bool CDOMDocument::loadXML(char *xml, unsigned long len)
{
bool ret = true;
// void *buff;
if(!this->parser)
return(false);
this->depth = -1;
this->State = CDOMDocument::INTO_UNKNOWN;
this->indexStart = 0;
this->indexEnd = 0;
this->tokenLen = 0;
this->tokenLCLen = 0;
this->tokenLCNDLen = 0;
this->wordIndex = 0;
this->parseText = true;
if(this->path)
_FREE(this->path);
this->path = (char *)_MALLOC_WHY(200, "dom.cpp:loadXML:path");
if(this->path)
{
this->path_msize = 200;
this->path[0] = '\0';
this->freepathoffset = 0;
}
if(this->upath)
_FREE(this->upath);
this->upath = (char *)_MALLOC_WHY(200, "dom.cpp:loadXML:upath");
if(this->upath)
{
this->upath_msize = 200;
this->upath[0] = '\0';
this->freeupathoffset = 0;
}
if(XML_Parse(this->parser, xml, len, true) != XML_STATUS_ERROR)
{
}
else
{
// handle parse error
zSyslog._log(CSyslog::LOGL_ERR, CSyslog::LOGC_XMLERR, "Parse error at line %u:\n%s\n",
XML_GetCurrentLineNumber(this->parser),
XML_ErrorString(XML_GetErrorCode(this->parser)));
ret = false;
}
return(ret);
}
CDOMElement::CDOMElement(const char *s, class CDOMDocument *owner)
{
int l;
this->ownerDocument = owner;
this->firstChild = this->lastChild = NULL;
this->nodeType = CDOMNode::XML_ELEMENT_NODE;
if( (this->tagName = (char *)_MALLOC_WHY(l = strlen(s)+1, "dom.h:CDOMElement:tagName")) )
memcpy(this->tagName, s, l);
this->valueLCND = NULL;
this->valueLCND_buffer_size = 0;
this->valueLCND_length = 0;
this->valueLC = NULL;
this->valueLC_buffer_size = 0;
this->valueLC_length = 0;
this->value = NULL;
this->value_buffer_size = 0;
this->value_length = 0;
this->value_end = 0;
this->index = 0;
this->pathoffset = 0;
this->field = NULL;
this->index_start = this->index_end = 0;
this->t0 = this->t1 = this->k0 = this->k1 = -1;
this->ink = 0;
this->lastFlags = CFLAG_ENDCHAR;
};
void CIndexer::flush()
{
if(!this->tRecord)
{
// printf("| nothing to flush.\n");
}
else
{
this->connbas->execute((char *)("START TRANSACTION"), 17);
int nrecs_flushed = 0;
// -----------------------------------------------------------------------------
// warning : before flushing thits, we check that the thesaurus/cterms hasn't changed
// else we will set the records as 'to-reindex-thesaurus' again
// -----------------------------------------------------------------------------
bool thesaurusChanged = false;
// --------------------------------------------------- flush xpath
CXPath *xp;
//start by counting unknown xpath
int nNewXPath = 0;
for(xp = this->tXPaths; xp; xp = xp->next)
{
if(xp->id == 0)
nNewXPath++;
}
if(nNewXPath > 0)
{
// there is some unknown, we get a uid and we write
unsigned int xpath_new_uid;
xpath_new_uid = this->connbas->getID("XPATH", nNewXPath);
for(xp = this->tXPaths; !this->connbas->crashed && xp; xp = xp->next)
{
if(xp->id == 0)
{
xp->new_id = xpath_new_uid++;
int r;
if((r = this->connbas->insertXPath(xp->upath, &(xp->new_id))) == 0)
{
// ok : we have created the xpath, or if it was existing the id is returned in xp->new_id
xp->id = xp->new_id;
}
else
{
// err : no way to create the xpath neither finding the existing one
}
}
}
}
// --------------------------------------------------- flush the record (delete idx, prop, thits)
// create a list of rids
CRecord *r;
int lrids_len = 0;
char *pbuff, *lrids_buff;
for(r=this->tRecord; r; r=r->next)
lrids_len += 34; // 33=lmax of itoa() + comma delimiter.
