// . returns false if blocked, true otherwise
// . sets g_errno on error
// . make a web page displaying the titleRec of "docId" given via cgi
// . call g_httpServer.sendDynamicPage() to send it
bool sendPageTitledb ( TcpSocket *s , HttpRequest *r ) {
// get the docId from the cgi vars
long long docId = r->getLongLong ("d", 0LL );
// set up a msg22 to get the next titleRec
State4 *st ;
try { st = new (State4); }
catch ( ... ) {
g_errno = ENOMEM;
log("PageTitledb: new(%i): %s",
(int)sizeof(State4),mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));}
mnew ( st , sizeof(State4) , "PageTitledb");
// save the socket
st->m_socket = s;
// copy it
st->m_r.copy ( r );
// remember if http request is internal/local or not
st->m_isRootAdmin = g_conf.isCollAdmin ( s , r );
st->m_isLocal = r->isLocal();
st->m_docId = docId;
// password, too
st->m_pwd = r->getString ( "pwd" );
// get the collection
long collLen = 0;
char *coll = st->m_r.getString("c",&collLen);
if ( ! coll || ! coll[0] ) {
//coll = g_conf.m_defaultColl;
coll = g_conf.getDefaultColl( r->getHost(), r->getHostLen() );
collLen = gbstrlen(coll);
}
st->m_coll = coll;
st->m_collLen = collLen;
// just print page if no docid provided
if ( ! docId ) return gotTitleRec ( st );
// get the handy XmlDoc
XmlDoc *xd = &st->m_xd;
// use 0 for niceness
xd->set3 ( docId , coll , 0 );
// callback
xd->setCallback ( st , gotTitleRec );
// . and tell it to load from old title rec
// . this sets all the member vars from it and also sets
// m_titleRecBuf to contain the actual compressed title rec
if ( ! xd->loadFromOldTitleRec ( ) ) return false;
// we got it without blocking. cached?
return gotTitleRec ( st );
}
// returns false if blocked, true otherwise
bool processLoop ( void *state ) {
// get it
State60 *st = (State60 *)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
// . if it returns false it blocked and will call our callback
// processLoop() when it completes
if ( ! xd->loadFromOldTitleRec ( ) ) return false;
}
if ( g_errno ) return sendErrorReply ( st , g_errno );
// 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) 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 ) return sendErrorReply ( st , EBADENGINEER );
// make it into an editable page now for the turk guy
sendTurkPageReply ( st );
}
// for procog
bool sendPageAnalyze ( TcpSocket *s , HttpRequest *r ) {
// make a state
State8 *st;
try { st = new (State8); }
catch ( ... ) {
g_errno = ENOMEM;
log("PageParser: new(%i): %s",
(int)sizeof(State8),mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,
mstrerror(g_errno));}
mnew ( st , sizeof(State8) , "PageParser" );
st->m_freeIt = true;
st->m_state = NULL;
//st->m_callback = callback;
//st->m_q = q;
//st->m_termFreqs = termFreqs;
//st->m_termFreqWeights = termFreqWeights;
//st->m_affWeights = affWeights;
//st->m_total = (score_t)-1;
st->m_indexCode = 0;
st->m_blocked = false;
st->m_didRootDom = false;
st->m_didRootWWW = false;
st->m_wasRootDom = false;
st->m_u = NULL;
// password, too
long pwdLen = 0;
char *pwd = r->getString ( "pwd" , &pwdLen );
if ( pwdLen > 31 ) pwdLen = 31;
if ( pwdLen > 0 ) strncpy ( st->m_pwd , pwd , pwdLen );
st->m_pwd[pwdLen]='\0';
// save socket ptr
st->m_s = s;
st->m_r.copy ( r );
// get the collection
char *coll = r->getString ( "c" , &st->m_collLen ,NULL /*default*/);
if ( ! coll ) coll = g_conf.m_defaultColl;
if ( ! coll ) coll = "main";
long collLen = gbstrlen(coll);
if ( collLen > MAX_COLL_LEN ) return sendErrorReply ( st , ENOBUFS );
strcpy ( st->m_coll , coll );
// version to use, if -1 use latest
st->m_titleRecVersion = r->getLong("version",-1);
if ( st->m_titleRecVersion == -1 )
st->m_titleRecVersion = TITLEREC_CURRENT_VERSION;
// default to 0 if not provided
st->m_hopCount = r->getLong("hc",0);
long old = r->getLong ( "old", 0 );
// set query
long qlen;
char *qs = r->getString("q",&qlen,NULL);
if ( qs ) st->m_tq.set2 ( qs , langUnknown , true );
// url will override docid if given
st->m_docId = r->getLongLong ("d",-1);
st->m_docId = r->getLongLong ("docid",st->m_docId);
long ulen;
char *u = st->m_r.getString("u",&ulen,NULL);
if ( ! u ) u = st->m_r.getString("url",&ulen,NULL);
if ( ! u && st->m_docId == -1LL )
return sendErrorReply ( st , EBADREQUEST );
// set url in state class (may have length 0)
//if ( u ) st->m_url.set ( u , ulen );
//st->m_urlLen = ulen;
st->m_u = u;
st->m_ulen = 0;
if ( u ) st->m_ulen = gbstrlen(u);
// should we recycle link info?
st->m_recycle = r->getLong("recycle",1);
st->m_recycle2 = r->getLong("recycleimp",0);
st->m_render = r->getLong("render" ,0);
st->m_recompute = r->getLong("recompute" ,0);
// for quality computation... takes way longer cuz we have to
// lookup the IP address of every outlink, so we can get its root
// quality using Msg25 which needs to filter out voters from that IP
// range.
st->m_oips = r->getLong("oips" ,0);
//st->m_page = r->getLong("page",1);
long linkInfoLen = 0;
// default is NULL
char *linkInfoColl = r->getString ( "oli" , &linkInfoLen, NULL );
if ( linkInfoColl ) strcpy ( st->m_linkInfoColl , linkInfoColl );
else st->m_linkInfoColl[0] = '\0';
// set the flag in our SafeBuf class so that Words.cpp knows to show
// html or html source depending on this value
//st->m_xbuf.m_renderHtml = st->m_render;
// should we use the old title rec?
st->m_old = old;
// are we coming from a local machine?
