void DataFeed::parse ( char *dataFeedPage,
long dataFeedPageLen ) {
// use Xml Class to parse up the page
Xml xml;
xml.set ( csUTF8, dataFeedPage, dataFeedPageLen, false, 0, false,
TITLEREC_CURRENT_VERSION );
// get the nodes
long numNodes = xml.getNumNodes();
XmlNode *nodes = xml.getNodes();
// to count the tiers, result levels, and level costs
long currTier = 0;
long currResultLevel = 0;
long currLevelCost = 0;
// pull out the keywords for the data feed
for (long i = 0; i < numNodes; i++) {
// skip if this isn't a meta tag, shouldn't happen
if (nodes[i].m_nodeId != 68)
continue;
// get the meta tag name
//long tagLen;
//char *tag = xml.getString(i, "name", &tagLen);
long ucTagLen;
char *ucTag = xml.getString(i, "name", &ucTagLen);
char tag[256];
long tagLen = utf16ToLatin1 ( tag, 256,
(UChar*)ucTag, ucTagLen>>1 );
// skip if empty
if (!tag || tagLen <= 0)
continue;
// get the content
long ucConLen;
char *ucCon = xml.getString(i, "content", &ucConLen);
char con[1024];
long conLen = utf16ToLatin1 ( con, 1024,
(UChar*)ucCon, ucConLen>>1 );
if (!con || conLen <= 0)
continue;
// match the meta tag to its local var and copy content
if (tagLen == 10 && strncasecmp(tag, "customerid", 10) == 0)
m_customerId = atoll(con);
else if (tagLen == 11 && strncasecmp(tag, "datafeedurl", 11) == 0)
setUrl(con, conLen);
else if (tagLen == 8 && strncasecmp(tag, "passcode", 8) == 0)
m_passcodeLen = setstr(m_passcode, MAX_PASSCODELEN, con, conLen);
else if (tagLen == 6 && strncasecmp(tag, "status", 6) == 0)
m_isActive = (bool)atoi(con);
else if (tagLen == 6 && strncasecmp(tag, "locked", 6) == 0)
m_isLocked = (bool)atoi(con);
else if (tagLen == 14 &&
strncasecmp(tag, "dfcreationtime", 14) == 0)
m_creationTime = atol(con);
else if (tagLen == 8 && strncasecmp(tag, "numtiers", 8) == 0)
m_priceTable.m_numTiers = atol(con);
else if (tagLen == 15 && strncasecmp(tag, "numresultlevels", 15) == 0)
m_priceTable.m_numResultLevels = atol(con);
else if (tagLen == 10 && strncasecmp(tag, "monthlyfee", 10) == 0)
m_priceTable.m_monthlyFee = atol(con);
else if (tagLen == 7 && strncasecmp(tag, "tiermax", 7) == 0) {
m_priceTable.m_tierMax[currTier] = (unsigned long)atol(con);
currTier++;
}
else if (tagLen == 11 && strncasecmp(tag, "resultlevel", 11) == 0) {
m_priceTable.m_resultLevels[currResultLevel] = (unsigned long)atol(con);
currResultLevel++;
}
else if (tagLen == 9 && strncasecmp(tag, "levelcost", 9) == 0) {
m_priceTable.m_levelCosts[currLevelCost] = (unsigned long)atol(con);
currLevelCost++;
}
else
log(LOG_INFO, "datafeed: Invalid Meta Tag Parsed [%li]:"
" %s", tagLen, tag);
}
}
// returns length of stripped content, but will set g_errno and return -1
// on error
int32_t stripHtml( char *content, int32_t contentLen, int32_t version, int32_t strip ) {
if ( !strip ) {
log( LOG_WARN, "query: html stripping not required!" );
return contentLen;
}
if ( ! content )
return 0;
if ( contentLen == 0 )
return 0;
// filter content if we should
// keep this on the big stack so "content" still references something
Xml tmpXml;
// . get the content as xhtml (should be NULL terminated)
// . parse as utf8 since all we are doing is messing with
// the tags...content manipulation comes later
if ( !tmpXml.set( content, contentLen, version, CT_HTML ) ) {
return -1;
}
//if( strip == 4 )
// return tmpXml.getText( content, contentLen );
// go tag by tag
int32_t n = tmpXml.getNumNodes();
XmlNode *nodes = tmpXml.getNodes();
// Xml class may have converted to utf16
content = tmpXml.getContent();
contentLen = tmpXml.getContentLen();
char *x = content;
char *xend = content + contentLen;
int32_t stackid = -1;
int32_t stackc = 0;
char skipIt = 0;
// . hack COL tag to NOT require a back tag
// . do not leave it that way as it could mess up our parsing
//g_nodes[25].m_hasBackTag = 0;
for ( int32_t i = 0 ; i < n ; i++ ) {
// get id of this node
int32_t id = nodes[i].m_nodeId;
// if strip is 4, just remove the script tag
if( strip == 4 ){
if ( id ){
if ( id == TAG_SCRIPT ){
skipIt ^= 1;
continue;
}
}
else if ( skipIt ) continue;
goto keepit;
}
// if strip is 3, ALL tags will be removed!
