@kujolang/paperclip 0.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +18 -0
- package/LICENSE +21 -0
- package/README.md +116 -0
- package/SECURITY.md +35 -0
- package/VERSION +1 -0
- package/bundled/components/changebucket/LICENSE +9 -0
- package/bundled/components/changebucket/changebucket.kujo +12 -0
- package/bundled/components/changebucket/src/analyze.kujo +240 -0
- package/bundled/components/changebucket/src/budget.kujo +92 -0
- package/bundled/components/changebucket/src/classify.kujo +221 -0
- package/bundled/components/changebucket/src/cli.kujo +324 -0
- package/bundled/components/changebucket/src/diffsrc.kujo +252 -0
- package/bundled/components/changebucket/src/render.kujo +346 -0
- package/bundled/components/changebucket/src/util.kujo +62 -0
- package/bundled/components/context/LICENSE +9 -0
- package/bundled/components/context/scent.kujo +3250 -0
- package/bundled/components/failure-evidence/LICENSE +9 -0
- package/bundled/components/failure-evidence/casefile.kujo +2061 -0
- package/bundled/components/patchbrief/LICENSE +9 -0
- package/bundled/components/patchbrief/patchbrief.kujo +153 -0
- package/bundled/components/patchbrief/schemas/patchbrief-handoff.schema.json +38 -0
- package/bundled/components/patchbrief/schemas/patchbrief-summary.schema.json +46 -0
- package/bundled/components/patchbrief/src/common.kujo +212 -0
- package/bundled/components/patchbrief/src/git.kujo +250 -0
- package/bundled/components/patchbrief/src/handoff.kujo +145 -0
- package/bundled/components/patchbrief/src/suggest_tests.kujo +137 -0
- package/bundled/components/patchbrief/src/summarize.kujo +309 -0
- package/bundled/kujo-components.lock.json +134 -0
- package/dist/manifest.js +14696 -0
- package/dist/ui/index.js +179 -0
- package/dist/worker.js +30536 -0
- package/docs/ARCHITECTURE.md +28 -0
- package/docs/CATALOG_SUBMISSION.md +25 -0
- package/docs/COMPATIBILITY.md +21 -0
- package/docs/CONFIGURATION.md +33 -0
- package/docs/INSTALLATION.md +53 -0
- package/docs/OPERATIONS.md +60 -0
- package/docs/README.md +18 -0
- package/docs/RELEASE_READINESS.md +46 -0
- package/docs/THREAT_MODEL.md +26 -0
- package/docs/TROUBLESHOOTING.md +12 -0
- package/docs/USAGE.md +95 -0
- package/examples/agent-workflow.md +20 -0
- package/package.json +71 -0
- package/schemas/changebucket-analysis.schema.json +25 -0
- package/schemas/context-pack.schema.json +23 -0
- package/schemas/failure-evidence.schema.json +18 -0
- package/schemas/review-pack.schema.json +19 -0
- package/skills/scoped-repository-context/SKILL.md +9 -0
|
@@ -0,0 +1,3250 @@
|
|
|
1
|
+
#!/usr/bin/env kujo
|
|
2
|
+
|
|
3
|
+
VERSION := "1.0.0"
|
|
4
|
+
|
|
5
|
+
func usage_text() {
|
|
6
|
+
return "Usage:\n scent pack --task <text> [--out <path>] [--budget <n>] [--target codex|claude|deepseek|generic] [--include <path>] [--exclude <path>] [--changed] [--staged] [--unstaged] [--max-files <n>] [--max-file-bytes <n>] [--format md|json|both] [--verbose] [--json] [--dry-run]"
|
|
7
|
+
}
|
|
8
|
+
|
|
9
|
+
func version_text() {
|
|
10
|
+
return "Scent " + VERSION
|
|
11
|
+
}
|
|
12
|
+
|
|
13
|
+
func is_help_arg(value) {
|
|
14
|
+
return value == "help" || value == "--help" || value == "-h"
|
|
15
|
+
}
|
|
16
|
+
|
|
17
|
+
func is_version_arg(value) {
|
|
18
|
+
return value == "version" || value == "--version"
|
|
19
|
+
}
|
|
20
|
+
|
|
21
|
+
func defaults() {
|
|
22
|
+
return {
|
|
23
|
+
"task": "",
|
|
24
|
+
"out": null,
|
|
25
|
+
"budget": 12000,
|
|
26
|
+
"target": "codex",
|
|
27
|
+
"include": [],
|
|
28
|
+
"exclude": [],
|
|
29
|
+
"changed": false,
|
|
30
|
+
"staged": false,
|
|
31
|
+
"unstaged": false,
|
|
32
|
+
"max_files": 40,
|
|
33
|
+
"max_file_bytes": 50000,
|
|
34
|
+
"format": "both",
|
|
35
|
+
"verbose": false,
|
|
36
|
+
"json_stdout": false,
|
|
37
|
+
"dry_run": false
|
|
38
|
+
}
|
|
39
|
+
}
|
|
40
|
+
|
|
41
|
+
func ok(value) {
|
|
42
|
+
return {"ok": true, "value": value}
|
|
43
|
+
}
|
|
44
|
+
|
|
45
|
+
func err(message, code) {
|
|
46
|
+
return {"ok": false, "error": message, "exit_code": code}
|
|
47
|
+
}
|
|
48
|
+
|
|
49
|
+
func to_int_safe(value, fallback) {
|
|
50
|
+
try {
|
|
51
|
+
return to_int(value)
|
|
52
|
+
} except e {
|
|
53
|
+
return fallback
|
|
54
|
+
}
|
|
55
|
+
}
|
|
56
|
+
|
|
57
|
+
func collect_repeated_flag_values(argv, flag) {
|
|
58
|
+
values := []
|
|
59
|
+
i := 0
|
|
60
|
+
while i < len(argv) {
|
|
61
|
+
arg := to_string(argv[i])
|
|
62
|
+
if arg == flag {
|
|
63
|
+
if i + 1 < len(argv) {
|
|
64
|
+
next := to_string(argv[i + 1])
|
|
65
|
+
if !starts_with(next, "-") {
|
|
66
|
+
values = push(values, next)
|
|
67
|
+
i = i + 2
|
|
68
|
+
continue
|
|
69
|
+
}
|
|
70
|
+
}
|
|
71
|
+
} else if starts_with(arg, flag + "=") {
|
|
72
|
+
values = push(values, substring(arg, len(flag) + 1, len(arg)))
|
|
73
|
+
}
|
|
74
|
+
i = i + 1
|
|
75
|
+
}
|
|
76
|
+
return values
|
|
77
|
+
}
|
|
78
|
+
|
|
79
|
+
func shell_quote(arg) {
|
|
80
|
+
s := to_string(arg)
|
|
81
|
+
return "'" + replace(s, "'", "'\"'\"'") + "'"
|
|
82
|
+
}
|
|
83
|
+
|
|
84
|
+
func parse_cli(argv) {
|
|
85
|
+
if len(argv) == 0 {
|
|
86
|
+
return err(usage_text(), 2)
|
|
87
|
+
}
|
|
88
|
+
|
|
89
|
+
parser := arg_parser()
|
|
90
|
+
parser := parser.add_argument("--task", "type", "string", "required", true)
|
|
91
|
+
parser := parser.add_argument("--out", "type", "string", "default", "")
|
|
92
|
+
parser := parser.add_argument("--budget", "type", "int", "default", "12000")
|
|
93
|
+
parser := parser.add_argument("--target", "type", "string", "default", "codex")
|
|
94
|
+
parser := parser.add_argument("--include", "type", "string", "default", "")
|
|
95
|
+
parser := parser.add_argument("--exclude", "type", "string", "default", "")
|
|
96
|
+
parser := parser.add_argument("--changed", "type", "bool")
|
|
97
|
+
parser := parser.add_argument("--staged", "type", "bool")
|
|
98
|
+
parser := parser.add_argument("--unstaged", "type", "bool")
|
|
99
|
+
parser := parser.add_argument("--max-files", "type", "int", "default", "40")
|
|
100
|
+
parser := parser.add_argument("--max-file-bytes", "type", "int", "default", "50000")
|
|
101
|
+
parser := parser.add_argument("--format", "type", "string", "default", "both")
|
|
102
|
+
parser := parser.add_argument("--verbose", "type", "bool")
|
|
103
|
+
parser := parser.add_argument("--json", "type", "bool")
|
|
104
|
+
parser := parser.add_argument("--dry-run", "type", "bool")
|
|
105
|
+
|
|
106
|
+
parsed := null
|
|
107
|
+
try {
|
|
108
|
+
parsed = parser.parse()
|
|
109
|
+
} except parse_error {
|
|
110
|
+
return err(usage_text() + "\n\n" + to_string(parse_error), 2)
|
|
111
|
+
}
|
|
112
|
+
|
|
113
|
+
positionals := []
|
|
114
|
+
if has_key(parsed, "_positional") && parsed["_positional"] != null {
|
|
115
|
+
positionals = parsed["_positional"]
|
|
116
|
+
}
|
|
117
|
+
|
|
118
|
+
if len(positionals) > 1 {
|
|
119
|
+
return err("expected a single subcommand: pack", 2)
|
|
120
|
+
}
|
|
121
|
+
if len(positionals) == 0 {
|
|
122
|
+
return err("expected subcommand: pack\n\n" + usage_text(), 2)
|
|
123
|
+
}
|
|
124
|
+
if len(positionals) == 1 {
|
|
125
|
+
cmd := positionals[0]
|
|
126
|
+
if is_help_arg(cmd) {
|
|
127
|
+
return err(usage_text(), 2)
|
|
128
|
+
}
|
|
129
|
+
if cmd != "pack" {
|
|
130
|
+
return err("unknown command: " + cmd + "\n\n" + usage_text(), 2)
|
|
131
|
+
}
|
|
132
|
+
}
|
|
133
|
+
|
|
134
|
+
opts := defaults()
|
|
135
|
+
opts["task"] = to_string(parsed["task"])
|
|
136
|
+
out_raw := to_string(parsed["out"])
|
|
137
|
+
if trim(out_raw) == "" {
|
|
138
|
+
opts["out"] = null
|
|
139
|
+
} else {
|
|
140
|
+
opts["out"] = out_raw
|
|
141
|
+
}
|
|
142
|
+
opts["budget"] = to_int(parsed["budget"])
|
|
143
|
+
opts["target"] = to_string(parsed["target"])
|
|
144
|
+
opts["changed"] = parsed["changed"]
|
|
145
|
+
opts["staged"] = parsed["staged"]
|
|
146
|
+
opts["unstaged"] = parsed["unstaged"]
|
|
147
|
+
opts["max_files"] = to_int(parsed["max-files"])
|
|
148
|
+
opts["max_file_bytes"] = to_int(parsed["max-file-bytes"])
|
|
149
|
+
opts["format"] = to_string(parsed["format"])
|
|
150
|
+
opts["verbose"] = parsed["verbose"]
|
|
151
|
+
opts["json_stdout"] = parsed["json"]
|
|
152
|
+
opts["dry_run"] = parsed["dry-run"]
|
|
153
|
+
|
|
154
|
+
repeated_includes := collect_repeated_flag_values(argv, "--include")
|
|
155
|
+
if len(repeated_includes) > 0 {
|
|
156
|
+
opts["include"] = repeated_includes
|
|
157
|
+
} else {
|
|
158
|
+
include_raw := to_string(parsed["include"])
|
|
159
|
+
if trim(include_raw) != "" {
|
|
160
|
+
opts["include"] = push(opts["include"], include_raw)
|
|
161
|
+
}
|
|
162
|
+
}
|
|
163
|
+
|
|
164
|
+
repeated_excludes := collect_repeated_flag_values(argv, "--exclude")
|
|
165
|
+
if len(repeated_excludes) > 0 {
|
|
166
|
+
opts["exclude"] = repeated_excludes
|
|
167
|
+
} else {
|
|
168
|
+
exclude_raw := to_string(parsed["exclude"])
|
|
169
|
+
if trim(exclude_raw) != "" {
|
|
170
|
+
opts["exclude"] = push(opts["exclude"], exclude_raw)
|
|
171
|
+
}
|
|
172
|
+
}
|
|
173
|
+
|
|
174
|
+
if trim(to_string(opts["task"])) == "" {
|
|
175
|
+
return err("--task is required", 2)
|
|
176
|
+
}
|
|
177
|
+
budget_value := to_int(opts["budget"])
|
|
178
|
+
max_files_value := to_int(opts["max_files"])
|
|
179
|
+
max_file_bytes_value := to_int(opts["max_file_bytes"])
|
|
180
|
+
opts["budget"] = budget_value
|
|
181
|
+
opts["max_files"] = max_files_value
|
|
182
|
+
opts["max_file_bytes"] = max_file_bytes_value
|
|
183
|
+
|
|
184
|
+
if budget_value <= 0 {
|
|
185
|
+
return err("--budget must be greater than 0", 2)
|
|
186
|
+
}
|
|
187
|
+
if max_files_value <= 0 {
|
|
188
|
+
return err("--max-files must be greater than 0", 2)
|
|
189
|
+
}
|
|
190
|
+
if max_file_bytes_value <= 0 {
|
|
191
|
+
return err("--max-file-bytes must be greater than 0", 2)
|
|
192
|
+
}
|
|
193
|
+
|
|
194
|
+
allowed_targets := {"codex": 1, "claude": 1, "deepseek": 1, "generic": 1}
|
|
195
|
+
if has_key(allowed_targets, opts["target"]) == 0 {
|
|
196
|
+
return err("--target must be one of: codex, claude, deepseek, generic", 2)
|
|
197
|
+
}
|
|
198
|
+
|
|
199
|
+
allowed_formats := {"md": 1, "json": 1, "both": 1}
|
|
200
|
+
if has_key(allowed_formats, opts["format"]) == 0 {
|
|
201
|
+
return err("--format must be one of: md, json, both", 2)
|
|
202
|
+
}
|
|
203
|
+
|
|
204
|
+
return ok(opts)
|
|
205
|
+
}
|
|
206
|
+
|
|
207
|
+
func is_absolute(path) {
|
|
208
|
+
return starts_with(path, "/")
|
|
209
|
+
}
|
|
210
|
+
|
|
211
|
+
func resolve_path(base, path) {
|
|
212
|
+
if is_absolute(path) {
|
|
213
|
+
return path
|
|
214
|
+
}
|
|
215
|
+
if path == "." {
|
|
216
|
+
return base
|
|
217
|
+
}
|
|
218
|
+
return path_join(base, path)
|
|
219
|
+
}
|
|
220
|
+
|
|
221
|
+
func path_rel(root, full) {
|
|
222
|
+
prefix := root + "/"
|
|
223
|
+
if starts_with(full, prefix) {
|
|
224
|
+
return substring(full, len(prefix), len(full))
|
|
225
|
+
}
|
|
226
|
+
if full == root {
|
|
227
|
+
return "."