st->m_isLocal = r->isLocal();
//no more setting the default root quality to 30, instead if we do not
// know it setting it to -1
st->m_rootQuality=-1;
// header
//xbuf->safePrintf("<meta http-equiv=\"Content-Type\" "
// "content=\"text/html; charset=utf-8\">\n");
XmlDoc *xd = &st->m_xd;
long isXml = r->getLong("xml",0);
// if got docid, use that
if ( st->m_docId != -1 ) {
if ( ! xd->set3 ( st->m_docId,
st->m_coll,
0 ) ) // niceness
// return error reply if g_errno is set
return sendErrorReply ( st , g_errno );
// make this our callback in case something blocks
xd->setCallback ( st , gotXmlDoc );
xd->m_pbuf = &st->m_wbuf;
// reset this flag
st->m_donePrinting = false;
// . set xd from the old title rec if recycle is true
// . can also use XmlDoc::m_loadFromOldTitleRec flag
//if ( st->m_recycle ) xd->m_recycleContent = true;
xd->m_recycleContent = true;
// force this on
//xd->m_useSiteLinkBuf = true;
//xd->m_usePageLinkBuf = true;
if ( isXml ) xd->m_printInXml = true;
// now tell it to fetch the old title rec
if ( ! xd->loadFromOldTitleRec () )
// return false if this blocks
return false;
return gotXmlDoc ( st );
}
// set this up
SpiderRequest sreq;
sreq.reset();
if ( st->m_u ) strcpy(sreq.m_url,st->m_u);
long firstIp = hash32n(st->m_u);
if ( firstIp == -1 || firstIp == 0 ) firstIp = 1;
// parentdocid of 0
sreq.setKey( firstIp, 0LL, false );
sreq.m_isPageParser = 1;
sreq.m_hopCount = st->m_hopCount;
sreq.m_hopCountValid = 1;
sreq.m_fakeFirstIp = 1;
sreq.m_firstIp = firstIp;
Url nu;
nu.set(sreq.m_url);
sreq.m_domHash32 = nu.getDomainHash32();
sreq.m_siteHash32 = nu.getHostHash32();
// . get provided content if any
// . will be NULL if none provided
// . "content" may contain a MIME
long contentLen = 0;
char *content = r->getString ( "content" , &contentLen , NULL );
// is the "content" url-encoded? default is true.
bool contentIsEncoded = true;
// mark doesn't like to url-encode his content
if ( ! content ) {
content = r->getUnencodedContent ();
contentLen = r->getUnencodedContentLen ();
contentIsEncoded = false;
}
// ensure null
if ( contentLen == 0 ) content = NULL;
//uint8_t contentType = CT_HTML;
//if ( isXml ) contentType = CT_XML;
long ctype = r->getLong("ctype",CT_HTML);
// . use the enormous power of our new XmlDoc class
// . this returns false if blocked
if ( ! xd->set4 ( &sreq ,
NULL ,
st->m_coll ,
// we need this so the term table is set!
&st->m_wbuf , // XmlDoc::m_pbuf
0, // try 0 now! 1 ,//PP_NICENESS ))
content ,
false, // deletefromindex
0, // forced ip
ctype ))
// return error reply if g_errno is set
return sendErrorReply ( st , g_errno );
// make this our callback in case something blocks
xd->setCallback ( st , gotXmlDoc );
// reset this flag
st->m_donePrinting = false;
// prevent a core here in the event we download the page content
xd->m_crawlDelayValid = true;
xd->m_crawlDelay = 0;
// . set xd from the old title rec if recycle is true
// . can also use XmlDoc::m_loadFromOldTitleRec flag
//if ( st->m_recycle ) xd->m_recycleContent = true;
// only recycle if docid is given!!
if ( st->m_recycle ) xd->m_recycleContent = true;
// force this on
//xd->m_useSiteLinkBuf = true;
//xd->m_usePageLinkBuf = true;
if ( isXml ) xd->m_printInXml = true;
return gotXmlDoc ( st );
}
// . a new interface so Msg3b can call this with "s" set to NULL
// . returns false if blocked, true otherwise
// . sets g_errno on error
bool sendPageParser2 ( TcpSocket *s ,
HttpRequest *r ,
State8 *st ,
long long docId ,
Query *q ,
// in query term space, not imap space
long long *termFreqs ,
// in imap space
float *termFreqWeights ,
// in imap space
float *affWeights ,
void *state ,
void (* callback)(void *state) ) {
//log("parser: read sock=%li",s->m_sd);
// might a simple request to addsomething to validated.*.txt file
// from XmlDoc::print() or XmlDoc::validateOutput()
char *add = r->getString("add",NULL);
//long long uh64 = r->getLongLong("uh64",0LL);
char *uh64str = r->getString("uh64",NULL);
//char *divTag = r->getString("div",NULL);
if ( uh64str ) {
// convert add to number
long addNum = 0;
if ( to_lower_a(add[0])=='t' ) // "true" or "false"?
addNum = 1;
// convert it. skip beginning "str" inserted to prevent
// javascript from messing with the long long since it
// was rounding it!
//long long uh64 = atoll(uh64str);//+3);
// urldecode that
//long divTagLen = gbstrlen(divTag);
//long newLen = urlDecode ( divTag , divTag , divTagLen );
// null term?
//divTag[newLen] = '\0';
// do it. this is defined in XmlDoc.cpp
//addCheckboxSpan ( uh64 , divTag , addNum );
// make basic reply
char *reply;
reply = "HTTP/1.0 200 OK\r\n"
"Connection: Close\r\n";
// that is it! send a basic reply ok
bool status = g_httpServer.sendDynamicPage( s ,
reply,
gbstrlen(reply),
-1, //cachtime
false ,//postreply?
