bool JsonItem::getCompoundName ( SafeBuf &nameBuf ) {
// reset, but don't free mem etc. just set m_length to 0
nameBuf.reset();
// get its full compound name like "meta.twitter.title"
JsonItem *p = this;//ji;
char *lastName = NULL;
char *nameArray[20];
int32_t numNames = 0;
for ( ; p ; p = p->m_parent ) {
// empty name?
if ( ! p->m_name ) continue;
if ( ! p->m_name[0] ) continue;
// dup? can happen with arrays. parent of string
// in object, has same name as his parent, the
// name of the array. "dupname":[{"a":"b"},{"c":"d"}]
if ( p->m_name == lastName ) continue;
// update
lastName = p->m_name;
// add it up
nameArray[numNames++] = p->m_name;
// breach?
if ( numNames < 15 ) continue;
log("build: too many names in json tag");
break;
}
// assemble the names in reverse order which is correct order
for ( int32_t i = 1 ; i <= numNames ; i++ ) {
// copy into our safebuf
if ( ! nameBuf.safeStrcpy ( nameArray[numNames-i]) )
return false;
// separate names with periods
if ( ! nameBuf.pushChar('.') ) return false;
}
// remove last period
nameBuf.removeLastChar('.');
// and null terminate
if ( ! nameBuf.nullTerm() ) return false;
// change all :'s in names to .'s since : is reserved!
char *px = nameBuf.getBufStart();
for ( ; *px ; px++ ) if ( *px == ':' ) *px = '.';
return true;
}
// returns false if blocked, true otherwise
bool processLoop ( void *state ) {
// get it
State2 *st = (State2 *)state;
// get the tcp socket from the state
TcpSocket *s = st->m_socket;
// get it
XmlDoc *xd = &st->m_xd;
if ( ! xd->m_loaded ) {
// setting just the docid. niceness is 0.
//xd->set3 ( st->m_docId , st->m_coll , 0 );
// callback
xd->setCallback ( state , processLoop );
// . and tell it to load from the old title rec
// . this sets xd->m_oldTitleRec/m_oldTitleRecSize
// . this sets xd->ptr_* and all other member vars from
// the old title rec if found in titledb.
if ( ! xd->loadFromOldTitleRec ( ) ) return false;
}
if ( g_errno ) return sendErrorReply ( st , g_errno );
// now force it to load old title rec
//char **tr = xd->getTitleRec();
SafeBuf *tr = xd->getTitleRecBuf();
// blocked? return false if so. it will call processLoop() when it rets
if ( tr == (void *)-1 ) return false;
// we did not block. check for error? this will free "st" too.
if ( ! tr ) return sendErrorReply ( st , g_errno );
// if title rec was empty, that is a problem
if ( xd->m_titleRecBuf.length() == 0 )
return sendErrorReply ( st , ENOTFOUND);
// set callback
char *na = xd->getIsNoArchive();
// wait if blocked
if ( na == (void *)-1 ) return false;
// error?
if ( ! na ) return sendErrorReply ( st , g_errno );
// forbidden? allow turkeys through though...
if ( ! st->m_isMasterAdmin && *na )
return sendErrorReply ( st , ENOCACHE );
SafeBuf *sb = &st->m_sb;
// &page=4 will print rainbow sections
if ( ! st->m_printed && st->m_r.getLong("page",0) ) {
// do not repeat this call
st->m_printed = true;
// this will call us again since we called
// xd->setCallback() above to us
if ( ! xd->printDocForProCog ( sb , &st->m_r ) )
return false;
}
char *contentType = "text/html";
char format = st->m_format;
if ( format == FORMAT_XML ) contentType = "text/xml";
if ( format == FORMAT_JSON ) contentType = "application/json";
// if we printed a special page (like rainbow sections) then return now
if ( st->m_printed ) {
bool status = g_httpServer.sendDynamicPage (s,
//buf,bufLen,
sb->getBufStart(),
sb->getLength(),
-1,false,
//"text/html",
contentType,
-1, NULL, "utf8" );
// nuke state2
mdelete ( st , sizeof(State2) , "PageGet1" );
delete (st);
return status;
}
/*
// this was calling XmlDoc and setting sections, etc. to
// get the SpiderReply junk... no no no
// is it banned or filtered? this ignores the TagRec in the titleRec
// and uses msg8a to get it fresh instead
char *vi = xd->getIsFiltered();//Visible( );
// wait if blocked
if ( vi == (void *)-1 ) return false;
// error?
if ( ! vi ) return sendErrorReply ( st , g_errno );
// banned?
if ( ! st->m_isMasterAdmin && ! *vi ) return sendErrorReply (st,EDOCBANNED);
*/
// get the utf8 content
char **utf8 = xd->getUtf8Content();
//int32_t len = xd->size_utf8Content - 1;
// wait if blocked???
if ( utf8 == (void *)-1 ) return false;
// strange
if ( xd->size_utf8Content<=0) {
log("pageget: utf8 content <= 0");
return sendErrorReply(st,EBADENGINEER );
}
// alloc error?
if ( ! utf8 ) return sendErrorReply ( st , g_errno );
// get this host
Host *h = g_hostdb.getHost ( g_hostdb.m_hostId );
if ( ! h ) {
log("pageget: hostid %"INT32" is bad",g_hostdb.m_hostId);
return sendErrorReply(st,EBADENGINEER );
}
char *content = xd->ptr_utf8Content;
int32_t contentLen = xd->size_utf8Content - 1;
// int16_tcut
char strip = st->m_strip;
// alloc buffer now
//char *buf = NULL;
//int32_t bufMaxSize = 0;
//bufMaxSize = len + ( 32 * 1024 ) ;
//bufMaxSize = contentLen + ( 32 * 1024 ) ;
//buf = (char *)mmalloc ( bufMaxSize , "PageGet2" );
//char *p = buf;
//char *bufEnd = buf + bufMaxSize;
//if ( ! buf ) {
// return sendErrorReply ( st , g_errno );
//}
// for undoing the header
//char *start1 = p;
int32_t startLen1 = sb->length();
// we are always utfu
if ( strip != 2 )
sb->safePrintf( "<meta http-equiv=\"Content-Type\" "
"content=\"text/html;charset=utf8\">\n");
// base href
//Url *base = &xd->m_firstUrl;
//if ( xd->ptr_redirUrl.m_url[0] )
// base = &xd->m_redirUrl;
char *base = xd->ptr_firstUrl;
if ( xd->ptr_redirUrl ) base = xd->ptr_redirUrl;
//Url *redir = *xd->getRedirUrl();
if ( strip != 2 ) {
sb->safePrintf ( "<BASE HREF=\"%s\">" , base );
//p += gbstrlen ( p );
}
// default colors in case css files missing
if ( strip != 2 ) {
sb->safePrintf( "\n<style type=\"text/css\">\n"
"body{background-color:white;color:black;}\n"
"</style>\n");
//p += gbstrlen ( p );
}
//char format = st->m_format;
if ( format == FORMAT_XML ) sb->reset();
if ( format == FORMAT_JSON ) sb->reset();
if ( xd->m_contentType == CT_JSON ) sb->reset();
if ( xd->m_contentType == CT_XML ) sb->reset();
if ( xd->m_contentType == CT_STATUS ) sb->reset();
// for undoing the stuff below
int32_t startLen2 = sb->length();//p;
// query should be NULL terminated
char *q = st->m_qsb.getBufStart();
int32_t qlen = st->m_qsb.getLength(); // m_qlen;
char styleTitle[128] = "font-size:14px;font-weight:600;"
"color:#000000;";
char styleText[128] = "font-size:14px;font-weight:400;"
"color:#000000;";
char styleLink[128] = "font-size:14px;font-weight:400;"
"color:#0000ff;";
char styleTell[128] = "font-size:14px;font-weight:600;"
"color:#cc0000;";
// get the url of the title rec
Url *f = xd->getFirstUrl();
bool printDisclaimer = st->m_printDisclaimer;
if ( xd->m_contentType == CT_JSON )
printDisclaimer = false;
if ( xd->m_contentType == CT_STATUS )
printDisclaimer = false;
if ( format == FORMAT_XML ) printDisclaimer = false;
if ( format == FORMAT_JSON ) printDisclaimer = false;
char tbuf[100];
tbuf[0] = 0;
time_t lastSpiderDate = xd->m_spideredTime;
if ( printDisclaimer ||
format == FORMAT_XML ||
format == FORMAT_JSON ) {
struct tm *timeStruct = gmtime ( &lastSpiderDate );
strftime ( tbuf, 100,"%b %d, %Y UTC", timeStruct);
}
// We should always be displaying this disclaimer.
