some parser optimizations, xmllint --memory --timing --repeat --stream
* dict.c parser.c xmlstring.c: some parser optimizations,
xmllint --memory --timing --repeat --stream ./db10000.xml
went down from 16.5 secs to 15.5 secs.
Daniel
diff --git a/ChangeLog b/ChangeLog
index ea18ee4..37fbdb1 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,9 @@
+Mon Jan 12 17:22:57 CET 2004 Daniel Veillard <daniel@veillard.com>
+
+ * dict.c parser.c xmlstring.c: some parser optimizations,
+ xmllint --memory --timing --repeat --stream ./db10000.xml
+ went down from 16.5 secs to 15.5 secs.
+
Thu Jan 8 17:57:50 CET 2004 Daniel Veillard <daniel@veillard.com>
* xmlschemas.c: removed a memory leak remaining from the switch
diff --git a/dict.c b/dict.c
index b56cef2..0a9aebe 100644
--- a/dict.c
+++ b/dict.c
@@ -484,14 +484,30 @@
} else {
for (insert = &(dict->dict[key]); insert->next != NULL;
insert = insert->next) {
+#ifdef __GNUC__
+ if (insert->len == len) {
+ register int tmp = memcmp(insert->name, name, len);
+ if (!tmp)
+ return(insert->name);
+ }
+#else
if ((insert->len == len) &&
(!xmlStrncmp(insert->name, name, len)))
return(insert->name);
+#endif
nbi++;
}
+#ifdef __GNUC__
+ if (insert->len == len) {
+ register int tmp = memcmp(insert->name, name, len);
+ if (!tmp)
+ return(insert->name);
+ }
+#else
if ((insert->len == len) &&
(!xmlStrncmp(insert->name, name, len)))
return(insert->name);
+#endif
}
ret = xmlDictAddString(dict, name, len);
diff --git a/parser.c b/parser.c
index ff04816..4c76bf9 100644
--- a/parser.c
+++ b/parser.c
@@ -2038,13 +2038,15 @@
* @ctxt: an XML parser context
* @str: a xmlChar *
* @len: the size of @str
+ * @blank_chars: we know the chars are blanks
*
* Is this a sequence of blank chars that one can ignore ?
*
* Returns 1 if ignorable 0 otherwise.
*/
-static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) {
+static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
+ int blank_chars) {
int i, ret;
xmlNodePtr lastChild;
@@ -2064,8 +2066,10 @@
/*
* Check that the string is made of blanks
*/
- for (i = 0;i < len;i++)
- if (!(IS_BLANK_CH(str[i]))) return(0);
+ if (blank_chars == 0) {
+ for (i = 0;i < len;i++)
+ if (!(IS_BLANK_CH(str[i]))) return(0);
+ }
/*
* Look if the element is mixed content in the DTD if available
@@ -2339,8 +2343,8 @@
static const xmlChar *
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
- const xmlChar *cmp = other;
- const xmlChar *in;
+ register const xmlChar *cmp = other;
+ register const xmlChar *in;
const xmlChar *ret;
GROW;
@@ -3106,10 +3110,50 @@
if (!cdata) {
in = ctxt->input->cur;
do {
+get_more_space:
+ while (*in == 0x20) in++;
+ if (*in == 0xA) {
+ ctxt->input->line++;
+ in++;
+ while (*in == 0xA) {
+ ctxt->input->line++;
+ in++;
+ }
+ goto get_more_space;
+ }
+ if (*in == '<') {
+ nbchar = in - ctxt->input->cur;
+ if (nbchar > 0) {
+ const xmlChar *tmp = ctxt->input->cur;
+ ctxt->input->cur = in;
+
+ if (ctxt->sax->ignorableWhitespace !=
+ ctxt->sax->characters) {
+ if (areBlanks(ctxt, tmp, nbchar, 1)) {
+ ctxt->sax->ignorableWhitespace(ctxt->userData,
+ tmp, nbchar);
+ } else if (ctxt->sax->characters != NULL)
+ ctxt->sax->characters(ctxt->userData,
+ tmp, nbchar);
+ } else if (ctxt->sax->characters != NULL) {
+ ctxt->sax->characters(ctxt->userData,
+ tmp, nbchar);
+ }
+ }
+ return;
+ }
get_more:
+#if 0
while (((*in >= 0x20) && (*in != '<') && (*in != ']') &&
(*in != '&') && (*in <= 0x7F)) || (*in == 0x09))
in++;
+#endif
+ while (((*in > ']') && (*in <= 0x7F)) ||
+ ((*in > '&') && (*in < '<')) ||
+ ((*in > '<') && (*in < ']')) ||
+ ((*in >= 0x20) && (*in < '&')) ||
+ (*in == 0x09))
+ in++;
if (*in == 0xA) {
ctxt->input->line++;
in++;
@@ -3136,7 +3180,7 @@
const xmlChar *tmp = ctxt->input->cur;
ctxt->input->cur = in;
- if (areBlanks(ctxt, tmp, nbchar)) {
+ if (areBlanks(ctxt, tmp, nbchar, 0)) {
ctxt->sax->ignorableWhitespace(ctxt->userData,
tmp, nbchar);
} else if (ctxt->sax->characters != NULL)
@@ -3217,7 +3261,7 @@
* OK the segment is to be consumed as chars.
