src.java.com.ctc.wstx.io.StreamBootstrapper Maven / Gradle / Ivy
package com.ctc.wstx.io;
import java.io.*;
import javax.xml.stream.Location;
import javax.xml.stream.XMLStreamException;
import com.ctc.wstx.api.ReaderConfig;
import com.ctc.wstx.cfg.ParsingErrorMsgs;
import com.ctc.wstx.cfg.XmlConsts;
import com.ctc.wstx.exc.*;
import com.ctc.wstx.util.StringUtil;
/**
* Input bootstrap class used with streams, when encoding is not known
* (when encoding is specified by application, a reader is constructed,
* and then reader-based bootstrapper is used).
* 1) {
total /= mBytesPerChar;
}
return total;
}
public int getInputColumn() {
int col = mInputPtr - mInputRowStart;
if (mBytesPerChar > 1) {
col /= mBytesPerChar;
}
return col;
}
/*
////////////////////////////////////////
// Internal methods, parsing
////////////////////////////////////////
*/
/**
* Method called to try to figure out physical encoding the underlying
* input stream uses.
*/
protected void resolveStreamEncoding()
throws IOException, WstxException
{
// Let's first set defaults:
mBytesPerChar = 0;
mBigEndian = true;
/* Ok; first just need 4 bytes for determining bytes-per-char from
* BOM or first char(s) of likely xml declaration:
*/
if (ensureLoaded(4)) {
bomblock:
do { // BOM/auto-detection block
int quartet = (mByteBuffer[0] << 24)
| ((mByteBuffer[1] & 0xFF) << 16)
| ((mByteBuffer[2] & 0xFF) << 8)
| (mByteBuffer[3] & 0xFF);
/* Handling of (usually) optional BOM (required for
* multi-byte formats); first 32-bit charsets:
*/
switch (quartet) {
case 0x0000FEFF:
mBigEndian = true;
mInputPtr = mBytesPerChar = 4;
break bomblock;
case 0xFFFE0000: // UCS-4, LE?
mInputPtr = mBytesPerChar = 4;
mBigEndian = false;
break bomblock;
case 0x0000FFFE: // UCS-4, in-order...
reportWeirdUCS4("2143");
break bomblock;
case 0x0FEFF0000: // UCS-4, in-order...
reportWeirdUCS4("3412");
break bomblock;
}
// Ok, if not, how about 16-bit encoding BOMs?
int msw = quartet >>> 16;
if (msw == 0xFEFF) { // UTF-16, BE
mInputPtr = mBytesPerChar = 2;
mBigEndian = true;
break;
}
if (msw == 0xFFFE) { // UTF-16, LE
mInputPtr = mBytesPerChar = 2;
mBigEndian = false;
break;
}
// And if not, then UTF-8 BOM?
if ((quartet >>> 8) == 0xEFBBBF) { // UTF-8
mInputPtr = 3;
mBytesPerChar = 1;
mBigEndian = true; // doesn't really matter
break;
}
/* And if that wasn't succesful, how about auto-detection
* for ' 0);
// Let's update location markers to ignore BOM.
mInputProcessed = -mInputPtr;
mInputRowStart = mInputPtr;
}
/* Hmmh. If we haven't figured it out, let's just assume
* UTF-8 as per XML specs:
*/
mByteSizeFound = (mBytesPerChar != 0);
if (!mByteSizeFound) {
mBytesPerChar = 1;
mBigEndian = true; // doesn't matter
}
}
/**
* @return Normalized encoding name
*/
protected String verifyXmlEncoding(String enc)
throws WstxException
{
enc = CharsetNames.normalize(enc);
// Let's actually verify we got matching information:
if (enc == CharsetNames.CS_UTF8) {
verifyEncoding(enc, 1);
} else if (enc == CharsetNames.CS_ISO_LATIN1) {
verifyEncoding(enc, 1);
} else if (enc == CharsetNames.CS_US_ASCII) {
verifyEncoding(enc, 1);
} else if (enc == CharsetNames.CS_UTF16) {
// BOM is obligatory, to know the ordering
/* 22-Mar-2005, TSa: Actually, since we don't have a
* custom decoder, so the underlying JDK Reader may
* have dealt with it transparently... so we can not
* really throw an exception here.
