michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2011-2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  ppucd.h
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2011dec11
michael@0: *   created by: Markus W. Scherer
michael@0: */
michael@0: 
michael@0: #ifndef __PPUCD_H__
michael@0: #define __PPUCD_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "unicode/uniset.h"
michael@0: #include "unicode/unistr.h"
michael@0: 
michael@0: #include <stdio.h>
michael@0: 
michael@0: /** Additions to the uchar.h enum UProperty. */
michael@0: enum {
michael@0:     /** Name_Alias */
michael@0:     PPUCD_NAME_ALIAS=UCHAR_STRING_LIMIT,
michael@0:     PPUCD_CONDITIONAL_CASE_MAPPINGS,
michael@0:     PPUCD_TURKIC_CASE_FOLDING
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: class U_TOOLUTIL_API PropertyNames {
michael@0: public:
michael@0:     virtual ~PropertyNames();
michael@0:     virtual int32_t getPropertyEnum(const char *name) const;
michael@0:     virtual int32_t getPropertyValueEnum(int32_t property, const char *name) const;
michael@0: };
michael@0: 
michael@0: struct U_TOOLUTIL_API UniProps {
michael@0:     UniProps();
michael@0:     ~UniProps();
michael@0: 
michael@0:     int32_t getIntProp(int32_t prop) const { return intProps[prop-UCHAR_INT_START]; }
michael@0: 
michael@0:     UChar32 start, end;
michael@0:     UBool binProps[UCHAR_BINARY_LIMIT];
michael@0:     int32_t intProps[UCHAR_INT_LIMIT-UCHAR_INT_START];
michael@0:     UVersionInfo age;
michael@0:     UChar32 bmg, bpb;
michael@0:     UChar32 scf, slc, stc, suc;
michael@0:     int32_t digitValue;
michael@0:     const char *numericValue;
michael@0:     const char *name;
michael@0:     const char *nameAlias;
michael@0:     UnicodeString cf, lc, tc, uc;
michael@0:     UnicodeSet scx;
michael@0: };
michael@0: 
michael@0: class U_TOOLUTIL_API PreparsedUCD {
michael@0: public:
michael@0:     enum LineType {
michael@0:         /** No line, end of file. */
michael@0:         NO_LINE,
michael@0:         /** Empty line. (Might contain a comment.) */
michael@0:         EMPTY_LINE,
michael@0: 
michael@0:         /** ucd;6.1.0 */
michael@0:         UNICODE_VERSION_LINE,
michael@0: 
michael@0:         /** property;Binary;Alpha;Alphabetic */
michael@0:         PROPERTY_LINE,
michael@0:         /** binary;N;No;F;False */
michael@0:         BINARY_LINE,
michael@0:         /** value;gc;Zs;Space_Separator */
michael@0:         VALUE_LINE,
michael@0: 
michael@0:         /** defaults;0000..10FFFF;age=NA;bc=L;... */
michael@0:         DEFAULTS_LINE,
michael@0:         /** block;0000..007F;age=1.1;blk=ASCII;ea=Na;... */
michael@0:         BLOCK_LINE,
michael@0:         /** cp;0030;AHex;bc=EN;gc=Nd;na=DIGIT ZERO;... */
michael@0:         CP_LINE,
michael@0: 
michael@0:         /** algnamesrange;4E00..9FCC;han;CJK UNIFIED IDEOGRAPH- */
michael@0:         ALG_NAMES_RANGE_LINE,
michael@0: 
michael@0:         LINE_TYPE_COUNT
michael@0:     };
michael@0: 
michael@0:     /**
michael@0:      * Constructor.
michael@0:      * Prepare this object for a new, empty package.
