Home | History | Annotate | Download | only in xml 
      1 // Copyright 2016 PDFium Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 #include "xfa/fde/xml/fde_xml_imp.h"
      6 
      7 #include <memory>
      8 
      9 #include "testing/gtest/include/gtest/gtest.h"
     10 #include "xfa/fgas/crt/fgas_stream.h"
     11 
     12 TEST(CFDE_XMLSyntaxParser, CData) {
     13   const FX_WCHAR* input =
     14       L"<script contentType=\"application/x-javascript\">\n"
     15       L"  <![CDATA[\n"
     16       L"    if (a[1] < 3)\n"
     17       L"      app.alert(\"Tclams\");\n"
     18       L"  ]]>\n"
     19       L"</script>";
     20 
     21   const FX_WCHAR* cdata =
     22       L"\n"
     23       L"    if (a[1] < 3)\n"
     24       L"      app.alert(\"Tclams\");\n"
     25       L"  ";
     26 
     27   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
     28   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
     29   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
     30       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
     31   CFDE_XMLSyntaxParser parser;
     32   parser.Init(stream, 256);
     33   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
     34   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
     35 
     36   CFX_WideString data;
     37   parser.GetTagName(data);
     38   EXPECT_EQ(L"script", data);
     39 
     40   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
     41   parser.GetAttributeName(data);
     42   EXPECT_EQ(L"contentType", data);
     43   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
     44   parser.GetAttributeValue(data);
     45   EXPECT_EQ(L"application/x-javascript", data);
     46 
     47   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
     48   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
     49   parser.GetTextData(data);
     50   EXPECT_EQ(L"\n  ", data);
     51 
     52   EXPECT_EQ(FDE_XmlSyntaxResult::CData, parser.DoSyntaxParse());
     53   parser.GetTextData(data);
     54   EXPECT_EQ(cdata, data);
     55 
     56   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
     57   parser.GetTextData(data);
     58   EXPECT_EQ(L"\n", data);
     59 
     60   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
     61   parser.GetTagName(data);
     62   EXPECT_EQ(L"script", data);
     63 
     64   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
     65 }
     66 
     67 TEST(CFDE_XMLSyntaxParser, CDataWithInnerScript) {
     68   const FX_WCHAR* input =
     69       L"<script contentType=\"application/x-javascript\">\n"
     70       L"  <![CDATA[\n"
     71       L"    if (a[1] < 3)\n"
     72       L"      app.alert(\"Tclams\");\n"
     73       L"    </script>\n"
     74       L"  ]]>\n"
     75       L"</script>";
     76 
     77   const FX_WCHAR* cdata =
     78       L"\n"
     79       L"    if (a[1] < 3)\n"
     80       L"      app.alert(\"Tclams\");\n"
     81       L"    </script>\n"
     82       L"  ";
     83 
     84   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
     85   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
     86   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
     87       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
     88   CFDE_XMLSyntaxParser parser;
     89   parser.Init(stream, 256);
     90   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
     91   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
     92 
     93   CFX_WideString data;
     94   parser.GetTagName(data);
     95   EXPECT_EQ(L"script", data);
     96 
     97   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
     98   parser.GetAttributeName(data);
     99   EXPECT_EQ(L"contentType", data);
    100   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    101   parser.GetAttributeValue(data);
    102   EXPECT_EQ(L"application/x-javascript", data);
    103 
    104   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    105   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    106   parser.GetTextData(data);
    107   EXPECT_EQ(L"\n  ", data);
    108 
    109   EXPECT_EQ(FDE_XmlSyntaxResult::CData, parser.DoSyntaxParse());
    110   parser.GetTextData(data);
    111   EXPECT_EQ(cdata, data);
    112 
    113   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    114   parser.GetTextData(data);
    115   EXPECT_EQ(L"\n", data);
    116 
    117   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    118   parser.GetTagName(data);
    119   EXPECT_EQ(L"script", data);
    120 
    121   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    122 }
    123 
    124 TEST(CFDE_XMLSyntaxParser, ArrowBangArrow) {
    125   const FX_WCHAR* input =
    126       L"<script contentType=\"application/x-javascript\">\n"
    127       L"  <!>\n"
    128       L"</script>";
    129 
    130   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    131   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    132   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    133       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    134   CFDE_XMLSyntaxParser parser;
    135   parser.Init(stream, 256);
    136   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    137   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    138 
