cannam@89: /*
cannam@89:  * A C++ I/O streams interface to the zlib gz* functions
cannam@89:  *
cannam@89:  * by Ludwig Schwardt <schwardt@sun.ac.za>
cannam@89:  * original version by Kevin Ruland <kevin@rodin.wustl.edu>
cannam@89:  *
cannam@89:  * This version is standard-compliant and compatible with gcc 3.x.
cannam@89:  */
cannam@89: 
cannam@89: #include "zfstream.h"
cannam@89: #include <cstring>          // for strcpy, strcat, strlen (mode strings)
cannam@89: #include <cstdio>           // for BUFSIZ
cannam@89: 
cannam@89: // Internal buffer sizes (default and "unbuffered" versions)
cannam@89: #define BIGBUFSIZE BUFSIZ
cannam@89: #define SMALLBUFSIZE 1
cannam@89: 
cannam@89: /*****************************************************************************/
cannam@89: 
cannam@89: // Default constructor
cannam@89: gzfilebuf::gzfilebuf()
cannam@89: : file(NULL), io_mode(std::ios_base::openmode(0)), own_fd(false),
cannam@89:   buffer(NULL), buffer_size(BIGBUFSIZE), own_buffer(true)
cannam@89: {
cannam@89:   // No buffers to start with
cannam@89:   this->disable_buffer();
cannam@89: }
cannam@89: 
cannam@89: // Destructor
cannam@89: gzfilebuf::~gzfilebuf()
cannam@89: {
cannam@89:   // Sync output buffer and close only if responsible for file
cannam@89:   // (i.e. attached streams should be left open at this stage)
cannam@89:   this->sync();
cannam@89:   if (own_fd)
cannam@89:     this->close();
cannam@89:   // Make sure internal buffer is deallocated
cannam@89:   this->disable_buffer();
cannam@89: }
cannam@89: 
cannam@89: // Set compression level and strategy
cannam@89: int
cannam@89: gzfilebuf::setcompression(int comp_level,
cannam@89:                           int comp_strategy)
cannam@89: {
cannam@89:   return gzsetparams(file, comp_level, comp_strategy);
cannam@89: }
cannam@89: 
cannam@89: // Open gzipped file
cannam@89: gzfilebuf*
cannam@89: gzfilebuf::open(const char *name,
cannam@89:                 std::ios_base::openmode mode)
cannam@89: {
cannam@89:   // Fail if file already open
cannam@89:   if (this->is_open())
cannam@89:     return NULL;
cannam@89:   // Don't support simultaneous read/write access (yet)
cannam@89:   if ((mode & std::ios_base::in) && (mode & std::ios_base::out))
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // Build mode string for gzopen and check it [27.8.1.3.2]
cannam@89:   char char_mode[6] = "\0\0\0\0\0";
cannam@89:   if (!this->open_mode(mode, char_mode))
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // Attempt to open file
cannam@89:   if ((file = gzopen(name, char_mode)) == NULL)
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // On success, allocate internal buffer and set flags
cannam@89:   this->enable_buffer();
cannam@89:   io_mode = mode;
cannam@89:   own_fd = true;
cannam@89:   return this;
cannam@89: }
cannam@89: 
cannam@89: // Attach to gzipped file
cannam@89: gzfilebuf*
cannam@89: gzfilebuf::attach(int fd,
cannam@89:                   std::ios_base::openmode mode)
cannam@89: {
cannam@89:   // Fail if file already open
cannam@89:   if (this->is_open())
cannam@89:     return NULL;
cannam@89:   // Don't support simultaneous read/write access (yet)
cannam@89:   if ((mode & std::ios_base::in) && (mode & std::ios_base::out))
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // Build mode string for gzdopen and check it [27.8.1.3.2]
cannam@89:   char char_mode[6] = "\0\0\0\0\0";
cannam@89:   if (!this->open_mode(mode, char_mode))
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // Attempt to attach to file
cannam@89:   if ((file = gzdopen(fd, char_mode)) == NULL)
cannam@89:     return NULL;
cannam@89: 
cannam@89:   // On success, allocate internal buffer and set flags
cannam@89:   this->enable_buffer();
cannam@89:   io_mode = mode;
cannam@89:   own_fd = false;
cannam@89:   return this;
cannam@89: }
cannam@89: 
cannam@89: // Close gzipped file
