Factor out same-ip-network checks in Linkdb.cpp

Implemented is_same_network_linkwise() and used that instead of having masking
with the magic constanct 0x0000ffff spread out all over the code.

IPAddressChecks.cpp

@@ -145,3 +145,43 @@ bool is_trusted_protocol_ip(uint32_t ip/*network-order*/)
 	//probably not an intranet host, so we err on the side of caution
 	return false;
 }
+
+
+//This is a replacement for the hardcoded checks on the /16 prefix of the ip addresses
+//This is slightly more intelligent.
+//The correct solution would be to have access to whois information and check the
+//identity of the owner, and have special cases for all hosting providers.
+bool is_same_network_linkwise(uint32_t ip_a/*network order*/, uint32_t ip_b/*network order*/)
+{
+	if(ip_a==ip_b) {
+		//same IP - same network
+		//This is inaccurate for some hosting providers, apache with virtual
+		//domains and general frontends eg. cloudflare. It is widly inaccurate for CDNs
+		//but there are rarely permalinks to CNDs so it's not a major problem.
+		return true;
+	}
+	if((ip_a&0x000000ff)!=(ip_b&0x000000ff)) {
+		//Top 8 bit are different so they are not even on the same class-A network. So must be different.
+		//This fails if a company has a fragmented address space due to mergers. Eg. HP now owns Compaqs
+		//networks and Compaq owned DEC's networks, so ... quite messy and inexact. But better than nothing.
+		return false;
+	}
+	//make educated guesses for some of the known class-A owners
+	//from the test above we know that the top 8 bites are quals
+	if((ip_a&0xff)==3)
+		return true; //General Electric
+	if((ip_a&0xff)==6)
+		return true; //DoD
+	if((ip_a&0xff)==7)
+		return true; //DoD
+	if((ip_a&0xff)==9)
+		return true; //IBM
+	if((ip_a&0xff)==11)
+		return true; //DoD
+	if((ip_a&0xff)==15)
+		return true; //HP
+	//make a wild guess
+	if((ip_a&0x0000ffff)==(ip_b&0x0000ffff)) //same /16 net?
+		return true;
+	return false;
+}

IPAddressChecks.h

@@ -19,5 +19,8 @@ bool is_internal_net_ip(uint32_t ip/*network-order*/);
 //Is the IP an internal-like that we allow to use the udp protocol without being registered as part of the cluster?
 bool is_trusted_protocol_ip(uint32_t ip/*network-order*/);
 
+//Make a guess if the two IPs are on the same network / controlled by the same
+//entity and links between them should be treated as internal links
+bool is_same_network_linkwise(uint32_t ip_a/*network order*/, uint32_t ip_b/*network order*/);
 
 #endif // GB_IPADDRESSCHECKS_H

Linkdb.cpp

@@ -7,6 +7,7 @@
 #include "Rebalance.h"
 #include "Process.h"
 #include "HashTable.h"
+#include "IPAddressChecks.h"
 #ifdef _VALGRIND_
 #include <valgrind/memcheck.h>
 #endif
@@ -2537,9 +2538,9 @@ bool Msg25::gotLinkText ( Msg20Request *req ) { // LinkTextReply *linkText ) {
 		Msg20Reply *r = m_replyPtrs[i];
 		// are we internal
 		bool internal = false;
-		if ( (r->m_ip&0x0000ffff) == (m_ip&0x0000ffff) ) 
+		if ( is_same_network_linkwise(r->m_ip,m_ip) )
 			internal = true;
-		if ( (r->m_firstIp&0x0000ffff) == (m_ip&0x0000ffff))
+		if ( is_same_network_linkwise(r->m_firstIp,m_ip))
 			internal = true;
 		// the "external" string
 		//char *ext = "Y"; if ( internal ) ext = "N";
@@ -3001,9 +3002,9 @@ static LinkInfo *makeLinkInfo ( const char        *coll                    ,
 		if ( ! replies[i] ) continue;
 		//if ( texts[i]->m_errno ) continue;
 		bool internal = false;
-		if ( (replies[i]->m_ip&0x0000ffff) == (ip&0x0000ffff) )
+		if ( is_same_network_linkwise(replies[i]->m_ip,ip) )
 			internal = true;
-		if ( (replies[i]->m_firstIp&0x0000ffff) == (ip&0x0000ffff) )
+		if ( is_same_network_linkwise(replies[i]->m_firstIp,ip) )
 			internal = true;
 		if ( internal )
 			icount++;
@@ -3041,9 +3042,9 @@ static LinkInfo *makeLinkInfo ( const char        *coll                    ,
 		}
 
 		bool internal = false;
-		if ((r->m_ip&0x0000ffff) == (ip & 0x0000ffff)) 
+		if ( is_same_network_linkwise(r->m_ip,ip))
 			internal = true;
-		if ((r->m_firstIp&0x0000ffff) == (ip & 0x0000ffff)) 
+		if ( is_same_network_linkwise(r->m_firstIp,ip))
 			internal = true;
 
 		// if its internal do not count towards good, but do
@@ -3136,9 +3137,9 @@ static LinkInfo *makeLinkInfo ( const char        *coll                    ,
 		if ( r->m_isLinkSpam && onlyNeedGoodInlinks ) continue;
 		// are we internal?
 		bool internal = false;
-		if ( (r->m_ip&0x0000ffff) == (ip & 0x0000ffff) ) 
+		if ( is_same_network_linkwise(r->m_ip,ip) )
 			internal = true;
-		if ( (r->m_firstIp&0x0000ffff) == (ip & 0x0000ffff) )
+		if ( is_same_network_linkwise(r->m_firstIp,ip) )
 			internal = true;
 		if ( internal ) icount3++;
 		// set the Inlink