Index: /branches/eam_branches/ipp-20130207/Ohana/src/delstar/Makefile
===================================================================
--- /branches/eam_branches/ipp-20130207/Ohana/src/delstar/Makefile	(revision 35228)
+++ /branches/eam_branches/ipp-20130207/Ohana/src/delstar/Makefile	(revision 35229)
@@ -29,4 +29,5 @@
 $(SRC)/delete_imagename.$(ARCH).o \
 $(SRC)/delete_imagefile.$(ARCH).o \
+$(SRC)/delete_duplicate_images.$(ARCH).o \
 $(SRC)/delete_image_photcodes.$(ARCH).o \
 $(SRC)/delete_photcodes.$(ARCH).o \
@@ -45,4 +46,5 @@
 $(SRC)/ConfigInit.$(ARCH).o 	  \
 $(SRC)/args.$(ARCH).o	          \
+$(SRC)/delete_duplicate_images.$(ARCH).o \
 $(SRC)/delete_photcodes.$(ARCH).o \
 $(SRC)/delete_photcodes_single.$(ARCH).o \
Index: /branches/eam_branches/ipp-20130207/Ohana/src/delstar/include/delstar.h
===================================================================
--- /branches/eam_branches/ipp-20130207/Ohana/src/delstar/include/delstar.h	(revision 35228)
+++ /branches/eam_branches/ipp-20130207/Ohana/src/delstar/include/delstar.h	(revision 35229)
@@ -2,4 +2,7 @@
 # include <dvo.h>
 # include <signal.h>
+
+// options for generating the IndexArray used to select images for deletion (delete_duplicate_images.c)
+enum {NONE, EXTERN_ID, IMAGE_ID};
 
 typedef struct {
@@ -10,4 +13,11 @@
   double Area, density, spacing;
 } CatStats;
+
+typedef struct {
+  off_t minID;
+  off_t maxID;
+  off_t range;
+  off_t *value;
+} IndexArray;
 
 /* global variables set in parameter file */
@@ -49,5 +59,5 @@
 
 int    MODE;
-enum {MODE_NONE, MODE_IMAGENAME, MODE_IMAGEFILE, MODE_TIME, MODE_ORPHAN, MODE_MISSED, MODE_PHOTCODES};
+enum {MODE_NONE, MODE_IMAGENAME, MODE_IMAGEFILE, MODE_TIME, MODE_ORPHAN, MODE_MISSED, MODE_PHOTCODES, MODE_DUP_IMAGES};
 
 char DateKeyword[64], DateMode[64], UTKeyword[64], MJDKeyword[64], JDKeyword[64];
@@ -101,2 +111,11 @@
 int delete_image_photcodes (FITS_DB *db);
 int delete_photcodes_single (char *cptname);
+
+int delete_duplicate_images (int hostID, char *hostpath, FITS_DB *db);
+int delete_duplicate_image_measures (int hostID, char *hostpath, IndexArray *imageID);
+int delete_duplicate_image_measures_parallel (SkyList *sky);
+int delete_duplicate_image_measures_catalog (Catalog *catalog, IndexArray *imageID);
+
+IndexArray *find_duplicates (Image *image, off_t Nimage);
+IndexArray *make_index_array (Image *image, off_t Nimage, int mode);
+
Index: /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/args.c
===================================================================
--- /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/args.c	(revision 35228)
+++ /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/args.c	(revision 35229)
@@ -126,4 +126,9 @@
     if (MODE != MODE_NONE) usage();
     MODE = MODE_MISSED;
+    remove_argument (N, &argc, argv);
+  }
+  if ((N = get_argument (argc, argv, "-dup-images"))) {
+    if (MODE != MODE_NONE) usage();
+    MODE = MODE_DUP_IMAGES;
     remove_argument (N, &argc, argv);
   }
Index: /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delete_duplicate_images.c
===================================================================
--- /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delete_duplicate_images.c	(revision 35229)
+++ /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delete_duplicate_images.c	(revision 35229)
@@ -0,0 +1,456 @@
+# include "delstar.h"
+
+// this function identifies the images to be deleted based on duplication of the 
