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

Backed out changeset 77af189b5c49 (bug 1373984) for build bustage in nsHtml5Parser.cpp a=backout CLOSED TREE

Browse Source 
MozReview-Commit-ID: 6kBmU71j2To
This commit is contained in:
Wes Kocher
2017-06-25 05:10:14 -07:00
parent de5721ea81
commit d214d1f5fa
81 changed files with 446 additions and 579 deletions
Download Patch File Download Diff File Expand all files Collapse all files

103
parser/html/nsHtml5StreamParser.cpp
View File

@@ -152,7 +152,6 @@ nsHtml5StreamParser::nsHtml5StreamParser(nsHtml5TreeOpExecutor* aExecutor,
: mSniffingLength(0)
, mBomState(eBomState::BOM_SNIFFING_NOT_STARTED)
, mCharsetSource(kCharsetUninitialized)
, mEncoding(WINDOWS_1252_ENCODING)
, mReparseForbidden(false)
, mLastBuffer(nullptr) // Will be filled when starting
, mExecutor(aExecutor)
@@ -254,21 +253,23 @@ nsHtml5StreamParser::Notify(const char* aCharset, nsDetectionConfident aConf)
NS_ASSERTION(IsParserThread(), "Wrong thread!");
if (aConf == eBestAnswer || aConf == eSureAnswer) {
mFeedChardet = false; // just in case
auto encoding = Encoding::ForLabelNoReplacement(
const Encoding* encoding = Encoding::ForLabelNoReplacement(
nsDependentCString(aCharset));
if (!encoding) {
return NS_OK;
}
nsAutoCString charset;
encoding->Name(charset);
if (HasDecoder()) {
if (mEncoding == encoding) {
if (mCharset.Equals(charset)) {
NS_ASSERTION(mCharsetSource < kCharsetFromAutoDetection,
"Why are we running chardet at all?");
mCharsetSource = kCharsetFromAutoDetection;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
} else {
// We've already committed to a decoder. Request a reload from the
// docshell.
mTreeBuilder->NeedsCharsetSwitchTo(WrapNotNull(encoding),
mTreeBuilder->NeedsCharsetSwitchTo(charset,
kCharsetFromAutoDetection,
0);
FlushTreeOpsAndDisarmTimer();
@@ -277,9 +278,9 @@ nsHtml5StreamParser::Notify(const char* aCharset, nsDetectionConfident aConf)
} else {
// Got a confident answer from the sniffing buffer. That code will
// take care of setting up the decoder.
mEncoding = WrapNotNull(encoding);
mCharset.Assign(charset);
mCharsetSource = kCharsetFromAutoDetection;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
}
}
return NS_OK;
@@ -320,7 +321,7 @@ nsHtml5StreamParser::SetupDecodingAndWriteSniffingBufferAndCurrentSegment(const
{
NS_ASSERTION(IsParserThread(), "Wrong thread!");
nsresult rv = NS_OK;
mUnicodeDecoder = mEncoding->NewDecoderWithBOMRemoval();
mUnicodeDecoder = Encoding::ForName(mCharset)->NewDecoderWithBOMRemoval();
if (mSniffingBuffer) {
uint32_t writeCount;
rv = WriteStreamBytes(mSniffingBuffer.get(), mSniffingLength, &writeCount);
@@ -335,14 +336,14 @@ nsHtml5StreamParser::SetupDecodingAndWriteSniffingBufferAndCurrentSegment(const
}
nsresult
nsHtml5StreamParser::SetupDecodingFromBom(NotNull<const Encoding*> aEncoding)
nsHtml5StreamParser::SetupDecodingFromBom(const char* aDecoderCharsetName)
{
NS_ASSERTION(IsParserThread(), "Wrong thread!");
mEncoding = aEncoding;
mUnicodeDecoder = mEncoding->NewDecoderWithBOMRemoval();
mCharset.Assign(aDecoderCharsetName);
mUnicodeDecoder = Encoding::ForName(mCharset)->NewDecoderWithBOMRemoval();
