void QueryDocument::parseQuery ( XProcessor& xproc, bool isQuery )
{
ElementRef cookiesList = queryElement.getDocument().createElement ( queryElement, xem_web.cookies() );
queryElement.appendLastChild ( cookiesList );
ElementRef headersList = queryElement.getDocument().createElement ( queryElement, xem_web.headers() );
queryElement.appendLastChild ( headersList );
QParseState state = QParseState_HeadLine;
String accu = "";
String method;
ElementRef headerElement(*this);
ElementRef responseElement(*this);
ElementRef urlElement(*this);
ElementRef urlParams(*this);
ElementRef urlParam(*this);
ElementRef cookieElement(*this);
__ui64 contentLength = 0;
bool transferEncodingChunked = false;
#define Invalid() throwException(Exception, "Invalid character %d at state %d\n", r, state );
while ( state != QParseState_FinishedHTTPHeader && !reader.isFinished())
{
int r = reader.getNextChar();
if ( r >= 0x80 )
{
throwException ( Exception, "Invalid char %d\n", r );
}
// Log_QParse ( "state=%d, r=%x (%d, %c)\n", state, r, r, r );
switch ( state )
{
case QParseState_HeadLine:
if ( r == ' ' )
{
Log_QParse ( "Method : '%s'\n", accu.c_str() );
if ( isQuery )
{
if ( accu == "GET" || accu == "POST" )
{
ElementRef methodElement = addTextualElement ( queryElement, xem_web.method(), accu );
method = methodElement.getText();
state = QParseState_URLStart;
accu = "";
continue;
}
else
Invalid();
}
else
{
if ( accu == "HTTP/1.1" )
{
Log_QParse ( "Got a Response, protocol=%s\n", accu.c_str() );
addTextualElement ( queryElement, xem_web.protocol(), accu );
accu = "";
state = QParseState_ResponseCode;
responseElement = createElement ( headersList, xem_web.response() );
headersList.appendLastChild ( responseElement );
}
else
{
Invalid();
}
}
}
else if ( r == '\r' || r == '\n' )
{
Invalid();
}
accu.appendUtf8(r);
break;
case QParseState_URLStart:
if ( r == '/' )
{
state = QParseState_URL;
continue;
}
Invalid();
break;
case QParseState_URL:
if ( r == ' ' || r == '?' )
{
Log_QParse ( "Url : '%s'\n", accu.c_str() );
urlElement = createElement ( queryElement, xem_web.url() );
queryElement.appendLastChild ( urlElement );
addTextualElement ( urlElement, xem_web.base(), accu );
accu = "";
if ( r == ' ' )
{
state = QParseState_Protocol;
}
else
{
state = QParseState_URLParam;
}
continue;
}
else if ( r == '?' )
{
Invalid();
}
else if ( r == '\r' || r == '\n' )
{
Invalid();
}
else if ( r == '%' )
{
Invalid ();
}
accu.appendUtf8(r);
break;
case QParseState_URLParam:
if ( r == '\n' || r == '%' || r == '\r' )
{
Invalid();
}
else if ( r == '=' || r == ' ' || r == '&' )
{
if ( ! urlParams )
{
urlParams = createElement ( urlElement, xem_web.parameters() );
urlElement.appendLastChild ( urlParams );
}
Log_QParse ( "Param : '%s'\n", accu.c_str() );
urlParam = createElement ( urlParams, xem_web.param() );
urlParams.appendLastChild ( urlParam );
urlParam.addAttr ( xem_web.name(), accu );
accu = "";
if ( r == '=' )
state = QParseState_URLParamValue;
else if ( r == '&' )
state = QParseState_URLParam;
else if ( r == ' ' )
state = QParseState_Protocol;
else
{ Invalid(); }
continue;
}
accu.appendUtf8(r);
break;
case QParseState_URLParamValue:
if ( r == '&' || r == ' ')
{
Log_QParse ( "Param Value : '%s'\n", accu.c_str() );
ElementRef valueNode = createTextNode ( urlParam, accu );
urlParam.appendLastChild ( valueNode );
urlParam = ElementRef(*this);
accu = "";
if ( r == '&' )
state = QParseState_URLParam;
else
state = QParseState_Protocol;
continue;
}
accu.appendUtf8(r);
break;
case QParseState_Protocol:
if ( r == ' ' )
{
Invalid();
}
else if ( r == '\r' )
{
continue;
}
else if ( r == '\n' )
{
Log_QParse ( "Protocol = '%s'\n", accu.c_str() );
addTextualElement ( queryElement, xem_web.protocol(), accu );
accu = "";
state = QParseState_BeginLine;
continue;
}
accu.appendUtf8(r);
break;
case QParseState_ResponseCode:
if ( r == ' ' )
{
Log_QParse ( "ResponseCode : %s\n", accu.c_str() );
responseElement.addAttr ( xem_web.response_code(), accu );
