// A cookie header looks like: Cookie: username=xaxxon; password=comoesta
// everything after the : gets passed in in irsData
int HttpRequest::ParseCookieData ( std::string & irsData )
{
#ifdef EHS_DEBUG
fprintf ( stderr, "looking for cookies in '%s'\n", irsData.c_str ( ) );
#endif
PME oCookieRegex ( "\\s*([^=]+)=([^;]+)(;|$)*", "g" );
int nNameValuePairsFound = 0;
while ( oCookieRegex.match ( irsData ) ) {
/* fprintf ( stderr, "mapping cookie data: %s => %s\n",
oCookieRegex [ 1 ].c_str ( ),
oCookieRegex [ 2 ].c_str ( ) );
*/
oCookieMap [ oCookieRegex [ 1 ] ] = oCookieRegex [ 2 ];
nNameValuePairsFound++;
}
return nNameValuePairsFound;
}
std::string PME::UpdateReplacementString ( const std::string & r ) {
std::string finalreplacement = r;
// search for each backref and store it out
PME dollars ( "\\$([0-9]+)", "g" );
// search each backref
while ( int numdollars = dollars.match ( r ) ) {
// create a regex to replace the backref
stringstream regextext;
regextext << "[$]" << dollars[1];
PME dollarsub ( regextext.str ( ) );
// do the replacement, telling it not to look for backref
finalreplacement = dollarsub.sub ( finalreplacement,
(*this)[atoi(dollars[1].c_str())],
0 );
}
return finalreplacement;
}
std::string GetNextPathPart ( std::string & irsUri ///< URI to look for next path part in
)
{
PME oNextPathPartRegex ( "^[/]{0,1}([^/]+)/(.*)$" );
if ( oNextPathPartRegex.match ( irsUri ) ) {
std::string sReturnValue = oNextPathPartRegex [ 1 ];
std::string sNewUri = oNextPathPartRegex [ 2 ];
irsUri = sNewUri;
return sReturnValue;
} else {
return "";
}
}
void HttpRequest::GetFormDataFromString ( const std::string & irsString ///< string to parse for form data
)
{
PME oNameValueRegex ( "[?]?([^?=]*)=([^&]*)&?", "g" );
while ( oNameValueRegex.match ( irsString ) ) {
ContentDisposition oContentDisposition;
std::string sName = oNameValueRegex [ 1 ];
std::string sValue = oNameValueRegex [ 2 ];
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Got form data: '%s' => '%s'\n",
sName.c_str ( ),
sValue.c_str ( ) );
#endif
oFormValueMap [ sName ] =
FormValue ( sValue, oContentDisposition );
}
}
// returns 1 if it got a line, 0 otherwise
int GetNextLine ( std::string & irsLine, ///< line removed from *ippsBuffer
std::string & irsBuffer ///< buffer from which to remove a line
)
{
int nResult = 0;
// can't use split, because we lose our \r\n
PME oLineRegex ( "^([^\\r]*\\r\\n)(.*)$", "sm" );
if ( oLineRegex.match ( irsBuffer ) == 3 ) {
irsLine = oLineRegex [ 1 ];
irsBuffer = oLineRegex [ 2 ];
} else {
irsLine = "";
}
return nResult;
}
int main ( )
{
PME regex ( ", " );
int i = regex.split ( "very, very2, tired" );
for ( int j = 0; j < i; j++ ) {
printf ( "%s\n", regex [ j ].c_str ( ) );
}
StringVector sv = regex.GetStringVector ( );
for ( StringVector::iterator i = sv.begin ( );
i != sv.end ( );
i++ ) {
printf ( "%s\n", i->c_str ( ) );
}
PME match ( "([a-z])([0-9])([A-Z])" );
match.match ( "b2Z" );
sv = match.GetStringVector ( );
for ( StringVector::iterator i = sv.begin ( );
i != sv.end ( );
i++ ) {
printf ( "%s\n", i->c_str ( ) );
}
}
HttpRequest::ParseSubbodyResult HttpRequest::ParseSubbody ( std::string isSubbody ///< string in which to look for subbody stuff
)
{
// find the spot after the headers in the body
std::string::size_type nBlankLinePosition = isSubbody.find ( "\r\n\r\n" );
// if there's no double blank line, then this isn't a valid subbody
if ( nBlankLinePosition == std::string::npos ) {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEUBG] Invalix subbody, couldn't find double blank line\n" );
#endif
return PARSESUBBODY_INVALIDSUBBODY;
}
// create a string from the beginning to the blank line -- OFF BY ONE?
