Add static isAscii checks to FileDescriptor class. Refs #7263

Code/Mantid/Framework/Kernel/inc/MantidKernel/FileDescriptor.h

@@ -40,6 +40,12 @@ namespace Mantid
      */
     class MANTID_KERNEL_DLL FileDescriptor
     {
+    public:
+      /// Returns true if the file is considered ascii
+      static bool isAscii(const std::string & filename, const size_t nbytes=256);
+      /// Returns true if the stream is considered ascii
+      static bool isAscii(std::istream & data, const size_t nbytes=256);
+
     public:
       /// Constructor accepting a filename
       FileDescriptor(const std::string & filename);

Code/Mantid/Framework/Kernel/src/FileDescriptor.cpp

@@ -8,6 +8,67 @@ namespace Mantid
 {
   namespace Kernel
   {
+    //----------------------------------------------------------------------------------------------
+    // Public static methods
+    //----------------------------------------------------------------------------------------------
+    /**
+     * Check if the given file appears to be an ASCII file. The check searches the
+     * first nbytes of the file and returns false if a non-ascii character is found.
+     * If the file is shorter than nbytes then it checks until the end of the stream.
+     * @param filename A string pointing to an existing file
+     * @param bytes The number of bytes of the file to check (Default=256)
+     * @returns True if the file is considered ASCII, false otherwise
+     * @throws std::invalid_argument if the file cannot be opened
+     * @throws std::runtime_error if an error is occurred while reading the stream
+     */
+    bool FileDescriptor::isAscii(const std::string & filename, const size_t nbytes)
+    {
+      std::ifstream data(filename.c_str(), std::ios::in | std::ios::binary);
+      if(!data)
+      {
+        throw std::invalid_argument("FileDescriptor::isAscii() - Unable to open file '"
+             + filename + "'");
+      }
+      return FileDescriptor::isAscii(data, nbytes);
+    }
+
+    /**
+     * Check if the given stream appears to point to an ASCII data. The check searches the
+     * next nbytes of the stream and returns false if a non-ascii character is found.
+     * If the stream is shorter than nbytes or a reading error occurs then it simply returns
+     * the result up to that point
+     * The stream is reset to the position is was at when entering the function
+     * @param data An input stream opened in binary mode
+     * @param bytes The number of bytes of the file to check (Default=256)
+     * @returns True if the stream is considered ASCII, false otherwise
+     */
+    bool FileDescriptor::isAscii(std::istream & data, const size_t nbytes)
+    {
+      const std::streampos startPos = data.tellg();
+      char byte('\0');
+      size_t counter(0);
+      bool result(true);
+      while(counter < nbytes)
+      {
+        data >> byte;
+        if(!data) // reading error
+        {
+          data.clear();
+          break;
+        }
+        unsigned long ch = static_cast<unsigned long>(byte);
+        if( !(ch <= 0x7F) ) // non-ascii
+        {
+          result = false;
+          break;
+        }
+        ++counter;
+      }
+
+      data.seekg(startPos);
+      return result;
+    }
+
     //----------------------------------------------------------------------------------------------
     // Public methods
     //----------------------------------------------------------------------------------------------

Code/Mantid/Framework/Kernel/test/FileDescriptorTest.h

@@ -29,17 +29,51 @@ class FileDescriptorTest : public CxxTest::TestSuite
       if(Poco::File(nxsPath).exists()) m_testNexusPath = nxsPath.toString();
       Poco::Path nonNxsPath(*it, "CSP79590.raw");
       if(Poco::File(nonNxsPath).exists()) m_testNonNexusPath = nonNxsPath.toString();
+      Poco::Path asciiPath(*it, "AsciiExample.txt");
+      if(Poco::File(asciiPath).exists()) m_testAsciiPath = asciiPath.toString();
 
-      if(!m_testNexusPath.empty() && !m_testNonNexusPath.empty()) break;
+      if(!m_testNexusPath.empty() && !m_testNonNexusPath.empty() &&
+         !m_testAsciiPath.empty()) break;
     }
-    if(m_testNexusPath.empty() || m_testNonNexusPath.empty())
+    if(m_testNexusPath.empty() || m_testNonNexusPath.empty() || m_testAsciiPath.empty())
     {
       throw std::runtime_error("Unable to find test files for FileDescriptorTest. "
           "The AutoTestData directory needs to be in the search path");
     }
   }
 
   //===================== Success cases ============================================
+  void test_isAscii_Returns_True_For_Ascii_Filename()
+  {
+    TS_ASSERT(FileDescriptor::isAscii(m_testAsciiPath));
+  }
+
+  void test_isAscii_Returns_False_For_Binary_Filename()
+  {
+    TS_ASSERT(!FileDescriptor::isAscii(m_testNonNexusPath));
+  }
+
+  void test_isAscii_Returns_True_For_Stream_Pointing_At_Ascii_File_And_Stream_Is_Returned_To_Position_On_Entry()
+  {
+    std::ifstream is(m_testAsciiPath, std::ios::in|std::ios::binary);
+    // move stream along one to check it is returned to here
+    is.seekg(1);
+
+    TS_ASSERT(FileDescriptor::isAscii(is));
+    TS_ASSERT_EQUALS(1, is.tellg());
+  }
+
+  void test_isAscii_Returns_False_For_Stream_Pointing_At_Ascii_File_And_Stream_Is_Returned_To_Position_On_Entry()
+  {
+    std::ifstream is(m_testNonNexusPath, std::ios::in|std::ios::binary);
+    // move stream along one to check it is returned to here
+    is.seekg(1);
+
+    TS_ASSERT(!FileDescriptor::isAscii(is));
+    TS_ASSERT_EQUALS(1, is.tellg());
+  }
+
+
   void test_Constructor_With_Existing_File_Initializes_Description_Fields()
   {
     const std::string filename = m_testNexusPath;
@@ -80,6 +114,27 @@ class FileDescriptorTest : public CxxTest::TestSuite
   }
 
   //===================== Failure cases ============================================
+  void test_IsAscii_Throws_For_Inaccessible_Filename()
+  {
+    TS_ASSERT_THROWS(FileDescriptor::isAscii(""), std::invalid_argument);
+    TS_ASSERT_THROWS(FileDescriptor::isAscii("__not_a_File.txt__"), std::invalid_argument);
+  }
+
+  void test_IsAscii_Returns_True_For_Ascii_Stream_Shorter_Than_NBytes_Requested_And_Clears_Error_Flags()
+  {
+    // Fake data
+    std::istringstream is;
+    is.str("abcdef"); // 6 bytes
+
+    TS_ASSERT(FileDescriptor::isAscii(is, 6)); // Equal to length
+    TS_ASSERT_EQUALS(0, is.tellg());
+    TS_ASSERT(is.good());
+
+    TS_ASSERT(FileDescriptor::isAscii(is, 10)); // Larger
+    TS_ASSERT_EQUALS(0, is.tellg());
+    TS_ASSERT(is.good());
+  }
+
   void test_Constructor_Throws_With_Empty_filename()
   {
     TS_ASSERT_THROWS(FileDescriptor(""), std::invalid_argument);
@@ -93,6 +148,7 @@ class FileDescriptorTest : public CxxTest::TestSuite
 private:
   std::string m_testNexusPath;
   std::string m_testNonNexusPath;
+  std::string m_testAsciiPath;
 };