Example #1
0
 public GenericUDAFEvaluator getGenericUDAFEvaluator() {
   if (genericUDAFEvaluator != null) {
     return genericUDAFEvaluator;
   }
   if (genericUDAFWritableEvaluator != null) {
     return genericUDAFEvaluator = genericUDAFWritableEvaluator;
   }
   try {
     return genericUDAFEvaluator =
         ReflectionUtils.newInstance(
             Class.forName(
                     genericUDAFEvaluatorClassName,
                     true,
                     Utilities.getSessionSpecifiedClassLoader())
                 .asSubclass(GenericUDAFEvaluator.class),
             null);
   } catch (ClassNotFoundException e) {
     throw new RuntimeException(e);
   }
 }
  /**
   * Pre-analyze hook called after compilation and before semantic analysis We extract things for to
   * Database and metadata level operations which are not capture in the input/output entities
   * during semantic analysis. Ideally it should be handled in Hive. We need to move most of these
   * into hive semantic analyzer and then remove it from the access hook.
   */
  @Override
  public ASTNode preAnalyze(HiveSemanticAnalyzerHookContext context, ASTNode ast)
      throws SemanticException {

    switch (ast.getToken().getType()) {
        // Hive parser doesn't capture the database name in output entity, so we store it here for
        // now
      case HiveParser.TOK_CREATEDATABASE:
      case HiveParser.TOK_ALTERDATABASE_PROPERTIES:
      case HiveParser.TOK_DROPDATABASE:
      case HiveParser.TOK_SWITCHDATABASE:
      case HiveParser.TOK_DESCDATABASE:
        currDB = new Database(BaseSemanticAnalyzer.unescapeIdentifier(ast.getChild(0).getText()));
        break;
      case HiveParser.TOK_CREATETABLE:
      case HiveParser.TOK_CREATEVIEW:
        /*
         * Compiler doesn't create read/write entities for create table.
         * Hence we need extract dbname from db.tab format, if applicable
         */
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        break;
      case HiveParser.TOK_DROPTABLE:
      case HiveParser.TOK_DROPVIEW:
      case HiveParser.TOK_SHOW_CREATETABLE:
      case HiveParser.TOK_ALTERTABLE_SERIALIZER:
      case HiveParser.TOK_ALTERVIEW_ADDPARTS:
      case HiveParser.TOK_ALTERVIEW_DROPPARTS:
      case HiveParser.TOK_ALTERVIEW_PROPERTIES:
      case HiveParser.TOK_ALTERVIEW_RENAME:
      case HiveParser.TOK_CREATEINDEX:
      case HiveParser.TOK_DROPINDEX:
      case HiveParser.TOK_LOCKTABLE:
      case HiveParser.TOK_UNLOCKTABLE:
        currTab = extractTable((ASTNode) ast.getFirstChildWithType(HiveParser.TOK_TABNAME));
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        break;
      case HiveParser.TOK_ALTERINDEX_REBUILD:
        currTab = extractTable((ASTNode) ast.getChild(0)); // type is not TOK_TABNAME
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        break;
      case HiveParser.TOK_SHOW_TABLESTATUS:
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        int children = ast.getChildCount();
        for (int i = 1; i < children; i++) {
          ASTNode child = (ASTNode) ast.getChild(i);
          if (child.getToken().getType() == HiveParser.Identifier) {
            currDB = new Database(child.getText());
            break;
          }
        }
        // loosing the requested privileges for possible wildcard tables, since
        // further authorization will be done at the filter step and those unwanted will
        // eventually be filtered out from the output
        currTab = Table.ALL;
        break;
      case HiveParser.TOK_ALTERTABLE_RENAME:
      case HiveParser.TOK_ALTERTABLE_PROPERTIES:
      case HiveParser.TOK_ALTERTABLE_DROPPARTS:
      case HiveParser.TOK_ALTERTABLE_RENAMECOL:
      case HiveParser.TOK_ALTERTABLE_ADDCOLS:
      case HiveParser.TOK_ALTERTABLE_REPLACECOLS:
      case HiveParser.TOK_SHOW_TBLPROPERTIES:
      case HiveParser.TOK_SHOWINDEXES:
      case HiveParser.TOK_SHOWPARTITIONS:
        // token name TOK_TABNAME is not properly set in this case
        currTab = extractTable((ASTNode) ast.getChild(0));
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        break;
      case HiveParser.TOK_MSCK:
        // token name TOK_TABNAME is not properly set in this case and child(0) does
        // not contain the table name.
        // TODO: Fix Hive to capture the table and DB name
        currOutTab = extractTable((ASTNode) ast.getChild(1));
        currOutDB = extractDatabase((ASTNode) ast.getChild(0));
        break;
      case HiveParser.TOK_ALTERTABLE_ADDPARTS:
        /*
         * Compiler doesn't create read/write entities for create table.
         * Hence we need extract dbname from db.tab format, if applicable
         */
        currTab = extractTable((ASTNode) ast.getChild(0));
        currDB = extractDatabase((ASTNode) ast.getChild(0));
        partitionURI = extractPartition(ast);
        break;
      case HiveParser.TOK_CREATEFUNCTION:
        String udfClassName = BaseSemanticAnalyzer.unescapeSQLString(ast.getChild(1).getText());
        try {
          CodeSource udfSrc =
              Class.forName(udfClassName, true, Utilities.getSessionSpecifiedClassLoader())
                  .getProtectionDomain()
                  .getCodeSource();
          if (udfSrc == null) {
            throw new SemanticException("Could not resolve the jar for UDF class " + udfClassName);
          }
          String udfJar = udfSrc.getLocation().getPath();
          if (udfJar == null || udfJar.isEmpty()) {
            throw new SemanticException(
                "Could not find the jar for UDF class " + udfClassName + "to validate privileges");
          }
          udfURI = parseURI(udfSrc.getLocation().toString(), true);
        } catch (ClassNotFoundException e) {
          throw new SemanticException("Error retrieving udf class:" + e.getMessage(), e);
        }
        // create/drop function is allowed with any database
        currDB = Database.ALL;
        break;
      case HiveParser.TOK_DROPFUNCTION:
        // create/drop function is allowed with any database
        currDB = Database.ALL;
        break;

      case HiveParser.TOK_LOAD:
        String dbName =
            BaseSemanticAnalyzer.unescapeIdentifier(
                ast.getChild(1).getChild(0).getChild(0).getText());
        currDB = new Database(dbName);
        break;
      case HiveParser.TOK_DESCTABLE:
        currDB = getCanonicalDb();
        // For DESCRIBE FORMATTED/EXTENDED ast will have an additional child node with value
        // "FORMATTED/EXTENDED".
        isDescTableBasic = (ast.getChildCount() == 1);
        break;
      case HiveParser.TOK_TRUNCATETABLE:
        // SENTRY-826:
        // Truncate empty partitioned table should throw SemanticException only if the
        // user does not have permission.
        // In postAnalyze, currOutDB and currOutTbl will be added into outputHierarchy
        // which will be validated in the hiveAuthzBinding.authorize method.
        Preconditions.checkArgument(ast.getChildCount() == 1);
        // childcount is 1 for table without partition, 2 for table with partitions
        Preconditions.checkArgument(ast.getChild(0).getChildCount() >= 1);
        Preconditions.checkArgument(ast.getChild(0).getChild(0).getChildCount() == 1);
        currOutDB = extractDatabase((ASTNode) ast.getChild(0));
        currOutTab = extractTable((ASTNode) ast.getChild(0).getChild(0).getChild(0));
        break;
      default:
        currDB = getCanonicalDb();
        break;
    }
    return ast;
  }