From f1a25f959e08888a6adc6f0c738c9c53f1c4866d Mon Sep 17 00:00:00 2001
From: Leopere <colin@nixc.us>
Date: Thu, 20 Mar 2025 22:25:37 -0400
Subject: [PATCH] Add randomized data to tests for better verification

---
 test/test.sh | 75 +++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 62 insertions(+), 13 deletions(-)

diff --git a/test/test.sh b/test/test.sh
index 13190be..90bbcec 100755
--- a/test/test.sh
+++ b/test/test.sh
@@ -6,6 +6,19 @@ TEST_TYPE=${1:-"all"}
 # Set to 1 to keep temporary files, 0 to clean them up
 KEEP_TEMP_FILES=${2:-0}
 
+# Function to generate a random file with predictable size
+generate_random_file() {
+  local OUTPUT=$1
+  local SIZE=$2
+  dd if=/dev/urandom bs=1 count=$SIZE of="$OUTPUT" 2>/dev/null
+}
+
+# Generate random string (for file contents)
+generate_random_string() {
+  local LENGTH=$1
+  tr -dc A-Za-z0-9 </dev/urandom | head -c $LENGTH
+}
+
 cleanup_files() {
   if [ "$KEEP_TEMP_FILES" -eq 0 ]; then
     echo "=== CLEANING UP TEMPORARY FILES ==="
@@ -32,13 +45,18 @@ run_basic_test() {
   mkdir -p /tmp/complex/dir1/subdir2
   mkdir -p /tmp/complex/dir2/subdir1
 
-  # Create files at different levels
-  echo 'root level file' > /tmp/complex/rootfile.txt
-  echo 'level 1 file in dir1' > /tmp/complex/dir1/file1.txt
-  echo 'level 1 file in dir2' > /tmp/complex/dir2/file2.txt
-  echo 'level 2 file in subdir1' > /tmp/complex/dir1/subdir1/file3.txt
-  echo 'level 2 file in subdir2' > /tmp/complex/dir1/subdir2/file4.txt
-  echo 'level 3 file in subsubdir1' > /tmp/complex/dir1/subdir1/subsubdir1/file5.txt
+  # Create files at different levels with random content
+  echo "root level file ($(generate_random_string 8))" > /tmp/complex/rootfile.txt
+  echo "level 1 file in dir1 ($(generate_random_string 12))" > /tmp/complex/dir1/file1.txt
+  echo "level 1 file in dir2 ($(generate_random_string 16))" > /tmp/complex/dir2/file2.txt
+  echo "level 2 file in subdir1 ($(generate_random_string 10))" > /tmp/complex/dir1/subdir1/file3.txt
+  echo "level 2 file in subdir2 ($(generate_random_string 14))" > /tmp/complex/dir1/subdir2/file4.txt
+  echo "level 3 file in subsubdir1 ($(generate_random_string 20))" > /tmp/complex/dir1/subdir1/subsubdir1/file5.txt
+  
+  # Add random binary files of different sizes
+  generate_random_file "/tmp/complex/random_binary_small.bin" 512
+  generate_random_file "/tmp/complex/dir1/random_binary_medium.bin" 2048
+  generate_random_file "/tmp/complex/dir2/random_binary_large.bin" 8192
 
   # Create a symbolic link with a relative path instead of absolute
   cd /tmp/complex/dir2 && ln -s ../rootfile.txt symlink.txt && cd /workdir
@@ -76,6 +94,17 @@ run_basic_test() {
   cat /tmp/complex-extracted/complex-app/dir1/subdir1/subsubdir1/file5.txt
   echo ' <-- Extracted: deep nested file5.txt'
 
+  # Verify binary file MD5 hashes specifically
+  echo '=== VERIFYING BINARY FILE MD5 HASHES ==='
+  md5sum /tmp/complex/random_binary_small.bin
+  md5sum /tmp/complex-extracted/complex-app/random_binary_small.bin
+  
+  md5sum /tmp/complex/dir1/random_binary_medium.bin
+  md5sum /tmp/complex-extracted/complex-app/dir1/random_binary_medium.bin
+  
+  md5sum /tmp/complex/dir2/random_binary_large.bin
+  md5sum /tmp/complex-extracted/complex-app/dir2/random_binary_large.bin
+
   # Test symlink
   echo '=== TESTING SYMLINK ==='
   ls -la /tmp/complex/dir2/symlink.txt
@@ -144,12 +173,32 @@ run_tar_comparison_test() {
   mkdir -p /tmp/standard-test/data/user/pictures
   mkdir -p /tmp/standard-test/logs
 
-  # Create various file types
-  echo '{"app": "tarballer", "version": "1.0"}' > /tmp/standard-test/config/settings/app.json
-  echo 'debug=true' > /tmp/standard-test/config/settings/debug.conf
-  dd if=/dev/urandom bs=1K count=10 of=/tmp/standard-test/data/user/documents/binary.dat 2>/dev/null
-  echo 'Test log entry 1' > /tmp/standard-test/logs/app.log
-  echo 'Test log entry 2' >> /tmp/standard-test/logs/app.log
+  # Create various file types with random content
+  echo "{\"app\": \"tarballer\", \"version\": \"1.0\", \"random_data\": \"$(generate_random_string 32)\"}" > /tmp/standard-test/config/settings/app.json
+  echo "debug=true\nlog_level=$(generate_random_string 5)\ndate_format=\"$(generate_random_string 8)\"" > /tmp/standard-test/config/settings/debug.conf
+  
+  # Create binary files of different sizes
+  generate_random_file "/tmp/standard-test/data/user/documents/binary.dat" 10240
+  generate_random_file "/tmp/standard-test/data/user/pictures/image1.raw" 5120
+  generate_random_file "/tmp/standard-test/data/user/pictures/image2.raw" 7168
+  
+  # Create log files with random entries
+  echo "Test log entry 1 - $(generate_random_string 16)" > /tmp/standard-test/logs/app.log
+  echo "Test log entry 2 - $(generate_random_string 24)" >> /tmp/standard-test/logs/app.log
+  echo "Test log entry 3 - $(generate_random_string 20)" >> /tmp/standard-test/logs/app.log
+  
+  # Create text file with random data
+  generate_random_string 1024 > /tmp/standard-test/data/user/documents/text_file.txt
+  
+  # Create config file with mixed content
+  cat << EOF > /tmp/standard-test/config/mixed_content.conf
+# Configuration file with mixed content
+SERVER_NAME=$(generate_random_string 12)
+PORT=8080
+MAX_CONNECTIONS=100
+TIMEOUT=30
+RANDOM_SEED=$(generate_random_string 64)
+EOF
 
   # Create symlinks
   ln -s ../config/settings/app.json /tmp/standard-test/data/config-link.json