Vutils

vic4key · Sep 14, 2023 · 647e726 · 647e726
1 parent 95b445d
commit 647e726
Show file tree

Hide file tree

Showing 14 changed files with 93 additions and 63 deletions.
diff --git a/Test/Sample.AsyncSocket.h b/Test/Sample.AsyncSocket.h
@@ -32,7 +32,7 @@ void example_binding(const vu::Socket::Endpoint& endpoint)
   {
     vu::Buffer data(KiB);
     client.recv(data);
-    printf("client %d recv `%s`\n", client.get_remote_sai().sin_port, data.to_string_A().c_str());
+    printf("client %d recv `%s`\n", client.get_remote_sai().sin_port, data.to_string_A()->c_str());
   });
 
   server.bind(endpoint);

diff --git a/Test/Sample.Misc.h b/Test/Sample.Misc.h
@@ -154,22 +154,22 @@ DEF_SAMPLE(Misc)
   std::string s = "0123456789";
   vu::Buffer  slicer(s.data(), s.size());
 
-  std::cout << slicer(0, 0).to_string_A() << std::endl;
-  std::cout << slicer(9, 9).to_string_A() << std::endl;
-  std::cout << slicer(0, 9).to_string_A() << std::endl;
+  //std::cout << slicer(0, 0)->to_string_A()->c_str() << std::endl;
+  //std::cout << slicer(9, 9)->to_string_A()->c_str() << std::endl;
+  std::cout << slicer(0, 9)->to_string_A()->c_str() << std::endl;
 
-  std::cout << slicer(3, 7).to_string_A() << std::endl;
+  std::cout << slicer(3, 7)->to_string_A()->c_str() << std::endl;
 
-  std::cout << slicer(-7, -2).to_string_A() << std::endl;
-  std::cout << slicer(2, -2).to_string_A() << std::endl;
+  std::cout << slicer(-7, -2)->to_string_A()->c_str() << std::endl;
+  std::cout << slicer(2, -2)->to_string_A()->c_str() << std::endl;
 
-  std::cout << slicer(0, -8).to_string_A() << std::endl;
-  std::cout << slicer(0, -9).to_string_A() << std::endl;
-  std::cout << slicer(0, -10).to_string_A() << std::endl;
+  std::cout << slicer(0, -8)->to_string_A()->c_str() << std::endl;
+  std::cout << slicer(0, -9)->to_string_A()->c_str() << std::endl;
+  //std::cout << slicer(0, -10)->to_string_A()->c_str() << std::endl;
 
-  std::cout << slicer(-8, 10).to_string_A() << std::endl;
-  std::cout << slicer(-9, 10).to_string_A() << std::endl;
-  std::cout << slicer(-10, 10).to_string_A() << std::endl;
+  std::cout << slicer(-8, 10)->to_string_A()->c_str() << std::endl;
+  std::cout << slicer(-9, 10)->to_string_A()->c_str() << std::endl;
+  std::cout << slicer(-10, 10)->to_string_A()->c_str() << std::endl;
 
   std::tcout << vu::undecorate_cpp_symbol(ts("?func1@a@@AAEXH@Z")) << std::endl;
 

diff --git a/Test/Sample.Socket.h b/Test/Sample.Socket.h
@@ -50,7 +50,7 @@ DEF_SAMPLE(Socket)
 
   // extract response header & body
 
-  const std::string first_response = response.to_string_A();
+  const std::string first_response = response.to_string_A()->c_str();
   const std::string http_header_sep = std::crlf;
   const std::string http_header_end = http_header_sep + http_header_sep;
 

diff --git a/Test/Text.Encoding/ansi.txt b/Test/Text.Encoding/ansi.txt
@@ -0,0 +1 @@
+this is an ansi text
diff --git a/Test/Text.Encoding/utf-16-be-bom.txt b/Test/Text.Encoding/utf-16-be-bom.txt
diff --git a/Test/Text.Encoding/utf-16-be.txt b/Test/Text.Encoding/utf-16-be.txt
diff --git a/Test/Text.Encoding/utf-16-le-bom.txt b/Test/Text.Encoding/utf-16-le-bom.txt
diff --git a/Test/Text.Encoding/utf-16-le.txt b/Test/Text.Encoding/utf-16-le.txt
diff --git a/Test/Text.Encoding/utf-8-bom.txt b/Test/Text.Encoding/utf-8-bom.txt
@@ -0,0 +1 @@
+đây là văn bản utf-8 bom
diff --git a/Test/Text.Encoding/utf-8.txt b/Test/Text.Encoding/utf-8.txt
@@ -0,0 +1 @@
+đây là văn bản utf-8
diff --git a/include/Vutils.h b/include/Vutils.h
@@ -1029,7 +1029,7 @@ class Buffer
   bool  operator==(const Buffer& right) const;
   bool  operator!=(const Buffer& right) const;
   byte& operator[](const size_t offset);
-  Buffer operator()(int begin, int end) const;
+  std::unique_ptr<Buffer> operator()(int begin, int end) const;
 
   byte*  bytes() const;
   void*  pointer() const;
@@ -1044,14 +1044,14 @@ class Buffer
   bool replace(const Buffer& right);
   bool match(const void* ptr, const size_t size) const;
   size_t find(const void* ptr, const size_t size) const;
-  Buffer till(const void* ptr, const size_t size) const;
-  Buffer slice(int begin, int end) const;
+  std::unique_ptr<Buffer> till(const void* ptr, const size_t size) const;
+  std::unique_ptr<Buffer> slice(int begin, int end) const;
 
   bool append(const void* ptr, const size_t size);
   bool append(const Buffer& right);
 
