///////////
//
// main > Basic > Status
//
///////////
bool sendPageBasicStatus ( TcpSocket *socket , HttpRequest *hr ) {
char buf [ 128000 ];
SafeBuf sb(buf,128000);
sb.reset();
char format = hr->getReplyFormat();
// true = usedefault coll?
CollectionRec *cr = g_collectiondb.getRec ( hr , true );
if ( ! cr ) {
g_httpServer.sendErrorReply(socket,500,"invalid collection");
return true;
}
if ( format == FORMAT_JSON || format == FORMAT_XML) {
// this is in PageCrawlBot.cpp
printCrawlDetails2 ( &sb , cr , format );
char *ct = "text/xml";
if ( format == FORMAT_JSON ) ct = "application/json";
return g_httpServer.sendDynamicPage (socket,
sb.getBufStart(),
sb.length(),
0, // cachetime
false,//POSTReply ,
ct);
}
// print standard header
if ( format == FORMAT_HTML )
// this prints the <form tag as well
g_pages.printAdminTop ( &sb , socket , hr );
// table to split between widget and stats in left and right panes
if ( format == FORMAT_HTML ) {
sb.safePrintf("<TABLE id=pane>"
"<TR><TD valign=top>");
}
int32_t savedLen1, savedLen2;
//
// widget
//
// put the widget in here, just sort results by spidered date
//
// the scripts do "infinite" scrolling both up and down.
// but if you are at the top then new results will load above
// you and we try to maintain your current visual state even though
// the scrollbar position will change.
//
if ( format == FORMAT_HTML ) {
// save position so we can output the widget code
// so user can embed it into their own web page
savedLen1 = sb.length();
printScrollingWidget ( &sb , cr );
savedLen2 = sb.length();
}
// the right table pane is the crawl stats
if ( format == FORMAT_HTML ) {
sb.safePrintf("</TD><TD valign=top>");
}
//
// show stats
//
if ( format == FORMAT_HTML ) {
char *seedStr = cr->m_diffbotSeeds.getBufStart();
if ( ! seedStr ) seedStr = "";
SafeBuf tmp;
int32_t crawlStatus = -1;
getSpiderStatusMsg ( cr , &tmp , &crawlStatus );
CrawlInfo *ci = &cr->m_localCrawlInfo;
int32_t sentAlert = (int32_t)ci->m_sentCrawlDoneAlert;
if ( sentAlert ) sentAlert = 1;
//sb.safePrintf(
// "<form method=get action=/crawlbot>"
// "%s"
// , sb.getBufStart() // hidden input token/name/..
// );
char *hurts = "No";
if ( cr->m_globalCrawlInfo.m_hasUrlsReadyToSpider )
hurts = "Yes";
sb.safePrintf(//"<TABLE border=0>"
//"<TR><TD valign=top>"
"<table id=stats border=0 cellpadding=5>"
"<tr>"
"<td><b>Crawl Status Code:</td>"
"<td>%" PRId32"</td>"
"</tr>"
"<tr>"
"<td><b>Crawl Status Msg:</td>"
"<td>%s</td>"
"</tr>"
//"<tr>"
//"<td><b>Rounds Completed:</td>"
//"<td>%" PRId32"</td>"
//"</tr>"
"<tr>"
"<td><b>Has Urls Ready to Spider:</td>"
"<td>%s</td>"
"</tr>"
// this will have to be in crawlinfo too!