+// externID values (ex 0).  The result is an array of imageID values to be deleted
+
+// find & delete duplicate images
+// duplicates based on externID (skip 0s)
+int delete_duplicate_images (int hostID, char *hostpath, FITS_DB *db) {
+
+  off_t Nimage;
+  Image *image;
+
+  image = gfits_table_get_Image (&db[0].ftable, &Nimage, &db[0].swapped);
+  if (!image) {
+    fprintf (stderr, "ERROR: failed to read images\n");
+    exit (2);
+  }
+
+  IndexArray *imageID = find_duplicates (image, Nimage);
+
+  delete_duplicate_image_measures (hostID, hostpath, imageID);
+
+  return TRUE;
+}
+
+int delete_duplicate_image_measures (int hostID, char *hostpath, IndexArray *imageID) {
+
+  int i;
+  Catalog catalog;
+
+  // load the current sky table (layout of all SkyRegions) 
+  SkyTable *sky = SkyTableLoadOptimal (CATDIR, NULL, NULL, TRUE, -1, VERBOSE);
+  if (!sky) {
+    fprintf (stderr, "ERROR running loading sky table from %s\n", CATDIR);
+    exit (2);
+  }
+  SkyTableSetFilenames (sky, CATDIR, "cpt");
+
+  SkyList *skylist = SkyListByPatch (sky, -1, &UserPatch);
+  if (!skylist) {
+    fprintf (stderr, "ERROR setting up skylist for %s\n", CATDIR);
+    exit (2);
+  }
+
+  // launch the remote jobs
+  if (PARALLEL && !hostID) {
+    int status = delete_duplicate_image_measures_parallel (skylist);
+    return status;
+  }
+
+  // delete detections from region
+  for (i = 0; i < skylist[0].Nregions; i++) {
+
+    // does this host ID match the desired location for the table?
+    if (!HostTableTestHost(skylist[0].regions[i], hostID)) continue;
+
+    // set up the basic catalog info
+    char hostfile[1024];
+    snprintf (hostfile, 1024, "%s/%s.cpt", hostpath, skylist[0].regions[i]->name);
+    catalog.filename  = hostID ? hostfile : skylist[0].filename[i];
+    catalog.Nsecfilt  = GetPhotcodeNsecfilt ();
+    catalog.catflags = LOAD_AVES | LOAD_MEAS | LOAD_MISS | LOAD_SECF;
+
+    if (VERBOSE) fprintf (stderr, "deleting from %s\n", catalog.filename);
+
+    // an error exit status here is a significant error
+    if (!dvo_catalog_open (&catalog, skylist[0].regions[i], VERBOSE2, "a")) {
+      fprintf (stderr, "ERROR: failure to open catalog file %s\n", catalog.filename);
+      exit (2);
+    }
+    if (!catalog.Naves_disk) {
+      if (VERBOSE2) fprintf (stderr, "no data in %s, skipping\n", catalog.filename);
+      dvo_catalog_unlock (&catalog);
+      dvo_catalog_free (&catalog);
+      continue;
+    }
+
+    delete_duplicate_image_measures_catalog (&catalog, imageID);
+    dvo_catalog_save_complete (&catalog, VERBOSE2);
+    dvo_catalog_unlock (&catalog);
+    dvo_catalog_free (&catalog);
+  }
+  return TRUE;
+}
+
+// CATDIR is supplied globally
+int delete_duplicate_image_measures_parallel (SkyList *sky) {
+
+  // launch the delstar_client jobs to the parallel hosts
+
+  // load the list of hosts
+  HostTable *table = HostTableLoad (CATDIR, sky->hosts);
+  if (!table) {
+    fprintf (stderr, "ERROR: failure reading Host Table %s for database %s\n", sky->hosts, CATDIR);
+    exit (1);
+  }    
+
+  int i;
+  for (i = 0; i < table->Nhosts; i++) {
+
+    // ensure that the paths are absolute path names
+    char *tmppath = abspath (table->hosts[i].pathname, DVO_MAX_PATH);
+    free (table->hosts[i].pathname);
+    table->hosts[i].pathname = tmppath;
+
+    char command[1024];
+    snprintf (command, 1024, "delstar_client -hostID %d -D CATDIR %s -hostdir %s -region %f %f %f %f -dup-images", 