accu = "";
state = QParseState_ResponseString;
continue;
}
if ( '0' <= r && r <= '9' )
{
accu.appendUtf8(r);
continue;
}
Invalid();
break;
case QParseState_ResponseString:
if ( r == '\r' )
continue;
else if ( r == '\n' )
{
Log_QParse ( "ResponseString : %s\n", accu.c_str() );
responseElement.addAttr ( xem_web.response_string(), accu );
accu = "";
state = QParseState_BeginLine;
continue;
}
accu.appendUtf8(r);
break;
case QParseState_BeginLine:
if ( r == '\r' )
continue;
else if ( r == '\n' )
{
state = QParseState_FinishedHTTPHeader;
break;
}
else if ( r == ' ' )
{
Invalid();
}
accu.appendUtf8(r);
state = QParseState_FieldName;
continue;
case QParseState_FieldName:
if ( r == ':' )
{
Log_QParse ( "Field : '%s'\n", accu.c_str() );
if ( accu == "Cookie" )
{
accu = "";
state = QParseState_CookieStart;
continue;
}
headerElement = createElement ( headersList, xem_web.param() );
headersList.appendLastChild ( headerElement );
headerElement.addAttr ( xem_web.name(), accu );
accu = "";
state = QParseState_PostFieldName;
continue;
}
else if ( r == ' ' || r == '\n' || r == '\r' )
{
Invalid();
}
accu.appendUtf8(r);
break;
case QParseState_PostFieldName:
if ( r == ' ' )
{
state = QParseState_Value;
continue;
}
Invalid();
case QParseState_Value:
if ( r == '\r')
continue;
else if ( r == '\n' )
{
Log_QParse ( "Field value : '%s'\n", accu.c_str() );
ElementRef valueNode = createTextNode ( headerElement, accu );
headerElement.appendLastChild ( valueNode );
String fieldName = headerElement.getAttr(xem_web.name());
#ifdef __XEM_WEBSERVER_QUERYDOCUMENT_HAS_HEADERFIELDSMAP
headerFieldsMap[fieldName] = valueNode.getText();
#endif
if ( fieldName == "Content-Length" )
{
contentLength = accu.toUI64();
Log_QParse ( "ContentLength : %llu\n", contentLength );
}
else if ( fieldName == "Transfer-Encoding" && accu == "chunked" )
{
Log_QParse ( "TransferEncoding chuncked !\n" );
transferEncodingChunked = true;
}
headerElement = ElementRef(*this);
accu = "";
state = QParseState_BeginLine;
continue;
}
accu.appendUtf8(r);
break;
case QParseState_CookieStart:
if ( r == ' ' )
continue;
accu.appendUtf8(r);
state = QParseState_CookieName;
break;
case QParseState_CookieName:
if ( r == '=' )
{
Log_QParse ( "Cookie name : '%s'\n", accu.c_str() );
cookieElement = createElement ( cookiesList, xem_web.cookie() );
cookieElement.addAttr ( xem_web.name(), accu );
cookiesList.appendLastChild ( cookieElement );
accu = "";
state = QParseState_CookieValue;
continue;
}
accu.appendUtf8(r);
break;
case QParseState_CookieValue:
if ( r == '\r' )
continue;
if ( r == ';' || r == '\n' )
{
Log_QParse ( "Cookie value : '%s'\n", accu.c_str() );
ElementRef cookieValueNode = createTextNode ( cookieElement, accu );
cookieElement.appendLastChild ( cookieValueNode );
accu = "";
if ( r == ';' )
state = QParseState_CookieStart;
else
state = QParseState_BeginLine;
continue;
}
accu.appendUtf8(r);
break;
default:
Bug ( "Case %d Not implemented !\n", state );
}
}
if ( state != QParseState_FinishedHTTPHeader )
{
throwException ( Exception, "HTTP reader not finished ! state=%d\n", state );
}
if ( contentLength || transferEncodingChunked )
{
if ( !isQuery || method == "POST" )
{
ElementRef content = createElement ( queryElement, xem_web.content() );
queryElement.appendLastChild ( content );
BlobRef blob = content.addBlob(xem_web.blob_contents());
if ( contentLength )
parseToBlob(blob, contentLength);
else
parseChunkedToBlob(blob);
}
else
{
throwException ( Exception, "Invalid content-length with method = %s\n", method.c_str() );
}
}
}
void Url::parseString(const char* urlString, UtlBoolean isAddrSpec)
{
// If isAddrSpec:
// userinfo@hostport;uriParameters?headerParameters
// If !isAddrSpec:
// DisplayName<userinfo@hostport;urlParameters?headerParameters>;fieldParameters
# ifdef TIME_PARSE
OsTimeLog timeLog;
LOG_TIME("start ");
# endif
// Try to catch when a name-addr is passed but we are expecting an
// addr-spec -- many name-addr's start with '<' or '"'.
