示例#1
0
static void
test_hl_selection(TestBatchRunner *runner) {
    Schema *schema = Schema_new();
    StandardTokenizer *tokenizer = StandardTokenizer_new();
    FullTextType *plain_type = FullTextType_new((Analyzer*)tokenizer);
    FullTextType_Set_Highlightable(plain_type, true);
    String *content = (String*)SSTR_WRAP_UTF8("content", 7);
    Schema_Spec_Field(schema, content, (FieldType*)plain_type);
    DECREF(plain_type);
    DECREF(tokenizer);

    RAMFolder *folder = RAMFolder_new(NULL);
    Indexer *indexer = Indexer_new(schema, (Obj*)folder, NULL, 0);

    static char test_string[] =
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla NNN bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla MMM bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. "
        "bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla bla. ";
    Doc *doc = Doc_new(NULL, 0);
    String *string = (String *)SSTR_WRAP_UTF8(test_string, strlen(test_string));
    Doc_Store(doc, content, (Obj*)string);
    Indexer_Add_Doc(indexer, doc, 1.0f);
    DECREF(doc);

    Indexer_Commit(indexer);
    DECREF(indexer);

    Searcher *searcher = (Searcher*)IxSearcher_new((Obj*)folder);
    Obj *query = (Obj*)SSTR_WRAP_UTF8("NNN MMM", 7);
    Highlighter *highlighter = Highlighter_new(searcher, query, content, 200);
    Hits *hits = Searcher_Hits(searcher, query, 0, 10, NULL);
    HitDoc *hit = Hits_Next(hits);
    String *excerpt = Highlighter_Create_Excerpt(highlighter, hit);
    String *mmm = (String*)SSTR_WRAP_UTF8("MMM", 3);
    String *nnn = (String*)SSTR_WRAP_UTF8("NNN", 3);
    TEST_TRUE(runner, Str_Find(excerpt, mmm) >= 0 || Str_Find(excerpt, nnn) >= 0,
              "Sentence boundary algo doesn't chop terms");

    DECREF(excerpt);
    DECREF(hit);
    DECREF(hits);
    DECREF(highlighter);
    DECREF(searcher);
    DECREF(folder);
    DECREF(schema);
}
示例#2
0
static int64_t
S_find(String *string, String *substring) {
    StringIterator *iter = Str_Find(string, substring);
    if (iter == NULL) { return -1; }
    size_t tick = StrIter_Recede(iter, SIZE_MAX);
    DECREF(iter);
    return (int64_t)tick;
}