Cleans up ASTStrList to handle frames with more than one vector durin…

…g column conversion. Checks types before converting. Adds several new column type conversions.
h2oai · Oct 13, 2015 · 836c32e · 836c32e
1 parent 44994ee
commit 836c32e
Show file tree

Hide file tree

Showing 3 changed files with 271 additions and 60 deletions.
diff --git a/h2o-core/src/main/java/water/parser/BufferedString.java b/h2o-core/src/main/java/water/parser/BufferedString.java
@@ -131,5 +131,24 @@ else if (o instanceof String) {
   public final byte [] getBuffer() {return _buf;}
   public final int getOffset() {return _off;}
   public final int length() {return _len;}
+
+  public static final byte NA  =  0;
+  public static final byte INT =  1;
+  public static final byte REAL=  2;
+  public final byte getNumericType() {
+    int i = 0;
+    int decimalCnt = 0;
+    if (_len == 0) return NA;
+    if (_buf[_off] == '+' || _buf[_off] == '-') i++;
+    while( i < _len) {
+      if (_buf[_off+i] == '.') decimalCnt++;
+      if (_buf[_off+i] < '0' || _buf[_off+i] > '9') return NA;
+      i++;
+    }
+    if (decimalCnt > 0)
+      if (decimalCnt == 1) return REAL;
+      else return NA; //more than one decimal, NaN
+    else return INT;
+  }
 }
 
diff --git a/h2o-core/src/main/java/water/rapids/ASTStrList.java b/h2o-core/src/main/java/water/rapids/ASTStrList.java
@@ -82,11 +82,18 @@ class ASTAsCharacter extends ASTPrim {
   public String str() { return "as.character"; }
   @Override Val apply( Env env, Env.StackHelp stk, AST asts[] ) {
     Frame ary = stk.track(asts[1].exec(env)).getFrame();
-    if( ary.numCols() != 1 ) throw new IllegalArgumentException("character requires a single column");
-    Vec v0 = ary.anyVec();
-    Vec v1 = v0.isString() ? null : VecUtils.toStringVec(v0); // toCategoricalVec() creates a new vec --> must be cleaned up!
-    Frame fr = new Frame(ary._names, new Vec[]{v1 == null ? v0.makeCopy(null) : v1});
-    return new ValFrame(fr);
+    Vec[] nvecs = new Vec[ary.numCols()];
+    Vec vv;
+    for(int c=0;c<nvecs.length;++c) {
+      vv = ary.vec(c);
+      try {
+        nvecs[c] = VecUtils.toStringVec(vv);
+      } catch (Exception e) {
+        VecUtils.deleteVecs(nvecs, c);
+        throw e;
+      }
+    }
+    return new ValFrame(new Frame(ary._names, nvecs));
   }
 }
 
@@ -98,11 +105,26 @@ class ASTAsFactor extends ASTPrim {
   @Override
   public String str() { return "as.factor"; }
   @Override Val apply( Env env, Env.StackHelp stk, AST asts[] ) {
-    Frame fr = stk.track(asts[1].exec(env)).getFrame();
-    if( fr.numCols() != 1 ) throw new IllegalArgumentException("as.factor requires a single column");
-    Vec v0 = fr.anyVec();
-    if( !v0.isCategorical() ) v0 = VecUtils.toCategoricalVec(v0);
-    return new ValFrame(new Frame(fr._names, new Vec[]{v0}));
+    Frame ary = stk.track(asts[1].exec(env)).getFrame();
+    Vec[] nvecs = new Vec[ary.numCols()];
+
+    // Type check  - prescreen for correct types
+    for (Vec v : ary.vecs())
+      if (!(v.isCategorical() || v.isString()|| v.isNumeric()))
+        throw new IllegalArgumentException("asfactor() requires a string, categorical, or numeric column. "
+            +"Received "+ary.anyVec().get_type_str()
+            +". Please convert column to a string or categorical first.");
+    Vec vv;
+    for(int c=0;c<nvecs.length;++c) {
+      vv = ary.vec(c);
+      try {
+        nvecs[c] = VecUtils.toCategoricalVec(vv);
+      } catch (Exception e) {
+        VecUtils.deleteVecs(nvecs, c);
+        throw e;
+      }
+    }
+    return new ValFrame(new Frame(ary._names, nvecs));
   }
 }
 
@@ -119,9 +141,12 @@ class ASTAsNumeric extends ASTPrim {
     Vec vv;
     for(int c=0;c<nvecs.length;++c) {
       vv = fr.vec(c);
-      if( vv.isCategorical() ) nvecs[c] = VecUtils.toIntVec(vv);
-      else if( vv.isString() ) nvecs[c] = VecUtils.toNumericVec(vv);
-      else                     nvecs[c] = vv.makeCopy();
+      try {
+        nvecs[c] = VecUtils.toNumericVec(vv);
+      } catch (Exception e) {
+        VecUtils.deleteVecs(nvecs, c);
+        throw e;
+      }
     }
     return new ValFrame(new Frame(fr._names, nvecs));
   }