diff reporter.h @ 277:abfb26e08d9c audiodb-debian

Merge trunk changes -r326:386 into audiodb-debian branch.

Plus new debian/changelog version.  (Should have used an epoch really,
but couldn't be bothered; TODO: work out a sane version numbering
policy).

--- a/reporter.h	Mon Dec 17 16:44:37 2007 +0000
+++ b/reporter.h	Tue Jul 01 09:12:40 2008 +0000
@@ -1,5 +1,6 @@
 #include <utility>
 #include <queue>
+#include <deque>
 #include <set>
 #include <functional>
 
@@ -88,7 +89,7 @@
   if(adbQueryResponse==0) {
     for(rit = v.rbegin(); rit < v.rend(); rit++) {
       r = *rit;
-      std::cout << fileTable + r.trackID*O2_FILETABLESIZE << " ";
+      std::cout << fileTable + r.trackID*O2_FILETABLE_ENTRY_SIZE << " ";
       std::cout << r.dist << " " << r.qpos << " " << r.spos << std::endl;
     }
   } else {
@@ -107,7 +108,7 @@
       adbQueryResponse->result.Dist[k] = r.dist;
       adbQueryResponse->result.Qpos[k] = r.qpos;
       adbQueryResponse->result.Spos[k] = r.spos;
-      snprintf(adbQueryResponse->result.Rlist[k], O2_MAXFILESTR, "%s", fileTable+r.trackID*O2_FILETABLESIZE);
+      snprintf(adbQueryResponse->result.Rlist[k], O2_MAXFILESTR, "%s", fileTable+r.trackID*O2_FILETABLE_ENTRY_SIZE);
     }
   }
 }
@@ -118,7 +119,7 @@
   ~trackAveragingReporter();
   void add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist);
   void report(char *fileTable, adb__queryResponse *adbQueryResponse);
- private:
+ protected:
   unsigned int pointNN;
   unsigned int trackNN;
   unsigned int numFiles;
@@ -189,7 +190,7 @@
   if(adbQueryResponse==0) {
     for(rit = v.rbegin(); rit < v.rend(); rit++) {
       r = *rit;
-      std::cout << fileTable + r.trackID*O2_FILETABLESIZE << " ";
+      std::cout << fileTable + r.trackID*O2_FILETABLE_ENTRY_SIZE << " ";
       std::cout << r.dist << " " << r.qpos << " " << r.spos << std::endl;
     }
   } else {
@@ -208,18 +209,20 @@
       adbQueryResponse->result.Dist[k] = r.dist;
       adbQueryResponse->result.Qpos[k] = r.qpos;
       adbQueryResponse->result.Spos[k] = r.spos;
-      snprintf(adbQueryResponse->result.Rlist[k], O2_MAXFILESTR, "%s", fileTable+r.trackID*O2_FILETABLESIZE);
+      snprintf(adbQueryResponse->result.Rlist[k], O2_MAXFILESTR, "%s", fileTable+r.trackID*O2_FILETABLE_ENTRY_SIZE);
     }
   }
 }
 
+// track Sequence Query Radius Reporter
+// only return tracks and retrieved point counts
 class trackSequenceQueryRadReporter : public Reporter { 
 public:
   trackSequenceQueryRadReporter(unsigned int trackNN, unsigned int numFiles);
   ~trackSequenceQueryRadReporter();
   void add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist);
   void report(char *fileTable, adb__queryResponse *adbQueryResponse);
-private:
+ protected:
   unsigned int trackNN;
   unsigned int numFiles;
   std::set<std::pair<unsigned int, unsigned int> > *set;
@@ -242,11 +245,11 @@
 
 void trackSequenceQueryRadReporter::add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist) {
   std::set<std::pair<unsigned int, unsigned int> >::iterator it;
-  std::pair<unsigned int, unsigned int> pair = std::make_pair(trackID, qpos);
+  std::pair<unsigned int, unsigned int> pair = std::make_pair(trackID, qpos); // only count this once
   it = set->find(pair);
   if (it == set->end()) {
     set->insert(pair);
-    count[trackID]++;
+    count[trackID]++; // only count if <tackID,qpos> pair is unique
   }
 }
 
