String*
Highlighter_Create_Excerpt_IMP(Highlighter *self, HitDoc *hit_doc) {
HighlighterIVARS *const ivars = Highlighter_IVARS(self);
String *field_val = (String*)HitDoc_Extract(hit_doc, ivars->field);
String *retval;
if (!field_val || !Obj_Is_A((Obj*)field_val, STRING)) {
retval = NULL;
}
else if (!Str_Get_Size(field_val)) {
// Empty string yields empty string.
retval = Str_new_from_trusted_utf8("", 0);
}
else {
DocVector *doc_vec
= Searcher_Fetch_Doc_Vec(ivars->searcher,
HitDoc_Get_Doc_ID(hit_doc));
VArray *maybe_spans
= Compiler_Highlight_Spans(ivars->compiler, ivars->searcher,
doc_vec, ivars->field);
VArray *score_spans = maybe_spans ? maybe_spans : VA_new(0);
VA_Sort(score_spans, NULL, NULL);
HeatMap *heat_map
= HeatMap_new(score_spans, (ivars->excerpt_length * 2) / 3);
int32_t top;
String *raw_excerpt
= Highlighter_Raw_Excerpt(self, field_val, &top, heat_map);
String *highlighted
= Highlighter_Highlight_Excerpt(self, score_spans, raw_excerpt,
top);
DECREF(raw_excerpt);
DECREF(heat_map);
DECREF(score_spans);
DECREF(doc_vec);
retval = highlighted;
}
DECREF(field_val);
return retval;
}
static void
test_Raw_Excerpt(TestBatchRunner *runner, Searcher *searcher, Obj *query) {
String *content = (String*)SSTR_WRAP_UTF8("content", 7);
Highlighter *highlighter = Highlighter_new(searcher, query, content, 6);
int32_t top;
String *raw_excerpt;
String *field_val = (String *)SSTR_WRAP_UTF8("Ook. Urk. Ick. ", 18);
Vector *spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(0, 18, 1.0f));
HeatMap *heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, "Ook.", 4),
"Raw_Excerpt at top %s", Str_Get_Ptr8(raw_excerpt));
TEST_TRUE(runner,
top == 0,
"top is 0");
DECREF(raw_excerpt);
DECREF(heat_map);
spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(6, 12, 1.0f));
heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, "Urk.", 4),
"Raw_Excerpt in middle, with 2 bounds");
TEST_TRUE(runner,
top == 6,
"top in the middle modified by Raw_Excerpt");
DECREF(raw_excerpt);
DECREF(heat_map);
field_val = (String *)SSTR_WRAP_UTF8("Ook urk ick i.", 14);
spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(12, 1, 1.0f));
heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, ELLIPSIS " i.", 6),
"Ellipsis at top");
TEST_TRUE(runner,
top == 10,
"top correct when leading ellipsis inserted");
DECREF(heat_map);
DECREF(raw_excerpt);
field_val = (String *)SSTR_WRAP_UTF8("Urk. Iz no good.", 17);
spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(6, 2, 1.0f));
heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, "Iz no" ELLIPSIS, 8),
"Ellipsis at end");
TEST_TRUE(runner,
top == 6,
"top trimmed");
DECREF(heat_map);
DECREF(raw_excerpt);
// Words longer than excerpt len
field_val = (String *)SSTR_WRAP_UTF8("abc/def/ghi/jkl/mno", 19);
spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(0, 3, 1.0f));
heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, "abc/d" ELLIPSIS, 8),
"Long word at top");
DECREF(heat_map);
DECREF(raw_excerpt);
spans = Vec_new(1);
Vec_Push(spans, (Obj*)Span_new(8, 3, 1.0f));
heat_map = HeatMap_new(spans, 133);
DECREF(spans);
raw_excerpt = Highlighter_Raw_Excerpt(highlighter, field_val, &top,
heat_map);
TEST_TRUE(runner,
Str_Equals_Utf8(raw_excerpt, ELLIPSIS " f/g" ELLIPSIS, 10),
"Long word in middle");
DECREF(heat_map);
DECREF(raw_excerpt);
DECREF(highlighter);
}