if (isAddrSpec && (urlString[0] == '<' || urlString[0] == '"'))
{
OsSysLog::add(FAC_SIP, PRI_ERR,
"Url::parseString Invalid addr-spec found (probably name-addr format): '%s'",
urlString);
}
int workingOffset = 0; // begin at the beginning...
size_t afterAngleBrackets = UTL_NOT_FOUND;
if (isAddrSpec)
{
mAngleBracketsIncluded = FALSE;
}
else // ! addr-spec
{
// Is there a display name on the front?
mDisplayName.remove(0);
LOG_TIME("display <");
RegEx displayName(DisplayName);
if (displayName.SearchAt(urlString, workingOffset))
{
LOG_TIME("display > ");
switch (displayName.Matches() /* number of substrings that matched */)
{
case 2: // matched unquoted sequence of tokens
displayName.MatchString(&mDisplayName, 1);
break;
case 3: // matched a double quoted string
// see performance note on DisplayName
mDisplayName.append("\"");
displayName.MatchString(&mDisplayName, 2);
mDisplayName.append("\"");
break;
default:
assert(false);
}
// does not include whitespace or the '<'
workingOffset = displayName.AfterMatch(0);
}
// Are there angle brackets around the URI?
LOG_TIME("angles < ");
RegEx angleBrackets(AngleBrackets);
if (angleBrackets.SearchAt(urlString, workingOffset))
{
LOG_TIME("angles > ");
// yes, there are angle brackets
workingOffset = angleBrackets.MatchStart(1); // inside the angle brackets
afterAngleBrackets = angleBrackets.AfterMatch(0); // following the '>'
/*
* Note: We do not set mAngleBracketsIncluded just because we saw them
* That is only used for explicit control from the outside.
* The local knowledge of whether or not there are angle brackets
* is whether or not afterAngleBrackets == UTL_NOT_FOUND
*/
}
}
/*
* AMBIGUITY - there is a potential ambiguity when parsing real URLs.
*
* Consider the url 'foo:333' - it could be:
* scheme 'foo' host '333' ('333' is a valid local host name - bad idea, but legal)
* or host 'foo' port '333' (and scheme 'sip' is implied)
*
* Now make it worse by using 'sips' as a hostname:
* 'sips:333'
* scheme 'sips' host '333'
* or host 'sips' port '333' (and scheme 'sip' is implied)
*
* We resolve the first case by treating anything left of the colon as a scheme if
* it is one of the supported schemes. Otherwise, we set the scheme to the
* default (sip) and go on so that it will be parsed as a hostname. This does not
* do the right thing for the (scheme 'sips' host '333') case, but they get what
* they deserve.
*/
// Parse the scheme (aka url type)
LOG_TIME("scheme < ");
RegEx supportedScheme(SupportedScheme);
if ( (supportedScheme.SearchAt(urlString,workingOffset))
&& (supportedScheme.MatchStart(0) == workingOffset)
)
{
LOG_TIME("scheme > ");
// the scheme name matches one of the supported schemes
mScheme = static_cast<Scheme>(supportedScheme.Matches()-1);
workingOffset = supportedScheme.AfterMatch(0); // past the ':'
}
else
{
/*
* It did not match one of the supported scheme names
* so proceed on the assumption that it's a host and "sip:" is implied
* Leave the workingOffset where it is (before the token).
* The code below, through the parsing of host and port
* treats this as an implicit 'sip:' url; if it parses ok
* up to that point, it resets the scheme to SipsUrlScheme
*/
mScheme = UnknownUrlScheme;
}
// skip over any '//' following the scheme for the ones we know use that
switch (mScheme)
{
case FileUrlScheme:
case FtpUrlScheme:
case HttpUrlScheme:
case HttpsUrlScheme:
case RtspUrlScheme:
if (0==strncmp("//", urlString+workingOffset, 2))
{
workingOffset += 2;
}
break;
case UnknownUrlScheme:
case SipUrlScheme:
case SipsUrlScheme:
case MailtoUrlScheme:
default:
break;
}
if (FileUrlScheme != mScheme) // no user part in file urls
{
// Parse the username and password
LOG_TIME("userpass < ");
RegEx usernameAndPassword(UsernameAndPassword);
if ( (usernameAndPassword.SearchAt(urlString, workingOffset))
&& usernameAndPassword.MatchStart(0) == workingOffset
)
{
LOG_TIME("userpass > ");
usernameAndPassword.MatchString(&mUserId, 1);
usernameAndPassword.MatchString(&mPassword, 2);
workingOffset = usernameAndPassword.AfterMatch(0);
}
else
{
// username and password are optional, so not finding them is ok
// leave workingOffset where it is
}
}
// Parse the hostname and port
LOG_TIME("hostport < ");
RegEx hostAndPort(HostAndPort);
if ( (hostAndPort.SearchAt(urlString,workingOffset))
&& (hostAndPort.MatchStart(0) == workingOffset)
)
{
LOG_TIME("hostport > ");
hostAndPort.MatchString(&mHostAddress,1);
UtlString portStr;
if (hostAndPort.MatchString(&portStr,2))
{
mHostPort = atoi(portStr.data());
}
workingOffset = hostAndPort.AfterMatch(0);
if (UnknownUrlScheme == mScheme)
{
/*
* Resolve AMBIGUITY
* Since we were able to parse this as a host and port, it is now safe to
* set the scheme to the implied 'sip:'.
