ICU-62107.0.1.tar.gz

[apple/icu.git] / icuSources / common / ucnv_ext.h

diff --git a/icuSources/common/ucnv_ext.h b/icuSources/common/ucnv_ext.h
index 6ccd7669c44e3422c6ec3c8f2aef6418977817b5..dceea7ef12617a3254855c7eeef31e60f809e7ba 100644 (file)
--- a/icuSources/common/ucnv_ext.h
+++ b/icuSources/common/ucnv_ext.h
@@ -1,12 +1,14 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
 /*
 ******************************************************************************
 *
-*   Copyright (C) 2003-2004, International Business Machines
+*   Copyright (C) 2003-2013, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 ******************************************************************************
 *   file name:  ucnv_ext.h
-*   encoding:   US-ASCII
+*   encoding:   UTF-8
 *   tab size:   8 (not used)
 *   indentation:4
 *
@@ -29,10 +31,12 @@
 /*
  * See icuhtml/design/conversion/conversion_extensions.html
  *
- * Conversion extensions serve two purposes:
+ * Conversion extensions serve three purposes:
  * 1. They support m:n mappings.
  * 2. They support extension-only conversion files that are used together
  *    with the regular conversion data in base files.
+ * 3. They support mappings with more complicated meta data,
+ *    for example "good one-way" mappings (|4).
  *
  * A base file may contain an extension table (explicitly requested or
  * implicitly generated for m:n mappings), but its extension table is not
@@ -229,11 +233,13 @@
  *         return no mapping, but request for <subchar1>;
  *       }
  *       if(bit 31 set) {
- *         roundtrip;
+ *         roundtrip (|0);
+ *       } else if(bit 30 set) {
+ *         "good one-way" mapping (|4); -- new in ICU4C 51, _MBCSHeader.version 5.4/4.4
  *       } else {
- *         fallback;
+ *         normal fallback (|1);
  *       }
- *       // bits 30..29 reserved, 0
+ *       // bit 29 reserved, 0
  *       length=(value>>24)&0x1f; (bits 28..24)
  *       if(length==1..3) {
  *         bits 23..0 contain 1..3 bytes, padded with 00s on the left;
@@ -382,10 +388,20 @@ ucnv_extContinueMatchFromU(UConverter *cnv,
                            UConverterFromUnicodeArgs *pArgs, int32_t srcIndex,
                            UErrorCode *pErrorCode);
 
+/*
+ * Add code points and strings to the set according to the extension mappings.
+ * Limitation on the UConverterSetFilter:
+ * The filters currently assume that they are used with 1:1 mappings.
+ * They only apply to single input code points, and then they pass through
+ * only mappings with single-charset-code results.
+ * For example, the Shift-JIS filter only works for 2-byte results and tests
+ * that those 2 bytes are in the JIS X 0208 range of Shift-JIS.
+ */
 U_CFUNC void
 ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData,
                       const USetAdder *sa,
                       UConverterUnicodeSet which,
+                      UConverterSetFilter filter,
                       UErrorCode *pErrorCode);
 
 /* toUnicode helpers -------------------------------------------------------- */
@@ -434,7 +450,9 @@ ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData,
 
 #define UCNV_EXT_FROM_U_LENGTH_SHIFT 24
 #define UCNV_EXT_FROM_U_ROUNDTRIP_FLAG ((uint32_t)1<<31)
-#define UCNV_EXT_FROM_U_RESERVED_MASK 0x60000000
+#define UCNV_EXT_FROM_U_GOOD_ONE_WAY_FLAG 0x40000000
+#define UCNV_EXT_FROM_U_STATUS_MASK 0xc0000000
+#define UCNV_EXT_FROM_U_RESERVED_MASK 0x20000000
 #define UCNV_EXT_FROM_U_DATA_MASK 0xffffff
 
 /* special value for "no mapping" to <subchar1> (impossible roundtrip to 0 bytes, value 01) */
@@ -452,7 +470,7 @@ ucnv_extGetUnicodeSet(const UConverterSharedData *sharedData,
 #define UCNV_EXT_FROM_U_IS_ROUNDTRIP(value) (((value)&UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)!=0)
 #define UCNV_EXT_FROM_U_MASK_ROUNDTRIP(value) ((value)&~UCNV_EXT_FROM_U_ROUNDTRIP_FLAG)
 
-/* use after masking off the roundtrip flag */
+/* get length; masks away all other bits */
 #define UCNV_EXT_FROM_U_GET_LENGTH(value) (int32_t)(((value)>>UCNV_EXT_FROM_U_LENGTH_SHIFT)&UCNV_EXT_MAX_BYTES)
 
 /* get bytes or bytes index */