*/
if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
- if (areBlanks(ctxt, buf, nbchar)) {
+ if (areBlanks(ctxt, buf, nbchar, 0)) {
if (ctxt->sax->ignorableWhitespace != NULL)
ctxt->sax->ignorableWhitespace(ctxt->userData,
buf, nbchar);
@@ -3242,7 +3286,7 @@
* OK the segment is to be consumed as chars.
*/
if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
- if (areBlanks(ctxt, buf, nbchar)) {
+ if (areBlanks(ctxt, buf, nbchar, 0)) {
if (ctxt->sax->ignorableWhitespace != NULL)
ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar);
} else {
@@ -6689,6 +6733,7 @@
/**
* xmlParseNCName:
* @ctxt: an XML parser context
+ * @len: lenght of the string parsed
*
* parse an XML name.
*
@@ -7142,7 +7187,7 @@
static const xmlChar *
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
- const xmlChar **URI) {
+ const xmlChar **URI, int *tlen) {
const xmlChar *localname;
const xmlChar *prefix;
const xmlChar *attname;
@@ -7182,6 +7227,7 @@
"StartTag: invalid element name\n");
return(NULL);
}
+ *tlen = ctxt->input->cur - ctxt->input->base - cur;
/*
* Now parse the attributes, it ends up with the ending
@@ -7507,7 +7553,7 @@
static void
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlChar *prefix,
- const xmlChar *URI, int line, int nsNr) {
+ const xmlChar *URI, int line, int nsNr, int tlen) {
const xmlChar *name;
GROW;
@@ -7517,7 +7563,19 @@
}
SKIP(2);
- name = xmlParseQNameAndCompare(ctxt, ctxt->name, prefix);
+ if ((tlen > 0) && (memcmp(ctxt->input->cur, ctxt->name, tlen) == 0)) {
+ if (ctxt->input->cur[tlen] == '>') {
+ ctxt->input->cur += tlen + 1;
+ goto done;
+ }
+ ctxt->input->cur += tlen;
+ name = (xmlChar*)1;
+ } else {
+ if (prefix == NULL)
+ name = xmlParseNameAndCompare(ctxt, ctxt->name);
+ else
+ name = xmlParseQNameAndCompare(ctxt, ctxt->name, prefix);
+ }
/*
* We should definitely be at the ending "S? '>'" part
@@ -7545,6 +7603,7 @@
/*
* SAX: End of Tag
*/
+done:
if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
(!ctxt->disableSAX))
ctxt->sax->endElementNs(ctxt->userData, ctxt->name, prefix, URI);
@@ -7753,7 +7812,7 @@
const xmlChar *prefix;
const xmlChar *URI;
xmlParserNodeInfo node_info;
- int line;
+ int line, tlen;
xmlNodePtr ret;
int nsNr = ctxt->nsNr;
@@ -7773,7 +7832,7 @@
#ifdef LIBXML_SAX1_ENABLED
if (ctxt->sax2)
#endif /* LIBXML_SAX1_ENABLED */
- name = xmlParseStartTag2(ctxt, &prefix, &URI);
+ name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen);
#ifdef LIBXML_SAX1_ENABLED
else
name = xmlParseStartTag(ctxt);
@@ -7878,7 +7937,7 @@
* parse the end of tag: '</' should be here.
*/
if (ctxt->sax2) {
- xmlParseEndTag2(ctxt, prefix, URI, line, ctxt->nsNr - nsNr);
+ xmlParseEndTag2(ctxt, prefix, URI, line, ctxt->nsNr - nsNr, tlen);
namePop(ctxt);
}
#ifdef LIBXML_SAX1_ENABLED
@@ -8734,7 +8793,7 @@
static int
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
int ret = 0;
- int avail;
+ int avail, tlen;
xmlChar cur, next;
const xmlChar *lastlt, *lastgt;
@@ -8984,7 +9043,7 @@
#ifdef LIBXML_SAX1_ENABLED
if (ctxt->sax2)
#endif /* LIBXML_SAX1_ENABLED */
- name = xmlParseStartTag2(ctxt, &prefix, &URI);
+ name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen);
#ifdef LIBXML_SAX1_ENABLED
else
name = xmlParseStartTag(ctxt);
@@ -9159,7 +9218,7 @@
xmlParseEndTag2(ctxt,
(void *) ctxt->pushTab[ctxt->nameNr * 3 - 3],
(void *) ctxt->pushTab[ctxt->nameNr * 3 - 2], 0,
- (int) (long) ctxt->pushTab[ctxt->nameNr * 3 - 1]);
+ (int) (long) ctxt->pushTab[ctxt->nameNr * 3 - 1], 0);
nameNsPop(ctxt);
}
#ifdef LIBXML_SAX1_ENABLED
diff --git a/xmlstring.c b/xmlstring.c
index b47e13e..5c64053 100644
--- a/xmlstring.c
+++ b/xmlstring.c
@@ -208,11 +208,16 @@
if (str1 == str2) return(0);
if (str1 == NULL) return(-1);
if (str2 == NULL) return(1);
+#ifdef __GNUC__
+ tmp = strncmp(str1, str2, len);
+ return tmp;
+#else
do {
tmp = *str1++ - *str2;
if (tmp != 0 || --len == 0) return(tmp);
} while (*str2++ != 0);
return 0;
+#endif
}
static const xmlChar casemap[256] = {