From ce2300b1163cbcab13b478902ee8e2f7a515ceca Mon Sep 17 00:00:00 2001
From: Martin Michelsen <fuzziqersoftware+g@gmail.com>
Date: Sun, 24 Mar 2024 21:59:28 -0700
Subject: [PATCH] add pessimal compression

---
 src/Compression.cc   | 39 +++++++++++++++++++++++++++++++++++++++
 src/Compression.hh   |  4 ++++
 src/Main.cc          |  3 +++
 tests/compression.sh | 12 ++++++++++++
 4 files changed, 58 insertions(+)

diff --git a/src/Compression.cc b/src/Compression.cc
index 75262373..6d88ea6f 100644
--- a/src/Compression.cc
+++ b/src/Compression.cc
@@ -438,6 +438,45 @@ string prs_compress_optimal(const string& data, ProgressCallback progress_fn) {
   return prs_compress_optimal(data.data(), data.size(), progress_fn);
 }
 
+string prs_compress_pessimal(const void* vdata, size_t size) {
+  const uint8_t* in_data = reinterpret_cast<const uint8_t*>(vdata);
+
+  // The worst possible encoding we can do is a literal byte when no byte with
+  // the same value is within the window, or an extended copy if there is a byte
+  // with the same value in the window.
+  WindowIndex<0x1FFF, 1> window(in_data, size);
+  LZSSInterleavedWriter w;
+  for (size_t z = 0; z < size; z++) {
+    auto match = window.get_best_match();
+    if (match.second >= 1) {
+      // Write extended copy
+      int16_t offset = match.first - window.offset;
+      w.write_control(false);
+      w.flush_if_ready();
+      w.write_control(true);
+      uint16_t a = (offset << 3);
+      w.write_data(a & 0xFF);
+      w.write_data(a >> 8);
+      w.write_data(0);
+    } else {
+      // Write literal
+      w.write_control(true);
+      w.write_data(in_data[z]);
+    }
+    w.flush_if_ready();
+    window.advance();
+  }
+
+  // Write stop command
+  w.write_control(false);
+  w.flush_if_ready();
+  w.write_control(true);
+  w.write_data(0);
+  w.write_data(0);
+
+  return std::move(w.close());
+}
+
 PRSCompressor::PRSCompressor(
     ssize_t compression_level, ProgressCallback progress_fn)
     : compression_level(compression_level),
diff --git a/src/Compression.hh b/src/Compression.hh
index 684691a5..efbd46a9 100644
--- a/src/Compression.hh
+++ b/src/Compression.hh
@@ -177,6 +177,10 @@ std::string prs_compress_indexed(
 std::string prs_compress_optimal(const void* vdata, size_t size, ProgressCallback progress_fn = nullptr);
 std::string prs_compress_optimal(const std::string& data, ProgressCallback progress_fn = nullptr);
 
+// Compresses data using PRS to the LARGEST possible output size. There is no
+// practical use for this function except for amusement.
+std::string prs_compress_pessimal(const void* vdata, size_t size);
+
 // Decompresses PRS-compressed data.
 struct PRSDecompressResult {
   std::string data;
diff --git a/src/Main.cc b/src/Main.cc
index 0ff3803d..1bf0b2b3 100644
--- a/src/Main.cc
+++ b/src/Main.cc
@@ -266,6 +266,7 @@ static void a_compress_decompress_fn(Arguments& args) {
   bool is_decompress = starts_with(action, "decompress-");
   bool is_big_endian = args.get<bool>("big-endian");
   bool is_optimal = args.get<bool>("optimal");
+  bool is_pessimal = args.get<bool>("pessimal");
   int8_t compression_level = args.get<int8_t>("compression-level", 0);
   size_t bytes = args.get<size_t>("bytes", 0);
   string seed = args.get<string>("seed");
@@ -298,6 +299,8 @@ static void a_compress_decompress_fn(Arguments& args) {
   if (!is_decompress && (is_prs || is_pr2 || is_prc)) {
     if (is_optimal) {
       data = prs_compress_optimal(data.data(), data.size(), optimal_progress_fn);
+    } else if (is_pessimal) {
+      data = prs_compress_pessimal(data.data(), data.size());
     } else {
       data = prs_compress(data, compression_level, progress_fn);
     }
diff --git a/tests/compression.sh b/tests/compression.sh
index ebe8e228..1ef63939 100755
--- a/tests/compression.sh
+++ b/tests/compression.sh
@@ -20,6 +20,10 @@ echo "... compress with level=0"
 $EXECUTABLE compress-$SCHEME --compression-level=0 $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.l0
 echo "... compress with level=1"
 $EXECUTABLE compress-$SCHEME --compression-level=1 $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.l1
+echo "... compress optimally"
+$EXECUTABLE compress-$SCHEME --optimal $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.lo
+echo "... compress pessimally"
+$EXECUTABLE compress-$SCHEME --pessimal $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.lp
 
 echo "... decompress from level=-1 (no compression)"
 $EXECUTABLE decompress-$SCHEME $BASENAME.mnrd.$SCHEME.lN $BASENAME.mnrd.$SCHEME.lN.dec
@@ -27,6 +31,10 @@ echo "... decompress from level=0"
 $EXECUTABLE decompress-$SCHEME $BASENAME.mnrd.$SCHEME.l0 $BASENAME.mnrd.$SCHEME.l0.dec
 echo "... decompress from level=1"
 $EXECUTABLE decompress-$SCHEME $BASENAME.mnrd.$SCHEME.l1 $BASENAME.mnrd.$SCHEME.l1.dec
+echo "... decompress from optimal"
+$EXECUTABLE decompress-$SCHEME $BASENAME.mnrd.$SCHEME.lo $BASENAME.mnrd.$SCHEME.lo.dec
+echo "... decompress from pessimal"
+$EXECUTABLE decompress-$SCHEME $BASENAME.mnrd.$SCHEME.lp $BASENAME.mnrd.$SCHEME.lp.dec
 
 echo "... check result from level=-1 (no compression)"
 diff $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.lN.dec
@@ -34,6 +42,10 @@ echo "... check result from level=0"
 diff $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.l0.dec
 echo "... check result from level=1"
 diff $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.l1.dec
+echo "... check result from optimal"
+diff $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.lo.dec
+echo "... check result from pessimal"
+diff $BASENAME.mnrd $BASENAME.mnrd.$SCHEME.lp.dec
 
 echo "... clean up"
 rm $BASENAME.mnrd \