typedef struct {
NAMED **v;
+ unsigned char power;
size_t size;
size_t used;
- size_t usedLim;
const XML_Memory_Handling_Suite *mem;
} HASH_TABLE;
+/* Basic character hash algorithm, taken from Python's string hash:
+ h = h * 1000003(prime number) ^ character.
+*/
+#ifdef XML_UNICODE
+#define CHAR_HASH(h, c) \
+ (((h) * 0xF4243) ^ (unsigned short)(c))
+#else
+#define CHAR_HASH(h, c) \
+ (((h) * 0xF4243) ^ (unsigned char)(c))
+#endif
+
+/* For probing (after a collision) we need a step size relative prime
+ to the hash table size, which is a power of 2. We use double-hashing,
+ since we can calculate a second hash value cheaply by taking those bits
+ of the first hash value that were discarded (masked out) when the table
+ index was calculated: index = hash & mask, where mask = table->size - 1.
+ We limit the maximum step size to table->size / 4 (mask >> 2) and make
+ it odd, since odd numbers are always relative prime to a power of 2.
+*/
+#define SECOND_HASH(hash, mask, power) \
+ ((((hash) & ~(mask)) >> ((power) - 1)) & ((mask) >> 2))
+#define PROBE_STEP(hash, mask, power) \
+ ((unsigned char)((SECOND_HASH(hash, mask, power)) | 1))
+
typedef struct {
NAMED **p;
NAMED **end;
#define INIT_TAG_BUF_SIZE 32 /* must be a multiple of sizeof(XML_Char) */
#define INIT_DATA_BUF_SIZE 1024
#define INIT_ATTS_SIZE 16
+#define INIT_ATTS_VERSION 0xFFFFFFFF
#define INIT_BLOCK_SIZE 1024
#define INIT_BUFFER_SIZE 1024
} DEFAULT_ATTRIBUTE;
typedef struct {
+ unsigned long version;
unsigned long hash;
const XML_Char *uriName;
} NS_ATT;
int m_idAttIndex;
ATTRIBUTE *m_atts;
NS_ATT *m_nsAtts;
- int m_nsAttsSize;
+ unsigned long m_nsAttsVersion;
+ unsigned char m_nsAttsPower;
POSITION m_position;
STRING_POOL m_tempPool;
STRING_POOL m_temp2Pool;
char *m_groupConnector;
- unsigned m_groupSize;
+ unsigned int m_groupSize;
XML_Char m_namespaceSeparator;
XML_Parser m_parentParser;
#ifdef XML_DTD
#define nSpecifiedAtts (parser->m_nSpecifiedAtts)
#define idAttIndex (parser->m_idAttIndex)
#define nsAtts (parser->m_nsAtts)
-#define nsAttsSize (parser->m_nsAttsSize)
+#define nsAttsVersion (parser->m_nsAttsVersion)
+#define nsAttsPower (parser->m_nsAttsPower)
#define tempPool (parser->m_tempPool)
#define temp2Pool (parser->m_temp2Pool)
#define groupConnector (parser->m_groupConnector)
ns_triplets = XML_FALSE;
nsAtts = NULL;
- nsAttsSize = 0;
+ nsAttsVersion = 0;
+ nsAttsPower = 0;
poolInit(&tempPool, &(parser->m_mem));
poolInit(&temp2Pool, &(parser->m_mem));
+ XmlNameLength(enc, atts[i].name));
if (!attId)
return XML_ERROR_NO_MEMORY;
- /* detect duplicate attributes by their QNames */
+ /* Detect duplicate attributes by their QNames. This does not work when
+ namespace processing is turned on and different prefixes for the same
+ namespace are used. For this case we have a check further down.
