Forks/privacore-open-source-search-engine
1
0
Fork 0
forked from Mirrors/privacore-open-source-search-engine
Code Pull Requests Activity

Fixed tmp-set-NUL-string hack in Words.

Browse Source 
Seems to only have had an effect when being set via an Xml instance.
This commit is contained in:
Ivan Skytte Jørgensen
2018-02-09 17:11:50 +01:00
parent f7fa5858fc
commit 8f41e07f62
1 changed files with 2 additions and 7 deletions

9
Words.cpp

@ -154,12 +154,7 @@ bool Words::set( Xml *xml, int32_t node1, int32_t node2 ) {
// is the kth node a tag?
if ( !xml->isTag( k ) ) {
/// @todo ALC why are we adding NULL and restoring it after?
/// addWords should be change to use nodeLen and not null terminated string
char c = node[nodeLen];
node[nodeLen] = '\0';
addWords(node, nodeLen);
node[nodeLen] = c;
continue;
}
@ -234,7 +229,7 @@ bool Words::addWords(char *s, int32_t nodeLen) {
// it is a punct word, find end of it
char *start = s+i;
for ( ; s[i] ; i += getUtf8CharSize(s+i)) {
for ( ; i<nodeLen && s[i] ; i += getUtf8CharSize(s+i)) {
// if we are simple ascii, skip quickly
if ( is_ascii(s[i]) ) {
// accumulate NON-alnum chars
@ -280,7 +275,7 @@ bool Words::addWords(char *s, int32_t nodeLen) {
// get an alnum word
j = i;
again:
for ( ; s[i] ; i += getUtf8CharSize(s+i) ) {
for ( ; i<nodeLen && s[i] ; i += getUtf8CharSize(s+i) ) {
// simple ascii?
if ( is_ascii(s[i]) ) {
// accumulate alnum chars