|
|
228
|
+
}
|
|
229
|
+
return full
|
|
230
|
+
}
|
|
231
|
+
|
|
232
|
+
func has_parent_path_segment(path) {
|
|
233
|
+
parts := split(to_string(path), "/")
|
|
234
|
+
i := 0
|
|
235
|
+
while i < len(parts) {
|
|
236
|
+
if parts[i] == ".." {
|
|
237
|
+
return true
|
|
238
|
+
}
|
|
239
|
+
i = i + 1
|
|
240
|
+
}
|
|
241
|
+
return false
|
|
242
|
+
}
|
|
243
|
+
|
|
244
|
+
func strip_trailing_slashes(path) {
|
|
245
|
+
out := to_string(path)
|
|
246
|
+
while len(out) > 1 && ends_with(out, "/") {
|
|
247
|
+
out = substring(out, 0, len(out) - 1)
|
|
248
|
+
}
|
|
249
|
+
return out
|
|
250
|
+
}
|
|
251
|
+
|
|
252
|
+
func normalize_selector(repo_root, path) {
|
|
253
|
+
raw := strip_trailing_slashes(trim(to_string(path)))
|
|
254
|
+
if is_absolute(raw) {
|
|
255
|
+
return path_rel(repo_root, raw)
|
|
256
|
+
}
|
|
257
|
+
while starts_with(raw, "./") {
|
|
258
|
+
raw = substring(raw, 2, len(raw))
|
|
259
|
+
}
|
|
260
|
+
raw = strip_trailing_slashes(raw)
|
|
261
|
+
if raw == "" || raw == "." {
|
|
262
|
+
return "."
|
|
263
|
+
}
|
|
264
|
+
return raw
|
|
265
|
+
}
|
|
266
|
+
|
|
267
|
+
func path_has_symlink_segment(path) {
|
|
268
|
+
full := to_string(path)
|
|
269
|
+
parts := split(full, "/")
|
|
270
|
+
current := "/"
|
|
271
|
+
i := 0
|
|
272
|
+
while i < len(parts) {
|
|
273
|
+
part := parts[i]
|
|
274
|
+
i = i + 1
|
|
275
|
+
if part == "" {
|
|
276
|
+
continue
|
|
277
|
+
}
|
|
278
|
+
current = path_join(current, part)
|
|
279
|
+
if path_exists(current) && path_is_symlink(current) {
|
|
280
|
+
return true
|
|
281
|
+
}
|
|
282
|
+
}
|
|
283
|
+
return false
|
|
284
|
+
}
|
|
285
|
+
|
|
286
|
+
func validate_scoped_paths(repo_root, paths, flag_name) {
|
|
287
|
+
normalized := []
|
|
288
|
+
i := 0
|
|
289
|
+
while i < len(paths) {
|
|
290
|
+
raw := trim(to_string(paths[i]))
|
|
291
|
+
if raw == "" {
|
|
292
|
+
i = i + 1
|
|
293
|
+
continue
|
|
294
|
+
}
|
|
295
|
+
if has_parent_path_segment(raw) {
|
|
296
|
+
return err(flag_name + " path cannot contain '..': " + raw, 2)
|
|
297
|
+
}
|
|
298
|
+
full := resolve_path(repo_root, raw)
|
|
299
|
+
root_prefix := repo_root + "/"
|
|
300
|
+
if full != repo_root && !starts_with(full, root_prefix) {
|
|
301
|
+
return err(flag_name + " path must stay inside the repository: " + raw, 2)
|
|
302
|
+
}
|
|
303
|
+
if path_has_symlink_segment(full) {
|
|
304
|
+
return err(flag_name + " path cannot contain a symlink: " + raw, 2)
|
|
305
|
+
}
|
|
306
|
+
normalized = push(normalized, normalize_selector(repo_root, raw))
|
|
307
|
+
i = i + 1
|
|
308
|
+
}
|
|
309
|
+
return ok(normalized)
|
|
310
|
+
}
|
|
311
|
+
|
|
312
|
+
func detect_repo_root(start) {
|
|
313
|
+
current := start
|
|
314
|
+
previous := ""
|
|
315
|
+
|
|
316
|
+
while current != previous {
|
|
317
|
+
git_dir := path_join(current, ".git")
|
|
318
|
+
if path_exists(git_dir) {
|
|
319
|
+
return current
|
|
320
|
+
}
|
|
321
|
+
previous = current
|
|
322
|
+
current = dirname(current)
|
|
323
|
+
}
|
|
324
|
+
|
|
325
|
+
return start
|
|
326
|
+
}
|
|
327
|
+
|
|
328
|
+
func estimate_tokens(text) {
|
|
329
|
+
chars := len(text)
|
|
330
|
+
return (chars + 3) / 4
|
|
331
|
+
}
|
|
332
|
+
|
|
333
|
+
func tokenize_task(task) {
|
|
334
|
+
tokens := []
|
|
335
|
+
seen := {}
|
|
336
|
+
|
|
337
|
+
parts := regex_split(lower(task), "[^a-z0-9_-]+")
|
|
338
|
+
i := 0
|
|
339
|
+
while i < len(parts) {
|
|
340
|
+
part := parts[i]
|
|
341
|
+
p := trim(part)
|
|
342
|
+
if len(p) > 2 && has_key(seen, p) == 0 {
|
|
343
|
+
seen[p] = 1
|
|
344
|
+
tokens = push(tokens, p)
|
|
345
|
+
}
|
|
346
|
+
i = i + 1
|
|
347
|
+
}
|
|
348
|
+
|
|
349
|
+
return tokens
|
|
350
|
+
}
|
|
351
|
+
|
|
352
|
+
func discover_instruction_files(repo_root) {
|
|
353
|
+
candidates := [
|
|
354
|
+
"AGENTS.md",
|
|
355
|
+
"AGENTS.override.md",
|
|
356
|
+
"CLAUDE.md",
|
|
357
|
+
"README.md",
|
|
358
|
+
"CONTRIBUTING.md",
|
|
359
|
+
"docs/README.md"
|
|
360
|
+
]
|
|
361
|
+
found := []
|
|
362
|
+
|
|
363
|
+
i := 0
|
|
364
|
+
while i < len(candidates) {
|
|
365
|
+
rel := candidates[i]
|
|
366
|
+
full := path_join(repo_root, rel)
|
|
367
|
+
if path_exists(full) && path_is_file(full) {
|
|
368
|
+
found = push(found, full)
|
|
369
|
+
}
|
|
370
|
+
i = i + 1
|
|
371
|
+
}
|
|
372
|
+
|
|
373
|
+
return found
|
|
374
|
+
}
|
|
375
|
+
|
|
376
|
+
func run_process(args) {
|
|
377
|
+
parts := []
|
|
378
|
+
i := 0
|
|
379
|
+
while i < len(args) {
|
|
380
|
+
parts = push(parts, shell_quote(args[i]))
|
|
381
|
+
i = i + 1
|
|
382
|
+
}
|
|
383
|
+
cmd := join(parts, " ")
|
|
384
|
+
|
|
385
|
+
try {
|
|
386
|
+
return execute_status(cmd)
|
|
387
|
+
} except e {
|
|
388
|
+
return {
|
|
389
|
+
"success": false,
|
|
390
|
+
"exitcode": 127,
|
|
391
|
+
"stdout": "",
|
|
392
|
+
"stderr": to_string(e),
|
|
393
|
+
"timed_out": false,
|
|
394
|
+
"stdout_truncated": false,
|
|
395
|
+
"stderr_truncated": false
|
|
396
|
+
}
|
|
397
|
+
}
|
|
398
|
+
}
|
|
399
|
+
|
|
400
|
+
func canonical_path_for_scope(path) {
|
|
401
|
+
raw := strip_trailing_slashes(to_string(path))
|
|
402
|
+
if path_exists(raw) {
|
|
403
|
+
resolved := run_process(["realpath", raw])
|
|
404
|
+
if resolved.success {
|
|
405
|
+
return strip_trailing_slashes(trim(to_string(resolved.stdout)))
|
|
406
|
+
}
|
|
407
|
+
}
|
|
408
|
+
|
|
409
|
+
parent := dirname(raw)
|
|
410
|
+
if path_exists(parent) {
|
|
411
|
+
resolved_parent := run_process(["realpath", parent])
|
|
412
|
+
if resolved_parent.success {
|
|
413
|
+
return path_join(strip_trailing_slashes(trim(to_string(resolved_parent.stdout))), basename(raw))
|
|
414
|
+
}
|
|
415
|
+
}
|
|
416
|
+
return raw
|
|
417
|
+
}
|
|
418
|
+
|
|
419
|
+
func split_lines(text) {
|
|
420
|
+
lines := split(text, "\n")
|
|
421
|
+
out := []
|
|
422
|
+
i := 0
|
|
423
|
+
while i < len(lines) {
|
|
424
|
+
line := lines[i]
|
|
425
|
+
t := trim(line)
|
|
426
|
+
if len(t) > 0 {
|
|
427
|
+
out = push(out, t)
|
|
428
|
+
}
|
|
429
|
+
i = i + 1
|
|
430
|
+
}
|
|
431
|
+
return out
|
|
432
|
+
}
|
|
433
|
+
|
|
434
|
+
func split_nul_records(text) {
|
|
435
|
+
nul := parse_json("\"\\u0000\"")
|
|
436
|
+
records := split(to_string(text), nul)
|
|
437
|
+
out := []
|
|
438
|
+
i := 0
|
|
439
|
+
while i < len(records) {
|
|
440
|
+
record := records[i]
|
|
441
|
+
if record != "" {
|
|
442
|
+
out = push(out, record)
|
|
443
|
+
}
|
|
444
|
+
i = i + 1
|
|
445
|
+
}
|
|
446
|
+
return out
|
|
447
|
+
}
|
|
448
|
+
|
|
449
|
+
func inspect_git(repo_root) {
|
|
450
|
+
result := {
|
|
451
|
+
"available": false,
|
|
452
|
+
"branch": "unknown",
|
|
453
|
+
"commit": "unknown",
|
|
454
|
+
"dirty": false,
|
|
455
|
+
"staged_files": [],
|
|
456
|
+
"unstaged_files": [],
|
|
457
|
+
"untracked_files": [],
|
|
458
|
+
"changed_files": [],
|
|
459
|
+
"diff_summary": []
|
|
460
|
+
}
|
|
461
|
+
|
|
462
|
+
staged_res := run_process(["git", "-C", repo_root, "diff", "--name-only", "--cached", "-z"])
|
|
463
|
+
if !staged_res.success {
|
|
464
|
+
return result
|
|
465
|
+
}
|
|
466
|
+
|
|
467
|
+
unstaged_res := run_process(["git", "-C", repo_root, "diff", "--name-only", "-z"])
|
|
468
|
+
if !unstaged_res.success {
|
|
469
|
+
return result
|
|
470
|
+
}
|
|
471
|
+
|
|
472
|
+
untracked_res := run_process(["git", "-C", repo_root, "ls-files", "--others", "--exclude-standard", "-z"])
|
|
473
|
+
if !untracked_res.success {
|
|
474
|
+
return result
|
|
475
|
+
}
|
|
476
|
+
|
|
477
|
+
result["available"] = true
|
|
478
|
+
|
|
479
|
+
branch_res := run_process(["git", "-C", repo_root, "rev-parse", "--abbrev-ref", "HEAD"])
|
|
480
|
+
if branch_res.success {
|
|
481
|
+
branch := trim(to_string(branch_res.stdout))
|
|
482
|
+
if branch != "" {
|
|
483
|
+
result["branch"] = branch
|
|
484
|
+
}
|
|
485
|
+
}
|
|
486
|
+
|
|
487
|
+
commit_res := run_process(["git", "-C", repo_root, "rev-parse", "--short", "HEAD"])
|
|
488
|
+
if commit_res.success {
|
|
489
|
+
commit := trim(to_string(commit_res.stdout))
|
|
490
|
+
if commit != "" {
|
|
491
|
+
result["commit"] = commit
|
|
492
|
+
}
|
|
493
|
+
}
|
|
494
|
+
|
|
495
|
+
staged_seen := {}
|
|
496
|
+
unstaged_seen := {}
|
|
497
|
+
untracked_seen := {}
|
|
498
|
+
changed_seen := {}
|
|
499
|
+
|
|
500
|
+
staged_files := split_nul_records(to_string(staged_res.stdout))
|
|
501
|
+
unstaged_files := split_nul_records(to_string(unstaged_res.stdout))
|
|
502
|
+
untracked_files := split_nul_records(to_string(untracked_res.stdout))
|
|
503
|
+
|
|
504
|
+
idx := 0
|
|
505
|
+
while idx < len(staged_files) {
|
|
506
|
+
path := staged_files[idx]
|
|
507
|
+
idx = idx + 1
|
|
508
|
+
if path == "" {
|
|
509
|
+
continue
|
|
510
|
+
}
|
|
511
|
+
if has_key(staged_seen, path) == 0 {
|
|
512
|
+
staged_seen[path] = 1
|
|
513
|
+
result["staged_files"] = push(result["staged_files"], path)
|
|
514
|
+
}
|
|
515
|
+
if has_key(changed_seen, path) == 0 {
|
|
516
|
+
changed_seen[path] = 1
|
|
517
|
+
result["changed_files"] = push(result["changed_files"], path)
|
|
518
|
+
}
|
|
519
|
+
result["diff_summary"] = push(result["diff_summary"], "staged: " + path)
|
|
520
|
+
}
|
|
521
|
+
|
|
522
|
+
idx := 0
|
|
523
|
+
while idx < len(unstaged_files) {
|
|
524
|
+
path := unstaged_files[idx]
|
|
525
|
+
idx = idx + 1
|
|
526
|
+
if path == "" {
|
|
527
|
+
continue
|
|
528
|
+
}
|
|
529
|
+
if has_key(unstaged_seen, path) == 0 {
|
|
530
|
+
unstaged_seen[path] = 1
|
|
531
|
+
result["unstaged_files"] = push(result["unstaged_files"], path)
|
|
532
|
+
}
|
|
533
|
+
if has_key(changed_seen, path) == 0 {
|
|
534
|
+
changed_seen[path] = 1
|
|
535
|
+
result["changed_files"] = push(result["changed_files"], path)
|
|
536
|
+
}
|
|
537
|
+
result["diff_summary"] = push(result["diff_summary"], "unstaged: " + path)
|
|
538
|
+
}
|
|
539
|
+
|
|
540
|
+
idx = 0
|
|
541
|
+
while idx < len(untracked_files) {
|
|
542
|
+
path := untracked_files[idx]
|
|
543
|
+
idx = idx + 1
|
|
544
|
+
if path == "" {
|
|
545
|
+
continue
|
|
546
|
+
}
|
|
547
|
+
if has_key(untracked_seen, path) == 0 {
|
|
548
|
+
untracked_seen[path] = 1
|
|
549
|
+
result["untracked_files"] = push(result["untracked_files"], path)
|
|
550
|
+
}
|
|
551
|
+
if has_key(changed_seen, path) == 0 {
|
|
552
|
+
changed_seen[path] = 1
|
|
553
|
+
result["changed_files"] = push(result["changed_files"], path)
|
|
554
|
+
}
|
|
555
|
+
result["diff_summary"] = push(result["diff_summary"], "untracked: " + path)
|
|
556
|
+
}
|
|
557
|
+
|
|
558
|
+
if len(result["staged_files"]) > 0 || len(result["unstaged_files"]) > 0 || len(result["untracked_files"]) > 0 {
