[apple/icu.git] / icuSources / test / cintltst / sprpdata.c

/*
 *******************************************************************************
 *
 *   Copyright (C) 2003-2012, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
 *   file name:  spreptst.c
 *   encoding:   US-ASCII
 *   tab size:   8 (not used)
 *   indentation:4
 *
 *   created on: 2003jul11
 *   created by: Ram Viswanadha
 */
#define USPREP_TYPE_NAMES_ARRAY

#include "unicode/utypes.h"

#if !UCONFIG_NO_IDNA

#include "unicode/ustring.h"
#include "unicode/putil.h"
#include "cintltst.h"
#include "unicode/usprep.h"
#include "unicode/utf16.h"
#include "sprpimpl.h"
#include "uparse.h"
#include "cmemory.h"
#include "ustr_imp.h"
#include "cstring.h"

static void
parseMappings(const char *filename, UStringPrepProfile* data, UBool reportError, UErrorCode *pErrorCode);

static void
compareMapping(UStringPrepProfile* data, uint32_t codepoint, uint32_t* mapping, int32_t mapLength,
               UStringPrepType option);

static void
compareFlagsForRange(UStringPrepProfile* data, uint32_t start, uint32_t end,UStringPrepType option);

void
doStringPrepTest(const char* binFileName, const char* txtFileName, int32_t options, UErrorCode* errorCode);

static void U_CALLCONV
strprepProfileLineFn(void *context,
              char *fields[][2], int32_t fieldCount,
              UErrorCode *pErrorCode) {
    uint32_t mapping[40];
    char *end, *map;
    uint32_t code;
    int32_t length;
    UStringPrepProfile* data = (UStringPrepProfile*) context;
    const char* typeName;
    uint32_t rangeStart=0,rangeEnd =0;

    typeName = fields[2][0];
    map = fields[1][0];

    if(strstr(typeName, usprepTypeNames[USPREP_UNASSIGNED])!=NULL){

        u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);

        /* store the range */
        compareFlagsForRange(data, rangeStart,rangeEnd,USPREP_UNASSIGNED);

    }else if(strstr(typeName, usprepTypeNames[USPREP_PROHIBITED])!=NULL){

        u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);

        /* store the range */
        compareFlagsForRange(data, rangeStart,rangeEnd,USPREP_PROHIBITED);

    }else if(strstr(typeName, usprepTypeNames[USPREP_MAP])!=NULL){
        /* get the character code, field 0 */
        code=(uint32_t)uprv_strtoul(fields[0][0], &end, 16);

        /* parse the mapping string */
        length=u_parseCodePoints(map, mapping, sizeof(mapping)/4, pErrorCode);

        /* compare the mapping */
        compareMapping(data, code,mapping, length,USPREP_MAP);
    }else{
        *pErrorCode = U_INVALID_FORMAT_ERROR;
    }

}


static void
parseMappings(const char *filename, UStringPrepProfile* data, UBool reportError, UErrorCode *pErrorCode) {
    char *fields[3][2];

    if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
        return;
    }

    u_parseDelimitedFile(filename, ';', fields, 3, strprepProfileLineFn, (void*)data, pErrorCode);

    /*fprintf(stdout,"Number of code points that have mappings with length >1 : %i\n",len);*/

    if(U_FAILURE(*pErrorCode) && (reportError || *pErrorCode!=U_FILE_ACCESS_ERROR)) {
        log_err( "testidn error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
    }
}


static UStringPrepType
getValues(uint32_t result, int32_t* value, UBool* isIndex){

    UStringPrepType type;
    if(result == 0){
        /*
         * Initial value stored in the mapping table
         * just return USPREP_TYPE_LIMIT .. so that
         * the source codepoint is copied to the destination
         */
        type = USPREP_TYPE_LIMIT;
    }else if(result >= _SPREP_TYPE_THRESHOLD){
        type = (UStringPrepType) (result - _SPREP_TYPE_THRESHOLD);
    }else{
        /* get the type */
        type = USPREP_MAP;
        /* ascertain if the value is index or delta */
        if(result & 0x02){
            *isIndex = TRUE;
            *value = result  >> 2;

        }else{
            *isIndex = FALSE;
            *value = (int16_t)result;
            *value =  (*value >> 2);

