// . 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 );
}
// . reset a collection
// . returns false if failed
bool Collectiondb::resetColl ( char *coll , bool resetTurkdb ) {
// ensure it's not NULL
if ( ! coll ) {
log(LOG_LOGIC,"admin: Collection name to delete is NULL.");
return false;
}
// now must be "test" only for now
if ( strcmp(coll,"test") ) { char *xx=NULL;*xx=0; }
// no spiders can be out. they may be referencing the CollectionRec
// in XmlDoc.cpp... quite likely.
if ( g_conf.m_spideringEnabled ||
g_spiderLoop.m_numSpidersOut > 0 ) {
log("admin: Can not delete collection while "
"spiders are enabled or active.");
return false;
}
// do not allow this if in repair mode
if ( g_repairMode > 0 ) {
log("admin: Can not delete collection while in repair mode.");
return false;
}
// get the CollectionRec for "test"
CollectionRec *cr = getRec ( "test" );
// must be there. if not, we create test i guess
if ( ! cr ) {
log("db: could not get test coll rec");
char *xx=NULL;*xx=0;
}
// make sure an update not in progress
if ( cr->m_inProgress ) { char *xx=NULL;*xx=0; }
CollectionRec tmp;
// copy it to "tmp"
long size = (char *)&(cr->m_END_COPY) - (char *)cr;
// do not copy the hashtable crap since you will have to re-init it!
memcpy ( &tmp , cr , size ); // sizeof(CollectionRec) );
// delete the test coll now
if ( ! deleteRec ( "test" , resetTurkdb ) )
return log("admin: reset coll failed");
// make a collection called "test2" so that we copy "test"'s parms
bool status = addRec ( "test" ,
NULL ,
0 ,
true , // bool isNew ,
(collnum_t) -1 ,
// not a dump
false ,
// do not save it!
false );
// bail on error
if ( ! status ) return log("admin: failed to add new coll for reset");
// get its rec
CollectionRec *nr = getRec ( "test" );
// must be there
if ( ! nr ) { char *xx=NULL;*xx=0; }
// save this though, this might have changed!
collnum_t cn = nr->m_collnum;
// overwrite its rec
memcpy ( nr , &tmp , size ) ; // sizeof(CollectionRec) );
// put that collnum back
nr->m_collnum = cn;
// set the flag
m_needsSave = true;
// save it again after copy
nr->save();
// and clear the robots.txt cache in case we recently spidered a
// robots.txt, we don't want to use it, we want to use the one we
// have in the test-parser subdir so we are consistent
RdbCache *robots = Msg13::getHttpCacheRobots();
RdbCache *others = Msg13::getHttpCacheOthers();
robots->clear ( cn );
others->clear ( cn );
//g_templateTable.reset();
//g_templateTable.save( g_hostdb.m_dir , "turkedtemplates.dat" );
// repopulate CollectionRec::m_sortByDateTable. should be empty
// since we are resetting here.
//initSortByDateTable ( coll );
// done
return true;
}
// . returns false if blocked, true otherwise
// . sets g_errno on error
bool sendPageCatdb ( TcpSocket *s , HttpRequest *r ) {
// are we the admin?
bool isAdmin = g_collectiondb.isAdmin ( r , s );
// get the collection record
CollectionRec *cr = g_collectiondb.getRec ( r );
if ( ! cr ) {
log("admin: No collection record found "
"for specified collection name. Could not add sites to "
"tagdb. Returning HTTP status of 500.");
return g_httpServer.sendErrorReply ( s , 500 ,
"collection does not exist");
}
bool isAssassin = cr->isAssassin ( s->m_ip );
if ( isAdmin ) isAssassin = true;
// bail if permission denied
if ( ! isAssassin && ! cr->hasPermission ( r , s ) ) {
log("admin: Bad collection name or password. Could not add "
"sites to tagdb. Permission denied.");
return sendPageLogin ( s , r ,
"Collection name or "
"password is incorrect");
}
// get the collection
long collLen = 0;
char *coll = r->getString("c", &collLen, NULL);
// check for generate catdb command