|
|
559
|
+
result["dirty"] = true
|
|
560
|
+
}
|
|
561
|
+
|
|
562
|
+
return result
|
|
563
|
+
}
|
|
564
|
+
|
|
565
|
+
func is_ignored_dir(name) {
|
|
566
|
+
ignored := {
|
|
567
|
+
".git": 1,
|
|
568
|
+
".scent": 1,
|
|
569
|
+
"node_modules": 1,
|
|
570
|
+
"vendor": 1,
|
|
571
|
+
"out": 1,
|
|
572
|
+
"dist": 1,
|
|
573
|
+
"build": 1,
|
|
574
|
+
"target": 1
|
|
575
|
+
}
|
|
576
|
+
return has_key(ignored, name)
|
|
577
|
+
}
|
|
578
|
+
|
|
579
|
+
func is_binary_extension(path) {
|
|
580
|
+
ext := lower(path_extension(path))
|
|
581
|
+
binary_exts := {
|
|
582
|
+
"png": 1,
|
|
583
|
+
"jpg": 1,
|
|
584
|
+
"jpeg": 1,
|
|
585
|
+
"gif": 1,
|
|
586
|
+
"webp": 1,
|
|
587
|
+
"pdf": 1,
|
|
588
|
+
"zip": 1,
|
|
589
|
+
"gz": 1,
|
|
590
|
+
"tar": 1,
|
|
591
|
+
"bz2": 1,
|
|
592
|
+
"xz": 1,
|
|
593
|
+
"mp3": 1,
|
|
594
|
+
"mp4": 1,
|
|
595
|
+
"mov": 1,
|
|
596
|
+
"avi": 1,
|
|
597
|
+
"so": 1,
|
|
598
|
+
"dll": 1,
|
|
599
|
+
"dylib": 1,
|
|
600
|
+
"exe": 1,
|
|
601
|
+
"bin": 1,
|
|
602
|
+
"db": 1,
|
|
603
|
+
"sqlite": 1,
|
|
604
|
+
"class": 1
|
|
605
|
+
}
|
|
606
|
+
return has_key(binary_exts, ext)
|
|
607
|
+
}
|
|
608
|
+
|
|
609
|
+
func should_skip_file(path) {
|
|
610
|
+
name := basename(path)
|
|
611
|
+
if starts_with(name, ".env") {
|
|
612
|
+
return true
|
|
613
|
+
}
|
|
614
|
+
return is_binary_extension(path)
|
|
615
|
+
}
|
|
616
|
+
|
|
617
|
+
func contains_nul(text) {
|
|
618
|
+
nul := parse_json("\"\\u0000\"")
|
|
619
|
+
return contains(to_string(text), nul)
|
|
620
|
+
}
|
|
621
|
+
|
|
622
|
+
func is_instruction_rel(rel) {
|
|
623
|
+
return rel == "AGENTS.md" || rel == "AGENTS.override.md" || rel == "CLAUDE.md" || rel == "README.md" || rel == "CONTRIBUTING.md" || rel == "docs/README.md"
|
|
624
|
+
}
|
|
625
|
+
|
|
626
|
+
func is_doc_rel(rel) {
|
|
627
|
+
return ends_with(lower(rel), ".md") || starts_with(lower(rel), "docs/")
|
|
628
|
+
}
|
|
629
|
+
|
|
630
|
+
func is_test_rel(rel) {
|
|
631
|
+
lower_rel := lower(rel)
|
|
632
|
+
return starts_with(lower_rel, "tests/") || contains(lower_rel, "/tests/") || ends_with(lower_rel, "_test.rs") || ends_with(lower_rel, "_test.py") || ends_with(lower_rel, ".test.ts") || ends_with(lower_rel, ".spec.ts") || ends_with(lower_rel, "_test.kujo")
|
|
633
|
+
}
|
|
634
|
+
|
|
635
|
+
func is_config_rel(rel) {
|
|
636
|
+
exact := {
|
|
637
|
+
"Cargo.toml": 1,
|
|
638
|
+
"Cargo.lock": 1,
|
|
639
|
+
"package.json": 1,
|
|
640
|
+
"pnpm-lock.yaml": 1,
|
|
641
|
+
"yarn.lock": 1,
|
|
642
|
+
"package-lock.json": 1,
|
|
643
|
+
"pyproject.toml": 1,
|
|
644
|
+
"setup.cfg": 1,
|
|
645
|
+
"setup.py": 1,
|
|
646
|
+
"Makefile": 1,
|
|
647
|
+
"justfile": 1,
|
|
648
|
+
"Justfile": 1,
|
|
649
|
+
"tsconfig.json": 1,
|
|
650
|
+
"kujo.toml": 1
|
|
651
|
+
}
|
|
652
|
+
if has_key(exact, rel) {
|
|
653
|
+
return true
|
|
654
|
+
}
|
|
655
|
+
lower_rel := lower(rel)
|
|
656
|
+
return ends_with(lower_rel, ".toml") || ends_with(lower_rel, ".yaml") || ends_with(lower_rel, ".yml") || ends_with(lower_rel, ".json")
|
|
657
|
+
}
|
|
658
|
+
|
|
659
|
+
func path_is_excluded(full, exclude_abs) {
|
|
660
|
+
i := 0
|
|
661
|
+
while i < len(exclude_abs) {
|
|
662
|
+
ex := exclude_abs[i]
|
|
663
|
+
if full == ex || starts_with(full, ex + "/") {
|
|
664
|
+
return true
|
|
665
|
+
}
|
|
666
|
+
i = i + 1
|
|
667
|
+
}
|
|
668
|
+
return false
|
|
669
|
+
}
|
|
670
|
+
|
|
671
|
+
func read_text_file(path) {
|
|
672
|
+
try {
|
|
673
|
+
content := read_file(path)
|
|
674
|
+
if contains_nul(content) {
|
|
675
|
+
return {"ok": false, "error": "binary file"}
|
|
676
|
+
}
|
|
677
|
+
return {"ok": true, "content": content}
|
|
678
|
+
} except e {
|
|
679
|
+
return {"ok": false, "error": to_string(e)}
|
|
680
|
+
}
|
|
681
|
+
}
|
|
682
|
+
|
|
683
|
+
func walk_collect(repo_root, current, max_candidates, visited_dirs, depth) {
|
|
684
|
+
files := []
|
|
685
|
+
if max_candidates <= 0 {
|
|
686
|
+
return files
|
|
687
|
+
}
|
|
688
|
+
if depth > 20 {
|
|
689
|
+
return files
|
|
690
|
+
}
|
|
691
|
+
|
|
692
|
+
if has_key(visited_dirs, current) == 1 {
|
|
693
|
+
return files
|
|
694
|
+
}
|
|
695
|
+
visited_dirs[current] = 1
|
|
696
|
+
|
|
697
|
+
entries := []
|
|
698
|
+
try {
|
|
699
|
+
entries = list_dir(current)
|
|
700
|
+
} except e {
|
|
701
|
+
return files
|
|
702
|
+
}
|
|
703
|
+
entries = sort(entries)
|
|
704
|
+
|
|
705
|
+
idx := 0
|
|
706
|
+
while idx < len(entries) {
|
|
707
|
+
name := entries[idx]
|
|
708
|
+
idx = idx + 1
|
|
709
|
+
|
|
710
|
+
if len(files) >= max_candidates {
|
|
711
|
+
break
|
|
712
|
+
}
|
|
713
|
+
|
|
714
|
+
if name == "." {
|
|
715
|
+
continue
|
|
716
|
+
}
|
|
717
|
+
if name == ".." {
|
|
718
|
+
continue
|
|
719
|
+
}
|
|
720
|
+
|
|
721
|
+
if is_ignored_dir(name) {
|
|
722
|
+
continue
|
|
723
|
+
}
|
|
724
|
+
|
|
725
|
+
full := path_join(current, name)
|
|
726
|
+
if path_is_symlink(full) {
|
|
727
|
+
continue
|
|
728
|
+
}
|
|
729
|
+
if path_is_dir(full) {
|
|
730
|
+
remaining := max_candidates - len(files)
|
|
731
|
+
child_files := walk_collect(repo_root, full, remaining, visited_dirs, depth + 1)
|
|
732
|
+
child_idx := 0
|
|
733
|
+
while child_idx < len(child_files) {
|
|
734
|
+
files = push(files, child_files[child_idx])
|
|
735
|
+
child_idx = child_idx + 1
|
|
736
|
+
if len(files) >= max_candidates {
|
|
737
|
+
break
|
|
738
|
+
}
|
|
739
|
+
}
|
|
740
|
+
} else if path_is_file(full) {
|
|
741
|
+
files = push(files, full)
|
|
742
|
+
}
|
|
743
|
+
}
|
|
744
|
+
|
|
745
|
+
return files
|
|
746
|
+
}
|
|
747
|
+
|
|
748
|
+
func collect_candidates(repo_root, git_info, instruction_files, includes, excludes, task_terms) {
|
|
749
|
+
visited_dirs := {}
|
|
750
|
+
all_paths := walk_collect(repo_root, repo_root, 2000, visited_dirs, 0)
|
|
751
|
+
|
|
752
|
+
i := 0
|
|
753
|
+
while i < len(includes) {
|
|
754
|
+
inc := includes[i]
|
|
755
|
+
inc_abs := resolve_path(repo_root, inc)
|
|
756
|
+
if path_exists(inc_abs) {
|
|
757
|
+
if path_is_file(inc_abs) {
|
|
758
|
+
all_paths = push(all_paths, inc_abs)
|
|
759
|
+
} else if path_is_dir(inc_abs) {
|
|
760
|
+
include_visited := {}
|
|
761
|
+
include_paths := walk_collect(repo_root, inc_abs, 2000, include_visited, 0)
|
|
762
|
+
add_idx := 0
|
|
763
|
+
while add_idx < len(include_paths) {
|
|
764
|
+
all_paths = push(all_paths, include_paths[add_idx])
|
|
765
|
+
add_idx = add_idx + 1
|
|
766
|
+
}
|
|
767
|
+
}
|
|
768
|
+
}
|
|
769
|
+
i = i + 1
|
|
770
|
+
}
|
|
771
|
+
|
|
772
|
+
instruction_rel := {}
|
|
773
|
+
i = 0
|
|
774
|
+
while i < len(instruction_files) {
|
|
775
|
+
inst := instruction_files[i]
|
|
776
|
+
instruction_rel[path_rel(repo_root, inst)] = 1
|
|
777
|
+
i = i + 1
|
|
778
|
+
}
|
|
779
|
+
|
|
780
|
+
changed_map := {}
|
|
781
|
+
i = 0
|
|
782
|
+
while i < len(git_info["changed_files"]) {
|
|
783
|
+
p := git_info["changed_files"][i]
|
|
784
|
+
changed_map[p] = 1
|
|
785
|
+
i = i + 1
|
|
786
|
+
}
|
|
787
|
+
staged_map := {}
|
|
788
|
+
i = 0
|
|
789
|
+
while i < len(git_info["staged_files"]) {
|
|
790
|
+
p := git_info["staged_files"][i]
|
|
791
|
+
staged_map[p] = 1
|
|
792
|
+
i = i + 1
|
|
793
|
+
}
|
|
794
|
+
unstaged_map := {}
|
|
795
|
+
i = 0
|
|
796
|
+
while i < len(git_info["unstaged_files"]) {
|
|
797
|
+
p := git_info["unstaged_files"][i]
|
|
798
|
+
unstaged_map[p] = 1
|
|
799
|
+
i = i + 1
|
|
800
|
+
}
|
|
801
|
+
|
|
802
|
+
exclude_abs := []
|
|
803
|
+
i = 0
|
|
804
|
+
while i < len(excludes) {
|
|
805
|
+
ex := excludes[i]
|
|
806
|
+
exclude_abs = push(exclude_abs, resolve_path(repo_root, ex))
|
|
807
|
+
i = i + 1
|
|
808
|
+
}
|
|
809
|
+
|
|
810
|
+
seen := {}
|
|
811
|
+
candidates := []
|
|
812
|
+
|
|
813
|
+
i = 0
|
|
814
|
+
while i < len(all_paths) {
|
|
815
|
+
full := all_paths[i]
|
|
816
|
+
i = i + 1
|
|
817
|
+
|
|
818
|
+
if !path_is_file(full) {
|
|
819
|
+
continue
|
|
820
|
+
}
|
|
821
|
+
if should_skip_file(full) {
|
|
822
|
+
continue
|
|
823
|
+
}
|
|
824
|
+
if path_is_excluded(full, exclude_abs) {
|
|
825
|
+
continue
|
|
826
|
+
}
|
|
827
|
+
|
|
828
|
+
rel := path_rel(repo_root, full)
|
|
829
|
+
if has_key(seen, rel) {
|
|
830
|
+
continue
|
|
831
|
+
}
|
|
832
|
+
seen[rel] = 1
|
|
833
|
+
|
|
834
|
+
matched_terms := []
|
|
835
|
+
rel_lower := lower(rel)
|
|
836
|
+
j := 0
|
|
837
|
+
while j < len(task_terms) {
|
|
838
|
+
term := task_terms[j]
|
|
839
|
+
if contains(rel_lower, term) {
|
|
840
|
+
matched_terms = push(matched_terms, term)
|
|
841
|
+
}
|
|
842
|
+
j = j + 1
|
|
843
|
+
}
|
|
844
|
+
|
|
845
|
+
size := to_int_safe(file_size(full), 0)
|
|
846
|
+
|
|
847
|
+
candidates = push(candidates, {
|
|
848
|
+
"path": full,
|
|
849
|
+
"rel_path": rel,
|
|
850
|
+
"size_bytes": size,
|
|
851
|
+
"is_changed": has_key(changed_map, rel),
|
|
852
|
+
"is_staged": has_key(staged_map, rel),
|
|
853
|
+
"is_unstaged": has_key(unstaged_map, rel),
|
|
854
|
+
"is_instruction": has_key(instruction_rel, rel),
|
|
855
|
+
"is_doc": is_doc_rel(rel),
|
|
856
|
+
"is_test": is_test_rel(rel),
|
|
857
|
+
"is_config": is_config_rel(rel),
|
|
858
|
+
"matched_terms": matched_terms,
|
|
859
|
+
"selected": false,
|
|
860
|
+
"selection_reason": ""
|
|
861
|
+
})
|
|
862
|
+
}
|
|
863
|
+
|
|
864
|
+
return candidates
|
|
865
|
+
}
|
|
866
|
+
|
|
867
|
+
func explicit_include_match(rel, includes, repo_root) {
|
|
868
|
+
i := 0
|
|
869
|
+
while i < len(includes) {
|
|
870
|
+
inc := includes[i]
|
|
871
|
+
inc_rel := inc
|
|
872
|
+
if is_absolute(inc_rel) {
|
|
873
|
+
inc_rel = path_rel(repo_root, inc_rel)
|
|
874
|
+
}
|
|
875
|
+
|
|
876
|
+
if inc_rel == "." || rel == inc_rel || starts_with(rel, inc_rel + "/") {
|
|
877
|
+
return true
|
|
878
|
+
}
|
|
879
|
+
i = i + 1
|
|
880
|
+
}
|
|
881
|
+
return false
|
|
882
|
+
}
|
|
883
|
+
|
|
884
|
+
func score_candidate(c, opts, repo_root) {
|
|
885
|
+
score := 0
|
|
886
|
+
reasons := []
|
|
887
|
+
|
|
888
|
+
if c["is_instruction"] {
|
|
889
|
+
score = score + 120
|
|
890
|
+
reasons = push(reasons, "instruction file")
|
|
891
|
+
}
|
|
892
|
+
|
|
893
|
+
if explicit_include_match(c["rel_path"], opts["include"], repo_root) {
|
|
894
|
+
score = score + 200
|
|
895
|
+
reasons = push(reasons, "explicit include")
|
|
896
|
+
}
|
|
897
|
+
|
|
898
|
+
if opts["changed"] && c["is_changed"] {
|
|
899
|
+
score = score + 110
|
|
900
|
+
reasons = push(reasons, "changed file")
|
|
901
|
+
}
|
|
902
|
+
|
|
903
|
+
if opts["staged"] && c["is_staged"] {
|
|
904