        }
        if((result>>2) == _SPREP_MAX_INDEX_VALUE){
            type = USPREP_DELETE;
            isIndex =FALSE;
            value = 0;
        }
    }
    return type;
}

static void
compareMapping(UStringPrepProfile* data, uint32_t codepoint, uint32_t* mapping,int32_t mapLength,
               UStringPrepType type){
    uint32_t result = 0;
    int32_t length=0;
    UBool isIndex = FALSE;
    UStringPrepType retType;
    int32_t value=0, idx=0, delta=0;
    int32_t* indexes = data->indexes;
    UTrie trie = data->sprepTrie;
    const uint16_t* mappingData = data->mappingData;
    int32_t realLength =0;
    int32_t j=0;
    int8_t i=0;

    UTRIE_GET16(&trie, codepoint, result);
    retType = getValues(result,&value,&isIndex);


    if(type != retType && retType != USPREP_DELETE){

        log_err( "Did not get the assigned type for codepoint 0x%08X. Expected: %i Got: %i\n",codepoint, USPREP_MAP, type);

    }

    if(isIndex){
        idx = value;
        if(idx >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
                 idx < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
            length = 1;
        }else if(idx >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
                 idx < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
            length = 2;
        }else if(idx >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
                 idx < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
            length = 3;
        }else{
            length = mappingData[idx++];
        }
    }else{
        delta = value;
        length = (retType == USPREP_DELETE)? 0 :  1;
    }

    /* figure out the real length */
    for(j=0; j<mapLength; j++){
        if(mapping[j] > 0xFFFF){
            realLength +=2;
        }else{
            realLength++;
        }
    }

    if(realLength != length){
        log_err( "Did not get the expected length. Expected: %i Got: %i\n", mapLength, length);
    }

    if(isIndex){
        for(i =0; i< mapLength; i++){
            if(mapping[i] <= 0xFFFF){
                if(mappingData[idx+i] != (uint16_t)mapping[i]){
                    log_err("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[i], mappingData[idx+i]);
                }
            }else{
                UChar lead  = U16_LEAD(mapping[i]);
                UChar trail = U16_TRAIL(mapping[i]);
                if(mappingData[idx+i] != lead ||
                    mappingData[idx+i+1] != trail){
                    log_err( "Did not get the expected result. Expected: 0x%04X 0x%04X  Got: 0x%04X 0x%04X\n", lead, trail, mappingData[idx+i], mappingData[idx+i+1]);
                }
            }
        }
    }else{
        if(retType!=USPREP_DELETE && (codepoint-delta) != (uint16_t)mapping[0]){
           log_err("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[0],(codepoint-delta));
        }
    }

}

static void
compareFlagsForRange(UStringPrepProfile* data,
                     uint32_t start, uint32_t end,
                     UStringPrepType type){

    uint32_t result =0 ;
    UStringPrepType retType;
    UBool isIndex=FALSE;
    int32_t value=0;
    UTrie trie = data->sprepTrie;
/*
    // supplementary code point
    UChar __lead16=U16_LEAD(0x2323E);
    int32_t __offset;

    // get data for lead surrogate
    (result)=_UTRIE_GET_RAW((&idnTrie), index, 0, (__lead16));
    __offset=(&idnTrie)->getFoldingOffset(result);

    // get the real data from the folded lead/trail units
    if(__offset>0) {
        (result)=_UTRIE_GET_RAW((&idnTrie), index, __offset, (0x2323E)&0x3ff);
    } else {
        (result)=(uint32_t)((&idnTrie)->initialValue);
    }

    UTRIE_GET16(&idnTrie,0x2323E, result);
*/
    while(start < end+1){
        UTRIE_GET16(&trie,start, result);
        retType = getValues(result, &value, &isIndex);
        if(result > _SPREP_TYPE_THRESHOLD){
            if(retType != type){
                log_err( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
            }
        }else{
            if(type == USPREP_PROHIBITED && ((result & 0x01) != 0x01)){
                log_err( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
            }
        }

        start++;
    }

}

void
doStringPrepTest(const char* binFileName, const char* txtFileName, int32_t options, UErrorCode* errorCode){

    const char *testdatapath = loadTestData(errorCode);
    const char *srcdatapath = NULL;
    const char *relativepath = NULL;
    char *filename = NULL;
    UStringPrepProfile* profile = NULL;

