/*
*******************************************************************************
*
-* Copyright (C) 2004, International Business Machines
+* Copyright (C) 2004-2012, International Business Machines
* Corporation and others. All Rights Reserved.
*
*******************************************************************************
#include "unicode/utypes.h"
#include "unicode/uset.h"
+#include "putilimp.h"
#include "uset_imp.h"
#include "udataswp.h"
-U_CDECL_BEGIN
+#ifdef __cplusplus
+U_NAMESPACE_BEGIN
-/* library API -------------------------------------------------------------- */
+class UnicodeString;
-struct UCaseProps;
-typedef struct UCaseProps UCaseProps;
+U_NAMESPACE_END
+#endif
-U_CAPI UCaseProps * U_EXPORT2
-ucase_open(UErrorCode *pErrorCode);
+/* library API -------------------------------------------------------------- */
-U_CAPI UCaseProps * U_EXPORT2
-ucase_openBinary(const uint8_t *bin, int32_t length, UErrorCode *pErrorCode);
+U_CDECL_BEGIN
-U_CAPI void U_EXPORT2
-ucase_close(UCaseProps *csp);
+struct UCaseProps;
+typedef struct UCaseProps UCaseProps;
+U_CDECL_END
-U_CAPI UCaseProps * U_EXPORT2
-ucase_getSingleton(UErrorCode *pErrorCode);
+U_CAPI const UCaseProps * U_EXPORT2
+ucase_getSingleton(void);
+U_CFUNC void U_EXPORT2
+ucase_addPropertyStarts(const UCaseProps *csp, const USetAdder *sa, UErrorCode *pErrorCode);
-U_CAPI int32_t U_EXPORT2
-ucase_swap(const UDataSwapper *ds,
- const void *inData, int32_t length, void *outData,
- UErrorCode *pErrorCode);
+/**
+ * Requires non-NULL locale ID but otherwise does the equivalent of
+ * checking for language codes as if uloc_getLanguage() were called:
+ * Accepts both 2- and 3-letter codes and accepts case variants.
+ */
+U_CFUNC int32_t
+ucase_getCaseLocale(const char *locale, int32_t *locCache);
-U_CAPI void U_EXPORT2
-ucase_addPropertyStarts(const UCaseProps *csp, USetAdder *sa, UErrorCode *pErrorCode);
+/* Casing locale types for ucase_getCaseLocale */
+enum {
+ UCASE_LOC_UNKNOWN,
+ UCASE_LOC_ROOT,
+ UCASE_LOC_TURKISH,
+ UCASE_LOC_LITHUANIAN,
+ UCASE_LOC_DUTCH
+};
/**
* Bit mask for getting just the options from a string compare options word
ucase_totitle(const UCaseProps *csp, UChar32 c);
U_CAPI UChar32 U_EXPORT2
-ucase_fold(UCaseProps *csp, UChar32 c, uint32_t options);
+ucase_fold(const UCaseProps *csp, UChar32 c, uint32_t options);
+
+/**
+ * Adds all simple case mappings and the full case folding for c to sa,
+ * and also adds special case closure mappings.
+ * c itself is not added.
+ * For example, the mappings
+ * - for s include long s
+ * - for sharp s include ss
+ * - for k include the Kelvin sign
+ */
+U_CFUNC void U_EXPORT2
+ucase_addCaseClosure(const UCaseProps *csp, UChar32 c, const USetAdder *sa);
+
+/**
+ * Maps the string to single code points and adds the associated case closure
+ * mappings.
+ * The string is mapped to code points if it is their full case folding string.
+ * In other words, this performs a reverse full case folding and then
+ * adds the case closure items of the resulting code points.
+ * If the string is found and its closure applied, then
+ * the string itself is added as well as part of its code points' closure.
+ * It must be length>=0.
+ *
+ * @return TRUE if the string was found
+ */
+U_CFUNC UBool U_EXPORT2
+ucase_addStringCaseClosure(const UCaseProps *csp, const UChar *s, int32_t length, const USetAdder *sa);
+
+#ifdef __cplusplus
+U_NAMESPACE_BEGIN
+
+/**
+ * Iterator over characters with more than one code point in the full default Case_Folding.
+ */
+class U_COMMON_API FullCaseFoldingIterator {
+public:
+ /** Constructor. */
+ FullCaseFoldingIterator();
+ /**
+ * Returns the next (cp, full) pair where "full" is cp's full default Case_Folding.
+ * Returns a negative cp value at the end of the iteration.
+ */
+ UChar32 next(UnicodeString &full);
+private:
+ FullCaseFoldingIterator(const FullCaseFoldingIterator &); // no copy
+ FullCaseFoldingIterator &operator=(const FullCaseFoldingIterator &); // no assignment
+
+ const UChar *unfold;
+ int32_t unfoldRows;
+ int32_t unfoldRowWidth;
+ int32_t unfoldStringWidth;
+ int32_t currentRow;
+ int32_t rowCpIndex;
+};
+
+U_NAMESPACE_END
+#endif
/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
U_CAPI int32_t U_EXPORT2
/* string case mapping functions */
+U_CDECL_BEGIN
+
/**
* Iterator function for string case mappings, which need to look at the
* context (surrounding text) of a given character for conditional mappings.
* It does not support random access or an arbitrary change of
* iteration direction.