|
+
score = score + 95
|
|
905
|
+
reasons = push(reasons, "staged file")
|
|
906
|
+
}
|
|
907
|
+
|
|
908
|
+
if opts["unstaged"] && c["is_unstaged"] {
|
|
909
|
+
score = score + 90
|
|
910
|
+
reasons = push(reasons, "unstaged file")
|
|
911
|
+
}
|
|
912
|
+
|
|
913
|
+
if len(c["matched_terms"]) > 0 {
|
|
914
|
+
score = score + (len(c["matched_terms"]) * 20)
|
|
915
|
+
reasons = push(reasons, "task-term match")
|
|
916
|
+
}
|
|
917
|
+
|
|
918
|
+
if c["is_config"] {
|
|
919
|
+
score = score + 12
|
|
920
|
+
reasons = push(reasons, "config/build file")
|
|
921
|
+
}
|
|
922
|
+
|
|
923
|
+
if c["is_doc"] {
|
|
924
|
+
score = score + 10
|
|
925
|
+
reasons = push(reasons, "docs/readme")
|
|
926
|
+
}
|
|
927
|
+
|
|
928
|
+
if c["is_test"] {
|
|
929
|
+
score = score + 15
|
|
930
|
+
reasons = push(reasons, "test file")
|
|
931
|
+
}
|
|
932
|
+
|
|
933
|
+
reason := ""
|
|
934
|
+
if len(reasons) > 0 {
|
|
935
|
+
reason = join(reasons, ", ")
|
|
936
|
+
} else {
|
|
937
|
+
reason = "did not meet selection threshold"
|
|
938
|
+
}
|
|
939
|
+
|
|
940
|
+
return {"score": score, "reason": reason}
|
|
941
|
+
}
|
|
942
|
+
|
|
943
|
+
func sort_scored(scored) {
|
|
944
|
+
n := len(scored)
|
|
945
|
+
i := 0
|
|
946
|
+
while i < n {
|
|
947
|
+
j := i + 1
|
|
948
|
+
while j < n {
|
|
949
|
+
left := scored[i]
|
|
950
|
+
right := scored[j]
|
|
951
|
+
swap := false
|
|
952
|
+
if right["score"] > left["score"] {
|
|
953
|
+
swap = true
|
|
954
|
+
} else if right["score"] == left["score"] && right["candidate"]["rel_path"] < left["candidate"]["rel_path"] {
|
|
955
|
+
swap = true
|
|
956
|
+
}
|
|
957
|
+
|
|
958
|
+
if swap {
|
|
959
|
+
tmp := scored[i]
|
|
960
|
+
scored[i] = scored[j]
|
|
961
|
+
scored[j] = tmp
|
|
962
|
+
}
|
|
963
|
+
j = j + 1
|
|
964
|
+
}
|
|
965
|
+
i = i + 1
|
|
966
|
+
}
|
|
967
|
+
return scored
|
|
968
|
+
}
|
|
969
|
+
|
|
970
|
+
func file_stem(path) {
|
|
971
|
+
name := basename(path)
|
|
972
|
+
if contains(name, ".") {
|
|
973
|
+
parts := split(name, ".")
|
|
974
|
+
if len(parts) > 1 {
|
|
975
|
+
return parts[0]
|
|
976
|
+
}
|
|
977
|
+
}
|
|
978
|
+
return name
|
|
979
|
+
}
|
|
980
|
+
|
|
981
|
+
func looks_like_pair(source_rel, test_rel) {
|
|
982
|
+
lhs := lower(file_stem(source_rel))
|
|
983
|
+
rhs := lower(file_stem(test_rel))
|
|
984
|
+
if lhs == rhs {
|
|
985
|
+
return true
|
|
986
|
+
}
|
|
987
|
+
if contains(rhs, lhs) || contains(lhs, rhs) {
|
|
988
|
+
return true
|
|
989
|
+
}
|
|
990
|
+
if contains(test_rel, "tests") {
|
|
991
|
+
lhs2 := replace(lhs, "_test", "")
|
|
992
|
+
rhs2 := replace(rhs, "_test", "")
|
|
993
|
+
if contains(rhs, lhs2) || contains(lhs, rhs2) {
|
|
994
|
+
return true
|
|
995
|
+
}
|
|
996
|
+
}
|
|
997
|
+
return false
|
|
998
|
+
}
|
|
999
|
+
|
|
1000
|
+
func select_candidates(opts, repo_root, candidates) {
|
|
1001
|
+
scored := []
|
|
1002
|
+
i := 0
|
|
1003
|
+
while i < len(candidates) {
|
|
1004
|
+
c := candidates[i]
|
|
1005
|
+
sr := score_candidate(c, opts, repo_root)
|
|
1006
|
+
if sr["score"] > 0 {
|
|
1007
|
+
scored = push(scored, {"score": sr["score"], "reason": sr["reason"], "candidate": c})
|
|
1008
|
+
}
|
|
1009
|
+
i = i + 1
|
|
1010
|
+
}
|
|
1011
|
+
|
|
1012
|
+
scored = sort_scored(scored)
|
|
1013
|
+
|
|
1014
|
+
selected := []
|
|
1015
|
+
seen := {}
|
|
1016
|
+
max_files := to_int(opts["max_files"])
|
|
1017
|
+
|
|
1018
|
+
i = 0
|
|
1019
|
+
while i < len(scored) {
|
|
1020
|
+
if to_int(len(selected)) >= to_int(max_files) {
|
|
1021
|
+
break
|
|
1022
|
+
}
|
|
1023
|
+
s := scored[i]
|
|
1024
|
+
c := s["candidate"]
|
|
1025
|
+
rel := c["rel_path"]
|
|
1026
|
+
if has_key(seen, rel) {
|
|
1027
|
+
i = i + 1
|
|
1028
|
+
continue
|
|
1029
|
+
}
|
|
1030
|
+
seen[rel] = 1
|
|
1031
|
+
c["selected"] = true
|
|
1032
|
+
c["selection_reason"] = s["reason"]
|
|
1033
|
+
selected = push(selected, c)
|
|
1034
|
+
i = i + 1
|
|
1035
|
+
}
|
|
1036
|
+
|
|
1037
|
+
sources := []
|
|
1038
|
+
i = 0
|
|
1039
|
+
while i < len(selected) {
|
|
1040
|
+
c := selected[i]
|
|
1041
|
+
if !c["is_test"] {
|
|
1042
|
+
sources = push(sources, c["rel_path"])
|
|
1043
|
+
}
|
|
1044
|
+
i = i + 1
|
|
1045
|
+
}
|
|
1046
|
+
|
|
1047
|
+
i = 0
|
|
1048
|
+
while i < len(candidates) {
|
|
1049
|
+
if to_int(len(selected)) >= to_int(max_files) {
|
|
1050
|
+
break
|
|
1051
|
+
}
|
|
1052
|
+
c := candidates[i]
|
|
1053
|
+
if !c["is_test"] {
|
|
1054
|
+
i = i + 1
|
|
1055
|
+
continue
|
|
1056
|
+
}
|
|
1057
|
+
if has_key(seen, c["rel_path"]) {
|
|
1058
|
+
i = i + 1
|
|
1059
|
+
continue
|
|
1060
|
+
}
|
|
1061
|
+
|
|
1062
|
+
matched := false
|
|
1063
|
+
j := 0
|
|
1064
|
+
while j < len(sources) {
|
|
1065
|
+
src := sources[j]
|
|
1066
|
+
if looks_like_pair(src, c["rel_path"]) {
|
|
1067
|
+
matched = true
|
|
1068
|
+
break
|
|
1069
|
+
}
|
|
1070
|
+
j = j + 1
|
|
1071
|
+
}
|
|
1072
|
+
|
|
1073
|
+
if matched {
|
|
1074
|
+
seen[c["rel_path"]] = 1
|
|
1075
|
+
c["selected"] = true
|
|
1076
|
+
c["selection_reason"] = "nearby test for selected source"
|
|
1077
|
+
selected = push(selected, c)
|
|
1078
|
+
}
|
|
1079
|
+
i = i + 1
|
|
1080
|
+
}
|
|
1081
|
+
|
|
1082
|
+
if len(selected) == 0 {
|
|
1083
|
+
i = 0
|
|
1084
|
+
while i < len(candidates) {
|
|
1085
|
+
if to_int(len(selected)) >= to_int(max_files) || to_int(len(selected)) >= 3 {
|
|
1086
|
+
break
|
|
1087
|
+
}
|
|
1088
|
+
c := candidates[i]
|
|
1089
|
+
rel := c["rel_path"]
|
|
1090
|
+
if has_key(seen, rel) {
|
|
1091
|
+
i = i + 1
|
|
1092
|
+
continue
|
|
1093
|
+
}
|
|
1094
|
+
seen[rel] = 1
|
|
1095
|
+
c["selected"] = true
|
|
1096
|
+
c["selection_reason"] = "fallback baseline selection"
|
|
1097
|
+
selected = push(selected, c)
|
|
1098
|
+
i = i + 1
|
|
1099
|
+
}
|
|
1100
|
+
}
|
|
1101
|
+
|
|
1102
|
+
return selected
|
|
1103
|
+
}
|
|
1104
|
+
|
|
1105
|
+
func utf8_byte_len(content) {
|
|
1106
|
+
encoded := encode_base64(to_string(content))
|
|
1107
|
+
padding := 0
|
|
1108
|
+
if ends_with(encoded, "==") {
|
|
1109
|
+
padding = 2
|
|
1110
|
+
} else if ends_with(encoded, "=") {
|
|
1111
|
+
padding = 1
|
|
1112
|
+
}
|
|
1113
|
+
return (len(encoded) / 4) * 3 - padding
|
|
1114
|
+
}
|
|
1115
|
+
|
|
1116
|
+
func clip_text(content, limit_bytes) {
|
|
1117
|
+
if utf8_byte_len(content) <= limit_bytes {
|
|
1118
|
+
return {"content": content, "truncated": false}
|
|
1119
|
+
}
|
|
1120
|
+
|
|
1121
|
+
low := 0
|
|
1122
|
+
high := len(content)
|
|
1123
|
+
while low < high {
|
|
1124
|
+
middle := (low + high + 1) / 2
|
|
1125
|
+
prefix := substring(content, 0, middle)
|
|
1126
|
+
if utf8_byte_len(prefix) <= limit_bytes {
|
|
1127
|
+
low = middle
|
|
1128
|
+
} else {
|
|
1129
|
+
high = middle - 1
|
|
1130
|
+
}
|
|
1131
|
+
}
|
|
1132
|
+
return {"content": substring(content, 0, low), "truncated": true}
|
|
1133
|
+
}
|
|
1134
|
+
|
|
1135
|
+
func normalize_credential_key(raw_key) {
|
|
1136
|
+
key := trim(lower(to_string(raw_key)))
|
|
1137
|
+
prefixes := ["export ", "let ", "const ", "var "]
|
|
1138
|
+
prefix_idx := 0
|
|
1139
|
+
while prefix_idx < len(prefixes) {
|
|
1140
|
+
prefix := prefixes[prefix_idx]
|
|
1141
|
+
if starts_with(key, prefix) {
|
|
1142
|
+
key = trim(substring(key, len(prefix), len(key)))
|
|
1143
|
+
}
|
|
1144
|
+
prefix_idx = prefix_idx + 1
|
|
1145
|
+
}
|
|
1146
|
+
|
|
1147
|
+
if len(key) > 1 && ((starts_with(key, "\"") && ends_with(key, "\"")) || (starts_with(key, "'") && ends_with(key, "'"))) {
|
|
1148
|
+
key = substring(key, 1, len(key) - 1)
|
|
1149
|
+
}
|
|
1150
|
+
if key == "" || contains(key, " ") || contains(key, "\t") {
|
|
1151
|
+
return ""
|
|
1152
|
+
}
|
|
1153
|
+
|
|
1154
|
+
if !regex_match(key, "^[a-z0-9_.-]+$") {
|
|
1155
|
+
return ""
|
|
1156
|
+
}
|
|
1157
|
+
return key
|
|
1158
|
+
}
|
|
1159
|
+
|
|
1160
|
+
func contains_credential_key_term(text) {
|
|
1161
|
+
terms := ["api_key", "apikey", "access_token", "refresh_token", "session_token", "password", "passwd", "secret", "token", "cookie", "webhook"]
|
|
1162
|
+
i := 0
|
|
1163
|
+
lower_text := lower(to_string(text))
|
|
1164
|
+
while i < len(terms) {
|
|
1165
|
+
if contains(lower_text, terms[i]) {
|
|
1166
|
+
return true
|
|
1167
|
+
}
|
|
1168
|
+
i = i + 1
|
|
1169
|
+
}
|
|
1170
|
+
return false
|
|
1171
|
+
}
|
|
1172
|
+
|
|
1173
|
+
func credential_key_matches(key, term) {
|
|
1174
|
+
return regex_match(key, "(^|[._-])" + term + "($|[._-])")
|
|
1175
|
+
}
|
|
1176
|
+
|
|
1177
|
+
func credential_assignment(line) {
|
|
1178
|
+
separators := ["=", ":"]
|
|
1179
|
+
separator_idx := 0
|
|
1180
|
+
while separator_idx < len(separators) {
|
|
1181
|
+
separator := separators[separator_idx]
|
|
1182
|
+
if contains(line, separator) {
|
|
1183
|
+
parts := split(line, separator)
|
|
1184
|
+
if contains_credential_key_term(parts[0]) {
|
|
1185
|
+
key := normalize_credential_key(parts[0])
|
|
1186
|
+
if key != "" {
|
|
1187
|
+
return {"ok": true, "key": key, "prefix": parts[0] + separator, "separator": separator}
|
|
1188
|
+
}
|
|
1189
|
+
}
|
|
1190
|
+
}
|
|
1191
|
+
separator_idx = separator_idx + 1
|
|
1192
|
+
}
|
|
1193
|
+
return {"ok": false, "key": "", "prefix": "", "separator": ""}
|
|
1194
|
+
}
|
|
1195
|
+
|
|
1196
|
+
func redact_text(file, content) {
|
|
1197
|
+
redactions := []
|
|
1198
|
+
text := to_string(content)
|
|
1199
|
+
lines := split(text, "\n")
|
|
1200
|
+
out := []
|
|
1201
|
+
in_private_key := false
|
|
1202
|
+
|
|
1203
|
+
idx := 0
|
|
1204
|
+
while idx < len(lines) {
|
|
1205
|
+
line := lines[idx]
|
|
1206
|
+
line_no := idx + 1
|
|
1207
|
+
lower_line := lower(line)
|
|
1208
|
+
|
|
1209
|
+
if contains(line, "-----BEGIN") && contains(line, "PRIVATE KEY-----") {
|
|
1210
|
+
in_private_key = true
|
|
1211
|
+
out = push(out, "[REDACTED:PRIVATE_KEY]")
|
|
1212
|
+
redactions = push(redactions, {"file": file, "line": line_no, "redaction_type": "private_key", "placeholder": "[REDACTED:PRIVATE_KEY]"})
|
|
1213
|
+
idx = idx + 1
|
|
1214
|
+
continue
|
|
1215
|
+
}
|
|
1216
|
+
|
|
1217
|
+
if in_private_key {
|
|
1218
|
+
if contains(line, "-----END") && contains(line, "PRIVATE KEY-----") {
|
|
1219
|
+
in_private_key = false
|
|
1220
|
+
}
|
|
1221
|
+
idx = idx + 1
|
|
1222
|
+
continue
|
|
1223
|
+
}
|
|
1224
|
+
|
|
1225
|
+
if contains(lower_line, "authorization:") {
|
|
1226
|
+
out = push(out, "authorization: [REDACTED:TOKEN]")
|
|
1227
|
+
redactions = push(redactions, {"file": file, "line": line_no, "redaction_type": "authorization_header", "placeholder": "[REDACTED:TOKEN]"})
