ICU-66108.tar.gz

[apple/icu.git] / icuSources / common / propname.cpp

diff --git a/icuSources/common/propname.cpp b/icuSources/common/propname.cpp
index 37a282344c068b1cd6a892b5660accb7d22a2a03..a12eb7d9134a2612350992efadf063e3a7093ae5 100644 (file)
--- a/icuSources/common/propname.cpp
+++ b/icuSources/common/propname.cpp
@@ -1,199 +1,328 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html

 /*
 **********************************************************************
 /*
 **********************************************************************

-* Copyright (c) 2002-2003, International Business Machines
+* Copyright (c) 2002-2014, International Business Machines

 * Corporation and others.  All Rights Reserved.
 **********************************************************************
 * Author: Alan Liu
 * Created: October 30 2002
 * Since: ICU 2.4
 * Corporation and others.  All Rights Reserved.
 **********************************************************************
 * Author: Alan Liu
 * Created: October 30 2002
 * Since: ICU 2.4

+* 2010nov19 Markus Scherer  Rewrite for formatVersion 2.

 **********************************************************************
 */
 #include "propname.h"
 #include "unicode/uchar.h"
 #include "unicode/udata.h"
 **********************************************************************
 */
 #include "propname.h"
 #include "unicode/uchar.h"
 #include "unicode/udata.h"

+#include "unicode/uscript.h"

 #include "umutex.h"
 #include "umutex.h"

+#include "cmemory.h"
+#include "cstring.h"
+#include "uarrsort.h"
+#include "uinvchar.h"

 
 

-U_NAMESPACE_BEGIN
+#define INCLUDED_FROM_PROPNAME_CPP
+#include "propname_data.h"

 
 

-//----------------------------------------------------------------------
-// PropertyAliases implementation
+U_CDECL_BEGIN

 
 

-const char*
-PropertyAliases::chooseNameInGroup(Offset offset,
-                                   UPropertyNameChoice choice) const {
-    int32_t c = choice;
-    if (!offset || c < 0) {
-        return NULL;
+/**
+ * Get the next non-ignorable ASCII character from a property name
+ * and lowercases it.
+ * @return ((advance count for the name)<<8)|character
+ */
+static inline int32_t
+getASCIIPropertyNameChar(const char *name) {
+    int32_t i;
+    char c;
+
+    /* Ignore delimiters '-', '_', and ASCII White_Space */
+    for(i=0;
+        (c=name[i++])==0x2d || c==0x5f ||
+        c==0x20 || (0x09<=c && c<=0x0d);
+    ) {}
+
+    if(c!=0) {
+        return (i<<8)|(uint8_t)uprv_asciitolower((char)c);
+    } else {
+        return i<<8;

     }
     }

-    const Offset* p = (const Offset*) getPointer(offset);
-    while (c-- > 0) {
-        if (*p++ < 0) return NULL;
+}
+
+/**
+ * Get the next non-ignorable EBCDIC character from a property name
+ * and lowercases it.
+ * @return ((advance count for the name)<<8)|character
+ */
+static inline int32_t
+getEBCDICPropertyNameChar(const char *name) {
+    int32_t i;
+    char c;
+
+    /* Ignore delimiters '-', '_', and EBCDIC White_Space */
+    for(i=0;
+        (c=name[i++])==0x60 || c==0x6d ||
+        c==0x40 || c==0x05 || c==0x15 || c==0x25 || c==0x0b || c==0x0c || c==0x0d;
+    ) {}
+
+    if(c!=0) {
+        return (i<<8)|(uint8_t)uprv_ebcdictolower((char)c);
+    } else {
+        return i<<8;

     }
     }

-    Offset a = *p;
-    if (a < 0) a = -a;
-    return (const char*) getPointerNull(a);

 }
 
 }
 

-const ValueMap*
-PropertyAliases::getValueMap(EnumValue prop) const {
-    NonContiguousEnumToOffset* e2o = (NonContiguousEnumToOffset*) getPointer(enumToValue_offset);
-    Offset a = e2o->getOffset(prop);
-    return (const ValueMap*) (a ? getPointerNull(a) : NULL);
+/**
+ * Unicode property names and property value names are compared "loosely".
+ *
+ * UCD.html 4.0.1 says:
+ *   For all property names, property value names, and for property values for
+ *   Enumerated, Binary, or Catalog properties, use the following
+ *   loose matching rule:
+ *
+ *   LM3. Ignore case, whitespace, underscore ('_'), and hyphens.
+ *
+ * This function does just that, for (char *) name strings.
+ * It is almost identical to ucnv_compareNames() but also ignores
+ * C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC).
+ *
+ * @internal
+ */
+
+U_CAPI int32_t U_EXPORT2
+uprv_compareASCIIPropertyNames(const char *name1, const char *name2) {
+    int32_t rc, r1, r2;
+
+    for(;;) {
+        r1=getASCIIPropertyNameChar(name1);
+        r2=getASCIIPropertyNameChar(name2);
+
+        /* If we reach the ends of both strings then they match */
+        if(((r1|r2)&0xff)==0) {
+            return 0;
+        }
+
+        /* Compare the lowercased characters */
+        if(r1!=r2) {
+            rc=(r1&0xff)-(r2&0xff);
+            if(rc!=0) {
+                return rc;
+            }
+        }
+
+        name1+=r1>>8;
+        name2+=r2>>8;
+    }