// - May eventually want to display this at a different location
// on the page, or on the click 'n' scroll browser page itself
// when this page is not being viewed solo.
// CNS: if ( ! st->m_clickNScroll ) {
if ( printDisclaimer ) {
sb->safePrintf(//sprintf ( p ,
//"<BASE HREF=\"%s\">"
//"<table border=1 width=100%%>"
//"<tr><td>"
"<table border=\"1\" bgcolor=\"#"
BGCOLOR
"\" cellpadding=\"10\" "
//"id=\"gbcnsdisctable\" class=\"gbcnsdisctable_v\""
"cellspacing=\"0\" width=\"100%%\" color=\"#ffffff\">"
"<tr"
//" id=\"gbcnsdisctr\" class=\"gbcnsdisctr_v\""
"><td>"
//"<font face=times,sans-serif color=black size=-1>"
"<span style=\"%s\">"
"This is Gigablast's cached page of </span>"
"<a href=\"%s\" style=\"%s\">%s</a>"
"" , styleTitle, f->getUrl(), styleLink,
f->getUrl() );
//p += gbstrlen ( p );
// then the rest
//sprintf(p ,
sb->safePrintf(
"<span style=\"%s\">. "
"Gigablast is not responsible for the content of "
"this page.</span>", styleTitle );
//p += gbstrlen ( p );
sb->safePrintf ( "<br/><span style=\"%s\">"
"Cached: </span>"
"<span style=\"%s\">",
styleTitle, styleText );
//p += gbstrlen ( p );
// then the spider date in GMT
// time_t lastSpiderDate = xd->m_spideredTime;
// struct tm *timeStruct = gmtime ( &lastSpiderDate );
// char tbuf[100];
// strftime ( tbuf, 100,"%b %d, %Y UTC", timeStruct);
//p += gbstrlen ( p );
sb->safeStrcpy(tbuf);
// Moved over from PageResults.cpp
sb->safePrintf( "</span> - <a href=\""
"/get?"
"q=%s&c=%s&rtq=%"INT32"&"
"d=%"INT64"&strip=1\""
" style=\"%s\">"
"[stripped]</a>",
q , st->m_coll ,
(int32_t)st->m_rtq,
st->m_docId, styleLink );
// a link to alexa
if ( f->getUrlLen() > 5 ) {
sb->safePrintf( " - <a href=\"http:"
"//web.archive.org/web/*/%s\""
" style=\"%s\">"
"[older copies]</a>" ,
f->getUrl(), styleLink );
}
if (st->m_noArchive){
sb->safePrintf( " - <span style=\"%s\"><b>"
"[NOARCHIVE]</b></span>",
styleTell );
}
if (st->m_isBanned){
sb->safePrintf(" - <span style=\"%s\"><b>"
"[BANNED]</b></span>",
styleTell );
}
// only print this if we got a query
if ( qlen > 0 ) {
sb->safePrintf("<br/><br/><span style=\"%s\"> "
"These search terms have been "
"highlighted: ",
styleText );
//p += gbstrlen ( p );
}
}
// how much space left in p?
//int32_t avail = bufEnd - p;
// . make the url that we're outputting for (like in PageResults.cpp)
// . "thisUrl" is the baseUrl for click & scroll
char thisUrl[MAX_URL_LEN];
char *thisUrlEnd = thisUrl + MAX_URL_LEN;
char *x = thisUrl;
// . use the external ip of our gateway
// . construct the NAT mapped port
// . you should have used iptables to map port to the correct
// internal ip:port
//uint32_t ip =g_conf.m_mainExternalIp ; // h->m_externalIp;
//uint16_t port=g_conf.m_mainExternalPort;//h->m_externalHttpPort
// local check
//if ( st->m_isLocal ) {
uint32_t ip = h->m_ip;
uint16_t port = h->m_httpPort;
//}
//sprintf ( x , "http://%s:%"INT32"/get?q=" , iptoa ( ip ) , port );
// . we no longer put the port in here
// . but still need http:// since we use <base href=>
if (port == 80) sprintf(x,"http://%s/get?q=",iptoa(ip));
else sprintf(x,"http://%s:%hu/get?q=",iptoa(ip),port);
x += gbstrlen ( x );
// the query url encoded
int32_t elen = urlEncode ( x , thisUrlEnd - x , q , qlen );
x += elen;
// separate cgi vars with a &
//sprintf ( x, "&seq=%"INT32"&rtq=%"INT32"d=%"INT64"",
// (int32_t)st->m_seq,(int32_t)st->m_rtq,st->m_msg22.getDocId());
sprintf ( x, "&d=%"INT64"",st->m_docId );
x += gbstrlen(x);
// set our query for highlighting
Query qq;
qq.set2 ( q, st->m_langId , true );
// print the query terms into our highlight buffer
Highlight hi;
// make words so we can set the scores to ignore fielded terms
Words qw;
qw.set ( q , // content being highlighted, utf8
qlen , // content being highlighted, utf8
TITLEREC_CURRENT_VERSION,
true , // computeIds
false ); // hasHtmlEntities?
// . assign scores of 0 to query words that should be ignored
// . TRICKY: loop over words in qq.m_qwords, but they should be 1-1
// with words in qw.
// . sanity check
//if ( qw.getNumWords() != qq.m_numWords ) { char *xx = NULL; *xx = 0;}
// declare up here
Matches m;
// do the loop
//Scores ss;
//ss.set ( &qw , NULL );
//for ( int32_t i = 0 ; i < qq.m_numWords ; i++ )
// if ( ! m.matchWord ( &qq.m_qwords[i],i ) ) ss.m_scores[i] = 0;
// now set m.m_matches[] to those words in qw that match a query word
// or phrase in qq.