@@ -279,9 +282,297 @@
   if(adbQueryResponse==0) {
     for(rit = v.rbegin(); rit < v.rend(); rit++) {
       r = *rit;
-      std::cout << fileTable + r.trackID*O2_FILETABLESIZE << " " << r.count << std::endl;
+      std::cout << fileTable + r.trackID*O2_FILETABLE_ENTRY_SIZE << " " << r.count << std::endl;
     }
   } else {
     // FIXME
   }
 }
+
+// Another type of trackAveragingReporter that reports all pointNN nearest neighbours
+template <class T> class trackSequenceQueryNNReporter : public trackAveragingReporter<T> {
+ protected:
+  using trackAveragingReporter<T>::numFiles;
+  using trackAveragingReporter<T>::queues;
+  using trackAveragingReporter<T>::trackNN;
+  using trackAveragingReporter<T>::pointNN;
+ public:
+  trackSequenceQueryNNReporter(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles);
+  void report(char *fileTable, adb__queryResponse *adbQueryResponse);
+};
+
+template <class T> trackSequenceQueryNNReporter<T>::trackSequenceQueryNNReporter(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles)
+:trackAveragingReporter<T>(pointNN, trackNN, numFiles){}
+
+template <class T> void trackSequenceQueryNNReporter<T>::report(char *fileTable, adb__queryResponse *adbQueryResponse) {
+  std::priority_queue < NNresult, std::vector< NNresult>, T> result;
+  std::priority_queue< NNresult, std::vector< NNresult>, std::less<NNresult> > *point_queues = new std::priority_queue< NNresult, std::vector< NNresult>, std::less<NNresult> >[numFiles];
+  
+  for (int i = numFiles-1; i >= 0; i--) {
+    unsigned int size = queues[i].size();
+    if (size > 0) {
+      NNresult r;
+      double dist = 0;
+      NNresult oldr = queues[i].top();
+      for (unsigned int j = 0; j < size; j++) {
+        r = queues[i].top();
+        dist += r.dist;
+	point_queues[i].push(r);
+	queues[i].pop();
+        if (r.dist == oldr.dist) {
+          r.qpos = oldr.qpos;
+          r.spos = oldr.spos;
+        } else {
+          oldr = r;
+        }
+      }
+      dist /= size;
+      r.dist = dist; // trackID, qpos and spos are magically right already.
+      result.push(r);
+      if (result.size() > trackNN) {
+        result.pop();
+      }
+    }
+  }
+
+  NNresult r;
+  std::vector<NNresult> v;
+  unsigned int size = result.size();
+  for(unsigned int k = 0; k < size; k++) {
+    r = result.top();
+    v.push_back(r);
+    result.pop();
+  }
+  std::vector<NNresult>::reverse_iterator rit;
+  std::priority_queue< NNresult, std::vector< NNresult>, std::greater<NNresult> > point_queue;      
+
+  if(adbQueryResponse==0) {
+    for(rit = v.rbegin(); rit < v.rend(); rit++) {
+      r = *rit;
+      std::cout << fileTable + r.trackID*O2_FILETABLE_ENTRY_SIZE << " " << r.dist << std::endl;
+      unsigned int qsize = point_queues[r.trackID].size();
+      // Reverse the order of the points stored in point_queues
+      for(unsigned int k=0; k < qsize; k++){
+	point_queue.push( point_queues[r.trackID].top() );
+	point_queues[r.trackID].pop();
+      }
+      
+      for(unsigned int k = 0; k < qsize; k++) {
+	NNresult rk = point_queue.top();
+	std::cout << rk.dist << " " << rk.qpos << " " << rk.spos << std::endl;
+	point_queue.pop();
+      }
+    }
+  } else {
+    adbQueryResponse->result.__sizeRlist=size;
+    adbQueryResponse->result.__sizeDist=size;
+    adbQueryResponse->result.__sizeQpos=size;
+    adbQueryResponse->result.__sizeSpos=size;