    139   CFX_WideString data;
    140   parser.GetTagName(data);
    141   EXPECT_EQ(L"script", data);
    142 
    143   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    144   parser.GetAttributeName(data);
    145   EXPECT_EQ(L"contentType", data);
    146   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    147   parser.GetAttributeValue(data);
    148   EXPECT_EQ(L"application/x-javascript", data);
    149 
    150   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    151   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    152   parser.GetTextData(data);
    153   EXPECT_EQ(L"\n  ", data);
    154 
    155   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    156   parser.GetTextData(data);
    157   EXPECT_EQ(L"\n", data);
    158 
    159   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    160   parser.GetTagName(data);
    161   EXPECT_EQ(L"script", data);
    162 
    163   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    164 }
    165 
    166 TEST(CFDE_XMLSyntaxParser, ArrowBangBracketArrow) {
    167   const FX_WCHAR* input =
    168       L"<script contentType=\"application/x-javascript\">\n"
    169       L"  <![>\n"
    170       L"</script>";
    171 
    172   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    173   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    174   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    175       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    176   CFDE_XMLSyntaxParser parser;
    177   parser.Init(stream, 256);
    178   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    179   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    180 
    181   CFX_WideString data;
    182   parser.GetTagName(data);
    183   EXPECT_EQ(L"script", data);
    184 
    185   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    186   parser.GetAttributeName(data);
    187   EXPECT_EQ(L"contentType", data);
    188   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    189   parser.GetAttributeValue(data);
    190   EXPECT_EQ(L"application/x-javascript", data);
    191 
    192   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    193   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    194   parser.GetTextData(data);
    195   EXPECT_EQ(L"\n  ", data);
    196 
    197   // Parser walks to end of input.
    198 
    199   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    200 }
    201 
    202 TEST(CFDE_XMLSyntaxParser, IncompleteCData) {
    203   const FX_WCHAR* input =
    204       L"<script contentType=\"application/x-javascript\">\n"
    205       L"  <![CDATA>\n"
    206       L"</script>";
    207 
    208   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    209   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    210   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    211       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    212   CFDE_XMLSyntaxParser parser;
    213   parser.Init(stream, 256);
    214   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    215   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    216 
    217   CFX_WideString data;
    218   parser.GetTagName(data);
    219   EXPECT_EQ(L"script", data);
    220 
    221   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    222   parser.GetAttributeName(data);
    223   EXPECT_EQ(L"contentType", data);
    224   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    225   parser.GetAttributeValue(data);
    226   EXPECT_EQ(L"application/x-javascript", data);
    227 
    228   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    229   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    230   parser.GetTextData(data);
    231   EXPECT_EQ(L"\n  ", data);
    232 
    233   // Parser walks to end of input.
    234 
    235   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    236 }
    237 
    238 TEST(CFDE_XMLSyntaxParser, UnClosedCData) {
    239   const FX_WCHAR* input =
    240       L"<script contentType=\"application/x-javascript\">\n"
    241       L"  <![CDATA[\n"
    242       L"</script>";
    243 
    244   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    245   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    246   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    247       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    248   CFDE_XMLSyntaxParser parser;
    249   parser.Init(stream, 256);
    250   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    251   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    252 
    253   CFX_WideString data;
    254   parser.GetTagName(data);
    255   EXPECT_EQ(L"script", data);
    256 
    257   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    258   parser.GetAttributeName(data);
    259   EXPECT_EQ(L"contentType", data);
    260   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    261   parser.GetAttributeValue(data);
    262   EXPECT_EQ(L"application/x-javascript", data);
    263 
    264   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    265   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    266   parser.GetTextData(data);
    267   EXPECT_EQ(L"\n  ", data);
    268 
    269   // Parser walks to end of input.