//"<tr>"
//"<td><b>pages indexed</b>"
//"<td>%" PRId64"</td>"
//"</tr>"
"<tr>"
"<td><b><nobr>URLs Harvested</b> "
"(may include dups)</nobr></td>"
"<td>%" PRId64"</td>"
"</tr>"
//"<tr>"
//"<td><b>URLs Examined</b></td>"
//"<td>%" PRId64"</td>"
//"</tr>"
"<tr>"
"<td><b>Page Crawl Attempts</b></td>"
"<td>%" PRId64"</td>"
"</tr>"
"<tr>"
"<td><b>Page Crawl Successes</b></td>"
"<td>%" PRId64"</td>"
"</tr>"
, crawlStatus
, tmp.getBufStart()
//, cr->m_spiderRoundNum
//, cr->m_globalCrawlInfo.m_hasUrlsReadyToSpider
, hurts
, cr->m_globalCrawlInfo.m_urlsHarvested
//, cr->m_globalCrawlInfo.m_urlsConsidered
, cr->m_globalCrawlInfo.m_pageDownloadAttempts
, cr->m_globalCrawlInfo.m_pageDownloadSuccesses
);
//
// begin status code breakdown
//
for ( int32_t i = 0 ; i < 65536 ; i++ ) {
if ( g_stats.m_allErrorsNew[i] == 0 &&
g_stats.m_allErrorsOld[i] == 0 )
continue;
sb.safePrintf (
"<tr>"
"<td><b> <a href=/search?c=%s&q="
"gbstatusmsg%%3A"
"%%22"
,
cr->m_coll );
sb.urlEncode(mstrerror(i));
sb.safePrintf ("%%22>"
"%s"
"</a>"
"</b></td>"
"<td>%" PRId64"</td>"
"</tr>\n" ,
mstrerror(i),
g_stats.m_allErrorsNew[i] +
g_stats.m_allErrorsOld[i] );
}
//
// end status code breakdown
//
char tmp3[64];
struct tm *timeStruct;
time_t tt = (time_t)cr->m_diffbotCrawlStartTime;
timeStruct = localtime(&tt);
// Jan 01 1970 at 10:30:00
strftime ( tmp3,64 , "%b %d %Y at %H:%M:%S",timeStruct);
sb.safePrintf("<tr><td><b>Collection Created</b></td>"
"<td>%s (local time)</td></tr>",tmp3);
// print link to embed the code in their own site
SafeBuf embed;
embed.htmlEncode(sb.getBufStart()+savedLen1,
savedLen2-savedLen1,
false); // encodePoundSign #?
// convert all ''s to "'s for php's echo ''; cmd
embed.replaceChar('\'','\"');
sb.safePrintf("<tr>"
"<td valign=top>"
"<a onclick=\""
"var dd=document.getElementById('hcode');"
"if ( dd.style.display=='none' ) "
"dd.style.display=''; "
"else "
"dd.style.display='none';"
"\" style=color:blue;>"
"<u>"
"show Widget HTML code"
"</u>"
"</a>"
"</td><td>"
"<div id=hcode style=display:none;"
"max-width:800px;>"
"%s"
"</div>"
"</td></tr>"
, embed.getBufStart() );
sb.safePrintf("<tr>"
"<td valign=top>"
"<a onclick=\""
"var dd=document.getElementById('pcode');"
"if ( dd.style.display=='none' ) "
"dd.style.display=''; "
"else "
"dd.style.display='none';"
"\" style=color:blue;>"
"<u>"
"show Widget PHP code"
"</u>"
"</a>"
"</td>"
"<td>"
"<div id=pcode style=display:none;"
"max-width:800px;>"
"<i>"
"echo '"
"%s"
"';"
"</i>"
"</div>"
"</td></tr>"
, embed.getBufStart() );
sb.safePrintf("</table>\n\n");
}
// end the right table pane
if ( format == FORMAT_HTML ) {
sb.safePrintf("</TD></TR></TABLE>");
}
//if ( format != FORMAT_JSON )
// // wrap up the form, print a submit button
// g_pages.printAdminBottom ( &sb );
return g_httpServer.sendDynamicPage (socket,
sb.getBufStart(),
sb.length(),
0); // cachetime
}
void processReply ( char *reply , long replyLen ) {
// store our current reply
SafeBuf fb2;