#ifdef U_TOPSRCDIR
    srcdatapath = U_TOPSRCDIR;
    relativepath = U_FILE_SEP_STRING"test"U_FILE_SEP_STRING"testdata"U_FILE_SEP_STRING;
#else
    srcdatapath = ctest_dataOutDir();
    relativepath = ".."U_FILE_SEP_STRING".."U_FILE_SEP_STRING"test"U_FILE_SEP_STRING"testdata"U_FILE_SEP_STRING;
#endif

    profile = usprep_open(testdatapath, binFileName, errorCode);

    if(*errorCode == U_FILE_ACCESS_ERROR) {
        log_data_err("Failed to load %s data file. Error: %s \n", binFileName, u_errorName(*errorCode));
        return;
    } else if(U_FAILURE(*errorCode)){
        log_err("Failed to load %s data file. Error: %s \n", binFileName, u_errorName(*errorCode));
        return;
    }
    filename = (char*) malloc(strlen(srcdatapath)+strlen(relativepath)+strlen(txtFileName)+10 );
    /* open and load the txt file */
    strcpy(filename,srcdatapath);
    strcat(filename,relativepath);
    strcat(filename,txtFileName);

    parseMappings(filename,profile, TRUE,errorCode);

    free(filename);
}
#endif
/*
 * Hey, Emacs, please set the following:
 *
 * Local Variables:
 * indent-tabs-mode: nil
 * End:
 *
 */
Commit	Line	Data
374ca955 A	1	/*
	2	*******************************************************************************
	3	*
51004dcb	4	* Copyright (C) 2003-2012, International Business Machines
374ca955 A	5	* Corporation and others. All Rights Reserved.
	6	*
	7	*******************************************************************************
	8	* file name: spreptst.c
	9	* encoding: US-ASCII
	10	* tab size: 8 (not used)
	11	* indentation:4
	12	*
	13	* created on: 2003jul11
	14	* created by: Ram Viswanadha
	15	*/
	16	#define USPREP_TYPE_NAMES_ARRAY
	17
	18	#include "unicode/utypes.h"
	19
	20	#if !UCONFIG_NO_IDNA
	21
	22	#include "unicode/ustring.h"
	23	#include "unicode/putil.h"
	24	#include "cintltst.h"
	25	#include "unicode/usprep.h"
4388f060	26	#include "unicode/utf16.h"
374ca955 A	27	#include "sprpimpl.h"
	28	#include "uparse.h"
	29	#include "cmemory.h"
	30	#include "ustr_imp.h"
	31	#include "cstring.h"
	32
	33	static void
	34	parseMappings(const char filename, UStringPrepProfile data, UBool reportError, UErrorCode *pErrorCode);
	35
	36	static void
	37	compareMapping(UStringPrepProfile* data, uint32_t codepoint, uint32_t* mapping, int32_t mapLength,
	38	UStringPrepType option);
	39
	40	static void
	41	compareFlagsForRange(UStringPrepProfile* data, uint32_t start, uint32_t end,UStringPrepType option);
	42
	43	void
	44	doStringPrepTest(const char* binFileName, const char* txtFileName, int32_t options, UErrorCode* errorCode);
	45
	46	static void U_CALLCONV
	47	strprepProfileLineFn(void *context,
	48	char *fields[][2], int32_t fieldCount,
	49	UErrorCode *pErrorCode) {
	50	uint32_t mapping[40];
	51	char end, map;
	52	uint32_t code;
	53	int32_t length;
	54	UStringPrepProfile* data = (UStringPrepProfile*) context;
	55	const char* typeName;
	56	uint32_t rangeStart=0,rangeEnd =0;
	57
	58	typeName = fields[2][0];
	59	map = fields[1][0];
	60
	61	if(strstr(typeName, usprepTypeNames[USPREP_UNASSIGNED])!=NULL){
	62
	63	u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);
	64
	65	/* store the range */
	66	compareFlagsForRange(data, rangeStart,rangeEnd,USPREP_UNASSIGNED);
	67
	68	}else if(strstr(typeName, usprepTypeNames[USPREP_PROHIBITED])!=NULL){
	69
	70	u_parseCodePointRange(fields[0][0], &rangeStart,&rangeEnd, pErrorCode);
	71
	72	/* store the range */
	73	compareFlagsForRange(data, rangeStart,rangeEnd,USPREP_PROHIBITED);
	74
	75	}else if(strstr(typeName, usprepTypeNames[USPREP_MAP])!=NULL){
	76	/* get the character code, field 0 */
	77	code=(uint32_t)uprv_strtoul(fields[0][0], &end, 16);
	78
	79	/* parse the mapping string */
	80	length=u_parseCodePoints(map, mapping, sizeof(mapping)/4, pErrorCode);