    270 
    271   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    272 }
    273 
    274 TEST(CFDE_XMLSyntaxParser, EmptyCData) {
    275   const FX_WCHAR* input =
    276       L"<script contentType=\"application/x-javascript\">\n"
    277       L"  <![CDATA[]]>\n"
    278       L"</script>";
    279 
    280   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    281   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    282   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    283       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    284   CFDE_XMLSyntaxParser parser;
    285   parser.Init(stream, 256);
    286   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    287   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    288 
    289   CFX_WideString data;
    290   parser.GetTagName(data);
    291   EXPECT_EQ(L"script", data);
    292 
    293   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    294   parser.GetAttributeName(data);
    295   EXPECT_EQ(L"contentType", data);
    296   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    297   parser.GetAttributeValue(data);
    298   EXPECT_EQ(L"application/x-javascript", data);
    299 
    300   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    301   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    302   parser.GetTextData(data);
    303   EXPECT_EQ(L"\n  ", data);
    304 
    305   EXPECT_EQ(FDE_XmlSyntaxResult::CData, parser.DoSyntaxParse());
    306   parser.GetTextData(data);
    307   EXPECT_EQ(L"", data);
    308 
    309   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    310   parser.GetTextData(data);
    311   EXPECT_EQ(L"\n", data);
    312 
    313   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    314   parser.GetTagName(data);
    315   EXPECT_EQ(L"script", data);
    316 
    317   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    318 }
    319 
    320 TEST(CFDE_XMLSyntaxParser, Comment) {
    321   const FX_WCHAR* input =
    322       L"<script contentType=\"application/x-javascript\">\n"
    323       L"  <!-- A Comment -->\n"
    324       L"</script>";
    325 
    326   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    327   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    328   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    329       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    330   CFDE_XMLSyntaxParser parser;
    331   parser.Init(stream, 256);
    332   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    333   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    334 
    335   CFX_WideString data;
    336   parser.GetTagName(data);
    337   EXPECT_EQ(L"script", data);
    338 
    339   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    340   parser.GetAttributeName(data);
    341   EXPECT_EQ(L"contentType", data);
    342   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    343   parser.GetAttributeValue(data);
    344   EXPECT_EQ(L"application/x-javascript", data);
    345 
    346   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    347   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    348   parser.GetTextData(data);
    349   EXPECT_EQ(L"\n  ", data);
    350 
    351   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    352   parser.GetTextData(data);
    353   EXPECT_EQ(L"\n", data);
    354 
    355   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    356   parser.GetTagName(data);
    357   EXPECT_EQ(L"script", data);
    358 
    359   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    360 }
    361 
    362 TEST(CFDE_XMLSyntaxParser, IncorrectCommentStart) {
    363   const FX_WCHAR* input =
    364       L"<script contentType=\"application/x-javascript\">\n"
    365       L"  <!- A Comment -->\n"
    366       L"</script>";
    367 
    368   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    369   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    370   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    371       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    372   CFDE_XMLSyntaxParser parser;
    373   parser.Init(stream, 256);
    374   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    375   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    376 
    377   CFX_WideString data;
    378   parser.GetTagName(data);
    379   EXPECT_EQ(L"script", data);
    380 
    381   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    382   parser.GetAttributeName(data);
    383   EXPECT_EQ(L"contentType", data);
    384   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    385   parser.GetAttributeValue(data);
    386   EXPECT_EQ(L"application/x-javascript", data);
    387 
    388   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    389   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    390   parser.GetTextData(data);
    391   EXPECT_EQ(L"\n  ", data);
    392 