|
|
1228
|
+
idx = idx + 1
|
|
1229
|
+
continue
|
|
1230
|
+
}
|
|
1231
|
+
|
|
1232
|
+
secret_line := false
|
|
1233
|
+
placeholder := ""
|
|
1234
|
+
redaction_type := ""
|
|
1235
|
+
assignment := credential_assignment(line)
|
|
1236
|
+
|
|
1237
|
+
patterns := [
|
|
1238
|
+
["api_key", "[REDACTED:API_KEY]", "api_key"],
|
|
1239
|
+
["apikey", "[REDACTED:API_KEY]", "api_key"],
|
|
1240
|
+
["access_token", "[REDACTED:TOKEN]", "access_token"],
|
|
1241
|
+
["refresh_token", "[REDACTED:TOKEN]", "refresh_token"],
|
|
1242
|
+
["session_token", "[REDACTED:TOKEN]", "session_token"],
|
|
1243
|
+
["password", "[REDACTED:PASSWORD]", "password"],
|
|
1244
|
+
["passwd", "[REDACTED:PASSWORD]", "password"],
|
|
1245
|
+
["secret", "[REDACTED:SECRET]", "secret"],
|
|
1246
|
+
["token", "[REDACTED:TOKEN]", "token"],
|
|
1247
|
+
["cookie", "[REDACTED:TOKEN]", "cookie"],
|
|
1248
|
+
["webhook", "[REDACTED:SECRET]", "webhook"]
|
|
1249
|
+
]
|
|
1250
|
+
|
|
1251
|
+
p_idx := 0
|
|
1252
|
+
while p_idx < len(patterns) {
|
|
1253
|
+
p := patterns[p_idx]
|
|
1254
|
+
if assignment["ok"] && credential_key_matches(assignment["key"], p[0]) {
|
|
1255
|
+
secret_line = true
|
|
1256
|
+
placeholder = p[1]
|
|
1257
|
+
redaction_type = p[2]
|
|
1258
|
+
break
|
|
1259
|
+
}
|
|
1260
|
+
p_idx = p_idx + 1
|
|
1261
|
+
}
|
|
1262
|
+
|
|
1263
|
+
if secret_line {
|
|
1264
|
+
separator_space := ""
|
|
1265
|
+
if assignment["separator"] == ":" {
|
|
1266
|
+
separator_space = " "
|
|
1267
|
+
}
|
|
1268
|
+
out = push(out, assignment["prefix"] + separator_space + placeholder)
|
|
1269
|
+
redactions = push(redactions, {"file": file, "line": line_no, "redaction_type": redaction_type, "placeholder": placeholder})
|
|
1270
|
+
idx = idx + 1
|
|
1271
|
+
continue
|
|
1272
|
+
}
|
|
1273
|
+
|
|
1274
|
+
token_patterns := [
|
|
1275
|
+
["sk-[A-Za-z0-9_-]{20,}", "[REDACTED:TOKEN]", "openai_token"],
|
|
1276
|
+
["ghp_[A-Za-z0-9_]{20,}", "[REDACTED:TOKEN]", "github_token"],
|
|
1277
|
+
["github_pat_[A-Za-z0-9_]{20,}", "[REDACTED:TOKEN]", "github_token"],
|
|
1278
|
+
["AKIA[0-9A-Z]{16}", "[REDACTED:ACCESS_KEY]", "aws_access_key"],
|
|
1279
|
+
["sk_live_[A-Za-z0-9]{16,}", "[REDACTED:TOKEN]", "stripe_token"]
|
|
1280
|
+
]
|
|
1281
|
+
token_line := line
|
|
1282
|
+
token_found := false
|
|
1283
|
+
token_idx := 0
|
|
1284
|
+
while token_idx < len(token_patterns) {
|
|
1285
|
+
tp := token_patterns[token_idx]
|
|
1286
|
+
if regex_match(token_line, tp[0]) {
|
|
1287
|
+
token_line = regex_replace(token_line, tp[0], tp[1])
|
|
1288
|
+
redactions = push(redactions, {"file": file, "line": line_no, "redaction_type": tp[2], "placeholder": tp[1]})
|
|
1289
|
+
token_found = true
|
|
1290
|
+
}
|
|
1291
|
+
token_idx = token_idx + 1
|
|
1292
|
+
}
|
|
1293
|
+
if token_found {
|
|
1294
|
+
out = push(out, token_line)
|
|
1295
|
+
idx = idx + 1
|
|
1296
|
+
continue
|
|
1297
|
+
}
|
|
1298
|
+
|
|
1299
|
+
if regex_match(line, "[A-Za-z0-9_-]{8,}[.][A-Za-z0-9_-]{8,}[.][A-Za-z0-9_-]{8,}") {
|
|
1300
|
+
redacted_line := regex_replace(line, "[A-Za-z0-9_-]{8,}[.][A-Za-z0-9_-]{8,}[.][A-Za-z0-9_-]{8,}", "[REDACTED:TOKEN]")
|
|
1301
|
+
out = push(out, redacted_line)
|
|
1302
|
+
redactions = push(redactions, {"file": file, "line": line_no, "redaction_type": "jwt", "placeholder": "[REDACTED:TOKEN]"})
|
|
1303
|
+
idx = idx + 1
|
|
1304
|
+
continue
|
|
1305
|
+
}
|
|
1306
|
+
|
|
1307
|
+
out = push(out, line)
|
|
1308
|
+
idx = idx + 1
|
|
1309
|
+
}
|
|
1310
|
+
|
|
1311
|
+
return {"content": join(out, "\n"), "redactions": redactions}
|
|
1312
|
+
}
|
|
1313
|
+
|
|
1314
|
+
func reserved_tokens(task) {
|
|
1315
|
+
return 1400 + estimate_tokens(task)
|
|
1316
|
+
}
|
|
1317
|
+
|
|
1318
|
+
func apply_budget(budget, reserved, selected_files) {
|
|
1319
|
+
budget_for_files := budget - reserved
|
|
1320
|
+
if budget_for_files < 0 {
|
|
1321
|
+
budget_for_files = 0
|
|
1322
|
+
}
|
|
1323
|
+
|
|
1324
|
+
manifest := []
|
|
1325
|
+
included := []
|
|
1326
|
+
used := 0
|
|
1327
|
+
|
|
1328
|
+
i := 0
|
|
1329
|
+
while i < len(selected_files) {
|
|
1330
|
+
item := selected_files[i]
|
|
1331
|
+
if used + item["estimated_tokens"] <= budget_for_files {
|
|
1332
|
+
decision := "include"
|
|
1333
|
+
if item["truncated"] {
|
|
1334
|
+
decision = "truncate"
|
|
1335
|
+
}
|
|
1336
|
+
manifest = push(manifest, {
|
|
1337
|
+
"path": item["path"],
|
|
1338
|
+
"type": "file",
|
|
1339
|
+
"decision": decision,
|
|
1340
|
+
"reason": item["reason"],
|
|
1341
|
+
"score": item["score"],
|
|
1342
|
+
"size_bytes": item["size_bytes"],
|
|
1343
|
+
"estimated_tokens": item["estimated_tokens"]
|
|
1344
|
+
})
|
|
1345
|
+
used = used + item["estimated_tokens"]
|
|
1346
|
+
included = push(included, item)
|
|
1347
|
+
i = i + 1
|
|
1348
|
+
continue
|
|
1349
|
+
}
|
|
1350
|
+
|
|
1351
|
+
remaining := budget_for_files - used
|
|
1352
|
+
if remaining > 120 {
|
|
1353
|
+
char_budget := remaining * 4
|
|
1354
|
+
if char_budget < 0 {
|
|
1355
|
+
char_budget = 0
|
|
1356
|
+
}
|
|
1357
|
+
item["content"] = substring(item["content"], 0, char_budget)
|
|
1358
|
+
item["truncated"] = true
|
|
1359
|
+
item["estimated_tokens"] = estimate_tokens(item["content"])
|
|
1360
|
+
item["content_hash"] = sha256(item["content"])
|
|
1361
|
+
|
|
1362
|
+
if item["estimated_tokens"] <= remaining {
|
|
1363
|
+
used = used + item["estimated_tokens"]
|
|
1364
|
+
manifest = push(manifest, {
|
|
1365
|
+
"path": item["path"],
|
|
1366
|
+
"type": "file",
|
|
1367
|
+
"decision": "truncate",
|
|
1368
|
+
"reason": "truncated to fit budget; " + item["reason"],
|
|
1369
|
+
"score": item["score"],
|
|
1370
|
+
"size_bytes": item["size_bytes"],
|
|
1371
|
+
"estimated_tokens": item["estimated_tokens"]
|
|
1372
|
+
})
|
|
1373
|
+
included = push(included, item)
|
|
1374
|
+
i = i + 1
|
|
1375
|
+
continue
|
|
1376
|
+
}
|
|
1377
|
+
}
|
|
1378
|
+
|
|
1379
|
+
manifest = push(manifest, {
|
|
1380
|
+
"path": item["path"],
|
|
1381
|
+
"type": "file",
|
|
1382
|
+
"decision": "exclude",
|
|
1383
|
+
"reason": "excluded due to budget; " + item["reason"],
|
|
1384
|
+
"score": item["score"],
|
|
1385
|
+
"size_bytes": item["size_bytes"],
|
|
1386
|
+
"estimated_tokens": item["estimated_tokens"]
|
|
1387
|
+
})
|
|
1388
|
+
i = i + 1
|
|
1389
|
+
}
|
|
1390
|
+
|
|
1391
|
+
return {"files": included, "manifest": manifest}
|
|
1392
|
+
}
|
|
1393
|
+
|
|
1394
|
+
func discover_commands(repo_root) {
|
|
1395
|
+
commands := []
|
|
1396
|
+
|
|
1397
|
+
cargo_path := path_join(repo_root, "Cargo.toml")
|
|
1398
|
+
if path_exists(cargo_path) {
|
|
1399
|
+
cargo_read := read_text_file(cargo_path)
|
|
1400
|
+
cargo_valid := false
|
|
1401
|
+
if cargo_read["ok"] {
|
|
1402
|
+
try {
|
|
1403
|
+
cargo_doc := parse_toml(cargo_read["content"])
|
|
1404
|
+
cargo_valid = type(cargo_doc) == "dict"
|
|
1405
|
+
} except cargo_parse_error {
|
|
1406
|
+
cargo_valid = false
|
|
1407
|
+
}
|
|
1408
|
+
}
|
|
1409
|
+
if cargo_valid {
|
|
1410
|
+
commands = push(commands, "cargo test")
|
|
1411
|
+
commands = push(commands, "cargo fmt -- --check")
|
|
1412
|
+
}
|
|
1413
|
+
}
|
|
1414
|
+
|
|
1415
|
+
package_path := path_join(repo_root, "package.json")
|
|
1416
|
+
if path_exists(package_path) {
|
|
1417
|
+
pkg_read := read_text_file(package_path)
|
|
1418
|
+
if pkg_read["ok"] {
|
|
1419
|
+
parsed := null
|
|
1420
|
+
parse_ok := false
|
|
1421
|
+
try {
|
|
1422
|
+
parsed = parse_json(pkg_read["content"])
|
|
1423
|
+
parse_ok = true
|
|
1424
|
+
} except parse_error {
|
|
1425
|
+
parse_ok = false
|
|
1426
|
+
}
|
|
1427
|
+
|
|
1428
|
+
if parse_ok && type(parsed) == "dict" && has_key(parsed, "scripts") {
|
|
1429
|
+
scripts := parsed["scripts"]
|
|
1430
|
+
if type(scripts) == "dict" {
|
|
1431
|
+
if has_key(scripts, "test") {
|
|
1432
|
+
commands = push(commands, "npm test")
|
|
1433
|
+
}
|
|
1434
|
+
if has_key(scripts, "lint") {
|
|
1435
|
+
commands = push(commands, "npm run lint")
|
|
1436
|
+
}
|
|
1437
|
+
if has_key(scripts, "build") {
|
|
1438
|
+
commands = push(commands, "npm run build")
|
|
1439
|
+
}
|
|
1440
|
+
}
|
|
1441
|
+
}
|
|
1442
|
+
}
|
|
1443
|
+
}
|
|
1444
|
+
|
|
1445
|
+
makefile_path := path_join(repo_root, "Makefile")
|
|
1446
|
+
if path_exists(makefile_path) {
|
|
1447
|
+
mf := read_text_file(makefile_path)
|
|
1448
|
+
if mf["ok"] {
|
|
1449
|
+
make_lines := split(mf["content"], "\n")
|
|
1450
|
+
i := 0
|
|
1451
|
+
while i < len(make_lines) {
|
|
1452
|
+
make_line := make_lines[i]
|
|
1453
|
+
if starts_with(make_line, "test:") {
|
|
1454
|
+
commands = push(commands, "make test")
|
|
1455
|
+
}
|
|
1456
|
+
if starts_with(make_line, "build:") {
|
|
1457
|
+
commands = push(commands, "make build")
|
|
1458
|
+
}
|
|
1459
|
+
if starts_with(make_line, "lint:") {
|
|
1460
|
+
commands = push(commands, "make lint")
|
|
1461
|
+
}
|
|
1462
|
+
i = i + 1
|
|
1463
|
+
}
|
|
1464
|
+
}
|
|
1465
|
+
}
|
|
1466
|
+
|
|
1467
|
+
justfile_path := path_join(repo_root, "justfile")
|
|
1468
|
+
if !path_exists(justfile_path) {
|
|
1469
|
+
justfile_path = path_join(repo_root, "Justfile")
|
|
1470
|
+
}
|
|
1471
|
+
if path_exists(justfile_path) {
|
|
1472
|
+
jf := read_text_file(justfile_path)
|
|
1473
|
+
if jf["ok"] {
|
|
1474
|
+
just_lines := split(jf["content"], "\n")
|
|
1475
|
+
j := 0
|
|
1476
|
+
while j < len(just_lines) {
|
|
1477
|
+
just_line := just_lines[j]
|
|
1478
|
+
if starts_with(just_line, "test:") {
|
|
1479
|
+
commands = push(commands, "just test")
|
|
1480
|
+
}
|
|
1481
|
+
if starts_with(just_line, "build:") {
|
|
1482
|
+
commands = push(commands, "just build")
|
|
1483
|
+
}
|
|
1484
|
+
if starts_with(just_line, "lint:") {
|
|
1485
|
+
commands = push(commands, "just lint")
|
|
1486
|
+
}
|
|
1487
|
+
j = j + 1
|
|
1488
|
+
}
|
|
1489
|
+
}
|
|
1490
|
+
}
|
|
1491
|
+
|
|
1492
|
+
# dedupe + sort
|
|
1493
|
+
seen := {}
|
|
1494
|
+
uniq := []
|
|
1495
|
+
k := 0
|
|
1496
|
+
while k < len(commands) {
|
|
1497
|
+
c := commands[k]
|
|
1498
|
+
if has_key(seen, c) == 0 {
|
|
1499
|
+
seen[c] = 1
|
|
1500
|
+
uniq = push(uniq, c)
|
|
1501
|
+
}
|
|
1502
|
+
k = k + 1
|
|
1503
|
+
}
|
|
1504
|
+
return sort(uniq)
|
|
1505
|
+
}
|
|
1506
|
+
|
|
1507
|
+
func target_notes(target) {
|
|
1508
|
+
if target == "codex" {
|
|
1509
|
+
return "Codex usage notes:\n- Follow AGENTS.md/project instructions first.\n- Use this pack as bounded context.\n- Run detected tests after changes.\n- Do not modify files outside the task scope unless necessary."