NULL, //ctype
-1 , //httpstatus
NULL,//cookie
"utf-8");
return status;
}
// make a state
if ( st ) st->m_freeIt = false;
if ( ! st ) {
try { st = new (State8); }
catch ( ... ) {
g_errno = ENOMEM;
log("PageParser: new(%i): %s",
(int)sizeof(State8),mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,
mstrerror(g_errno));}
mnew ( st , sizeof(State8) , "PageParser" );
st->m_freeIt = true;
}
// msg3b uses this to get a score from the query
st->m_state = state;
st->m_callback = callback;
st->m_q = q;
st->m_termFreqs = termFreqs;
st->m_termFreqWeights = termFreqWeights;
st->m_affWeights = affWeights;
//st->m_total = (score_t)-1;
st->m_indexCode = 0;
st->m_blocked = false;
st->m_didRootDom = false;
st->m_didRootWWW = false;
st->m_wasRootDom = false;
st->m_u = NULL;
st->m_recompute = false;
//st->m_url.reset();
// do not allow more than one to be launched at a time if in
// a quickpoll. will cause quickpoll in quickpoll.
g_inPageParser = true;
// password, too
long pwdLen = 0;
char *pwd = r->getString ( "pwd" , &pwdLen );
if ( pwdLen > 31 ) pwdLen = 31;
if ( pwdLen > 0 ) strncpy ( st->m_pwd , pwd , pwdLen );
st->m_pwd[pwdLen]='\0';
// save socket ptr
st->m_s = s;
st->m_r.copy ( r );
// get the collection
char *coll = r->getString ( "c" , &st->m_collLen ,NULL /*default*/);
if ( st->m_collLen > MAX_COLL_LEN )
return sendErrorReply ( st , ENOBUFS );
if ( ! coll )
return sendErrorReply ( st , ENOCOLLREC );
strcpy ( st->m_coll , coll );
// version to use, if -1 use latest
st->m_titleRecVersion = r->getLong("version",-1);
if ( st->m_titleRecVersion == -1 )
st->m_titleRecVersion = TITLEREC_CURRENT_VERSION;
// default to 0 if not provided
st->m_hopCount = r->getLong("hc",0);
//long ulen = 0;
//char *u = r->getString ( "u" , &ulen , NULL /*default*/);
long old = r->getLong ( "old", 0 );
// set query
long qlen;
char *qs = r->getString("q",&qlen,NULL);
if ( qs ) st->m_tq.set2 ( qs , langUnknown , true );
// url will override docid if given
if ( ! st->m_u || ! st->m_u[0] )
st->m_docId = r->getLongLong ("docid",-1);
else
st->m_docId = -1;
// set url in state class (may have length 0)
//if ( u ) st->m_url.set ( u , ulen );
//st->m_urlLen = ulen;
st->m_u = st->m_r.getString("u",&st->m_ulen,NULL);
// should we recycle link info?
st->m_recycle = r->getLong("recycle",0);
st->m_recycle2 = r->getLong("recycleimp",0);
st->m_render = r->getLong("render" ,0);
// for quality computation... takes way longer cuz we have to
// lookup the IP address of every outlink, so we can get its root
// quality using Msg25 which needs to filter out voters from that IP
// range.
st->m_oips = r->getLong("oips" ,0);
long linkInfoLen = 0;
// default is NULL
char *linkInfoColl = r->getString ( "oli" , &linkInfoLen, NULL );
if ( linkInfoColl ) strcpy ( st->m_linkInfoColl , linkInfoColl );
else st->m_linkInfoColl[0] = '\0';
// set the flag in our SafeBuf class so that Words.cpp knows to show
// html or html source depending on this value
st->m_xbuf.m_renderHtml = st->m_render;
// should we use the old title rec?
st->m_old = old;
// are we coming from a local machine?
st->m_isLocal = r->isLocal();
//no more setting the default root quality to 30, instead if we do not
// know it setting it to -1
st->m_rootQuality=-1;
// header
SafeBuf *xbuf = &st->m_xbuf;
xbuf->safePrintf("<meta http-equiv=\"Content-Type\" "
"content=\"text/html; charset=utf-8\">\n");
// print standard header
g_pages.printAdminTop ( xbuf , st->m_s , &st->m_r );
// print the standard header for admin pages
char *dd = "";
char *rr = "";
char *rr2 = "";
char *render = "";
char *oips = "";
char *us = "";
if ( st->m_u && st->m_u[0] ) us = st->m_u;
//if ( st->m_sfn != -1 ) sprintf ( rtu , "%li",st->m_sfn );
if ( st->m_old ) dd = " checked";
if ( st->m_recycle ) rr = " checked";
if ( st->m_recycle2 ) rr2 = " checked";
if ( st->m_render ) render = " checked";
if ( st->m_oips ) oips = " checked";
xbuf->safePrintf(
"<style>"
".poo { background-color:#%s;}\n"
"</style>\n" ,
LIGHT_BLUE );
long clen;
char *contentParm = r->getString("content",&clen,"");
// print the input form
xbuf->safePrintf (
"<style>\n"
"h2{font-size: 12px; color: #666666;}\n"
".gbtag { border: 1px solid gray;"
"background: #ffffef;display:inline;}\n"
".gbcomment { border: 1px solid gray;"
"color: #888888; font-style:italic; "
"background: #ffffef;display:inline;}\n"
".token { border: 1px solid gray;"
"background: #f0ffff;display:inline;}\n"
".spam { border: 1px solid gray;"
"background: #af0000;"
"color: #ffffa0;}"
".hs {color: #009900;}"
"</style>\n"
"<center>"
"<table %s>"
"<tr><td colspan=5><center><b>"
"Parser"
"</b></center></td></tr>\n"
"<tr class=poo>"
"<td>"
"<b>url</b>"
"<br><font size=-2>"
"Type in <b>FULL</b> url to parse."
"</font>"
"</td>"
"</td>"
"<td>"
"<input type=text name=u value=\"%s\" size=\"40\">\n"
"</td>"
"</tr>"
/*
"<tr class=poo>"
"<td>"
"Parser version to use: "
"</td>"
"<td>"
"<input type=text name=\"version\" size=\"4\" value=\"-1\"> "
"</td>"
"<td>"
"(-1 means to use latest title rec version)<br>"
"</td>"
"</tr>"
*/
/*
"<tr class=poo>"
"<td>"
"Hop count to use: "
"</td>"
"<td>"
"<input type=text name=\"hc\" size=\"4\" value=\"%li\"> "
"</td>"
"<td>"
"(-1 is unknown. For root urls hopcount is always 0)<br>"
"</td>"
"</tr>"
*/
"<tr class=poo>"
"<td>"
"<b>use cached</b>"
"<br><font size=-2>"
"Load page from cache (titledb)?"
"</font>"
"</td>"
"<td>"
"<input type=checkbox name=old value=1%s> "
"</td>"
"</tr>"
/*
"<tr class=poo>"
"<td>"
"Reparse root:"
"</td>"
"<td>"
"<input type=checkbox name=artr value=1%s> "
"</td>"
"<td>"
"Apply selected ruleset to root to update quality"
"</td>"
"</tr>"
*/
"<tr class=poo>"
"<td>"
"<b>recycle link info</b>"
"<br><font size=-2>"
"Recycle the link info from the title rec"
"Load page from cache (titledb)?"