fb2.safeMemcpy(reply,replyLen );
fb2.nullTerm();
// log that we got the reply
log("qa: got reply(len=%li)(errno=%s)=%s",
replyLen,mstrerror(g_errno),reply);
char *content = NULL;
long contentLen = 0;
// get mime
if ( reply ) {
HttpMime mime;
mime.set ( reply, replyLen , NULL );
// only hash content since mime has a timestamp in it
content = mime.getContent();
contentLen = mime.getContentLen();
if ( content && contentLen>0 && content[contentLen] ) {
char *xx=NULL;*xx=0; }
}
if ( ! content ) {
content = "";
contentLen = 0;
}
s_content = content;
// take out <responseTimeMS>
markOut ( content , "<currentTimeUTC>");
markOut ( content , "<responseTimeMS>");
// until i figure this one out, take it out
markOut ( content , "<docsInCollection>");
// until i figure this one out, take it out
markOut ( content , "<hits>");
// for those links in the html pages
markOut ( content, "rand64=");
// for json
markOut ( content , "\"currentTimeUTC\":" );
markOut ( content , "\"responseTimeMS\":");
markOut ( content , "\"docsInCollection\":");
// for xml
markOut ( content , "<currentTimeUTC>" );
markOut ( content , "<responseTimeMS>");
markOut ( content , "<docsInCollection>");
// indexed 1 day ago
markOut ( content,"indexed:");
// modified 1 day ago
markOut ( content,"modified:");
// s_gigabitCount... it is perpetually incrementing static counter
// in PageResults.cpp
markOut(content,"ccc(");
markOut(content,"id=fd");
markOut(content,"id=sd");
// for some reason the term freq seems to change a little in
// the scoring table
markOut(content,"id=tf");
// make checksum. we ignore back to back spaces so this
// hash works for <docsInCollection>10 vs <docsInCollection>9
long contentCRC = 0;
if ( content ) contentCRC = qa_hash32 ( content );
// note it
log("qa: got contentCRC of %lu",contentCRC);
// if what we expected, save to disk if not there yet, then
// call s_callback() to resume the qa pipeline
/*
if ( contentCRC == s_expectedCRC ) {
// save content if good
char fn3[1024];
sprintf(fn3,"%sqa/content.%lu",g_hostdb.m_dir,contentCRC);
File ff; ff.set ( fn3 );
if ( ! ff.doesExist() ) {
// if not there yet then save it
fb2.save(fn3);
}
// . continue on with the qa process
// . which qa function that may be
//s_callback();
return;
}
*/
//
// if crc of content does not match what was expected then do a diff
// so we can see why not
//
// this means caller does not care about the response
if ( ! s_checkCRC ) {
//s_callback();
return;
}
//const char *emsg = "qa: bad contentCRC of %li should be %li "
// "\n";//"phase=%li\n";
//fprintf(stderr,emsg,contentCRC,s_expectedCRC);//,s_phase-1);
// hash url
long urlHash32 = hash32n ( s_url.getUrl() );
// combine test function too since two tests may use the same url
long nameHash = hash32n ( s_qt->m_testName );
// combine together
urlHash32 = hash32h ( nameHash , urlHash32 );
static bool s_init = false;
if ( ! s_init ) {
s_init = true;
s_ht.set(4,4,1024,NULL,0,false,0,"qaht");
// make symlink