std::string sHeaders ( isSubbody, 0, nBlankLinePosition - 1 );
// first line MUST be the content-disposition header line, so that
// we know what the name of the field is.. otherwise, we're in trouble
int nMatchResult = 0;
PME oContentDispositionRegex ( "Content-Disposition:[ ]?([^;]+);[ ]?(.*)" );
nMatchResult = oContentDispositionRegex.match ( sHeaders );
if ( nMatchResult == 3 ) {
std::string sContentDisposition ( oContentDispositionRegex [ 1 ] );
std::string sNameValuePairs ( oContentDispositionRegex [ 2 ] );
/* fprintf ( stderr, "cont. disp. '%s', name/value pairs '%s'\n",
sContentDisposition.c_str ( ),
sNameValuePairs.c_str ( ) );
*/
StringMap oStringMap;
PME oContentDispositionNameValueRegex ( "[ ]?([^= ]+)=\"([^\"]+)\"[;]?", "g" );
// go through the sNameValuePairs string and grab out the pieces
nMatchResult = 3;
while ( nMatchResult == 3 ) {
nMatchResult = oContentDispositionNameValueRegex.match ( sNameValuePairs );
if ( nMatchResult == 3 ) {
std::string sName = oContentDispositionNameValueRegex [ 1 ];
std::string sValue = oContentDispositionNameValueRegex [ 2 ];
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Subbody header found: '%s' => '%s' with %d matches\n",
sName.c_str ( ),
sValue.c_str ( ),
nMatchResult );
#endif
oStringMap [ sName ] = sValue;
} else if ( nMatchResult == 0 ) {
; // this is okay -- just done with name value pairs
} else {
//fprintf ( stderr, "Unexpected number of matches, '%d' != 3\n", nMatchResult );
assert ( 0 );
}
//fprintf ( stderr, "nMatchResult = %d\n", nMatchResult );
}
// fprintf ( stderr, "done looking for headers\n" );
// take oStringMap and actually fill the right object with its data
FormValue & roFormValue = oFormValueMap [ oStringMap [ "name" ] ];
// copy the headers in now that we know the name
roFormValue.oContentDisposition.oContentDispositionHeaders = oStringMap;
// grab out the body
roFormValue.sBody = isSubbody.substr ( nBlankLinePosition + 4);
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Subbody body (in binary):\n---\n" );
fwrite ( roFormValue.sBody.c_str ( ), 1, roFormValue.sBody.length ( ), stderr );
fprintf ( stderr, "\n---\n" );
#endif
}
// couldn't find content-disposition line -- FATAL ERROR
else {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Error: Couldn't find content-disposition line\n" );
#endif
return PARSESUBBODY_INVALIDSUBBODY;
}
return PARSESUBBODY_SUCCESS;
}
// takes data and tries to figure out what it is. it will loop through
// irsData as many times as it can as long as it gets a full line each time.
// Depending on how much data it's gotten already, it will handle a line
// differently..