"</font>"
"</td>"
"<td>"
"<input type=checkbox name=recycle value=1%s> "
"</td>"
"</tr>"
/*
"<tr class=poo>"
"<td>"
"Recycle Link Info Imported:"
"</td>"
"<td>"
"<input type=checkbox name=recycleimp value=1%s> "
"</td>"
"<td>"
"Recycle the link info imported from other coll"
"</td>"
"</tr>"
*/
"<tr class=poo>"
"<td>"
"<b>render html</b>"
"<br><font size=-2>"
"Render document content as HTML"
"</font>"
"</td>"
"<td>"
"<input type=checkbox name=render value=1%s> "
"</td>"
"</tr>"
/*
"<tr class=poo>"
"<td>"
"Lookup outlinks' ruleset, ips, quality:"
"</td>"
"<td>"
"<input type=checkbox name=oips value=1%s> "
"</td>"
"<td>"
"To compute quality lookup IP addresses of roots "
"of outlinks."
"</td>"
"</tr>"
"<tr class=poo>"
"<td>"
"LinkInfo Coll:"
"</td>"
"<td>"
"<input type=text name=\"oli\" size=\"10\" value=\"\"> "
"</td>"
"<td>"
"Leave empty usually. Uses this coll to lookup link info."
"</td>"
"</tr>"
*/
"<tr class=poo>"
"<td>"
"<b>optional query</b>"
"<br><font size=-2>"
"Leave empty usually. For title generation only."
"</font>"
"</td>"
"<td>"
"<input type=text name=\"q\" size=\"20\" value=\"\"> "
"</td>"
"</tr>",
TABLE_STYLE,
us ,
dd,
rr,
render
);
xbuf->safePrintf(
"<tr class=poo>"
"<td>"
"<b>content type below is</b>"
"<br><font size=-2>"
"Is the content below HTML? XML? JSON?"
"</font>"
"</td>"
"<td>"
//"<input type=checkbox name=xml value=1> "
"<select name=ctype>\n"
"<option value=%li selected>HTML</option>\n"
"<option value=%li selected>XML</option>\n"
"<option value=%li selected>JSON</option>\n"
"</select>\n"
"</td>"
"</tr>",
(long)CT_HTML,
(long)CT_XML,
(long)CT_JSON
);
xbuf->safePrintf(
"<tr class=poo>"
"<td><b>content</b>"
"<br><font size=-2>"
"Use this content for the provided <i>url</i> "
"rather than downloading it from the web."
"</td>"
"<td>"
"<textarea rows=10 cols=80 name=content>"
"%s"
"</textarea>"
"</td>"
"</tr>"
"</table>"
"</center>"
"</form>"
"<br>",
//oips ,
contentParm );
xbuf->safePrintf(
"<center>"
"<input type=submit value=Submit>"
"</center>"
);
// just print the page if no url given
if ( ! st->m_u || ! st->m_u[0] ) return processLoop ( st );
XmlDoc *xd = &st->m_xd;
// set this up
SpiderRequest sreq;
sreq.reset();
strcpy(sreq.m_url,st->m_u);
long firstIp = hash32n(st->m_u);
if ( firstIp == -1 || firstIp == 0 ) firstIp = 1;
// parentdocid of 0
sreq.setKey( firstIp, 0LL, false );
sreq.m_isPageParser = 1;
sreq.m_hopCount = st->m_hopCount;
sreq.m_hopCountValid = 1;
sreq.m_fakeFirstIp = 1;
sreq.m_firstIp = firstIp;
Url nu;
nu.set(sreq.m_url);
sreq.m_domHash32 = nu.getDomainHash32();
sreq.m_siteHash32 = nu.getHostHash32();
// . get provided content if any
// . will be NULL if none provided
// . "content" may contain a MIME
long contentLen = 0;
char *content = r->getString ( "content" , &contentLen , NULL );
// is the "content" url-encoded? default is true.
bool contentIsEncoded = true;
// mark doesn't like to url-encode his content
if ( ! content ) {
content = r->getUnencodedContent ();
contentLen = r->getUnencodedContentLen ();
contentIsEncoded = false;
}
// ensure null
if ( contentLen == 0 ) content = NULL;
uint8_t contentType = CT_HTML;
if ( r->getBool("xml",0) ) contentType = CT_XML;
contentType = r->getLong("ctype",contentType);//CT_HTML);
// if facebook, load xml content from title rec...
bool isFacebook = (bool)strstr(st->m_u,"http://www.facebook.com/");
if ( isFacebook && ! content ) {
long long docId = g_titledb.getProbableDocId(st->m_u);
sprintf(sreq.m_url ,"%llu", docId );
sreq.m_isPageReindex = true;
}
// hack
if ( content ) {
st->m_dbuf.purge();
st->m_dbuf.safeStrcpy(content);
//char *data = strstr(content,"\r\n\r\n");
//long dataPos = 0;
//if ( data ) dataPos = (data + 4) - content;
//st->m_dbuf.convertJSONtoXML(0,dataPos);
//st->m_dbuf.decodeJSON(0);
content = st->m_dbuf.getBufStart();
}
// . use the enormous power of our new XmlDoc class
// . this returns false if blocked
if ( ! xd->set4 ( &sreq ,
NULL ,
st->m_coll ,
&st->m_wbuf ,
0 ,//PP_NICENESS ))
content ,
false, // deletefromindex
0, // forced ip
contentType ))
// return error reply if g_errno is set
return sendErrorReply ( st , g_errno );
// make this our callback in case something blocks
xd->setCallback ( st , processLoop );
// . set xd from the old title rec if recycle is true
// . can also use XmlDoc::m_loadFromOldTitleRec flag
if ( st->m_recycle ) xd->m_recycleContent = true;
return processLoop ( st );
}
// . returns false if blocked, true otherwise
// . sets g_errno on error
bool sendPageGet ( TcpSocket *s , HttpRequest *r ) {
// get the collection
long collLen = 0;
char *coll = r->getString("c",&collLen);
if ( ! coll || ! coll[0] ) {
//coll = g_conf.m_defaultColl;
coll = g_conf.getDefaultColl( r->getHost(), r->getHostLen() );
collLen = gbstrlen(coll);
}
// ensure collection not too big
if ( collLen >= MAX_COLL_LEN ) {
g_errno = ECOLLTOOBIG;
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));
}
// get the collection rec
CollectionRec *cr = g_collectiondb.getRec ( coll );
if ( ! cr ) {
g_errno = ENOCOLLREC;
log("query: Archived copy retrieval failed. "
"No collection record found for "
"collection \"%s\".",coll);
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));
}
// does this collection ban this IP?
