harn-lint 0.10.141

Linter for the Harn programming language
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137

use super::*;

const RULE: &str = "unbounded-native-decision-state";

const POLICY: &str = r#"{
    backend: "native_decision",
    provider: "typesafe",
    model: "jev-1",
    effort: "low",
    temperature: 0.0,
    threshold: 0.5,
    evaluation_cost_limit: 0.01,
    run_cost_limit: 0.1,
  }"#;

fn source(state_type: &str, backend: &str) -> String {
    format!(
        r#"
pipeline main(harness: Harness) {{
  const state: {state_type} = fetch_state()
  const policy = {policy}
  return harness.llm.evaluate("site.v1", state, {{}}, policy)
}}
"#,
        policy = POLICY.replace("native_decision", backend),
    )
}

#[test]
fn a_list_state_on_a_native_decision_route_is_reported() {
    let diagnostics = lint_source(&source("list<string>", "native_decision"));
    assert_eq!(count_rule(&diagnostics, RULE), 1);
}

#[test]
fn a_string_state_on_a_native_decision_route_is_reported() {
    let diagnostics = lint_source(&source("string", "native_decision"));
    assert_eq!(count_rule(&diagnostics, RULE), 1);
}

#[test]
fn a_record_carrying_an_unbounded_field_is_reported() {
    let diagnostics = lint_source(&source("{id: int, notes: list<string>}", "native_decision"));
    assert_eq!(count_rule(&diagnostics, RULE), 1);
}

#[test]
fn an_alias_of_an_unbounded_type_is_reported_through_the_alias() {
    let diagnostics = lint_source(
        r#"
type Transcript = list<string>

pipeline main(harness: Harness) {
  const state: Transcript = fetch_state()
  const policy = {
    backend: "native_decision",
    provider: "typesafe",
    model: "jev-1",
    effort: "low",
    temperature: 0.0,
    threshold: 0.5,
    evaluation_cost_limit: 0.01,
    run_cost_limit: 0.1,
  }
  return harness.llm.evaluate("site.v1", state, {}, policy)
}
"#,
    );
    assert_eq!(count_rule(&diagnostics, RULE), 1);
}

#[test]
fn a_bounded_record_is_not_reported() {
    let diagnostics = lint_source(&source(
        r#"{severity: "low" | "high", reopened: bool, age_days: int}"#,
        "native_decision",
    ));
    assert!(!has_rule(&diagnostics, RULE));
}

/// The negative control on the trigger. The same unbounded state is fine on a
/// structured-LLM route, which refuses an oversized state before dispatch. A
/// rule that reported both would be reporting the input type, not the
/// combination this warning is about.
#[test]
fn an_unbounded_state_on_a_structured_llm_route_is_not_reported() {
    let diagnostics = lint_source(&source("list<string>", "structured_llm"));
    assert!(!has_rule(&diagnostics, RULE));
}

/// Silence when the type is not visible is deliberate and stated on the rule,
/// so it is pinned rather than left to be discovered as a surprise.
#[test]
fn an_undeclared_state_is_not_reported() {
    let diagnostics = lint_source(
        r#"
pipeline main(harness: Harness) {
  const state = fetch_state()
  const policy = {
    backend: "native_decision",
    provider: "typesafe",
    model: "jev-1",
    effort: "low",
    temperature: 0.0,
    threshold: 0.5,
    evaluation_cost_limit: 0.01,
    run_cost_limit: 0.1,
  }
  return harness.llm.evaluate("site.v1", state, {}, policy)
}
"#,
    );
    assert!(!has_rule(&diagnostics, RULE));
}

#[test]
fn the_single_boolean_projection_reads_its_own_input_position() {
    let diagnostics = lint_source(
        r#"
pipeline main(harness: Harness) {
  const input: list<string> = fetch_state()
  const policy = {
    backend: "native_decision",
    provider: "typesafe",
    model: "jev-1",
    effort: "low",
    temperature: 0.0,
    threshold: 0.5,
    evaluation_cost_limit: 0.01,
    run_cost_limit: 0.1,
  }
  return harness.llm.evaluate_predicate("site.v1", "Safe?", input, policy)
}
"#,
    );
    assert_eq!(count_rule(&diagnostics, RULE), 1);
}