|
|
1510
|
+
}
|
|
1511
|
+
if target == "claude" {
|
|
1512
|
+
return "Usage notes:\n- Follow project instructions first.\n- Use this pack as bounded context.\n- Run detected tests after changes.\n- Avoid changes outside task scope unless needed."
|
|
1513
|
+
}
|
|
1514
|
+
if target == "deepseek" {
|
|
1515
|
+
return "Implementation notes:\n- Treat this as bounded task context.\n- Prefer minimal diff and deterministic changes.\n- Run detected tests before handoff."
|
|
1516
|
+
}
|
|
1517
|
+
return "Usage notes:\n- Use this pack as bounded task context.\n- Follow project instructions and constraints.\n- Validate with detected commands before handoff."
|
|
1518
|
+
}
|
|
1519
|
+
|
|
1520
|
+
func markdown_fence(content) {
|
|
1521
|
+
fence := "```"
|
|
1522
|
+
while contains(content, fence) {
|
|
1523
|
+
fence = fence + "`"
|
|
1524
|
+
}
|
|
1525
|
+
return fence
|
|
1526
|
+
}
|
|
1527
|
+
|
|
1528
|
+
func markdown_section_list(title, items, empty_line) {
|
|
1529
|
+
out := "## " + title + "\n\n"
|
|
1530
|
+
if len(items) == 0 {
|
|
1531
|
+
out = out + empty_line + "\n\n"
|
|
1532
|
+
return out
|
|
1533
|
+
}
|
|
1534
|
+
i := 0
|
|
1535
|
+
while i < len(items) {
|
|
1536
|
+
item := items[i]
|
|
1537
|
+
out = out + "- " + item + "\n"
|
|
1538
|
+
i = i + 1
|
|
1539
|
+
}
|
|
1540
|
+
out = out + "\n"
|
|
1541
|
+
return out
|
|
1542
|
+
}
|
|
1543
|
+
|
|
1544
|
+
func build_markdown(task, target, budget, generated_at, repo_root, git_info, instruction_files, selected_files, changed_files, commands, constraints, excluded_manifest, redactions, manifest, estimated_tokens) {
|
|
1545
|
+
out := "# Scent Context Pack\n\n"
|
|
1546
|
+
out = out + "Task: " + task + "\n\n"
|
|
1547
|
+
out = out + "Target Agent: " + target + "\n\n"
|
|
1548
|
+
out = out + "Budget: " + to_string(budget) + "\n\n"
|
|
1549
|
+
out = out + "Generated: " + generated_at + "\n\n"
|
|
1550
|
+
out = out + "Repository: " + repo_root + "\n\n"
|
|
1551
|
+
out = out + "Git Branch: " + git_info["branch"] + "\n\n"
|
|
1552
|
+
|
|
1553
|
+
status_summary := "dirty=" + to_string(git_info["dirty"]) + " staged=" + to_string(len(git_info["staged_files"])) + " unstaged=" + to_string(len(git_info["unstaged_files"])) + " untracked=" + to_string(len(git_info["untracked_files"]))
|
|
1554
|
+
out = out + "Git Status Summary: " + status_summary + "\n\n"
|
|
1555
|
+
|
|
1556
|
+
out = out + "## How To Use This Pack\n\n"
|
|
1557
|
+
out = out + "Use this context as the bounded task context.\n"
|
|
1558
|
+
out = out + "Do not assume omitted files are irrelevant if tests or errors point elsewhere.\n"
|
|
1559
|
+
out = out + "Follow project instructions first.\n"
|
|
1560
|
+
out = out + "Respect the constraints and guardrails.\n"
|
|
1561
|
+
out = out + "Run the listed validation commands when making changes.\n\n"
|
|
1562
|
+
out = out + target_notes(target) + "\n\n"
|
|
1563
|
+
|
|
1564
|
+
inst_rel := []
|
|
1565
|
+
idx := 0
|
|
1566
|
+
while idx < len(instruction_files) {
|
|
1567
|
+
inst_path := instruction_files[idx]
|
|
1568
|
+
inst_rel = push(inst_rel, path_rel(repo_root, inst_path))
|
|
1569
|
+
idx = idx + 1
|
|
1570
|
+
}
|
|
1571
|
+
out = out + markdown_section_list("Project Instructions", inst_rel, "No instruction files discovered.")
|
|
1572
|
+
|
|
1573
|
+
out = out + "## Task-Relevant Files\n\n"
|
|
1574
|
+
idx = 0
|
|
1575
|
+
while idx < len(selected_files) {
|
|
1576
|
+
f := selected_files[idx]
|
|
1577
|
+
out = out + "- " + f["path"] + " (" + f["reason"] + ", " + to_string(f["estimated_tokens"]) + " tokens)\n"
|
|
1578
|
+
idx = idx + 1
|
|
1579
|
+
}
|
|
1580
|
+
out = out + "\n"
|
|
1581
|
+
|
|
1582
|
+
out = out + markdown_section_list("Changed Files", changed_files, "No changed files detected.")
|
|
1583
|
+
|
|
1584
|
+
test_commands := []
|
|
1585
|
+
build_commands := []
|
|
1586
|
+
idx = 0
|
|
1587
|
+
while idx < len(commands) {
|
|
1588
|
+
c := commands[idx]
|
|
1589
|
+
if contains(c, "test") {
|
|
1590
|
+
test_commands = push(test_commands, "`" + c + "`")
|
|
1591
|
+
} else {
|
|
1592
|
+
build_commands = push(build_commands, "`" + c + "`")
|
|
1593
|
+
}
|
|
1594
|
+
idx = idx + 1
|
|
1595
|
+
}
|
|
1596
|
+
out = out + markdown_section_list("Tests And Validation Commands", test_commands, "No validation commands were confidently detected.")
|
|
1597
|
+
out = out + markdown_section_list("Build/Run Commands", build_commands, "No build/run commands were confidently detected.")
|
|
1598
|
+
|
|
1599
|
+
out = out + "## Constraints And Guardrails\n\n"
|
|
1600
|
+
idx = 0
|
|
1601
|
+
while idx < len(constraints) {
|
|
1602
|
+
c := constraints[idx]
|
|
1603
|
+
out = out + "- " + c + "\n"
|
|
1604
|
+
idx = idx + 1
|
|
1605
|
+
}
|
|
1606
|
+
out = out + "\n"
|
|
1607
|
+
|
|
1608
|
+
out = out + "## Included File Summaries\n\n"
|
|
1609
|
+
idx = 0
|
|
1610
|
+
while idx < len(selected_files) {
|
|
1611
|
+
f := selected_files[idx]
|
|
1612
|
+
out = out + "- " + f["path"] + ": reason=" + f["reason"] + ", size_bytes=" + to_string(f["size_bytes"]) + ", estimated_tokens=" + to_string(f["estimated_tokens"]) + "\n"
|
|
1613
|
+
idx = idx + 1
|
|
1614
|
+
}
|
|
1615
|
+
out = out + "\n"
|
|
1616
|
+
|
|
1617
|
+
out = out + "## Included File Contents\n\n"
|
|
1618
|
+
idx = 0
|
|
1619
|
+
while idx < len(selected_files) {
|
|
1620
|
+
f := selected_files[idx]
|
|
1621
|
+
out = out + "### " + f["path"] + "\n\n"
|
|
1622
|
+
fence := markdown_fence(f["content"])
|
|
1623
|
+
out = out + fence + "text\n"
|
|
1624
|
+
out = out + f["content"]
|
|
1625
|
+
if !ends_with(f["content"], "\n") {
|
|
1626
|
+
out = out + "\n"
|
|
1627
|
+
}
|
|
1628
|
+
out = out + fence + "\n\n"
|
|
1629
|
+
idx = idx + 1
|
|
1630
|
+
}
|
|
1631
|
+
|
|
1632
|
+
out = out + "## Excluded Or Truncated Content\n\n"
|
|
1633
|
+
if len(excluded_manifest) == 0 {
|
|
1634
|
+
out = out + "None.\n\n"
|
|
1635
|
+
} else {
|
|
1636
|
+
idx = 0
|
|
1637
|
+
while idx < len(excluded_manifest) {
|
|
1638
|
+
e := excluded_manifest[idx]
|
|
1639
|
+
out = out + "- " + e["path"] + " [" + e["decision"] + "] reason=" + e["reason"] + "\n"
|
|
1640
|
+
idx = idx + 1
|
|
1641
|
+
}
|
|
1642
|
+
out = out + "\n"
|
|
1643
|
+
}
|
|
1644
|
+
|
|
1645
|
+
out = out + "## Redactions\n\n"
|
|
1646
|
+
if len(redactions) == 0 {
|
|
1647
|
+
out = out + "No redactions were required.\n\n"
|
|
1648
|
+
} else {
|
|
1649
|
+
idx = 0
|
|
1650
|
+
while idx < len(redactions) {
|
|
1651
|
+
r := redactions[idx]
|
|
1652
|
+
line_text := "n/a"
|
|
1653
|
+
if has_key(r, "line") {
|
|
1654
|
+
line_text = to_string(r["line"])
|
|
1655
|
+
}
|
|
1656
|
+
out = out + "- file=" + r["file"] + " line=" + line_text + " type=" + r["redaction_type"] + " placeholder=" + r["placeholder"] + "\n"
|
|
1657
|
+
idx = idx + 1
|
|
1658
|
+
}
|
|
1659
|
+
out = out + "\n"
|
|
1660
|
+
}
|
|
1661
|
+
|
|
1662
|
+
out = out + "## Manifest Summary\n\n"
|
|
1663
|
+
out = out + "Included files: " + to_string(len(selected_files)) + "\n\n"
|
|
1664
|
+
out = out + "Estimated tokens: " + to_string(estimated_tokens) + " / " + to_string(budget) + "\n\n"
|
|
1665
|
+
|
|
1666
|
+
idx = 0
|
|
1667
|
+
while idx < len(manifest) {
|
|
1668
|
+
m := manifest[idx]
|
|
1669
|
+
out = out + "- " + m["path"] + " => " + m["decision"] + " (" + m["reason"] + ", score=" + to_string(m["score"]) + ", tokens=" + to_string(m["estimated_tokens"]) + ")\n"
|
|
1670
|
+
idx = idx + 1
|
|
1671
|
+
}
|
|
1672
|
+
|
|
1673
|
+
return out
|
|
1674
|
+
}
|
|
1675
|
+
|
|
1676
|
+
func ensure_dir(path) {
|
|
1677
|
+
if path_exists(path) {
|
|
1678
|
+
return true
|
|
1679
|
+
}
|
|
1680
|
+
|
|
1681
|
+
parent := dirname(path)
|
|
1682
|
+
if parent != path && !path_exists(parent) {
|
|
1683
|
+
ensure_dir(parent)
|
|
1684
|
+
}
|
|
1685
|
+
|
|
1686
|
+
result := create_dir(path)
|
|
1687
|
+
if !result {
|
|
1688
|
+
return false
|
|
1689
|
+
}
|
|
1690
|
+
return true
|
|
1691
|
+
}
|
|
1692
|
+
|
|
1693
|
+
func path_entry_exists(path) {
|
|
1694
|
+
parent := dirname(path)
|
|
1695
|
+
if !path_exists(parent) || !path_is_dir(parent) {
|
|
1696
|
+
return false
|
|
1697
|
+
}
|
|
1698
|
+
try {
|
|
1699
|
+
return contains(list_dir(parent), basename(path))
|
|
1700
|
+
} except e {
|
|
1701
|
+
return false
|
|
1702
|
+
}
|
|
1703
|
+
}
|
|
1704
|
+
|
|
1705
|
+
func write_text(path, content) {
|
|
1706
|
+
if path_entry_exists(path) && path_is_symlink(path) {
|
|
1707
|
+
try {
|
|
1708
|
+
removed := delete_file(path)
|
|
1709
|
+
if !removed {
|
|
1710
|
+
return err("failed to replace symlink artifact: " + path, 4)
|
|
1711
|
+
}
|
|
1712
|
+
} except e {
|
|
1713
|
+
return err("failed to replace symlink artifact " + path + ": " + to_string(e), 4)
|
|
1714
|
+
}
|
|
1715
|
+
}
|
|
1716
|
+
|
|
1717
|
+
parent := dirname(path)
|
|
1718
|
+
if !ensure_dir(parent) {
|
|
1719
|
+
return err("failed to create parent directory for " + path, 4)
|
|
1720
|
+
}
|
|
1721
|
+
|
|
1722
|
+
wr := write_file(path, content, true)
|
|
1723
|
+
if !wr {
|
|
1724
|
+