m.setQuery ( &qq );
//m.addMatches ( &qw , &ss , true );
m.addMatches ( &qw );
int32_t hilen = 0;
// CNS: if ( ! st->m_clickNScroll ) {
// and highlight the matches
if ( printDisclaimer ) {
hilen = hi.set ( //p ,
//avail ,
sb ,
&qw , // words to highlight
&m , // matches relative to qw
false , // doSteming
false , // st->m_clickAndScroll ,
(char *)thisUrl );// base url for ClcknScrll
//p += hilen;
// now an hr
//gbmemcpy ( p , "</span></table></table>\n" , 24 ); p += 24;
sb->safeStrcpy("</span></table></table>\n");
}
bool includeHeader = st->m_includeHeader;
// do not show header for json object display
if ( xd->m_contentType == CT_JSON )
includeHeader = false;
if ( xd->m_contentType == CT_XML )
includeHeader = false;
if ( xd->m_contentType == CT_STATUS )
includeHeader = false;
if ( format == FORMAT_XML ) includeHeader = false;
if ( format == FORMAT_JSON ) includeHeader = false;
//mfree(uq, uqCapacity, "PageGet");
// undo the header writes if we should
if ( ! includeHeader ) {
// including base href is off by default when not including
// the header, so the caller must explicitly turn it back on
if ( st->m_includeBaseHref ) sb->m_length=startLen2;//p=start2;
else sb->m_length=startLen1;//p=start1;
}
//sb->safeStrcpy(tbuf);
if ( format == FORMAT_XML ) {
sb->safePrintf("<response>\n");
sb->safePrintf("<statusCode>0</statusCode>\n");
sb->safePrintf("<statusMsg>Success</statusMsg>\n");
sb->safePrintf("<url><![CDATA[");
sb->cdataEncode(xd->m_firstUrl.m_url);
sb->safePrintf("]]></url>\n");
sb->safePrintf("<docId>%"UINT64"</docId>\n",xd->m_docId);
sb->safePrintf("\t<cachedTimeUTC>%"INT32"</cachedTimeUTC>\n",
(int32_t)lastSpiderDate);
sb->safePrintf("\t<cachedTimeStr>%s</cachedTimeStr>\n",tbuf);
}
if ( format == FORMAT_JSON ) {
sb->safePrintf("{\"response\":{\n");
sb->safePrintf("\t\"statusCode\":0,\n");
sb->safePrintf("\t\"statusMsg\":\"Success\",\n");
sb->safePrintf("\t\"url\":\"");
sb->jsonEncode(xd->m_firstUrl.m_url);
sb->safePrintf("\",\n");
sb->safePrintf("\t\"docId\":%"UINT64",\n",xd->m_docId);
sb->safePrintf("\t\"cachedTimeUTC\":%"INT32",\n",
(int32_t)lastSpiderDate);
sb->safePrintf("\t\"cachedTimeStr\":\"%s\",\n",tbuf);
}
// identify start of <title> tag we wrote out
char *sbstart = sb->getBufStart();
char *sbend = sb->getBufEnd();
char *titleStart = NULL;
char *titleEnd = NULL;
char ctype = (char)xd->m_contentType;
// do not calc title or print it if doc is xml or json
if ( ctype == CT_XML ) sbend = sbstart;
if ( ctype == CT_JSON ) sbend = sbstart;
if ( ctype == CT_STATUS ) sbend = sbstart;
for ( char *t = sbstart ; t < sbend ; t++ ) {
// title tag?
if ( t[0]!='<' ) continue;
if ( to_lower_a(t[1])!='t' ) continue;
if ( to_lower_a(t[2])!='i' ) continue;
if ( to_lower_a(t[3])!='t' ) continue;
if ( to_lower_a(t[4])!='l' ) continue;
if ( to_lower_a(t[5])!='e' ) continue;
// point to it
char *x = t + 5;
// max - to keep things fast
char *max = x + 500;
for ( ; *x && *x != '>' && x < max ; x++ );
x++;
// find end
char *e = x;
for ( ; *e && e < max ; e++ ) {
if ( e[0]=='<' &&
to_lower_a(e[1])=='/' &&
to_lower_a(e[2])=='t' &&
to_lower_a(e[3])=='i' &&
to_lower_a(e[4])=='t' &&
to_lower_a(e[5])=='l' &&
to_lower_a(e[6])=='e' )
break;
}
if ( e < max ) {
titleStart = x;
titleEnd = e;
}
break;
}
// . print title at top!
// . consider moving
if ( titleStart ) {
char *ebuf = st->m_r.getString("eb");
if ( ! ebuf ) ebuf = "";
//p += sprintf ( p ,
sb->safePrintf(
"<table border=1 "
"cellpadding=10 "
"cellspacing=0 "
"width=100%% "
"color=#ffffff>" );
int32_t printLinks = st->m_r.getLong("links",0);
if ( ! printDisclaimer && printLinks )
sb->safePrintf(//p += sprintf ( p ,
// first put cached and live link
"<tr>"
"<td bgcolor=lightyellow>"
// print cached link
//"<center>"
" "
"<b>"
"<a "
"style=\"font-size:18px;font-weight:600;"
"color:#000000;\" "
"href=\""
"/get?"
"c=%s&d=%"INT64"&qh=0&cnsp=1&eb=%s\">"
"cached link</a>"
" "
"<a "
"style=\"font-size:18px;font-weight:600;"
"color:#000000;\" "
"href=%s>live link</a>"
"</b>"
//"</center>"
"</td>"
"</tr>\n"
,st->m_coll
,st->m_docId
,ebuf
,thisUrl // st->ptr_ubuf
);
if ( printLinks ) {
sb->safePrintf(//p += sprintf ( p ,
"<tr><td bgcolor=pink>"
"<span style=\"font-size:18px;"
"font-weight:600;"
"color:#000000;\">"
" "
"<b>PAGE TITLE:</b> "
);
int32_t tlen = titleEnd - titleStart;
sb->safeMemcpy ( titleStart , tlen );
sb->safePrintf ( "</span></td></tr>" );
}
sb->safePrintf( "</table><br>\n" );
}
// is the content preformatted?
bool pre = false;
if ( ctype == CT_TEXT ) pre = true ; // text/plain
if ( ctype == CT_DOC ) pre = true ; // filtered msword
if ( ctype == CT_PS ) pre = true ; // filtered postscript
if ( format == FORMAT_XML ) pre = false;
if ( format == FORMAT_JSON ) pre = false;
// if it is content-type text, add a <pre>
if ( pre ) {//p + 5 < bufEnd && pre ) {
sb->safePrintf("<pre>");
//p += 5;
}
if ( st->m_strip == 1 )
contentLen = stripHtml( content, contentLen,
(int32_t)xd->m_version, st->m_strip );
// it returns -1 and sets g_errno on error, line OOM
if ( contentLen == -1 ) {
//if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
return sendErrorReply ( st , g_errno );
}
Xml xml;
Words ww;
// if no highlighting, skip it
bool queryHighlighting = st->m_queryHighlighting;
if ( st->m_strip == 2 ) queryHighlighting = false;
// do not do term highlighting if json
if ( xd->m_contentType == CT_JSON )
queryHighlighting = false;
if ( xd->m_contentType == CT_STATUS )
queryHighlighting = false;
SafeBuf tmp;
SafeBuf *xb = sb;
if ( format == FORMAT_XML ) xb = &tmp;
if ( format == FORMAT_JSON ) xb = &tmp;
if ( ! queryHighlighting ) {
xb->safeMemcpy ( content , contentLen );
xb->nullTerm();
//p += contentLen ;
}
else {
// get the content as xhtml (should be NULL terminated)
//Words *ww = xd->getWords();
if ( ! xml.set ( content , contentLen , false ,
0 , false , TITLEREC_CURRENT_VERSION ,
false , 0 , CT_HTML ) ) { // niceness is 0
//if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
return sendErrorReply ( st , g_errno );
}
if ( ! ww.set ( &xml , true , 0 ) ) { // niceness is 0
//if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
return sendErrorReply ( st , g_errno );
}
// sanity check
//if ( ! xd->m_wordsValid ) { char *xx=NULL;*xx=0; }
// how much space left in p?