-  std::string  to_string_A() const;
-  std::wstring to_string_W() const;
+  std::unique_ptr<std::string>  to_string_A() const;
+  std::unique_ptr<std::wstring> to_string_W() const;
 
   bool save_to_file(const std::string&  file_path);
   bool save_to_file(const std::wstring& file_path);
@@ -1718,9 +1718,9 @@ class FileSystemA : public FileSystemX
     fs_share fs_flags = fs_share::FS_ALLACCESS,
     fs_attribute fa_flags = fs_attribute::FA_NORMAL);
 
-  std::unique_ptr<std::string> vuapi read_as_string();
+  std::unique_ptr<std::string> vuapi read_as_text();
 
-  static std::unique_ptr<std::string> vuapi quick_read_as_string(const std::string& file_path);
+  static std::unique_ptr<std::string> vuapi quick_read_as_text(const std::string& file_path);
 
   static std::unique_ptr<Buffer> quick_read_as_buffer(const std::string& file_path);
 
@@ -1749,9 +1749,9 @@ class FileSystemW : public FileSystemX
     fs_share fs_flags = fs_share::FS_ALLACCESS,
     fs_attribute fa_flags = fs_attribute::FA_NORMAL);
 
-  std::unique_ptr<std::wstring> vuapi read_as_string();
+  std::unique_ptr<std::wstring> vuapi read_as_text();
 
-  static std::unique_ptr<std::wstring>vuapi quick_read_as_string(const std::wstring& file_path);
+  static std::unique_ptr<std::wstring>vuapi quick_read_as_text(const std::wstring& file_path);
 
   static std::unique_ptr<Buffer> vuapi quick_read_as_buffer(const std::wstring& file_path);
 

diff --git a/src/details/filesys.cpp b/src/details/filesys.cpp
@@ -212,7 +212,7 @@ bool vuapi FileSystemA::initialize(
   return true;
 }
 
-std::unique_ptr<std::string> vuapi FileSystemA::read_as_string()
+std::unique_ptr<std::string> vuapi FileSystemA::read_as_text()
 {
   auto buffer = this->read_as_buffer();
   if (buffer == nullptr)
@@ -254,10 +254,10 @@ std::unique_ptr<std::string> vuapi FileSystemA::read_as_string()
   return result;
 }
 
-std::unique_ptr<std::string> vuapi FileSystemA::quick_read_as_string(const std::string& file_path)
+std::unique_ptr<std::string> vuapi FileSystemA::quick_read_as_text(const std::string& file_path)
 {
   FileSystemA file(file_path, vu::fs_mode::FM_OPENEXISTING, fs_generic::FG_READ, fs_share::FS_READ);
-  return file.read_as_string();
+  return file.read_as_text();
 }
 
 std::unique_ptr<Buffer> FileSystemA::quick_read_as_buffer(const std::string& file_path)
@@ -368,7 +368,7 @@ bool vuapi FileSystemW::initialize(
   return true;
 }
 
-std::unique_ptr<std::wstring> vuapi FileSystemW::read_as_string()
+std::unique_ptr<std::wstring> vuapi FileSystemW::read_as_text()
 {
   auto buffer = this->read_as_buffer();
   if (buffer == nullptr)
@@ -410,10 +410,10 @@ std::unique_ptr<std::wstring> vuapi FileSystemW::read_as_string()
   return result;
 }
 
-std::unique_ptr<std::wstring> vuapi FileSystemW::quick_read_as_string(const std::wstring& file_path)
+std::unique_ptr<std::wstring> vuapi FileSystemW::quick_read_as_text(const std::wstring& file_path)
 {
   FileSystemW file(file_path, vu::fs_mode::FM_OPENEXISTING, fs_generic::FG_READ, fs_share::FS_READ);
-  return file.read_as_string();
+  return file.read_as_text();
 }
 
 std::unique_ptr<Buffer> FileSystemW::quick_read_as_buffer(const std::wstring& file_path)

diff --git a/src/details/mbuffer.cpp b/src/details/mbuffer.cpp
@@ -80,7 +80,7 @@ byte& Buffer::operator[](const size_t offset)
   return static_cast<byte*>(m_ptr)[offset];
 }
 
-Buffer Buffer::operator()(int begin, int end) const
+std::unique_ptr<Buffer> Buffer::operator()(int begin, int end) const
 {
   return this->slice(begin, end);
 }
@@ -113,26 +113,24 @@ bool Buffer::match(const void* ptr, const size_t size) const
   return this->find(ptr, size) != -1;
 }
 