michael@0:      */
michael@0:     PreparsedUCD(const char *filename, UErrorCode &errorCode);
michael@0: 
michael@0:     /** Destructor. */
michael@0:     ~PreparsedUCD();
michael@0: 
michael@0:     /** Sets (aliases) a non-standard PropertyNames implementation. Caller retains ownership. */
michael@0:     void setPropertyNames(const PropertyNames *pn) { pnames=pn; }
michael@0: 
michael@0:     /**
michael@0:      * Reads a line from the preparsed UCD file.
michael@0:      * Splits the line by replacing each ';' with a NUL.
michael@0:      */
michael@0:     LineType readLine(UErrorCode &errorCode);
michael@0: 
michael@0:     /** Returns the number of the line read by readLine(). */
michael@0:     int32_t getLineNumber() const { return lineNumber; }
michael@0: 
michael@0:     /** Returns the line's next field, or NULL. */
michael@0:     const char *nextField();
michael@0: 
michael@0:     /** Returns the Unicode version when or after the UNICODE_VERSION_LINE has been read. */
michael@0:     const UVersionInfo &getUnicodeVersion() const { return ucdVersion; }
michael@0: 
michael@0:     /** Returns TRUE if the current line has property values. */
michael@0:     UBool lineHasPropertyValues() const { return DEFAULTS_LINE<=lineType && lineType<=CP_LINE; }
michael@0: 
michael@0:     /**
michael@0:      * Parses properties from the current line.
michael@0:      * Clears newValues and sets UProperty codes for property values mentioned
michael@0:      * on the current line (as opposed to being inherited).
michael@0:      * Returns a pointer to the filled-in UniProps, or NULL if something went wrong.
michael@0:      * The returned UniProps are usable until the next line of the same type is read.
michael@0:      */
michael@0:     const UniProps *getProps(UnicodeSet &newValues, UErrorCode &errorCode);
michael@0: 
michael@0:     /**
michael@0:      * Returns the code point range for the current algnamesrange line.
michael@0:      * Calls & parses nextField().
michael@0:      * Further nextField() calls will yield the range's type & prefix string.
michael@0:      * Returns U_SUCCESS(errorCode).
michael@0:      */
michael@0:     UBool getRangeForAlgNames(UChar32 &start, UChar32 &end, UErrorCode &errorCode);
michael@0: 
michael@0: private:
michael@0:     UBool isLineBufferAvailable(int32_t i) {
michael@0:         return defaultLineIndex!=i && blockLineIndex!=i;
michael@0:     }
michael@0: 
michael@0:     /** Resets the field iterator and returns the line's first field (the line type field). */
michael@0:     const char *firstField();
michael@0: 
michael@0:     UBool parseProperty(UniProps &props, const char *field, UnicodeSet &newValues,
michael@0:                         UErrorCode &errorCode);
michael@0:     UChar32 parseCodePoint(const char *s, UErrorCode &errorCode);
michael@0:     UBool parseCodePointRange(const char *s, UChar32 &start, UChar32 &end, UErrorCode &errorCode);
michael@0:     void parseString(const char *s, UnicodeString &uni, UErrorCode &errorCode);
michael@0:     void parseScriptExtensions(const char *s, UnicodeSet &scx, UErrorCode &errorCode);
michael@0: 
michael@0:     static const int32_t kNumLineBuffers=3;
michael@0: 
michael@0:     PropertyNames *icuPnames;  // owned
michael@0:     const PropertyNames *pnames;  // aliased
michael@0:     FILE *file;
michael@0:     int32_t defaultLineIndex, blockLineIndex, lineIndex;
michael@0:     int32_t lineNumber;
michael@0:     LineType lineType;
michael@0:     char *fieldLimit;
michael@0:     char *lineLimit;
michael@0: 
michael@0:     UVersionInfo ucdVersion;
michael@0:     UniProps defaultProps, blockProps, cpProps;
michael@0:     // Multiple lines so that default and block properties can maintain pointers
michael@0:     // into their line buffers.
michael@0:     char lines[kNumLineBuffers][4096];
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif  // __PPUCD_H__