//avail = bufEnd - p;
Matches m;
m.setQuery ( &qq );
m.addMatches ( &ww );
hilen = hi.set ( xb , // p , avail ,
&ww , &m ,
false /*doStemming?*/ ,
st->m_clickAndScroll ,
thisUrl /*base url for click & scroll*/);
//p += hilen;
log(LOG_DEBUG, "query: Done highlighting cached page content");
}
if ( format == FORMAT_XML ) {
sb->safePrintf("\t<content><![CDATA[");
sb->cdataEncode ( xb->getBufStart() );
sb->safePrintf("]]></content>\n");
sb->safePrintf("</response>\n");
}
if ( format == FORMAT_JSON ) {
sb->safePrintf("\t\"content\":\"\n");
sb->jsonEncode ( xb->getBufStart() );
sb->safePrintf("\"\n}\n}\n");
}
// if it is content-type text, add a </pre>
if ( pre ) { // p + 6 < bufEnd && pre ) {
sb->safeMemcpy ( "</pre>" , 6 );
//p += 6;
}
// calculate bufLen
//int32_t bufLen = p - buf;
/*
MDW: return the xml page as is now. 9/28/2014
int32_t ct = xd->m_contentType;
// now filter the entire buffer to escape out the xml tags
// so it is displayed nice
SafeBuf newbuf;
if ( ct == CT_XML ) {
// encode the xml tags into <tagname> sequences
if ( !newbuf.htmlEncodeXmlTags ( sb->getBufStart() ,
sb->getLength(),
0)){// niceness=0
//if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
return sendErrorReply ( st , g_errno );
}
// free out buffer that we alloc'd before returning since this
// should have copied it into another buffer
//if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
// reassign
//buf = newbuf.getBufStart();
//bufLen = newbuf.length();
sb->stealBuf ( &newbuf );
}
*/
// now encapsulate it in html head/tail and send it off
// sendErr:
contentType = "text/html";
if ( strip == 2 ) contentType = "text/xml";
// xml is usually buggy and this throws browser off
//if ( ctype == CT_XML ) contentType = "text/xml";
if ( xd->m_contentType == CT_JSON )
contentType = "application/json";
if ( xd->m_contentType == CT_STATUS )
contentType = "application/json";
if ( xd->m_contentType == CT_XML )
contentType = "test/xml";
if ( format == FORMAT_XML ) contentType = "text/xml";
if ( format == FORMAT_JSON ) contentType = "application/json";
// safebuf, sb, is a member of "st" so this should copy the buffer
// when it constructs the http reply, and we gotta call delete(st)
// AFTER this so sb is still valid.
bool status = g_httpServer.sendDynamicPage (s,
//buf,bufLen,
sb->getBufStart(),
sb->getLength(),
-1,false,
contentType,
-1, NULL, "utf8" );
// nuke state2
mdelete ( st , sizeof(State2) , "PageGet1" );
delete (st);
// free out buffer that we alloc'd before returning since this
// should have copied it into another buffer
//if ( ct == CT_XML ) newbuf.purge();
//else if ( buf ) mfree ( buf , bufMaxSize , "PageGet2" );
// and convey the status
return status;
}
bool qaspider1 ( ) {
//
// delete the 'qatest123' collection
//
//static bool s_x1 = false;
if ( ! s_flags[0] ) {
s_flags[0] = true;
if ( ! getUrl ( "/admin/delcoll?xml=1&delcoll=qatest123" ) )
return false;
}
//
// add the 'qatest123' collection
//
//static bool s_x2 = false;
if ( ! s_flags[1] ) {
s_flags[1] = true;
if ( ! getUrl ( "/admin/addcoll?addcoll=qatest123&xml=1" ,
// checksum of reply expected
238170006 ) )
return false;
}
// restrict hopcount to 0 or 1 in url filters so we do not spider
// too deep
//static bool s_z1 = false;
if ( ! s_flags[2] ) {
s_flags[2] = true;
SafeBuf sb;
sb.safePrintf("&c=qatest123&"
// make it the custom filter
"ufp=0&"
"fe=%%21ismanualadd+%%26%%26+%%21insitelist&hspl=0&hspl=1&fsf=0.000000&mspr=0&mspi=1&xg=1000&fsp=-3&"
// take out hopcount for now, just test quotas
// "fe1=tag%%3Ashallow+%%26%%26+hopcount%%3C%%3D1&hspl1=0&hspl1=1&fsf1=1.000000&mspr1=1&mspi1=1&xg1=1000&fsp1=3&"
// just one spider out allowed for consistency
"fe1=tag%%3Ashallow+%%26%%26+sitepages%%3C%%3D20&hspl1=0&hspl1=1&fsf1=1.000000&mspr1=1&mspi1=1&xg1=1000&fsp1=45&"
"fe2=default&hspl2=0&hspl2=1&fsf2=1.000000&mspr2=0&mspi2=1&xg2=1000&fsp2=45&"
);
if ( ! getUrl ( "/admin/filters",0,sb.getBufStart()) )
return false;
}
// set the site list to
// a few sites
//static bool s_z2 = false;
if ( ! s_flags[3] ) {
s_flags[3] = true;
SafeBuf sb;
sb.safePrintf("&c=qatest123&format=xml&sitelist=");
sb.urlEncode("tag:shallow site:www.walmart.com\r\n"
"tag:shallow site:http://www.ibm.com/\r\n");
sb.nullTerm();
if ( ! getUrl ("/admin/settings",0,sb.getBufStart() ) )
return false;
}
//
// use the add url interface now
// walmart.com above was not seeded because of the site: directive
// so this will seed it.
//
//static bool s_y2 = false;
if ( ! s_flags[4] ) {
s_flags[4] = true;
SafeBuf sb;
// delim=+++URL:
sb.safePrintf("&c=qatest123"
"&format=json"
"&strip=1"
"&spiderlinks=1"
"&urls=www.walmart.com+ibm.com"
);
// . now a list of websites we want to spider
// . the space is already encoded as +
//sb.urlEncode(s_urls1);
if ( ! getUrl ( "/admin/addurl",0,sb.getBufStart()) )
return false;
}
//
// wait for spidering to stop
//
checkagain:
// wait until spider finishes. check the spider status page
// in json to see when completed
//static bool s_k1 = false;
if ( ! s_flags[5] ) {
// wait 5 seconds, call sleep timer... then call qatest()
//usleep(5000000); // 5 seconds
wait(3.0);
s_flags[5] = true;
return false;
}
if ( ! s_flags[15] ) {
s_flags[15] = true;
if ( ! getUrl ( "/admin/status?format=json&c=qatest123",0) )
return false;
}
//static bool s_k2 = false;
if ( ! s_flags[6] ) {
// ensure spiders are done.