 }
 
 }
 

-inline const char*
-PropertyAliases::getPropertyName(EnumValue prop,
-                                 UPropertyNameChoice choice) const {
-    NonContiguousEnumToOffset* e2n = (NonContiguousEnumToOffset*) getPointer(enumToName_offset);
-    return chooseNameInGroup(e2n->getOffset(prop), choice);
+U_CAPI int32_t U_EXPORT2
+uprv_compareEBCDICPropertyNames(const char *name1, const char *name2) {
+    int32_t rc, r1, r2;
+
+    for(;;) {
+        r1=getEBCDICPropertyNameChar(name1);
+        r2=getEBCDICPropertyNameChar(name2);
+
+        /* If we reach the ends of both strings then they match */
+        if(((r1|r2)&0xff)==0) {
+            return 0;
+        }
+
+        /* Compare the lowercased characters */
+        if(r1!=r2) {
+            rc=(r1&0xff)-(r2&0xff);
+            if(rc!=0) {
+                return rc;
+            }
+        }
+
+        name1+=r1>>8;
+        name2+=r2>>8;
+    }

 }
 
 }
 

-inline EnumValue
-PropertyAliases::getPropertyEnum(const char* alias) const {
-    NameToEnum* n2e = (NameToEnum*) getPointer(nameToEnum_offset);
-    return n2e->getEnum(alias, *this);
+U_CDECL_END
+
+U_NAMESPACE_BEGIN
+
+int32_t PropNameData::findProperty(int32_t property) {
+    int32_t i=1;  // valueMaps index, initially after numRanges
+    for(int32_t numRanges=valueMaps[0]; numRanges>0; --numRanges) {
+        // Read and skip the start and limit of this range.
+        int32_t start=valueMaps[i];
+        int32_t limit=valueMaps[i+1];
+        i+=2;
+        if(property<start) {
+            break;
+        }
+        if(property<limit) {
+            return i+(property-start)*2;
+        }
+        i+=(limit-start)*2;  // Skip all entries for this range.
+    }
+    return 0;

 }
 
 }
 

-inline const char*
-PropertyAliases::getPropertyValueName(EnumValue prop,
-                                      EnumValue value,
-                                      UPropertyNameChoice choice) const {
-    const ValueMap* vm = getValueMap(prop);
-    if (!vm) return NULL;
-    Offset a;
-    if (vm->enumToName_offset) {
-        a = ((EnumToOffset*) getPointer(vm->enumToName_offset))->
-            getOffset(value);
+int32_t PropNameData::findPropertyValueNameGroup(int32_t valueMapIndex, int32_t value) {
+    if(valueMapIndex==0) {
+        return 0;  // The property does not have named values.
+    }
+    ++valueMapIndex;  // Skip the BytesTrie offset.
+    int32_t numRanges=valueMaps[valueMapIndex++];
+    if(numRanges<0x10) {
+        // Ranges of values.
+        for(; numRanges>0; --numRanges) {
+            // Read and skip the start and limit of this range.
+            int32_t start=valueMaps[valueMapIndex];
+            int32_t limit=valueMaps[valueMapIndex+1];
+            valueMapIndex+=2;
+            if(value<start) {
+                break;
+            }
+            if(value<limit) {
+                return valueMaps[valueMapIndex+value-start];
+            }
+            valueMapIndex+=limit-start;  // Skip all entries for this range.
+        }

     } else {
     } else {

-        a = ((NonContiguousEnumToOffset*) getPointer(vm->ncEnumToName_offset))->
-            getOffset(value);
+        // List of values.
+        int32_t valuesStart=valueMapIndex;
+        int32_t nameGroupOffsetsStart=valueMapIndex+numRanges-0x10;
+        do {
+            int32_t v=valueMaps[valueMapIndex];
+            if(value<v) {
+                break;
+            }
+            if(value==v) {
+                return valueMaps[nameGroupOffsetsStart+valueMapIndex-valuesStart];
+            }
+        } while(++valueMapIndex<nameGroupOffsetsStart);

