Commit 7402c02c808224932bd766a18f22b780a4d30b63
1 parent
74162a2d
Combine stacks in QPDFParser::parse
Part of #729
Showing
1 changed file
with
35 additions
and
31 deletions
libqpdf/QPDFParser.cc
| @@ -7,6 +7,24 @@ | @@ -7,6 +7,24 @@ | ||
| 7 | #include <qpdf/QUtil.hh> | 7 | #include <qpdf/QUtil.hh> |
| 8 | #include <qpdf/SparseOHArray.hh> | 8 | #include <qpdf/SparseOHArray.hh> |
| 9 | 9 | ||
| 10 | +namespace | ||
| 11 | +{ | ||
| 12 | + struct StackFrame | ||
| 13 | + { | ||
| 14 | + StackFrame(std::shared_ptr<InputSource> input) : | ||
| 15 | + offset(input->tell()), | ||
| 16 | + contents_string(""), | ||
| 17 | + contents_offset(-1) | ||
| 18 | + { | ||
| 19 | + } | ||
| 20 | + | ||
| 21 | + std::vector<QPDFObjectHandle> olist; | ||
| 22 | + qpdf_offset_t offset; | ||
| 23 | + std::string contents_string; | ||
| 24 | + qpdf_offset_t contents_offset; | ||
| 25 | + }; | ||
| 26 | +} // namespace | ||
| 27 | + | ||
| 10 | QPDFObjectHandle | 28 | QPDFObjectHandle |
| 11 | QPDFParser::parse(bool& empty, bool content_stream) | 29 | QPDFParser::parse(bool& empty, bool content_stream) |
| 12 | { | 30 | { |
| @@ -17,8 +35,6 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -17,8 +35,6 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 17 | // this, it will cause a logic error to be thrown from | 35 | // this, it will cause a logic error to be thrown from |
| 18 | // QPDF::inParse(). | 36 | // QPDF::inParse(). |
| 19 | 37 | ||
| 20 | - using OHVector = std::vector<QPDFObjectHandle>; | ||
| 21 | - | ||
| 22 | QPDF::ParseGuard pg(context); | 38 | QPDF::ParseGuard pg(context); |
| 23 | 39 | ||
| 24 | empty = false; | 40 | empty = false; |
| @@ -26,28 +42,22 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -26,28 +42,22 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 26 | QPDFObjectHandle object; | 42 | QPDFObjectHandle object; |
| 27 | bool set_offset = false; | 43 | bool set_offset = false; |
| 28 | 44 | ||
| 29 | - std::vector<OHVector> olist_stack; | ||
| 30 | - olist_stack.push_back(OHVector()); | 45 | + std::vector<StackFrame> stack; |
| 46 | + stack.push_back(StackFrame(input)); | ||
| 31 | std::vector<parser_state_e> state_stack; | 47 | std::vector<parser_state_e> state_stack; |
| 32 | state_stack.push_back(st_top); | 48 | state_stack.push_back(st_top); |
| 33 | - std::vector<qpdf_offset_t> offset_stack; | ||
| 34 | - qpdf_offset_t offset = input->tell(); | ||
| 35 | - offset_stack.push_back(offset); | 49 | + qpdf_offset_t offset; |
| 36 | bool done = false; | 50 | bool done = false; |
| 37 | int bad_count = 0; | 51 | int bad_count = 0; |
| 38 | int good_count = 0; | 52 | int good_count = 0; |
| 39 | bool b_contents = false; | 53 | bool b_contents = false; |
| 40 | - std::vector<std::string> contents_string_stack; | ||
| 41 | - contents_string_stack.push_back(""); | ||
| 42 | - std::vector<qpdf_offset_t> contents_offset_stack; | ||
| 43 | - contents_offset_stack.push_back(-1); | 54 | + |
| 44 | while (!done) { | 55 | while (!done) { |
| 45 | bool bad = false; | 56 | bool bad = false; |
| 46 | - auto& olist = olist_stack.back(); | 57 | + auto& frame = stack.back(); |
| 58 | + auto& olist = frame.olist; | ||
| 47 | parser_state_e state = state_stack.back(); | 59 | parser_state_e state = state_stack.back(); |
| 48 | - offset = offset_stack.back(); | ||
| 49 | - std::string& contents_string = contents_string_stack.back(); | ||
| 50 | - qpdf_offset_t& contents_offset = contents_offset_stack.back(); | 60 | + offset = frame.offset; |
| 51 | 61 | ||
| 52 | object = QPDFObjectHandle(); | 62 | object = QPDFObjectHandle(); |
| 53 | set_offset = false; | 63 | set_offset = false; |
| @@ -108,23 +118,20 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -108,23 +118,20 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 108 | 118 | ||
| 109 | case QPDFTokenizer::tt_array_open: | 119 | case QPDFTokenizer::tt_array_open: |
| 110 | case QPDFTokenizer::tt_dict_open: | 120 | case QPDFTokenizer::tt_dict_open: |
| 111 | - if (olist_stack.size() > 500) { | 121 | + if (stack.size() > 500) { |
| 112 | QTC::TC("qpdf", "QPDFParser too deep"); | 122 | QTC::TC("qpdf", "QPDFParser too deep"); |
| 113 | warn("ignoring excessively deeply nested data structure"); | 123 | warn("ignoring excessively deeply nested data structure"); |
| 114 | bad = true; | 124 | bad = true; |