long genCatdb = r->getLong("gencatdb", 0);
// check for a lookup url
long urlLen = 0;
char *url = r->getString("caturl", &urlLen, NULL);
// create the State
StateCatdb *st;
try { st = new (StateCatdb); }
catch ( ... ) {
g_errno = ENOMEM;
log("catdb: Unable to allocate %i bytes for StateCatdb",
sizeof(StateCatdb) );
return true;
}
mnew ( st, sizeof(StateCatdb), "PageCatdb" );
// fill the state
st->m_socket = s;
st->m_r.copy(r);
// copy collection
if (collLen > MAX_COLL_LEN) collLen = MAX_COLL_LEN - 1;
memcpy(st->m_coll, coll, collLen);
st->m_coll[collLen] = '\0';
st->m_collLen = collLen;
// defaults
st->m_catLookup = false;
st->m_genCatdb = false;
st->m_startTime = gettimeofdayInMilliseconds();
// generate catdb if requested
if (genCatdb == 1) {
st->m_genCatdb = true;
if (!st->m_msg2a.makeCatdb ( st->m_coll,
st->m_collLen,
false,
st,
sendReplyWrapper ) )
return false;
}
// update catdb from .new files
else if (genCatdb == 2) {
st->m_genCatdb = true;
if (!st->m_msg2a.makeCatdb ( st->m_coll,
st->m_collLen,
true,
st,
sendReplyWrapper ) )
return false;
}
// lookup a url if requested
else if (url && urlLen > 0) {
st->m_catLookup = true;
// set the url
st->m_url.set(url, urlLen);
// call msg8b to lookup in catdb
if (!st->m_msg8b.getCatRec ( &st->m_url,
NULL,//st->m_coll,
0,//st->m_collLen,
true,
1,
&st->m_catRec,
st,
gotCatInfoWrapper))
//RDB_CATDB ) )
//RDB_TAGDB ) )
return false;
}
// otherwise return the regular page
return sendReply ( st );
}
// . returns false if blocked, true otherwise
// . sets g_errno on error
bool sendPageAddUrl ( TcpSocket *s , HttpRequest *r ) {
// . get fields from cgi field of the requested url
// . get the search query
long urlLen = 0;
char *url = r->getString ( "u" , &urlLen , NULL /*default*/);
// see if they provided a url of a file of urls if they did not
// provide a url to add directly
//bool isAdmin = g_collectiondb.isAdmin ( r , s );
bool isAdmin = r->getIsLocal();
long ufuLen = 0;
char *ufu = NULL;
if ( isAdmin )
// get the url of a file of urls (ufu)
ufu = r->getString ( "ufu" , &ufuLen , NULL );
// can't be too long, that's obnoxious
if ( urlLen > MAX_URL_LEN || ufuLen > MAX_URL_LEN ) {
g_errno = EBUFTOOSMALL;
g_msg = " (error: url too long)";
return g_httpServer.sendErrorReply(s,500,"url too long");
}
// 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);
}
// get collection rec
CollectionRec *cr = g_collectiondb.getRec ( coll );
// bitch if no collection rec found
if ( ! cr ) {
g_errno = ENOCOLLREC;
g_msg = " (error: no collection)";
return g_httpServer.sendErrorReply(s,500,"no coll rec");
}
// . make sure the ip is not banned
// . we may also have an exclusive list of IPs for private collections
if ( ! cr->hasSearchPermission ( s ) ) {
g_errno = ENOPERM;
g_msg = " (error: permission denied)";
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno));
}
// make a new state
State1 *st1 ;
try { st1 = new (State1); }
catch ( ... ) {
g_errno = ENOMEM;
log("PageAddUrl: new(%i): %s",
sizeof(State1),mstrerror(g_errno));
return g_httpServer.sendErrorReply(s,500,mstrerror(g_errno)); }
mnew ( st1 , sizeof(State1) , "PageAddUrl" );
// save socket and isAdmin
st1->m_socket = s;
st1->m_isAdmin = isAdmin;
// assume no url buf yet, set below
//st1->m_ubuf = NULL;
//st1->m_ubufAlloc = NULL;
//st1->m_metaList = NULL;
// save the url
st1->m_url[0] = '\0';
if ( 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
st1->m_urlLen=cleanInput ( st1->m_url,
MAX_URL_LEN,
uu.getUrl(),
uu.getUrlLen() );
// point to that as the url "buf" to add
//st1->m_ubuf = st1->m_url;
//st1->m_ubufSize = urlLen;
//st1->m_ubufAlloc = NULL; // do not free it!