+    adbQueryResponse->result.Rlist= new char*[size];
+    adbQueryResponse->result.Dist = new double[size];
+    adbQueryResponse->result.Qpos = new unsigned int[size];
+    adbQueryResponse->result.Spos = new unsigned int[size];
+    unsigned int k = 0;
+    for(rit = v.rbegin(); rit < v.rend(); rit++, k++) {
+      r = *rit;
+      adbQueryResponse->result.Rlist[k] = new char[O2_MAXFILESTR];
+      adbQueryResponse->result.Dist[k] = r.dist;
+      adbQueryResponse->result.Qpos[k] = r.qpos;
+      adbQueryResponse->result.Spos[k] = r.spos;
+      snprintf(adbQueryResponse->result.Rlist[k], O2_MAXFILESTR, "%s", fileTable+r.trackID*O2_FILETABLE_ENTRY_SIZE);
+    }
+  }
+  // clean up
+  delete[] point_queues;
+}
+
+
+// track Sequence Query Radius NN Reporter
+// retrieve tracks ordered by query-point matches (one per track per query point)
+//
+// as well as sorted n-NN points per retrieved track
+class trackSequenceQueryRadNNReporter : public Reporter { 
+public:
+  trackSequenceQueryRadNNReporter(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles);
+  ~trackSequenceQueryRadNNReporter();
+  void add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist);
+  void report(char *fileTable, adb__queryResponse *adbQueryResponse);
+ protected:
+  unsigned int pointNN;
+  unsigned int trackNN;
+  unsigned int numFiles;
+  std::set< NNresult > *set;
+  std::priority_queue< NNresult, std::vector< NNresult>, std::less<NNresult> > *point_queues;
+  unsigned int *count;
+};
+
+trackSequenceQueryRadNNReporter::trackSequenceQueryRadNNReporter(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles):
+pointNN(pointNN), trackNN(trackNN), numFiles(numFiles) {
+  // Where to count Radius track matches (one-to-one)
+  set = new std::set< NNresult >; 
+  // Where to insert individual point matches (one-to-many)
+  point_queues = new std::priority_queue< NNresult, std::vector< NNresult>, std::less<NNresult> >[numFiles];
+  
+  count = new unsigned int[numFiles];
+  for (unsigned i = 0; i < numFiles; i++) {
+    count[i] = 0;
+  }
+}
+
+trackSequenceQueryRadNNReporter::~trackSequenceQueryRadNNReporter() {
+  delete set;
+  delete [] count;
+}
+
+void trackSequenceQueryRadNNReporter::add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist) {
+  std::set< NNresult >::iterator it;
+  NNresult r;
+  r.trackID = trackID;
+  r.qpos = qpos;
+  r.dist = dist;
+  r.spos = spos;
+
+  // Record all matching points (within radius)
+  if (!isnan(dist)) {
+    point_queues[trackID].push(r);
+    if(point_queues[trackID].size() > pointNN)
+      point_queues[trackID].pop();
+  }
+
+  // Record counts of <trackID,qpos> pairs
+  it = set->find(r);
+  if (it == set->end()) {
+    set->insert(r);
+    count[trackID]++;
+  }
+}
+
+void trackSequenceQueryRadNNReporter::report(char *fileTable, adb__queryResponse *adbQueryResponse) {
+  std::priority_queue < Radresult > result;
+  // KLUDGE: doing this backwards in an attempt to get the same
+  // tiebreak behaviour as before.
+  for (int i = numFiles-1; i >= 0; i--) {
+    Radresult r;
+    r.trackID = i;
+    r.count = count[i];
+    if(r.count > 0) {
+      result.push(r);
+      if (result.size() > trackNN) {
+        result.pop();
+      }
+    }
+  }
+
+  Radresult r;
+  std::vector<Radresult> v;
+  unsigned int size = result.size();
+  for(unsigned int k = 0; k < size; k++) {
+    r = result.top();