if ( ! cr->hasSearchPermission ( s ) ) {
g_errno = ENOPERM;
//log("PageGet::sendDynamicReply0: permission denied for %s",
// iptoa(s->m_ip) );
g_msg = " (error: permission denied)";
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));
}
// . get fields from cgi field of the requested url
// . get the search query
long qlen = 0;
char *q = r->getString ( "q" , &qlen , NULL /*default*/);
// ensure query not too big
if ( qlen >= MAX_QUERY_LEN-1 ) {
g_errno=EQUERYTOOBIG;
return g_httpServer.sendErrorReply (s,500 ,mstrerror(g_errno));
}
// the docId
long long docId = r->getLongLong ( "d" , 0LL /*default*/ );
// get url
char *url = r->getString ( "u",NULL);
if ( docId == 0 && ! url ) {
g_errno = EMISSINGINPUT;
return g_httpServer.sendErrorReply (s,500 ,mstrerror(g_errno));
}
// . should we do a sequential lookup?
// . we need to match summary here so we need to know this
//bool seq = r->getLong ( "seq" , false );
// restrict to root file?
bool rtq = r->getLong ( "rtq" , false );
// . get the titleRec
// . TODO: redirect client to a better http server to save bandwidth
State2 *st ;
try { st = new (State2); }
catch (... ) {
g_errno = ENOMEM;
log("PageGet: new(%i): %s",
(int)sizeof(State2),mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));}
mnew ( st , sizeof(State2) , "PageGet1" );
// save the socket and if Host: is local in the Http request Mime
st->m_socket = s;
st->m_isAdmin = g_conf.isCollAdmin ( s , r );
st->m_isLocal = r->isLocal();
st->m_docId = docId;
st->m_printed = false;
// include header ... "this page cached by Gigablast on..."
st->m_includeHeader = r->getLong ("ih" , true );
st->m_includeBaseHref = r->getLong ("ibh" , false );
st->m_queryHighlighting = r->getLong ("qh" , true );
st->m_strip = r->getLong ("strip" , 0 );
st->m_clickAndScroll = r->getLong ("cas" , true );
st->m_cnsPage = r->getLong ("cnsp" , true );
char *langAbbr = r->getString("qlang",NULL);
st->m_langId = langUnknown;
if ( langAbbr ) {
uint8_t langId = getLangIdFromAbbr ( langAbbr );
st->m_langId = langId;
}
strncpy ( st->m_coll , coll , MAX_COLL_LEN+1 );
// store query for query highlighting
st->m_netTestResults = r->getLong ("rnettest", false );
//if( st->m_netTestResults ) {
// mdelete ( st , sizeof(State2) , "PageGet1" );
// delete ( st );
// return sendPageNetResult( s );
//}
if ( q && qlen > 0 ) strcpy ( st->m_q , q );
else st->m_q[0] = '\0';
st->m_qlen = qlen;
//st->m_seq = seq;
st->m_rtq = rtq;
st->m_boolFlag = r->getLong ("bq", 2 /*default is 2*/ );
st->m_isBanned = false;
st->m_noArchive = false;
st->m_socket = s;
st->m_format = r->getReplyFormat();
// default to 0 niceness
st->m_niceness = 0;
st->m_r.copy ( r );
//st->m_cr = cr;
st->m_printDisclaimer = true;
if ( st->m_cnsPage )
st->m_printDisclaimer = false;
if ( st->m_strip ) // ! st->m_evbits.isEmpty() )
st->m_printDisclaimer = false;
// should we cache it?
char useCache = r->getLong ( "usecache" , 1 );
char rcache = r->getLong ( "rcache" , 1 );
char wcache = r->getLong ( "wcache" , 1 );
long cacheAge = r->getLong ( "cacheAge" , 60*60 ); // default one hour
if ( useCache == 0 ) { cacheAge = 0; wcache = 0; }
if ( rcache == 0 ) cacheAge = 0;
// . fetch the TitleRec
// . a max cache age of 0 means not to read from the cache
XmlDoc *xd = &st->m_xd;
// url based?
if ( url ) {
SpiderRequest sreq;
sreq.reset();
strcpy(sreq.m_url, url );
sreq.setDataSize();
// this returns false if "coll" is invalid
if ( ! xd->set4 ( &sreq , NULL , coll , NULL , st->m_niceness ) )
goto hadSetError;
}
// . when getTitleRec() is called it will load the old one
// since XmlDoc::m_setFromTitleRec will be true
// . niceness is 0
// . use st->m_coll since XmlDoc just points to it!
// . this returns false if "coll" is invalid
else if ( ! xd->set3 ( docId , st->m_coll , 0 ) ) {
hadSetError:
mdelete ( st , sizeof(State2) , "PageGet1" );
delete ( st );
g_errno = ENOMEM;
log("PageGet: set3: %s", mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));
}
// if it blocks while it loads title rec, it will re-call this routine
xd->setCallback ( st , processLoopWrapper );
// good to go!
return processLoop ( st );
}
// 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_isAdmin && *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_isAdmin && ! *vi ) return sendErrorReply (st,EDOCBANNED);
*/
// get the utf8 content
char **utf8 = xd->getUtf8Content();
//long 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 %li is bad",g_hostdb.m_hostId);
return sendErrorReply(st,EBADENGINEER );
}
char *content = xd->ptr_utf8Content;
long contentLen = xd->size_utf8Content - 1;
// shortcut
char strip = st->m_strip;
// alloc buffer now
//char *buf = NULL;
//long 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;
long 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();
// for undoing the stuff below
long startLen2 = sb->length();//p;
// query should be NULL terminated
char *q = st->m_q;
long qlen = st->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 ( 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=%li&"
"d=%lli&strip=1\""
" style=\"%s\">"
"[stripped]</a>",
q , st->m_coll ,
(long)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?