mCharsetSource = kCharsetFromByteOrderMark;
mFeedChardet = false;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
mSniffingBuffer = nullptr;
mMetaScanner = nullptr;
mBomState = BOM_SNIFFING_OVER;
@@ -397,12 +398,12 @@ nsHtml5StreamParser::SniffBOMlessUTF16BasicLatin(const uint8_t* aFromSegment,
}
if (byteNonZero[0]) {
mEncoding = UTF_16LE_ENCODING;
mCharset.AssignLiteral("UTF-16LE");
} else {
mEncoding = UTF_16BE_ENCODING;
mCharset.AssignLiteral("UTF-16BE");
}
mCharsetSource = kCharsetFromIrreversibleAutoDetection;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
mFeedChardet = false;
mTreeBuilder->MaybeComplainAboutCharset("EncBomlessUtf16",
true,
@@ -417,9 +418,8 @@ nsHtml5StreamParser::SetEncodingFromExpat(const char16_t* aEncoding)
nsDependentString utf16(aEncoding);
nsAutoCString utf8;
CopyUTF16toUTF8(utf16, utf8);
auto encoding = PreferredForInternalEncodingDecl(utf8);
if (encoding) {
mEncoding = WrapNotNull(encoding);
if (PreferredForInternalEncodingDecl(utf8)) {
mCharset.Assign(utf8);
mCharsetSource = kCharsetFromMetaTag; // closest for XML
return;
}
@@ -428,7 +428,7 @@ nsHtml5StreamParser::SetEncodingFromExpat(const char16_t* aEncoding)
// right away and let the encoding be set to UTF-8 which we'd default to
// anyway.
}
mEncoding = UTF_8_ENCODING; // XML defaults to UTF-8 without a BOM
mCharset.AssignLiteral("UTF-8"); // XML defaults to UTF-8 without a BOM
mCharsetSource = kCharsetFromMetaTag; // means confident
}
@@ -560,7 +560,7 @@ nsHtml5StreamParser::FinalizeSniffing(const uint8_t* aFromSegment, // can be nul
// confidently to UTF-8 in this case.
// It is also possible that the document has an XML declaration that is
// longer than 1024 bytes, but that case is not worth worrying about.
mEncoding = UTF_8_ENCODING;
mCharset.AssignLiteral("UTF-8");
mCharsetSource = kCharsetFromMetaTag; // means confident
}
@@ -612,18 +612,18 @@ nsHtml5StreamParser::FinalizeSniffing(const uint8_t* aFromSegment, // can be nul
}
if (mCharsetSource == kCharsetUninitialized) {
// Hopefully this case is never needed, but dealing with it anyway
mEncoding = WINDOWS_1252_ENCODING;
mCharset.AssignLiteral("windows-1252");
mCharsetSource = kCharsetFromFallback;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
} else if (mMode == LOAD_AS_DATA &&
mCharsetSource == kCharsetFromFallback) {
NS_ASSERTION(mReparseForbidden, "Reparse should be forbidden for XHR");
NS_ASSERTION(!mFeedChardet, "Should not feed chardet for XHR");
NS_ASSERTION(mEncoding == UTF_8_ENCODING,
NS_ASSERTION(mCharset.EqualsLiteral("UTF-8"),
"XHR should default to UTF-8");
// Now mark charset source as non-weak to signal that we have a decision
mCharsetSource = kCharsetFromDocTypeDefault;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
}
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment, aCount, aWriteCount);
}
@@ -637,9 +637,9 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
nsresult rv = NS_OK;
uint32_t writeCount;
// mEncoding and mCharsetSource potentially have come from channel or higher
// mCharset and mCharsetSource potentially have come from channel or higher
// by now. If we find a BOM, SetupDecodingFromBom() will overwrite them.