+    v.push_back(r);
+    result.pop();
+  }
+
+
+  // Traverse tracks in descending order of count cardinality
+  std::vector<Radresult>::reverse_iterator rit;
+  std::priority_queue< NNresult, std::vector< NNresult>, std::greater<NNresult> > point_queue;
+
+  if(adbQueryResponse==0) {
+    for(rit = v.rbegin(); rit < v.rend(); rit++) {
+      r = *rit;
+      std::cout << fileTable + r.trackID*O2_FILETABLE_ENTRY_SIZE << " " << r.count << std::endl;
+
+      // Reverse the order of the points stored in point_queues
+      unsigned int qsize=point_queues[r.trackID].size();
+      for(unsigned int k=0; k < qsize; k++){
+	point_queue.push(point_queues[r.trackID].top());
+	point_queues[r.trackID].pop();
+      }
+
+      for(unsigned int k=0; k < qsize; k++){
+	NNresult rk = point_queue.top();
+	std::cout << rk.dist << " " << rk.qpos << " " << rk.spos << std::endl;
+	point_queue.pop();
+      }
+    }
+  } else {
+    // FIXME
+  }
+  delete[] point_queues;
+}
+
+
+/********** ONE-TO-ONE REPORTERS *****************/
+
+// track Sequence Query Radius NN Reporter One-to-One
+// for each query point find the single best matching target point in all database
+// report qpos, spos and trackID
+class trackSequenceQueryRadNNReporterOneToOne : public Reporter { 
+public:
+  trackSequenceQueryRadNNReporterOneToOne(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles);
+  ~trackSequenceQueryRadNNReporterOneToOne();
+  void add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist);
+  void report(char *fileTable, adb__queryResponse *adbQueryResponse);
+ protected:
+  unsigned int pointNN;
+  unsigned int trackNN;
+  unsigned int numFiles;
+  std::set< NNresult > *set;
+  std::vector< NNresult> *point_queue;
+  unsigned int *count;
+
+};
+
+trackSequenceQueryRadNNReporterOneToOne::trackSequenceQueryRadNNReporterOneToOne(unsigned int pointNN, unsigned int trackNN, unsigned int numFiles):
+pointNN(pointNN), trackNN(trackNN), numFiles(numFiles) {
+  // Where to count Radius track matches (one-to-one)
+  set = new std::set< NNresult >; 
+  // Where to insert individual point matches (one-to-many)
+  point_queue = new std::vector< NNresult >;
+  
+  count = new unsigned int[numFiles];
+  for (unsigned i = 0; i < numFiles; i++) {
+    count[i] = 0;
+  }
+}
+
+trackSequenceQueryRadNNReporterOneToOne::~trackSequenceQueryRadNNReporterOneToOne() {
+  delete set;
+  delete [] count;
+}
+
+void trackSequenceQueryRadNNReporterOneToOne::add_point(unsigned int trackID, unsigned int qpos, unsigned int spos, double dist) {
+  std::set< NNresult >::iterator it;
+  NNresult r;
+
+  r.qpos = qpos;
+  r.trackID = trackID;
+  r.spos = spos;
+  r.dist = dist;
+
+  if(point_queue->size() < r.qpos + 1){
+    point_queue->resize( r.qpos + 1 );
+    (*point_queue)[r.qpos].dist = 1e6;
+  }
+
+  if (r.dist < (*point_queue)[r.qpos].dist)
+    (*point_queue)[r.qpos] = r;
+
+}
+
+void trackSequenceQueryRadNNReporterOneToOne::report(char *fileTable, adb__queryResponse *adbQueryResponse) {
+  if(adbQueryResponse==0) {
+    std::vector< NNresult >::iterator vit;
+    NNresult rk;
+    for( vit = point_queue->begin() ; vit < point_queue->end() ; vit++ ){
+      rk = *vit;
+      std::cout << rk.dist << " " 
+		<< rk.qpos << " " 
+		<< rk.spos << " " 
+		<< fileTable + rk.trackID*O2_FILETABLE_ENTRY_SIZE 
+		<< std::endl;
+      }
+  } else {
+    // FIXME
+  }
+}