OpenSystemsDevelopment / qpdf

Commit 31372edce0b60211c7af98340b3afa054f414ca4

Authored by Jay Berkenbilt
1 parent c1363563

Inline image token value ends with EI, not delimiter 

The inline image token erroneously included the delimiter that
followed EI. The ObjectHandle created from it was correct.
Inline Side-by-side
Showing 6 changed files with 13 additions and 10 deletions
ChangeLog
  View file @31372ed
1 2019-01-29 Jay Berkenbilt <ejb@ql.org> 1 2019-01-29 Jay Berkenbilt <ejb@ql.org>
2 2
  3 + * Bug fix: when returning an inline image token, the tokenizer no
  4 + longer includes the delimiter that follows EI. The
  5 + QPDFObjectHandle created from the token was correct.
  6 +
3 * Handle files with direct page objects, which is not allowed by 7 * Handle files with direct page objects, which is not allowed by
4 the PDF spec but has been seen in the wild. Fixes #164. 8 the PDF spec but has been seen in the wild. Fixes #164.
5 9
libqpdf/QPDFObjectHandle.cc
  View file @31372ed
@@ -1571,9 +1571,9 @@ QPDFObjectHandle::parseContentStream_data( @@ -1571,9 +1571,9 @@ QPDFObjectHandle::parseContentStream_data(
1571 else 1571 else
1572 { 1572 {
1573 // Skip back over EI 1573 // Skip back over EI
1574 - input->seek(-3, SEEK_CUR); 1574 + input->seek(-2, SEEK_CUR);
1575 std::string inline_image = t.getRawValue(); 1575 std::string inline_image = t.getRawValue();
1576 - for (int i = 0; i < 4; ++i) 1576 + for (int i = 0; i < 3; ++i)
1577 { 1577 {
1578 if (inline_image.length() > 0) 1578 if (inline_image.length() > 0)
1579 { 1579 {
libqpdf/QPDFTokenizer.cc
  View file @31372ed
@@ -468,6 +468,7 @@ QPDFTokenizer::presentCharacter(char ch) @@ -468,6 +468,7 @@ QPDFTokenizer::presentCharacter(char ch)
468 } 468 }
469 else if (this->m->state == st_inline_image) 469 else if (this->m->state == st_inline_image)
470 { 470 {
  471 + this->m->val += ch;
471 size_t len = this->m->val.length(); 472 size_t len = this->m->val.length();
472 if ((len >= 4) && 473 if ((len >= 4) &&
473 isDelimiter(this->m->val.at(len-4)) && 474 isDelimiter(this->m->val.at(len-4)) &&
@@ -475,22 +476,18 @@ QPDFTokenizer::presentCharacter(char ch) @@ -475,22 +476,18 @@ QPDFTokenizer::presentCharacter(char ch)
475 (this->m->val.at(len-2) == 'I') && 476 (this->m->val.at(len-2) == 'I') &&
476 isDelimiter(this->m->val.at(len-1))) 477 isDelimiter(this->m->val.at(len-1)))
477 { 478 {
  479 + this->m->val.erase(len - 1);
478 this->m->type = tt_inline_image; 480 this->m->type = tt_inline_image;
479 this->m->unread_char = true; 481 this->m->unread_char = true;
480 this->m->char_to_unread = ch; 482 this->m->char_to_unread = ch;
481 this->m->state = st_token_ready; 483 this->m->state = st_token_ready;
482 } 484 }
483 - else  
484 - {  
485 - this->m->val += ch;  
486 - }  
487 } 485 }
488 else 486 else
489 { 487 {
490 handled = false; 488 handled = false;
491 } 489 }
492 490
493 -  
494 if (handled) 491 if (handled)
495 { 492 {
496 // okay 493 // okay
qpdf/qtest/qpdf/tokens-maxlen.out
  View file @31372ed
@@ -599,7 +599,8 @@ skipping to endstream @@ -599,7 +599,8 @@ skipping to endstream
599 103: dict_close: >> 599 103: dict_close: >>
600 105: space: \x0a 600 105: space: \x0a
601 106: word: ID 601 106: word: ID
602 -108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI\x0a 602 +108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI
  603 +354: space: \x0a
603 355: word: BT 604 355: word: BT
604 357: space: \x0a 605 357: space: \x0a
605 360: name: /F1 606 360: name: /F1
qpdf/qtest/qpdf/tokens-no-ignorable.out
  View file @31372ed
@@ -291,7 +291,7 @@ skipping to endstream @@ -291,7 +291,7 @@ skipping to endstream
291 101: integer: 66 291 101: integer: 66
292 103: dict_close: >> 292 103: dict_close: >>
293 106: word: ID 293 106: word: ID
294 -108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI\x0a 294 +108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI
295 355: word: BT 295 355: word: BT
296 360: name: /F1 296 360: name: /F1
297 364: integer: 24 297 364: integer: 24
qpdf/qtest/qpdf/tokens.out
  View file @31372ed
@@ -599,7 +599,8 @@ skipping to endstream @@ -599,7 +599,8 @@ skipping to endstream
599 103: dict_close: >> 599 103: dict_close: >>
600 105: space: \x0a 600 105: space: \x0a
601 106: word: ID 601 106: word: ID
602 -108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI\x0a 602 +108: inline-image: x\x9c\xc5\xd6I\x0e\xc3 \x0c\x05P|\xffC;U\xc8`\xc0\xd37\x91Z\xa9\x0b\xa6\x17\x02\xc4\x98\xda\xe6\x8f\x1b}D\xf0\xef_\xb4\xf8\x1c\xc9W\xa9\x84\x9c\xc4-\x94\x88>\xff\x87\xc0\x8d>\x94^\x01&\xae\xa1u\xe2]\x80"!\x87\x95\x08\x96\x05*\xac&\x8fE|Sy\xae \xf0d-\x80<\x9d\x19B\x010B\x05\xfa@N\x11\xea+<\x1fhl\xe8K\xd0\xee/56L\xa0\x89\x90\xe3\x19\x1e \xa3\x96\xb9\xa6>0\x06>\x15Y\x81\xf9!c\xec\\x0eY\x0c\xd8\x0f%Y\xf0\x01\xa5\xd68?&\xa0\xd6\xeb\x88}j\x92\xfb\xe8\x1d;\xab\x8d3\x9d\xc2\xd6l\x14p\xdbsH\xf6\xfbt\xfa\x01Q\x02\xd8Tt*h\xccU\xfa\xe3w\x07\xcd\xd5\xd0%\xa8)p\x96\xb3"\x95DiRj\xb9\x96D\x18YNU\x11\xd3\xd9Av\x92F\xe0&\x0d\x90\xcd\xd4u#c\x95\xc6W\x09\xf4\xdf\x89\x03W\x93O\x0d\x0aEI
  603 +354: space: \x0a
603 355: word: BT 604 355: word: BT
604 357: space: \x0a 605 357: space: \x0a
605 360: name: /F1 606 360: name: /F1