*/
//if (!mHadBOM) {
//reportMissingBOM(enc);
//}
verifyEncoding(enc, 2);
} else if (enc == CharsetNames.CS_UTF16LE) {
verifyEncoding(enc, 2, false);
} else if (enc == CharsetNames.CS_UTF16BE) {
verifyEncoding(enc, 2, true);
} else if (enc == CharsetNames.CS_UTF32) {
// Do we require a BOM here? we can live without it...
//if (!mHadBOM) {
// reportMissingBOM(enc);
//}
verifyEncoding(enc, 4);
} else if (enc == CharsetNames.CS_UTF32LE) {
verifyEncoding(enc, 4, false);
} else if (enc == CharsetNames.CS_UTF32BE) {
verifyEncoding(enc, 4, true);
}
return enc;
}
/*
/////////////////////////////////////////////////////
// Internal methods, loading input data
/////////////////////////////////////////////////////
*/
protected boolean ensureLoaded(int minimum)
throws IOException
{
/* Let's assume here buffer has enough room -- this will always
* be true for the limited used this method gets
*/
int gotten = (mInputLen - mInputPtr);
while (gotten < minimum) {
int count = mIn.read(mByteBuffer, mInputLen,
mByteBuffer.length - mInputLen);
if (count < 1) {
return false;
}
mInputLen += count;
gotten += count;
}
return true;
}
protected void loadMore()
throws IOException, WstxException
{
/* Need to make sure offsets are properly updated for error
* reporting purposes, and do this now while previous amounts
* are still known.
*/
/* Note: at this point these are all in bytes, not chars (for multibyte
* encodings)
*/
mInputProcessed += mInputLen;
mInputRowStart -= mInputLen;
mInputPtr = 0;
mInputLen = mIn.read(mByteBuffer, 0, mByteBuffer.length);
if (mInputLen < 1) {
throw new WstxEOFException(ParsingErrorMsgs.SUFFIX_IN_XML_DECL,
getLocation());
}
}
/*
/////////////////////////////////////////////////////
// Implementations of abstract parsing methods
/////////////////////////////////////////////////////
*/
protected void pushback() {
if (mBytesPerChar < 0) {
mInputPtr += mBytesPerChar;
} else {
mInputPtr -= mBytesPerChar;
}
}
protected int getNext()
throws IOException, WstxException
{
if (mBytesPerChar != 1) {
if (mBytesPerChar == -1) { // need to translate
return nextTranslated();
}
return nextMultiByte();
}
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
return (b & 0xFF);
}
protected int getNextAfterWs(boolean reqWs)
throws IOException, WstxException
{
int count;
if (mBytesPerChar == 1) { // single byte
count = skipSbWs();
} else {
if (mBytesPerChar == -1) { // translated
count = skipTranslatedWs();
} else { // multi byte
count = skipMbWs();
}
}
if (reqWs && count == 0) {
reportUnexpectedChar(getNext(), ERR_XMLDECL_EXP_SPACE);
}
// inlined getNext()
if (mBytesPerChar != 1) {
if (mBytesPerChar == -1) { // translated
return nextTranslated();
}
return nextMultiByte();
}
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
return (b & 0xFF);
}
/**
* @return First character that does not match expected, if any;
* CHAR_NULL if match succeeded
*/
protected int checkKeyword(String exp)
throws IOException, WstxException
{
if (mBytesPerChar != 1) {
if (mBytesPerChar == -1) {
return checkTranslatedKeyword(exp);
}
return checkMbKeyword(exp);
}
return checkSbKeyword(exp);
}
protected int readQuotedValue(char[] kw, int quoteChar)
throws IOException, WstxException
{
int i = 0;
int len = kw.length;
boolean simple = (mBytesPerChar == 1);
boolean mb = !simple && (mBytesPerChar > 1);
while (i < len) {
int c;
if (simple) {
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
if (b == BYTE_NULL) {
reportNull();
}
if (b == BYTE_CR || b == BYTE_LF) {
skipSbLF(b);
b = BYTE_LF;
}
c = (b & 0xFF);
} else {
if (mb) {
c = nextMultiByte();
if (c == CHAR_CR || c == CHAR_LF) {
skipMbLF(c);
c = CHAR_LF;
}
} else {
c = nextTranslated();
if (c == CHAR_CR || c == CHAR_LF) {
skipTranslatedLF(c);
c = CHAR_LF;
}
}
}
if (c == quoteChar) {
return (i < len) ? i : -1;
}
if (i < len) {
kw[i++] = (char) c;
}
}
/* If we end up this far, we ran out of buffer space... let's let
* caller figure that out, though
*/
return -1;
}
protected boolean hasXmlDecl()
throws IOException, WstxException
{
/* Separate handling for common and fast case; 1/variable byte
* encodings that have ASCII subset:
*/
if (mBytesPerChar == 1) {
/* However... there has to be at least 6 bytes available; and if
* so, can check the 'signature' easily:
*/
if (ensureLoaded(6)) {
if (mByteBuffer[mInputPtr] == '<'
&& mByteBuffer[mInputPtr+1] == '?'
