Changeset 34461 for trunk/Ohana/src/opihi/cmd.data

Timestamp:

Sep 25, 2012, 6:39:11 AM (14 years ago)

Author:

eugene

Message:

add :char type to read vectors (very crude: sets vector to ascii value of first char); update reindex to handle duplicates; add mmatch function (equiv to avmatch)

Location:

trunk/Ohana/src/opihi/cmd.data

Files:

• read_vectors.c (modified) (5 diffs)
• reindex.c (modified) (5 diffs)
• subset.c (modified) (1 diff)

trunk/Ohana/src/opihi/cmd.data/read_vectors.c

@@ -22 +22 @@
 
 // vector types
-enum {COLTYPE_NONE, COLTYPE_FLT, COLTYPE_INT, COLTYPE_TIME};
+enum {COLTYPE_NONE, COLTYPE_FLT, COLTYPE_INT, COLTYPE_TIME, COLTYPE_CHAR};
 
 int read_vectors (int argc, char **argv) {
@@ -101 +101 @@
       if (!strcasecmp(ptr, "float")) { coltype[i] = COLTYPE_FLT; }
       if (!strcasecmp(ptr, "int"))   { coltype[i] = COLTYPE_INT; }
+      if (!strcasecmp(ptr, "char"))  { coltype[i] = COLTYPE_CHAR; }
       if (!strcasecmp(ptr, "time"))  { coltype[i] = COLTYPE_TIME; }
       if (!coltype[i]) goto bad_colname;
@@ -143 +144 @@
   NELEM = 1000;
   for (i = 0; i < Nvec; i++) {
-    if (coltype[i] == COLTYPE_INT) {
+    if ((coltype[i] == COLTYPE_INT) || (coltype[i] == COLTYPE_CHAR)) {
       ResetVector (vec[i], OPIHI_INT, NELEM);
     } else {
@@ -195 +196 @@
       for (i = 0; i < Nvec; i++) {
         int ivalue;
+        char cvalue;
         double dvalue;
         time_t tvalue;
@@ -203 +205 @@
             readStatus = IsCSV ? iparse_csv (&ivalue, col[i], c0) : iparse (&ivalue, col[i], c0);
             vec[i][0].elements.Int[Nelem] = readStatus ? ivalue : 0;
+            break;
+          case COLTYPE_CHAR:
+            readStatus = IsCSV ? charparse_csv (&cvalue, col[i], c0) : charparse (&cvalue, col[i], c0);
+            vec[i][0].elements.Int[Nelem] = readStatus ? cvalue : 0;
             break;
           case COLTYPE_FLT:

trunk/Ohana/src/opihi/cmd.data/reindex.c

@@ -5 +5 @@
 int reindex (int argc, char **argv) {
   
-  int  i, Npts, Nmax;
+  int  i, Nmax, N;
   Vector *ivec, *ovec, *xvec;
 
   ivec = ovec = xvec = NULL;
-  Npts = 0;
+  int Npts = 0;
+
+  int KEEP_UNMATCH = FALSE;
+  if ((N = get_argument (argc, argv, "-keep-unmatched"))) {
+    remove_argument (N, &argc, argv);
+    KEEP_UNMATCH = TRUE;
+  }
 
   if (argc != 6) goto usage;
@@ -22 +28 @@
 
   // ovec matches ivec in type and xvec in size (xvec need not have all ivec elements, and may have duplicates
+  int NPTS = xvec[0].Nelements;
   ResetVector (ovec, ivec->type, xvec[0].Nelements);
 
@@ -31 +38 @@
     opihi_int *vx = xvec[0].elements.Int;
     for (Npts = i = 0; i < xvec[0].Nelements; i++, vx++) {
-      if (*vx == -1) continue;
-      if (*vx < 0) ESCAPE("unexpected value in index: %d (%d)\n", *vx, i);
+      if (Npts >= NPTS) {
+        NPTS += 2000;
+        REALLOCATE (ovec[0].elements.Flt, opihi_flt, NPTS);
+      }
+      if (*vx < 0) {
+        if (KEEP_UNMATCH) {
+          ovec[0].elements.Flt[Npts] = NAN;
+          Npts++;
+        } 
+        continue;
+      }
       if (*vx > Nmax) ESCAPE("unexpected value in index: %d (%d)\n", *vx, i);
       ovec[0].elements.Flt[Npts] = vi[*vx];
@@ -42 +58 @@
     opihi_int *vx = xvec[0].elements.Int;
     for (Npts = i = 0; i < xvec[0].Nelements; i++, vx++) {
-      if (*vx == -1) continue;
-      if (*vx < 0) ESCAPE("unexpected value in index: %d (%d)\n", *vx, i);
+      if (Npts >= NPTS) {
+        NPTS += 2000;
+        REALLOCATE (ovec[0].elements.Int, opihi_int, NPTS);
+      }
+      if (*vx < 0) {
+        if (KEEP_UNMATCH) {
+          ovec[0].elements.Flt[Npts] = -1;
+          Npts++;
+        } 
+        continue;
+      }
       if (*vx > Nmax) ESCAPE("unexpected value in index: %d (%d)\n", *vx, i);
       ovec[0].elements.Int[Npts] = vi[*vx];
@@ -60 +85 @@
 usage:
     gprint (GP_ERR, "USAGE: reindex (out) = (in) using (index)\n");
-    gprint (GP_ERR, "  creates a new vectors (out) from (in) based on sequence in (index)\n");
+    gprint (GP_ERR, "  Creates a new vector (out) from (in) based on sequence in (index)\n");
+    gprint (GP_ERR, "  output[i] = input[index[i]]\n");
+    gprint (GP_ERR, "  If -keep-unmatched is provided, elements of index with negative values will be set to NaN / -1,\n");
+    gprint (GP_ERR, "    otherwise they will be skipped in the output.\n");
+    gprint (GP_ERR, "  The output vector has the type of the input vector and the length of the index (if -keep-unmatched).\n");
+    gprint (GP_ERR, "  The index vector may have duplicates\n");
     return (FALSE);
 }

trunk/Ohana/src/opihi/cmd.data/subset.c

@@ -13 +13 @@
   Npts = 0;
 
+  if (argc < 6) {
+    gprint (GP_ERR, "SYNTAX: subset vec = vec [if/where] (logic expression)\n");
+    return (FALSE);
+  }
+
   valid = TRUE;
-  valid &= (argc >= 6);
   valid &= !strcmp(argv[2], "=");
   valid &= !strcmp(argv[4], "if") || !strcmp (argv[4], "where");