Generalized comparator, fixed some bugs in record length handling.

git-svn-id: svn+ssh://svn.corp.yahoo.com/yahoo/yrl/labs/pnuts/code/logstore@613 8dad8b1f-cf64-0410-95b6-bcf113ffbcfe

datapage.cpp

@@ -276,7 +276,7 @@ bool DataPage<TUPLE>::recordRead(typename TUPLE::key_t key, size_t keySize,  TUP
     int match = -1;
     while((*buf=itr.getnext()) != 0)
         {
-            match = TUPLE::compare((*buf)->key(), key);
+	  match = TUPLE::compare((*buf)->key(), (*buf)->keylen(), key, keySize);
             
             if(match<0) //keep searching
             {

datatuple.h

@@ -49,7 +49,7 @@ public:
 
     //this is used by the stl set
     bool operator() (const datatuple* lhs, const datatuple* rhs) const {
-		return compare(lhs->key(), rhs->key()) < 0; //strcmp((char*)lhs.key(),(char*)rhs.key()) < 0;
+		return compare(lhs->key(), lhs->keylen(), rhs->key(), rhs->keylen()) < 0; //strcmp((char*)lhs.key(),(char*)rhs.key()) < 0;
 	}
 
     /**
@@ -57,14 +57,27 @@ public:
      * 0 if k1 == k2
      * 1 of k1 > k2
     **/
-    static int compare(const byte* k1,const byte* k2) {
+    static int compare(const byte* k1,size_t k1l, const byte* k2, size_t k2l) {
-		// XXX string comparison is probably not the right approach.
+		// This function handles ASCII and UTF-8 correctly.
-		//for char* ending with \0
+    	// It also handles a Sherpa LSM-Tree specific encoding, where multiple utf-8 strings
-		return strcmp((char*)k1,(char*)k2);
+    	// are concatenated with \254., and the \254 is replaced with a \255 for 'max value in range'.
+
+    	size_t min_l = k1l < k2l ? k1l : k2l;
+
+    	int ret = memcmp(k1,k2, min_l);
+    	if(ret)        return ret;
+    	if(k1l < k2l)  return -1;
+    	if(k1l == k2l) return 0;
+    	return 1;
+
+		//for testing with char* ending with \0
+    	/*assert(strlen((char*)k1) == k1l - 1);
+		  assert(strlen((char*)k2) == k2l - 1);
+	      return strcmp((char*)k1,(char*)k2); */
 	}
 
     static int compare_obj(const datatuple * a, const datatuple* b) {
-    	return compare(a->key(), b->key());
+    	return compare(a->key(), a->keylen(), b->key(), b->keylen());
     }
 
     inline void setDelete() {

diskTreeComponent.cpp

@@ -726,7 +726,8 @@ recordid diskTreeComponent::lookup(int xid,
     for(int i = FIRST_SLOT+1; i < *stasis_page_slotted_numslots_ptr(node); i++)
     {
         rec = (const indexnode_rec*)readRecord(xid,node,i,0);
-        int cmpval = datatuple::compare((datatuple::key_t) (rec+1),(datatuple::key_t) key);
+        int cmpval = datatuple::compare((datatuple::key_t) (rec+1), *stasis_page_slotted_slot_length_ptr(node, i)-sizeof(*rec),
+					(datatuple::key_t) key, keySize);
         if(cmpval>0) //changed it from >
             break;
         match = i;
@@ -738,7 +739,7 @@ recordid diskTreeComponent::lookup(int xid,
         pageid_t child_id = ((const indexnode_rec*)readRecord(xid,node,match,0))->ptr;
         Page* child_page = loadPage(xid, child_id);
         readlock(child_page->rwlatch,0);
-        recordid ret = lookup(xid,child_page,depth-1,key,0);
+        recordid ret = lookup(xid,child_page,depth-1,key,keySize);
         unlock(child_page->rwlatch);
         releasePage(child_page);
         return ret;

logserver.cpp

@@ -263,6 +263,7 @@ void logserver::eventLoop()
                     //wake up the thread to do work
                     pthread_mutex_lock(idle_th.data->th_mut);
                     //set the job of the idle thread
+		    assert(currsock != -1);
                     *(idle_th.data->workitem) = currsock;
                     pthread_cond_signal(idle_th.data->th_cond);
                     pthread_mutex_unlock(idle_th.data->th_mut);
@@ -362,9 +363,10 @@ void *serverLoop(void *args)
         //insert the given element to the ready queue
         sdata->ready_queue->push(newsockfd);
 
-        if(sdata->ready_queue->size() == 1) //signal the event loop
+   /*     if(sdata->ready_queue->size() == 1) //signal the event loop
             pthread_cond_signal(sdata->selcond);
-        
+        else */ if(sdata->ready_queue->size()) //in case multiple events were received in race.
+        	pthread_cond_broadcast(sdata->selcond);
         pthread_mutex_unlock(sdata->qlock);
     }
 }
@@ -445,6 +447,10 @@ void * thread_work_fn( void * args)
         if(!err) { tuple2 = readtuplefromsocket(*(item->data->workitem), &err); }
         //step 3: process the tuple
 
+	if(tuple) {
+	  printf("Tuple req = %d key = >%s<\n", opcode, tuple->key());
+	}
+
         if(opcode == OP_INSERT)
         {
             //insert/update/delete
@@ -464,7 +470,7 @@ void * thread_work_fn( void * args)
             } else if( dt->datalen() != 1024) {
                 DEBUG("data len for\t%s:\t%d\n", datatuple::key_to_str(tuple.key()).c_str(),
                        dt->datalen);
-                if(datatuple::compare(tuple->key(), dt->key()) != 0) {
+                if(datatuple::compare(tuple->key(), tuple->keylen(), dt->key(), dt->keylen()) != 0) {
                     DEBUG("key not equal:\t%s\t%s\n", datatuple::key_to_str(tuple.key()).c_str(),
                            datatuple::key_to_str(dt->key).c_str());
                 }
@@ -481,7 +487,7 @@ void * thread_work_fn( void * args)
             } else {
             	dt_needs_free = true;
             }
-
+	    fprintf(stderr, "find result: %s\n", dt->isDelete() ? "not found" : "found");
             //send the reply code
             int err = writeoptosocket(*(item->data->workitem), LOGSTORE_RESPONSE_SENDING_TUPLES);
             if(!err) {

merger.cpp

@@ -488,7 +488,7 @@ int64_t merge_iterators(int xid,
         DEBUG("tuple\t%lld: keylen %d datalen %d\n",
                ntuples, *(t2->keylen),*(t2->datalen) );        
 
-        while(t1 != 0 && datatuple::compare(t1->key(), t2->key()) < 0) // t1 is less than t2
+        while(t1 != 0 && datatuple::compare(t1->key(), t1->keylen(), t2->key(), t2->keylen()) < 0) // t1 is less than t2
         {
             //insert t1
             dp = insertTuple(xid, dp, t1, ltable, scratch_tree,
@@ -500,7 +500,7 @@ int64_t merge_iterators(int xid,
             t1 = itrA->getnext();
         }
 
-        if(t1 != 0 && datatuple::compare(t1->key(), t2->key()) == 0)
+        if(t1 != 0 && datatuple::compare(t1->key(), t1->keylen(), t2->key(), t2->keylen()) == 0)
         {
             datatuple *mtuple = ltable->gettuplemerger()->merge(t1,t2);