// "Nothing currently available to spider"
if ( s_content&&!strstr(s_content,"Nothing currently avail")){
s_flags[5] = false;
s_flags[15] = false;
goto checkagain;
}
s_flags[6] = true;
}
// wait for index msg4 to not be cached to ensure all results indexed
if ( ! s_flags[22] ) {
s_flags[22] = true;
wait(1.5);
}
// verify no results for gbhopcount:2 query
//static bool s_y4 = false;
if ( ! s_flags[7] ) {
s_flags[7] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"q=gbhopcount%3A2",
-1672870556 ) )
return false;
}
// but some for gbhopcount:0 query
//static bool s_t0 = false;
if ( ! s_flags[8] ) {
s_flags[8] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"q=gbhopcount%3A0",
908338607 ) )
return false;
}
// check facet sections query for walmart
//static bool s_y5 = false;
if ( ! s_flags[9] ) {
s_flags[9] = true;
if ( ! getUrl ( "/search?c=qatest123&format=json&stream=1&"
"q=gbfacetstr%3Agbxpathsitehash2492664135",
55157060 ) )
return false;
}
//static bool s_y6 = false;
if ( ! s_flags[10] ) {
s_flags[10] = true;
if ( ! getUrl ( "/get?page=4&q=gbfacetstr:gbxpathsitehash2492664135&qlang=xx&c=qatest123&d=9861563119&cnsp=0" , 999 ) )
return false;
}
// in xml
//static bool s_y7 = false;
if ( ! s_flags[11] ) {
s_flags[11] = true;
if ( ! getUrl ( "/get?xml=1&page=4&q=gbfacetstr:gbxpathsitehash2492664135&qlang=xx&c=qatest123&d=9861563119&cnsp=0" , 999 ) )
return false;
}
// and json
//static bool s_y8 = false;
if ( ! s_flags[12] ) {
s_flags[12] = true;
if ( ! getUrl ( "/get?json=1&page=4&q=gbfacetstr:gbxpathsitehash2492664135&qlang=xx&c=qatest123&d=9861563119&cnsp=0" , 999 ) )
return false;
}
// delete the collection
//static bool s_fee = false;
// if ( ! s_flags[13] ) {
// s_flags[13] = true;
// if ( ! getUrl ( "/admin/delcoll?delcoll=qatest123" ) )
// return false;
// }
if ( ! s_flags[17] ) {
s_flags[17] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"q=site2%3Awww.walmart.com+"
"gbsortby%3Agbspiderdate",
999 ) )
return false;
}
// xpath is like a title here i think. check the returned
// facet table in the left column
if ( ! s_flags[18] ) {
s_flags[18] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=html&"
"q=gbfacetstr%3Agbxpathsitehash3624590799"
, 999 ) )
return false;
}
if ( ! s_flags[19] ) {
s_flags[19] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&xml=1&"
"q=gbfacetint%3Agbhopcount"
, 999 ) )
return false;
}
if ( ! s_flags[20] ) {
s_flags[20] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&json=1&"
"q=gbfacetint%3Alog.score"
, 999 ) )
return false;
}
if ( ! s_flags[21] ) {
s_flags[21] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&xml=1&"
"q=gbfacetfloat%3Atalks.rating"
, 999 ) )
return false;
}
if ( ! s_flags[23] ) {
s_flags[23] = true;
// test facets mixed with gigabits in left hand column
if ( ! getUrl ( "/search?c=qatest123&qa=1&html=1&"
"q=gbfacetint%3Agbhopcount+walmart"
, 999 ) )
return false;
}
//static bool s_fee2 = false;
if ( ! s_flags[14] ) {
s_flags[14] = true;
log("qa: SUCCESSFULLY COMPLETED "
"QA SPIDER1 TEST");
return true;
}
return true;
}
//
// the injection qa test suite
//
bool qainject1 ( ) {
//if ( ! s_callback ) s_callback = qainject1;
//
// delete the 'qatest123' collection
//
//static bool s_x1 = false;
if ( ! s_flags[0] ) {
s_flags[0] = true;
if ( ! getUrl ( "/admin/delcoll?xml=1&delcoll=qatest123" ) )
return false;
}
//
// add the 'qatest123' collection
//
//static bool s_x2 = false;
if ( ! s_flags[1] ) {
s_flags[1] = true;
if ( ! getUrl ( "/admin/addcoll?addcoll=qatest123&xml=1" ,
// checksum of reply expected
238170006 ) )
return false;
}
// this only loads once
loadUrls();
long max = s_ubuf2.length()/(long)sizeof(char *);
//max = 1;
//
// inject urls, return false if not done yet
//
//static bool s_x4 = false;
if ( ! s_flags[2] ) {
// TODO: try delimeter based injection too
//static long s_ii = 0;
for ( ; s_flags[20] < max ; ) {
// inject using html api
SafeBuf sb;
sb.safePrintf("&c=qatest123&deleteurl=0&"
"format=xml&u=");
sb.urlEncode ( s_urlPtrs[s_flags[20]] );
// the content
sb.safePrintf("&hasmime=1");
// sanity
//if ( strstr(s_urlPtrs[s_flags[20]],"wdc.htm") )
// log("hey");
sb.safePrintf("&content=");
sb.urlEncode(s_contentPtrs[s_flags[20]] );
sb.nullTerm();
// pre-inc it in case getUrl() blocks
s_flags[20]++;//ii++;
if ( ! getUrl("/admin/inject",
0, // no idea what crc to expect
sb.getBufStart()) )
return false;
}
s_flags[2] = true;
}
// +the
//static bool s_x5 = false;
if ( ! s_flags[3] ) {
wait(1.5);
s_flags[3] = true;
return false;
}
if ( ! s_flags[16] ) {
s_flags[16] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&q=%2Bthe",
702467314 ) )
return false;
}
// sports news
//static bool s_x7 = false;
if ( ! s_flags[4] ) {
s_flags[4] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"q=sports+news",2009472889 ) )
return false;
}
// 'washer & dryer' does some algorithmic synonyms 'washer and dryer'
if ( ! s_flags[15] ) {
s_flags[15] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"debug=1&q=washer+%26+dryer",9999 ) )
return false;
}
//
// mdw: query reindex test
//
// if ( ! s_flags[30] ) {
// s_flags[30] = true;
// if ( ! getUrl ( "/admin/reindex?c=qatest123&qa=1&format=xml&"
// "debug=1&q=sports",9999 ) )
// return false;
// }
// // temp end it here
// return true;
//
// eject/delete the urls
//
//static long s_ii2 = 0;
for ( ; s_flags[5] < max ; ) {
// reject using html api
SafeBuf sb;
sb.safePrintf( "/admin/inject?c=qatest123&deleteurl=1&"
"format=xml&u=");
sb.urlEncode ( s_urlPtrs[s_flags[5]] );
sb.nullTerm();
// pre-inc it in case getUrl() blocks
//s_ii2++;
s_flags[5]++;
if ( ! getUrl ( sb.getBufStart() , 0 ) )
return false;
}
//
// make sure no results left, +the
//
if ( ! s_flags[6] ) {
wait(1.5);
s_flags[6] = true;
return false;
}
if ( ! s_flags[14] ) {
s_flags[14] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=2&format=xml&q=%2Bthe",
-1672870556 ) )
return false;
}
//static bool s_fee2 = false;
if ( ! s_flags[13] ) {
s_flags[13] = true;
log("qa: SUCCESSFULLY COMPLETED "
"QA INJECT TEST 1");
//if ( s_callback == qainject ) exit(0);
return true;
}
return true;
}
bool qaspider2 ( ) {
//
// delete the 'qatest123' collection
//
//static bool s_x1 = false;
if ( ! s_flags[0] ) {
s_flags[0] = true;
if ( ! getUrl ( "/admin/delcoll?xml=1&delcoll=qatest123" ) )
return false;
}
//
// add the 'qatest123' collection
//
//static bool s_x2 = false;
if ( ! s_flags[1] ) {
s_flags[1] = true;
if ( ! getUrl ( "/admin/addcoll?addcoll=qatest123&xml=1" ,
// checksum of reply expected
238170006 ) )
return false;
}
// restrict hopcount to 0 or 1 in url filters so we do not spider
// too deep
//static bool s_z1 = false;
if ( ! s_flags[2] ) {
s_flags[2] = true;
SafeBuf sb;
sb.safePrintf("&c=qatest123&"
// make it the custom filter
"ufp=0&"
"fe=%%21ismanualadd+%%26%%26+%%21insitelist&hspl=0&hspl=1&fsf=0.000000&mspr=0&mspi=1&xg=1000&fsp=-3&"
// take out hopcount for now, just test quotas
// "fe1=tag%%3Ashallow+%%26%%26+hopcount%%3C%%3D1&hspl1=0&hspl1=1&fsf1=1.000000&mspr1=1&mspi1=1&xg1=1000&fsp1=3&"
// sitepages is a little fuzzy so take it
// out for this test and use hopcount!!!