// If we don't find a BOM, the previously set values of mEncoding and
// If we don't find a BOM, the previously set values of mCharset and
// mCharsetSource are not modified by the BOM sniffing here.
for (uint32_t i = 0; i < aCount && mBomState != BOM_SNIFFING_OVER; i++) {
switch (mBomState) {
@@ -662,7 +662,7 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
break;
case SEEN_UTF_16_LE_FIRST_BYTE:
if (aFromSegment[i] == 0xFE) {
rv = SetupDecodingFromBom(UTF_16LE_ENCODING); // upper case is the raw form
rv = SetupDecodingFromBom("UTF-16LE"); // upper case is the raw form
NS_ENSURE_SUCCESS(rv, rv);
uint32_t count = aCount - (i + 1);
rv = WriteStreamBytes(aFromSegment + (i + 1), count, &writeCount);
@@ -674,7 +674,7 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
break;
case SEEN_UTF_16_BE_FIRST_BYTE:
if (aFromSegment[i] == 0xFF) {
rv = SetupDecodingFromBom(UTF_16BE_ENCODING); // upper case is the raw form
rv = SetupDecodingFromBom("UTF-16BE"); // upper case is the raw form
NS_ENSURE_SUCCESS(rv, rv);
uint32_t count = aCount - (i + 1);
rv = WriteStreamBytes(aFromSegment + (i + 1), count, &writeCount);
@@ -693,7 +693,7 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
break;
case SEEN_UTF_8_SECOND_BYTE:
if (aFromSegment[i] == 0xBF) {
rv = SetupDecodingFromBom(UTF_8_ENCODING); // upper case is the raw form
rv = SetupDecodingFromBom("UTF-8"); // upper case is the raw form
NS_ENSURE_SUCCESS(rv, rv);
uint32_t count = aCount - (i + 1);
rv = WriteStreamBytes(aFromSegment + (i + 1), count, &writeCount);
@@ -718,13 +718,13 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
if (mBomState == BOM_SNIFFING_OVER &&
mCharsetSource == kCharsetFromChannel) {
// There was no BOM and the charset came from channel. mEncoding
// There was no BOM and the charset came from channel. mCharset
// still contains the charset from the channel as set by an
// earlier call to SetDocumentCharset(), since we didn't find a BOM and
// overwrite mEncoding. (Note that if the user has overridden the charset,
// overwrite mCharset. (Note that if the user has overridden the charset,
// we don't come here but check <meta> for XSS-dangerous charsets first.)
mFeedChardet = false;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment,
aCount, aWriteCount);
}
@@ -743,14 +743,15 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
nsHtml5ByteReadable readable(aFromSegment, aFromSegment +
countToSniffingLimit);
nsAutoCString charset;
auto encoding = mMetaScanner->sniff(&readable);
mMetaScanner->sniff(&readable, charset);
// Due to the way nsHtml5Portability reports OOM, ask the tree buider
nsresult rv;
if (NS_FAILED((rv = mTreeBuilder->IsBroken()))) {
MarkAsBroken(rv);
return rv;
}
if (encoding) {
if (!charset.IsEmpty()) {
const Encoding* encoding = Encoding::ForName(charset);
// meta scan successful; honor overrides unless meta is XSS-dangerous
if ((mCharsetSource == kCharsetFromParentForced ||
mCharsetSource == kCharsetFromUserForced) &&
@@ -760,10 +761,10 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(
aFromSegment, aCount, aWriteCount);
}
mEncoding = WrapNotNull(encoding);
mCharset.Assign(charset);
mCharsetSource = kCharsetFromMetaPrescan;
mFeedChardet = false;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(
aFromSegment, aCount, aWriteCount);
}
@@ -781,14 +782,16 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
// not the last buffer
if (mMode == NORMAL || mMode == VIEW_SOURCE_HTML || mMode == LOAD_AS_DATA) {
nsHtml5ByteReadable readable(aFromSegment, aFromSegment + aCount);
auto encoding = mMetaScanner->sniff(&readable);
nsAutoCString charset;
mMetaScanner->sniff(&readable, charset);
// Due to the way nsHtml5Portability reports OOM, ask the tree buider
nsresult rv;
if (NS_FAILED((rv = mTreeBuilder->IsBroken()))) {
MarkAsBroken(rv);
return rv;
}
if (encoding) {
if (!charset.IsEmpty()) {
const Encoding* encoding = Encoding::ForName(charset);
// meta scan successful; honor overrides unless meta is XSS-dangerous
if ((mCharsetSource == kCharsetFromParentForced ||
mCharsetSource == kCharsetFromUserForced) &&
@@ -798,10 +801,10 @@ nsHtml5StreamParser::SniffStreamBytes(const uint8_t* aFromSegment,
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment,
aCount, aWriteCount);
}
mEncoding = WrapNotNull(encoding);
mCharset.Assign(charset);
mCharsetSource = kCharsetFromMetaPrescan;
mFeedChardet = false;
mTreeBuilder->SetDocumentCharset(mEncoding, mCharsetSource);
mTreeBuilder->SetDocumentCharset(mCharset, mCharsetSource);
return SetupDecodingAndWriteSniffingBufferAndCurrentSegment(aFromSegment,
aCount, aWriteCount);
}
@@ -1004,7 +1007,7 @@ nsHtml5StreamParser::OnStartRequest(nsIRequest* aRequest, nsISupports* aContext)
mFeedChardet = false;
// Instantiate the converter here to avoid BOM sniffing.