//long 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
//unsigned long ip =g_conf.m_mainExternalIp ; // h->m_externalIp;
//unsigned short port=g_conf.m_mainExternalPort;//h->m_externalHttpPort
// local check
//if ( st->m_isLocal ) {
unsigned long ip = h->m_ip;
unsigned short port = h->m_httpPort;
//}
//sprintf ( x , "http://%s:%li/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
long elen = urlEncode ( x , thisUrlEnd - x , q , qlen );
x += elen;
// separate cgi vars with a &
//sprintf ( x, "&seq=%li&rtq=%lid=%lli",
// (long)st->m_seq,(long)st->m_rtq,st->m_msg22.getDocId());
sprintf ( x, "&d=%lli",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 ( long 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 );
long 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
//memcpy ( 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 ( 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>%llu</docId>\n",xd->m_docId);
sb->safePrintf("\t<cachedTimeUTC>%lu</cachedTimeUTC>\n",
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\":%llu,\n",xd->m_docId);
sb->safePrintf("\t\"cachedTimeUTC\":%lu,\n",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;
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>" );
long 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=%lli&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> "
);
long tlen = titleEnd - titleStart;
sb->safeMemcpy ( titleStart , tlen );
sb->safePrintf ( "</span></td></tr>" );
}
sb->safePrintf( "</table><br>\n" );
}
// is the content preformatted?
bool pre = false;
char ctype = (char)xd->m_contentType;
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,
(long)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;
SafeBuf tmp;
SafeBuf *xb = sb;
if ( format == FORMAT_XML ) xb = &tmp;
if ( format == FORMAT_JSON ) xb = &tmp;
if ( ! queryHighlighting ) {
xb->safeMemcpy ( content , contentLen );
//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
//long bufLen = p - buf;
long 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 ( 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 Msg7::inject ( char *url ,
long forcedIp ,
char *content ,
long contentLen ,
bool recycleContent,
uint8_t contentType,
char *coll ,
bool quickReply ,
char *username ,
char *pwd ,
long niceness,
void *state ,
void (*callback)(void *state),
long firstIndexed,
long lastSpidered,
long hopCount,
char newOnly,
short charset,
char spiderLinks,
char deleteIt,
char hasMime,
bool doConsistencyTesting
) {
m_quickReply = quickReply;
// store coll
if ( ! coll ) { g_errno = ENOCOLLREC; return true; }
long collLen = gbstrlen ( coll );
if ( collLen > MAX_COLL_LEN ) collLen = MAX_COLL_LEN;
strncpy ( m_coll , coll , collLen );
m_coll [ collLen ] = '\0';
// store user
//long ulen = 0;
//if ( username ) ulen = gbstrlen(username);
//if ( ulen >= MAX_USER_SIZE-1 ) {g_errno = EBUFOVERFLOW; return true;}
//if ( username ) strcpy( m_username, username );
// store password
//long pwdLen = 0;
//if ( pwd ) pwdLen = gbstrlen(pwd);
//m_pwd [ 0 ] ='\0';
//if ( pwdLen > 31 ) pwdLen = 31;
//if ( pwdLen > 0 ) strncpy ( m_pwd , pwd , pwdLen );
//m_pwd [ pwdLen ] = '\0';
// store url
if ( ! url ) { g_errno = 0; return true; }
long urlLen = gbstrlen(url);
if ( urlLen > MAX_URL_LEN ) {g_errno = EBADENGINEER; return true; }
// skip injecting if no url given! just print the admin page.
if ( urlLen <= 0 ) return true;
//strcpy ( m_url , url );
if ( g_repairMode ) { g_errno = EREPAIRING; return true; }
// send template reply if no content supplied
if ( ! content && ! recycleContent ) {
log("inject: no content supplied to inject command and "
"recycleContent is false.");
//return true;
}
// clean url?
// normalize and add www. if it needs it
Url uu;
uu.set ( url , gbstrlen(url) , true );
// remove >'s i guess and store in st1->m_url[] buffer
char cleanUrl[MAX_URL_LEN+1];
urlLen = cleanInput ( cleanUrl,
MAX_URL_LEN,
uu.getUrl(),
uu.getUrlLen() );
// this can go on the stack since set4() copies it
SpiderRequest sreq;
sreq.reset();
strcpy(sreq.m_url, cleanUrl );
// parentdocid of 0
long firstIp = hash32n(cleanUrl);
if ( firstIp == -1 || firstIp == 0 ) firstIp = 1;
sreq.setKey( firstIp,0LL, false );
sreq.m_isInjecting = 1;
sreq.m_isPageInject = 1;
sreq.m_hopCount = hopCount;
sreq.m_hopCountValid = 1;
sreq.m_fakeFirstIp = 1;
sreq.m_firstIp = firstIp;
// shortcut
XmlDoc *xd = &m_xd;
// log it now
//log("inject: injecting doc %s",cleanUrl);
static char s_dummy[3];
// sometims the content is indeed NULL...
if ( newOnly && ! content ) {
// don't let it be NULL because then xmldoc will
// try to download the page!
s_dummy[0] = '\0';
content = s_dummy;
//char *xx=NULL;*xx=0; }
}
// . use the enormous power of our new XmlDoc class
// . this returns false with g_errno set on error
if ( //m_needsSet &&
! xd->set4 ( &sreq ,
NULL ,
m_coll ,
NULL , // pbuf
// give it a niceness of 1, we have to be
// careful since we are a niceness of 0!!!!
niceness, // 1 ,
// inject this content
content ,
deleteIt, // false, // deleteFromIndex ,
forcedIp ,
contentType ,
lastSpidered ,
hasMime )) {
// g_errno should be set if that returned false
if ( ! g_errno ) { char *xx=NULL;*xx=0; }
return true;
}
// do not re-call the set
//m_needsSet = false;
// make this our callback in case something blocks
xd->setCallback ( state , callback );
xd->m_doConsistencyTesting = doConsistencyTesting;
// . set xd from the old title rec if recycle is true
// . can also use XmlDoc::m_loadFromOldTitleRec flag
if ( recycleContent ) xd->m_recycleContent = true;
// othercrap
if ( firstIndexed ) {
xd->m_firstIndexedDate = firstIndexed;
xd->m_firstIndexedDateValid = true;
}
if ( lastSpidered ) {
xd->m_spideredTime = lastSpidered;
xd->m_spideredTimeValid = true;
}
if ( hopCount != -1 ) {
xd->m_hopCount = hopCount;
xd->m_hopCountValid = true;
}
if ( charset != -1 && charset != csUnknown ) {
xd->m_charset = charset;
xd->m_charsetValid = true;
}
// avoid looking up ip of each outlink to add "firstip" tag to tagdb
// because that can be slow!!!!!!!
xd->m_spiderLinks = spiderLinks;
xd->m_spiderLinks2 = spiderLinks;
xd->m_spiderLinksValid = true;
// . newOnly is true --> do not inject if document is already indexed!