+	      table->hosts[i].hostID, CATDIR, table->hosts[i].pathname, 
+	      UserPatch.Rmin, UserPatch.Rmax, UserPatch.Dmin, UserPatch.Dmax);
+
+    char tmpline[1024];
+    if (VERBOSE)       { snprintf (tmpline, 1024, "%s -v",              command);                    strcpy (command, tmpline); }
+    if (VERBOSE2)      { snprintf (tmpline, 1024, "%s -vv",             command);                    strcpy (command, tmpline); }
+
+    fprintf (stderr, "command: %s\n", command);
+
+    fprintf (stderr, "parallel version is not ready (need to save / load the imageID array\n");
+
+    if (PARALLEL_MANUAL) continue;
+
+    if (PARALLEL_SERIAL) {
+      int status = system (command);
+      if (status) {
+	fprintf (stderr, "ERROR running delstar_client\n");
+	exit (2);
+      }
+    } else {
+      // launch the job on the remote machine (no handshake)
+      int errorInfo = 0;
+      int pid = rconnect ("ssh", table->hosts[i].hostname, command, table->hosts[i].stdio, &errorInfo, FALSE);
+      if (!pid) {
+	fprintf (stderr, "failure to start %s (error %d)\n", table->hosts[i].hostname, errorInfo);
+	exit (1);
+      }
+      table->hosts[i].pid = pid; // save for future reference
+    }
+  }
+
+  if (PARALLEL_MANUAL) {
+    fprintf (stderr, "run the client commands above.  when these are done, hit return\n");
+    getchar();
+  }
+  if (!PARALLEL_MANUAL && !PARALLEL_SERIAL) {
+    HostTableWaitJobsGetIO (table, __FILE__, __LINE__, VERBOSE);
+  }
+
+  return TRUE;
+}
+
+int delete_duplicate_image_measures_catalog (Catalog *catalog, IndexArray *imageID) {
+
+  off_t i, j, n, m, N, D, currentAve;
+
+  Measure *measureOut = NULL;
+  Average *averageOut = NULL;
+  SecFilt *secfiltOut = NULL;
+
+  off_t *measureDrop, *measureSeqRaw, *measureSeqOut, *measureRefOut, *measureAveRaw, *averageNmeas, *averageDmeas, *averageSeqOut, *averageRefOut, *measureAveOut;
+
+  int Nsecfilt = GetPhotcodeNsecfilt ();
+
+  /* internal counters */
+  off_t Nmeasure = catalog[0].Nmeasure;
+  off_t Naverage = catalog[0].Naverage;
+
+  Measure *measure = catalog[0].measure;
+  Average *average = catalog[0].average;
+  SecFilt *secfilt = catalog[0].secfilt;
+  
+  if (VERBOSE) fprintf (stderr, "starting with Nave, Nmeas: "OFF_T_FMT" "OFF_T_FMT"\n",  catalog[0].Naverage,  catalog[0].Nmeasure);
+
+  // we have a table of average objects and an unsorted table of measurements.  each measurement
+  // has a reference to the average object sequence (as well as an ID)
+  // measure[i].averef -> average[averef]
+  // measure[i].objID = average[averef].objID
+  // measure[i].catID = average[averef].catID
+
+  // we want a sorted measure array with all averef entries in sequence, skipping the entries which match our criteria
+
+  // arrays 
+  ALLOCATE (measureDrop,   off_t, Nmeasure);
+  ALLOCATE (measureSeqRaw, off_t, Nmeasure);
+  ALLOCATE (measureSeqOut, off_t, Nmeasure);
+  ALLOCATE (measureRefOut, off_t, Nmeasure);
+  ALLOCATE (measureAveRaw, off_t, Nmeasure);
+  ALLOCATE (averageNmeas,  off_t, Naverage);
+  ALLOCATE (averageDmeas,  off_t, Naverage);
+  ALLOCATE (averageSeqOut, off_t, Naverage);
+  ALLOCATE (averageRefOut, off_t, Naverage);
+
+  // mark the measures to be dropped
+  for (i = 0; i < Nmeasure; i++) {
+    if (measure[i].imageID == 0) continue;
+    off_t Ni = measure[i].imageID - imageID->minID;
+    myAssert (Ni >= 0, "oops");
+    myAssert (Ni < imageID->range, "oops");
+    measureDrop[i] = imageID->value[Ni];
+  }
+
+  // set up the measure sequence lists