HttpRequest::HttpParseStates HttpRequest::ParseData ( std::string & irsData ///< buffer to look in for more data
)
{
std::string sLine;
int nDoneWithCurrentData = 0;
PME oHeaderRegex ( "^([^:]*):\\s+(.*)\\r\\n$" );
while ( ! nDoneWithCurrentData &&
nCurrentHttpParseState != HTTPPARSESTATE_INVALIDREQUEST &&
nCurrentHttpParseState != HTTPPARSESTATE_COMPLETEREQUEST &&
nCurrentHttpParseState != HTTPPARSESTATE_INVALID ) {
switch ( nCurrentHttpParseState ) {
case HTTPPARSESTATE_REQUEST:
// get the request line
GetNextLine ( sLine, irsData );
// if we got a line, parse out the data..
if ( sLine.length ( ) == 0 ) {
nDoneWithCurrentData = 1;
}
// if we got a line, look for a request line
else {
// everything must be uppercase according to rfc2616
PME oRequestLineRegex ( "^(OPTIONS|GET|HEAD|POST|PUT|DELETE|TRACE|CONNECT) ([^ ]*) HTTP/([^ ]+)\\r\\n$" );
if ( oRequestLineRegex.match ( sLine ) ) {
// get the info from the request line
nRequestMethod = GetRequestMethodFromString ( oRequestLineRegex [ 1 ] );
sUri = oRequestLineRegex [ 2 ];
sOriginalUri = oRequestLineRegex [ 2 ];
sHttpVersionNumber = oRequestLineRegex [ 3 ];
// check to see if the uri appeared to have form data in it
GetFormDataFromString ( sUri );
// on to the headers
nCurrentHttpParseState = HTTPPARSESTATE_HEADERS;
}
// if the regex failed
else {
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
} // end match on request line
} // end whether we got a line
break;
case HTTPPARSESTATE_HEADERS:
// get the next line
GetNextLine ( sLine, irsData );
// if we didn't get a full line of data
if ( sLine.length ( ) == 0 ) {
nDoneWithCurrentData = 1;
}
// check to see if we're done with headers
else if ( sLine == "\r\n" ) {
// if content length is found
if ( oRequestHeaders.find ( "content-length" ) !=
oRequestHeaders.end ( ) ) {
nCurrentHttpParseState = HTTPPARSESTATE_BODY;
} else {
nCurrentHttpParseState = HTTPPARSESTATE_COMPLETEREQUEST;
}
// if this is an HTTP/1.1 request, then it had better have a Host: header
if ( sHttpVersionNumber == "1.1" &&
oRequestHeaders [ "host" ].length ( ) == 0 ) {
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
}
}
// else if there is still data
else if ( oHeaderRegex.match ( sLine ) ) {
std::string sName = oHeaderRegex [ 1 ];
std::string sValue = oHeaderRegex [ 2 ];
if ( sName == "Transfer-Encoding" &&
sValue == "chunked" ) {
fprintf ( stderr, "EHS DOES NOT SUPPORT CHUNKED ENCODING. Send an email to [email protected] and tell him you want chunked encoding (or send a patch)\n" );
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
}
sName = mytolower ( sName );
if ( sName == "cookie" ) {
ParseCookieData ( sValue );
}
oRequestHeaders [ sName ] = sValue;
}
// else we had some sort of error -- bail out
else {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Error: Invalid header line: '%s'\n",
sLine.c_str ( ) );
#endif
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
nDoneWithCurrentData = 1;
}
break;
case HTTPPARSESTATE_BODY:
{
// if a content length wasn't specified, we can't be here (we
// don't know chunked encoding)
if ( oRequestHeaders.find ( "content-length" ) ==
oRequestHeaders.end ( ) ) {
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
continue;
}
// get the content length
unsigned int nContentLength =
atoi ( oRequestHeaders [ "content-length" ].c_str ( ) );
// else if we haven't gotten all the data we're looking for,
// just hold off and try again when we get more
if ( irsData.length ( ) < nContentLength ) {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Not enough data yet -- %d < %d\n",
irsData.length ( ), nContentLength );
#endif
