fix mem leak of LinkInfo.

fixed json output from injecting url.

Linkdb.cpp

@@ -415,8 +415,9 @@ void Msg25::reset() {
 	m_numReplyPtrs = 0;
 	// . free the linkinfo if we are responsible for it
 	// . if someone "steals" it from us, they should set this to NULL
-	if ( m_linkInfo ) 
-		mfree ( m_linkInfo , m_linkInfo->getStoredSize(),"msg25s");
+	//if ( m_linkInfo ) 
+	//	mfree ( m_linkInfo , m_linkInfo->getStoredSize(),"msg25s");
+	// this now points into m_linkInfoBuf safebuf, just NULL it
 	m_linkInfo = NULL;
 
 	m_table.reset();
@@ -468,7 +469,8 @@ bool Msg25::getLinkInfo ( char      *site                ,
 			  bool       onlyNeedGoodInlinks  ,
 			  bool       getLinkerTitles ,
 			  long       ourHostHash32 ,
-			  long       ourDomHash32 ) {
+			  long       ourDomHash32 ,
+			  SafeBuf   *linkInfoBuf ) {
 
 	// reset the ip table
 	reset();
@@ -483,6 +485,9 @@ bool Msg25::getLinkInfo ( char      *site                ,
 	if ( ! coll ) { char *xx=NULL; *xx=0; }
 	m_onlyNeedGoodInlinks = onlyNeedGoodInlinks;
 	m_getLinkerTitles     = getLinkerTitles;
+	// save safebuf ptr, where we store the link info
+	m_linkInfoBuf = linkInfoBuf;
+	if ( ! linkInfoBuf ) { char *xx=NULL;*xx=0; }
 	// sanity check
 	if ( m_mode == MODE_PAGELINKINFO && ! docId ) {char *xx=NULL; *xx=0; }
 	// must have a valid ip
@@ -1903,7 +1908,8 @@ bool Msg25::gotLinkText ( Msg20Request *req ) { // LinkTextReply *linkText ) {
 				    m_lastUpdateTime  ,
 				    m_onlyNeedGoodInlinks ,
 				    m_niceness        ,
-				    this );
+				    this ,
+				    m_linkInfoBuf );
 	// return true with g_errno set on error
 	if ( ! m_linkInfo ) {
 		log("build: msg25 linkinfo set: %s",mstrerror(g_errno));
@@ -3182,7 +3188,8 @@ LinkInfo *makeLinkInfo ( char        *coll                    ,
 			 long         lastUpdateTime          ,
 			 bool         onlyNeedGoodInlinks      ,
 			 long         niceness                ,
-			 Msg25 *msg25 ) {
+			 Msg25 *msg25 ,
+			 SafeBuf *linkInfoBuf ) {
 
 	// for parsing the link text
 	Words words;
@@ -3449,10 +3456,11 @@ LinkInfo *makeLinkInfo ( char        *coll                    ,
 	// we need space for our header
 	need += sizeof(LinkInfo);
 	// alloc the buffer
-	char *buf = (char *)mmalloc ( need,"LinkInfo");
-	if ( ! buf ) return NULL;
+	//char *buf = (char *)mmalloc ( need,"LinkInfo");
+	//if ( ! buf ) return NULL;
+	if ( ! linkInfoBuf->reserve ( need , "LinkInfo" ) ) return NULL;
 	// set ourselves to this new buffer
-	LinkInfo *info = (LinkInfo *)buf;
+	LinkInfo *info = (LinkInfo *)(linkInfoBuf->getBufStart());
 	
 	// set our header
 	info->m_version                = 0;
@@ -3487,7 +3495,7 @@ LinkInfo *makeLinkInfo ( char        *coll                    ,
 
 	// point to our buf
 	char *p    = info->m_buf;
-	char *pend = buf + need;
+	char *pend = linkInfoBuf->getBufStart() + need;
 	// count the ones we store that are internal
 	long  icount3 = 0;
 	// now set each inlink