+ */
if ((attId->name)[-1]) {
if (enc == encoding)
eventPtr = atts[i].name;
}
appAtts[attIndex] = 0;
- /* expand prefixed attribute names and
- clear flags that say whether attributes were specified */
+ /* expand prefixed attribute names, check for duplicates,
+ and clear flags that say whether attributes were specified */
i = 0;
if (nPrefixes) {
- int j;
- if ((nPrefixes * 2) > nsAttsSize) {
- NS_ATT *temp = (NS_ATT *)REALLOC(nsAtts, nPrefixes * 2 * sizeof(NS_ATT));
+ int j; /* hash table index */
+ unsigned long version = nsAttsVersion;
+ int nsAttsSize = (int)1 << nsAttsPower;
+ /* size of hash table must be at least 2 * (# of prefixed attributes) */
+ if ((nPrefixes << 1) >> nsAttsPower) { /* true for nsAttsPower = 0 */
+ NS_ATT *temp;
+ /* hash table size must also be a power of 2 and >= 8 */
+ while (nPrefixes >> nsAttsPower++);
+ if (nsAttsPower < 3)
+ nsAttsPower = 3;
+ nsAttsSize = (int)1 << nsAttsPower;
+ temp = (NS_ATT *)REALLOC(nsAtts, nsAttsSize * sizeof(NS_ATT));
if (!temp)
return XML_ERROR_NO_MEMORY;
nsAtts = temp;
- nsAttsSize = nPrefixes * 2;
+ version = 0; /* force re-initialization of nsAtts hash table */
}
- /* clear nsAtts hash table */
- for (j = 0; j < nsAttsSize; j++)
- nsAtts[j].uriName = NULL;
+ /* using a version flag saves us from initializing nsAtts every time */
+ if (!version) { /* initialize version flags when version wraps around */
+ version = INIT_ATTS_VERSION;
+ for (j = nsAttsSize; j != 0; )
+ nsAtts[--j].version = version;
+ }
+ nsAttsVersion = --version;
+ /* expand prefixed names and check for duplicates */
for (; i < attIndex; i += 2) {
const XML_Char *s = appAtts[i];
- if (s[-1] == 2) {
+ if (s[-1] == 2) { /* prefixed */
ATTRIBUTE_ID *id;
const BINDING *b;
unsigned long uriHash = 0;
- ((XML_Char *)s)[-1] = 0;
+ ((XML_Char *)s)[-1] = 0; /* clear flag */
id = (ATTRIBUTE_ID *)lookup(&dtd->attributeIds, s, 0);
b = id->prefix->binding;
if (!b)
return XML_ERROR_UNBOUND_PREFIX;
- /* b->uri includes namespace separator */
+ /* as we expand the name we also calculate its hash value */
for (j = 0; j < b->uriLen; j++) {
const XML_Char c = b->uri[j];
if (!poolAppendChar(&tempPool, c))
return XML_ERROR_NO_MEMORY;
- uriHash = (uriHash << 5) + uriHash + (unsigned char)c;
+ uriHash = CHAR_HASH(uriHash, c);
}
while (*s++ != XML_T(':'))
;
- do {
+ do { /* copies null terminator */
const XML_Char c = *s;
if (!poolAppendChar(&tempPool, *s))
return XML_ERROR_NO_MEMORY;
- uriHash = (uriHash << 5) + uriHash + (unsigned char)c;
+ uriHash = CHAR_HASH(uriHash, c);
} while (*s++);
- /* detect duplicate attributes based on uriName = uri + local name */
- for (j = uriHash & (nsAttsSize - 1);
- nsAtts[j].uriName;
- j == 0 ? j = nsAttsSize - 1 : --j) {
- if (uriHash == nsAtts[j].hash) {
- const XML_Char *s1 = poolStart(&tempPool); /* null-terminated */
- const XML_Char *s2 = nsAtts[j].uriName;
- for (; *s1 == *s2 && *s1 != 0; s1++, s2++);
- if (*s1 == 0)
- return XML_ERROR_DUPLICATE_ATTRIBUTE;
+ { /* Check hash table for duplicate of expanded name (uriName).
+ Derived from code in lookup(HASH_TABLE *table, ...).