	81
	82	/* compare the mapping */
	83	compareMapping(data, code,mapping, length,USPREP_MAP);
	84	}else{
	85	*pErrorCode = U_INVALID_FORMAT_ERROR;
	86	}
	87
	88	}
	89
	90
91
92	static void
93	parseMappings(const char filename, UStringPrepProfile data, UBool reportError, UErrorCode *pErrorCode) {
94	char *fields[3][2];
95
96	if(pErrorCode==NULL \|\| U_FAILURE(*pErrorCode)) {
97	return;
98	}
99
100	u_parseDelimitedFile(filename, ';', fields, 3, strprepProfileLineFn, (void*)data, pErrorCode);
101
102	/fprintf(stdout,"Number of code points that have mappings with length >1 : %i\n",len);/
103
104	if(U_FAILURE(pErrorCode) && (reportError \|\| pErrorCode!=U_FILE_ACCESS_ERROR)) {
105	log_err( "testidn error: u_parseDelimitedFile(\"%s\") failed - %s\n", filename, u_errorName(*pErrorCode));
106	}
107	}
108
109
110	static UStringPrepType
111	getValues(uint32_t result, int32_t* value, UBool* isIndex){
112
113	UStringPrepType type;
114	if(result == 0){
115	/*
116	* Initial value stored in the mapping table
117	* just return USPREP_TYPE_LIMIT .. so that
118	* the source codepoint is copied to the destination
119	*/
120	type = USPREP_TYPE_LIMIT;
121	}else if(result >= _SPREP_TYPE_THRESHOLD){
122	type = (UStringPrepType) (result - _SPREP_TYPE_THRESHOLD);
123	}else{
124	/* get the type */
125	type = USPREP_MAP;
126	/* ascertain if the value is index or delta */
127	if(result & 0x02){
128	*isIndex = TRUE;
129	*value = result >> 2;
130
131	}else{
132	*isIndex = FALSE;
133	*value = (int16_t)result;
134	value = (value >> 2);
135
136	}
137	if((result>>2) == _SPREP_MAX_INDEX_VALUE){
138	type = USPREP_DELETE;
139	isIndex =FALSE;
140	value = 0;
141	}
142	}
143	return type;
144	}
145
146	static void
147	compareMapping(UStringPrepProfile* data, uint32_t codepoint, uint32_t* mapping,int32_t mapLength,
148	UStringPrepType type){
149	uint32_t result = 0;
150	int32_t length=0;
151	UBool isIndex = FALSE;
152	UStringPrepType retType;
51004dcb	153	int32_t value=0, idx=0, delta=0;
374ca955 A	154	int32_t* indexes = data->indexes;
	155	UTrie trie = data->sprepTrie;
	156	const uint16_t* mappingData = data->mappingData;
	157	int32_t realLength =0;
	158	int32_t j=0;
	159	int8_t i=0;
	160
	161	UTRIE_GET16(&trie, codepoint, result);
	162	retType = getValues(result,&value,&isIndex);
	163
	164
	165	if(type != retType && retType != USPREP_DELETE){
	166
	167	log_err( "Did not get the assigned type for codepoint 0x%08X. Expected: %i Got: %i\n",codepoint, USPREP_MAP, type);
	168
	169	}
	170
	171	if(isIndex){
51004dcb A	172	idx = value;
	173	if(idx >= indexes[_SPREP_ONE_UCHAR_MAPPING_INDEX_START] &&
	174	idx < indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START]){
374ca955	175	length = 1;
51004dcb A	176	}else if(idx >= indexes[_SPREP_TWO_UCHARS_MAPPING_INDEX_START] &&
51004dcb A	177	idx < indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START]){
374ca955	178	length = 2;
51004dcb A	179	}else if(idx >= indexes[_SPREP_THREE_UCHARS_MAPPING_INDEX_START] &&
51004dcb A	180	idx < indexes[_SPREP_FOUR_UCHARS_MAPPING_INDEX_START]){
374ca955 A	181	length = 3;
374ca955 A	182	}else{
51004dcb	183	length = mappingData[idx++];
374ca955 A	184	}
	185	}else{
	186	delta = value;
	187	length = (retType == USPREP_DELETE)? 0 : 1;
	188	}
	189
	190	/* figure out the real length */
	191	for(j=0; j<mapLength; j++){
	192	if(mapping[j] > 0xFFFF){
	193	realLength +=2;
	194	}else{
	195	realLength++;
	196	}
	197	}
	198
	199	if(realLength != length){
	200	log_err( "Did not get the expected length. Expected: %i Got: %i\n", mapLength, length);
	201	}
	202
	203	if(isIndex){
	204	for(i =0; i< mapLength; i++){
	205	if(mapping[i] <= 0xFFFF){
51004dcb A	206	if(mappingData[idx+i] != (uint16_t)mapping[i]){