Linkdb.h

@@ -341,7 +341,8 @@ class Msg25 {
 			   // to not perform this algo in handleRequest20()'s
 			   // call to XmlDoc::getMsg20Reply().
 			   long       ourHostHash32 , // = 0 ,
-			   long       ourDomHash32 ); // = 0 );
+			   long       ourDomHash32 , // = 0 );
+			   SafeBuf *myLinkInfoBuf );
 	Msg25();
 	~Msg25();
 	void reset();
@@ -370,6 +371,9 @@ class Msg25 {
 
 	class LinkInfo *getLinkInfo () { return m_linkInfo; };
 
+	// m_linkInfo ptr references into here. provided by caller.
+	SafeBuf *m_linkInfoBuf;
+
 	// private:
 	// these need to be public for wrappers to call:
 	bool gotTermFreq ( bool msg42Called ) ;
@@ -886,7 +890,8 @@ LinkInfo *makeLinkInfo ( char        *coll                    ,
 			 long         lastUpdateTime          ,
 			 bool         onlyNeedGoodInlinks      ,
 			 long         niceness                ,
-			 class Msg25 *msg25 ) ;
+			 class Msg25 *msg25 ,
+			 SafeBuf *linkInfoBuf ) ;
 
 // . set from the Msg20 replies in MsgE
 // . Msg20 uses this to set the LinkInfo class to the "outlinks"

PageCrawlBot.cpp

@@ -2119,7 +2119,7 @@ bool sendPageCrawlbot ( TcpSocket *socket , HttpRequest *hr ) {
 
 	if ( ! name ) {
 		log("crawlbot: no crawl name given");
-		char *msg = "invalid or missing \"name\"";
+		char *msg = "invalid or missing name";
 		return sendErrorReply2 (socket,fmt,msg);
 	}
 
@@ -2548,13 +2548,16 @@ bool printCrawlBotPage2 ( TcpSocket *socket ,
 	if ( fmt == FMT_HTML )
 		sb.safePrintf ( "</center><br/>" );
 
+	// the ROOT JSON {
+	if ( fmt == FMT_JSON )
+		sb.safePrintf("{\n");
 
 	if ( fmt == FMT_JSON && injectionResponse )
-		sb.safePrintf("{\"seedResponse\":\"%s\"},\n\n"
+		sb.safePrintf("\"seedResponse\":\"%s\",\n\n"
 			      , injectionResponse->getBufStart() );
 
 	if ( fmt == FMT_JSON && urlUploadResponse )
-		sb.safePrintf("{\"addUrlsResponse\":\"%s\"},\n\n"
+		sb.safePrintf("\"addUrlsResponse\":\"%s\",\n\n"
 			      , urlUploadResponse->getBufStart() );
 
 
@@ -2565,7 +2568,7 @@ bool printCrawlBotPage2 ( TcpSocket *socket ,
 	//////
 
 	if ( fmt == FMT_JSON )
-		sb.safePrintf("{\"crawls\":[");//\"collections\":");
+		sb.safePrintf("\"crawls\":[");//\"collections\":");
 
 	long summary = hr->getLong("summary",0);
 	// enter summary mode for json
@@ -2692,7 +2695,7 @@ bool printCrawlBotPage2 ( TcpSocket *socket ,
 
 	if ( fmt == FMT_JSON ) 
 		// end the array of collection objects
-		sb.safePrintf("\n]}\n");
+		sb.safePrintf("\n]\n");
 
 	///////
 	//
@@ -3363,6 +3366,9 @@ bool printCrawlBotPage2 ( TcpSocket *socket ,
 	}
 
 
+	// the ROOT JSON }
+	if ( fmt == FMT_JSON )
+		sb.safePrintf("}\n");
 
 	char *ct = "text/html";
 	if ( fmt == FMT_JSON ) ct = "application/json";

XmlDoc.cpp

@@ -180,6 +180,10 @@ static long long s_lastTimeStart = 0LL;
 
 void XmlDoc::reset ( ) {
 
+	m_mySiteLinkInfoBuf.purge();
+	m_myPageLinkInfoBuf.purge();
+	m_myTempLinkInfoBuf.purge();
+
 	// reset count for nukeJSONObjects() function
 	m_joc = 0;
 