pbuff = lrids_buff = (char *)(_MALLOC_WHY(lrids_len, "indexer.cpp:flush:pbuff"));
while( (r = this->tRecord) )
{
if(pbuff)
{
pbuff += sprintf(pbuff, "%d", r->id);
if(r->next)
*pbuff++ = ',';
}
this->tRecord = r->next;
delete r;
nrecs_flushed++;
}
lrids_len = pbuff-lrids_buff; // ajuste la longueur
if(lrids_buff)
{
// delete idx, prop, thits for those records
this->connbas->delRecRefs2(lrids_buff, lrids_len);
// lock prefs and thits
// if(!this->connbas->crashed && (this->connbas->lockPref() == 0))
// {
// check if something has changed in the thesaurus
time_t struct_moddate, thesaurus_moddate, cterms_moddate;
this->connbas->selectPref_moddates(&struct_moddate, &thesaurus_moddate, &cterms_moddate);
thesaurusChanged = (thesaurus_moddate > this->current_thesaurus_moddate || cterms_moddate > this->current_cterms_moddate);
if(!thesaurusChanged)
{
// thesaurus hasn't change
if(!this->connbas->crashed && this->ctermsChanged)
{
// cterms has changed, let's save
saveCterms(this);
this->ctermsChanged = false;
}
}
// --------------------------------------------------- flush thit
CTHit *th;
while(!this->connbas->crashed && (th = this->firstTHit) )
{
if(!thesaurusChanged)
{
// if the th/ct hasn't chnaged, we can flush thits
this->connbas->insertTHit(th->record_id, th->pxpath->id, th->pxpath->field->name, th->value, th->hitstart, th->hitlen, th->business);
}
this->firstTHit = th->next;
delete th;
}
// we can unlock
// this->connbas->unlockTables();
// flag records 'to-reindex-thesaurus'
if(!this->connbas->crashed && thesaurusChanged)
{
// this->connbas->execute(ibuf, pibuf-ibuf);
// _FREE(ibuf);
this->connbas->setRecordsToReindexTh2(lrids_buff, lrids_len);
}
// }
// --------------------------------------------------- flush prop
CProp *p;
while(!this->connbas->crashed && (p = this->firstProp) )
{
this->connbas->insertProp(p->record_id, p->pxpath->id, p->pxpath->field->uname, p->value, p->type, p->business);
this->firstProp = p->next;
delete p;
}
// --------------------------------------------------- flush kword and idx
unsigned int kword_new_uid = 0;
if(this->nNewKeywords > 0)
kword_new_uid = this->connbas->getID("KEYWORDS", this->nNewKeywords);
CKword *k;
CHit *h;
for(int hash=0; !this->connbas->crashed && (hash<KWORD_HASHSIZE); hash++)
{
for(k=this->tKeywords[hash]; !this->connbas->crashed && k; k=k->next)
{
// save new kwords
if(k->id == 0)
{
k->new_id = kword_new_uid++;
int r;
if((r = this->connbas->insertKword(k->kword, k->l, &(k->new_id))) == 0)
{
// ok : we have created the kword with new_id, or if it was existing the id is returned in k->new_id
k->id = k->new_id;
}
else
{
}
}
// on save les hits
if(k->id > 0)
{
while(!this->connbas->crashed && (h = k->firsthit) )
{
this->connbas->insertIdx(h->record_id, k->id, h->index, h->pxpath->id, h->pos, h->len, h->business);
k->firsthit = h->next;
delete h;
}
}
}
}
this->nNewKeywords = 0;
// flag the records as 'indexed' (status-bit 2,1,0 to '1')
this->connbas->updateRecord_unlock2(lrids_buff, lrids_len);
this->nrecsIndexed += nrecs_flushed;
_FREE(lrids_buff);
}
zSyslog._log(CSyslog::LOGL_INFO, CSyslog::LOGC_INDEXING, "#%d : %d records flushed", this->connbas->sbas_id, nrecs_flushed);
this->connbas->execute((char *)("COMMIT"), 6);
for(xp = this->tXPaths; xp; xp = xp->next)
{
if(xp->id == 0 && xp->new_id != 0)
{
xp->id = xp->new_id;
}
}
for(int hash=0; hash<KWORD_HASHSIZE; hash++)
{
for(CKword *k=this->tKeywords[hash]; k; k=k->next)
{
// save new kwords
if(k->id == 0 && k->new_id != 0)
{
k->id = k->new_id;
}
}
}
}
}