if( strip == 3 ) {
if( id ) {
// . we dont want anything in between:
// - script tags (83)
// - style tags (111)
if ((id == TAG_SCRIPT) || (id == TAG_STYLE)) skipIt ^= 1;
// save img to have alt text kept.
if ( id == TAG_IMG ) goto keepit;
continue;
}
else {
if( skipIt ) continue;
goto keepit;
}
}
// get it
int32_t fk;
if ( strip == 1 ) fk = g_nodes[id].m_filterKeep1;
else fk = g_nodes[id].m_filterKeep2;
// if tag is <link ...> only keep it if it has
// rel="stylesheet" or rel=stylesheet
if ( strip == 2 && id == TAG_LINK ) { // <link> tag id
int32_t fflen;
char *ff = nodes[i].getFieldValue ( "rel" , &fflen );
if ( ff && fflen == 10 &&
strncmp(ff,"stylesheet",10) == 0 )
goto keepit;
}
// just remove just the tag if this is 2
if ( fk == 2 ) continue;
// keep it if not in a stack
if ( ! stackc && fk ) goto keepit;
// if no front/back for tag, just skip it
if ( ! nodes[i].m_hasBackTag ) continue;
// start stack if none
if ( stackc == 0 ) {
// but not if this is a back tag
if ( nodes[i].m_node[1] == '/' ) continue;
// now start the stack
stackid = id;
stackc = 1;
continue;
}
// skip if this tag does not match what is on stack
if ( id != stackid ) continue;
// if ANOTHER front tag, inc stack
if ( nodes[i].m_node[1] != '/' ) stackc++;
// otherwise, dec the stack count
else stackc--;
// . ensure not negative from excess back tags
// . reset stackid to -1 to indicate no stack
if ( stackc <= 0 ) { stackid= -1; stackc = 0; }
// skip it
continue;
keepit:
// replace images with their alt text
int32_t vlen;
char *v;
if ( id == TAG_IMG ) {
v = nodes[i].getFieldValue("alt", &vlen );
// try title if no alt text
if ( ! v )
v = nodes[i].getFieldValue("title", &vlen );
if ( v ) { gbmemcpy ( x, v, vlen ); x += vlen; }
continue;
}
// remove background image from body,table,td tags
if ( id == TAG_BODY || id == TAG_TABLE || id == TAG_TD ) {
v = nodes[i].getFieldValue("background", &vlen);
// remove background, just sabotage it
if ( v ) v[-4] = 'x';
}
// store it
gbmemcpy ( x , nodes[i].m_node , nodes[i].m_nodeLen );
x += nodes[i].m_nodeLen;
// sanity check
if ( x > xend ) { g_process.shutdownAbort(true);}
}
contentLen = x - content;
content [ contentLen ] = '\0';
// unhack COL tag
//g_nodes[25].m_hasBackTag = 1;
return contentLen;
}