@@ -489,12 +493,14 @@ void XmlDoc::reset ( ) {
 	if ( m_ahrefsDocValid  ) nukeDoc ( m_ahrefsDoc  );
 
 	if ( m_linkInfo1Valid && ptr_linkInfo1 && m_freeLinkInfo1 ) {
-		mfree ( ptr_linkInfo1 , size_linkInfo1, "LinkInfo1");
+		// it now points into m_myPageLinkInfoBuf !
+		//mfree ( ptr_linkInfo1 , size_linkInfo1, "LinkInfo1");
 		ptr_linkInfo1    = NULL;
 		m_linkInfo1Valid = false;
 	}
 	if ( m_linkInfo2Valid && ptr_linkInfo2 && m_freeLinkInfo2 ) {
-		mfree ( ptr_linkInfo2 , size_linkInfo2, "LinkInfo2");
+		// should point into a safebuf as well
+		//mfree ( ptr_linkInfo2 , size_linkInfo2, "LinkInfo2");
 		ptr_linkInfo2    = NULL;
 		m_linkInfo1Valid = false;
 	}
@@ -11302,7 +11308,9 @@ LinkInfo *XmlDoc::getSiteLinkInfo() {
 
 	setStatus ( "getting site link info" );
 
-	if ( m_siteLinkInfoValid ) return m_msg25.m_linkInfo;
+	if ( m_siteLinkInfoValid ) 
+		//return msg25.m_linkInfo;
+		return (LinkInfo *)m_mySiteLinkInfoBuf.getBufStart();
 	char *mysite = getSite();
 	if ( ! mysite || mysite == (void *)-1 ) return (LinkInfo *)mysite;
 	long *fip = getFirstIp();
@@ -11366,7 +11374,9 @@ LinkInfo *XmlDoc::getSiteLinkInfo() {
 				onlyNeedGoodInlinks ,
 				false,
 				0,
-				0) )
+				0,
+				// it will store the linkinfo into this safebuf
+				&m_mySiteLinkInfoBuf) )
 		// return -1 if it blocked
 		return (LinkInfo *)-1;
 	// sanity check
@@ -12171,7 +12181,8 @@ LinkInfo *XmlDoc::getLinkInfo1 ( ) {
 					onlyNeedGoodInlinks ,
 					false, // getlinkertitles
 					0, // ourhosthash32 (special)
-					0  // ourdomhash32 (special)
+					0,  // ourdomhash32 (special)
+					&m_myPageLinkInfoBuf
 					) )
 			// blocked
 			return (LinkInfo *)-1;
@@ -12185,7 +12196,8 @@ LinkInfo *XmlDoc::getLinkInfo1 ( ) {
 
 	// at this point assume its valid
 	m_linkInfo1Valid = true;
-	// get the link info we got set
+	// . get the link info we got set
+	// . this ptr references into m_myPageLinkInfoBuf safebuf
 	ptr_linkInfo1  = m_msg25.m_linkInfo;
 	size_linkInfo1 = m_msg25.m_linkInfo->getSize();
 	// we should free it
@@ -41359,7 +41371,8 @@ Msg25 *XmlDoc::getAllInlinks ( bool forSite ) {
 					      false ,//onlyneedgoodinlinks?
 					      false,//getlinkertitles?
 					      0, // ourhosthash32 (special)
-					      0)) // ourdomhash32 (special)
+					      0, // ourdomhash32 (special)
+					      &m_myTempLinkInfoBuf ) )
 			// blocked?
 			return (Msg25 *)-1;
 	}

XmlDoc.h

@@ -1423,6 +1423,9 @@ class XmlDoc {
 	uint8_t m_siteNumInlinks8;
 	//long m_siteNumInlinks;
 	LinkInfo m_siteLinkInfo;
+	SafeBuf m_mySiteLinkInfoBuf;
+	SafeBuf m_myPageLinkInfoBuf;
+	SafeBuf m_myTempLinkInfoBuf;
 	char m_isInjecting;
 	char m_useFakeMime;
 	char m_useSiteLinkBuf;