    393   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    394   parser.GetTextData(data);
    395   EXPECT_EQ(L"\n", data);
    396 
    397   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    398   parser.GetTagName(data);
    399   EXPECT_EQ(L"script", data);
    400 
    401   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    402 }
    403 
    404 TEST(CFDE_XMLSyntaxParser, CommentEmpty) {
    405   const FX_WCHAR* input =
    406       L"<script contentType=\"application/x-javascript\">\n"
    407       L"  <!---->\n"
    408       L"</script>";
    409 
    410   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    411   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    412   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    413       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    414   CFDE_XMLSyntaxParser parser;
    415   parser.Init(stream, 256);
    416   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    417   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    418 
    419   CFX_WideString data;
    420   parser.GetTagName(data);
    421   EXPECT_EQ(L"script", data);
    422 
    423   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    424   parser.GetAttributeName(data);
    425   EXPECT_EQ(L"contentType", data);
    426   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    427   parser.GetAttributeValue(data);
    428   EXPECT_EQ(L"application/x-javascript", data);
    429 
    430   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    431   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    432   parser.GetTextData(data);
    433   EXPECT_EQ(L"\n  ", data);
    434 
    435   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    436   parser.GetTextData(data);
    437   EXPECT_EQ(L"\n", data);
    438 
    439   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    440   parser.GetTagName(data);
    441   EXPECT_EQ(L"script", data);
    442 
    443   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    444 }
    445 
    446 TEST(CFDE_XMLSyntaxParser, CommentThreeDash) {
    447   const FX_WCHAR* input =
    448       L"<script contentType=\"application/x-javascript\">\n"
    449       L"  <!--->\n"
    450       L"</script>";
    451 
    452   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    453   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    454   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    455       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    456   CFDE_XMLSyntaxParser parser;
    457   parser.Init(stream, 256);
    458   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    459   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    460 
    461   CFX_WideString data;
    462   parser.GetTagName(data);
    463   EXPECT_EQ(L"script", data);
    464 
    465   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    466   parser.GetAttributeName(data);
    467   EXPECT_EQ(L"contentType", data);
    468   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    469   parser.GetAttributeValue(data);
    470   EXPECT_EQ(L"application/x-javascript", data);
    471 
    472   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    473   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    474   parser.GetTextData(data);
    475   EXPECT_EQ(L"\n  ", data);
    476 
    477   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    478 }
    479 
    480 TEST(CFDE_XMLSyntaxParser, CommentTwoDash) {
    481   const FX_WCHAR* input =
    482       L"<script contentType=\"application/x-javascript\">\n"
    483       L"  <!-->\n"
    484       L"</script>";
    485 
    486   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    487   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    488   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    489       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    490   CFDE_XMLSyntaxParser parser;
    491   parser.Init(stream, 256);
    492   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    493   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    494 
    495   CFX_WideString data;
    496   parser.GetTagName(data);
    497   EXPECT_EQ(L"script", data);
    498 
    499   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    500   parser.GetAttributeName(data);
    501   EXPECT_EQ(L"contentType", data);
    502   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    503   parser.GetAttributeValue(data);
    504   EXPECT_EQ(L"application/x-javascript", data);
    505 
    506   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    507   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    508   parser.GetTextData(data);
    509   EXPECT_EQ(L"\n  ", data);
    510 
    511   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    512 }
    513 
    514 TEST(CFDE_XMLSyntaxParser, Entities) {
    515   const FX_WCHAR* input =
