size_t *FrTextSpans::wordMapping(bool skip_white) const
{
if (!originalString() || textLength() == 0)
return 0 ;
size_t *map = FrNewN(size_t,textLength()+1) ;
if (map)
{
size_t count = 0 ;
map[0] = count ;
if (skip_white)
{
size_t i ;
for (i = 1 ; i <= textLength() ; i++)
{
if (m_positions[i] != m_positions[i-1] + 1)
count++ ;
map[i] = count ;
}
}
else
{
for (size_t i = 1 ; i <= textLength() ; i++)
{
if (Fr_isspace(m_text[i-1]) && !Fr_isspace(m_text[i]))
count++ ;
map[i] = count ;
}
}
}
return map ;
}
void FrTextSpans::makeWordSpans(const char *text, FrCharEncoding enc,
const char *word_delim)
{
m_text = FrDupString(text) ;
if (m_text)
{
if (!setPositionMap())
{
m_textlength = 0 ;
return ;
}
char *canon = FrCanonicalizeSentence(m_text,enc,false,word_delim) ;
if (canon && *canon)
{
m_spancount = 1 ;
for (char *cptr = canon ; *cptr ; cptr++)
{
if (' ' == *cptr)
m_spancount++ ;
}
m_spans = FrNewN(FrTextSpan,m_spancount) ;
if (m_spans)
{
size_t tpos = 0 ;
size_t cpos = 0 ;
size_t start = 0 ;
size_t end = 0 ;
for (size_t i = 0 ; i < m_spancount ; i++)
{
// scan over the nonwhitespace chars at the current location
for ( ; canon[cpos] && canon[cpos] != ' ' ; cpos++)
{
tpos++ ;
end++ ; // counts toward m_positions index
}
// skip over any trailing whitespace
while (m_text[tpos] && Fr_isspace(m_text[tpos]))
tpos++ ;
if (canon[cpos] == ' ')
cpos++ ;
m_spans[i].init(start,end-1,DEFAULT_SCORE,DEFAULT_WEIGHT,
0,this) ;
m_sorted = false ;
start = end ;
}
}
}
FrFree(canon) ;
}
return ;
}
void FrTextSpans::makeWordSpans(const FrList *defn)
{
FrString *concat = new FrString(defn) ;
m_text = FrDupString(concat->stringValue()) ;
free_object(concat) ;
if (setPositionMap())
{
m_spancount = defn->simplelistlength() ;
m_spans = FrNewN(FrTextSpan,m_spancount) ;
if (!m_spans)
{
m_spancount = 0 ;
return ;
}
size_t spannum = 0 ;
size_t start = 0 ;
size_t end = 0 ;
for ( ; defn ; defn = defn->rest())
{
FrObject *def = defn->first() ;
if (def && def->structp())
{
addMetaData((FrStruct*)def) ;
continue ;
}
const char *item = FrPrintableName(def) ;
if (!item)
item = "" ;
for (const char *i = item ; *i ; i++)
{
if (!Fr_isspace(*i))
end++ ;
}
if (end > 0)
m_spans[spannum++].init(start,end-1,DEFAULT_SCORE,DEFAULT_WEIGHT,
item,this) ;
m_sorted = false ;
start = end ;
}
m_spancount = spannum ;
}
return ;
}
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 ;
}
WildcardCollection::WildcardCollection(const WildcardCollection *orig,
bool allow_all_if_empty)
{
if (!orig)
{
m_numsets = 0 ;
return ;
}
m_wildcards = FrNewN(WildcardSet,orig->numSets()) ;
if (!m_wildcards)
{
m_numsets = 0 ;
return ;
}
else
{
m_numsets = orig->numSets() ;
copy(orig,allow_all_if_empty) ;
}
return ;
}
WildcardCollection::WildcardCollection(unsigned max_ref, bool allow_all)
{
m_wildcards = FrNewN(WildcardSet,max_ref) ;
if (!m_wildcards)
{
m_numsets = 0 ;
return ;
}
else
{
m_numsets = max_ref ;
if (allow_all)
{
for (unsigned i = 0 ; i < numSets() ; i++)
m_wildcards[i].addAll() ;
}
else
{
for (unsigned i = 0 ; i < numSets() ; i++)
m_wildcards[i].removeAll() ;
}
}
return ;
}
