void FrTextSpans::parseSpans(const FrList *spans, size_t numstrings)
{
for (const FrList *sp = spans ; sp ; sp = sp->rest())
{
const FrObject *item = sp->first() ;
if (item && (item->stringp() || item->symbolp()))
{
m_text = FrDupString(FrPrintableName(item)) ;
if (m_text && setPositionMap())
break ;
return ;
}
}
m_spancount = spans->simplelistlength() - numstrings ;
m_spans = FrNewN(FrTextSpan,m_spancount) ;
if (!m_spans)
{
m_spancount = 0 ;
return ;
}
size_t count = 0 ;
for ( ; spans ; spans = spans->rest())
{
FrList *span = (FrList*)spans->first() ;
if (!span)
continue ;
if (span->consp() && m_spans[count].parse(span,this))
{
count++ ;
m_sorted = false ;
}
else if (span->structp())
{
FrStruct *meta = (FrStruct*)span ;
if (meta->typeName() &&
Fr_stricmp(FrPrintableName(meta->typeName()),"META") == 0)
addMetaData(meta) ;
}
}
m_spancount = count ;
return ;
}
bool FrTextSpan::parse(const FrList *span, FrTextSpans *contain)
{
if (span && span->consp() && span->simplelistlength() >= 2 &&
span->first() && span->second() &&
span->first()->numberp() && span->second()->numberp())
{
size_t sp_start = span->first()->intValue() ;
size_t sp_end = span->second()->intValue() ;
span = span->rest()->rest() ;
// we'll allow rather free-form input from the rest of the span's
// description: the first two numbers are the score and weight,
// respectively, the first string is the span's text, the second
// string (if present) becomes the INIT_TEXT metadata. Then,
// the first structure (if present) is the metadata, and any lists
// starting with a symbol become additional metadata fields
double sp_score = DEFAULT_SCORE ;
double sp_weight = DEFAULT_WEIGHT ;
// scan for the first two numbers
for (const FrList *sp = span ; sp ; sp = sp->rest())
{
if (sp->first() && sp->first()->numberp())
{
sp_score = sp->first()->floatValue() ;
for (sp = sp->rest() ; sp ; sp = sp->rest())
{
if (sp->first() && sp->first()->numberp())
{
sp_weight = sp->first()->floatValue() ;
break ;
}
}
break ;
}
}
const char *curr_text = 0 ;
const char *orig_text = 0 ;
// scan for the first two strings or symbols
for (const FrList *sp = span ; sp ; sp = sp->rest())
{
FrObject *item = sp->first() ;
if (item && (item->stringp() || item->symbolp()))
{
curr_text = item->printableName() ;
for (sp = sp->rest() ; sp ; sp = sp->rest())
{
item = sp->first() ;
if (item && (item->stringp() || item->symbolp()))
{
orig_text = item->printableName() ;
break ;
}
}
break ;
}
}
if (curr_text)
(void)FrSkipWhitespace(curr_text) ;
if (orig_text)
(void)FrSkipWhitespace(orig_text) ;
// scan for the first structure
const FrStruct *meta = 0 ;
for (const FrList *sp = span ; sp ; sp = sp->rest())
{
if (sp->first() && sp->first()->structp())
{
meta = (FrStruct*)sp->first() ;
break ;
}
}
init(sp_start,sp_end,sp_score,sp_weight,curr_text,contain) ;
free_object(m_metadata) ;
if (meta)
{
FrSymbol *symMETATYPE = FrSymbolTable::add(METADATA_TYPENAME) ;
if (meta->typeName() == symMETATYPE)
m_metadata = (FrStruct*)meta->deepcopy() ;
else
{
// copy the keywords one by one
FrList *keys = meta->fieldNames() ;
while (keys)
{
FrSymbol *key = (FrSymbol*)poplist(keys) ;
setMetaData(key,meta->get(key)) ;
}
}
}
if (orig_text)
setMetaData(FrSymbolTable::add(init_text_tag),
new FrString(orig_text),false) ;
// finally, scan for any embedded lists and add them as metadata fields
for (const FrList *sp = span ; sp ; sp = sp->rest())
{
FrList *item = (FrList*)sp->first() ;
if (item && item->consp() && item->first() &&
item->first()->symbolp())
{
FrSymbol *key = (FrSymbol*)item->first() ;
setMetaData(key,item->rest()) ;
}
}
return true ;
}
return false ;
}