+  for (i = 0; i < Nmeasure; i++) {
+    measureSeqRaw[i] = i;
+    measureAveRaw[i] = measure[i].averef;
+  }
+  // sort measureSeqRaw and measureAveRaw in order of measureAveRaw
+  SortAveMeasMatch(measureSeqRaw, measureAveRaw, Nmeasure);
+
+  // generate the measureSeqOut array
+  off_t NmeasOut = 0;
+  for (i = 0; i < Nmeasure; i++) {
+    j = measureSeqRaw[i];
+    if (measureDrop[j]) continue;
+    measureSeqOut[NmeasOut] = j;
+    measureRefOut[j] = NmeasOut;
+    NmeasOut ++;
+  }
+  // n = measureSeqOut[i] : measureOut[i] = measure[n]  (i = 0 -- NmeasOut, n = 0 -- Nmeasure)
+  // n = measureRefOut[i] : measureOut[n] = measure[i]
+  ALLOCATE (measureAveOut, off_t, NmeasOut);
+
+  // count the number of measures for each averef
+  N =  0;
+  D = -1;
+  currentAve = measureAveRaw[0];
+  for (i = 0; i < Nmeasure; i++) {
+    if (measureAveRaw[i] != currentAve) {
+      // we have hit the next entry in the list
+      averageNmeas[currentAve] = N; // number of measures
+      averageDmeas[currentAve] = D; // first measure 
+      N =  0;
+      D = -1;
+      currentAve = measureAveRaw[i];
+    }
+    j = measureSeqRaw[i];
+    if (measureDrop[j]) continue;
+    N++;
+    if (D == -1) { 
+      // first valid measureOut for this averef
+      D = measureRefOut[j];
+    }
+  }
+  averageNmeas[currentAve] = N; // number of measures
+  averageDmeas[currentAve] = D; // first measure 
+
+  // generate the new average sequence, skipping entries with no measurements
+  off_t NaveOut = 0;
+  for (i = 0; i < Naverage; i++) {
+    if (averageNmeas[i] == 0) continue;
+    averageSeqOut[NaveOut] = i;
+    averageRefOut[i] = NaveOut;
+    NaveOut ++;
+  }
+  // n = averageSeqOut[i] : averageOut[i] = average[n]  (i = 0 -- NaveOut, n = 0 -- Naverage)
+  // n = averageRefOut[i] : averageOut[n] = average[i]
+
+  // generate the output averefs
+  for (i = 0; i < NmeasOut; i++) {
+    j = measureSeqOut[i]; // measure[j] = measureOut[i]
+    n = measureAveRaw[j]; // average[n] : measure[j]
+    N = averageRefOut[n]; // averageOut[N] = average[n];
+    measureAveOut[i] = N; // measureOut[i].averef = measureAveOut[i] 
+  }
+
+  // copy the (kept) measurements in the sorted order
+  ALLOCATE (measureOut, Measure, NmeasOut);
+  for (i = 0; i < NmeasOut; i++) {
+    j = measureSeqOut[i];
+    measureOut[i] = measure[j];
+    measureOut[i].averef = measureAveOut[i];
+  }
+
+  // copy the (kept) average entries
+  ALLOCATE (averageOut, Average, NaveOut);
+  for (i = 0; i < NaveOut; i++) {
+    j = averageSeqOut[i];
+    averageOut[i] = average[j];
+    averageOut[i].Nmeasure      = averageNmeas[j]; 
+    averageOut[i].measureOffset = averageDmeas[j];
+  }
+
+  // copy the secfilt entries for the (kept) average entries
+  ALLOCATE (secfiltOut, SecFilt, NaveOut*Nsecfilt);
+  for (i = 0; i < NaveOut; i++) {
+    j = averageSeqOut[i];
+    for (m = 0; m < Nsecfilt; m++) {
+      secfiltOut[i*Nsecfilt + m] = secfilt[j*Nsecfilt + m];
+    }
+  }
+
+  // update the values of average.measureOffset and average.Nmeasure
+  FREE(measure);
+  FREE(average);
+  FREE(secfilt);
+  catalog[0].measure = measureOut;
+  catalog[0].average = averageOut;
+  catalog[0].secfilt = secfiltOut;
+  catalog[0].Nmeasure = NmeasOut;
+  catalog[0].Naverage = NaveOut;
+
+  // update catalog.Nmeasure and catalog.Naverage, double check 
+  int NmeasureTotal = 0;
+  int measureOffsetOK = TRUE;
+  int averefOK = TRUE;
+  for (i = 0; i < NaveOut; i++) {
