qpdf/libqpdf/Pl_Flate.cc

#include <qpdf/Pl_Flate.hh>
#include <zlib.h>
#include <string.h>
#include <limits.h>

#include <qpdf/QUtil.hh>
#include <qpdf/QIntC.hh>

Pl_Flate::Pl_Flate(char const* identifier, Pipeline* next,
		   action_e action, unsigned int out_bufsize_int) :
    Pipeline(identifier, next),
    out_bufsize(QIntC::to_size(out_bufsize_int)),
    action(action),
    initialized(false)
{
    this->outbuf = new unsigned char[out_bufsize];
    // Indirect through zdata to reach the z_stream so we don't have
    // to include zlib.h in Pl_Flate.hh.  This means people using
    // shared library versions of qpdf don't have to have zlib
    // development files available, which particularly helps in a
    // Windows environment.
    this->zdata = new z_stream;

    if (out_bufsize > UINT_MAX)
    {
        throw std::runtime_error(
            "Pl_Flate: zlib doesn't support buffer"
            " sizes larger than unsigned int");
    }

    z_stream& zstream = *(static_cast<z_stream*>(this->zdata));
    zstream.zalloc = 0;
    zstream.zfree = 0;
    zstream.opaque = 0;
    zstream.next_in = 0;
    zstream.avail_in = 0;
    zstream.next_out = this->outbuf;
    zstream.avail_out = QIntC::to_uint(out_bufsize);
}

Pl_Flate::~Pl_Flate()
{
    delete [] this->outbuf;
    this->outbuf = 0;

    if (this->initialized)
    {
        z_stream& zstream = *(static_cast<z_stream*>(this->zdata));
        if (action == a_deflate)
        {
            deflateEnd(&zstream);
        }
        else
        {
            inflateEnd(&zstream);
        }
    }

    delete static_cast<z_stream*>(this->zdata);
    this->zdata = 0;
}

void
Pl_Flate::write(unsigned char* data, size_t len)
{
    if (this->outbuf == 0)
    {
	throw std::logic_error(
	    this->identifier +
	    ": Pl_Flate: write() called after finish() called");
    }

    // Write in chunks in case len is too big to fit in an int.
    // Assume int is at least 32 bits.
    static size_t const max_bytes = 1 << 30;
    size_t bytes_left = len;
    unsigned char* buf = data;
    while (bytes_left > 0)
    {
	size_t bytes = (bytes_left >= max_bytes ? max_bytes : bytes_left);
        handleData(buf, bytes,
                   (action == a_inflate ? Z_SYNC_FLUSH : Z_NO_FLUSH));
	bytes_left -= bytes;
        buf += bytes;
    }
}

void
Pl_Flate::handleData(unsigned char* data, size_t len, int flush)
{
    if (len > UINT_MAX)
    {
        throw std::runtime_error(
            "Pl_Flate: zlib doesn't support data"
            " blocks larger than int");
    }
    z_stream& zstream = *(static_cast<z_stream*>(this->zdata));
    zstream.next_in = data;
    zstream.avail_in = QIntC::to_uint(len);

    if (! this->initialized)
    {
	int err = Z_OK;

        // deflateInit and inflateInit are macros that use old-style
        // casts.
#if ((defined(__GNUC__) && ((__GNUC__ * 100) + __GNUC_MINOR__) >= 406) || \
     defined(__clang__))
#       pragma GCC diagnostic push
#       pragma GCC diagnostic ignored "-Wold-style-cast"
#endif
	if (this->action == a_deflate)
	{
	    err = deflateInit(&zstream, Z_DEFAULT_COMPRESSION);
	}
	else
	{
	    err = inflateInit(&zstream);
	}
#if ((defined(__GNUC__) && ((__GNUC__ * 100) + __GNUC_MINOR__) >= 406) || \
     defined(__clang__))
#       pragma GCC diagnostic pop
#endif

	checkError("Init", err);
	this->initialized = true;
    }

    int err = Z_OK;

    bool done = false;
    while (! done)
    {
	if (action == a_deflate)
	{
	    err = deflate(&zstream, flush);
	}
	else
	{
	    err = inflate(&zstream, flush);
	}
        if ((action == a_inflate) && (err != Z_OK) && zstream.msg &&
            (strcmp(zstream.msg, "incorrect data check") == 0))
        {
            // Other PDF readers ignore this specific error. Combining
            // this with Z_SYNC_FLUSH enables qpdf to handle some
            // broken zlib streams without losing data.
            err = Z_STREAM_END;
        }
	switch (err)
	{
	  case Z_BUF_ERROR:
	    // Probably shouldn't be able to happen, but possible as a
	    // boundary condition: if the last call to inflate exactly
	    // filled the output buffer, it's possible that the next
	    // call to inflate could have nothing to do.
	    done = true;
	    break;

	  case Z_STREAM_END:
	    done = true;
	    // fall through

	  case Z_OK:
	    {
		if ((zstream.avail_in == 0) &&
		    (zstream.avail_out > 0))
		{
		    // There is nothing left to read, and there was
		    // sufficient buffer space to write everything we
		    // needed, so we're done for now.
		    done = true;
		}
		uLong ready =
                    QIntC::to_ulong(this->out_bufsize - zstream.avail_out);
		if (ready > 0)
		{
		    this->getNext()->write(this->outbuf, ready);
		    zstream.next_out = this->outbuf;
		    zstream.avail_out = QIntC::to_uint(this->out_bufsize);
		}
	    }
	    break;

	  default:
	    this->checkError("data", err);
	    break;
	}
    }
}

void
Pl_Flate::finish()
{
    try
    {
        if (this->outbuf)
        {
            if (this->initialized)
            {
                z_stream& zstream = *(static_cast<z_stream*>(this->zdata));
                unsigned char buf[1];
                buf[0] = '\0';
                handleData(buf, 0, Z_FINISH);
                int err = Z_OK;
                if (action == a_deflate)
                {
                    err = deflateEnd(&zstream);
                }
                else
                {
                    err = inflateEnd(&zstream);
                }
                this->initialized = false;
                checkError("End", err);
            }

            delete [] this->outbuf;
            this->outbuf = 0;
        }
    }
    catch (std::exception& e)
    {
        this->getNext()->finish();
        throw e;
    }
    this->getNext()->finish();
}

void
Pl_Flate::checkError(char const* prefix, int error_code)
{
    z_stream& zstream = *(static_cast<z_stream*>(this->zdata));
    if (error_code != Z_OK)
    {
	char const* action_str = (action == a_deflate ? "deflate" : "inflate");
	std::string msg =
	    this->identifier + ": " + action_str + ": " + prefix + ": ";

	if (zstream.msg)
	{
	    msg += zstream.msg;
	}
	else
	{
	    switch (error_code)
	    {
	      case Z_ERRNO:
		msg += "zlib system error";
		break;

	      case Z_STREAM_ERROR:
		msg += "zlib stream error";
		break;

	      case Z_DATA_ERROR:
		msg += "zlib data error";
		break;

	      case Z_MEM_ERROR:
		msg += "zlib memory error";
		break;

	      case Z_BUF_ERROR:
		msg += "zlib buffer error";
		break;

	      case Z_VERSION_ERROR:
		msg += "zlib version error";
		break;

	      default:
		msg += std::string("zlib unknown error (") +
		    QUtil::int_to_string(error_code) + ")";
		break;
	    }
	}

	throw std::runtime_error(msg);
    }
}