/src/icu/icu4c/source/common/propname.cpp

Source
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
**********************************************************************
* Copyright (c) 2002-2014, International Business Machines
* 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 "unicode/uscript.h"
#include "umutex.h"
#include "cmemory.h"
#include "cstring.h"
#include "uarrsort.h"
#include "uinvchar.h"

#define INCLUDED_FROM_PROPNAME_CPP
#include "propname_data.h"

U_CDECL_BEGIN

/**
 * 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) | static_cast<uint8_t>(uprv_asciitolower(c));
    } else {
        return i<<8;
    }
}

/**
 * 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) | static_cast<uint8_t>(uprv_ebcdictolower(c));
    } else {
        return i<<8;
    }
}

/**
 * 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;
    }
}

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;
    }
}

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;
}

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 {
        // 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 0;
}

const char *PropNameData::getName(const char *nameGroup, int32_t nameIndex) {
    int32_t numNames=*nameGroup++;
    if(nameIndex<0 || numNames<=nameIndex) {
        return nullptr;
    }
    // Skip nameIndex names.
    for(; nameIndex>0; --nameIndex) {
        nameGroup=uprv_strchr(nameGroup, 0)+1;
    }
    if(*nameGroup==0) {
        return nullptr;  // no name (Property[Value]Aliases.txt has "n/a")
    }
    return nameGroup;
}

UBool PropNameData::containsName(BytesTrie &trie, const char *name) {
    if(name==nullptr) {
        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(static_cast<uint8_t>(c));
    }
    return USTRINGTRIE_HAS_VALUE(result);
}

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

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

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;
    }
}

int32_t PropNameData::getPropertyEnum(const char *alias) {
    return getPropertyOrValueEnum(0, alias);
}

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

U_CAPI const char* U_EXPORT2
u_getPropertyName(UProperty property,
                  UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED {
    // The nameChoice is really an integer with a couple of named constants.
    // Unicode allows for names other than short and long ones.
    // If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,...
    U_NAMESPACE_USE
    return PropNameData::getPropertyName(property, nameChoice);
}

U_CAPI UProperty U_EXPORT2
u_getPropertyEnum(const char* alias) {
    U_NAMESPACE_USE
    return (UProperty)PropNameData::getPropertyEnum(alias);
}

U_CAPI const char* U_EXPORT2
u_getPropertyValueName(UProperty property,
                       int32_t value,
                       UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED {
    // The nameChoice is really an integer with a couple of named constants.
    // Unicode allows for names other than short and long ones.
    // If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,...
    U_NAMESPACE_USE
    return PropNameData::getPropertyValueName(property, value, nameChoice);
}

U_CAPI int32_t U_EXPORT2
u_getPropertyValueEnum(UProperty property,
                       const char* alias) {
    U_NAMESPACE_USE
    return PropNameData::getPropertyValueEnum(property, alias);
}

