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

Backed out changeset dc6b9ca8f3fa (bug 1727491) for causing mochitest failures on test_bug631751be.html. CLOSED TREE

Browse Source
This commit is contained in:
criss
2021-08-30 11:14:38 +03:00
parent b40a4cb123
commit c360a77168
8 changed files with 64 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

55
parser/html/nsHtml5StreamParser.cpp
View File

@@ -503,6 +503,44 @@ 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);
@@ -696,9 +734,15 @@ 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) {
if (mForceAutoDetection &&
mCharsetSource != kCharsetFromIrreversibleAutoDetection) {
// neither meta nor XML declaration found, honor override
FinalizeSniffingWithDetector(aFromSegment, aCountToSniffingLimit, false);
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment);
@@ -893,10 +937,11 @@ nsresult nsHtml5StreamParser::SniffStreamBytes(
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment);
}
MOZ_ASSERT(!(mBomState == BOM_SNIFFING_OVER && mChannelHadCharset &&
!mForceAutoDetection),
"How come we're running post-BOM sniffing with channel charset unless "
"we're also processing forced detection?");
MOZ_ASSERT(
!(mBomState == BOM_SNIFFING_OVER && mChannelHadCharset &&
!mForceAutoDetection),
"How come we're running post-BOM sniffing with channel charset unless "
"we're also processing forced detection?");
if (!mMetaScanner &&
(mMode == NORMAL || mMode == VIEW_SOURCE_HTML || mMode == LOAD_AS_DATA)) {