selfware 0.6.2

Your personal AI workshop — software you own, software that lasts
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
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271

use std::fs;

use axum::http::StatusCode;
use serde_json::{json, Value};

use selfware::config::Config;
use selfware::evolve::{EvolveServer, Graph, Node};

use crate::{get_json, post_json};

/// Build a temp project with one real Rust file and a matching graph node.
fn fixture(tokens: usize) -> (tempfile::TempDir, Graph) {
    let dir = tempfile::tempdir().unwrap();
    let src_dir = dir.path().join("src");
    fs::create_dir_all(&src_dir).unwrap();
    let body = "pub fn f() {}\n".repeat((tokens / 4).max(1));
    fs::write(src_dir.join("a.rs"), &body).unwrap();
    let mut node = Node::code("crate::a", "src/a.rs");
    node.tokens = selfware::token_count::estimate_content_tokens(&body);
    (
        dir,
        Graph {
            nodes: vec![node],
            edges: vec![],
        },
    )
}

/// POST /api/context/mode with the session header and return the JSON body.
async fn post_mode(server: &EvolveServer, mode: &str) -> Value {
    let (status, body) = post_json(server, "/api/context/mode", json!({ "mode": mode })).await;
    assert_eq!(status, StatusCode::OK);
    serde_json::from_str(&body).unwrap()
}

/// POST /api/context/mode and return the raw status + JSON body, so tests can
/// assert on non-200 responses (e.g. the over-budget 422).
async fn post_mode_raw(server: &EvolveServer, mode: &str) -> (StatusCode, Value) {
    let (status, body) = post_json(server, "/api/context/mode", json!({ "mode": mode })).await;
    (status, serde_json::from_str(&body).unwrap())
}

#[tokio::test]
async fn auto_mode_resolves_full_on_large_window_and_degrades_on_small() {
    let (dir, graph) = fixture(20_000);

    // Large window: auto resolves to a full tier and fits.
    let config = Config {
        context_length: 1_000_000,
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph.clone(), dir.path(), &config).unwrap();
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert!(["full", "full_extended"].contains(&json["mode"].as_str().unwrap()));
    assert_eq!(json["requested_mode"].as_str().unwrap(), "auto");
    assert!(json["fits_context_window"].as_bool().unwrap());

    // 8k window: auto degrades below full, still reports coherently.
    let config = Config {
        context_length: 8_192,
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert!(["map", "lite", "compact"].contains(&json["mode"].as_str().unwrap()));
    assert_eq!(json["requested_mode"].as_str().unwrap(), "auto");
}

#[tokio::test]
async fn auto_mode_reports_fit_and_pinned_mode_clears_it() {
    let (dir, graph) = fixture(20_000);
    let config = Config {
        context_length: 1_000_000,
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph.clone(), dir.path(), &config).unwrap();

    // Auto: the fit outcome rides along with the context payload.
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    let fit = &json["auto_fit"];
    let expected_budget = (0.7
        * (config.context_length - config.max_tokens.min(config.context_length / 4)) as f64)
        as usize;
    assert_eq!(
        fit["budget_tokens"].as_u64().unwrap() as usize,
        expected_budget
    );
    assert!(fit["fits"].as_bool().unwrap());
    // The resolved tier is measured, so its size matches the fixture exactly.
    let tier_tokens: usize = graph.nodes.iter().map(|n| n.tokens).sum();
    assert_eq!(
        fit["measured_tokens"].as_u64().unwrap() as usize,
        tier_tokens
    );
    assert!(fit["measured_tokens"].as_u64().unwrap() as usize <= expected_budget);

    // Pinned: auto_fit clears to null.
    let json = post_mode(&server, "lite").await;
    assert!(json["auto_fit"].is_null());
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert!(json["auto_fit"].is_null());

    // Back to auto: the fit outcome is reported again.
    let json = post_mode(&server, "auto").await;
    assert!(json["auto_fit"].is_object());
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert!(json["auto_fit"].is_object());
}

/// Temp project with two real Rust files and matching graph nodes.
/// Sized so the verbatim (full tier) payload is ~20k tokens — comfortably over
/// the usable budget of an 8k context window.
fn fixture_two_files() -> (tempfile::TempDir, Graph) {
    let dir = tempfile::tempdir().unwrap();
    let src_dir = dir.path().join("src");
    fs::create_dir_all(&src_dir).unwrap();
    let a = "pub fn a() {}\n".repeat(1_000);
    let b = "pub fn b() {}\n".repeat(2_000);
    fs::write(src_dir.join("a.rs"), &a).unwrap();
    fs::write(src_dir.join("b.rs"), &b).unwrap();
    let mut node_a = Node::code("crate::a", "src/a.rs");
    node_a.tokens = selfware::token_count::estimate_content_tokens(&a);
    let mut node_b = Node::code("crate::b", "src/b.rs");
    node_b.tokens = selfware::token_count::estimate_content_tokens(&b);
    (
        dir,
        Graph {
            nodes: vec![node_a, node_b],
            edges: vec![],
        },
    )
}