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);
}

Coverage Report

Created: 2025-12-07 06:36

Line	Count	Source
1		// © 2016 and later: Unicode, Inc. and others.
2		// License & terms of use: http://www.unicode.org/copyright.html
3		/*
4		**********************************************************************
5		* Copyright (c) 2002-2014, International Business Machines
6		* Corporation and others. All Rights Reserved.
7		**********************************************************************
8		* Author: Alan Liu
9		* Created: October 30 2002
10		* Since: ICU 2.4
11		* 2010nov19 Markus Scherer Rewrite for formatVersion 2.
12		**********************************************************************
13		*/
14		#include "propname.h"
15		#include "unicode/uchar.h"
16		#include "unicode/udata.h"
17		#include "unicode/uscript.h"
18		#include "umutex.h"
19		#include "cmemory.h"
20		#include "cstring.h"
21		#include "uarrsort.h"
22		#include "uinvchar.h"
23
24		#define INCLUDED_FROM_PROPNAME_CPP
25		#include "propname_data.h"
26
27		U_CDECL_BEGIN
28
29		/**
30		* Get the next non-ignorable ASCII character from a property name
31		* and lowercases it.
32		* @return ((advance count for the name)<<8)\|character
33		*/
34		static inline int32_t
35	184k	getASCIIPropertyNameChar(const char *name) {
36	184k	int32_t i;
37	184k	char c;
38
39		/* Ignore delimiters '-', '_', and ASCII White_Space */
40	184k	for(i=0;
41	228k	(c=name[i++])==0x2d \|\| c==0x5f \|\|
42	218k	c==0x20 \|\| (0x09<=c && c<=0x0d);
43	184k	) {}
44
45	184k	if(c!=0) {
46	157k	return (i << 8) \| static_cast<uint8_t>(uprv_asciitolower(c));
47	157k	} else {
48	27.2k	return i<<8;
49	27.2k	}
50	184k	}
51
52		/**
53		* Get the next non-ignorable EBCDIC character from a property name
54		* and lowercases it.
55		* @return ((advance count for the name)<<8)\|character
56		*/
57		static inline int32_t
58	0	getEBCDICPropertyNameChar(const char *name) {
59	0	int32_t i;
60	0	char c;
61
62		/* Ignore delimiters '-', '_', and EBCDIC White_Space */
63	0	for(i=0;
64	0	(c=name[i++])==0x60 \|\| c==0x6d \|\|
65	0	c==0x40 \|\| c==0x05 \|\| c==0x15 \|\| c==0x25 \|\| c==0x0b \|\| c==0x0c \|\| c==0x0d;
66	0	) {}
67
68	0	if(c!=0) {
69	0	return (i << 8) \| static_cast<uint8_t>(uprv_ebcdictolower(c));
70	0	} else {
71	0	return i<<8;
72	0	}
73	0	}
74
75		/**
76		* Unicode property names and property value names are compared "loosely".
77		*
78		* UCD.html 4.0.1 says:
79		* For all property names, property value names, and for property values for
80		* Enumerated, Binary, or Catalog properties, use the following
81		* loose matching rule:
82		*
83		* LM3. Ignore case, whitespace, underscore ('_'), and hyphens.
84		*
85		* This function does just that, for (char *) name strings.
86		* It is almost identical to ucnv_compareNames() but also ignores
87		* C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC).
88		*
89		* @internal
90		*/
91
92		U_CAPI int32_t U_EXPORT2
93	45.3k	uprv_compareASCIIPropertyNames(const char name1, const char name2) {
94	45.3k	int32_t rc, r1, r2;
95
96	92.3k	for(;;) {
97	92.3k	r1=getASCIIPropertyNameChar(name1);
98	92.3k	r2=getASCIIPropertyNameChar(name2);
99
100		/* If we reach the ends of both strings then they match */
101	92.3k	if(((r1\|r2)&0xff)==0) {
102	13.0k	return 0;
103	13.0k	}
104
105		/* Compare the lowercased characters */
106	79.3k	if(r1!=r2) {
107	35.9k	rc=(r1&0xff)-(r2&0xff);
108	35.9k	if(rc!=0) {
109	32.2k	return rc;
110	32.2k	}
111	35.9k	}
112
113	47.0k	name1+=r1>>8;
114	47.0k	name2+=r2>>8;
115	47.0k	}
116	45.3k	}
117
118		U_CAPI int32_t U_EXPORT2
119	0	uprv_compareEBCDICPropertyNames(const char name1, const char name2) {