//"fe1=tag%%3Ashallow+%%26%%26+sitepages%%3C%%3D20&hspl1=0&hspl1=1&fsf1=1.000000&mspr1=1&mspi1=1&xg1=1000&fsp1=45&"
"fe1=tag%%3Ashallow+%%26%%26+hopcount<%%3D1&hspl1=0&hspl1=1&fsf1=1.000000&mspr1=1&mspi1=1&xg1=1000&fsp1=45&"
"fe2=default&hspl2=0&hspl2=1&fsf2=1.000000&mspr2=0&mspi2=1&xg2=1000&fsp2=45&"
);
if ( ! getUrl ( "/admin/filters",0,sb.getBufStart()) )
return false;
}
// set the site list to
// a few sites
// these should auto seed so no need to use addurl
//static bool s_z2 = false;
if ( ! s_flags[3] ) {
s_flags[3] = true;
SafeBuf sb;
sb.safePrintf("&c=qatest123&format=xml&sitelist=");
sb.urlEncode(//walmart has too many pages at depth 1, so remove it
//"tag:shallow www.walmart.com\r\n"
"tag:shallow http://www.ibm.com/\r\n");
sb.nullTerm();
if ( ! getUrl ("/admin/settings",0,sb.getBufStart() ) )
return false;
}
//
// wait for spidering to stop
//
checkagain:
// wait until spider finishes. check the spider status page
// in json to see when completed
//static bool s_k1 = false;
if ( ! s_flags[4] ) {
//usleep(5000000); // 5 seconds
s_flags[4] = true;
wait(3.0);
return false;
}
if ( ! s_flags[14] ) {
s_flags[14] = true;
if ( ! getUrl ( "/admin/status?format=json&c=qatest123",0) )
return false;
}
//static bool s_k2 = false;
if ( ! s_flags[5] ) {
// ensure spiders are done.
// "Nothing currently available to spider"
if ( s_content&&!strstr(s_content,"Nothing currently avail")){
s_flags[4] = false;
s_flags[14] = false;
goto checkagain;
}
s_flags[5] = true;
}
// verify no results for gbhopcount:2 query
//static bool s_y4 = false;
if ( ! s_flags[6] ) {
s_flags[6] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&"
"q=gbhopcount%3A2",
-1310551262 ) )
return false;
}
// but some for gbhopcount:0 query
//static bool s_t0 = false;
if ( ! s_flags[7] ) {
s_flags[7] = true;
if ( ! getUrl ( "/search?c=qatest123&qa=1&format=xml&n=500&"
"q=gbhopcount%3A0",
999 ) )
return false;
}
// check facet sections query for walmart
//static bool s_y5 = false;
if ( ! s_flags[8] ) {
s_flags[8] = true;
if ( ! getUrl ( "/search?c=qatest123&format=json&stream=0&"
"q=gbfacetstr%3Agbxpathsitehash3311332088",
999 ) )
return false;
}
// wait for some reason
if ( ! s_flags[15] ) {
s_flags[15] = true;
wait(1.5);
return false;
}
//static bool s_y6 = false;
if ( ! s_flags[9] ) {
s_flags[9] = true;
if ( ! getUrl ( "/get?page=4&q=gbfacetstr:gbxpathsitehash3311332088&qlang=xx&c=qatest123&d=9577169402&cnsp=0" , 999 ) )
return false;
}
// in xml
//static bool s_y7 = false;
if ( ! s_flags[10] ) {
s_flags[10] = true;
if ( ! getUrl ( "/get?xml=1&page=4&q=gbfacetstr:gbxpathsitehash2492664135&qlang=xx&c=qatest123&d=9577169402&cnsp=0" , 999 ) )
return false;
}
// and json
//static bool s_y8 = false;
if ( ! s_flags[11] ) {
s_flags[11] = true;
if ( ! getUrl ( "/get?json=1&page=4&q=gbfacetstr:gbxpathsitehash2492664135&qlang=xx&c=qatest123&d=9577169402&cnsp=0" , 999 ) )
return false;
}
// delete the collection
//static bool s_fee = false;
// if ( ! s_flags[12] ) {
// s_flags[12] = true;
// if ( ! getUrl ( "/admin/delcoll?delcoll=qatest123" ) )
// return false;
// }
//static bool s_fee2 = false;
if ( ! s_flags[13] ) {
s_flags[13] = true;
log("qa: SUCCESSFULLY COMPLETED "
"QA SPIDER2 TEST");
return true;
}
return true;
}
// . return the score of the highest-scoring window containing match #m
// . window is defined by the half-open interval [a,b) where a and b are
// word #'s in the Words array indicated by match #m
// . return -1 and set g_errno on error
int64_t Summary::getBestWindow ( Matches *matches, int32_t mm, int32_t *lasta,
int32_t *besta, int32_t *bestb, char *gotIt,
char *retired, int32_t maxExcerptLen ) {
// get the window around match #mm
Match *m = &matches->m_matches[mm];
// what is the word # of match #mm?
int32_t matchWordNum = m->m_wordNum;
// what Words/Pos/Bits classes is this match in?
Words *words = m->m_words;
Section **sp = NULL;
int32_t *pos = m->m_pos->m_pos;
// use "m_swbits" not "m_bits", that is what Bits::setForSummary() uses
const swbit_t *bb = m->m_bits->m_swbits;
// shortcut
if ( m->m_sections ) {
sp = m->m_sections->m_sectionPtrs;
}
int32_t nw = words->getNumWords();
int64_t *wids = words->getWordIds();
nodeid_t *tids = words->getTagIds();
// . sanity check
// . this prevents a core i've seen
if ( matchWordNum >= nw ) {
log("summary: got overflow condition for q=%s",m_q->m_orig);
// assume no best window
*besta = -1;
*bestb = -1;
*lasta = matchWordNum;
return 0;
}
// . we NULLify the section ptrs if we already used the word in another summary.
int32_t badFlags = SEC_SCRIPT|SEC_STYLE|SEC_SELECT|SEC_IN_TITLE;
if ( (bb[matchWordNum] & D_USED) || ( sp && (sp[matchWordNum]->m_flags & badFlags) ) ) {
// assume no best window
*besta = -1;
*bestb = -1;
*lasta = matchWordNum;
return 0;
}
// . "a" is the left fence post of the window (it is a word # in Words)
// . go to the left as far as we can
// . thus we decrement "a"
int32_t a = matchWordNum;
// "posa" is the character position of the END of word #a
int32_t posa = pos[a+1];
int32_t firstFrag = -1;
bool startOnQuote = false;
bool goodStart = false;
int32_t wordCount = 0;
// . decrease "a" as int32_t as we stay within maxNumCharsPerLine
// . avoid duplicating windows by using "lasta", the last "a" of the
// previous call to getBestWindow(). This can happen if our last
// central query term was close to this one.