//char cmd[512];
//snprintf(cmd,"cd %s/html ;ln -s ../qa ./qa", g_hostdb.m_dir);
//system(cmd);
char dir[1024];
snprintf(dir,1000,"%sqa",g_hostdb.m_dir);
long status = ::mkdir ( dir ,
S_IRUSR | S_IWUSR | S_IXUSR |
S_IRGRP | S_IWGRP | S_IXGRP |
S_IROTH | S_IXOTH );
if ( status == -1 && errno != EEXIST && errno )
log("qa: Failed to make directory %s: %s.",
dir,mstrerror(errno));
// try to load from disk
SafeBuf fn;
fn.safePrintf("%s/qa/",g_hostdb.m_dir);
log("qa: loading crctable.dat");
s_ht.load ( fn.getBufStart() , "crctable.dat" );
}
// break up into lines
char fn2[1024];
sprintf(fn2,"%sqa/content.%lu",g_hostdb.m_dir,contentCRC);
fb2.save ( fn2 );
// look up in hashtable to see what reply crc should be
long *val = (long *)s_ht.getValue ( &urlHash32 );
// just return if the same
if ( val && contentCRC == *val ) {
g_qaOutput.safePrintf("<b style=color:green;>"
"passed test</b><br>%s : "
"<a href=%s>%s</a> (urlhash=%lu "
"crc=<a href=/qa/content.%lu>"
"%lu</a>)<br>"
"<hr>",
s_qt->m_testName,
s_url.getUrl(),
s_url.getUrl(),
urlHash32,
contentCRC,
contentCRC);
return;
}
if ( ! val ) {
// add it so we know
s_ht.addKey ( &urlHash32 , &contentCRC );
g_qaOutput.safePrintf("<b style=color:blue;>"
"first time testing</b><br>%s : "
"<a href=%s>%s</a> "
"(urlhash=%lu "
"crc=<a href=/qa/content.%lu>%lu"
"</a>)<br>"
"<hr>",
s_qt->m_testName,
s_url.getUrl(),
s_url.getUrl(),
urlHash32,
contentCRC,
contentCRC);
return;
}
log("qa: crc changed for url %s from %li to %li",
s_url.getUrl(),*val,contentCRC);
// get response on file
SafeBuf fb1;
char fn1[1024];
sprintf(fn1,"%sqa/content.%lu",g_hostdb.m_dir, *val);
fb1.load(fn1);
fb1.nullTerm();
// do the diff between the two replies so we can see what changed
char cmd[1024];
sprintf(cmd,"diff %s %s > /tmp/diffout",fn1,fn2);
log("qa: %s\n",cmd);
system(cmd);
g_numErrors++;
g_qaOutput.safePrintf("<b style=color:red;>FAILED TEST</b><br>%s : "
"<a href=%s>%s</a> (urlhash=%lu)<br>"
"<input type=checkbox name=urlhash%lu value=1 "
// use ajax to update test crc. if you undo your
// check then it should put the old val back.
// when you first click the checkbox it should
// gray out the diff i guess.
"onclick=submitchanges(%lu,%lu);> "
"Accept changes"
"<br>"
"original on left, new on right. "
"oldcrc = <a href=/qa/content.%lu>%lu</a>"
" != <a href=/qa/content.%lu>%lu</a> = newcrc"
"<br>diff output follows:<br>"
"<pre id=%lu style=background-color:0xffffff;>",
s_qt->m_testName,
s_url.getUrl(),
s_url.getUrl(),
urlHash32,
// input checkbox name field
urlHash32,
// submitchanges() parms
urlHash32,
contentCRC,
// original/old content.%lu
*val,
*val,
// new content.%lu
contentCRC,
contentCRC,
// for the pre tag id:
urlHash32);
// store in output
SafeBuf sb;
sb.load("/tmp/diffout");
g_qaOutput.htmlEncode ( sb.getBufStart() );
g_qaOutput.safePrintf("</pre><br><hr>");
// if this is zero allow it to slide by. it is learning mode i guess.
// so we can learn what crc we need to use.