120	0	int32_t rc, r1, r2;
121
122	0	for(;;) {
123	0	r1=getEBCDICPropertyNameChar(name1);
124	0	r2=getEBCDICPropertyNameChar(name2);
125
126		/* If we reach the ends of both strings then they match */
127	0	if(((r1\|r2)&0xff)==0) {
128	0	return 0;
129	0	}
130
131		/* Compare the lowercased characters */
132	0	if(r1!=r2) {
133	0	rc=(r1&0xff)-(r2&0xff);
134	0	if(rc!=0) {
135	0	return rc;
136	0	}
137	0	}
138
139	0	name1+=r1>>8;
140	0	name2+=r2>>8;
141	0	}
142	0	}
143
144		U_CDECL_END
145
146		U_NAMESPACE_BEGIN
147
148	571k	int32_t PropNameData::findProperty(int32_t property) {
149	571k	int32_t i=1; // valueMaps index, initially after numRanges
150	1.33M	for(int32_t numRanges=valueMaps[0]; numRanges>0; --numRanges) {
151		// Read and skip the start and limit of this range.
152	1.33M	int32_t start=valueMaps[i];
153	1.33M	int32_t limit=valueMaps[i+1];
154	1.33M	i+=2;
155	1.33M	if(property<start) {
156	0	break;
157	0	}
158	1.33M	if(property<limit) {
159	571k	return i+(property-start)*2;
160	571k	}
161	761k	i+=(limit-start)*2; // Skip all entries for this range.
162	761k	}
163	0	return 0;
164	571k	}
165
166	252k	int32_t PropNameData::findPropertyValueNameGroup(int32_t valueMapIndex, int32_t value) {
167	252k	if(valueMapIndex==0) {
168	0	return 0; // The property does not have named values.
169	0	}
170	252k	++valueMapIndex; // Skip the BytesTrie offset.
171	252k	int32_t numRanges=valueMaps[valueMapIndex++];
172	252k	if(numRanges<0x10) {
173		// Ranges of values.
174	252k	for(; numRanges>0; --numRanges) {
175		// Read and skip the start and limit of this range.
176	252k	int32_t start=valueMaps[valueMapIndex];
177	252k	int32_t limit=valueMaps[valueMapIndex+1];
178	252k	valueMapIndex+=2;
179	252k	if(value<start) {
180	0	break;
181	0	}
182	252k	if(value<limit) {
183	252k	return valueMaps[valueMapIndex+value-start];
184	252k	}
185	0	valueMapIndex+=limit-start; // Skip all entries for this range.
186	0	}
187	252k	} else {
188		// List of values.
189	0	int32_t valuesStart=valueMapIndex;
190	0	int32_t nameGroupOffsetsStart=valueMapIndex+numRanges-0x10;
191	0	do {
192	0	int32_t v=valueMaps[valueMapIndex];
193	0	if(value<v) {
194	0	break;
195	0	}
196	0	if(value==v) {
197	0	return valueMaps[nameGroupOffsetsStart+valueMapIndex-valuesStart];
198	0	}
199	0	} while(++valueMapIndex<nameGroupOffsetsStart);
200	0	}
201	0	return 0;
202	252k	}
203
204	252k	const char PropNameData::getName(const char nameGroup, int32_t nameIndex) {
205	252k	int32_t numNames=*nameGroup++;
206	252k	if(nameIndex<0 \|\| numNames<=nameIndex) {
207	0	return nullptr;
208	0	}
209		// Skip nameIndex names.
210	252k	for(; nameIndex>0; --nameIndex) {
211	0	nameGroup=uprv_strchr(nameGroup, 0)+1;
212	0	}
213	252k	if(*nameGroup==0) {
214	0	return nullptr; // no name (Property[Value]Aliases.txt has "n/a")
215	0	}
216	252k	return nameGroup;
217	252k	}
218
219	466k	UBool PropNameData::containsName(BytesTrie &trie, const char *name) {
220	466k	if(name==nullptr) {
221	0	return false;
222	0	}
223	466k	UStringTrieResult result=USTRINGTRIE_NO_VALUE;
224	466k	char c;
225	1.54M	while((c=*name++)!=0) {
226	1.15M	c=uprv_invCharToLowercaseAscii(c);
227		// Ignore delimiters '-', '_', and ASCII White_Space.
228	1.15M	if(c==0x2d \|\| c==0x5f \|\| c==0x20 \|\| (0x09<=c && c<=0x0d)) {
229	99.2k	continue;
230	99.2k	}
231	1.05M	if(!USTRINGTRIE_HAS_NEXT(result)) {
232	80.8k	return false;
233	80.8k	}
234	974k	result = trie.next(static_cast<uint8_t>(c));
235	974k	}
236	385k	return USTRINGTRIE_HAS_VALUE(result);
237	466k	}
238
239	0	const char *PropNameData::getPropertyName(int32_t property, int32_t nameChoice) {