}
// save the "ufu" (url of file of urls)
st1->m_ufu[0] = '\0';
st1->m_ufuLen = ufuLen;
memcpy ( st1->m_ufu , ufu , ufuLen );
st1->m_ufu[ufuLen] = '\0';
st1->m_doTuringTest = cr->m_doTuringTest;
char *username = g_users.getUsername(r);
if(username) strcpy(st1->m_username,username);
//st1->m_user = g_pages.getUserType ( s , r );
st1->m_spiderLinks = true;
st1->m_strip = true;
//st1->m_raw = r->getLong("raw",0);
// init state2
for ( long i = 0; i < 5; i++ ){
st1->m_state2[i].m_buf = NULL;
st1->m_state2[i].m_bufLen = 0;
st1->m_state2[i].m_bufMaxLen = 0;
}
// save the collection name in the State1 class
if ( collLen > MAX_COLL_LEN ) collLen = MAX_COLL_LEN;
strncpy ( st1->m_coll , coll , collLen );
st1->m_coll [ collLen ] = '\0';
// assume they answered turing test correctly
st1->m_goodAnswer = true;
// if addurl is turned off, just print "disabled" msg
if ( ! g_conf.m_addUrlEnabled ) return sendReply ( st1 , false );
// can also be turned off in the collection rec
if ( ! cr->m_addUrlEnabled ) return sendReply ( st1 , false );
// or if in read-only mode
if ( g_conf.m_readOnlyMode ) return sendReply ( st1 , false );
// cannot add if another Msg10 from here is still in progress
if ( s_inprogress ) return sendReply ( st1 , true );
// use now as the spiderTime
// get ip of submitter
//unsigned long h = ipdom ( s->m_ip );
// . use top 2 bytes now, some isps have large blocks
// . if this causes problems, then they can do pay for inclusion
unsigned long h = iptop ( s->m_ip );
long codeLen;
char* code = r->getString("code", &codeLen);
if(g_autoBan.hasCode(code, codeLen, s->m_ip)) {
long uipLen = 0;
char* uip = r->getString("uip",&uipLen);
long hip = 0;
//use the uip when we have a raw query to test if
//we can submit
if(uip) {
hip = atoip(uip, uipLen);
h = iptop( hip );
}
}
st1->m_strip = r->getLong("strip",0);
// Remember, for cgi, if the box is not checked, then it is not
// reported in the request, so set default return value to 0
long spiderLinks = r->getLong("spiderLinks",-1);
// also support all lowercase like PageInject.cpp uses
if ( spiderLinks == -1 )
spiderLinks = r->getLong("spiderlinks",0);
// . should we force it into spiderdb even if already in there
// . use to manually update spider times for a url
// . however, will not remove old scheduled spider times
// . mdw: made force on the default
st1->m_forceRespider = r->getLong("force",1); // 0);
long now = getTimeGlobal();
// . allow 1 submit every 1 hour
// . restrict by submitter domain ip
if ( ! st1->m_isAdmin &&
! canSubmit ( h , now , cr->m_maxAddUrlsPerIpDomPerDay ) ) {
// return error page
g_errno = ETOOEARLY;
return sendReply ( st1 , true );
}
//st1->m_query = r->getString( "qts", &st1->m_queryLen );
// check it, if turing test is enabled for this collection
if ( ! st1->m_isAdmin && cr->m_doTuringTest &&
! g_turingTest.isHuman(r) ) {
// log note so we know it didn't make it
g_msg = " (error: bad answer)";
//log("PageAddUrl:: addurl failed for %s : bad answer",
// iptoa(s->m_ip));
st1->m_goodAnswer = false;
return sendReply ( st1 , true /*addUrl enabled?*/ );
}
//if ( st1->m_queryLen > 0 )
// return getPages( st1 );
// if no url given, just print a blank page
if ( ! url ) return sendReply ( st1 , true );
//
// make a SpiderRequest
//
SpiderRequest *sreq = &st1->m_sreq;
// reset it
sreq->reset();
// make the probable docid
long long probDocId = g_titledb.getProbableDocId ( st1->m_url );
// make one up, like we do in PageReindex.cpp
long firstIp = (probDocId & 0xffffffff);
// . now fill it up
// . TODO: calculate the other values... lazy!!! (m_isRSSExt,
// m_siteNumInlinks,...)
sreq->m_isNewOutlink = 1;
sreq->m_isAddUrl = 1;
sreq->m_addedTime = now;
sreq->m_fakeFirstIp = 1;
sreq->m_probDocId = probDocId;
sreq->m_firstIp = firstIp;
sreq->m_hopCount = 0;
// its valid if root
Url uu; uu.set ( st1->m_url );
if ( uu.isRoot() ) sreq->m_hopCountValid = true;
// too big?
//long len = st1->m_urlLen;
// the url! includes \0
strcpy ( sreq->m_url , st1->m_url );
// call this to set sreq->m_dataSize now
sreq->setDataSize();
// make the key dude -- after setting url
sreq->setKey ( firstIp , 0LL, false );
// need a fake first ip lest we core!
//sreq->m_firstIp = (pdocId & 0xffffffff);
// how to set m_firstIp? i guess addurl can be throttled independently
// of the other urls??? use the hash of the domain for it!
long dlen;
char *dom = getDomFast ( st1->m_url , &dlen );
// fake it for this...
//sreq->m_firstIp = hash32 ( dom , dlen );
// sanity
if ( ! dom ) {
g_errno = EBADURL;
return sendReply ( st1 , true );
}
// shortcut
Msg4 *m = &st1->m_msg4;
// now add that to spiderdb using msg4
if ( ! m->addMetaList ( (char *)sreq ,
sreq->getRecSize() ,
coll ,
st1 , // state
addedStuff ,
MAX_NICENESS ,
RDB_SPIDERDB ) )
// we blocked
return false;
// send back the reply
return sendReply ( st1 , true );
}
