From 67a88aaf91040ce7b685b1b67471ef1204eef768 Mon Sep 17 00:00:00 2001
From: Phuntsok Drak-pa <phundrak@phundrak.fr>
Date: Mon, 11 Jun 2018 21:02:59 +0200
Subject: [PATCH] bug fixed as input stream was char and not unsigned char

---
 src/bitpack.cc  |  5 ++++-
 src/common.cc   | 11 ++++++++++-
 src/compress.cc | 26 ++++++++++++++++----------
 src/compress.hh |  3 ++-
 4 files changed, 32 insertions(+), 13 deletions(-)

diff --git a/src/bitpack.cc b/src/bitpack.cc
index d9f4421..fcc819f 100644
--- a/src/bitpack.cc
+++ b/src/bitpack.cc
@@ -29,7 +29,10 @@ using ustring = std::basic_string<unsigned char>;
  */
 [[nodiscard]] vuchar
 pack_n(const vuint16::const_iterator t_input_begin,
-       const vuint16::const_iterator t_input_end, const int t_n) {
+           const vuint16::const_iterator t_input_end, const int t_n) {
+#ifdef Debug
+  std::printf("%d bits!\n", t_n);
+#endif
   if (t_n == 16) {
     return pack_16(t_input_begin, t_input_end);
   }
diff --git a/src/common.cc b/src/common.cc
index 056a01d..3124caf 100644
--- a/src/common.cc
+++ b/src/common.cc
@@ -65,8 +65,11 @@ ustring dico_uncompress(std::map<uint16_t, ustring> &t_dict,
                         const uint16_t t_code, const uint16_t t_old) {
   // le code existe dans le dictionnaire s’il est < 256
   if (t_code < 256) {
-
+#ifdef Debug
+    std::printf("Code : %d\tNOT EMPTY\n", t_code);
+#endif
     ustring e{static_cast<unsigned char>(t_code)};
+    // 256 car on n'a pas encore tenté d'insérer de nouveau caractère
     if (t_old < 256) {
       t_dict[static_cast<uint16_t>(t_dict.size() + 256)] =
           static_cast<unsigned char>(t_old) + e;
@@ -82,6 +85,9 @@ ustring dico_uncompress(std::map<uint16_t, ustring> &t_dict,
 
   // le code existe dans le dictionnaire
   if (!e.empty()) {
+#ifdef Debug
+    std::printf("Code : %d\tNOT EMPTY\n", t_code);
+#endif
     str += e[0];
     const uint16_t index = static_cast<uint16_t>(t_dict.size() + 256);
     t_dict[index] = str;
@@ -89,6 +95,9 @@ ustring dico_uncompress(std::map<uint16_t, ustring> &t_dict,
   }
 
   // le code n'existe pas encore dans le dictionnaire
+#ifdef Debug
+  std::printf("Code : %d\tEMPTY\n", t_code);
+#endif
   str += str[0];
   e = str;
   t_dict[t_code] = e;
diff --git a/src/compress.cc b/src/compress.cc
index a8ea9cf..50360ec 100644
--- a/src/compress.cc
+++ b/src/compress.cc
@@ -15,7 +15,7 @@ using std::uint8_t;
 using std::vector;
 using vuint16 = vector<uint16_t>;
 using vvuint16 = vector<vuint16>;
-using std::string;
+using ustring = std::basic_string<unsigned char>;
 
 using dict_t = std::map<std::pair<uint16_t, uint8_t>, uint16_t>;
 using std::printf;
@@ -33,7 +33,7 @@ const size_t DICT_MAX = static_cast<size_t>(ipow(2, 17) - 256); /* 16 bits */
  *  \param t_text Chaîne de caractères uint8_t représentant le fichier d'entrée
  *  \return Vecteur de chunks (vecteurs de uint16_t)
  */
-vvuint16 lzw_compress(string &&t_text) {
+vvuint16 lzw_compress(ustring &&t_text) {
   std::puts("Compressing...");
   uint16_t w = 0xFFFF;
   vuint16 chunk{};
@@ -49,7 +49,7 @@ vvuint16 lzw_compress(string &&t_text) {
     if (const auto &[yes, pos] = dico(dict, w, static_cast<uint8_t>(c)); yes) {
       w = pos;
     } else {
-      chunk.push_back(static_cast<uint16_t>(w));
+      chunk.push_back(w);
       w = static_cast<uint16_t>(c);
     }
   }
@@ -70,20 +70,26 @@ vvuint16 lzw_compress(string &&t_text) {
  *  \param[in] t_out_file Chemin vers le fichier de sortie
  */
 void compress(const std::string &t_in_file, const char *t_out_file) {
-  std::ifstream input_file{t_in_file};
-  assert(input_file.is_open());
+  FILE *const input_file = fopen(t_in_file.c_str(), "rb");
+  assert(input_file);
   FILE *const out = (t_out_file != nullptr) ? fopen(t_out_file, "wb")
                                             : fopen("output.lzw", "wb");
   if (out == nullptr) {
     std::cerr << "Error at " << __FILE__ << ":" << __LINE__ - 4
               << ": could not open output file. Aborting...\n";
-    input_file.close();
+    // input_file.close();
+    std::fclose(input_file);
     exit(1);
   }
-  const auto compressed_text{
-      lzw_compress(std::string{std::istreambuf_iterator<char>(input_file),
-                               std::istreambuf_iterator<char>()})};
+
+  std::fseek(input_file, 0L, SEEK_END);
+  const auto file_size = static_cast<size_t>(ftell(input_file));
+  std::rewind(input_file);
+
+  auto raw_text = std::make_unique<unsigned char[]>(file_size);
+  std::fread(raw_text.get(), sizeof(unsigned char), file_size, input_file);
+  const auto compressed_text(lzw_compress(ustring{raw_text.get(), &raw_text[file_size]}));
   write_file(out, compressed_text);
   fclose(out);
-  input_file.close();
+  fclose(input_file);
 }
diff --git a/src/compress.hh b/src/compress.hh
index 7e0089d..4aa4c75 100644
--- a/src/compress.hh
+++ b/src/compress.hh
@@ -12,7 +12,8 @@
 #include <thread>
 
 /// \brief Compression d'une chaine de caractères
-std::vector<std::vector<std::uint16_t>> lzw_compress(std::string &&);
+std::vector<std::vector<std::uint16_t>>
+lzw_compress(std::basic_string<unsigned char> &&);
 
 /// \brief Wrapper de \ref lzw_compress
 void compress(const std::string &, const char *);