+    NmeasureTotal += catalog[0].average[i].Nmeasure;
+    if (VERBOSE2 && !(NmeasureTotal <= catalog[0].Nmeasure)) {
+      fprintf (stderr, "too few measurements: %d %d %d\n", (int) i, NmeasureTotal, (int) catalog[0].Nmeasure);
+    }
+    measureOffsetOK &= (catalog[0].average[i].measureOffset < catalog[0].Nmeasure);
+    if (VERBOSE2 && !(catalog[0].average[i].measureOffset < catalog[0].Nmeasure)) {
+      fprintf (stderr, "offset too large: %d %d %d\n", (int) i, catalog[0].average[i].Nmeasure, (int) catalog[0].Nmeasure);
+    }
+    measureOffsetOK &= (catalog[0].average[i].measureOffset + catalog[0].average[i].Nmeasure <= catalog[0].Nmeasure);
+    if (VERBOSE2 && !(catalog[0].average[i].measureOffset + catalog[0].average[i].Nmeasure <= catalog[0].Nmeasure)) {
+      fprintf (stderr, "orrset + Nmeasure too large: %d + %d > %d %d\n", (int) i, catalog[0].average[i].measureOffset, catalog[0].average[i].Nmeasure, (int) catalog[0].Nmeasure);
+    }
+    m = catalog[0].average[i].measureOffset;
+    for (j = 0; j < catalog[0].average[i].Nmeasure; j++) {
+      averefOK &= (catalog[0].measure[m+j].averef == i);
+      if (VERBOSE2 && !(catalog[0].measure[m+j].averef == i)) {
+	fprintf (stderr, "averef broken: %d vs %d (measure %d)\n", (int) i, catalog[0].measure[m+j].averef, (int) (m+j));
+      }
+    }
+  }
+
+  if (!measureOffsetOK) {
+    fprintf (stderr, "ERROR: catalog %s has an invalid measureOffset\n", catalog[0].filename);
+  }
+
+  if (!averefOK) {
+    fprintf (stderr, "ERROR: catalog %s has invalid averefs\n", catalog[0].filename);
+  }
+
+  if (NmeasureTotal != catalog[0].Nmeasure) {
+    fprintf (stderr, "ERROR: catalog %s has an invalid Nmeasure\n", catalog[0].filename);
+  }
+
+  // MARKTIME("  match time %9.4f sec for %7lld measures, %6lld average\n", dtime, (long long) Nmeasure, (long long) Naverage);
+
+  catalog[0].sorted = TRUE;
+
+  if (VERBOSE) fprintf (stderr, "ending with Nave, Nmeas: "OFF_T_FMT" "OFF_T_FMT"\n",  catalog[0].Naverage,  catalog[0].Nmeasure);
+
+  FREE (measureDrop);
+  FREE (measureSeqRaw);
+  FREE (measureSeqOut);
+  FREE (measureRefOut);
+  FREE (measureAveRaw);
+  FREE (averageNmeas);
+  FREE (averageDmeas);
+  FREE (averageSeqOut);
+  FREE (averageRefOut);
+  FREE (measureAveOut);
+
+  return TRUE;
+}
+
+IndexArray *find_duplicates (Image *image, off_t Nimage) {
+
+  off_t i, *Ndelete;
+
+  // how to find duplicates:
+  // generate an index array of externID
+  //  * find min & max value
+  //  * array has length max - min
+  // generate the count of externID values
+  // mark those with N > 1
+
+  // index->value array is set to 0 initially
+  IndexArray *extID = make_index_array (image, Nimage, EXTERN_ID);
+
+  ALLOCATE (Ndelete, off_t, extID->range);
+  for (i = 0; i < extID->range; i++) Ndelete[i] = 0;
+
+  // set extID->value to have the number of images with a given externID
+  for (i = 0; i < Nimage; i++) {
+    if (image[i].externID == 0) continue;
+    off_t N = image[i].externID - extID->minID;
+    myAssert (N >= 0, "oops");
+    myAssert (N < extID->range, "oops");
+    extID->value[N] ++;
+  }
+  // count has the number for each image.externID
+  // now find things to be deleted:
+
+  // now I know the externID values to delete, I need to get a list of imageID values
+  // more specifically, i need an index of the imageIDs against an array saying delete or not
+
+  // somewhere downstream I want to be able to say:
+  // if (delete[measure.imageID]) delete_measure...