// otherwise, stop right there for debugging
//if ( s_expectedCRC != 0 ) exit(1);
// keep on going
//s_callback();
}
bool sendPageThesaurus( TcpSocket *s, HttpRequest *r ) {
SafeBuf p;
char getBuf[64]; // holds extra values for GET method
char formBuf[256]; // holds extra values for forms
snprintf(getBuf, 64, "c=%s",
r->getString("c", 0, ""));
snprintf(formBuf, 256,
"<input type=hidden name=\"c\" value=\"%s\">",
//"<input type=hidden name=\"pwd\" value=\"%s\">",
r->getString("c", 0, ""));
g_pages.printAdminTop( &p, s, r);
if (r->getLong("cancel", 0) != 0) {
g_thesaurus.cancelRebuild();
p.safePrintf("<br><br>\n");
p.safePrintf(
"<center><b><font color=#ff0000>"
"rebuild canceled"
"</font></b></center>");
}
if (r->getLong("rebuild", 0) != 0) {
bool full = r->getLong("full", 0);
p.safePrintf("<br><br>\n");
if (g_thesaurus.rebuild(0, full)) {
p.safePrintf(
"<center><b><font color=#ff0000>"
"error starting rebuild, check log for details"
"</font></b></center>");
} else {
p.safePrintf(
"<center><b><font color=#ff0000>"
"rebuild started"
"</font></b></center>");
}
}
if (r->getLong("rebuildaff", 0) != 0) {
bool full = r->getLong("full", 0);
p.safePrintf("<br><br>\n");
if (g_thesaurus.rebuildAffinity(0, full)) {
p.safePrintf(
"<center><b><font color=#ff0000>"
"error starting rebuild, check log for details"
"</font></b></center>");
} else {
p.safePrintf(
"<center><b><font color=#ff0000>"
"rebuild started"
"</font></b></center>");
}
}
if (r->getLong("distribute", 0) != 0) {
char cmd[1024];
p.safePrintf("<br><br>\n");
if (g_thesaurus.m_affinityState) {
p.safePrintf(
"<center><b><font color=#ff0000>"
"cannot distribute during rebuild"
"</font></b></center>");
} else {
for ( long i = 0; i < g_hostdb.getNumHosts() ; i++ ) {
Host *h = g_hostdb.getHost(i);
snprintf(cmd, 512,
"rcp -r "
"./dict/thesaurus.* "
"%s:%s/dict/ &",
iptoa(h->m_ip),
h->m_dir);
log(LOG_INFO, "admin: %s", cmd);
system( cmd );
}
p.safePrintf(
"<center><b><font color=#ff0000>"
"data distributed"
"</font></b></center>");
}
}
if (r->getLong("reload", 0) != 0) {
p.safePrintf("<br><br>\n");
if (r->getLong("cast", 0) != 0) {
p.safePrintf(
"<center><b><font color=#ff0000>"
"reload command broadcast"
"</font></b></center>");
} else if (g_thesaurus.init()) {
p.safePrintf(
"<center><b><font color=#ff0000>"
"thesaurus data reloaded"
"</font></b></center>");
} else {
p.safePrintf(
"<center><b><font color=#ff0000>"
"error reloading thesaurus data"
"</font></b></center>");
}
}
long manualAddLen = 0;
char *manualAdd = NULL;
SafeBuf manualAddBuf;
if ((manualAdd = r->getString("manualadd", &manualAddLen))) {
trimWhite(manualAdd);
manualAddLen = gbstrlen(manualAdd);
File manualFile;
manualFile.set(g_hostdb.m_dir, "dict/thesaurus-manual.txt");
if (manualFile.open(O_WRONLY | O_CREAT | O_TRUNC) &&
(manualFile.write(manualAdd, manualAddLen, 0) ==
manualAddLen)) {
char newl = '\n'; // for write()
if (manualAdd[manualAddLen-1] != '\n')
manualFile.write(&newl, 1, manualAddLen);
p.safePrintf(
"<center><b><font color=#ff0000>"
"updated manual add file sucessfully"
"</font></b></center>");
} else {
p.safePrintf(
"<center><b><font color=#ff0000>"
"error writing manual add file"
"</font></b></center>");
}
} else {
char ff[PATH_MAX];
snprintf(ff, PATH_MAX, "%sdict/thesaurus-manual.txt",
g_hostdb.m_dir);
if (manualAddBuf.fillFromFile(ff)) {
if (*(manualAddBuf.getBuf()-1) != '\n')
manualAddBuf.pushChar('\n');
manualAdd = manualAddBuf.getBufStart();
manualAddLen = manualAddBuf.length();