+ */
+ unsigned char step = 0;
+ unsigned long mask = nsAttsSize - 1;
+ j = uriHash & mask; /* index into hash table */
+ while (nsAtts[j].version == version) {
+ /* for speed we compare stored hash values first */
+ if (uriHash == nsAtts[j].hash) {
+ const XML_Char *s1 = poolStart(&tempPool);
+ const XML_Char *s2 = nsAtts[j].uriName;
+ /* s1 is null terminated, but not s2 */
+ for (; *s1 == *s2 && *s1 != 0; s1++, s2++);
+ if (*s1 == 0)
+ return XML_ERROR_DUPLICATE_ATTRIBUTE;
+ }
+ if (!step)
+ step = PROBE_STEP(uriHash, mask, nsAttsPower);
+ j < step ? ( j += nsAttsSize - step) : (j -= step);
}
}
- if (ns_triplets) {
+ if (ns_triplets) { /* append namespace separator and prefix */
tempPool.ptr[-1] = namespaceSeparator;
s = b->prefix->name;
do {
} while (*s++);
}
+ /* store expanded name in attribute list */
s = poolStart(&tempPool);
- appAtts[i] = s;
poolFinish(&tempPool);
+ appAtts[i] = s;
- /* fill empty slot with new attribute */
+ /* fill empty slot with new version, uriName and hash value */
+ nsAtts[j].version = version;
nsAtts[j].hash = uriHash;
nsAtts[j].uriName = s;
if (!--nPrefixes)
break;
}
- else
- ((XML_Char *)s)[-1] = 0;
+ else /* not prefixed */
+ ((XML_Char *)s)[-1] = 0; /* clear flag */
}
}
/* clear flags for the remaining attributes */
return 1;
}
-#define INIT_SIZE 64
+#define INIT_POWER 6
static XML_Bool FASTCALL
keyeq(KEY s1, KEY s2)
{
unsigned long h = 0;
while (*s)
- h = (h << 5) + h + (unsigned char)*s++;
+ h = CHAR_HASH(h, *s++);
return h;
}
size_t i;
if (table->size == 0) {
size_t tsize;
-
if (!createSize)
return NULL;
- tsize = INIT_SIZE * sizeof(NAMED *);
+ table->power = INIT_POWER;
+ /* table->size is a power of 2 */
+ table->size = (size_t)1 << INIT_POWER;
+ tsize = table->size * sizeof(NAMED *);
table->v = (NAMED **)table->mem->malloc_fcn(tsize);
if (!table->v)
return NULL;
memset(table->v, 0, tsize);
- table->size = INIT_SIZE;
- table->usedLim = INIT_SIZE / 2;
- i = hash(name) & (table->size - 1);
+ i = hash(name) & ((unsigned long)table->size - 1);
}
else {
unsigned long h = hash(name);
- for (i = h & (table->size - 1);
- table->v[i];
- i == 0 ? i = table->size - 1 : --i) {
+ unsigned long mask = (unsigned long)table->size - 1;
+ unsigned char step = 0;
+ i = h & mask;
+ while (table->v[i]) {
if (keyeq(name, table->v[i]->name))
return table->v[i];
+ if (!step)
+ step = PROBE_STEP(h, mask, table->power);
+ i < step ? (i += table->size - step) : (i -= step);
}
if (!createSize)
return NULL;
- if (table->used == table->usedLim) {
- /* check for overflow */
- size_t newSize = table->size * 2;
+
+ /* check for overflow (table is half full) */
+ if (table->used >> (table->power - 1)) {
+ unsigned char newPower = table->power + 1;
+ size_t newSize = (size_t)1 << newPower;
+ unsigned long newMask = (unsigned long)newSize - 1;
size_t tsize = newSize * sizeof(NAMED *);
NAMED **newV = (NAMED **)table->mem->malloc_fcn(tsize);
if (!newV)
memset(newV, 0, tsize);
for (i = 0; i < table->size; i++)
if (table->v[i]) {
- size_t j;
- for (j = hash(table->v[i]->name) & (newSize - 1);
- newV[j];
- j == 0 ? j = newSize - 1 : --j)
- ;
+ unsigned long newHash = hash(table->v[i]->name);
+ size_t j = newHash & newMask;
+ step = 0;
+ while (newV[j]) {
+ if (!step)
+ step = PROBE_STEP(newHash, newMask, newPower);
+ j < step ? (j += newSize - step) : (j -= step);
+ }
newV[j] = table->v[i];
}
table->mem->free_fcn(table->v);
table->v = newV;
+ table->power = newPower;
table->size = newSize;
- table->usedLim = newSize/2;
- for (i = h & (table->size - 1);
- table->v[i];
- i == 0 ? i = table->size - 1 : --i)
- ;
+ i = h & newMask;
+ step = 0;
+ while (table->v[i]) {
+ if (!step)
+ step = PROBE_STEP(h, newMask, newPower);
+ i < step ? (i += newSize - step) : (i -= step);
+ }
}
}
table->v[i] = (NAMED *)table->mem->malloc_fcn(createSize);
table->mem->free_fcn(table->v[i]);
table->v[i] = NULL;
}
- table->usedLim = table->size / 2;
table->used = 0;
}
static void FASTCALL
hashTableInit(HASH_TABLE *p, const XML_Memory_Handling_Suite *ms)
{
+ p->power = 0;
p->size = 0;
- p->usedLim = 0;
p->used = 0;
p->v = NULL;
p->mem = ms;
return ret;
}
+
projects / libexpat / commitdiff