script: add print-mzcat-checksums.sh

build-all.sh

@@ -106,6 +106,18 @@ set -e
 
 # ----
 
+echo "Running  gen/bin/example-mzcat"
+set +e
+script/print-mzcat-checksums.sh | \
+    diff --unified test/mzcat-checksums-of-data.txt /dev/stdin
+if [ $? != 0 ]; then
+  echo "Unexpected change in test/mzcat-checksums-of-data.txt"
+  exit 1
+fi
+set -e
+
+# ----
+
 for f in gen/bin/fuzz-*; do
   echo "Running  $f"
   $f test/data > /dev/null

script/print-mzcat-checksums.sh

@@ -0,0 +1,62 @@
+#!/bin/bash -eu
+# Copyright 2023 The Wuffs Authors.
+#
+# Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
+# https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
+# <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
+# option. This file may not be copied, modified, or distributed
+# except according to those terms.
+#
+# SPDX-License-Identifier: Apache-2.0 OR MIT
+
+# ----------------
+
+# This script prints the CRC-32 checksum of the decompressed data of every file
+# given as a command line argument: if a file then itself, if a directory then
+# the files under it. It skips any non-compressed file found (non-compressed
+# meaning not decodable by Wuffs' standard library).
+#
+# It is not perfect. It can have false positives and false negatives.
+# Nonetheless, running it regularly (compiled against the in-development
+# release/c/wuffs-unsupported-snapshot.c) can help detect regressions.
+
+if [ ! -e wuffs-root-directory.txt ]; then
+  echo "$0 should be run from the Wuffs root directory."
+  exit 1
+elif [ ! -e gen/bin/example-mzcat ]; then
+  echo "Run \"./build-example.sh example/mzcat\" first."
+  exit 1
+elif [ ! -e gen/bin/example-crc32 ]; then
+  echo "Run \"./build-example.sh example/crc32\" first."
+  exit 1
+fi
+
+sources=$@
+if [ $# -eq 0 ]; then
+  sources=test/data
+fi
+
+# ----
+
+handle() {
+  local c=$(gen/bin/example-mzcat <$1 2>/dev/null | gen/bin/example-crc32)
+  if [ "$c" != "00000000" ]; then
+    echo $c $1
+  fi
+}
+
+# ----
+
+echo "# Generated by script/print-mzcat-checksums.sh"
+for f in $sources; do
+  if [ -f $f ]; then
+    handle $f
+  elif [ -d $f ]; then
+    for g in `find $f -type f | LANG=C sort`; do
+      handle $g
+    done
+  else
+    echo "Could not open $f"
+    exit 1
+  fi
+done

test/mzcat-checksums-of-data.txt

@@ -0,0 +1,19 @@
+# Generated by script/print-mzcat-checksums.sh
+29058c73 test/data/256.bytes.bz2
+29058c73 test/data/256.bytes.gz
+6ddef9db test/data/abraca.txt.bz2
+b19df306 test/data/archive.tar.bz2
+b19df306 test/data/archive.tar.gz
+6ddef9db test/data/artificial-bzip2/abraca.txt.bz2
+6ddef9db test/data/artificial-bzip2/huffman-258.bz2
+acef87f1 test/data/enwik5.lzma
+acef87f1 test/data/enwik5.xz
+3db2cdc6 test/data/midsummer.txt.bz2
+3db2cdc6 test/data/midsummer.txt.gz
+519e8b87 test/data/pi.txt.bz2
+519e8b87 test/data/pi.txt.gz
+abe507ef test/data/romeo.txt.bz2
+abe507ef test/data/romeo.txt.gz
+abe507ef test/data/romeo.txt.litonlylzma
+abe507ef test/data/romeo.txt.lzma
+abe507ef test/data/romeo.txt.xz