}
}
long affinityAddLen = 0;
char *affinityAdd = NULL;
SafeBuf affinityAddBuf;
if ((affinityAdd = r->getString("affinityadd", &affinityAddLen))) {
trimWhite(affinityAdd);
affinityAddLen = gbstrlen(affinityAdd);
File affinityFile;
affinityFile.set(g_hostdb.m_dir,
"dict/thesaurus-affinity.txt");
if (affinityFile.open(O_WRONLY | O_CREAT | O_TRUNC) &&
(affinityFile.write(affinityAdd, affinityAddLen, 0) ==
affinityAddLen)) {
char newl = '\n'; // for write()
if (affinityAdd[affinityAddLen-1] != '\n')
affinityFile.write(&newl, 1, affinityAddLen);
p.safePrintf(
"<center><b><font color=#ff0000>"
"updated affinity add file sucessfully"
"</font></b></center>");
} else {
p.safePrintf(
"<center><b><font color=#ff0000>"
"error writing affinity add file"
"</font></b></center>");
}
} else {
char ff[PATH_MAX];
snprintf(ff, PATH_MAX, "%sdict/thesaurus-affinity.txt",
g_hostdb.m_dir);
if (affinityAddBuf.fillFromFile(ff)) {
if (*(affinityAddBuf.getBuf()-1) != '\n')
affinityAddBuf.pushChar('\n');
affinityAdd = affinityAddBuf.getBufStart();
affinityAddLen = affinityAddBuf.length();
}
}
char *syn = r->getString("synonym");
long len = 0;
if (syn) len = gbstrlen(syn);
if (len) {
SynonymInfo info;
bool r = g_thesaurus.getAllInfo(syn, &info, len, SYNBIT_ALL);
p.safePrintf("<br><br>\n");
p.safePrintf (
"<table cellpadding=4 width=100%% bgcolor=#%s border=1>"
"<tr>"
"<td colspan=2 bgcolor=#%s>"
"<center><b>Synonym List (%ld)</b></center>"
"</td>"
"</tr>\n",
LIGHT_BLUE, DARK_BLUE, info.m_numSyns);
if (r) {
p.safePrintf("<tr>"
"<td align=right><tt>%s</tt></td>"
"<td align=left>"
"<tt>1.000/%08lX (1.000/%08lX)</tt>"
"</td>"
"</tr>\n", syn, MAX_AFFINITY, MAX_AFFINITY);
for (long i = 0; i < info.m_numSyns; i++) {
// get the reverse affinity as well
long aff = g_thesaurus.getAffinity(
info.m_syn[i], syn,
info.m_len[i], len);
p.safePrintf(
"<tr>"
"<td width=40%% align=right>"
"<tt>");
p.safeMemcpy(info.m_syn[i], info.m_len[i]);
p.safePrintf("</tt>"
"</td>"
"<td width=60%% align=left>"
"<tt>");
if (info.m_affinity[i] >= 0) {
p.safePrintf("%0.3f/%08lX ",
(float)info.m_affinity[i]
/ MAX_AFFINITY,
info.m_affinity[i]);
} else {
p.safePrintf("u ");
}
if (aff >= 0) {
p.safePrintf("(%0.3f/%08lX) ",
(float)aff / MAX_AFFINITY,
aff);
} else {
p.safePrintf("(u) ");
}
p.safePrintf("(%ld) (%ld) (%ld) (%ld) "
"(%lld) (%lld)",
(long)info.m_type[i], (long)info.m_sort[i],
info.m_firstId[i], info.m_lastId[i],
info.m_leftSynHash[i],
info.m_rightSynHash[i]);
for (int j = info.m_firstId[i];
j <= info.m_lastId[i];
j++) {
p.safePrintf(" (%lld)",
info.m_termId[j]);
}
p.safePrintf(
"</tt>"
"</td>"
"</tr>\n");
}
p.safePrintf("</table>");
} else {
p.safePrintf("<tr>"
"<td align=center><font color=#FF0000>"
"synonym not found: %s"
"</font></td>"
"</tr>\n",
syn);
}
}
p.safePrintf ( "<br><br>\n" );
p.safePrintf (
"<table cellpadding=4 width=100%% bgcolor=#%s border=1>"
"<tr>"
"<td colspan=2 bgcolor=#%s>"
"<center><b>Thesaurus Controls"
"</b></center></td>"
"</tr>\n",
LIGHT_BLUE, DARK_BLUE);
p.safePrintf (
"<tr>"
"<td width=37%%><b>rebuild all data</b><br>"
"<font size=1>"
"rebuilds synonyms and then begins the rebuild process for "
"affinity data; this should only be run on one host, as the "
"data is copied when the process is finished; full rebuild "
"does not use existing affinity data"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b><a href=\"/master/thesaurus?rebuild=1&%s\">"
"rebuild all data</a> <a href=\"/master/thesaurus?"
