corysanin/tubestation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Bug 1727491 - Remove support for BOMless unlabeled Latin1 Supplement-range UTF-16LE|BE. r=emk

Browse Source 
Differential Revision: https://phabricator.services.mozilla.com/D123596
This commit is contained in:
Henri Sivonen
2021-09-01 09:13:29 +00:00
parent 49c5ba8880
commit ae6cd4b683
10 changed files with 3 additions and 61 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
parser/html/nsHtml5StreamParser.cpp
View File

@@ -503,44 +503,6 @@ void nsHtml5StreamParser::SetupDecodingFromUtf16BogoXml(
mLastBuffer->AdvanceEnd(3);
}
void nsHtml5StreamParser::SniffBOMlessUTF16BasicLatin(const uint8_t* aBuf,
size_t aBufLen) {
// Avoid underspecified heuristic craziness for XHR
if (mMode == LOAD_AS_DATA) {
return;
}
// Make sure there's enough data. Require room for "<title></title>"
if (aBufLen < 30) {
return;
}
// even-numbered bytes tracked at 0, odd-numbered bytes tracked at 1
bool byteZero[2] = {false, false};
bool byteNonZero[2] = {false, false};
uint32_t i = 0;
for (; i < aBufLen; ++i) {
if (aBuf[i]) {
if (byteNonZero[1 - (i % 2)]) {
return;
}
byteNonZero[i % 2] = true;
} else {
if (byteZero[1 - (i % 2)]) {
return;
}
byteZero[i % 2] = true;
}
}
if (byteNonZero[0]) {
mEncoding = UTF_16LE_ENCODING;
} else {
mEncoding = UTF_16BE_ENCODING;
}
mCharsetSource = kCharsetFromIrreversibleAutoDetection;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
DontGuessEncoding();
mTreeBuilder->MaybeComplainAboutCharset("EncBomlessUtf16", true, 0);
}
void nsHtml5StreamParser::SetEncodingFromExpat(const char16_t* aEncoding) {
if (aEncoding) {
nsDependentString utf16(aEncoding);
@@ -734,15 +696,9 @@ nsresult nsHtml5StreamParser::FinalizeSniffing(Span<const uint8_t> aFromSegment,
mEncoding = WrapNotNull(encoding);
mCharsetSource = kCharsetFromXmlDeclaration;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
} else if (mCharsetSource < kCharsetFromIrreversibleAutoDetection) {
// meta scan and XML declaration check failed.
// Check for BOMless UTF-16 with Basic
// Latin content for compat with IE. See bug 631751.
SniffBOMlessUTF16BasicLatin(buf, bufLen);
}
}
if (mForceAutoDetection &&
mCharsetSource != kCharsetFromIrreversibleAutoDetection) {
if (mForceAutoDetection) {
// neither meta nor XML declaration found, honor override
FinalizeSniffingWithDetector(aFromSegment, aCountToSniffingLimit, false);
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment);