1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/tools/toolutil/ppucd.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,174 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +* Copyright (C) 2011-2013, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +*******************************************************************************
1.9 +* file name: ppucd.h
1.10 +* encoding: US-ASCII
1.11 +* tab size: 8 (not used)
1.12 +* indentation:4
1.13 +*
1.14 +* created on: 2011dec11
1.15 +* created by: Markus W. Scherer
1.16 +*/
1.17 +
1.18 +#ifndef __PPUCD_H__
1.19 +#define __PPUCD_H__
1.20 +
1.21 +#include "unicode/utypes.h"
1.22 +#include "unicode/uniset.h"
1.23 +#include "unicode/unistr.h"
1.24 +
1.25 +#include <stdio.h>
1.26 +
1.27 +/** Additions to the uchar.h enum UProperty. */
1.28 +enum {
1.29 + /** Name_Alias */
1.30 + PPUCD_NAME_ALIAS=UCHAR_STRING_LIMIT,
1.31 + PPUCD_CONDITIONAL_CASE_MAPPINGS,
1.32 + PPUCD_TURKIC_CASE_FOLDING
1.33 +};
1.34 +
1.35 +U_NAMESPACE_BEGIN
1.36 +
1.37 +class U_TOOLUTIL_API PropertyNames {
1.38 +public:
1.39 + virtual ~PropertyNames();
1.40 + virtual int32_t getPropertyEnum(const char *name) const;
1.41 + virtual int32_t getPropertyValueEnum(int32_t property, const char *name) const;
1.42 +};
1.43 +
1.44 +struct U_TOOLUTIL_API UniProps {
1.45 + UniProps();
1.46 + ~UniProps();
1.47 +
1.48 + int32_t getIntProp(int32_t prop) const { return intProps[prop-UCHAR_INT_START]; }
1.49 +
1.50 + UChar32 start, end;
1.51 + UBool binProps[UCHAR_BINARY_LIMIT];
1.52 + int32_t intProps[UCHAR_INT_LIMIT-UCHAR_INT_START];
1.53 + UVersionInfo age;
1.54 + UChar32 bmg, bpb;
1.55 + UChar32 scf, slc, stc, suc;
1.56 + int32_t digitValue;
1.57 + const char *numericValue;
1.58 + const char *name;
1.59 + const char *nameAlias;
1.60 + UnicodeString cf, lc, tc, uc;
1.61 + UnicodeSet scx;
1.62 +};
1.63 +
1.64 +class U_TOOLUTIL_API PreparsedUCD {
1.65 +public:
1.66 + enum LineType {
1.67 + /** No line, end of file. */
1.68 + NO_LINE,
1.69 + /** Empty line. (Might contain a comment.) */
1.70 + EMPTY_LINE,
1.71 +
1.72 + /** ucd;6.1.0 */
1.73 + UNICODE_VERSION_LINE,
1.74 +
1.75 + /** property;Binary;Alpha;Alphabetic */
1.76 + PROPERTY_LINE,
1.77 + /** binary;N;No;F;False */
1.78 + BINARY_LINE,
1.79 + /** value;gc;Zs;Space_Separator */
1.80 + VALUE_LINE,
1.81 +
1.82 + /** defaults;0000..10FFFF;age=NA;bc=L;... */
1.83 + DEFAULTS_LINE,
1.84 + /** block;0000..007F;age=1.1;blk=ASCII;ea=Na;... */
1.85 + BLOCK_LINE,
1.86 + /** cp;0030;AHex;bc=EN;gc=Nd;na=DIGIT ZERO;... */
1.87 + CP_LINE,
1.88 +
1.89 + /** algnamesrange;4E00..9FCC;han;CJK UNIFIED IDEOGRAPH- */
1.90 + ALG_NAMES_RANGE_LINE,
1.91 +
1.92 + LINE_TYPE_COUNT
1.93 + };
1.94 +
1.95 + /**
1.96 + * Constructor.
1.97 + * Prepare this object for a new, empty package.
1.98 + */
1.99 + PreparsedUCD(const char *filename, UErrorCode &errorCode);
1.100 +
1.101 + /** Destructor. */
1.102 + ~PreparsedUCD();
1.103 +
1.104 + /** Sets (aliases) a non-standard PropertyNames implementation. Caller retains ownership. */
1.105 + void setPropertyNames(const PropertyNames *pn) { pnames=pn; }
1.106 +
1.107 + /**
1.108 + * Reads a line from the preparsed UCD file.
1.109 + * Splits the line by replacing each ';' with a NUL.
1.110 + */
1.111 + LineType readLine(UErrorCode &errorCode);
1.112 +
1.113 + /** Returns the number of the line read by readLine(). */
1.114 + int32_t getLineNumber() const { return lineNumber; }
1.115 +
1.116 + /** Returns the line's next field, or NULL. */
1.117 + const char *nextField();
1.118 +
1.119 + /** Returns the Unicode version when or after the UNICODE_VERSION_LINE has been read. */
1.120 + const UVersionInfo &getUnicodeVersion() const { return ucdVersion; }
1.121 +
1.122 + /** Returns TRUE if the current line has property values. */
1.123 + UBool lineHasPropertyValues() const { return DEFAULTS_LINE<=lineType && lineType<=CP_LINE; }
1.124 +
1.125 + /**
1.126 + * Parses properties from the current line.
1.127 + * Clears newValues and sets UProperty codes for property values mentioned
1.128 + * on the current line (as opposed to being inherited).
1.129 + * Returns a pointer to the filled-in UniProps, or NULL if something went wrong.
1.130 + * The returned UniProps are usable until the next line of the same type is read.
1.131 + */
1.132 + const UniProps *getProps(UnicodeSet &newValues, UErrorCode &errorCode);
1.133 +
1.134 + /**
1.135 + * Returns the code point range for the current algnamesrange line.
1.136 + * Calls & parses nextField().
1.137 + * Further nextField() calls will yield the range's type & prefix string.
1.138 + * Returns U_SUCCESS(errorCode).
1.139 + */
1.140 + UBool getRangeForAlgNames(UChar32 &start, UChar32 &end, UErrorCode &errorCode);
1.141 +
1.142 +private:
1.143 + UBool isLineBufferAvailable(int32_t i) {
1.144 + return defaultLineIndex!=i && blockLineIndex!=i;
1.145 + }
1.146 +
1.147 + /** Resets the field iterator and returns the line's first field (the line type field). */
1.148 + const char *firstField();
1.149 +
1.150 + UBool parseProperty(UniProps &props, const char *field, UnicodeSet &newValues,
1.151 + UErrorCode &errorCode);
1.152 + UChar32 parseCodePoint(const char *s, UErrorCode &errorCode);
1.153 + UBool parseCodePointRange(const char *s, UChar32 &start, UChar32 &end, UErrorCode &errorCode);
1.154 + void parseString(const char *s, UnicodeString &uni, UErrorCode &errorCode);
1.155 + void parseScriptExtensions(const char *s, UnicodeSet &scx, UErrorCode &errorCode);
1.156 +
1.157 + static const int32_t kNumLineBuffers=3;
1.158 +
1.159 + PropertyNames *icuPnames; // owned
1.160 + const PropertyNames *pnames; // aliased
1.161 + FILE *file;
1.162 + int32_t defaultLineIndex, blockLineIndex, lineIndex;
1.163 + int32_t lineNumber;
1.164 + LineType lineType;
1.165 + char *fieldLimit;
1.166 + char *lineLimit;
1.167 +
1.168 + UVersionInfo ucdVersion;
1.169 + UniProps defaultProps, blockProps, cpProps;
1.170 + // Multiple lines so that default and block properties can maintain pointers
1.171 + // into their line buffers.
1.172 + char lines[kNumLineBuffers][4096];
1.173 +};
1.174 +
1.175 +U_NAMESPACE_END
1.176 +
1.177 +#endif // __PPUCD_H__