     }
     }

-    return chooseNameInGroup(a, choice);
+    return 0;

 }
 
 }
 

-inline EnumValue
-PropertyAliases::getPropertyValueEnum(EnumValue prop,
-                                      const char* alias) const {
-    const ValueMap* vm = getValueMap(prop);
-    if (!vm) return UCHAR_INVALID_CODE;
-    NameToEnum* n2e = (NameToEnum*) getPointer(vm->nameToEnum_offset);
-    return n2e->getEnum(alias, *this);
+const char *PropNameData::getName(const char *nameGroup, int32_t nameIndex) {
+    int32_t numNames=*nameGroup++;
+    if(nameIndex<0 || numNames<=nameIndex) {
+        return NULL;
+    }
+    // Skip nameIndex names.
+    for(; nameIndex>0; --nameIndex) {
+        nameGroup=uprv_strchr(nameGroup, 0)+1;
+    }
+    if(*nameGroup==0) {
+        return NULL;  // no name (Property[Value]Aliases.txt has "n/a")
+    }
+    return nameGroup;

 }
 
 }
 

-U_NAMESPACE_END
+UBool PropNameData::containsName(BytesTrie &trie, const char *name) {
+    if(name==NULL) {
+        return FALSE;
+    }
+    UStringTrieResult result=USTRINGTRIE_NO_VALUE;
+    char c;
+    while((c=*name++)!=0) {
+        c=uprv_invCharToLowercaseAscii(c);
+        // Ignore delimiters '-', '_', and ASCII White_Space.
+        if(c==0x2d || c==0x5f || c==0x20 || (0x09<=c && c<=0x0d)) {
+            continue;
+        }
+        if(!USTRINGTRIE_HAS_NEXT(result)) {
+            return FALSE;
+        }
+        result=trie.next((uint8_t)c);
+    }
+    return USTRINGTRIE_HAS_VALUE(result);
+}

 
 

-//----------------------------------------------------------------------
-// UDataMemory structures
+const char *PropNameData::getPropertyName(int32_t property, int32_t nameChoice) {
+    int32_t valueMapIndex=findProperty(property);
+    if(valueMapIndex==0) {
+        return NULL;  // Not a known property.
+    }
+    return getName(nameGroups+valueMaps[valueMapIndex], nameChoice);
+}

 
 

-static const PropertyAliases* PNAME = NULL;
-static UDataMemory* UDATA = NULL;
+const char *PropNameData::getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice) {
+    int32_t valueMapIndex=findProperty(property);
+    if(valueMapIndex==0) {
+        return NULL;  // Not a known property.
+    }
+    int32_t nameGroupOffset=findPropertyValueNameGroup(valueMaps[valueMapIndex+1], value);
+    if(nameGroupOffset==0) {
+        return NULL;
+    }
+    return getName(nameGroups+nameGroupOffset, nameChoice);
+}

 
 

-//----------------------------------------------------------------------
-// UDataMemory loading/unloading
+int32_t PropNameData::getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias) {
+    BytesTrie trie(bytesTries+bytesTrieOffset);
+    if(containsName(trie, alias)) {
+        return trie.getValue();
+    } else {
+        return UCHAR_INVALID_CODE;
+    }
+}

 
 

-/**
- * udata callback to verify the zone data.
- */
-U_CDECL_BEGIN
-static UBool U_CALLCONV
-isAcceptable(void* /*context*/,
-             const char* /*type*/, const char* /*name*/,
-             const UDataInfo* info) {
-    return
-        info->size >= sizeof(UDataInfo) &&
-        info->isBigEndian == U_IS_BIG_ENDIAN &&
-        info->charsetFamily == U_CHARSET_FAMILY &&
-        info->dataFormat[0] == PNAME_SIG_0 &&
-        info->dataFormat[1] == PNAME_SIG_1 &&
-        info->dataFormat[2] == PNAME_SIG_2 &&
-        info->dataFormat[3] == PNAME_SIG_3 &&
-        info->formatVersion[0] == PNAME_FORMAT_VERSION;
-}
-
-UBool
-pname_cleanup() {
-    if (UDATA) {
-        udata_close(UDATA);
-        UDATA = NULL;
-    }
-    PNAME = NULL;
-    return TRUE;
+int32_t PropNameData::getPropertyEnum(const char *alias) {
+    return getPropertyOrValueEnum(0, alias);

 }
 }

-U_CDECL_END

 
 