cannam@89: gzfilebuf*
cannam@89: gzfilebuf::close()
cannam@89: {
cannam@89:   // Fail immediately if no file is open
cannam@89:   if (!this->is_open())
cannam@89:     return NULL;
cannam@89:   // Assume success
cannam@89:   gzfilebuf* retval = this;
cannam@89:   // Attempt to sync and close gzipped file
cannam@89:   if (this->sync() == -1)
cannam@89:     retval = NULL;
cannam@89:   if (gzclose(file) < 0)
cannam@89:     retval = NULL;
cannam@89:   // File is now gone anyway (postcondition [27.8.1.3.8])
cannam@89:   file = NULL;
cannam@89:   own_fd = false;
cannam@89:   // Destroy internal buffer if it exists
cannam@89:   this->disable_buffer();
cannam@89:   return retval;
cannam@89: }
cannam@89: 
cannam@89: /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
cannam@89: 
cannam@89: // Convert int open mode to mode string
cannam@89: bool
cannam@89: gzfilebuf::open_mode(std::ios_base::openmode mode,
cannam@89:                      char* c_mode) const
cannam@89: {
cannam@89:   bool testb = mode & std::ios_base::binary;
cannam@89:   bool testi = mode & std::ios_base::in;
cannam@89:   bool testo = mode & std::ios_base::out;
cannam@89:   bool testt = mode & std::ios_base::trunc;
cannam@89:   bool testa = mode & std::ios_base::app;
cannam@89: 
cannam@89:   // Check for valid flag combinations - see [27.8.1.3.2] (Table 92)
cannam@89:   // Original zfstream hardcoded the compression level to maximum here...
cannam@89:   // Double the time for less than 1% size improvement seems
cannam@89:   // excessive though - keeping it at the default level
cannam@89:   // To change back, just append "9" to the next three mode strings
cannam@89:   if (!testi && testo && !testt && !testa)
cannam@89:     strcpy(c_mode, "w");
cannam@89:   if (!testi && testo && !testt && testa)
cannam@89:     strcpy(c_mode, "a");
cannam@89:   if (!testi && testo && testt && !testa)
cannam@89:     strcpy(c_mode, "w");
cannam@89:   if (testi && !testo && !testt && !testa)
cannam@89:     strcpy(c_mode, "r");
cannam@89:   // No read/write mode yet
cannam@89: //  if (testi && testo && !testt && !testa)
cannam@89: //    strcpy(c_mode, "r+");
cannam@89: //  if (testi && testo && testt && !testa)
cannam@89: //    strcpy(c_mode, "w+");
cannam@89: 
cannam@89:   // Mode string should be empty for invalid combination of flags
cannam@89:   if (strlen(c_mode) == 0)
cannam@89:     return false;
cannam@89:   if (testb)
cannam@89:     strcat(c_mode, "b");
cannam@89:   return true;
cannam@89: }
cannam@89: 
cannam@89: // Determine number of characters in internal get buffer
cannam@89: std::streamsize
cannam@89: gzfilebuf::showmanyc()
cannam@89: {
cannam@89:   // Calls to underflow will fail if file not opened for reading
cannam@89:   if (!this->is_open() || !(io_mode & std::ios_base::in))
cannam@89:     return -1;
cannam@89:   // Make sure get area is in use
cannam@89:   if (this->gptr() && (this->gptr() < this->egptr()))
cannam@89:     return std::streamsize(this->egptr() - this->gptr());
cannam@89:   else
cannam@89:     return 0;
cannam@89: }
cannam@89: 
cannam@89: // Fill get area from gzipped file
cannam@89: gzfilebuf::int_type
cannam@89: gzfilebuf::underflow()
cannam@89: {
cannam@89:   // If something is left in the get area by chance, return it
cannam@89:   // (this shouldn't normally happen, as underflow is only supposed
cannam@89:   // to be called when gptr >= egptr, but it serves as error check)
cannam@89:   if (this->gptr() && (this->gptr() < this->egptr()))
cannam@89:     return traits_type::to_int_type(*(this->gptr()));
cannam@89: 
cannam@89:   // If the file hasn't been opened for reading, produce error
cannam@89:   if (!this->is_open() || !(io_mode & std::ios_base::in))
cannam@89:     return traits_type::eof();
cannam@89: 
cannam@89:   // Attempt to fill internal buffer from gzipped file
cannam@89:   // (buffer must be guaranteed to exist...)
