ICU-2758 add APIs for FCD properties lccc & tccc

X-SVN-Rev: 14894

icu4c/source/common/unicode/uchar.h

@@ -351,6 +351,20 @@ typedef enum UProperty {
     /** Enumerated property NFKC_Quick_Check.
         Returns UNormalizationCheckResult values. @draft ICU 3.0 */
     UCHAR_NFKC_QUICK_CHECK,
+    /** Enumerated property Lead_Canonical_Combining_Class.
+        ICU-specific property for the ccc of the first code point
+        of the decomposition, or lccc(c)=ccc(NFD(c)[0]).
+        Useful for checking for canonically ordered text;
+        see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD .
+        Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @draft ICU 3.0 */
+    UCHAR_LEAD_CANONICAL_COMBINING_CLASS,
+    /** Enumerated property Trail_Canonical_Combining_Class.
+        ICU-specific property for the ccc of the last code point
+        of the decomposition, or tccc(c)=ccc(NFD(c)[last]).
+        Useful for checking for canonically ordered text;
+        see UNORM_FCD and http://www.unicode.org/notes/tn5/#FCD .
+        Returns 8-bit numeric values like UCHAR_CANONICAL_COMBINING_CLASS. @draft ICU 3.0 */
+    UCHAR_TRAIL_CANONICAL_COMBINING_CLASS,
     /** One more than the last constant for enumerated/integer Unicode properties. @stable ICU 2.2 */
     UCHAR_INT_LIMIT,
 

icu4c/source/common/unorm.cpp

@@ -1189,6 +1189,20 @@ unorm_getQuickCheck(UChar32 c, UNormalizationMode mode) {
     }
 }
 
+U_CAPI uint16_t U_EXPORT2
+unorm_getFCD16FromCodePoint(UChar32 c) {
+    UErrorCode errorCode;
+    uint16_t fcd;
+
+    errorCode=U_ZERO_ERROR;
+    if(!_haveData(errorCode)) {
+        return 0;
+    }
+
+    UTRIE_GET16(&fcdTrie, c, fcd);
+    return fcd;
+}
+
 /* reorder UTF-16 in-place -------------------------------------------------- */
 
 /*

icu4c/source/common/unormimp.h

@@ -265,6 +265,14 @@ unorm_cmpEquivFold(const UChar *s1, int32_t length1,
 
 #if !UCONFIG_NO_NORMALIZATION
 
+/**
+ * Internal API to get the 16-bit FCD value (lccc + tccc) for c,
+ * for u_getIntPropertyValue().
+ * @internal
+ */
+U_CAPI uint16_t U_EXPORT2
+unorm_getFCD16FromCodePoint(UChar32 c);
+
 /**
  * Internal API, used by collation code.
  * Get access to the internal FCD trie table to be able to perform

icu4c/source/common/uprops.c

@@ -354,6 +354,10 @@ u_getIntPropertyValue(UChar32 c, UProperty which) {
         case UCHAR_NFC_QUICK_CHECK:
         case UCHAR_NFKC_QUICK_CHECK:
             return (int32_t)unorm_getQuickCheck(c, (UNormalizationMode)(which-UCHAR_NFD_QUICK_CHECK)+UNORM_NFD);
+        case UCHAR_LEAD_CANONICAL_COMBINING_CLASS:
+            return unorm_getFCD16FromCodePoint(c)>>8;
+        case UCHAR_TRAIL_CANONICAL_COMBINING_CLASS:
+            return unorm_getFCD16FromCodePoint(c)&0xff;
         default:
             return 0; /* undefined */
         }
@@ -387,6 +391,8 @@ u_getIntPropertyMaxValue(UProperty which) {
             max=(uprv_getMaxValues(0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT;
             return max!=0 ? max : (int32_t)UBLOCK_COUNT-1;
         case UCHAR_CANONICAL_COMBINING_CLASS:
+        case UCHAR_LEAD_CANONICAL_COMBINING_CLASS:
+        case UCHAR_TRAIL_CANONICAL_COMBINING_CLASS:
             return 0xff; /* TODO do we need to be more precise, getting the actual maximum? */
         case UCHAR_DECOMPOSITION_TYPE:
             max=uprv_getMaxValues(2)&UPROPS_DT_MASK;

icu4c/source/test/cintltst/cnormtst.c

@@ -1388,18 +1388,20 @@ TestFCNFKCClosure(void) {
 static void
 TestQuickCheckPerCP() {
     UErrorCode errorCode;
-    UChar32 c;
-    UChar s[U16_MAX_LENGTH];
-    int32_t length;
+    UChar32 c, lead, trail;
+    UChar s[U16_MAX_LENGTH], nfd[16];
+    int32_t length, lccc1, lccc2, tccc1, tccc2;
     UNormalizationCheckResult qc1, qc2;
 
