|
321f2f5…
|
leo
|
1 |
"""Tests for robust JSON parsing from LLM responses.""" |
|
321f2f5…
|
leo
|
2 |
|
|
321f2f5…
|
leo
|
3 |
from video_processor.utils.json_parsing import parse_json_from_response |
|
321f2f5…
|
leo
|
4 |
|
|
321f2f5…
|
leo
|
5 |
|
|
321f2f5…
|
leo
|
6 |
class TestParseJsonFromResponse: |
|
321f2f5…
|
leo
|
7 |
def test_direct_dict(self): |
|
321f2f5…
|
leo
|
8 |
assert parse_json_from_response('{"key": "value"}') == {"key": "value"} |
|
321f2f5…
|
leo
|
9 |
|
|
321f2f5…
|
leo
|
10 |
def test_direct_array(self): |
|
829e24a…
|
leo
|
11 |
assert parse_json_from_response("[1, 2, 3]") == [1, 2, 3] |
|
321f2f5…
|
leo
|
12 |
|
|
321f2f5…
|
leo
|
13 |
def test_markdown_fenced_json(self): |
|
321f2f5…
|
leo
|
14 |
text = '```json\n{"key": "value"}\n```' |
|
321f2f5…
|
leo
|
15 |
assert parse_json_from_response(text) == {"key": "value"} |
|
321f2f5…
|
leo
|
16 |
|
|
321f2f5…
|
leo
|
17 |
def test_markdown_fenced_no_lang(self): |
|
829e24a…
|
leo
|
18 |
text = "```\n[1, 2]\n```" |
|
321f2f5…
|
leo
|
19 |
assert parse_json_from_response(text) == [1, 2] |
|
321f2f5…
|
leo
|
20 |
|
|
321f2f5…
|
leo
|
21 |
def test_json_embedded_in_text(self): |
|
321f2f5…
|
leo
|
22 |
text = 'Here is the result:\n{"name": "test", "value": 42}\nEnd of result.' |
|
321f2f5…
|
leo
|
23 |
result = parse_json_from_response(text) |
|
321f2f5…
|
leo
|
24 |
assert result == {"name": "test", "value": 42} |
|
321f2f5…
|
leo
|
25 |
|
|
321f2f5…
|
leo
|
26 |
def test_array_embedded_in_text(self): |
|
321f2f5…
|
leo
|
27 |
text = 'The entities are:\n[{"name": "Alice"}, {"name": "Bob"}]\nThat is all.' |
|
321f2f5…
|
leo
|
28 |
result = parse_json_from_response(text) |
|
321f2f5…
|
leo
|
29 |
assert len(result) == 2 |
|
321f2f5…
|
leo
|
30 |
assert result[0]["name"] == "Alice" |
|
321f2f5…
|
leo
|
31 |
|
|
321f2f5…
|
leo
|
32 |
def test_nested_json(self): |
|
321f2f5…
|
leo
|
33 |
text = '{"outer": {"inner": [1, 2, 3]}}' |
|
321f2f5…
|
leo
|
34 |
result = parse_json_from_response(text) |
|
321f2f5…
|
leo
|
35 |
assert result["outer"]["inner"] == [1, 2, 3] |
|
321f2f5…
|
leo
|
36 |
|
|
321f2f5…
|
leo
|
37 |
def test_empty_string(self): |
|
321f2f5…
|
leo
|
38 |
assert parse_json_from_response("") is None |
|
321f2f5…
|
leo
|
39 |
|
|
321f2f5…
|
leo
|
40 |
def test_none_input(self): |
|
321f2f5…
|
leo
|
41 |
assert parse_json_from_response(None) is None |
|
321f2f5…
|
leo
|
42 |
|
|
321f2f5…
|
leo
|
43 |
def test_whitespace_only(self): |
|
321f2f5…
|
leo
|
44 |
assert parse_json_from_response(" \n ") is None |
|
321f2f5…
|
leo
|
45 |
|
|
321f2f5…
|
leo
|
46 |
def test_no_json(self): |
|
321f2f5…
|
leo
|
47 |
assert parse_json_from_response("This is just plain text.") is None |
|
321f2f5…
|
leo
|
48 |
|
|
321f2f5…
|
leo
|
49 |
def test_invalid_json(self): |
|
321f2f5…
|
leo
|
50 |
assert parse_json_from_response("{invalid json}") is None |
|
321f2f5…
|
leo
|
51 |
|
|
321f2f5…
|
leo
|
52 |
def test_multiple_json_objects_picks_first(self): |
|
321f2f5…
|
leo
|
53 |
text = '{"a": 1} and {"b": 2}' |
|
321f2f5…
|
leo
|
54 |
result = parse_json_from_response(text) |
|
321f2f5…
|
leo
|
55 |
assert result == {"a": 1} |
|
321f2f5…
|
leo
|
56 |
|
|
321f2f5…
|
leo
|
57 |
def test_complex_fenced(self): |
|
321f2f5…
|
leo
|
58 |
text = """Here is the analysis: |
|
321f2f5…
|
leo
|
59 |
|
|
321f2f5…
|
leo
|
60 |
```json |
|
321f2f5…
|
leo
|
61 |
[ |
|
321f2f5…
|
leo
|
62 |
{"point": "Architecture uses microservices", "topic": "Architecture"}, |
|
321f2f5…
|
leo
|
63 |
{"point": "Deployment is automated", "topic": "DevOps"} |
|
321f2f5…
|
leo
|
64 |
] |
|
321f2f5…
|
leo
|
65 |
``` |
|
321f2f5…
|
leo
|
66 |
|
|
321f2f5…
|
leo
|
67 |
I hope this helps!""" |
|
321f2f5…
|
leo
|
68 |
result = parse_json_from_response(text) |
|
321f2f5…
|
leo
|
69 |
assert len(result) == 2 |
|
321f2f5…
|
leo
|
70 |
assert result[0]["topic"] == "Architecture" |