nDoneWithCurrentData = 1;
}
// otherwise, we've gotten enough data from the client, handle it now
else {
// grab out the actual body from the request and leave the rest
sBody = irsData.substr ( 0, nContentLength );
irsData = irsData.substr ( nContentLength );
// if we're dealing with multi-part form attachments
if ( oRequestHeaders [ "content-type" ].substr ( 0, 9 ) ==
"multipart" ) {
// handle the body as if it's multipart form data
if ( ParseMultipartFormData ( ) ==
PARSEMULTIPARTFORMDATA_SUCCESS ) {
nCurrentHttpParseState = HTTPPARSESTATE_COMPLETEREQUEST;
} else {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Error: Mishandled multi-part form data for unknown reason\n" );
#endif
nCurrentHttpParseState = HTTPPARSESTATE_INVALIDREQUEST;
}
}
// else the body is just one piece
else {
// check for any form data
GetFormDataFromString ( sBody );
#ifdef EHS_DEBUG
fprintf ( stderr, "Done with body, done with entire request\n" );
#endif
nCurrentHttpParseState = HTTPPARSESTATE_COMPLETEREQUEST;
}
}
nDoneWithCurrentData = 1;
}
break;
case HTTPPARSESTATE_INVALID:
default:
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Critical error: Invalid internal state: %d. Aborting\n", nCurrentHttpParseState );
#endif
assert ( 0 );
break;
}
}
return nCurrentHttpParseState;
}
HttpRequest::ParseMultipartFormDataResult
HttpRequest::ParseMultipartFormData ( )
{
assert ( !oRequestHeaders [ "content-type" ].empty ( ) );
// find the boundary string
int nMatchResult = 0;
PME oMultipartFormDataContentTypeValueRegex ( "multipart/[^;]+;[ ]*boundary=([^\"]+)$" );
#ifdef EHS_DEBUG
fprintf ( stderr, "looking for boundary in '%s'\n",oRequestHeaders [ "content-type" ].c_str ( ) );
#endif
if ( ( nMatchResult = oMultipartFormDataContentTypeValueRegex.match ( oRequestHeaders [ "content-type" ] ) ) ) {
assert ( nMatchResult == 2 );
std::string sBoundaryString = oMultipartFormDataContentTypeValueRegex [ 1 ];
// the actual boundary has two dashes prepended to it
std::string sActualBoundary = std::string ("--") + sBoundaryString;
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Found boundary of '%s'\n", sBoundaryString.c_str ( ) );
fprintf ( stderr, "[EHS_DEBUG] Info: Looking for boundary to match (%d) '%s'\n", sBody.length ( ),
sBody.substr ( 0, sActualBoundary.length ( ) ).c_str ( ) );
#endif
// check to make sure we started at the boundary
if ( sBody.substr ( 0, sActualBoundary.length ( ) ) != sActualBoundary ) {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Error: Misparsed multi-part form data for unknown reason - first bytes weren't the boundary string\n" );
#endif
return PARSEMULTIPARTFORMDATA_FAILED;
}
// go past the initial boundary
std::string sRemainingBody = sBody.substr ( sActualBoundary.length ( ) );
// while we're at a boundary after we grab a part, keep going
std::string::size_type nNextPartPosition;
while ( ( nNextPartPosition = sRemainingBody.find ( std::string ( "\r\n" ) + sActualBoundary ) ) !=
std::string::npos ) {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Info: Found subbody from pos %d to %d\n",
0, nNextPartPosition );
#endif
assert ( (unsigned int) sRemainingBody.length ( ) >= sActualBoundary.length ( ) );
ParseSubbody ( sRemainingBody.substr ( 0, nNextPartPosition ) );
// skip past the boundary at the end and look for the next one
nNextPartPosition += sActualBoundary.length ( );
sRemainingBody = sRemainingBody.substr ( nNextPartPosition );
}
} else {
#ifdef EHS_DEBUG
fprintf ( stderr, "[EHS_DEBUG] Error: Couldn't find boundary specification in content-type header\n" );
#endif
return PARSEMULTIPARTFORMDATA_FAILED;
}
return PARSEMULTIPARTFORMDATA_SUCCESS;
}