     if(
         u_getIntPropertyMaxValue(UCHAR_NFD_QUICK_CHECK)!=(int32_t)UNORM_YES ||
         u_getIntPropertyMaxValue(UCHAR_NFKD_QUICK_CHECK)!=(int32_t)UNORM_YES ||
         u_getIntPropertyMaxValue(UCHAR_NFC_QUICK_CHECK)!=(int32_t)UNORM_MAYBE ||
-        u_getIntPropertyMaxValue(UCHAR_NFKC_QUICK_CHECK)!=(int32_t)UNORM_MAYBE
+        u_getIntPropertyMaxValue(UCHAR_NFKC_QUICK_CHECK)!=(int32_t)UNORM_MAYBE ||
+        u_getIntPropertyMaxValue(UCHAR_LEAD_CANONICAL_COMBINING_CLASS)!=u_getIntPropertyMaxValue(UCHAR_CANONICAL_COMBINING_CLASS) ||
+        u_getIntPropertyMaxValue(UCHAR_TRAIL_CANONICAL_COMBINING_CLASS)!=u_getIntPropertyMaxValue(UCHAR_CANONICAL_COMBINING_CLASS)
     ) {
-        log_err("wrong result from one of the u_getIntPropertyMaxValue(UCHAR_NF*_QUICK_CHECK)\n");
+        log_err("wrong result from one of the u_getIntPropertyMaxValue(UCHAR_NF*_QUICK_CHECK) or UCHAR_*_CANONICAL_COMBINING_CLASS\n");
     }
 
     /*
@@ -1436,6 +1438,24 @@ TestQuickCheckPerCP() {
             log_err("u_getIntPropertyValue(NFKD)=%d != %d=unorm_quickCheck(NFKD) for U+%04x\n", qc1, qc2, c);
         }
 
+        length=unorm_normalize(s, length, UNORM_NFD, 0, nfd, LENGTHOF(nfd), &errorCode);
+        U16_GET(nfd, 0, 0, length, lead);
+        U16_GET(nfd, 0, length-1, length, trail);
+
+        lccc1=u_getIntPropertyValue(c, UCHAR_LEAD_CANONICAL_COMBINING_CLASS);
+        lccc2=u_getCombiningClass(lead);
+        tccc1=u_getIntPropertyValue(c, UCHAR_TRAIL_CANONICAL_COMBINING_CLASS);
+        tccc2=u_getCombiningClass(trail);
+
+        if(lccc1!=lccc2) {
+            log_err("u_getIntPropertyValue(lccc)=%d != %d=u_getCombiningClass(lead) for U+%04x\n",
+                    lccc1, lccc2, c);
+        }
+        if(tccc1!=tccc2) {
+            log_err("u_getIntPropertyValue(tccc)=%d != %d=u_getCombiningClass(trail) for U+%04x\n",
+                    tccc1, tccc2, c);
+        }
+
         /* skip some code points */
         c=(20*c)/19+1;
     }

icu4c/source/tools/genpname/SyntheticPropertyAliases.txt

@@ -1,5 +1,5 @@
 ######################################################################
-# Copyright (c) 2003, International Business Machines
+# Copyright (c) 2004, International Business Machines
 # Corporation and others.  All Rights Reserved.
 ######################################################################
 # Author: Alan Liu
@@ -21,6 +21,11 @@
 # Enumerated Non-Binary Properties
 # ================================================
 
+# lccc(c)=ccc(NFD(c)[0])
+# tccc(c)=ccc(NFD(c)[last])
+lccc; Lead_Canonical_Combining_Class
+tccc; Trail_Canonical_Combining_Class
+
 # ================================================
 # Bitmask Properties
 # ================================================

icu4c/source/tools/genpname/preparse.pl

@@ -688,11 +688,14 @@ sub merge_PropertyValueAliases {
     }
 
     # Merge the combining class values in manually
+    # Add the same values to the synthetic lccc and tccc properties
     die "Error: No ccc data"
         unless exists $va->{'ccc'};
     for my $ccc (keys %{$va->{'ccc'}}) {
         die "Error: Can't overwrite ccc $ccc"
             if (exists $h->{'ccc'}->{$ccc});
+        $h->{'lccc'}->{$ccc} =
+        $h->{'tccc'}->{$ccc} =
         $h->{'ccc'}->{$ccc} = $va->{'ccc'}->{$ccc};
     }
     delete $va->{'ccc'};