for ( ; a > 0 && posa - pos[a-1] < maxExcerptLen && a > *lasta; a-- ) {
// . don't include any "dead zone",
// . dead zones have already been used for the summary, and
// we are getting a second/third/... excerpt here now then
// stop if its the start of a sentence, too
// stop before title word
if ( (bb[a-1] & D_USED) || (bb[a] & D_STARTS_SENTENCE) || ( bb[a-1] & D_IN_TITLE )) {
goodStart = true;
break;
}
// don't go beyond an LI, TR, P tag
if ( tids && ( tids[a-1] == TAG_LI ||
tids[a-1] == TAG_TR ||
tids[a-1] == TAG_P ||
tids[a-1] == TAG_DIV ) ) {
goodStart = true;
break;
}
// stop if its the start of a quoted sentence
if ( a+1<nw && (bb[a+1] & D_IN_QUOTES) &&
words->getWord(a)[0] == '\"' ){
startOnQuote = true;
goodStart = true;
break;
}
// find out the first instance of a fragment (comma, etc)
// watch out! because frag also means 's' in there's
if ( ( bb[a] & D_STARTS_FRAG ) && !(bb[a-1] & D_IS_STRONG_CONNECTOR) && firstFrag == -1 ) {
firstFrag = a;
}
if ( wids[a] ) {
wordCount++;
}
}
// if didn't find a good start, then start at the start of the frag
if ( !goodStart && firstFrag != -1 ) {
a = firstFrag;
}
// don't let punct or tag word start a line, unless a quote
if ( a < matchWordNum && !wids[a] && words->getWord(a)[0] != '\"' ){
while ( a < matchWordNum && !wids[a] ) a++;
// do not break right after a "strong connector", like
// apostrophe
while ( a < matchWordNum && a > 0 &&
( bb[a-1] & D_IS_STRONG_CONNECTOR ) )
a++;
// don't let punct or tag word start a line
while ( a < matchWordNum && !wids[a] ) a++;
}
// remember, b is not included in the summary, the summary is [a,b-1]
// remember to include all words in a matched phrase
int32_t b = matchWordNum + m->m_numWords ;
int32_t endQuoteWordNum = -1;
int32_t numTagsCrossed = 0;
for ( ; b <= nw; b++ ) {
if ( b == nw ) {
break;
}
if ( pos[b+1] - pos[a] >= maxExcerptLen ) {
break;
}
if ( startOnQuote && words->getWord(b)[0] == '\"' ) {
endQuoteWordNum = b;
}
// don't include any dead zone, those are already-used samples
if ( bb[b] & D_USED ) {
break;
}
// stop on a title word
if ( bb[b] & D_IN_TITLE ) {
break;
}
if ( wids[b] ) {
wordCount++;
}
// don't go beyond an LI or TR backtag
if ( tids && ( tids[b] == (BACKBIT|TAG_LI) ||
tids[b] == (BACKBIT|TAG_TR) ) ) {
numTagsCrossed++;
// try to have atleast 10 words in the summary
if ( wordCount > 10 ) {
break;
}
}
// go beyond a P or DIV backtag in case the earlier char is a
// ':'. This came from a special case for wikipedia pages
// eg. http://en.wikipedia.org/wiki/Flyover
if ( tids && ( tids[b] == (BACKBIT|TAG_P) ||
tids[b] == (BACKBIT|TAG_DIV) )) {
numTagsCrossed++;
// try to have atleast 10 words in the summary
if ( wordCount > 10 && words->getWord(b-1)[0] != ':' ) {
break;
}
}
}
// don't end on a lot of punct words
if ( b > matchWordNum && !wids[b-1]){
// remove more than one punct words. if we're ending on a quote
// keep it
while ( b > matchWordNum && !wids[b-2] && endQuoteWordNum != -1 && b > endQuoteWordNum ) {
b--;
}
// do not break right after a "strong connector", like apostrophe
while ( b > matchWordNum && (bb[b-2] & D_IS_STRONG_CONNECTOR) ) {
b--;
}
}
Match *ms = matches->m_matches;
// make m_matches.m_matches[mi] the first match in our [a,b) window
int32_t mi ;
// . the match at the center of the window is match #"mm", so that
// matches->m_matches[mm] is the Match class
// . set "mi" to it and back up "mi" as int32_t as >= a
for ( mi = mm ; mi > 0 && ms[mi-1].m_wordNum >=a ; mi-- )
;
// now get the score of this excerpt. Also mark all the represented
// query words. Mark the represented query words in the array that
// comes to us. also mark how many times the same word is repeated in
// this summary.
int64_t score = 0LL;
// is a url contained in the summary, that looks bad! punish!
bool hasUrl = false;
// the word count we did above was just an approximate. count it right
wordCount = 0;
// for debug
//char buf[5000];
//char *xp = buf;
SafeBuf xp;
// wtf?
if ( b > nw ) {
b = nw;
}
// first score from the starting match down to a, including match
for ( int32_t i = a ; i < b ; i++ ) {
// debug print out
if ( g_conf.m_logDebugSummary ) {
int32_t len = words->getWordLen(i);
char cs;
for (int32_t k=0;k<len; k+=cs ) {
const char *c = words->getWord(i)+k;
cs = getUtf8CharSize(c);
if ( is_binary_utf8 ( c ) ) {
continue;
}
xp.safeMemcpy ( c , cs );
xp.nullTerm();
}
}
// skip if in bad section, marquee, select, script, style
if ( sp && (sp[i]->m_flags & badFlags) ) {
continue;
}
// don't count just numeric words
if ( words->isNum(i) ) {
continue;
}
// check if there is a url. best way to check for '://'
if ( wids && !wids[i] ) {
const char *wrd = words->getWord(i);
int32_t wrdLen = words->getWordLen(i);
if ( wrdLen == 3 && wrd[0] == ':' && wrd[1] == '/' && wrd[2] == '/' ) {
hasUrl = true;
}
}
// skip if not wid
if ( ! wids[i] ) {
continue;
}
// just make every word 100 pts
int32_t t = 100;
// penalize it if in one of these sections
if ( bb[i] & ( D_IN_PARENS | D_IN_SUP | D_IN_LIST ) ) {
t /= 2;
}
// boost it if in bold or italics
if ( bb[i] & D_IN_BOLDORITALICS ) {
t *= 2;
}
// add the score for this word
score += t;
// print the score, "t"
if ( g_conf.m_logDebugSummary ) {
xp.safePrintf("(%" PRId32")",t);
}
// count the alpha words we got
wordCount++;
// if no matches left, skip
if ( mi >= matches->m_numMatches ) {
continue;
}
// get the match
Match *next = &ms[mi];
// skip if not a match
if ( i != next->m_wordNum ) {
continue;
}
// must be a match in this class
if ( next->m_words != words ) {
continue;
}
// advance it
mi++;
// which query word # does it match
int32_t qwn = next->m_qwordNum;
if ( qwn < 0 || qwn >= m_q->m_numWords ){g_process.shutdownAbort(true);}
// undo old score
score -= t;
// add 100000 per match
t = 100000;
// weight based on tf, goes from 0.1 to 1.0
t = (int32_t)((float)t * m_wordWeights [ qwn ]);
// if it is a query stop word, make it 10000 pts
if ( m_q->m_qwords[qwn].m_isQueryStopWord ) {
t = 0;//10000;
}
// penalize it if in one of these sections
if ( bb[i] & ( D_IN_PARENS | D_IN_SUP | D_IN_LIST ) ) {
t /= 2;
}
if ( gotIt[qwn] > 0 ) {
// have we matched it in this [a,b) already?
if ( gotIt[qwn] == 1 ) {
t /= 15;
} else {
// if we have more than 2 matches in the same window,
// it may not give a good summary. give a heavy penalty
t -= 200000;
}
} else if ( retired [qwn] > 0 ) {
// have we matched it already in a winning window?
t /= 12;
}
// add it back
score += t;
if ( g_conf.m_logDebugSummary ) {
xp.safePrintf ("[%" PRId32"]{qwn=%" PRId32",ww=%f}",t,qwn,
m_wordWeights[qwn]);
}
// inc the query word count for this window
if ( gotIt[qwn] < 100 ) {
gotIt[qwn]++;
}
}
int32_t oldScore = score;
// apply the bonus if it starts or a sentence
// only apply if the score is positive and if the wordcount is decent
if ( score > 0 && wordCount > 7 ){
// a match can give us 10k to 100k pts based on the tf weights
// so we don't want to overwhelm that too much, so let's make
// this a 20k bonus if it starts a sentence
if ( bb[a] & D_STARTS_SENTENCE ) {
score += 8000;
} else if ( bb[a] & D_STARTS_FRAG ) {
// likewise, a fragment, like after a comma
score += 4000;
}
// 1k if the match word is very close to the
// start of a sentence, lets say 3 alphawords
if ( matchWordNum - a < 7 ) {
score += 1000;
}
}
// a summary isn't really a summary if its less than 7 words.
// reduce the score, but still give it a decent score.