// . maybe just set indexCode
xd->m_newOnly = newOnly;
// do not re-lookup the robots.txt
xd->m_isAllowed = true;
xd->m_isAllowedValid = true;
xd->m_crawlDelay = -1; // unknown
xd->m_crawlDelayValid = true;
// set this now
g_inPageInject = true;
// log it now
//log("inject: indexing injected doc %s",cleanUrl);
// . now tell it to index
// . this returns false if blocked
bool status = xd->indexDoc ( );
// log it. i guess only for errors when it does not block?
// because xmldoc.cpp::indexDoc calls logIt()
if ( status ) xd->logIt();
// undo it
g_inPageInject = false;
// note that it blocked
//if ( ! status ) log("inject: blocked for %s",cleanUrl);
// return false if it blocked
return status;
}
bool sendTurkPageReply ( State60 *st ) {
XmlDoc *xd = &st->m_xd;
//char *content = xd->ptr_utf8Content;
//int32_t contentLen = xd->size_utf8Content - 1;
// count the total number of EventDesc classes for all evids
//char *evd = xd->ptr_eventData;
//EventDisplay *ed = (EventDisplay *)evd;
//char *addr = evd + (int32_t)ed->m_addr;
//char timeZoneOffset = getTimeZoneFromAddr ( addr );
// in case getSections() block come right back in
xd->setCallback ( st , xdcallback );
// . set niceness to 1 so all this processing doesn't slow queries down
// . however, g_niceness should still be zero... hmmm...
xd->m_niceness = 1;
// default to 1 niceness
st->m_niceness = 1;
// now set the sections class
Sections *ss = xd->getSections();
// now for each section with alnum text, telescope up as far as
// possible without containing anymore alnum text than what it
// contained. set SEC_CONTROL bit. such sections will have the
// 2 green/blue dots, that are used for turning on/off title/desc.
// but really the indians will only turn off sections that should
// not have a title/desc.
for ( Section *si = ss->m_rootSection ; si ; si = si->m_next ) {
// breathe
QUICKPOLL(st->m_niceness);
// skip if does not have text
if ( si->m_firstWordPos < 0 ) continue;
// otherwise, find biggest parent that contains just that text
Section *p = si->m_parent;
Section *last = si;
for ( ; p ; p = p->m_parent ) {
if ( p->m_firstWordPos != si->m_firstWordPos ) break;
if ( p->m_lastWordPos != si->m_lastWordPos ) break;
last = p;
}
// set that bit then
last->m_flags |= SEC_CONTROL;
// and speed up the loop
si = last;
}
// * now each SEC_CONTROL sections have a fence activated by a turker
// * an event title or description can not span a fence. it must be
// confined within a fence. however, it is allowed to include
// title or description from a "title section".
// * hold shift down to designate as title section when clicking it
// * show the raw text of each event changing as you fence
// sections in or out. show in a right frame.
// * show list of events on page in the top frame. can toggle them
// all individually.
// * and remove no-display from all tags so we can see everything.
// * highlight addresses, not just dates.
// * each section hash has its own unique bg color when activated
// * with a single click, completely reject an event because:
// contains bad time, address, title or desc. specify which so
// we can improve our algo.
// * when selecting an individual event, scroll to its tod...
// * remove all color from webpage that we can so our colors show up
// * remove all imgs. just src them to dev null.
// * allow for entering a custom title for an event or all events
// that are or will ever appear on the page.
// * when displaying the text of the events, use hyphens to
// delineate the section topology. strike out text as a section
// fence is activated.
// * when a section is activated is it easier to just redownload
// the whole text of the page? maybe just the text frame?
// * clicking on an individual sentence section should just remove
// that sentence. that is kinda a special content hash removal
// tag. like "Click here for video."
// * when an event id is selected i guess activate its bgcolor to
// be light blue for all sentences currently in the event that
// are not in activated sections. (make exception for designated
// title sections). so we need multiple tags for each events
// sentence div section. if sentence is split use multiple div tags
// then to keep the order. so each event sentence would have
// <div ev1=1 ev2=1 ev10=1>...</div> if it is in event ids 1,2 and
// 10. that way we can activate it when one of those event ids is
// activated.
SafeBuf sb;
// int16_tcuts
if ( ! xd->m_wordsValid ) { char *xx=NULL;*xx=0; }
Words *words = &xd->m_words;
int32_t nw = words->getNumWords();
char **wptrs = words->getWords();
int32_t *wlens = words->getWordLens();
nodeid_t *tids = words->getTagIds();
// a special array for printing </div> tags
char *endCounts = (char *)mcalloc ( nw ,"endcounts");
if ( ! endCounts ) return sendErrorReply ( st , g_errno );
//
// now loop over all the words. if word starts a section that has
// SEC_CONTROL bit set, and print out the section hash and a color
// tag to be activated if the turkey activates us.
// CAUTION: word may start multiple sections.
//
for ( int32_t i = 0 ; i < nw ; i++ ) {
// get section ptr
Section *sj = ss->m_sectionPtrs[i];
// sanity check. sj must be first section ptr that starts @ a
if ( sj && sj->m_a==i && sj->m_prev && sj->m_prev->m_a==i ) {
char *xx=NULL;*xx=0; }
// . does word #i start a section?
// . if section is control, print out the control
while ( sj && sj->m_a == i ) {
// print this section's hash
if ( sj->m_flags & SEC_CONTROL) {
// after the turkeys have made all the edits
// they need to submit the changes they made.
