/** Converts an array of strings to a CBLIST. */
CBLIST *array_2_CBLIST(VALUE ary)
{
long i = 0;
CBLIST *result = cblistopen();
VALUE str;
for(i = 0; (str = rb_ary_entry(ary, i)) != Qnil; i++) {
cblistpush(result, RSTRING(str)->ptr, RSTRING(str)->len);
}
return result;
}
/**
* call-seq:
* document.add_content(index, content) -> document
*
* Takes the contents, breaks the words up, and then puts them in the document
* in normalized form. This is the common pattern that people use a Document
* with. You may also use Document.addword to add one word a time, and
* Document.add_word_list to add a list of words.
*
* It uses the default odanalyzetext method to break up the text,
* which means you can use the Index::setcharclass method to configure
* what is a DELIM, GLUE, and SPACE character. The default is the same
* as Odeum::breaktext.
*
* If the process of normalizing a word creates an empty word, then it
* is not added to the document's words. This usually happens for
* punctation that isn't usualy searched for anyway.
*
* The Index used with this document is now required since that object holds
* the information about how text is broken via the Index::setcharclass method.
*/
VALUE Document_add_content(VALUE self, VALUE index, VALUE content) {
CBLIST *asis_words = NULL;
CBLIST *norm_words = NULL;
const char *asis = NULL;
const char *norm = NULL;
int asis_len = 0;
int norm_len = 0;
int i = 0;
int count = 0;
ODDOC *oddoc = NULL;
ODEUM *odeum = NULL;
DATA_GET(self,ODDOC, oddoc);
DATA_GET(index,ODEUM, odeum);
REQUIRE_TYPE(content, T_STRING);
asis_words = cblistopen();
norm_words = cblistopen();
odanalyzetext(odeum, RSTRING(content)->ptr, asis_words, norm_words);
// go through words and add them
count = cblistnum(asis_words);
for(i = 0; i < count; i++) {
asis = cblistval(asis_words, i, &asis_len);
norm = cblistval(norm_words, i, &norm_len);
// only add words that normalize to some content
oddocaddword(oddoc, norm, asis);
}
cblistclose(asis_words);
cblistclose(norm_words);
return self;
}
/**
* call-seq:
* index.query(query) -> [[id,score], ... ]
*
* An implementation of a basic query language for Odeum. The query language
* allows boolean expressions of search terms and '&', '|', '!' with parenthesis
* as sub-expressions. The '!' operator implements NOTAND so that you can say,
* "this AND NOT that" using "this ! that". Consecutive words are assumed to
* have an implicit '&' between them.
*
* An example expression is: "Zed & shaw ! (frank blank)". The (frank blank)
* part actually is interpreted as (frank & blank).
*
* It returns the same ResultSet as Index.search does.
*/
VALUE Index_query(VALUE self, VALUE word) {
CBLIST *errors = NULL;
int num_returned = 0;
ODEUM *odeum = NULL;
DATA_GET(self, ODEUM, odeum);
REQUIRE_TYPE(word, T_STRING);
errors = cblistopen();
ODPAIR *pairs = odquery(odeum, RSTRING(word)->ptr, &num_returned, errors);
if(pairs == NULL) {
// nothing found
rb_raise(rb_eStandardError, "Query failure.");
}
return ResultSet_create(pairs, num_returned, errors);
}
CBLIST *objtocblist(JNIEnv *env, jobject obj){
jclass list, it;
jmethodID midit, midhn, midn;
jobject itobj, eobj;
jboolean icelem;
CBLIST *tlist;
const char *telem;
assert(obj);
tlist = cblistopen();
list = (*env)->GetObjectClass(env, obj);
midit = (*env)->GetMethodID(env, list, "iterator", "()L" CLSITERATOR ";");
itobj = (*env)->CallObjectMethod(env, obj, midit);
it = (*env)->GetObjectClass(env, itobj);
midhn = (*env)->GetMethodID(env, it, "hasNext", "()Z");
midn = (*env)->GetMethodID(env, it, "next", "()L" CLSOBJECT ";");
while((*env)->CallBooleanMethod(env, itobj, midhn)){
eobj = (*env)->CallObjectMethod(env, itobj, midn);
if(!isinstanceof(env, eobj, CLSSTRING)) continue;
if(!(telem = (*env)->GetStringUTFChars(env, eobj, &icelem))) continue;
cblistpush(tlist, telem, -1);
if(icelem == JNI_TRUE) (*env)->ReleaseStringUTFChars(env, eobj, telem);
}
return tlist;
}
