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

META tags enclosed in a comment should be ignored. b=124904, r=dp/heikki, sr=darin, a=asa

Browse Source
This commit is contained in:
harishd@netscape.com
2002-03-19 21:41:41 +00:00
parent 38e7007990
commit 6ff5f64660
2 changed files with 192 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files

158
parser/htmlparser/src/nsParser.cpp
View File

@@ -2110,6 +2110,12 @@ static const PRInt32 kContentStrLen = sizeof(kContentStr)-1;
static const char kCharsetStr[] = "charset";
static const PRInt32 kCharsetStrLen = sizeof(kCharsetStr)-1;
inline const char GetNextChar(nsReadingIterator<char>& aStart,
nsReadingIterator<char>& aEnd)
{
return (aStart != aEnd) ? *(++aStart) : '\0';
}
PRBool
nsParser::DetectMetaTag(const char* aBytes,
PRInt32 aLen,
@@ -2136,77 +2142,105 @@ nsParser::DetectMetaTag(const char* aBytes,
str.EndReading(end);
nsReadingIterator<char> tagStart(begin);
nsReadingIterator<char> tagEnd(end);
nsReadingIterator<char> ltPos;
do {
// Find the string META and make sure it's not right at the beginning
if (CaseInsensitiveFindInReadable(NS_LITERAL_CSTRING("META"), tagStart, tagEnd) &&
(tagStart != begin)) {
// Back up one to confirm that this is a tag
if (*--tagStart == '<') {
const char* attrStart = tagEnd.get();
const char* attrEnd;
// Find the end of the tag
FindCharInReadable('>', tagEnd, end);
attrEnd = tagEnd.get();
CWordTokenizer<char> tokenizer(attrStart, 0, attrEnd-attrStart);
PRInt32 offset;
// Start looking at the attributes
while ((offset = tokenizer.GetNextWord()) != kNotFound) {
// We need to have a HTTP-EQUIV attribute whose value is
// "Content-Type"
if ((tokenizer.GetLength() >= kHTTPEquivStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kHTTPEquivStr, kHTTPEquivStrLen) == 0)) {
if (((offset = tokenizer.GetNextWord(PR_TRUE)) != kNotFound) &&
(tokenizer.GetLength() >= kContentTypeStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kContentTypeStr, kContentTypeStrLen) == 0)) {
foundContentType = PR_TRUE;
}
else {
break;
}
while (tagStart != end) {
if (FindCharInReadable('<', tagStart, end)) {
ltPos = tagEnd = tagStart;
if (GetNextChar(ltPos, end) == '!' &&
GetNextChar(ltPos, end) == '-' &&
GetNextChar(ltPos, end) == '-') {
// Found MDO ( <!-- ). Now search for MDC ( --[*s]> )
PRBool foundMDC = PR_FALSE;
PRBool foundMatch = PR_FALSE;
while (!foundMDC) {
if (GetNextChar(ltPos, end) == '-' &&
GetNextChar(ltPos, end) == '-') {
foundMatch = !foundMatch; // toggle until we've matching "--"
}
// And a CONTENT attribute
else if ((tokenizer.GetLength() >= kContentStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kContentStr, kContentStrLen) == 0)) {
// The next word is the value which itself needs to be parsed
if ((offset = tokenizer.GetNextWord(PR_TRUE)) != kNotFound) {
const char* contentStart = attrStart+offset;
CWordTokenizer<char> contentTokenizer(contentStart, 0,
attrEnd-contentStart);
else if (foundMatch && *ltPos == '>') {
foundMDC = PR_TRUE; // found comment end delimiter.
tagStart = (ltPos != end) ? ++ltPos : ltPos;
}
else if (ltPos == end) {
return PR_FALSE; // Couldn't find --[*s]> in this buffer
}
}
continue; // continue searching for META tag.
}
else if (!FindCharInReadable('>', tagEnd, end)) {
// Can't be sure if this is a real tag since only
// a part of the tag is in this buffer.
return PR_FALSE;
}
}
else {
return PR_FALSE; // META not found in this buffer
}
// Read the content type
if (contentTokenizer.GetNextWord() != kNotFound) {
// Now see if we have a charset
if (((offset = contentTokenizer.GetNextWord()) != kNotFound) &&
(contentTokenizer.GetLength() >= kCharsetStrLen) &&
(nsCRT::strncasecmp(contentStart+offset,
kCharsetStr, kCharsetStrLen) == 0)) {
// The next word is the charset. PR_TRUE => That we should skip quotes - Bug 88746
if ((offset = contentTokenizer.GetNextWord(PR_TRUE)) != kNotFound) {
aCharset.Assign(NS_ConvertASCIItoUCS2(contentStart+offset,
contentTokenizer.GetLength()));
}
// Find the string META and make sure it's not right at the beginning
if (CaseInsensitiveFindInReadable(NS_LITERAL_CSTRING("META"), tagStart, tagEnd)) {
const char* attrStart = tagEnd.get();
const char* attrEnd;
// Find the end of the tag
FindCharInReadable('>', tagEnd, end);
attrEnd = tagEnd.get();
CWordTokenizer<char> tokenizer(attrStart, 0, attrEnd-attrStart);
PRInt32 offset;
// Start looking at the attributes
while ((offset = tokenizer.GetNextWord()) != kNotFound) {
// We need to have a HTTP-EQUIV attribute whose value is
// "Content-Type"
if ((tokenizer.GetLength() >= kHTTPEquivStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kHTTPEquivStr, kHTTPEquivStrLen) == 0)) {
if (((offset = tokenizer.GetNextWord(PR_TRUE)) != kNotFound) &&
(tokenizer.GetLength() >= kContentTypeStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kContentTypeStr, kContentTypeStrLen) == 0)) {
foundContentType = PR_TRUE;
}
else {
break;
}
}
// And a CONTENT attribute
else if ((tokenizer.GetLength() >= kContentStrLen) &&
(nsCRT::strncasecmp(attrStart+offset,
kContentStr, kContentStrLen) == 0)) {
// The next word is the value which itself needs to be parsed
if ((offset = tokenizer.GetNextWord(PR_TRUE)) != kNotFound) {
const char* contentStart = attrStart+offset;
CWordTokenizer<char> contentTokenizer(contentStart, 0,
attrEnd-contentStart);
// Read the content type
if (contentTokenizer.GetNextWord() != kNotFound) {
// Now see if we have a charset
if (((offset = contentTokenizer.GetNextWord()) != kNotFound) &&
(contentTokenizer.GetLength() >= kCharsetStrLen) &&
(nsCRT::strncasecmp(contentStart+offset,
kCharsetStr, kCharsetStrLen) == 0)) {
// The next word is the charset. PR_TRUE => That we should skip quotes - Bug 88746
if ((offset = contentTokenizer.GetNextWord(PR_TRUE)) != kNotFound) {
aCharset.Assign(NS_ConvertASCIItoUCS2(contentStart+offset,
contentTokenizer.GetLength()));
}
}
}
}
}
if (foundContentType && (aCharset.Length() > 0)) {
return PR_TRUE;
}
}
}
tagStart = tagEnd;
tagEnd = end;
} while (tagStart != end);
if (foundContentType && (aCharset.Length() > 0)) {
return PR_TRUE;
}
}
}
return PR_FALSE;
}