Changeset 41269 for trunk/Ohana/src/opihi/cmd.data/read_vectors.c

Timestamp:

Feb 24, 2020, 3:19:09 PM (6 years ago)

Author:

eugene

Message:

adding string-type vectors to opihi

File:

• trunk/Ohana/src/opihi/cmd.data/read_vectors.c (modified) (10 diffs)

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

@@ -35 +35 @@
 
 // vector types
-enum {COLTYPE_NONE, COLTYPE_FLT, COLTYPE_INT, COLTYPE_TIME, COLTYPE_CHAR};
+enum {COLTYPE_NONE, COLTYPE_FLT, COLTYPE_INT, COLTYPE_TIME, COLTYPE_CHAR, COLTYPE_STR};
 
 static int      Nvec     = 0;
@@ -91 +91 @@
     gprint (GP_ERR, "     -skip N : skip N lines before reading\n");
     gprint (GP_ERR, "     column names may include a type: name:type\n");
-    gprint (GP_ERR, "       type is int, float, char, time\n");
+    gprint (GP_ERR, "       type is int, float, char, str, time\n");
     gprint (GP_ERR, "       for char, values are placed into a list $name:0 - $name:n\n");
+    gprint (GP_ERR, "       for str, values are placed into a string-typed vector\n");
     gprint (GP_ERR, "       for time, values are human-readable date/time strings YYYY/MM/DD,hh:mm:ss\n");
     gprint (GP_ERR, "       the resulting vector is a float based on the TIMEFORMAT, TIMEREF values\n");
@@ -152 +153 @@
       if (!strcasecmp(ptr, "int"))   { coltype[i] = COLTYPE_INT; }
       if (!strcasecmp(ptr, "char"))  { coltype[i] = COLTYPE_CHAR; }
+      if (!strcasecmp(ptr, "str"))   { coltype[i] = COLTYPE_STR; }
       if (!strcasecmp(ptr, "time"))  { coltype[i] = COLTYPE_TIME; }
       if (!coltype[i]) goto bad_colname;
@@ -213 +215 @@
         // note that COLTYPE_TIME is a type of float
         ResetVector (vec[i], OPIHI_FLT, NELEM);
+        break;
+      case COLTYPE_STR:
+        ResetVector (vec[i], OPIHI_STR, NELEM);
         break;
       case COLTYPE_CHAR:
@@ -311 +316 @@
               break;
             }
+          case COLTYPE_STR:
+            {
+              // I need to get an isolated word in 'value' with the string value of this field 
+              char *ptr = IsCSV ? ptrparse_csv (col[i], c0) : ptrparse (col[i], c0);
+              char *value = NULL;
+              if (IsCSV) {
+                char *end = parse_nextword_csv (ptr);
+                if (end) {
+                  value = end ? strncreate (ptr, end - ptr) : strcreate (ptr);
+                }
+              } else {
+                value = thisword(ptr);
+              }
+              vec[i][0].elements.Str[Nelem] = value; // needs to be freed later.
+              break;
+            }
           case COLTYPE_FLT:
             readStatus = IsCSV ? dparse_csv (&dvalue, col[i], c0) : dparse (&dvalue, col[i], c0);
@@ -348 +369 @@
               ResetVector (vec[i], OPIHI_FLT, NELEM);
               break;
+            case COLTYPE_STR:
+              ResetVector (vec[i], OPIHI_STR, NELEM);
+              break;
             case COLTYPE_CHAR:
             default:
@@ -369 +393 @@
         ResetVector (vec[i], OPIHI_FLT, Nelem);
         break;
+      case COLTYPE_STR:
+        ResetVector (vec[i], OPIHI_STR, Nelem);
+        break;
       case COLTYPE_CHAR:
         sprintf (varname, "%s:n", listname[i]);
@@ -469 +496 @@
   }
 
-  // if CharAsVectors, char fields will be saved as vectors NAME:0 -- NAME:n for n characters
-  // else char fields will be saved as $NAME:0 - $NAME:m for m rows
+  // by default, we now store a string-type field as a string-type vector.
+
+  // if CharAsList is selected (and My < 10000), char fields will be saved as $NAME:0 - $NAME:m for m rows
+
+  // if CharAsVectors is selected, char fields will be saved as vectors NAME:0 -- NAME:n for n characters
+
   // if (Ny > 10000), force CharAsVectors
   int CharAsVectors = FALSE;
@@ -476 +507 @@
     remove_argument (N, &argc, argv);
     CharAsVectors = TRUE;
+  }
+  int CharAsList = FALSE;
+  if ((N = get_argument (argc, argv, "-char-list"))) {
+    remove_argument (N, &argc, argv);
+    CharAsList = TRUE;
   }
 
@@ -651 +687 @@
     if (!FITS_TRANSPOSE) {
       // read string column into a list rather than a vector
-      if (!strcmp (type, "char") && !CharAsVectors && (Ny < 3000)) {
-        char *fieldName = argv[i];
-        char *Ptr = data;
-        char varname[1024];  // used as a buffer for the names of string fields
-        for (j = 0; j < Ny; j++) {
-          set_list_varname (varname, fieldName, j, FALSE);
-          char *value = strncreate (&Ptr[j*Nval], Nval);
-          // replace instances of $ with _
-          char *p = strchr (value, '$');
-          while (p) { 
-            *p = '_';
-            p = strchr (p, '$');
+      if (!strcmp (type, "char")) {
+        // save char-type field as a List:
+        if (CharAsList && (Ny < 3000)) {
+          char *fieldName = argv[i];
+          char *Ptr = data;
+          char varname[1024];  // used as a buffer for the names of string fields
+          for (j = 0; j < Ny; j++) {
+            set_list_varname (varname, fieldName, j, FALSE);
+            char *value = strncreate (&Ptr[j*Nval], Nval);
+            // replace instances of $ with _
+            char *p = strchr (value, '$');
+            while (p) { 
+              *p = '_';
+              p = strchr (p, '$');
+            }
+            set_str_variable (varname, value);
+            free (value);
           }
-          set_str_variable (varname, value);
-          free (value);
+          sprintf (varname, "%s:n", fieldName);
+          set_int_variable (varname, Ny);
+          continue;
         }
-        sprintf (varname, "%s:n", fieldName);
-        set_int_variable (varname, Ny);
-        continue;
+        // save char-type field as a string-type vector:
+        if (!CharAsVectors) {
+          Vector *myVector = NULL;
+          sprintf (name, "%s", argv[i]);
+          if ((myVector = SelectVector (name, ANYVECTOR, TRUE)) == NULL) ESCAPE ("bad vector name\n");
+          ResetVector (myVector, OPIHI_STR, Ny);
+
+          char *Ptr = data;
+          for (j = 0; j < Ny; j++) {
+            myVector[0].elements.Str[j] = strncreate (&Ptr[j*Nval], Nval);
+            // replace instances of $ with _
+            char *p = strchr (myVector[0].elements.Str[j], '$');
+            while (p) { *p = '_'; p = strchr (p, '$'); }
+          }
+          continue;
+        }
       }
 
       // define the multifield vector names (Nval vectors x Ny elements)
+      // CharAsVectors is handled below automatically
       ALLOCATE (vec, Vector *, Nval);
       for (j = 0; j < Nval; j++) {