mUnicodeDecoder = mEncoding->NewDecoderWithBOMRemoval();
mUnicodeDecoder = Encoding::ForName(mCharset)->NewDecoderWithBOMRemoval();
return NS_OK;
}
@@ -1228,8 +1231,8 @@ nsHtml5StreamParser::CopySegmentsToParser(nsIInputStream *aInStream,
return NS_OK;
}
const Encoding*
nsHtml5StreamParser::PreferredForInternalEncodingDecl(const nsACString& aEncoding)
bool
nsHtml5StreamParser::PreferredForInternalEncodingDecl(nsACString& aEncoding)
{
const Encoding* newEncoding = Encoding::ForLabel(aEncoding);
if (!newEncoding) {
@@ -1237,7 +1240,7 @@ nsHtml5StreamParser::PreferredForInternalEncodingDecl(const nsACString& aEncodin
mTreeBuilder->MaybeComplainAboutCharset("EncMetaUnsupported",
true,
mTokenizer->getLineNumber());
return nullptr;
return false;
}
if (newEncoding == UTF_16BE_ENCODING ||
@@ -1256,7 +1259,7 @@ nsHtml5StreamParser::PreferredForInternalEncodingDecl(const nsACString& aEncodin
newEncoding = WINDOWS_1252_ENCODING;
}
if (newEncoding == mEncoding) {
if (newEncoding == Encoding::ForName(mCharset)) {
if (mCharsetSource < kCharsetFromMetaPrescan) {
if (mInitialEncodingWasFromParentFrame) {
mTreeBuilder->MaybeComplainAboutCharset("EncLateMetaFrame",
@@ -1270,10 +1273,11 @@ nsHtml5StreamParser::PreferredForInternalEncodingDecl(const nsACString& aEncodin
}
mCharsetSource = kCharsetFromMetaTag; // become confident
mFeedChardet = false; // don't feed chardet when confident
return nullptr;
return false;
}
return newEncoding;
newEncoding->Name(aEncoding);
return true;
}
bool
@@ -1292,8 +1296,7 @@ nsHtml5StreamParser::internalEncodingDeclaration(nsHtml5String aEncoding)
nsAutoCString newEncoding;
CopyUTF16toUTF8(newEncoding16, newEncoding);
auto encoding = PreferredForInternalEncodingDecl(newEncoding);
if (!encoding) {
if (!PreferredForInternalEncodingDecl(newEncoding)) {
return false;
}
@@ -1311,7 +1314,7 @@ nsHtml5StreamParser::internalEncodingDeclaration(nsHtml5String aEncoding)
// Avoid having the chardet ask for another restart after this restart
// request.
mFeedChardet = false;
mTreeBuilder->NeedsCharsetSwitchTo(WrapNotNull(encoding),
mTreeBuilder->NeedsCharsetSwitchTo(newEncoding,
kCharsetFromMetaTag,
mTokenizer->getLineNumber());
FlushTreeOpsAndDisarmTimer();