// minus 5M.
if ( wordCount < 7 ) {
score -= 20000;
}
// summaries that cross a lot of tags are usually bad, penalize them
if ( numTagsCrossed > 1 ) {
score -= (numTagsCrossed * 20000);
}
if ( hasUrl ) {
score -= 8000;
}
// show it
if ( g_conf.m_logDebugSummary ) {
log(LOG_DEBUG, "sum: score=%08" PRId32" prescore=%08" PRId32" a=%05" PRId32" b=%05" PRId32" %s",
(int32_t)score,oldScore,(int32_t)a,(int32_t)b,
xp.getBufStart());
}
// set lasta, besta, bestb
*lasta = a;
*besta = a;
*bestb = b;
return score;
}
// . "uf" is printf url format to scrape with a %s for the query
// . example: uf="http://www.google.com/search?num=50&q=%s&scoring=d&filter=0";
bool Msg7::scrapeQuery ( ) {
// advance round now in case we return early
m_round++;
GigablastRequest *gr = &m_gr;
// error?
char *qts = gr->m_queryToScrape;
if ( ! qts ) { char *xx=NULL;*xx=0; }
if ( gbstrlen(qts) > 500 ) {
g_errno = EQUERYTOOBIG;
return true;
}
// first encode the query
SafeBuf ebuf;
ebuf.urlEncode ( qts ); // queryUNEncoded );
ebuf.nullTerm();
char *uf;
if ( m_round == 1 )
// set to 1 for debugging
uf="http://www.google.com/search?num=20&"
"q=%s&scoring=d&filter=0";
//uf = "https://startpage.com/do/search?q=%s";
//uf = "http://www.google.com/"
// "/cse?cx=013269018370076798483%3A8eec3papwpi&"
// "ie=UTF-8&q=%s&"
// "num=20";
else
uf="http://www.bing.com/search?q=%s";
// skip bing for now
//if ( m_round == 2 )
// return true;
//if ( m_round == 1 )
// return true;
// make the url we will download
char ubuf[2048];
sprintf ( ubuf , uf , ebuf.getBufStart() );
// log it
log("inject: SCRAPING %s",ubuf);
SpiderRequest sreq;
sreq.reset();
// set the SpiderRequest
strcpy(sreq.m_url, ubuf);
// . tell it to only add the hosts of each outlink for now!
// . that will be passed on to when XmlDoc calls Links::set() i guess
// . xd will not reschedule the scraped url into spiderdb either
sreq.m_isScraping = 1;
sreq.m_fakeFirstIp = 1;
long firstIp = hash32n(ubuf);
if ( firstIp == 0 || firstIp == -1 ) firstIp = 1;
sreq.m_firstIp = firstIp;
// parent docid is 0
sreq.setKey(firstIp,0LL,false);
char *coll2 = gr->m_coll;
CollectionRec *cr = g_collectiondb.getRec ( coll2 );
// forceDEl = false, niceness = 0
m_xd.set4 ( &sreq , NULL , cr->m_coll , NULL , 0 );
//m_xd.m_isScraping = true;
// download without throttling
//m_xd.m_throttleDownload = false;
// disregard this
m_xd.m_useRobotsTxt = false;
// this will tell it to index ahrefs first before indexing
// the doc. but do NOT do this if we are from ahrefs.com
// ourselves to avoid recursive explosion!!
if ( m_useAhrefs )
m_xd.m_useAhrefs = true;
m_xd.m_reallyInjectLinks = true;//gr->m_injectLinks;
//
// rather than just add the links of the page to spiderdb,
// let's inject them!
//
m_xd.setCallback ( this , doneInjectingLinksWrapper );
// niceness is 0
m_linkDedupTable.set(4,0,512,NULL,0,false,0,"ldtab2");
// do we actually inject the links, or just scrape?
if ( ! m_xd.injectLinks ( &m_linkDedupTable ,
NULL,
this ,
doneInjectingLinksWrapper ) )
return false;
// otherwise, just download the google/bing search results so we
// can display them in xml
//else if ( m_xd.getUtf8Content() == (char **)-1 )
// return false;
// print reply..
//printReply();
return true;
}
bool HttpMime::addToCookieJar(Url *currentUrl, SafeBuf *sb) {
/// @note Slightly modified from Netscape HTTP Cookie File format
/// Difference is we only have one column for name/value
// http://www.cookiecentral.com/faq/#3.5
// The layout of Netscape's cookies.txt file is such that each line contains one name-value pair.
// An example cookies.txt file may have an entry that looks like this:
// .netscape.com TRUE / FALSE 946684799 NETSCAPE_ID 100103
//
// Each line represents a single piece of stored information. A tab is inserted between each of the fields.
// From left-to-right, here is what each field represents:
//
// domain - The domain that created AND that can read the variable.
// flag - A TRUE/FALSE value indicating if all machines within a given domain can access the variable. This value is set automatically by the browser, depending on the value you set for domain.
// path - The path within the domain that the variable is valid for.
// secure - A TRUE/FALSE value indicating if a secure connection with the domain is needed to access the variable.
// expiration - The UNIX time that the variable will expire on. UNIX time is defined as the number of seconds since Jan 1, 1970 00:00:00 GMT.
// name/value - The name/value of the variable.
/// @todo ALC we should sort cookie-list
// The user agent SHOULD sort the cookie-list in the following order:
// * Cookies with longer paths are listed before cookies with shorter paths.
// * Among cookies that have equal-length path fields, cookies with earlier creation-times are listed
// before cookies with later creation-times.
// fill in cookies from cookieJar
std::map<std::string, httpcookie_t> oldCookies;
const char *cookieJar = sb->getBufStart();
int32_t cookieJarLen = sb->length();
const char *lineStartPos = cookieJar;
const char *lineEndPos = NULL;
while ((lineEndPos = (const char*)memchr(lineStartPos, '\n', cookieJarLen - (lineStartPos - cookieJar))) != NULL) {
const char *currentPos = lineStartPos;
const char *tabPos = NULL;
unsigned fieldCount = 0;
httpcookie_t cookie = {};
while (fieldCount < 5 && (tabPos = (const char*)memchr(currentPos, '\t', lineEndPos - currentPos)) != NULL) {
switch (fieldCount) {
case 0:
// domain
cookie.m_domain = currentPos;
cookie.m_domainLen = tabPos - currentPos;
break;
case 1:
// flag
if (memcmp(currentPos, "TRUE", 4) != 0) {
cookie.m_defaultDomain = true;
}
break;
case 2: {
// path
cookie.m_path = currentPos;
cookie.m_pathLen = tabPos - currentPos;
} break;
case 3:
// secure
cookie.m_secure = (memcmp(currentPos, "TRUE", 4) == 0);
break;
case 4:
// expiration
break;
}
currentPos = tabPos + 1;
++fieldCount;
}
cookie.m_cookie = currentPos;
cookie.m_cookieLen = lineEndPos - currentPos;
const char *equalPos = (const char *)memchr(cookie.m_cookie, '=', cookie.m_cookieLen);
if (equalPos) {
cookie.m_nameLen = equalPos - cookie.m_cookie;
oldCookies[std::string(cookie.m_cookie, cookie.m_nameLen)] = cookie;
}
lineStartPos = lineEndPos + 1;
}
// we don't need to care about the last line (we always end on \n)
SafeBuf newCookieJar;
// add old cookies
for (auto &pair : oldCookies) {
if (m_cookies.find(pair.first) == m_cookies.end()) {
addCookie(pair.second, *currentUrl, &newCookieJar);
}
}
// add new cookies
for (auto &pair : m_cookies) {
addCookie(pair.second, *currentUrl, &newCookieJar);
}
newCookieJar.nullTerm();
// replace old with new
sb->reset();
sb->safeMemcpy(&newCookieJar);
sb->nullTerm();
return true;
}