return err("failed to write " + path, 4)
|
|
1725
|
+
}
|
|
1726
|
+
return ok(true)
|
|
1727
|
+
}
|
|
1728
|
+
|
|
1729
|
+
func delete_stale_file(path) {
|
|
1730
|
+
if !path_entry_exists(path) {
|
|
1731
|
+
return ok(true)
|
|
1732
|
+
}
|
|
1733
|
+
if !path_is_file(path) && !path_is_symlink(path) {
|
|
1734
|
+
return err("cannot replace stale artifact because it is not a file: " + path, 4)
|
|
1735
|
+
}
|
|
1736
|
+
|
|
1737
|
+
try {
|
|
1738
|
+
removed := delete_file(path)
|
|
1739
|
+
if removed {
|
|
1740
|
+
return ok(true)
|
|
1741
|
+
}
|
|
1742
|
+
return err("failed to delete stale artifact: " + path, 4)
|
|
1743
|
+
} except e {
|
|
1744
|
+
return err("failed to delete stale artifact " + path + ": " + to_string(e), 4)
|
|
1745
|
+
}
|
|
1746
|
+
}
|
|
1747
|
+
|
|
1748
|
+
func summary_payload(result) {
|
|
1749
|
+
return {
|
|
1750
|
+
"output_dir": result["output_dir"],
|
|
1751
|
+
"context_md": result["context_md"],
|
|
1752
|
+
"context_json": result["context_json"],
|
|
1753
|
+
"estimated_tokens": result["estimated_tokens"],
|
|
1754
|
+
"budget": result["budget"],
|
|
1755
|
+
"included_files": result["included_files"],
|
|
1756
|
+
"warnings": result["warnings"]
|
|
1757
|
+
}
|
|
1758
|
+
}
|
|
1759
|
+
|
|
1760
|
+
func run_pack(opts, raw_flags) {
|
|
1761
|
+
started := now()
|
|
1762
|
+
cwd := os_getcwd()
|
|
1763
|
+
repo_root := detect_repo_root(cwd)
|
|
1764
|
+
git_info := inspect_git(repo_root)
|
|
1765
|
+
instruction_files := discover_instruction_files(repo_root)
|
|
1766
|
+
task_redacted := redact_text("[task]", to_string(opts["task"]))
|
|
1767
|
+
task_text := task_redacted["content"]
|
|
1768
|
+
budget_value := to_int(opts["budget"])
|
|
1769
|
+
max_file_bytes_value := to_int(opts["max_file_bytes"])
|
|
1770
|
+
target_value := to_string(opts["target"])
|
|
1771
|
+
format_value := to_string(opts["format"])
|
|
1772
|
+
|
|
1773
|
+
include_scope := validate_scoped_paths(repo_root, opts["include"], "--include")
|
|
1774
|
+
if !include_scope["ok"] {
|
|
1775
|
+
return include_scope
|
|
1776
|
+
}
|
|
1777
|
+
opts["include"] = include_scope["value"]
|
|
1778
|
+
exclude_scope := validate_scoped_paths(repo_root, opts["exclude"], "--exclude")
|
|
1779
|
+
if !exclude_scope["ok"] {
|
|
1780
|
+
return exclude_scope
|
|
1781
|
+
}
|
|
1782
|
+
opts["exclude"] = exclude_scope["value"]
|
|
1783
|
+
|
|
1784
|
+
stamp := "unix-" + to_string(to_int(now() * 1000))
|
|
1785
|
+
out_dir := opts["out"]
|
|
1786
|
+
if out_dir == null {
|
|
1787
|
+
out_dir = path_join(path_join(path_join(repo_root, ".scent"), "packs"), stamp)
|
|
1788
|
+
} else {
|
|
1789
|
+
out_dir = resolve_path(cwd, out_dir)
|
|
1790
|
+
}
|
|
1791
|
+
|
|
1792
|
+
candidate_excludes := opts["exclude"]
|
|
1793
|
+
normalized_out_dir := canonical_path_for_scope(out_dir)
|
|
1794
|
+
if normalized_out_dir != repo_root && starts_with(normalized_out_dir, repo_root + "/") {
|
|
1795
|
+
candidate_excludes = push(candidate_excludes, normalized_out_dir)
|
|
1796
|
+
}
|
|
1797
|
+
|
|
1798
|
+
task_terms := tokenize_task(task_text)
|
|
1799
|
+
|
|
1800
|
+
reserve := reserved_tokens(task_text)
|
|
1801
|
+
if budget_value <= reserve {
|
|
1802
|
+
return err("budget too small: " + to_string(budget_value) + ". minimum required is " + to_string(reserve + 1), 5)
|
|
1803
|
+
}
|
|
1804
|
+
|
|
1805
|
+
warnings := []
|
|
1806
|
+
if !git_info["available"] {
|
|
1807
|
+
warnings = push(warnings, "git unavailable; running in degraded mode")
|
|
1808
|
+
}
|
|
1809
|
+
|
|
1810
|
+
candidates := collect_candidates(repo_root, git_info, instruction_files, opts["include"], candidate_excludes, task_terms)
|
|
1811
|
+
selected_candidates := select_candidates(opts, repo_root, candidates)
|
|
1812
|
+
|
|
1813
|
+
selected_files := []
|
|
1814
|
+
redactions_by_path := {}
|
|
1815
|
+
read_failures := {}
|
|
1816
|
+
|
|
1817
|
+
loop_idx := 0
|
|
1818
|
+
while loop_idx < len(selected_candidates) {
|
|
1819
|
+
selected_candidate := selected_candidates[loop_idx]
|
|
1820
|
+
read_res := read_text_file(selected_candidate["path"])
|
|
1821
|
+
if !read_res["ok"] {
|
|
1822
|
+
read_failures[selected_candidate["rel_path"]] = read_res["error"]
|
|
1823
|
+
loop_idx = loop_idx + 1
|
|
1824
|
+
continue
|
|
1825
|
+
}
|
|
1826
|
+
|
|
1827
|
+
redacted := redact_text(selected_candidate["rel_path"], read_res["content"])
|
|
1828
|
+
clipped := clip_text(redacted["content"], max_file_bytes_value)
|
|
1829
|
+
|
|
1830
|
+
redactions_by_path[selected_candidate["rel_path"]] = redacted["redactions"]
|
|
1831
|
+
|
|
1832
|
+
content := clipped["content"]
|
|
1833
|
+
selected_files = push(selected_files, {
|
|
1834
|
+
"path": selected_candidate["rel_path"],
|
|
1835
|
+
"reason": selected_candidate["selection_reason"],
|
|
1836
|
+
"score": score_candidate(selected_candidate, opts, repo_root)["score"],
|
|
1837
|
+
"size_bytes": selected_candidate["size_bytes"],
|
|
1838
|
+
"estimated_tokens": estimate_tokens(content),
|
|
1839
|
+
"included": true,
|
|
1840
|
+
"truncated": clipped["truncated"],
|
|
1841
|
+
"content": content,
|
|
1842
|
+
"content_hash": sha256(content)
|
|
1843
|
+
})
|
|
1844
|
+
loop_idx = loop_idx + 1
|
|
1845
|
+
}
|
|
1846
|
+
|
|
1847
|
+
budget_result := apply_budget(budget_value, reserve, selected_files)
|
|
1848
|
+
included_files := budget_result["files"]
|
|
1849
|
+
manifest := budget_result["manifest"]
|
|
1850
|
+
|
|
1851
|
+
redactions := task_redacted["redactions"]
|
|
1852
|
+
loop_idx = 0
|
|
1853
|
+
while loop_idx < len(included_files) {
|
|
1854
|
+
final_file := included_files[loop_idx]
|
|
1855
|
+
final_lines := split(final_file["content"], "\n")
|
|
1856
|
+
if has_key(redactions_by_path, final_file["path"]) {
|
|
1857
|
+
file_redactions := redactions_by_path[final_file["path"]]
|
|
1858
|
+
red_idx := 0
|
|
1859
|
+
while red_idx < len(file_redactions) {
|
|
1860
|
+
redaction_item := file_redactions[red_idx]
|
|
1861
|
+
line_index := to_int(redaction_item["line"]) - 1
|
|
1862
|
+
if line_index >= 0 && line_index < len(final_lines) && contains(final_lines[line_index], redaction_item["placeholder"]) {
|
|
1863
|
+
redactions = push(redactions, redaction_item)
|
|
1864
|
+
}
|
|
1865
|
+
red_idx = red_idx + 1
|
|
1866
|
+
}
|
|
1867
|
+
}
|
|
1868
|
+
loop_idx = loop_idx + 1
|
|
1869
|
+
}
|
|
1870
|
+
|
|
1871
|
+
if len(included_files) == 0 {
|
|
1872
|
+
warnings = push(warnings, "no relevant files selected; output contains metadata only")
|
|
1873
|
+
}
|
|
1874
|
+
|
|
1875
|
+
selected_map := {}
|
|
1876
|
+
selected_reason_map := {}
|
|
1877
|
+
loop_idx = 0
|
|
1878
|
+
while loop_idx < len(included_files) {
|
|
1879
|
+
included_file := included_files[loop_idx]
|
|
1880
|
+
selected_map[included_file["path"]] = 1
|
|
1881
|
+
selected_reason_map[included_file["path"]] = included_file["reason"]
|
|
1882
|
+
loop_idx = loop_idx + 1
|
|
1883
|
+
}
|
|
1884
|
+
|
|
1885
|
+
manifested_file_map := {}
|
|
1886
|
+
loop_idx = 0
|
|
1887
|
+
while loop_idx < len(manifest) {
|
|
1888
|
+
existing_manifest := manifest[loop_idx]
|
|
1889
|
+
if existing_manifest["type"] == "file" {
|
|
1890
|
+
manifested_file_map[existing_manifest["path"]] = 1
|
|
1891
|
+
}
|
|
1892
|
+
loop_idx = loop_idx + 1
|
|
1893
|
+
}
|
|
1894
|
+
|
|
1895
|
+
loop_idx = 0
|
|
1896
|
+
while loop_idx < len(candidates) {
|
|
1897
|
+
candidate_item := candidates[loop_idx]
|
|
1898
|
+
if has_key(selected_map, candidate_item["rel_path"]) {
|
|
1899
|
+
candidate_item["selected"] = true
|
|
1900
|
+
candidate_item["selection_reason"] = selected_reason_map[candidate_item["rel_path"]]
|
|
1901
|
+
} else {
|
|
1902
|
+
candidate_item["selected"] = false
|
|
1903
|
+
if has_key(read_failures, candidate_item["rel_path"]) {
|
|
1904
|
+
candidate_item["selection_reason"] = "excluded because file could not be read: " + to_string(read_failures[candidate_item["rel_path"]])
|
|
1905
|
+
} else if candidate_item["selection_reason"] == "" {
|
|
1906
|
+
candidate_item["selection_reason"] = "not selected by deterministic relevance"
|
|
1907
|
+
}
|
|
1908
|
+
if has_key(manifested_file_map, candidate_item["rel_path"]) == 0 {
|
|
1909
|
+
manifest = push(manifest, {
|
|
1910
|
+
"path": candidate_item["rel_path"],
|
|
1911
|
+
"type": "file",
|
|
1912
|
+
"decision": "exclude",
|
|
1913
|
+
"reason": candidate_item["selection_reason"],
|
|
1914
|
+
"score": 0,
|
|
1915
|
+
"size_bytes": candidate_item["size_bytes"],
|
|
1916
|
+
"estimated_tokens": 0
|
|
1917
|
+
})
|
|
1918
|
+
}
|
|
1919
|
+
}
|
|
1920
|
+
candidates[loop_idx] = candidate_item
|
|
1921
|
+
loop_idx = loop_idx + 1
|
|
1922
|
+
}
|
|
1923
|
+
|
|
1924
|
+
commands := discover_commands(repo_root)
|
|
1925
|
+
|
|
1926
|
+
constraints := [
|
|
1927
|
+
"Keep scope tight and deterministic.",
|
|
1928
|
+
"Do not assume omitted files are irrelevant if errors point elsewhere.",
|
|
1929
|
+
"Avoid touching files outside the task scope unless necessary.",
|
|
1930
|
+
"Do not include secrets in output artifacts."