51004dcb A	207	log_err("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[i], mappingData[idx+i]);
374ca955 A	208	}
374ca955 A	209	}else{
4388f060 A	210	UChar lead = U16_LEAD(mapping[i]);
4388f060 A	211	UChar trail = U16_TRAIL(mapping[i]);
51004dcb A	212	if(mappingData[idx+i] != lead \|\|
	213	mappingData[idx+i+1] != trail){
	214	log_err( "Did not get the expected result. Expected: 0x%04X 0x%04X Got: 0x%04X 0x%04X\n", lead, trail, mappingData[idx+i], mappingData[idx+i+1]);
374ca955 A	215	}
	216	}
	217	}
	218	}else{
	219	if(retType!=USPREP_DELETE && (codepoint-delta) != (uint16_t)mapping[0]){
	220	log_err("Did not get the expected result. Expected: 0x%04X Got: 0x%04X \n", mapping[0],(codepoint-delta));
	221	}
	222	}
	223
	224	}
	225
	226	static void
	227	compareFlagsForRange(UStringPrepProfile* data,
	228	uint32_t start, uint32_t end,
	229	UStringPrepType type){
	230
	231	uint32_t result =0 ;
	232	UStringPrepType retType;
	233	UBool isIndex=FALSE;
	234	int32_t value=0;
	235	UTrie trie = data->sprepTrie;
	236	/*
	237	// supplementary code point
4388f060	238	UChar __lead16=U16_LEAD(0x2323E);
374ca955 A	239	int32_t __offset;
	240
	241	// get data for lead surrogate
	242	(result)=_UTRIE_GET_RAW((&idnTrie), index, 0, (__lead16));
	243	__offset=(&idnTrie)->getFoldingOffset(result);
	244
	245	// get the real data from the folded lead/trail units
	246	if(__offset>0) {
	247	(result)=_UTRIE_GET_RAW((&idnTrie), index, __offset, (0x2323E)&0x3ff);
	248	} else {
	249	(result)=(uint32_t)((&idnTrie)->initialValue);
	250	}
	251
	252	UTRIE_GET16(&idnTrie,0x2323E, result);
	253	*/
	254	while(start < end+1){
	255	UTRIE_GET16(&trie,start, result);
	256	retType = getValues(result, &value, &isIndex);
	257	if(result > _SPREP_TYPE_THRESHOLD){
	258	if(retType != type){
	259	log_err( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
	260	}
	261	}else{
	262	if(type == USPREP_PROHIBITED && ((result & 0x01) != 0x01)){
	263	log_err( "FAIL: Did not get the expected type for 0x%06X. Expected: %s Got: %s\n",start,usprepTypeNames[type], usprepTypeNames[retType]);
	264	}
	265	}
	266
	267	start++;
	268	}
	269
	270	}
	271
	272	void
	273	doStringPrepTest(const char* binFileName, const char* txtFileName, int32_t options, UErrorCode* errorCode){
	274
	275	const char *testdatapath = loadTestData(errorCode);
	276	const char *srcdatapath = NULL;
	277	const char *relativepath = NULL;
	278	char *filename = NULL;
	279	UStringPrepProfile* profile = NULL;
	280
	281	#ifdef U_TOPSRCDIR
	282	srcdatapath = U_TOPSRCDIR;
	283	relativepath = U_FILE_SEP_STRING"test"U_FILE_SEP_STRING"testdata"U_FILE_SEP_STRING;
	284	#else
	285	srcdatapath = ctest_dataOutDir();
	286	relativepath = ".."U_FILE_SEP_STRING".."U_FILE_SEP_STRING"test"U_FILE_SEP_STRING"testdata"U_FILE_SEP_STRING;
	287	#endif
	288
374ca955 A	289	profile = usprep_open(testdatapath, binFileName, errorCode);
374ca955 A	290
46f4442e A	291	if(*errorCode == U_FILE_ACCESS_ERROR) {
	292	log_data_err("Failed to load %s data file. Error: %s \n", binFileName, u_errorName(*errorCode));
	293	return;
	294	} else if(U_FAILURE(*errorCode)){
374ca955 A	295	log_err("Failed to load %s data file. Error: %s \n", binFileName, u_errorName(*errorCode));
	296	return;
	297	}
4388f060	298	filename = (char*) malloc(strlen(srcdatapath)+strlen(relativepath)+strlen(txtFileName)+10 );
374ca955 A	299	/* open and load the txt file */
	300	strcpy(filename,srcdatapath);
	301	strcat(filename,relativepath);
	302	strcat(filename,txtFileName);
	303
	304	parseMappings(filename,profile, TRUE,errorCode);
	305
	306	free(filename);
	307	}
	308	#endif
	309	/*
	310	* Hey, Emacs, please set the following:
	311	*
	312	* Local Variables:
	313	* indent-tabs-mode: nil
	314	* End:
	315	*
	316	*/