csvreader.java

数据仓库工具

    }
  }


  /**
   *
   * Parse csv line with columnTypes.
   *
   * @param line
   * @return array with values or column names.
   * @throws SQLException
   */
  protected String[] parseCsvLine(String line) throws SQLException
  {
    ArrayList values = new ArrayList();
    boolean inQuotedString = false;
    String value = "";
    String orgLine = line;
    int currentPos = 0;
    int fullLine = 0;
    int currentColumn = 0;
    int indexOfBinaryObject = 0;
    char currentChar;

      line += separator;
      long lineLength = line.length();
      while (fullLine == 0) {
        currentPos = 0;
        while (currentPos < lineLength) {

//handle BINARY columns
          if( !(this.columnTypes.length <= currentColumn ) ) {
          if (this.columnTypes[currentColumn].equals(CsvDriver.BINARY_TYPE)) {
            String binaryValue = "";
            currentChar = line.charAt(currentPos);
            if (currentChar == ',') {
              values.add(binaryValue); //binary value is null;
              currentPos ++;
            }
            else if (currentChar == '"') {
              if (line.charAt(currentPos + 1) == '"') {
                values.add(binaryValue); //binary value is null
                currentPos = currentPos + 3;
              }
              else {
                // take all until next separator, and that is value
                // do not insert BinaryObject+index into line, just set right currentPos
                // and insert value into vector
                // binary value is always beteween quotes (")
                binaryValue = line.substring(currentPos);
                binaryValue = binaryValue.substring(1,
                                                    binaryValue.indexOf(separator) -
                                                    1);
                values.add(binaryValue);
                currentPos += binaryValue.length() + 3;
              }
            }
            //set currentColumn++
            currentColumn++;
            continue;
            }
          } else {
            throw new SQLException("Invalid csv format : file = "+new File(fileName).getAbsolutePath()+", line = "+line);
          }


//parse one by one character
          currentChar = line.charAt(currentPos);
          if (value.length() == 0 && currentChar == '"' && !inQuotedString) {
//enter here if we are at start of column value
            currentPos++;
            inQuotedString = true;
            continue;
          }

          if (currentChar == '"') {
//get next character
            char nextChar = line.charAt(currentPos + 1);
//if we have "", consider it as ", and add it to value
            if (nextChar == '"') {
              value += currentChar;
              currentPos++;
            }
            else {
//enter here if we are at end of column value
              if (!inQuotedString) {
                throw new SQLException("Unexpected '\"' in position " +
                                       currentPos + ". Line=" + orgLine);
              }
              if (inQuotedString && nextChar != separator) {
                throw new SQLException("Expecting " + separator +
                                       " in position " + (currentPos + 1) +
                                       ". Line=" + orgLine);
              }

//set currentPos to comma after value
              currentPos++;
//if value is empty string between double quotes consider it as empty string
//else if value is empty string between commas consider it as null value
              values.add(value);
              currentColumn++;
              value = "";
              inQuotedString = false;
            }
          }

          else {
//when we are at end of column value, and value is not inside of double quotes
            if (currentChar == separator) {
//when have separator in data
              if (inQuotedString) {
                value += currentChar;
              }
              else {
//if value is empty string between double quotes consider it as empty string
//else if value is empty string between commas consider it as null value
                if( value.equals("") )
                  value = null;
                values.add(value);
                currentColumn++;
                value = "";
              }
            }
            else {
              value += currentChar;
            }
          }

          currentPos++;
        } //end while

        if (inQuotedString) {
          // Remove extra , added at start
          value = value.substring(0, value.length() - 1);
          try {
            line = input.readLine();
          }
          catch (IOException e) {
            throw new SQLException(e.toString());
          }
        }
        else {
          fullLine = 1;
        }

      }// end while( fullLine == 0 )
    String[] retVal = new String[values.size()];
    values.toArray(retVal);

    return retVal;
  }


   /**
   *
   * Parse csv line, whithout columnTypes. 
   *
   * @param line
   * @return array with values or column names.
   * @throws SQLException
   */
  protected String[] parseCsvLineAsHeader(String line) throws SQLException
  {
    Vector values = new Vector();
    ArrayList columnTypesList = new ArrayList();
    boolean inQuotedString = false;
    String value = "";
    String orgLine = line;
    int currentPos = 0;
    int fullLine = 0;

    while (fullLine == 0) {
      currentPos = 0;
      line += separator;
      while (currentPos < line.length()) {
        char currentChar = line.charAt(currentPos);
        if (value.length() == 0 && currentChar == '"' && !inQuotedString) {
          currentPos++;
          inQuotedString = true;
          continue;
        }
        if (currentChar == '"') {
          char nextChar = line.charAt(currentPos + 1);
          if (nextChar == '"') {
            value += currentChar;
            currentPos++;
          }
          else {
            if (!inQuotedString) {
              throw new SQLException("Unexpected '\"' in position " +
                                     currentPos + ". Line=" + orgLine);
            }
            if (inQuotedString && nextChar != separator) {
              throw new SQLException("Expecting " + separator + " in position " +
                                     (currentPos + 1) + ". Line=" + orgLine);
            }
            if (value.endsWith("-"+CsvDriver.BINARY_TYPE)) {
              columnTypesList.add(CsvDriver.BINARY_TYPE);
              value = value.substring(0,value.indexOf("-"+CsvDriver.BINARY_TYPE));
            }
            else
              columnTypesList.add(CsvDriver.VARCHAR_TYPE);
            values.add(value);
            value = "";
            inQuotedString = false;
            currentPos++;
          }
        }
        else {
          if (currentChar == separator) {
            if (inQuotedString) {
              value += currentChar;
            }
            else {
              if (value.endsWith("-"+CsvDriver.BINARY_TYPE)) {
                columnTypesList.add(CsvDriver.BINARY_TYPE);
                value = value.substring(0,
                                        value.indexOf("-"+CsvDriver.BINARY_TYPE));
              }
              else
                columnTypesList.add(CsvDriver.VARCHAR_TYPE);
              values.add(value);
              value = "";
            }
          }
          else {
            value += currentChar;
          }
        }
        currentPos++;
      }
      if (inQuotedString) {
        value = value.substring(0, value.length() - 1);
        try {
          line = input.readLine();
        }
        catch (IOException e) {
          throw new SQLException(e.toString());
        }
      }
      else {
        fullLine = 1;
      }
    }
    String[] retVal = new String[values.size()];
    values.copyInto(retVal);

    this.columnTypes = new String[columnTypesList.size()];
    columnTypesList.toArray(columnTypes);

    return retVal;

  }

  private String formatString(String str) throws SQLException {
     String retValue = str;
     try {
       //replace spec. characters
       retValue = Utils.replaceAll(retValue,this.lineBreakEscape, "\n");
       retValue = Utils.replaceAll(retValue,this.carriageReturnEscape, "\r");
//       retValue = CsvSqlParser.replaceAll(retValue,this.doubleQuoteEscape, "\"");
     }catch(Exception e) {
       throw new SQLException("Error while reformat string ! : "+str);
     }
     return retValue;
   }


}