#!/bin/bash
# Diff two composed site outputs modulo the build clock.
#
# Normalised away, and only these:
#   - "generatedAt": "<iso>"  in every JSON (compact and pretty forms)
#   - the archive zip's own bytes: entry mtimes are the build clock, so a zip
#     built twice differs by a byte or two even when its ENTRIES are identical.
#     Zips are therefore compared by unpacked, generatedAt-normalised contents,
#     and archives/manifest.json's `bytes` (the zip's size) with them.
A=$1; B=$2
norm() { sed -E -e 's/"generatedAt": *"[^"]*"/"generatedAt":"X"/g' "$1"; }
normarch() { sed -E -e 's/"generatedAt": *"[^"]*"/"generatedAt":"X"/g' -e 's/"bytes": *[0-9]+/"bytes":N/g' "$1"; }
rc=0
( cd "$A" && find . -type f | sort ) > /tmp/fa.$$
( cd "$B" && find . -type f | sort ) > /tmp/fb.$$
if ! diff -q /tmp/fa.$$ /tmp/fb.$$ >/dev/null; then echo "FILE LIST DIFFERS"; diff /tmp/fa.$$ /tmp/fb.$$; rc=1; fi
while read -r f; do
  if [[ "$f" == *.zip ]]; then
    ta=$(mktemp -d); tb=$(mktemp -d)
    unzip -qq -o "$A/$f" -d "$ta"; unzip -qq -o "$B/$f" -d "$tb"
    while read -r e; do
      if ! diff <(norm "$ta/$e") <(norm "$tb/$e") >/dev/null; then
        echo "ZIP ENTRY DIFFERS: $f!$e"; rc=1
      fi
    done < <(cd "$ta" && find . -type f | sort)
    if ! diff <(cd "$ta" && find . -type f|sort) <(cd "$tb" && find . -type f|sort) >/dev/null; then
      echo "ZIP ENTRY LIST DIFFERS: $f"; rc=1; fi
    rm -rf "$ta" "$tb"
  elif [[ "$f" == ./archives/manifest.json ]]; then
    if ! diff <(normarch "$A/$f") <(normarch "$B/$f") >/dev/null; then
      echo "DIFFERS: $f"; diff <(normarch "$A/$f") <(normarch "$B/$f") | head -8; rc=1; fi
  elif ! diff <(norm "$A/$f") <(norm "$B/$f") >/dev/null; then
    echo "DIFFERS: $f"; diff <(norm "$A/$f") <(norm "$B/$f") | head -8; rc=1
  fi
done < /tmp/fa.$$
rm -f /tmp/fa.$$ /tmp/fb.$$
[ $rc -eq 0 ] && echo "IDENTICAL modulo the build clock ($(cd "$A" && find . -type f | wc -l) files)"
exit $rc