*
- * The direction parameter either starts
+ * The code point being case-mapped itself is never returned by
+ * this iterator.
*
* @param context A pointer to the iterator's working data.
* @param dir If <0 then start iterating backward from the character;
};
typedef struct UCaseContext UCaseContext;
+U_CDECL_END
+
+#define UCASECONTEXT_INITIALIZER { NULL, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+
enum {
/**
* For string case mappings, a single character (a code point) is mapped
const UChar **pString,
uint32_t options);
+U_CFUNC int32_t U_EXPORT2
+ucase_hasBinaryProperty(UChar32 c, UProperty which);
+
+
+U_CDECL_BEGIN
+
+/**
+ * @internal
+ */
+typedef int32_t U_CALLCONV
+UCaseMapFull(const UCaseProps *csp, UChar32 c,
+ UCaseContextIterator *iter, void *context,
+ const UChar **pString,
+ const char *locale, int32_t *locCache);
+
+U_CDECL_END
+
/* file definitions --------------------------------------------------------- */
#define UCASE_DATA_NAME "ucase"
UCASE_IX_LENGTH,
UCASE_IX_TRIE_SIZE,
UCASE_IX_EXC_LENGTH,
+ UCASE_IX_UNFOLD_LENGTH,
UCASE_IX_MAX_FULL_LENGTH=15,
UCASE_IX_TOP=16
UCASE_TITLE
};
-#define UCASE_SENSITIVE 4
-#define UCASE_EXCEPTION 8
+#define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
+#define UCASE_GET_TYPE_AND_IGNORABLE(props) ((props)&7)
+
+#define UCASE_IGNORABLE 4
+#define UCASE_SENSITIVE 8
+#define UCASE_EXCEPTION 0x10
-#define UCASE_DOT_MASK 0x30
+#define UCASE_DOT_MASK 0x60
enum {
UCASE_NO_DOT=0, /* normal characters with cc=0 */
- UCASE_SOFT_DOTTED=0x10, /* soft-dotted characters with cc=0 */
- UCASE_ABOVE=0x20, /* "above" accents with cc=230 */
- UCASE_OTHER_ACCENT=0x30 /* other accent character (0<cc!=230) */
+ UCASE_SOFT_DOTTED=0x20, /* soft-dotted characters with cc=0 */
+ UCASE_ABOVE=0x40, /* "above" accents with cc=230 */
+ UCASE_OTHER_ACCENT=0x60 /* other accent character (0<cc!=230) */
};
-/* no exception: bits 15..6 are a 10-bit signed case mapping delta */
-#define UCASE_DELTA_SHIFT 6
-#define UCASE_DELTA_MASK 0xffc0
-#define UCASE_MAX_DELTA 0x1ff
+/* no exception: bits 15..7 are a 9-bit signed case mapping delta */
+#define UCASE_DELTA_SHIFT 7
+#define UCASE_DELTA_MASK 0xff80
+#define UCASE_MAX_DELTA 0xff
#define UCASE_MIN_DELTA (-UCASE_MAX_DELTA-1)
-#define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
-
-/* case-ignorable uses one of the delta bits, see gencase/store.c */
-#define UCASE_CASE_IGNORABLE 0x40
+#if U_SIGNED_RIGHT_SHIFT_IS_ARITHMETIC
+# define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
+#else
+# define UCASE_GET_DELTA(props) (int16_t)(((props)&0x8000) ? (((props)>>UCASE_DELTA_SHIFT)|0xfe00) : ((uint16_t)(props)>>UCASE_DELTA_SHIFT))
+#endif
-/* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */
-#define UCASE_EXC_SHIFT 4
-#define UCASE_EXC_MASK 0xfff0
-#define UCASE_MAX_EXCEPTIONS 0x1000
+/* exception: bits 15..5 are an unsigned 11-bit index into the exceptions array */
+#define UCASE_EXC_SHIFT 5
+#define UCASE_EXC_MASK 0xffe0
+#define UCASE_MAX_EXCEPTIONS ((UCASE_EXC_MASK>>UCASE_EXC_SHIFT)+1)
/* definitions for 16-bit main exceptions word ------------------------------ */
UCASE_EXC_TITLE,
UCASE_EXC_4, /* reserved */
UCASE_EXC_5, /* reserved */
- UCASE_EXC_6, /* reserved */
+ UCASE_EXC_CLOSURE,
UCASE_EXC_FULL_MAPPINGS,
UCASE_EXC_ALL_SLOTS /* one past the last slot */
};
/* reserved: exception bits 11..9 */
/* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
-#define UCASE_EXC_DOT_SHIFT 8
+#define UCASE_EXC_DOT_SHIFT 7
/* normally stored in the main word, but pushed out for larger exception indexes */
#define UCASE_EXC_DOT_MASK 0x3000
#define UCASE_FULL_UPPER 0xf00
#define UCASE_FULL_TITLE 0xf000
-U_CDECL_END
+/* maximum lengths */
+#define UCASE_FULL_MAPPINGS_MAX_LENGTH (4*0xf)
+#define UCASE_CLOSURE_MAX_LENGTH 0xf
+
+/* constants for reverse case folding ("unfold") data */
+enum {
+ UCASE_UNFOLD_ROWS,
+ UCASE_UNFOLD_ROW_WIDTH,
+ UCASE_UNFOLD_STRING_WIDTH
+};
#endif
projects / apple / icu.git / blobdiff