&& mByteBuffer[mInputPtr+2] == 'x'
&& mByteBuffer[mInputPtr+3] == 'm'
&& mByteBuffer[mInputPtr+4] == 'l'
&& ((mByteBuffer[mInputPtr+5] & 0xFF) <= CHAR_SPACE)) {
// Let's skip stuff so far:
mInputPtr += 6;
return true;
}
}
} else if (mBytesPerChar == -1) { // translated (EBCDIC)
if (ensureLoaded(6)) {
int start = mInputPtr; // if we have to 'unread' chars
if (nextTranslated() == '<'
&& nextTranslated() == '?'
&& nextTranslated() == 'x'
&& nextTranslated() == 'm'
&& nextTranslated() == 'l'
&& nextTranslated() <= CHAR_SPACE) {
return true;
}
mInputPtr = start; // push data back
}
} else {
// ... and then for slower fixed-multibyte encodings:
// Is there enough data for checks?
if (ensureLoaded (6 * mBytesPerChar)) {
int start = mInputPtr; // if we have to 'unread' chars
if (nextMultiByte() == '<'
&& nextMultiByte() == '?'
&& nextMultiByte() == 'x'
&& nextMultiByte() == 'm'
&& nextMultiByte() == 'l'
&& nextMultiByte() <= CHAR_SPACE) {
return true;
}
mInputPtr = start; // push data back
}
}
return false;
}
protected Location getLocation()
{
/* Ok; for fixed-size multi-byte encodings, need to divide numbers
* to get character locations. For variable-length encodings the
* good thing is that xml declaration only uses shortest codepoints,
* ie. char count == byte count.
*/
int total = mInputProcessed + mInputPtr;
int col = mInputPtr - mInputRowStart;
if (mBytesPerChar > 1) {
total /= mBytesPerChar;
col /= mBytesPerChar;
}
return new WstxInputLocation(null, mPublicId, mSystemId,
total - 1, // 0-based
mInputRow, col);
}
/*
/////////////////////////////////////////////////////
// Internal methods, single-byte access methods
/////////////////////////////////////////////////////
*/
protected byte nextByte()
throws IOException, WstxException
{
if (mInputPtr >= mInputLen) {
loadMore();
}
return mByteBuffer[mInputPtr++];
}
protected int skipSbWs()
throws IOException, WstxException
{
int count = 0;
while (true) {
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
if ((b & 0xFF) > CHAR_SPACE) {
--mInputPtr;
break;
}
if (b == BYTE_CR || b == BYTE_LF) {
skipSbLF(b);
} else if (b == BYTE_NULL) {
reportNull();
}
++count;
}
return count;
}
protected void skipSbLF(byte lfByte)
throws IOException, WstxException
{
if (lfByte == BYTE_CR) {
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
if (b != BYTE_LF) {
--mInputPtr; // pushback if not 2-char/byte lf
}
}
++mInputRow;
mInputRowStart = mInputPtr;
}
/**
* @return First character that does not match expected, if any;
* CHAR_NULL if match succeeded
*/
protected int checkSbKeyword(String expected)
throws IOException, WstxException
{
int len = expected.length();
for (int ptr = 1; ptr < len; ++ptr) {
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
if (b == BYTE_NULL) {
reportNull();
}
if ((b & 0xFF) != expected.charAt(ptr)) {
return (b & 0xFF);
}
}
return CHAR_NULL;
}
/*
/////////////////////////////////////////////////////
// Internal methods, multi-byte/translated access/checks
/////////////////////////////////////////////////////
*/
protected int nextMultiByte()
throws IOException, WstxException
{
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
byte b2 = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
int c;
if (mBytesPerChar == 2) {
if (mBigEndian) {
c = ((b & 0xFF) << 8) | (b2 & 0xFF);
} else {
c = (b & 0xFF) | ((b2 & 0xFF) << 8);
}
} else {
// Has to be 4 bytes
byte b3 = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
byte b4 = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
if (mBigEndian) {
c = (b << 24) | ((b2 & 0xFF) << 16)
| ((b3 & 0xFF) << 8) | (b4 & 0xFF);
} else {
c = (b4 << 24) | ((b3 & 0xFF) << 16)
| ((b2 & 0xFF) << 8) | (b & 0xFF);
}
}
// Let's catch null chars early
if (c == 0) {
reportNull();
}
return c;
}
protected int nextTranslated()
throws IOException, WstxException
{
byte b = (mInputPtr < mInputLen) ?