+
+  IndexArray *imageID = make_index_array (image, Nimage, IMAGE_ID);
+
+  // set imageID->value to TRUE for images we want to delete
+  for (i = 0; i < Nimage; i++) {
+    if (image[i].externID == 0) continue;
+    off_t Nx = image[i].externID - extID->minID;
+    myAssert (Nx >= 0, "oops");
+    myAssert (Nx < extID->range, "oops");
+    if (extID->value[Nx] < 2) continue;
+    if (Ndelete[Nx] == extID->value[Nx] - 1) continue;
+
+    off_t Ni = image[i].imageID - imageID->minID;
+    myAssert (Ni >= 0, "oops");
+    myAssert (Ni < imageID->range, "oops");
+    imageID->value[Ni] = TRUE;
+    Ndelete[Nx] ++;
+    // mark the parentID for deletion as well..
+  }
+
+  for (i = 0; i < Nimage; i++) {
+    off_t Ni = image[i].imageID - imageID->minID;
+    if (!imageID->value[Ni]) continue;
+    fprintf (stderr, "delete image (i) " OFF_T_FMT ", extID = %d\n", i, image[i].externID);
+  }
+
+  free (extID->value);
+  free (extID);
+
+  return imageID;;
+}
+
+// find the min & max values of the given ID (externID or imageID)
+// construct an empty array with length needed to fit IDs
+IndexArray *make_index_array (Image *image, off_t Nimage, int mode) {
+
+  off_t i, value;
+
+  IndexArray *idxarray = NULL;
+  ALLOCATE (idxarray, IndexArray, 1);
+
+  idxarray->minID = -1;
+  idxarray->maxID = -1;
+  for (i = 0; i < Nimage; i++) {
+    if (mode == EXTERN_ID) {
+      value = image[i].externID;
+      if (value == 0) continue;
+    }
+    if (mode == IMAGE_ID) {
+      value = image[i].imageID;
+      if (value == 0) continue;
+    }
+    if (idxarray->minID == -1) idxarray->minID = value;
+    idxarray->minID = MIN(idxarray->minID, value);
+    idxarray->maxID = MAX(idxarray->maxID, value);
+  }
+  idxarray->range = idxarray->maxID- idxarray->minID + 1;
+
+  // check if this will be too large?
+  ALLOCATE (idxarray->value, off_t, idxarray->range);
+
+  for (i = 0; i < idxarray->range; i++) idxarray->value[i] = 0;
+
+  return idxarray;
+}
+
Index: /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delstar.c
===================================================================
--- /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delstar.c	(revision 35228)
+++ /branches/eam_branches/ipp-20130207/Ohana/src/delstar/src/delstar.c	(revision 35229)
@@ -20,4 +20,7 @@
 
   switch (MODE) {
+    case MODE_DUP_IMAGES:
+      delete_duplicate_images (0, NULL, &db);
+      break;
     case MODE_IMAGEFILE:
       delete_imagefile (&db);