[GitHub] [carbondata] zzcclp commented on a change in pull request #3504: [CARBONDATA-3614] Support Alter table properties set/unset for longstring columns

[GitBox]

zzcclp commented on a change in pull request #3504: [CARBONDATA-3614] Support Alter table properties set/unset for longstring columns
URL: https://github.com/apache/carbondata/pull/3504#discussion_r356983517
 
 

 ##########
 File path: core/src/main/java/org/apache/carbondata/core/util/CarbonUtil.java
 ##########
 @@ -3396,4 +3398,104 @@ public static CarbonFile createTempFolderForIndexServer(String tablePath, String
       return file;
     }
   }
+
+  /**
+   * For alter table tableProperties of long string, modify ColumnSchema
+   * set operation:
+   * a. reset the current varchar datatype to string
+   * b. reorder back to original order (consider sort column and original schema ordinal)
+   * c. set the new columns to varchar
+   * d. reorder new varchar columns to be after dimensions
+   * <p>
+   * unset operation:
+   * a. reset the current varchar datatype to string
+   * b. reorder back to original order (consider sort column and original schema ordinal)
+   *
+   * @param columns                 list of column schema from thrift
+   * @param longStringColumnsString comma separated long string columns to be set, empty for unset
+   * @return updated columnSchema list
+   */
+  public static List<org.apache.carbondata.format.ColumnSchema> reorderColumnsForLongString(
+      List<org.apache.carbondata.format.ColumnSchema> columns, String longStringColumnsString) {
+    // schema will be like
+    // sortColumns-otherDimensions-varchar-[complexColumns + complex_child - measures]
+    List<org.apache.carbondata.format.ColumnSchema> sortColumns = new ArrayList<>();
+    List<org.apache.carbondata.format.ColumnSchema> dimColumns = new ArrayList<>();
+    List<org.apache.carbondata.format.ColumnSchema> otherColumns = new ArrayList<>();
+    // true if cleared the varchar type in dims to string
+    boolean isResetDone = false;
+    // complex type child also looks like dimension or measure,
+    // so just take all other columns at once.
+    int otherColumnStartIndex = -1;
+    for (int i = 0; i < columns.size(); i++) {
+      if (columns.get(i).getColumnProperties() != null) {
+        String isSortColumn =
+            columns.get(i).getColumnProperties().get(CarbonCommonConstants.SORT_COLUMNS);
+        if ((isSortColumn != null) && (isSortColumn.equalsIgnoreCase("true"))) {
+          // add sort column dimensions
+          sortColumns.add(columns.get(i));
+        }
+      } else if ((columns.get(i).getData_type() == org.apache.carbondata.format.DataType.ARRAY
+          || columns.get(i).getData_type() == org.apache.carbondata.format.DataType.STRUCT
+          || columns.get(i).getData_type() == org.apache.carbondata.format.DataType.MAP || (!columns
+          .get(i).isDimension()))) {
+        // complex type or measure starts here and processed all sort columns and other dimensions
+        otherColumnStartIndex = i;
+        break;
+      } else {
+        // if it is varchar, reset to string as new set and unset needs to be done.
+        // if not, just add it to dimColumns
+        org.apache.carbondata.format.ColumnSchema col = columns.get(i);
+        if (col.data_type == org.apache.carbondata.format.DataType.VARCHAR) {
+          col.data_type = org.apache.carbondata.format.DataType.STRING;
+          isResetDone = true;
+        }
+        dimColumns.add(col);
+      }
+    }
+    if (otherColumnStartIndex != -1) {
+      otherColumns.addAll(columns.subList(otherColumnStartIndex, columns.size()));
+    }
+    if (isResetDone) {
+      // reorder the dims based on original schema ordinal
+      dimColumns.sort(new Comparator<org.apache.carbondata.format.ColumnSchema>() {
+        @Override
+        public int compare(org.apache.carbondata.format.ColumnSchema o1,
+            org.apache.carbondata.format.ColumnSchema o2) {
+          return o1.getSchemaOrdinal() - o2.getSchemaOrdinal();
+        }
+      });
+    }
+    List<org.apache.carbondata.format.ColumnSchema> nonVarCharDims = new ArrayList<>();
+    // for setting long string columns
+    if (!longStringColumnsString.isEmpty()) {
+      String[] inputColumns = longStringColumnsString.split(",");
+      Set<String> longStringSet = new HashSet<>(Arrays.asList(inputColumns));
+      List<org.apache.carbondata.format.ColumnSchema> varCharColumns = new ArrayList<>();
+      // change data type to varchar and extract the varchar columns
+      for (org.apache.carbondata.format.ColumnSchema dim : dimColumns) {
+        if (longStringSet.contains(dim.getColumn_name())) {
+          dim.data_type = org.apache.carbondata.format.DataType.VARCHAR;
+          // extract varchar dimensions
+          varCharColumns.add(dim);
+        } else {
+          // extract non varchar, non sort dimensions
+          nonVarCharDims.add(dim);
+        }
+      }
+      // append varchar in the end of dimensions
+      nonVarCharDims.addAll(varCharColumns);
+    } else {
+      nonVarCharDims = dimColumns;
+    }
+    // combine all columns and return
+    if (otherColumns.size() != 0) {
+      nonVarCharDims.addAll(otherColumns);
+    }
+    if (sortColumns.size() == 0) {
 
 Review comment:
   this 'if' branch can be removed.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[hidden email]


With regards,
Apache Git Services