.
+set -eu
+
+HERE="$(dirname "$(readlink -f "$0")")"
+RESOURCES="$HERE/resources"
+SITE="$RESOURCES/site"
+EXPECTED="$RESOURCES/expected"
+
+WORK="$(mktemp -d -t mkwb-integration.XXXXXX)"
+trap 'rm -rf "$WORK"' EXIT INT TERM
+
+cp -r "$SITE/." "$WORK/"
+
+CONTENT="$WORK/src/content"
+ACTUAL="$WORK/actual"
+mkdir -p "$ACTUAL"
+
+# --- single-page subcommands -------------------------------------
+# htmlbody: .adoc -> HTML body fragment
+mkwb htmlbody "$CONTENT/hello.adoc" > "$ACTUAL/hello.htmlbody"
+
+# links: extract :attr: link values from the .adoc
+mkwb links    "$CONTENT/hello.adoc" > "$ACTUAL/hello.links"
+
+# snippets: emits one .txt per "----" block next to the .adoc, and
+# prints the list of filenames to stdout.  Run from $CONTENT so the
+# filenames are relative.
+( cd "$CONTENT" && mkwb snippets hello.adoc ) > "$ACTUAL/hello.snippets"
+cp "$CONTENT/hello.html.0.txt" "$ACTUAL/hello.html.0.txt"
+cp "$CONTENT/hello.html.1.txt" "$ACTUAL/hello.html.1.txt"
+
+# --- sortdata: per-collection chronological sort -----------------
+# Each <collection>/YYYY/MM/DD/X.conf gets a sibling sentinel three
+# levels up at <collection>/${date_iso}-${sort}.sortdata pointing
+# back to the original .sortdata.  Sorting these sentinel filenames
+# alphabetically yields the chronological order per collection.
+for conf in \
+		"$CONTENT/en/blog/2024/01/15/launch.conf"      \
+		"$CONTENT/en/blog/2025/03/20/feature.conf"     \
+		"$CONTENT/en/news/2024/06/01/founding.conf"    \
+		"$CONTENT/en/news/2025/06/01/anniversary.conf" \
+		; do
+	( cd "$WORK" && mkwb sortdata "${conf#$WORK/}" ) > /dev/null
+done
+
+# --- diff every captured output against goldens ------------------
+fail=0
+diff_one() {
+	if diff -u "$EXPECTED/$1" "$2"; then
+		echo "PASS: $1"
+	else
+		echo "FAIL: $1 differs from expected" >&2
+		fail=1
+	fi
+}
+
+for f in hello.htmlbody hello.links hello.snippets \
+         hello.html.0.txt hello.html.1.txt; do
+	diff_one "$f" "$ACTUAL/$f"
+done
+
+# Diff each sentinel file content (points back to .sortdata).
+for f in src/content/en/blog/2024-01-15-launch.sortdata    \
+         src/content/en/blog/2025-03-20-feature.sortdata   \
+         src/content/en/news/2024-06-01-founding.sortdata  \
+         src/content/en/news/2025-06-01-anniversary.sortdata; do
+	diff_one "$f" "$WORK/$f"
+done
+
+# Verify the FILENAME order per collection encodes chronological
+# order -- sort all *.sortdata under each collection and check the
+# listing matches expected.
+for col in blog news; do
+	want="$(cd "$EXPECTED/src/content/en/$col" && \
+		ls *.sortdata 2>/dev/null | sort)"
+	got="$(cd "$WORK/src/content/en/$col" && \
+		ls *.sortdata 2>/dev/null | sort)"
+	if [ "$want" = "$got" ]; then
+		echo "PASS: sortdata order in collection '$col'"
+	else
+		echo "FAIL: sortdata order in collection '$col'" >&2
+		echo "  expected:"; echo "$want" | sed 's/^/    /'
+		echo "  actual:  "; echo "$got"  | sed 's/^/    /'
+		fail=1
+	fi
+done
+
+exit "$fail"