/// POST /api/context/custom with the session header and return the JSON body.
async fn post_custom(server: &EvolveServer, components: Vec<&str>) -> Value {
    let (status, body) = post_json(
        server,
        "/api/context/custom",
        json!({ "components": components }),
    )
    .await;
    assert_eq!(status, StatusCode::OK);
    serde_json::from_str(&body).unwrap()
}

#[tokio::test]
async fn custom_mode_applies_handpicked_selection_and_drops_unknown() {
    let (dir, graph) = fixture_two_files();
    let config = Config {
        context_length: 1_000_000,
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();

    let json = post_custom(&server, vec!["crate::a", "crate::bogus", "crate::b"]).await;
    assert_eq!(json["mode"].as_str().unwrap(), "custom");
    assert_eq!(json["requested_mode"].as_str().unwrap(), "custom");
    let included: Vec<&str> = json["included"]
        .as_array()
        .unwrap()
        .iter()
        .map(|id| id.as_str().unwrap())
        .collect();
    assert_eq!(included, ["crate::a", "crate::b"], "unknown ids dropped");

    // The selection sticks for subsequent context reads.
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert_eq!(json["mode"].as_str().unwrap(), "custom");
    assert_eq!(json["requested_mode"].as_str().unwrap(), "custom");
}

#[tokio::test]
async fn custom_mode_empty_selection_clears_back_to_auto() {
    let (dir, graph) = fixture_two_files();
    let config = Config {
        context_length: 1_000_000,
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();

    let json = post_custom(&server, vec!["crate::a"]).await;
    assert_eq!(json["requested_mode"].as_str().unwrap(), "custom");

    let json = post_custom(&server, vec![]).await;
    assert_eq!(json["requested_mode"].as_str().unwrap(), "auto");
    assert!(
        ["map", "lite", "compact", "full", "full_extended"]
            .contains(&json["mode"].as_str().unwrap()),
        "cleared custom re-fits to a concrete ladder tier, got {}",
        json["mode"]
    );
    assert!(json["auto_fit"].is_object());
}

#[tokio::test]
async fn invalid_context_mode_in_config_is_an_error() {
    let (dir, graph) = fixture(1_000);
    let config = Config {
        context_mode: "bogus".to_string(),
        ..Default::default()
    };
    assert!(EvolveServer::with_config(graph, dir.path(), &config).is_err());
}

#[tokio::test]
async fn post_mode_auto_refits_and_pinned_mode_sticks() {
    let (dir, graph) = fixture(20_000);
    let config = Config {
        context_length: 1_000_000,
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();

    let json = post_mode(&server, "auto").await;
    assert_eq!(json["requested_mode"].as_str().unwrap(), "auto");

    let json = post_mode(&server, "lite").await;
    assert_eq!(json["mode"].as_str().unwrap(), "lite");
    assert_eq!(json["requested_mode"].as_str().unwrap(), "lite");
}

#[tokio::test]
async fn context_json_reports_envelope_fields() {
    let (dir, graph) = fixture_two_files();
    let config = Config {
        context_length: 1_000_000,
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();
    let (status, json) = get_json(&server, "/api/context").await;
    assert_eq!(status, StatusCode::OK);
    assert!(json["envelope_hash"].is_string());
    assert!(json["envelope_tokens"].as_u64().unwrap() > 0);
    // Auto on a huge window resolves to full/full_extended: envelope tokens
    // must equal the composer's own estimate (verbatim source).
    assert!(
        json["envelope_tokens"].as_u64().unwrap()
            >= json["production"]["tokens"].as_u64().unwrap() / 2
    );
}

#[tokio::test]
async fn pinned_over_budget_mode_is_rejected_with_422() {
    let (dir, graph) = fixture_two_files();
    let config = Config {
        context_length: 8_192, // usable budget = 0.7 * (8192 - 2048) = 4300 tokens
        context_mode: "auto".to_string(),
        ..Default::default()
    };
    let server = EvolveServer::with_config(graph, dir.path(), &config).unwrap();

    let (status, body) = post_mode_raw(&server, "full").await;
    assert_eq!(status, StatusCode::UNPROCESSABLE_ENTITY);
    assert_eq!(body["error"], "context_over_budget");
    assert!(body["measured_tokens"].as_u64().unwrap() > body["budget_tokens"].as_u64().unwrap());

    // Auto remains accepted on the same server.
    let (status, _) = post_mode_raw(&server, "auto").await;
    assert_eq!(status, StatusCode::OK);
}