    516       L"<script contentType=\"application/x-javascript\">"
    517       L"&#66;"
    518       L"&#x54;"
    519       L"&#x00000000000000000048;"
    520       L"&#x0000000000000000AB48;"
    521       L"&#x0000000000000000000;"
    522       L"</script>";
    523 
    524   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    525   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    526   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    527       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    528   CFDE_XMLSyntaxParser parser;
    529   parser.Init(stream, 256);
    530   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    531   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    532 
    533   CFX_WideString data;
    534   parser.GetTagName(data);
    535   EXPECT_EQ(L"script", data);
    536 
    537   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    538   parser.GetAttributeName(data);
    539   EXPECT_EQ(L"contentType", data);
    540   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    541   parser.GetAttributeValue(data);
    542   EXPECT_EQ(L"application/x-javascript", data);
    543 
    544   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    545   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    546   parser.GetTextData(data);
    547   EXPECT_EQ(L"BTH\xab48", data);
    548 
    549   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    550   parser.GetTagName(data);
    551   EXPECT_EQ(L"script", data);
    552 
    553   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    554 }
    555 
    556 TEST(CFDE_XMLSyntaxParser, EntityOverflowHex) {
    557   const FX_WCHAR* input =
    558       L"<script contentType=\"application/x-javascript\">"
    559       L"&#xaDBDFFFFF;"
    560       L"&#xafffffffffffffffffffffffffffffffff;"
    561       L"</script>";
    562 
    563   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    564   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    565   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    566       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    567   CFDE_XMLSyntaxParser parser;
    568   parser.Init(stream, 256);
    569   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    570   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    571 
    572   CFX_WideString data;
    573   parser.GetTagName(data);
    574   EXPECT_EQ(L"script", data);
    575 
    576   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    577   parser.GetAttributeName(data);
    578   EXPECT_EQ(L"contentType", data);
    579   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    580   parser.GetAttributeValue(data);
    581   EXPECT_EQ(L"application/x-javascript", data);
    582 
    583   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    584   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    585   parser.GetTextData(data);
    586   EXPECT_EQ(L"  ", data);
    587 
    588   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    589   parser.GetTagName(data);
    590   EXPECT_EQ(L"script", data);
    591 
    592   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    593 }
    594 
    595 TEST(CFDE_XMLSyntaxParser, EntityOverflowDecimal) {
    596   const FX_WCHAR* input =
    597       L"<script contentType=\"application/x-javascript\">"
    598       L"&#2914910205;"
    599       L"&#29149102052342342134521341234512351234213452315;"
    600       L"</script>";
    601 
    602   // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
    603   size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
    604   CFX_RetainPtr<IFGAS_Stream> stream = IFGAS_Stream::CreateStream(
    605       reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0);
    606   CFDE_XMLSyntaxParser parser;
    607   parser.Init(stream, 256);
    608   EXPECT_EQ(FDE_XmlSyntaxResult::ElementOpen, parser.DoSyntaxParse());
    609   EXPECT_EQ(FDE_XmlSyntaxResult::TagName, parser.DoSyntaxParse());
    610 
    611   CFX_WideString data;
    612   parser.GetTagName(data);
    613   EXPECT_EQ(L"script", data);
    614 
    615   EXPECT_EQ(FDE_XmlSyntaxResult::AttriName, parser.DoSyntaxParse());
    616   parser.GetAttributeName(data);
    617   EXPECT_EQ(L"contentType", data);
    618   EXPECT_EQ(FDE_XmlSyntaxResult::AttriValue, parser.DoSyntaxParse());
    619   parser.GetAttributeValue(data);
    620   EXPECT_EQ(L"application/x-javascript", data);
    621 
    622   EXPECT_EQ(FDE_XmlSyntaxResult::ElementBreak, parser.DoSyntaxParse());
    623   EXPECT_EQ(FDE_XmlSyntaxResult::Text, parser.DoSyntaxParse());
    624   parser.GetTextData(data);
    625   EXPECT_EQ(L"  ", data);
    626 
    627   EXPECT_EQ(FDE_XmlSyntaxResult::ElementClose, parser.DoSyntaxParse());
    628   parser.GetTagName(data);
    629   EXPECT_EQ(L"script", data);
    630 
    631   EXPECT_EQ(FDE_XmlSyntaxResult::EndOfString, parser.DoSyntaxParse());
    632 }
    633