mByteBuffer[mInputPtr++] : nextByte();
int ch = mSingleByteTranslation[b & 0xFF];
if (ch < 0) { // special char... won't care for now
ch = -ch;
}
return ch;
}
protected int skipMbWs()
throws IOException, WstxException
{
int count = 0;
while (true) {
int c = nextMultiByte();
if (c > CHAR_SPACE) {
mInputPtr -= mBytesPerChar;
break;
}
if (c == CHAR_CR || c == CHAR_LF) {
skipMbLF(c);
} else if (c == CHAR_NULL) {
reportNull();
}
++count;
}
return count;
}
protected int skipTranslatedWs()
throws IOException, WstxException
{
int count = 0;
while (true) {
int c = nextTranslated();
// Hmmh. Are we to accept NEL (0x85)?
if (c > CHAR_SPACE && c != CHAR_NEL) {
--mInputPtr;
break;
}
if (c == CHAR_CR || c == CHAR_LF) {
skipTranslatedLF(c);
} else if (c == CHAR_NULL) {
reportNull();
}
++count;
}
return count;
}
protected void skipMbLF(int lf)
throws IOException, WstxException
{
if (lf == CHAR_CR) {
int c = nextMultiByte();
if (c != CHAR_LF) {
mInputPtr -= mBytesPerChar;
}
}
++mInputRow;
mInputRowStart = mInputPtr;
}
protected void skipTranslatedLF(int lf)
throws IOException, WstxException
{
if (lf == CHAR_CR) {
int c = nextTranslated();
if (c != CHAR_LF) {
mInputPtr -= 1;
}
}
++mInputRow;
mInputRowStart = mInputPtr;
}
/**
* @return First character that does not match expected, if any;
* CHAR_NULL if match succeeded
*/
protected int checkMbKeyword(String expected)
throws IOException, WstxException
{
int len = expected.length();
for (int ptr = 1; ptr < len; ++ptr) {
int c = nextMultiByte();
if (c == BYTE_NULL) {
reportNull();
}
if (c != expected.charAt(ptr)) {
return c;
}
}
return CHAR_NULL;
}
protected int checkTranslatedKeyword(String expected)
throws IOException, WstxException
{
int len = expected.length();
for (int ptr = 1; ptr < len; ++ptr) {
int c = nextTranslated();
if (c == BYTE_NULL) {
reportNull();
}
if (c != expected.charAt(ptr)) {
return c;
}
}
return CHAR_NULL;
}
/*
////////////////////////////////////////
// Other private methods:
////////////////////////////////////////
*/
private void verifyEncoding(String id, int bpc)
throws WstxException
{
if (mByteSizeFound) {
/* Let's verify that if we matched an encoding, it's the same
* as what was declared...
*/
if (bpc != mBytesPerChar) {
// [WSTX-138]: Needs to detect EBCDIC discrepancy
if (mEBCDIC) {
reportXmlProblem("Declared encoding '"+id+"' incompatible with auto-detected physical encoding (EBCDIC variant), can not decode input since actual code page not known");
}
reportXmlProblem("Declared encoding '"+id+"' uses "+bpc
+" bytes per character; but physical encoding appeared to use "+mBytesPerChar+"; cannot decode");
}
}
}
private void verifyEncoding(String id, int bpc, boolean bigEndian)
throws WstxException
{
if (mByteSizeFound) {
verifyEncoding(id, bpc);
if (bigEndian != mBigEndian) {
String bigStr = bigEndian ? "big" : "little";
reportXmlProblem
("Declared encoding '"+id+"' has different endianness ("
+bigStr+" endian) than what physical ordering appeared to be; cannot decode");
}
}
}
private void reportWeirdUCS4(String type)
throws IOException
{
throw new CharConversionException("Unsupported UCS-4 endianness ("+type+") detected");
}
private void reportMissingBOM(String enc)
throws WstxException
{
throw new WstxException("Missing BOM for encoding '"+enc+"'; can not be omitted",
getLocation());
}
}