*/
mScheme = SipUrlScheme;
}
}
else
{
if (FileUrlScheme != mScheme) // no host is ok in a file URL
{
/*
* This is not a file URL, so not having a recognized host name is invalid.
*
* Since we may have been called from a constructor, there is no way to
* return an error, but at this point we know this is bad, so instead
* we just log an error and set the scheme to the unknown url type and
* clear any components that might have been set.
*/
OsSysLog::add(FAC_SIP, PRI_ERR,
"Url::parseString no valid host found at char %d in '%s', "
"isAddrSpec = %d",
workingOffset, urlString, isAddrSpec
);
mScheme = UnknownUrlScheme;
mDisplayName.remove(0);
mUserId.remove(0);
mPassword.remove(0);
}
}
// Next is a path if http, https, or ftp,
// OR url parameters if sip or sips.
// There can be no Url parameters for http, https, or ftp
// because semicolon is a valid part of the path value
switch ( mScheme )
{
case FileUrlScheme:
case FtpUrlScheme:
case HttpUrlScheme:
case HttpsUrlScheme:
case RtspUrlScheme:
{
// this is an http, https, or ftp URL, so get the path
LOG_TIME("path < ");
RegEx urlPath(UrlPath);
if ( (urlPath.SearchAt(urlString, workingOffset))
&& (urlPath.MatchStart(0) == workingOffset)
)
{
LOG_TIME("path > ");
urlPath.MatchString(&mPath,1);
workingOffset = urlPath.AfterMatch(1);
}
}
break;
case SipUrlScheme:
case SipsUrlScheme:
{
// it may have url parameters of the form ";" param "=" value ...
// if it meets the right conditions:
if ( isAddrSpec // in addr-spec, any param is a url param
|| afterAngleBrackets != UTL_NOT_FOUND // inside angle brackets there may be a url param
)
{
LOG_TIME("urlparm < ");
RegEx urlParams(UrlParams);
if ( (urlParams.SearchAt(urlString, workingOffset))
&& (urlParams.MatchStart(0) == workingOffset)
)
{
LOG_TIME("urlparm > ");
urlParams.MatchString(&mRawUrlParameters, 1);
workingOffset = urlParams.AfterMatch(1);
// actual parsing of the parameters is in parseUrlParameters
// so that it only happens if someone asks for them.
}
}
}
break;
case MailtoUrlScheme:
default:
// no path component
break;
}
if (UnknownUrlScheme != mScheme)
{
// Parse any header or query parameters
LOG_TIME("hdrparm < ");
RegEx headerOrQueryParams(HeaderOrQueryParams);
if( (headerOrQueryParams.SearchAt(urlString, workingOffset))
&& (headerOrQueryParams.MatchStart(0) == workingOffset)
)
{
LOG_TIME("hdrparm > ");
headerOrQueryParams.MatchString(&mRawHeaderOrQueryParameters, 1);
workingOffset = headerOrQueryParams.AfterMatch(0);
// actual parsing of the parameters is in parseHeaderOrQueryParameters
// so that it only happens if someone asks for them.
}
// Parse the field parameters
if (!isAddrSpec) // can't have field parameters in an addrspec
{
if (afterAngleBrackets != UTL_NOT_FOUND)
{
workingOffset = afterAngleBrackets;
}
LOG_TIME("fldparm < ");
RegEx fieldParameters(FieldParams);
if ( (fieldParameters.SearchAt(urlString, workingOffset))
&& (fieldParameters.MatchStart(0) == workingOffset)
)
{
LOG_TIME("fldparm > ");
fieldParameters.MatchString(&mRawFieldParameters, 1);
// actual parsing of the parameters is in parseFieldParameters
// so that it only happens if someone asks for them.
}
}
}
# ifdef TIME_PARSE
UtlString timeDump;
timeLog.getLogString(timeDump);
printf("\n%s\n", timeDump.data());
# endif
}