-Buffer Buffer::till(const void* ptr, const size_t size) const
+std::unique_ptr<Buffer> Buffer::till(const void* ptr, const size_t size) const
 {
-  Buffer result;
-
   size_t offset = this->find(ptr, size);
-  if (offset > 0)
+  if (offset == 0)
   {
-    result.create(m_ptr, offset);
+    return nullptr;
   }
 
+  std::unique_ptr<Buffer> result(new Buffer);
+  result->create(m_ptr, offset);
   return result;
 }
 
-Buffer Buffer::slice(int begin, int end) const
+std::unique_ptr<Buffer> Buffer::slice(int begin, int end) const
 {
-  Buffer result;
-
   if (m_ptr == nullptr || m_size == 0)
   {
-    return result;
+    return nullptr;
   }
 
   if (begin < 0)
@@ -147,17 +145,18 @@ Buffer Buffer::slice(int begin, int end) const
 
   if (begin < 0 || end < 0 || begin > int(m_size) || end > int(m_size) || begin > end)
   {
-    return result;
+    return nullptr;
   }
 
   int size = end - begin;
 
   if (size <= 0 || size > int(m_size))
   {
-    return result;
+    return nullptr;
   }
 
-  result.create(this->bytes() + begin, size);
+  std::unique_ptr<Buffer> result(new Buffer);
+  result->create(this->bytes() + begin, size);
 
   return result;
 }
@@ -295,14 +294,28 @@ bool Buffer::append(const Buffer& right)
   return this->append(right.pointer(), right.size());
 }
 
-std::string Buffer::to_string_A() const
+std::unique_ptr<std::string> Buffer::to_string_A() const
 {
-  return std::string(reinterpret_cast<const char*>(m_ptr), m_size / sizeof(char));
+  std::unique_ptr<std::string> result(new std::string);
+  if (m_ptr == nullptr || m_size == 0)
+  {
+    return result;
+  }
+
+  result.reset(new std::string(reinterpret_cast<const char*>(m_ptr), m_size / sizeof(char)));
+  return result;
 }
 
-std::wstring Buffer::to_string_W() const
+std::unique_ptr<std::wstring> Buffer::to_string_W() const
 {
-  return std::wstring(reinterpret_cast<const wchar*>(m_ptr), m_size / sizeof(wchar));
+  std::unique_ptr<std::wstring> result(new std::wstring);
+  if (m_ptr == nullptr || m_size == 0)
+  {
+    return result;
+  }
+
+  result.reset(new std::wstring(reinterpret_cast<const wchar*>(m_ptr), m_size / sizeof(wchar)));
+  return result;
 }
 
 bool Buffer::save_to_file(const std::string& file_path)
@@ -323,8 +336,18 @@ bool Buffer::save_to_file(const std::string& file_path)
 
 bool Buffer::save_to_file(const std::wstring& file_path)
 {
-  const auto s = vu::to_string_A(file_path);
-  return this->save_to_file(s);
+  if (file_path.empty())
+  {
+    return false;
+  }
+
+  bool result = true;
+
+  FileSystemW file(file_path, fs_mode::FM_CREATEALWAY);
+  result &= file.write(m_ptr, ulong(m_size));
+  result &= file.close();
+
+  return result;
 }
 
 } // namespace vu
diff --git a/src/details/string.cpp b/src/details/string.cpp
@@ -76,22 +76,26 @@ text_encoding vuapi detect_text_encoding(const void* data, const size_t size)
       return text_encoding::TE_UTF16_BE_BOM;
     }
 
-    /* UTF-16 LE */
-    if ((p[0] >= SCHAR_MIN && p[0] <= SCHAR_MAX) && p[1] == 0x00)
+    /* Without BOM */
+    if (p[0] >= SCHAR_MIN && p[0] <= SCHAR_MAX)
     {
-      return text_encoding::TE_UTF16_LE;
-    }
-
-    /* UTF-16 BE */
-    if ((p[1] >= SCHAR_MIN && p[1] <= SCHAR_MAX) && p[0] == 0x00)
-    {
-      return text_encoding::TE_UTF16_BE;
-    }
-
-    /* UTF-8 */
-    if ((p[0] >= SCHAR_MIN && p[0] <= SCHAR_MAX) && (p[1] >= SCHAR_MIN && p[1] <= SCHAR_MAX))
-    {
-      return text_encoding::TE_UTF8;
+      // /* UTF-16 LE */
+      // ...
+      // {
+      //   return text_encoding::TE_UTF16_LE;
+      // }
+
+      // /* UTF-16 BE */
+      // ...
+      // {
+      //   return text_encoding::TE_UTF16_BE;
+      // }
+
+      /* UTF-8 */
+      if (p[1] >= SCHAR_MIN && p[1] <= SCHAR_MAX)
+      {
+        return text_encoding::TE_UTF8;
+      }
     }
   }
   catch (int)