"rebuild=1&full=1&%s\">(full)</a></b></center>"
"</td>"
"</tr>\n", getBuf, getBuf);
p.safePrintf (
"<tr>"
"<td width=37%%><b>distribute data</b><br>"
"<font size=1>"
"distributes all thesaurus data to all hosts, this is "
"normally done automatically but if there was a problem "
"with the copy, this lets you do it manually"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b><a href=\"/master/thesaurus?distribute=1&%s\">"
"distribute data</a></b></center>"
"</td>"
"</tr>\n", getBuf);
p.safePrintf (
"<tr>"
"<td width=37%%><b>reload data</b><br>"
"<font size=1>"
"reloads the synonyms and affinity table on this host only"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b>"
"<a href=\"/master/thesaurus?reload=1&cast=0&%s\">"
"reload data</a></b></center>"
"</td>"
"</tr>\n", getBuf);
p.safePrintf (
"<tr>"
"<td width=37%%><b>reload data (all hosts)</b><br>"
"<font size=1>"
"reloads the synonyms and affinity table on all hosts"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b>"
"<a href=\"/master/thesaurus?reload=1&cast=1&%s\">"
"reload data (all hosts)</a></b></center>"
"</td>"
"</tr>\n", getBuf);
p.safePrintf (
"<tr>"
"<td width=37%%><b>list synonyms</b><br>"
"<font size=1>"
"enter a word here to list all synonym entries and their "
"affinities"
"</font>"
"</td>"
"<td width=12%%>"
"<form action=\"/master/thesaurus>\">"
"<input type=text name=synonym size=20>"
"<input type=submit value=Submit>"
"%s"
"</form></td>"
"</tr>\n", formBuf);
p.safePrintf (
"<tr>"
"<td colspan=2 bgcolor=#%s>"
"<center><b>Affinity Controls"
"</b></center></td>"
"</tr>\n",
DARK_BLUE);
p.safePrintf (
"<tr>"
"<td width=37%%><b>cancel running rebuild</b><br>"
"<font size=1>"
"cancels the rebuild and throws all intermediate data away"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b><a href=\"/master/thesaurus?cancel=1&%s\">"
"cancel running rebuild</a></b></center>"
"</td>"
"</tr>\n", getBuf);
p.safePrintf (
"<tr>"
"<td width=37%%><b>rebuild affinity only</b><br>"
"<font size=1>"
"begins the rebuild process for affinity data, has no "
"effect if a rebuild is already in progress; full rebuild "
"does not reuse existing affinity data"
"</font>"
"</td>"
"<td width=12%% bgcolor=#0000ff>"
"<center><b><a href=\"/master/thesaurus?rebuildaff=1&%s\">"
"rebuild affinity</a> <a href=\"/master/thesaurus?"
