/** ******************************************************************************* * Copyright (C) 1996-2010, International Business Machines Corporation and * * others. All Rights Reserved. * ******************************************************************************* */ package com.ibm.icu.text; import java.io.BufferedInputStream; import java.io.DataInputStream; import java.io.IOException; import java.io.InputStream; import com.ibm.icu.impl.CharTrie; import com.ibm.icu.impl.Trie; /** *
Internal class used for Rule Based Break Iterators
*This class provides access to the compiled break rule data, as
* it is stored in a .brk file.
*/
final class RBBIDataWrapper {
//
// These fields are the ready-to-use compiled rule data, as
// read from the file.
//
RBBIDataHeader fHeader;
short fFTable[];
short fRTable[];
short fSFTable[];
short fSRTable[];
CharTrie fTrie;
String fRuleSource;
int fStatusTable[];
//
// Indexes to fields in the ICU4C style binary form of the RBBI Data Header
// Used by the rule compiler when flattening the data.
//
final static int DH_SIZE = 24;
final static int DH_MAGIC = 0;
final static int DH_FORMATVERSION = 1;
final static int DH_LENGTH = 2;
final static int DH_CATCOUNT = 3;
final static int DH_FTABLE = 4;
final static int DH_FTABLELEN = 5;
final static int DH_RTABLE = 6;
final static int DH_RTABLELEN = 7;
final static int DH_SFTABLE = 8;
final static int DH_SFTABLELEN = 9;
final static int DH_SRTABLE = 10;
final static int DH_SRTABLELEN = 11;
final static int DH_TRIE = 12;
final static int DH_TRIELEN = 13;
final static int DH_RULESOURCE = 14;
final static int DH_RULESOURCELEN = 15;
final static int DH_STATUSTABLE = 16;
final static int DH_STATUSTABLELEN = 17;
// Index offsets to the fields in a state table row.
// Corresponds to struct RBBIStateTableRow in the C version.
//
final static int ACCEPTING = 0;
final static int LOOKAHEAD = 1;
final static int TAGIDX = 2;
final static int RESERVED = 3;
final static int NEXTSTATES = 4;
// Index offsets to header fields of a state table
// struct RBBIStateTable {... in the C version.
//
final static int NUMSTATES = 0;
final static int ROWLEN = 2;
final static int FLAGS = 4;
final static int RESERVED_2 = 6;
final static int ROW_DATA = 8;
// Bit selectors for the "FLAGS" field of the state table header
// enum RBBIStateTableFlags in the C version.
//
final static int RBBI_LOOKAHEAD_HARD_BREAK = 1;
final static int RBBI_BOF_REQUIRED = 2;
/**
* Data Header. A struct-like class with the fields from the RBBI data file header.
*/
final static class RBBIDataHeader {
int fMagic; // == 0xbla0
int fVersion; // == 1 (for ICU 3.2 and earlier.
byte[] fFormatVersion; // For ICU 3.4 and later.
int fLength; // Total length in bytes of this RBBI Data,
// including all sections, not just the header.
int fCatCount; // Number of character categories.
//
// Offsets and sizes of each of the subsections within the RBBI data.
// All offsets are bytes from the start of the RBBIDataHeader.
// All sizes are in bytes.
//
int fFTable; // forward state transition table.
int fFTableLen;
int fRTable; // Offset to the reverse state transition table.
int fRTableLen;
int fSFTable; // safe point forward transition table
int fSFTableLen;
int fSRTable; // safe point reverse transition table
int fSRTableLen;
int fTrie; // Offset to Trie data for character categories
int fTrieLen;
int fRuleSource; // Offset to the source for for the break
int fRuleSourceLen; // rules. Stored UChar *.
int fStatusTable; // Offset to the table of rule status values
int fStatusTableLen;
public RBBIDataHeader() {
fMagic = 0;
fFormatVersion = new byte[4];
}
}
/**
* RBBI State Table Indexing Function. Given a state number, return the
* array index of the start of the state table row for that state.
*
*/
int getRowIndex(int state){
return ROW_DATA + state * (fHeader.fCatCount + 4);
}
static class TrieFoldingFunc implements Trie.DataManipulate {
public int getFoldingOffset(int data) {
if ((data & 0x8000) != 0) {
return data & 0x7fff;
} else {
return 0;
}
}
}
static TrieFoldingFunc fTrieFoldingFunc = new TrieFoldingFunc();
RBBIDataWrapper() {
}
/*
* Get an RBBIDataWrapper from an InputStream onto a pre-compiled set
* of RBBI rules.
*/
static RBBIDataWrapper get(InputStream is) throws IOException {
int i;
DataInputStream dis = new DataInputStream(new BufferedInputStream(is));
RBBIDataWrapper This = new RBBIDataWrapper();
// Seek past the ICU data header.
// TODO: verify that the header looks good.
dis.skip(0x80);
// Read in the RBBI data header...
This.fHeader = new RBBIDataHeader();
This.fHeader.fMagic = dis.readInt();
This.fHeader.fVersion = dis.readInt();
This.fHeader.fFormatVersion[0] = (byte) (This.fHeader.fVersion >> 24);
This.fHeader.fFormatVersion[1] = (byte) (This.fHeader.fVersion >> 16);
This.fHeader.fFormatVersion[2] = (byte) (This.fHeader.fVersion >> 8);
This.fHeader.fFormatVersion[3] = (byte) (This.fHeader.fVersion);
This.fHeader.fLength = dis.readInt();
This.fHeader.fCatCount = dis.readInt();
This.fHeader.fFTable = dis.readInt();
This.fHeader.fFTableLen = dis.readInt();
This.fHeader.fRTable = dis.readInt();
This.fHeader.fRTableLen = dis.readInt();
This.fHeader.fSFTable = dis.readInt();
This.fHeader.fSFTableLen = dis.readInt();
This.fHeader.fSRTable = dis.readInt();
This.fHeader.fSRTableLen = dis.readInt();
This.fHeader.fTrie = dis.readInt();
This.fHeader.fTrieLen = dis.readInt();
This.fHeader.fRuleSource = dis.readInt();
This.fHeader.fRuleSourceLen = dis.readInt();
This.fHeader.fStatusTable = dis.readInt();
This.fHeader.fStatusTableLen = dis.readInt();
dis.skip(6 * 4); // uint32_t fReserved[6];
if (This.fHeader.fMagic != 0xb1a0 ||
! (This.fHeader.fVersion == 1 || // ICU 3.2 and earlier
This.fHeader.fFormatVersion[0] == 3) // ICU 3.4
) {
throw new IOException("Break Iterator Rule Data Magic Number Incorrect, or unsupported data version.");
}
// Current position in input stream.
int pos = 24 * 4; // offset of end of header, which has 24 fields, all int32_t (4 bytes)
//
// Read in the Forward state transition table as an array of shorts.
//
// Quick Sanity Check
if (This.fHeader.fFTable < pos || This.fHeader.fFTable > This.fHeader.fLength) {
throw new IOException("Break iterator Rule data corrupt");
}
// Skip over any padding preceding this table
dis.skip(This.fHeader.fFTable - pos);
pos = This.fHeader.fFTable;
This.fFTable = new short[This.fHeader.fFTableLen / 2];
for ( i=0; i