Add unescapeCsv to StringUtil

Motivation:

See #3435

Modifications:

Add unescapeCsv to StringUtil

Result:

StringUtil has the counter part of escapeCsv: unescapeCsv
This commit is contained in:
Xiaoyan Lin 2016-02-06 14:21:07 -08:00 committed by Scott Mitchell
parent 1bd4b702be
commit 85d0c9b573
2 changed files with 116 additions and 0 deletions

View File

@ -385,6 +385,67 @@ public final class StringUtil {
escaped.append(DOUBLE_QUOTE) : value; escaped.append(DOUBLE_QUOTE) : value;
} }
/**
* Unescapes the specified escaped CSV field, if necessary according to
* <a href="https://tools.ietf.org/html/rfc4180#section-2">RFC-4180</a>.
*
* @param value The escaped CSV field which will be unescaped according to
* <a href="https://tools.ietf.org/html/rfc4180#section-2">RFC-4180</a>
* @return {@link CharSequence} the unescaped value if necessary, or the value unchanged
*/
public static CharSequence unescapeCsv(CharSequence value) {
int length = checkNotNull(value, "value").length();
if (length == 0) {
return value;
}
int last = length - 1;
boolean quoted = isDoubleQuote(value.charAt(0)) && isDoubleQuote(value.charAt(last)) && length != 1;
if (!quoted) {
validateCsvFormat(value);
return value;
}
StringBuilder unescaped = InternalThreadLocalMap.get().stringBuilder();
for (int i = 1; i < last; i++) {
char current = value.charAt(i);
if (current == DOUBLE_QUOTE) {
if (isDoubleQuote(value.charAt(i + 1)) && (i + 1) != last) {
// Followed by a double-quote but not the last character
// Just skip the next double-quote
i++;
} else {
// Not followed by a double-quote or the following double-quote is the last character
throw newInvalidEscapedCsvFieldException(value, i);
}
}
unescaped.append(current);
}
return unescaped.toString();
}
/**
* Validate if {@code value} is a valid csv field without double-quotes.
*
* @throws IllegalArgumentException if {@code value} needs to be encoded with double-quotes.
*/
private static void validateCsvFormat(CharSequence value) {
int length = value.length();
for (int i = 0; i < length; i++) {
switch (value.charAt(i)) {
case DOUBLE_QUOTE:
case LINE_FEED:
case CARRIAGE_RETURN:
case COMMA:
// If value contains any special character, it should be enclosed with double-quotes
throw newInvalidEscapedCsvFieldException(value, i);
default:
}
}
}
private static IllegalArgumentException newInvalidEscapedCsvFieldException(CharSequence value, int index) {
return new IllegalArgumentException("invalid escaped CSV field: " + value + " index: " + index);
}
/** /**
* Get the length of a string, {@code null} input is considered {@code 0} length. * Get the length of a string, {@code null} input is considered {@code 0} length.
*/ */

View File

@ -321,6 +321,61 @@ public class StringUtilTest {
} }
} }
@Test
public void testUnescapeCsv() {
assertEquals("", unescapeCsv(""));
assertEquals("\"", unescapeCsv("\"\"\"\""));
assertEquals("\"\"", unescapeCsv("\"\"\"\"\"\""));
assertEquals("\"\"\"", unescapeCsv("\"\"\"\"\"\"\"\""));
assertEquals("\"netty\"", unescapeCsv("\"\"\"netty\"\"\""));
assertEquals("netty", unescapeCsv("netty"));
assertEquals("netty", unescapeCsv("\"netty\""));
assertEquals("\r", unescapeCsv("\"\r\""));
assertEquals("\n", unescapeCsv("\"\n\""));
assertEquals("hello,netty", unescapeCsv("\"hello,netty\""));
}
@Test(expected = IllegalArgumentException.class)
public void unescapeCsvWithSingleQuote() {
unescapeCsv("\"");
}
@Test(expected = IllegalArgumentException.class)
public void unescapeCsvWithOddQuote() {
unescapeCsv("\"\"\"");
}
@Test(expected = IllegalArgumentException.class)
public void unescapeCsvWithCRAndWithoutQuote() {
unescapeCsv("\r");
}
@Test(expected = IllegalArgumentException.class)
public void unescapeCsvWithLFAndWithoutQuote() {
unescapeCsv("\n");
}
@Test(expected = IllegalArgumentException.class)
public void unescapeCsvWithCommaAndWithoutQuote() {
unescapeCsv(",");
}
@Test
public void escapeCsvAndUnEscapeCsv() {
assertEscapeCsvAndUnEscapeCsv("");
assertEscapeCsvAndUnEscapeCsv("netty");
assertEscapeCsvAndUnEscapeCsv("hello,netty");
assertEscapeCsvAndUnEscapeCsv("hello,\"netty\"");
assertEscapeCsvAndUnEscapeCsv("\"");
assertEscapeCsvAndUnEscapeCsv(",");
assertEscapeCsvAndUnEscapeCsv("\r");
assertEscapeCsvAndUnEscapeCsv("\n");
}
private void assertEscapeCsvAndUnEscapeCsv(String value) {
assertEquals(value, unescapeCsv(StringUtil.escapeCsv(value)));
}
@Test @Test
public void testSimpleClassName() throws Exception { public void testSimpleClassName() throws Exception {
testSimpleClassName(String.class); testSimpleClassName(String.class);