org.databene.formats.csv.CSVTokenizer Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of databene-formats Show documentation
Show all versions of databene-formats Show documentation
'Databene Formats' is an open source software library for supporting
data file and other formats like CSV, fixed width files, XLS, Properties and Regex.
It is designed for multithreaded use and high performance.
The newest version!
/*
* Copyright (C) 2011-2015 Volker Bergmann ([email protected]).
* All rights reserved.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
* http://www.apache.org/licenses/LICENSE-2.0
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.databene.formats.csv;
import static org.databene.formats.csv.CSVTokenType.*;
import org.databene.commons.IOUtil;
import org.databene.commons.SystemInfo;
import java.io.*;
/**
* Parses a CSV file token by token as specified in RFC 4180.
* It returns parsed values as CSVTokensType of type CELL, EOL and EOF.
* The current cell content is accessible by the public attribute 'cell'.
*
* Created: 26.08.2006 17:19:35
* @see CSVTokenType
*/
public class CSVTokenizer implements Closeable {
/** The default separator to use */
public static final char DEFAULT_SEPARATOR = ',';
private static final char DEFAULT_LINE_COMMENT = '#';
/** the source to read from */
private PushbackReader reader;
/** the actual separator */
private char separator;
/** The token at the cursor position */
public CSVTokenType ttype;
public CSVTokenType lastType;
/** String representation of the cell at the cursor position.
* If the cursor is at a EOL/EOF position, this is null */
public String cell;
public int line;
// constructors ----------------------------------------------------------------------------------------------------
/**
* Creates a tokenizer that reads from a URL.
* @param uri the URL to read from
* @throws IOException if stream access fails
*/
public CSVTokenizer(String uri) throws IOException {
this(uri, DEFAULT_SEPARATOR);
}
/**
* Creates a tokenizer that reads from a uri.
* @param uri the uri to read from
* @param separator character used for separating CSV cells
* @throws IOException if stream access fails
*/
public CSVTokenizer(String uri, char separator) throws IOException {
this(uri, separator, SystemInfo.getFileEncoding());
}
public CSVTokenizer(String uri, char separator, String encoding) throws IOException {
this(IOUtil.getReaderForURI(uri, encoding), separator);
}
/**
* Creates a tokenizer that reads from a java.io.Reader.
* @param reader the reader to use as input
*/
public CSVTokenizer(Reader reader) {
this(reader, DEFAULT_SEPARATOR);
}
/**
* Creates a tokenizer that reads from a java.io.Reader.
* @param reader the reader to use as input
* @param separator character used for separating CSV cells
*/
public CSVTokenizer(Reader reader, char separator) {
this.reader = new PushbackReader(new BufferedReader(reader));
this.separator = separator;
this.line = 1;
}
// interface -------------------------------------------------------------------------------------------------------
/**
* Returns the next token.
* @return the next token
* @throws IOException if source access fails
*/
public CSVTokenType next() throws IOException {
this.lastType = this.ttype;
if (reader == null) // if closed, return EOF
return setState(EOF, null);
if (lastType == EOL) { // on line separator increase line count
line++;
}
int c = read();
if (c == -1) { // if end of file is reached, close and signal EOF
close();
return setState(EOF, null);
} else if (this.lastType == EOL && c == DEFAULT_LINE_COMMENT) {
skipLine();
return next();
} else if (c == separator && lastType == CELL) {
c = read();
}
if (c == -1) { // if end of file is reached, close and signal EOF
close();
return setState(CELL, null);
}
if (c == separator) {
unread(c);
return setState(CELL, null);
} else if (c == '\r') { // handle \r\n or \r
if ((c = read()) != '\n')
unread(c);
return setState(EOL, null);
} else if (c == '\n') { // handle \n
return setState(EOL, null);
} else if (c == '"') {
unread(c);
return parseQuotes();
} else {
return parseSimpleCell(c);
}
}
public void skipLine() throws IOException {
int c;
// go to end of line
while ((c = read()) != -1 && c != '\r' && c != '\n') {
// skip EOL characters
}
switch (c) {
case -1 :
return;
case '\n' :
return;
case '\r' :
int c2 = read();
if (c2 != '\n')
unread(c2);
return;
default :
throw new IllegalStateException();
}
}
/** Closes the source */
@Override
public void close() {
if (reader != null)
IOUtil.close(reader);
reader = null;
}
public CSVTokenType lastTtype() {
return lastType;
}
// private helpers -------------------------------------------------------------------------------------------------
/**
* sets the state of the tokenizer to the given tokenType and cell content.
* @param tokenType the tokenType to use
* @param cell the cell content
* @return the token type
*/
private CSVTokenType setState(CSVTokenType tokenType, String cell) {
this.cell = cell;
this.ttype = tokenType;
return this.ttype;
}
private void unread(int c) throws IOException {
reader.unread(c);
}
private int read() throws IOException {
int c = reader.read();
return c;
}
private CSVTokenType parseSimpleCell(int c) throws IOException {
StringBuilder buffer = new StringBuilder().append((char) c);
boolean escapeMode = false;
while ((c = read()) != -1 && c != '\r' && c != '\n') {
if (escapeMode) {
c = unescape((char) c);
escapeMode = false;
} else if (c == '\\') {
escapeMode = true;
continue;
}
if (c == separator) {
unread(c);
return setState(CELL, buffer.toString());
}
buffer.append((char) c);
}
if (c == '\r' || c == '\n' || c == separator)
unread(c);
return setState(CELL, buffer.toString());
}
private static char unescape(char c) { // this is more efficient than StringUtil.unescape(String)
switch (c) {
case 't': return '\t';
case 'r': return '\r';
case 'n': return '\n';
default: return c;
}
}
private CSVTokenType parseQuotes() throws IOException {
read(); // skip leading quote
StringBuilder buffer = new StringBuilder();
int c;
boolean escapeMode = false;
boolean done;
do {
while ((c = read()) != -1 && c != '"') {
if (escapeMode) {
c = unescape((char) c);
escapeMode = false;
} else if (c == '\\') {
escapeMode = true;
continue;
}
buffer.append((char) c);
}
if (c == '"') {
c = read();
if (c == '"') {
// escaped quote
buffer.append('"');
done = false;
} else
done = true;
} else
done = true;
} while (!done);
if (c == '\r' || c == '\n' || c == separator) {
unread(c);
}
return setState(CELL, buffer.toString());
}
}