cannam@89:   int bytes_read = gzread(file, buffer, buffer_size);
cannam@89:   // Indicates error or EOF
cannam@89:   if (bytes_read <= 0)
cannam@89:   {
cannam@89:     // Reset get area
cannam@89:     this->setg(buffer, buffer, buffer);
cannam@89:     return traits_type::eof();
cannam@89:   }
cannam@89:   // Make all bytes read from file available as get area
cannam@89:   this->setg(buffer, buffer, buffer + bytes_read);
cannam@89: 
cannam@89:   // Return next character in get area
cannam@89:   return traits_type::to_int_type(*(this->gptr()));
cannam@89: }
cannam@89: 
cannam@89: // Write put area to gzipped file
cannam@89: gzfilebuf::int_type
cannam@89: gzfilebuf::overflow(int_type c)
cannam@89: {
cannam@89:   // Determine whether put area is in use
cannam@89:   if (this->pbase())
cannam@89:   {
cannam@89:     // Double-check pointer range
cannam@89:     if (this->pptr() > this->epptr() || this->pptr() < this->pbase())
cannam@89:       return traits_type::eof();
cannam@89:     // Add extra character to buffer if not EOF
cannam@89:     if (!traits_type::eq_int_type(c, traits_type::eof()))
cannam@89:     {
cannam@89:       *(this->pptr()) = traits_type::to_char_type(c);
cannam@89:       this->pbump(1);
cannam@89:     }
cannam@89:     // Number of characters to write to file
cannam@89:     int bytes_to_write = this->pptr() - this->pbase();
cannam@89:     // Overflow doesn't fail if nothing is to be written
cannam@89:     if (bytes_to_write > 0)
cannam@89:     {
cannam@89:       // If the file hasn't been opened for writing, produce error
cannam@89:       if (!this->is_open() || !(io_mode & std::ios_base::out))
cannam@89:         return traits_type::eof();
cannam@89:       // If gzipped file won't accept all bytes written to it, fail
cannam@89:       if (gzwrite(file, this->pbase(), bytes_to_write) != bytes_to_write)
cannam@89:         return traits_type::eof();
cannam@89:       // Reset next pointer to point to pbase on success
cannam@89:       this->pbump(-bytes_to_write);
cannam@89:     }
cannam@89:   }
cannam@89:   // Write extra character to file if not EOF
cannam@89:   else if (!traits_type::eq_int_type(c, traits_type::eof()))
cannam@89:   {
cannam@89:     // If the file hasn't been opened for writing, produce error
cannam@89:     if (!this->is_open() || !(io_mode & std::ios_base::out))
cannam@89:       return traits_type::eof();
cannam@89:     // Impromptu char buffer (allows "unbuffered" output)
cannam@89:     char_type last_char = traits_type::to_char_type(c);
cannam@89:     // If gzipped file won't accept this character, fail
cannam@89:     if (gzwrite(file, &last_char, 1) != 1)
cannam@89:       return traits_type::eof();
cannam@89:   }
cannam@89: 
cannam@89:   // If you got here, you have succeeded (even if c was EOF)
cannam@89:   // The return value should therefore be non-EOF
cannam@89:   if (traits_type::eq_int_type(c, traits_type::eof()))
cannam@89:     return traits_type::not_eof(c);
cannam@89:   else
cannam@89:     return c;
cannam@89: }
cannam@89: 
cannam@89: // Assign new buffer
cannam@89: std::streambuf*
cannam@89: gzfilebuf::setbuf(char_type* p,
cannam@89:                   std::streamsize n)
cannam@89: {
cannam@89:   // First make sure stuff is sync'ed, for safety
cannam@89:   if (this->sync() == -1)
cannam@89:     return NULL;
cannam@89:   // If buffering is turned off on purpose via setbuf(0,0), still allocate one...
