2
1
mirror of https://github.com/qpdf/qpdf.git synced 2024-11-16 17:45:09 +00:00
qpdf/libqpdf/Pl_QPDFTokenizer.cc
Jay Berkenbilt 5d4cad9c02 ABI change: fix use of off_t, size_t, and integer types
Significantly improve the code's use of off_t for file offsets, size_t
for memory sizes, and integer types in cases where there has to be
compatibility with external interfaces.  Rework sections of the code
that would have prevented qpdf from working on files larger than 2 (or
maybe 4) GB in size.
2012-06-20 15:20:26 -04:00

199 lines
4.1 KiB
C++

#include <qpdf/Pl_QPDFTokenizer.hh>
#include <qpdf/QPDF_String.hh>
#include <qpdf/QPDF_Name.hh>
#include <qpdf/QTC.hh>
#include <stdexcept>
#include <string.h>
Pl_QPDFTokenizer::Pl_QPDFTokenizer(char const* identifier, Pipeline* next) :
Pipeline(identifier, next),
newline_after_next_token(false),
just_wrote_nl(false),
last_char_was_cr(false),
unread_char(false),
char_to_unread('\0'),
in_inline_image(false)
{
memset(this->image_buf, 0, IMAGE_BUF_SIZE);
}
Pl_QPDFTokenizer::~Pl_QPDFTokenizer()
{
}
void
Pl_QPDFTokenizer::writeNext(char const* buf, size_t len)
{
if (len)
{
unsigned char* t = new unsigned char[len];
memcpy(t, buf, len);
getNext()->write(t, len);
delete [] t;
this->just_wrote_nl = (buf[len-1] == '\n');
}
}
void
Pl_QPDFTokenizer::writeToken(QPDFTokenizer::Token& token)
{
std::string value = token.getRawValue();
switch (token.getType())
{
case QPDFTokenizer::tt_string:
value = QPDF_String(token.getValue()).unparse();
break;
case QPDFTokenizer::tt_name:
value = QPDF_Name(token.getValue()).unparse();
break;
default:
break;
}
writeNext(value.c_str(), value.length());
}
void
Pl_QPDFTokenizer::processChar(char ch)
{
if (this->in_inline_image)
{
// Scan through the input looking for EI surrounded by
// whitespace. If that pattern appears in the inline image's
// representation, we're hosed, but this situation seems
// excessively unlikely, and this code path is only followed
// during content stream normalization, which is pretty much
// used for debugging and human inspection of PDF files.
memmove(this->image_buf,
this->image_buf + 1,
IMAGE_BUF_SIZE - 1);
this->image_buf[IMAGE_BUF_SIZE - 1] = ch;
if (strchr(" \t\n\v\f\r", this->image_buf[0]) &&
(this->image_buf[1] == 'E') &&
(this->image_buf[2] == 'I') &&
strchr(" \t\n\v\f\r", this->image_buf[3]))
{
// We've found an EI operator. We've already written the
// EI operator to output; terminate with a newline
// character and resume normal processing.
writeNext("\n", 1);
this->in_inline_image = false;
QTC::TC("qpdf", "Pl_QPDFTokenizer found EI");
}
else
{
writeNext(&ch, 1);
}
return;
}
tokenizer.presentCharacter(ch);
QPDFTokenizer::Token token;
if (tokenizer.getToken(token, this->unread_char, this->char_to_unread))
{
writeToken(token);
if (this->newline_after_next_token)
{
writeNext("\n", 1);
this->newline_after_next_token = false;
}
if ((token.getType() == QPDFTokenizer::tt_word) &&
(token.getValue() == "ID"))
{
// Suspend normal scanning until we find an EI token.
this->in_inline_image = true;
if (this->unread_char)
{
writeNext(&this->char_to_unread, 1);
this->unread_char = false;
}
}
}
else
{
bool suppress = false;
if ((ch == '\n') && (this->last_char_was_cr))
{
// Always ignore \n following \r
suppress = true;
}
if ((this->last_char_was_cr = (ch == '\r')))
{
ch = '\n';
}
if (this->tokenizer.betweenTokens())
{
if (! suppress)
{
writeNext(&ch, 1);
}
}
else
{
if (ch == '\n')
{
this->newline_after_next_token = true;
}
}
}
}
void
Pl_QPDFTokenizer::checkUnread()
{
if (this->unread_char)
{
processChar(this->char_to_unread);
if (this->unread_char)
{
throw std::logic_error(
"INTERNAL ERROR: unread_char still true after processing "
"unread character");
}
}
}
void
Pl_QPDFTokenizer::write(unsigned char* buf, size_t len)
{
checkUnread();
for (size_t i = 0; i < len; ++i)
{
processChar(buf[i]);
checkUnread();
}
}
void
Pl_QPDFTokenizer::finish()
{
this->tokenizer.presentEOF();
if (! this->in_inline_image)
{
QPDFTokenizer::Token token;
if (tokenizer.getToken(token, this->unread_char, this->char_to_unread))
{
writeToken(token);
if (unread_char)
{
if (this->char_to_unread == '\r')
{
this->char_to_unread = '\n';
}
writeNext(&this->char_to_unread, 1);
}
}
}
if (! this->just_wrote_nl)
{
writeNext("\n", 1);
}
getNext()->finish();
}