| 115 | object = QPDFObjectHandle::newNull(); | 125 | object = QPDFObjectHandle::newNull(); |
| 116 | state = st_top; | 126 | state = st_top; |
| 117 | } else { | 127 | } else { |
| 118 | - olist_stack.push_back(OHVector()); | ||
| 119 | state = st_start; | 128 | state = st_start; |
| 120 | - offset_stack.push_back(input->tell()); | ||
| 121 | state_stack.push_back( | 129 | state_stack.push_back( |
| 122 | (token.getType() == QPDFTokenizer::tt_array_open) | 130 | (token.getType() == QPDFTokenizer::tt_array_open) |
| 123 | ? st_array | 131 | ? st_array |
| 124 | : st_dictionary); | 132 | : st_dictionary); |
| 125 | b_contents = false; | 133 | b_contents = false; |
| 126 | - contents_string_stack.push_back(""); | ||
| 127 | - contents_offset_stack.push_back(-1); | 134 | + stack.push_back(StackFrame(input)); |
| 128 | } | 135 | } |
| 129 | break; | 136 | break; |
| 130 | 137 | ||
| @@ -206,8 +213,8 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -206,8 +213,8 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 206 | std::string val = token.getValue(); | 213 | std::string val = token.getValue(); |
| 207 | if (decrypter) { | 214 | if (decrypter) { |
| 208 | if (b_contents) { | 215 | if (b_contents) { |
| 209 | - contents_string = val; | ||
| 210 | - contents_offset = input->getLastOffset(); | 216 | + frame.contents_string = val; |
| 217 | + frame.contents_offset = input->getLastOffset(); | ||
| 211 | b_contents = false; | 218 | b_contents = false; |
| 212 | } | 219 | } |
| 213 | decrypter->decryptString(val); | 220 | decrypter->decryptString(val); |
| @@ -279,7 +286,7 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -279,7 +286,7 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 279 | break; | 286 | break; |
| 280 | 287 | ||
| 281 | case st_stop: | 288 | case st_stop: |
| 282 | - if ((state_stack.size() < 2) || (olist_stack.size() < 2)) { | 289 | + if ((state_stack.size() < 2) || (stack.size() < 2)) { |
| 283 | throw std::logic_error( | 290 | throw std::logic_error( |
| 284 | "QPDFObjectHandle::parseInternal: st_stop encountered" | 291 | "QPDFObjectHandle::parseInternal: st_stop encountered" |
| 285 | " with insufficient elements in stack"); | 292 | " with insufficient elements in stack"); |
| @@ -354,13 +361,13 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -354,13 +361,13 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 354 | } | 361 | } |
| 355 | dict[key] = val; | 362 | dict[key] = val; |
| 356 | } | 363 | } |
| 357 | - if (!contents_string.empty() && dict.count("/Type") && | 364 | + if (!frame.contents_string.empty() && dict.count("/Type") && |
| 358 | dict["/Type"].isNameAndEquals("/Sig") && | 365 | dict["/Type"].isNameAndEquals("/Sig") && |
| 359 | dict.count("/ByteRange") && dict.count("/Contents") && | 366 | dict.count("/ByteRange") && dict.count("/Contents") && |
| 360 | dict["/Contents"].isString()) { | 367 | dict["/Contents"].isString()) { |
| 361 | dict["/Contents"] = | 368 | dict["/Contents"] = |
| 362 | - QPDFObjectHandle::newString(contents_string); | ||
| 363 | - dict["/Contents"].setParsedOffset(contents_offset); | 369 | + QPDFObjectHandle::newString(frame.contents_string); |
| 370 | + dict["/Contents"].setParsedOffset(frame.contents_offset); | ||
| 364 | } | 371 | } |
| 365 | object = QPDFObjectHandle::newDictionary(dict); | 372 | object = QPDFObjectHandle::newDictionary(dict); |
| 366 | setDescriptionFromInput(object, offset); | 373 | setDescriptionFromInput(object, offset); |
| @@ -372,15 +379,12 @@ QPDFParser::parse(bool& empty, bool content_stream) | @@ -372,15 +379,12 @@ QPDFParser::parse(bool& empty, bool content_stream) | ||
| 372 | object.setParsedOffset(offset - 2); | 379 | object.setParsedOffset(offset - 2); |
| 373 | set_offset = true; | 380 | set_offset = true; |
| 374 | } | 381 | } |
| 375 | - olist_stack.pop_back(); | ||
| 376 | - offset_stack.pop_back(); | 382 | + stack.pop_back(); |
| 377 | if (state_stack.back() == st_top) { | 383 | if (state_stack.back() == st_top) { |
| 378 | done = true; | 384 | done = true; |
| 379 | } else { | 385 | } else { |
| 380 | - olist_stack.back().push_back(object); | 386 | + stack.back().olist.push_back(object); |
| 381 | } | 387 | } |
| 382 | - contents_string_stack.pop_back(); | ||
| 383 | - contents_offset_stack.pop_back(); | ||
| 384 | } | 388 | } |
| 385 | } | 389 | } |
| 386 | 390 |