cannam@89:   // "Unbuffered" only really refers to put [27.8.1.4.10], while get needs at
cannam@89:   // least a buffer of size 1 (very inefficient though, therefore make it bigger?)
cannam@89:   // This follows from [27.5.2.4.3]/12 (gptr needs to point at something, it seems)
cannam@89:   if (!p || !n)
cannam@89:   {
cannam@89:     // Replace existing buffer (if any) with small internal buffer
cannam@89:     this->disable_buffer();
cannam@89:     buffer = NULL;
cannam@89:     buffer_size = 0;
cannam@89:     own_buffer = true;
cannam@89:     this->enable_buffer();
cannam@89:   }
cannam@89:   else
cannam@89:   {
cannam@89:     // Replace existing buffer (if any) with external buffer
cannam@89:     this->disable_buffer();
cannam@89:     buffer = p;
cannam@89:     buffer_size = n;
cannam@89:     own_buffer = false;
cannam@89:     this->enable_buffer();
cannam@89:   }
cannam@89:   return this;
cannam@89: }
cannam@89: 
cannam@89: // Write put area to gzipped file (i.e. ensures that put area is empty)
cannam@89: int
cannam@89: gzfilebuf::sync()
cannam@89: {
cannam@89:   return traits_type::eq_int_type(this->overflow(), traits_type::eof()) ? -1 : 0;
cannam@89: }
cannam@89: 
cannam@89: /* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */
cannam@89: 
cannam@89: // Allocate internal buffer
cannam@89: void
cannam@89: gzfilebuf::enable_buffer()
cannam@89: {
cannam@89:   // If internal buffer required, allocate one
cannam@89:   if (own_buffer && !buffer)
cannam@89:   {
cannam@89:     // Check for buffered vs. "unbuffered"
cannam@89:     if (buffer_size > 0)
cannam@89:     {
cannam@89:       // Allocate internal buffer
cannam@89:       buffer = new char_type[buffer_size];
cannam@89:       // Get area starts empty and will be expanded by underflow as need arises
cannam@89:       this->setg(buffer, buffer, buffer);
cannam@89:       // Setup entire internal buffer as put area.
cannam@89:       // The one-past-end pointer actually points to the last element of the buffer,
cannam@89:       // so that overflow(c) can safely add the extra character c to the sequence.
cannam@89:       // These pointers remain in place for the duration of the buffer
cannam@89:       this->setp(buffer, buffer + buffer_size - 1);
cannam@89:     }
cannam@89:     else
cannam@89:     {
cannam@89:       // Even in "unbuffered" case, (small?) get buffer is still required
cannam@89:       buffer_size = SMALLBUFSIZE;
cannam@89:       buffer = new char_type[buffer_size];
cannam@89:       this->setg(buffer, buffer, buffer);
cannam@89:       // "Unbuffered" means no put buffer
cannam@89:       this->setp(0, 0);
cannam@89:     }
cannam@89:   }
cannam@89:   else
cannam@89:   {
cannam@89:     // If buffer already allocated, reset buffer pointers just to make sure no
cannam@89:     // stale chars are lying around
cannam@89:     this->setg(buffer, buffer, buffer);
cannam@89:     this->setp(buffer, buffer + buffer_size - 1);
cannam@89:   }
cannam@89: }
cannam@89: 
cannam@89: // Destroy internal buffer
cannam@89: void
cannam@89: gzfilebuf::disable_buffer()
cannam@89: {
cannam@89:   // If internal buffer exists, deallocate it
cannam@89:   if (own_buffer && buffer)
cannam@89:   {
cannam@89:     // Preserve unbuffered status by zeroing size
cannam@89:     if (!this->pbase())
cannam@89:       buffer_size = 0;
cannam@89:     delete[] buffer;
cannam@89:     buffer = NULL;
cannam@89:     this->setg(0, 0, 0);
cannam@89:     this->setp(0, 0);
cannam@89:   }
cannam@89:   else
cannam@89:   {
cannam@89:     // Reset buffer pointers to initial state if external buffer exists
cannam@89:     this->setg(buffer, buffer, buffer);
cannam@89:     if (buffer)
cannam@89:       this->setp(buffer, buffer + buffer_size - 1);
cannam@89:     else