-/**
- * Load the property names data.  Caller should check that data is
- * not loaded BEFORE calling this function.  Returns TRUE if the load
- * succeeds.
- */
-static UBool _load() {
-    UErrorCode ec = U_ZERO_ERROR;
-    UDataMemory* data =
-        udata_openChoice(0, PNAME_DATA_TYPE, PNAME_DATA_NAME,
-                         isAcceptable, 0, &ec);
-    if (U_SUCCESS(ec)) {
-        umtx_lock(NULL);
-        if (UDATA == NULL) {
-            UDATA = data;
-            PNAME = (const PropertyAliases*) udata_getMemory(UDATA);
-            data = NULL;
-        }
-        umtx_unlock(NULL);
+int32_t PropNameData::getPropertyValueEnum(int32_t property, const char *alias) {
+    int32_t valueMapIndex=findProperty(property);
+    if(valueMapIndex==0) {
+        return UCHAR_INVALID_CODE;  // Not a known property.

     }
     }

-    if (data) {
-        udata_close(data);
+    valueMapIndex=valueMaps[valueMapIndex+1];
+    if(valueMapIndex==0) {
+        return UCHAR_INVALID_CODE;  // The property does not have named values.

     }
     }

-    return PNAME!=NULL;
-}
-
-/**
- * Inline function that expands to code that does a lazy load of the
- * property names data.  If the data is already loaded, avoids an
- * unnecessary function call.  If the data is not loaded, call _load()
- * to load it, and return TRUE if the load succeeds.
- */
-static inline UBool load() {
-    umtx_lock(NULL);
-    UBool f = (PNAME!=NULL);
-    umtx_unlock(NULL);
-    return f || _load();
+    // valueMapIndex is the start of the property's valueMap,
+    // where the first word is the BytesTrie offset.
+    return getPropertyOrValueEnum(valueMaps[valueMapIndex], alias);

 }
 }

+U_NAMESPACE_END

 
 //----------------------------------------------------------------------
 // Public API implementation
 
 
 //----------------------------------------------------------------------
 // Public API implementation
 

-// The C API is just a thin wrapper.  Each function obtains a pointer
-// to the singleton PropertyAliases, and calls the appropriate method
-// on it.  If it cannot obtain a pointer, because valid data is not
-// available, then it returns NULL or UCHAR_INVALID_CODE.
-

 U_CAPI const char* U_EXPORT2
 u_getPropertyName(UProperty property,
                   UPropertyNameChoice nameChoice) {
 U_CAPI const char* U_EXPORT2
 u_getPropertyName(UProperty property,
                   UPropertyNameChoice nameChoice) {

-    return load() ? PNAME->getPropertyName(property, nameChoice)
-                  : NULL;
+    U_NAMESPACE_USE
+    return PropNameData::getPropertyName(property, nameChoice);

 }
 
 U_CAPI UProperty U_EXPORT2
 u_getPropertyEnum(const char* alias) {
 }
 
 U_CAPI UProperty U_EXPORT2
 u_getPropertyEnum(const char* alias) {

-    UProperty p = load() ? (UProperty) PNAME->getPropertyEnum(alias)
-                         : UCHAR_INVALID_CODE;
-    return p;
+    U_NAMESPACE_USE
+    return (UProperty)PropNameData::getPropertyEnum(alias);

 }
 
 U_CAPI const char* U_EXPORT2
 u_getPropertyValueName(UProperty property,
                        int32_t value,
                        UPropertyNameChoice nameChoice) {
 }
 
 U_CAPI const char* U_EXPORT2
 u_getPropertyValueName(UProperty property,
                        int32_t value,
                        UPropertyNameChoice nameChoice) {

-    return load() ? PNAME->getPropertyValueName(property, value, nameChoice)
-                  : NULL;
+    U_NAMESPACE_USE
+    return PropNameData::getPropertyValueName(property, value, nameChoice);

 }
 
 U_CAPI int32_t U_EXPORT2
 u_getPropertyValueEnum(UProperty property,
                        const char* alias) {
 }
 
 U_CAPI int32_t U_EXPORT2
 u_getPropertyValueEnum(UProperty property,
                        const char* alias) {

-    return load() ? PNAME->getPropertyValueEnum(property, alias)
-                  : UCHAR_INVALID_CODE;
+    U_NAMESPACE_USE
+    return PropNameData::getPropertyValueEnum(property, alias);

 }
 
 }
 

-//eof
+U_CAPI const char*  U_EXPORT2
+uscript_getName(UScriptCode scriptCode){
+    return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
+                                  U_LONG_PROPERTY_NAME);
+}
+
+U_CAPI const char*  U_EXPORT2
+uscript_getShortName(UScriptCode scriptCode){
+    return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
+                                  U_SHORT_PROPERTY_NAME);
+}