"rebuildaff=1&full=1&%s\">(full)</a></b></center>"
"</td>"
"</tr>\n", getBuf, getBuf);
p.safePrintf (
"<tr>"
"<td colspan=2 bgcolor=#%s>"
"<center><b>Manual File Controls"
"</b></td>"
"</tr>\n",
DARK_BLUE);
p.safePrintf (
"<tr>"
"<td align=center colspan=2>");
p.safePrintf(
"<b>manually added pairs</b><br>\n"
"<font size=1>place word pairs here that should be linked "
"as synonyms, one pair per line, seperated by a pipe '|' "
"character, optionally followed by another pipe and a type "
"designation; any badly formatted lines will be silently "
"ignored</font><br>\n"
"<form action=\"/master/thesaurus\" method=post>"
"<textarea name=\"manualadd\" rows=20 cols=80>");
if (manualAdd && manualAddLen) {
p.htmlEncode(manualAdd, manualAddLen, true);
}
p.safePrintf (
"</textarea><br>"
"<input type=submit value=Submit>"
"<input type=reset value=Reset>"
"%s"
"</form></td>"
"</tr>\n",
formBuf);
p.safePrintf (
"<tr>"
"<td align=center colspan=2>"
"<b>affinity value overrides</b><br>\n"
"<font size=1>place word/phrase pairs here that should have "
"there affinity values overridden, format is "
"\"word1|word2|value\", where value is a floating point, "
"integer (either decimal or hex), or the word \"max\"; "
"any badly formatted lines will be silently ignored; note "
"that these pairs will only work if the thesaurus otherwise "
"has an entry for them, so add them to the manual add file "
"above if need be</font><br>\n"
"<form action=\"/master/thesaurus\" method=post>"
"<textarea name=\"affinityadd\" rows=20 cols=80>");
if (affinityAdd && affinityAddLen) {
p.htmlEncode(affinityAdd, affinityAddLen, true);
}
p.safePrintf (
"</textarea><br>"
"<input type=submit value=Submit>"
"<input type=reset value=Reset>"
"%s"
"</form></td>"
"</tr>\n",
formBuf);
p.safePrintf ( "</table>\n" );
p.safePrintf ( "<br><br>\n" );
p.safePrintf (
"<table cellpadding=4 width=100%% bgcolor=#%s border=1>"
"<tr>"
"<td colspan=2 bgcolor=#%s>"
"<center><b>Affinity Builder Status"
"</b></td>"
"</tr>\n",
LIGHT_BLUE, DARK_BLUE);
long long a, b, c, d, e, f, g, h, i, j, k;
StateAffinity *aff = g_thesaurus.m_affinityState;
if (!aff) {
p.safePrintf (
"<tr><td colspan=2>"
"<center><b>Not running</b></center>"
"</td></tr>\n");
a = b = c = d = e = f = g = h = i = j = k = 0;
} else {
a = aff->m_oldTable->getNumSlotsUsed();
b = aff->m_oldTable->getNumSlotsUsed() - aff->m_n;
c = aff->m_n;
d = (gettimeofdayInMilliseconds() - aff->m_time) / 1000;
if (!d || !(c / d)) {
e = 0;
} else {
e = b / (c / d);
}
f = aff->m_sent;
g = aff->m_recv;
h = aff->m_errors;
i = aff->m_old;
j = aff->m_cache;
k = aff->m_hitsTable.getNumSlotsUsed();
}
p.safePrintf (
"<tr><td><b># of total pairs</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of pairs remaining</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of pairs processed</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b>elapsed time in seconds</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b>estimated remaining time in seconds</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of requests sent</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of requests received</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of request errors</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of old values reused</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b># of cache hits</b></td>"
"<td>%lli</td></tr>\n"
"<tr><td><b>cache size</b></td>"
"<td>%lli</td></tr>\n",
a, b, c, d, e, f, g, h, i, j, k);
p.safePrintf ( "</table>\n" );
return g_httpServer.sendDynamicPage ( s, p.getBufStart(), p.length() );
}