240	0	int32_t valueMapIndex=findProperty(property);
241	0	if(valueMapIndex==0) {
242	0	return nullptr; // Not a known property.
243	0	}
244	0	return getName(nameGroups+valueMaps[valueMapIndex], nameChoice);
245	0	}
246
247	252k	const char *PropNameData::getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice) {
248	252k	int32_t valueMapIndex=findProperty(property);
249	252k	if(valueMapIndex==0) {
250	0	return nullptr; // Not a known property.
251	0	}
252	252k	int32_t nameGroupOffset=findPropertyValueNameGroup(valueMaps[valueMapIndex+1], value);
253	252k	if(nameGroupOffset==0) {
254	0	return nullptr;
255	0	}
256	252k	return getName(nameGroups+nameGroupOffset, nameChoice);
257	252k	}
258
259	466k	int32_t PropNameData::getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias) {
260	466k	BytesTrie trie(bytesTries+bytesTrieOffset);
261	466k	if(containsName(trie, alias)) {
262	321k	return trie.getValue();
263	321k	} else {
264	145k	return UCHAR_INVALID_CODE;
265	145k	}
266	466k	}
267
268	147k	int32_t PropNameData::getPropertyEnum(const char *alias) {
269	147k	return getPropertyOrValueEnum(0, alias);
270	147k	}
271
272	319k	int32_t PropNameData::getPropertyValueEnum(int32_t property, const char *alias) {
273	319k	int32_t valueMapIndex=findProperty(property);
274	319k	if(valueMapIndex==0) {
275	0	return UCHAR_INVALID_CODE; // Not a known property.
276	0	}
277	319k	valueMapIndex=valueMaps[valueMapIndex+1];
278	319k	if(valueMapIndex==0) {
279	0	return UCHAR_INVALID_CODE; // The property does not have named values.
280	0	}
281		// valueMapIndex is the start of the property's valueMap,
282		// where the first word is the BytesTrie offset.
283	319k	return getPropertyOrValueEnum(valueMaps[valueMapIndex], alias);
284	319k	}
285		U_NAMESPACE_END
286
287		//----------------------------------------------------------------------
288		// Public API implementation
289
290		U_CAPI const char* U_EXPORT2
291		u_getPropertyName(UProperty property,
292	0	UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED {
293		// The nameChoice is really an integer with a couple of named constants.
294		// Unicode allows for names other than short and long ones.
295		// If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,...
296	0	U_NAMESPACE_USE
297	0	return PropNameData::getPropertyName(property, nameChoice);
298	0	}
299
300		U_CAPI UProperty U_EXPORT2
301	147k	u_getPropertyEnum(const char* alias) {
302	147k	U_NAMESPACE_USE
303	147k	return (UProperty)PropNameData::getPropertyEnum(alias);
304	147k	}
305
306		U_CAPI const char* U_EXPORT2
307		u_getPropertyValueName(UProperty property,
308		int32_t value,
309	252k	UPropertyNameChoice nameChoice) UPRV_NO_SANITIZE_UNDEFINED {
310		// The nameChoice is really an integer with a couple of named constants.
311		// Unicode allows for names other than short and long ones.
312		// If present, these will be returned for U_LONG_PROPERTY_NAME + i, where i=1, 2,...
313	252k	U_NAMESPACE_USE
314	252k	return PropNameData::getPropertyValueName(property, value, nameChoice);
315	252k	}
316
317		U_CAPI int32_t U_EXPORT2
318		u_getPropertyValueEnum(UProperty property,
319	319k	const char* alias) {
320	319k	U_NAMESPACE_USE
321	319k	return PropNameData::getPropertyValueEnum(property, alias);
322	319k	}
323
324		U_CAPI const char* U_EXPORT2
325	0	uscript_getName(UScriptCode scriptCode){
326	0	return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
327	0	U_LONG_PROPERTY_NAME);
328	0	}
329
330		U_CAPI const char* U_EXPORT2
331	252k	uscript_getShortName(UScriptCode scriptCode){
332	252k	return u_getPropertyValueName(UCHAR_SCRIPT, scriptCode,
333	252k	U_SHORT_PROPERTY_NAME);
334	252k	}