1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
4 *******************************************************************************
5 * Copyright (C) 2011-2013, International Business Machines
6 * Corporation and others. All Rights Reserved.
7 *******************************************************************************
10 * tab size: 8 (not used)
13 * created on: 2011dec11
14 * created by: Markus W. Scherer
20 #include "unicode/utypes.h"
21 #include "unicode/uniset.h"
22 #include "unicode/unistr.h"
26 /** Additions to the uchar.h enum UProperty. */
29 PPUCD_NAME_ALIAS=UCHAR_STRING_LIMIT,
30 PPUCD_CONDITIONAL_CASE_MAPPINGS,
31 PPUCD_TURKIC_CASE_FOLDING
36 class U_TOOLUTIL_API PropertyNames {
38 virtual ~PropertyNames();
39 virtual int32_t getPropertyEnum(const char *name) const;
40 virtual int32_t getPropertyValueEnum(int32_t property, const char *name) const;
43 struct U_TOOLUTIL_API UniProps {
47 int32_t getIntProp(int32_t prop) const { return intProps[prop-UCHAR_INT_START]; }
50 UBool binProps[UCHAR_BINARY_LIMIT];
51 int32_t intProps[UCHAR_INT_LIMIT-UCHAR_INT_START];
54 UChar32 scf, slc, stc, suc;
56 const char *numericValue;
58 const char *nameAlias;
59 UnicodeString cf, lc, tc, uc;
63 class U_TOOLUTIL_API PreparsedUCD {
66 /** No line, end of file. */
68 /** Empty line. (Might contain a comment.) */
74 /** property;Binary;Alpha;Alphabetic */
76 /** binary;N;No;F;False */
78 /** value;gc;Zs;Space_Separator */
81 /** defaults;0000..10FFFF;age=NA;bc=L;... */
83 /** block;0000..007F;age=1.1;blk=ASCII;ea=Na;... */
85 /** cp;0030;AHex;bc=EN;gc=Nd;na=DIGIT ZERO;... */
88 /** algnamesrange;4E00..9FCC;han;CJK UNIFIED IDEOGRAPH- */
96 * Prepare this object for a new, empty package.
98 PreparsedUCD(const char *filename, UErrorCode &errorCode);
103 /** Sets (aliases) a non-standard PropertyNames implementation. Caller retains ownership. */
104 void setPropertyNames(const PropertyNames *pn) { pnames=pn; }
107 * Reads a line from the preparsed UCD file.
108 * Splits the line by replacing each ';' with a NUL.
110 LineType readLine(UErrorCode &errorCode);
112 /** Returns the number of the line read by readLine(). */
113 int32_t getLineNumber() const { return lineNumber; }
115 /** Returns the line's next field, or NULL. */
116 const char *nextField();
118 /** Returns the Unicode version when or after the UNICODE_VERSION_LINE has been read. */
119 const UVersionInfo &getUnicodeVersion() const { return ucdVersion; }
121 /** Returns TRUE if the current line has property values. */
122 UBool lineHasPropertyValues() const { return DEFAULTS_LINE<=lineType && lineType<=CP_LINE; }
125 * Parses properties from the current line.
126 * Clears newValues and sets UProperty codes for property values mentioned
127 * on the current line (as opposed to being inherited).
128 * Returns a pointer to the filled-in UniProps, or NULL if something went wrong.
129 * The returned UniProps are usable until the next line of the same type is read.
131 const UniProps *getProps(UnicodeSet &newValues, UErrorCode &errorCode);
134 * Returns the code point range for the current algnamesrange line.
135 * Calls & parses nextField().
136 * Further nextField() calls will yield the range's type & prefix string.
137 * Returns U_SUCCESS(errorCode).
139 UBool getRangeForAlgNames(UChar32 &start, UChar32 &end, UErrorCode &errorCode);
142 UBool isLineBufferAvailable(int32_t i) {
143 return defaultLineIndex!=i && blockLineIndex!=i;
146 /** Resets the field iterator and returns the line's first field (the line type field). */
147 const char *firstField();
149 UBool parseProperty(UniProps &props, const char *field, UnicodeSet &newValues,
150 UErrorCode &errorCode);
151 UChar32 parseCodePoint(const char *s, UErrorCode &errorCode);
152 UBool parseCodePointRange(const char *s, UChar32 &start, UChar32 &end, UErrorCode &errorCode);
153 void parseString(const char *s, UnicodeString &uni, UErrorCode &errorCode);
154 void parseScriptExtensions(const char *s, UnicodeSet &scx, UErrorCode &errorCode);
156 static const int32_t kNumLineBuffers=3;
158 PropertyNames *icuPnames; // owned
159 const PropertyNames *pnames; // aliased
161 int32_t defaultLineIndex, blockLineIndex, lineIndex;
167 UVersionInfo ucdVersion;
168 UniProps defaultProps, blockProps, cpProps;
169 // Multiple lines so that default and block properties can maintain pointers
170 // into their line buffers.
171 char lines[kNumLineBuffers][4096];
176 #endif // __PPUCD_H__