cannam@89:       this->setp(0, 0);
cannam@89:   }
cannam@89: }
cannam@89: 
cannam@89: /*****************************************************************************/
cannam@89: 
cannam@89: // Default constructor initializes stream buffer
cannam@89: gzifstream::gzifstream()
cannam@89: : std::istream(NULL), sb()
cannam@89: { this->init(&sb); }
cannam@89: 
cannam@89: // Initialize stream buffer and open file
cannam@89: gzifstream::gzifstream(const char* name,
cannam@89:                        std::ios_base::openmode mode)
cannam@89: : std::istream(NULL), sb()
cannam@89: {
cannam@89:   this->init(&sb);
cannam@89:   this->open(name, mode);
cannam@89: }
cannam@89: 
cannam@89: // Initialize stream buffer and attach to file
cannam@89: gzifstream::gzifstream(int fd,
cannam@89:                        std::ios_base::openmode mode)
cannam@89: : std::istream(NULL), sb()
cannam@89: {
cannam@89:   this->init(&sb);
cannam@89:   this->attach(fd, mode);
cannam@89: }
cannam@89: 
cannam@89: // Open file and go into fail() state if unsuccessful
cannam@89: void
cannam@89: gzifstream::open(const char* name,
cannam@89:                  std::ios_base::openmode mode)
cannam@89: {
cannam@89:   if (!sb.open(name, mode | std::ios_base::in))
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89:   else
cannam@89:     this->clear();
cannam@89: }
cannam@89: 
cannam@89: // Attach to file and go into fail() state if unsuccessful
cannam@89: void
cannam@89: gzifstream::attach(int fd,
cannam@89:                    std::ios_base::openmode mode)
cannam@89: {
cannam@89:   if (!sb.attach(fd, mode | std::ios_base::in))
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89:   else
cannam@89:     this->clear();
cannam@89: }
cannam@89: 
cannam@89: // Close file
cannam@89: void
cannam@89: gzifstream::close()
cannam@89: {
cannam@89:   if (!sb.close())
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89: }
cannam@89: 
cannam@89: /*****************************************************************************/
cannam@89: 
cannam@89: // Default constructor initializes stream buffer
cannam@89: gzofstream::gzofstream()
cannam@89: : std::ostream(NULL), sb()
cannam@89: { this->init(&sb); }
cannam@89: 
cannam@89: // Initialize stream buffer and open file
cannam@89: gzofstream::gzofstream(const char* name,
cannam@89:                        std::ios_base::openmode mode)
cannam@89: : std::ostream(NULL), sb()
cannam@89: {
cannam@89:   this->init(&sb);
cannam@89:   this->open(name, mode);
cannam@89: }
cannam@89: 
cannam@89: // Initialize stream buffer and attach to file
cannam@89: gzofstream::gzofstream(int fd,
cannam@89:                        std::ios_base::openmode mode)
cannam@89: : std::ostream(NULL), sb()
cannam@89: {
cannam@89:   this->init(&sb);
cannam@89:   this->attach(fd, mode);
cannam@89: }
cannam@89: 
cannam@89: // Open file and go into fail() state if unsuccessful
cannam@89: void
cannam@89: gzofstream::open(const char* name,
cannam@89:                  std::ios_base::openmode mode)
cannam@89: {
cannam@89:   if (!sb.open(name, mode | std::ios_base::out))
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89:   else
cannam@89:     this->clear();
cannam@89: }
cannam@89: 
cannam@89: // Attach to file and go into fail() state if unsuccessful
cannam@89: void
cannam@89: gzofstream::attach(int fd,
cannam@89:                    std::ios_base::openmode mode)
cannam@89: {
cannam@89:   if (!sb.attach(fd, mode | std::ios_base::out))
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89:   else
cannam@89:     this->clear();
cannam@89: }
cannam@89: 
cannam@89: // Close file
cannam@89: void
cannam@89: gzofstream::close()
cannam@89: {
cannam@89:   if (!sb.close())
cannam@89:     this->setstate(std::ios_base::failbit);
cannam@89: }