// how can we get that data sent back to the
// back end? we need to send back the colors
// of the sections that have been activated
// i guess. just do a loop over them.
sb.safePrintf("<div nobreak gbsecid=%"UINT32" "
"bgcolor=#%"XINT32" "
"onclick=gbtogglecolor()>",
(uint32_t)sj->m_tagHash,
(uint32_t)sj->m_tagHash);
// sanity check
if ( sj->m_b < 0 ) { char *xx=NULL;*xx=0; }
if ( sj->m_b > nw ) { char *xx=NULL;*xx=0; }
// and inc the /div count for that word
endCounts[sj->m_b-1]++;
}
// try next section too
sj = sj->m_next;
}
// if this is a tag, remove any coloring
if ( tids[i] ) {
}
// print the word, be it a tag, alnum, punct
sb.safeMemcpy ( wptrs[i] , wlens[i] );
// end a div tag?
if ( ! endCounts[i] ) continue;
// might be many so loop it
for ( int32_t j = 0 ; j < endCounts[i] ; j++ )
sb.safePrintf("</div>");
}
return false;
}
void gotDatedbList ( State60 *st ) {
// must only be run on host #0 since we need just one lock table
if ( g_hostdb.m_myHost->m_hostId != 0 ) { char *xx=NULL;*xx=0; }
// load turk lock table if we need to
bool s_init = false;
if ( ! s_init ) {
s_init = true;
if ( ! g_turkLocks.set(8,sizeof(TurkLock),256) )
log("turk: failed to init turk lock table");
if ( ! g_turkLocks.load(g_conf.m_dir,"turkdir/docidlocks.dat"))
log("turk: failed to load turk lock table");
}
time_t now = getTimeGlobal();
// int16_tcut
RdbList *list = &st->m_list;
// the best docid
int64_t best = 0LL;
// scan the list to get urls/docids to turk out
for ( ; ! list->isExhausted() ; ) {
// get rec
char *k = list->getCurrentKey();
// skip that
list->skipCurrentRecord();
// skip if negative
if ( (k[0] & 0x01) == 0x00 ) continue;
// get the docid
int64_t docid = g_datedb.getDocId ( k );
// skip if locked
TurkLock *tt = (TurkLock *)g_turkLock.getValue(&docid);
// if there check time
if ( tt && now - tt->m_lockTime > 3600 ) {
// remove it
g_turkLock.removeKey(&docId);
// nuke tt
tt = NULL;
}
// if still there, skip it and try next one
if ( tt ) continue;
// ok, we got a good docid to dish out
best = docId;
break;
}
SafeBuf sb;
// print description so they can clikc a button to start the turk
sb.safePrintf("<html>\n"
"<title>Event Editor</title>\n"
"<body>\n"
"<table width=\"100%%\" border=\"0\">\n"
"<tr><td style=\"background-color:#0079ba;\">\n"
"<center><font color=#00000>"
"<h2>Event Editor</h2>\n"
"</font></center></td>"
"</tr></table>");
// if we had no docid, give user an empty msg
if ( ! best ) {
sb.safePrintf("<center>Nothing currently available to edit. "
"Please try again later.</center>"
"</body></html>\n");
sendReply ( &sb );
return;
}
// lock it!
TurkLock tt;
strcpy ( tt.m_user , st->m_user );
tt.m_lockTime = now;
if ( ! g_lockTable.addLock ( &tt ) ) {
sendErrorReply ( st , g_errno );
return;
}
// . fetch the TitleRec
// . a max cache age of 0 means not to read from the cache
XmlDoc *xd = &st->m_xd;
// . when getTitleRec() is called it will load the old one
// since XmlDoc::m_setFromTitleRec will be true
// . niceness is 0
xd->set3 ( best , st->m_coll , 0 );
// if it blocks while it loads title rec, it will re-call this routine
xd->setCallback ( st , processLoopWrapper );
// good to go!
return processLoop ( st );
}
// . this is called
// . destroys the UdpSlot if false is returned
void handleRequest20 ( UdpSlot *slot , long netnice ) {
// . check g_errno
// . before, we were not sending a reply back here and we continued
// to process the request, even though it was empty. the slot
// had a NULL m_readBuf because it could not alloc mem for the read
// buf i'm assuming. and the slot was saved in a line below here...
// state20->m_msg22.m_parent = slot;
if ( g_errno ) {
log("net: Msg20 handler got error: %s.",mstrerror(g_errno));
g_udpServer.sendErrorReply ( slot , g_errno );
return;
}
// ensure request is big enough
if ( slot->m_readBufSize < (long)sizeof(Msg20Request) ) {
g_udpServer.sendErrorReply ( slot , EBADREQUESTSIZE );
return;
}
// parse the request
Msg20Request *req = (Msg20Request *)slot->m_readBuf;
// . turn the string offsets into ptrs in the request
// . this is "destructive" on "request"
long nb = req->deserialize();
// sanity check
if ( nb != slot->m_readBufSize ) { char *xx = NULL; *xx = 0; }
// sanity check, the size include the \0
if ( req->size_coll <= 1 || *req->ptr_coll == '\0' ) {
log("query: Got empty collection in msg20 handler. FIX!");
char *xx =NULL; *xx = 0;
}
// if it's not stored locally that's an error
if ( req->m_docId >= 0 && ! g_titledb.isLocal ( req->m_docId ) ) {
log("query: Got msg20 request for non-local docId %lli",
req->m_docId);
g_udpServer.sendErrorReply ( slot , ENOTLOCAL );
return;
}
// sanity
if ( req->m_docId == 0 && ! req->ptr_ubuf ) { char *xx=NULL;*xx=0; }
long long startTime = gettimeofdayInMilliseconds();
// alloc a new state to get the titlerec
XmlDoc *xd;
try { xd = new (XmlDoc); }
catch ( ... ) {
g_errno = ENOMEM;
log("query: msg20 new(%i): %s", sizeof(XmlDoc),
mstrerror(g_errno));
g_udpServer.sendErrorReply ( slot, g_errno );
return;
}
mnew ( xd , sizeof(XmlDoc) , "xd20" );
// ok, let's use the new XmlDoc.cpp class now!
xd->set20 ( req );
// encode slot
xd->m_slot = slot;
// set the callback
xd->setCallback ( xd , gotReplyWrapperxd );
// set set time
xd->m_setTime = startTime;
xd->m_cpuSummaryStartTime = 0;
// . now as for the msg20 reply!
// . TODO: move the parse state cache into just a cache of the
// XmlDoc itself, and put that cache logic into XmlDoc.cpp so
// it can be used more generally.
Msg20Reply *reply = xd->getMsg20Reply ( );
// this is just blocked
if ( reply == (void *)-1 ) return;
// got it?
gotReplyWrapperxd ( xd );
}