robustness: throw exception on gzstream error

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@22 ec762483-ff6d-05da-a07a-a48fb63a330f

2 files changed, 35 insertions, 14 deletions

diff --git a/decoder/gzstream.cc b/decoder/gzstream.cc
index 9703e6ad..aa015bff 100644
--- a/decoder/gzstream.cc
+++ b/decoder/gzstream.cc
@@ -21,14 +21,15 @@
 // Revision      : $Revision: 1.1 $
 // Revision_date : $Date: 2006/03/30 04:05:52 $
 // Author(s)     : Deepak Bandyopadhyay, Lutz Kettner
-// 
-// Standard streambuf implementation following Nicolai Josuttis, "The 
+//
+// Standard streambuf implementation following Nicolai Josuttis, "The
 // Standard C++ Library".
 // ============================================================================
 
 #include "gzstream.h"
 #include <iostream>
 #include <cstring>
+#include <stdexcept>
 
 #ifdef GZSTREAM_NAMESPACE
 namespace GZSTREAM_NAMESPACE {
@@ -59,6 +60,7 @@ gzstreambuf* gzstreambuf::open( const char* name, int open_mode) {
     *fmodeptr++ = 'b';
     *fmodeptr = '\0';
     file = gzopen( name, fmode);
+    if (!file) handle_gzerror();
     if (file == 0)
         return (gzstreambuf*)0;
     opened = 1;
@@ -71,10 +73,19 @@ gzstreambuf * gzstreambuf::close() {
         opened = 0;
         if ( gzclose( file) == Z_OK)
             return this;
+        else
+            handle_gzerror();
     }
     return (gzstreambuf*)0;
 }
 
+void gzstreambuf::handle_gzerror() {
+    int errnum;
+    const char *errmsg=gzerror(file,&errnum);
+    if (errnum==Z_DATA_ERROR) errmsg="CRC error reading gzip";
+    throw std::runtime_error(std::string("gzstreambuf error: ")+errmsg);
+}
+
 int gzstreambuf::underflow() { // used for input buffer only
     if ( gptr() && ( gptr() < egptr()))
         return * reinterpret_cast<unsigned char *>( gptr());
@@ -89,7 +100,11 @@ int gzstreambuf::underflow() { // used for input buffer only
 
     int num = gzread( file, buffer+4, bufferSize-4);
     if (num <= 0) // ERROR or EOF
-        return EOF;
+    {
+        if (gzeof(file))
+            return EOF;
+        handle_gzerror();
+    }
 
     // reset buffer pointers
     setg( buffer + (4 - n_putback),   // beginning of putback area
@@ -97,7 +112,7 @@ int gzstreambuf::underflow() { // used for input buffer only
           buffer + 4 + num);          // end of buffer
 
     // return next character
-    return * reinterpret_cast<unsigned char *>( gptr());    
+    return * reinterpret_cast<unsigned char *>( gptr());
 }
 
 int gzstreambuf::flush_buffer() {
@@ -105,7 +120,7 @@ int gzstreambuf::flush_buffer() {
     // sync() operation.
     int w = pptr() - pbase();
     if ( gzwrite( file, pbase(), w) != w)
-        return EOF;
+        handle_gzerror();
     pbump( -w);
     return w;
 }
diff --git a/decoder/gzstream.h b/decoder/gzstream.h
index ad9785fd..5c709490 100644
--- a/decoder/gzstream.h
+++ b/decoder/gzstream.h
@@ -21,8 +21,8 @@
 // Revision      : $Revision: 1.1 $
 // Revision_date : $Date: 2006/03/30 04:05:52 $
 // Author(s)     : Deepak Bandyopadhyay, Lutz Kettner
-// 
-// Standard streambuf implementation following Nicolai Josuttis, "The 
+//
+// Standard streambuf implementation following Nicolai Josuttis, "The
 // Standard C++ Library".
 // ============================================================================
 
@@ -44,7 +44,7 @@ namespace GZSTREAM_NAMESPACE {
 
 class gzstreambuf : public std::streambuf {
 private:
-    static const int bufferSize = 47+256;    // size of data buff
+  static const int bufferSize = 47+(1024*256);    // size of data buff
     // totals 512 bytes under g++ for igzstream at the end.
 
     gzFile           file;               // file handle for compressed file
@@ -53,19 +53,25 @@ private:
     int              mode;               // I/O mode
 
     int flush_buffer();
+    void handle_gzerror(); // throws exception
 public:
+#if defined(_WIN32) && !defined(CYGWIN) && !defined(EOF)
+	enum {
+		EOF = -1
+	};
+#endif
     gzstreambuf() : opened(0) {
         setp( buffer, buffer + (bufferSize-1));
         setg( buffer + 4,     // beginning of putback area
               buffer + 4,     // read position
-              buffer + 4);    // end position      
+              buffer + 4);    // end position
         // ASSERT: both input & output capabilities will not be used together
     }
     int is_open() { return opened; }
     gzstreambuf* open( const char* name, int open_mode);
     gzstreambuf* close();
     ~gzstreambuf() { close(); }
-    
+
     virtual int     overflow( int c = EOF);
     virtual int     underflow();
     virtual int     sync();
@@ -85,15 +91,15 @@ public:
 
 // ----------------------------------------------------------------------------
 // User classes. Use igzstream and ogzstream analogously to ifstream and
-// ofstream respectively. They read and write files based on the gz* 
+// ofstream respectively. They read and write files based on the gz*
 // function interface of the zlib. Files are compatible with gzip compression.
 // ----------------------------------------------------------------------------
 
 class igzstream : public gzstreambase, public std::istream {
 public:
-    igzstream() : std::istream( &buf) {} 
+    igzstream() : std::istream( &buf) {}
     igzstream( const char* name, int open_mode = std::ios::in)
-        : gzstreambase( name, open_mode), std::istream( &buf) {}  
+        : gzstreambase( name, std::ios::in | open_mode), std::istream( &buf) {}
     gzstreambuf* rdbuf() { return gzstreambase::rdbuf(); }
     void open( const char* name, int open_mode = std::ios::in) {
         gzstreambase::open( name, open_mode);
@@ -104,7 +110,7 @@ class ogzstream : public gzstreambase, public std::ostream {
 public:
     ogzstream() : std::ostream( &buf) {}
     ogzstream( const char* name, int mode = std::ios::out)
-        : gzstreambase( name, mode), std::ostream( &buf) {}  
+        : gzstreambase( name, mode), std::ostream( &buf) {}
     gzstreambuf* rdbuf() { return gzstreambase::rdbuf(); }
     void open( const char* name, int open_mode = std::ios::out) {
         gzstreambase::open( name, open_mode);