|
|
1931
|
+
]
|
|
1932
|
+
|
|
1933
|
+
estimated_tokens := estimate_tokens(task_text) + 2
|
|
1934
|
+
loop_idx = 0
|
|
1935
|
+
while loop_idx < len(included_files) {
|
|
1936
|
+
estimated_tokens = estimated_tokens + included_files[loop_idx]["estimated_tokens"]
|
|
1937
|
+
loop_idx = loop_idx + 1
|
|
1938
|
+
}
|
|
1939
|
+
|
|
1940
|
+
generated_at := "unix-" + to_string(to_int(now()))
|
|
1941
|
+
|
|
1942
|
+
artifact_names := []
|
|
1943
|
+
if format_value == "md" || format_value == "both" {
|
|
1944
|
+
artifact_names = push(artifact_names, "context.md")
|
|
1945
|
+
}
|
|
1946
|
+
if format_value == "json" || format_value == "both" {
|
|
1947
|
+
artifact_names = push(artifact_names, "context.json")
|
|
1948
|
+
}
|
|
1949
|
+
artifact_names = push(artifact_names, "manifest.json")
|
|
1950
|
+
artifact_names = push(artifact_names, "files.json")
|
|
1951
|
+
artifact_names = push(artifact_names, "redactions.json")
|
|
1952
|
+
artifact_names = push(artifact_names, "metadata.json")
|
|
1953
|
+
|
|
1954
|
+
loop_idx = 0
|
|
1955
|
+
while loop_idx < len(artifact_names) {
|
|
1956
|
+
art := artifact_names[loop_idx]
|
|
1957
|
+
decision := "include"
|
|
1958
|
+
reason := "generated artifact"
|
|
1959
|
+
if opts["dry_run"] {
|
|
1960
|
+
decision = "exclude"
|
|
1961
|
+
reason = "dry-run; artifact not written"
|
|
1962
|
+
}
|
|
1963
|
+
|
|
1964
|
+
manifest = push(manifest, {
|
|
1965
|
+
"path": art,
|
|
1966
|
+
"type": "artifact",
|
|
1967
|
+
"decision": decision,
|
|
1968
|
+
"reason": reason,
|
|
1969
|
+
"score": 0,
|
|
1970
|
+
"size_bytes": 0,
|
|
1971
|
+
"estimated_tokens": 0
|
|
1972
|
+
})
|
|
1973
|
+
loop_idx = loop_idx + 1
|
|
1974
|
+
}
|
|
1975
|
+
|
|
1976
|
+
excluded_manifest := []
|
|
1977
|
+
loop_idx = 0
|
|
1978
|
+
while loop_idx < len(manifest) {
|
|
1979
|
+
manifest_item := manifest[loop_idx]
|
|
1980
|
+
if manifest_item["decision"] == "exclude" || manifest_item["decision"] == "truncate" {
|
|
1981
|
+
excluded_manifest = push(excluded_manifest, manifest_item)
|
|
1982
|
+
}
|
|
1983
|
+
loop_idx = loop_idx + 1
|
|
1984
|
+
}
|
|
1985
|
+
|
|
1986
|
+
markdown := build_markdown(
|
|
1987
|
+
task_text,
|
|
1988
|
+
target_value,
|
|
1989
|
+
budget_value,
|
|
1990
|
+
generated_at,
|
|
1991
|
+
repo_root,
|
|
1992
|
+
git_info,
|
|
1993
|
+
instruction_files,
|
|
1994
|
+
included_files,
|
|
1995
|
+
git_info["changed_files"],
|
|
1996
|
+
commands,
|
|
1997
|
+
constraints,
|
|
1998
|
+
excluded_manifest,
|
|
1999
|
+
redactions,
|
|
2000
|
+
manifest,
|
|
2001
|
+
estimated_tokens
|
|
2002
|
+
)
|
|
2003
|
+
|
|
2004
|
+
instruction_rels := []
|
|
2005
|
+
loop_idx = 0
|
|
2006
|
+
while loop_idx < len(instruction_files) {
|
|
2007
|
+
instruction_rels = push(instruction_rels, path_rel(repo_root, instruction_files[loop_idx]))
|
|
2008
|
+
loop_idx = loop_idx + 1
|
|
2009
|
+
}
|
|
2010
|
+
|
|
2011
|
+
context_json_obj := {
|
|
2012
|
+
"schema_version": "1.0.0",
|
|
2013
|
+
"task": task_text,
|
|
2014
|
+
"target": target_value,
|
|
2015
|
+
"budget": budget_value,
|
|
2016
|
+
"estimated_tokens": estimated_tokens,
|
|
2017
|
+
"generated_at": generated_at,
|
|
2018
|
+
"repo": repo_root,
|
|
2019
|
+
"git": git_info,
|
|
2020
|
+
"instructions": instruction_rels,
|
|
2021
|
+
"selected_files": included_files,
|
|
2022
|
+
"changed_files": git_info["changed_files"],
|
|
2023
|
+
"commands": commands,
|
|
2024
|
+
"constraints": constraints,
|
|
2025
|
+
"redactions": redactions,
|
|
2026
|
+
"excluded": excluded_manifest,
|
|
2027
|
+
"artifacts": artifact_names
|
|
2028
|
+
}
|
|
2029
|
+
|
|
2030
|
+
files_json := []
|
|
2031
|
+
loop_idx = 0
|
|
2032
|
+
while loop_idx < len(candidates) {
|
|
2033
|
+
cand := candidates[loop_idx]
|
|
2034
|
+
files_json = push(files_json, {
|
|
2035
|
+
"path": cand["rel_path"],
|
|
2036
|
+
"size_bytes": cand["size_bytes"],
|
|
2037
|
+
"is_changed": cand["is_changed"],
|
|
2038
|
+
"is_staged": cand["is_staged"],
|
|
2039
|
+
"is_unstaged": cand["is_unstaged"],
|
|
2040
|
+
"is_instruction": cand["is_instruction"],
|
|
2041
|
+
"is_doc": cand["is_doc"],
|
|
2042
|
+
"is_test": cand["is_test"],
|
|
2043
|
+
"is_config": cand["is_config"],
|
|
2044
|
+
"matched_terms": cand["matched_terms"],
|
|
2045
|
+
"selected": cand["selected"],
|
|
2046
|
+
"selection_reason": cand["selection_reason"]
|
|
2047
|
+
})
|
|
2048
|
+
loop_idx = loop_idx + 1
|
|
2049
|
+
}
|
|
2050
|
+
|
|
2051
|
+
exit_code := 0
|
|
2052
|
+
if len(warnings) > 0 {
|
|
2053
|
+
exit_code = 1
|
|
2054
|
+
}
|
|
2055
|
+
|
|
2056
|
+
sanitized_flags := []
|
|
2057
|
+
loop_idx = 0
|
|
2058
|
+
while loop_idx < len(raw_flags) {
|
|
2059
|
+
sanitized_flag := redact_text("[command]", to_string(raw_flags[loop_idx]))
|
|
2060
|
+
sanitized_flags = push(sanitized_flags, sanitized_flag["content"])
|
|
2061
|
+
loop_idx = loop_idx + 1
|
|
2062
|
+
}
|
|
2063
|
+
|
|
2064
|
+
metadata := {
|
|
2065
|
+
"tool": "scent",
|
|
2066
|
+
"version": VERSION,
|
|
2067
|
+
"generated_at": generated_at,
|
|
2068
|
+
"duration_ms": to_int((now() - started) * 1000),
|
|
2069
|
+
"repo_root": repo_root,
|
|
2070
|
+
"git_branch": git_info["branch"],
|
|
2071
|
+
"git_commit": git_info["commit"],
|
|
2072
|
+
"dirty": git_info["dirty"],
|
|
2073
|
+
"command": "scent pack",
|
|
2074
|
+
"flags": sanitized_flags,
|
|
2075
|
+
"exit_code": exit_code
|
|
2076
|
+
}
|
|
2077
|
+
|
|
2078
|
+
context_md_path := null
|
|
2079
|
+
context_json_path := null
|
|
2080
|
+
|
|
2081
|
+
if !opts["dry_run"] {
|
|
2082
|
+
if format_value == "md" {
|
|
2083
|
+
stale_json := delete_stale_file(path_join(out_dir, "context.json"))
|
|
2084
|
+
if !stale_json["ok"] {
|
|
2085
|
+
return stale_json
|
|
2086
|
+
}
|
|
2087
|
+
} else if format_value == "json" {
|
|
2088
|
+
stale_markdown := delete_stale_file(path_join(out_dir, "context.md"))
|
|
2089
|
+
if !stale_markdown["ok"] {
|
|
2090
|
+
return stale_markdown
|
|
2091
|
+
}
|
|
2092
|
+
}
|
|
2093
|
+
|
|
2094
|
+
if format_value == "md" || format_value == "both" {
|
|
2095
|
+
context_md_path = path_join(out_dir, "context.md")
|
|
2096
|
+
w1 := write_text(context_md_path, markdown)
|
|
2097
|
+
if !w1["ok"] {
|
|
2098
|
+
return w1
|
|
2099
|
+
}
|
|
2100
|
+
}
|
|
2101
|
+
if format_value == "json" || format_value == "both" {
|
|
2102
|
+
context_json_path = path_join(out_dir, "context.json")
|
|
2103
|
+
w2 := write_text(context_json_path, to_json_pretty(context_json_obj))
|
|
2104
|
+
if !w2["ok"] {
|
|
2105
|
+
return w2
|
|
2106
|
+
}
|
|
2107
|
+
}
|
|
2108
|
+
|
|
2109
|
+
w3 := write_text(path_join(out_dir, "manifest.json"), to_json_pretty(manifest))
|
|
2110
|
+
if !w3["ok"] {
|
|
2111
|
+
return w3
|
|
2112
|
+
}
|
|
2113
|
+
|
|
2114
|
+
w4 := write_text(path_join(out_dir, "files.json"), to_json_pretty(files_json))
|
|
2115
|
+
if !w4["ok"] {
|
|
2116
|
+
return w4
|
|
2117
|
+
}
|
|
2118
|
+
|
|
2119
|
+
w5 := write_text(path_join(out_dir, "redactions.json"), to_json_pretty(redactions))
|
|
2120
|
+
if !w5["ok"] {
|
|
2121
|
+
return w5
|
|
2122
|
+
}
|
|
2123
|
+
|
|
2124
|
+
w6 := write_text(path_join(out_dir, "metadata.json"), to_json_pretty(metadata))
|
|
2125
|
+
if !w6["ok"] {
|
|
2126
|
+
return w6
|
|
2127
|
+
}
|
|
2128
|
+
}
|
|
2129
|
+
|
|
2130
|
+
return ok({
|
|
2131
|
+
"output_dir": out_dir,
|
|
2132
|
+
"context_md": context_md_path,
|
|
2133
|
+
"context_json": context_json_path,
|
|
2134
|
+
"estimated_tokens": estimated_tokens,
|
|
2135
|
+
"budget": budget_value,
|
|
2136
|
+
"included_files": len(included_files),
|
|
2137
|
+
"warnings": warnings,
|
|
2138
|
+
"selected_files": included_files,
|
|
2139
|
+
"manifest": manifest,
|
|
2140
|
+
"redactions": redactions,
|
|
2141
|
+
"exit_code": exit_code
|
|
2142
|
+
})
|
|
2143
|
+
}
|
|
2144
|
+
|
|
2145
|
+
func print_lines(lines) {
|
|
2146
|
+
i := 0
|
|
2147
|
+
while i < len(lines) {
|
|
2148
|
+
print(lines[i])
|
|
2149
|
+
i = i + 1
|
|
2150
|
+
}
|
|
2151
|
+
return null
|
|
2152
|
+
}
|
|
2153
|
+
|
|
2154
|
+
func print_kv(label, value) {
|
|
2155
|
+
print(label + ": " + to_string(value))
|
|
2156
|
+
return null
|
|
2157
|
+
}
|
|
2158
|
+
|
|
2159
|
+
func print_human_result(result, opts) {
|
|
2160
|
+
payload := summary_payload(result)
|
|
2161
|
+
|
|
2162
|
+
if opts["json_stdout"] {
|
|
2163
|
+
print(to_json(payload))
|
|
2164
|
+
return null
|
|
2165
|
+
}
|
|
2166
|
+
|
|
2167
|
+
print("scent pack complete")
|
|
2168
|
+
print_kv("output_dir", result["output_dir"])
|
|
2169
|
+
print_kv("included_files", result["included_files"])
|
|
2170
|
+
print_kv("estimated_tokens", to_string(result["estimated_tokens"]) + " / " + to_string(result["budget"]))
|
|
2171
|
+
if len(result["warnings"]) > 0 {
|
|
2172
|
+
warning_lines := ["warnings:"]
|
|
2173
|
+
i := 0
|
|
2174
|
+
while i < len(result["warnings"]) {
|
|
2175
|
+
warning_lines = push(warning_lines, "- " + result["warnings"][i])
|
|
2176
|
+
i = i + 1
|
|
2177
|
+
}
|
|
2178
|
+
print_lines(warning_lines)
|
|
2179
|
+
}
|
|
2180
|
+
|
|
2181
|
+
if opts["verbose"] {
|
|
2182
|
+
print(to_json(payload))
|
|
2183
|
+
}
|
|
2184
|
+
return null
|
|
2185
|
+
}
|
|
2186
|
+
|
|
2187
|
+
func main() {
|
|
2188
|
+
argv := args()
|
|
2189
|
+
|
|
2190
|
+
if len(argv) > 0 {
|
|
2191
|
+
first := argv[0]
|
|
2192
|
+
if is_help_arg(first) {
|
|
2193
|
+
print(usage_text())
|
|
2194
|
+
exit(0)
|
|
2195
|
+
}
|
|
2196
|
+
if is_version_arg(first) {
|
|
2197
|
+
print(version_text())
|
|
2198
|
+
exit(0)
|
|
2199
|
+
}
|
|
2200
|
+
if first == "pack" && len(argv) > 1 {
|
|
2201
|
+
second := argv[1]
|
|
2202
|
+
if is_help_arg(second) {
|
|
2203
|
+
print(usage_text())
|
|
2204
|
+
exit(0)
|
|
2205
|
+
}
|
|
2206
|
+
if is_version_arg(second) {
|
|
2207
|
+
print(version_text())
|
|
2208
|
+
exit(0)
|
|
2209
|
+
}
|
|
2210
|
+
}
|
|
2211
|
+
}
|
|
2212
|
+
|
|
2213
|
+
parsed := parse_cli(argv)
|
|
2214
|
+
|
|
2215
|
+
if !parsed["ok"] {
|
|
2216
|
+
if parsed["exit_code"] == 2 {
|
|
2217
|
+
print(parsed["error"])
|
|
2218
|
+
} else {
|
|
2219
|
+
print("error: " + parsed["error"])
|
|
2220
|
+
}
|
|
2221
|
+
exit(parsed["exit_code"])
|
|
2222
|
+
}
|
|
2223
|
+
|
|
2224
|
+
result := run_pack(parsed["value"], argv)
|
|
2225
|
+
if !result["ok"] {
|
|
2226
|
+
print("error: " + result["error"])
|
|
2227
|
+
exit(result["exit_code"])
|
|
2228
|
+
}
|
|
2229
|
+
|
|
2230
|
+
print_human_result(result["value"], parsed["value"])
|
|
2231
|
+
exit(result["value"]["exit_code"])
|
|
2232
|
+
}
|
|
2233
|
+
|
|
2234
|
+
# =============================
|
|
2235
|
+
# Tests (kujo test-run scent.kujo)
|
|
2236
|
+
# =============================
|
|
2237
|
+
|
|
2238
|
+
test "help and version surface" {
|
|
2239
|
+
cwd := os_getcwd()
|
|
2240
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2241
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2242
|
+
|
|
2243
|
+
func shq(value) {
|
|
2244
|
+
text := to_string(value)
|
|
2245
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2246
|
+
}
|
|
2247
|
+
|
|
2248
|
+
func run_cmd(command) {
|
|
2249
|
+
res := execute_status(command, {"timeout_ms": 120000})
|
|
2250
|
+
if type(res) == "struct" {
|
|
2251
|
+
return res
|
|
2252
|
+
}
|
|
2253
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2254
|
+
}
|
|
2255
|
+
|
|
2256
|
+
func run_scent(args) {
|
|
2257
|
+
command := shq(kujo_bin) + " run " + shq(scent_script)
|
|
2258
|
+
i := 0
|
|
2259
|
+
while i < len(args) {
|
|
2260
|
+
command = command + " " + shq(args[i])
|
|
2261
|
+
i = i + 1
|
|
2262
|
+
}
|
|
2263
|
+
return run_cmd(command)
|
|
2264
|
+
}
|
|
2265
|
+
|
|
2266
|
+
help_res := run_scent(["help"])
|
|
2267
|
+
assert_equal(help_res.exitcode, 0)
|
|
2268
|
+
assert_equal(contains(help_res.stdout, "Usage:"), 1)
|
|
2269
|
+
assert_equal(contains(help_res.stdout, "Unknown argument"), 0)
|
|
2270
|
+
assert_equal(contains(help_res.stderr, "Unknown argument"), 0)
|
|
2271
|
+
|
|
2272
|
+
top_help_res := run_scent(["--help"])
|
|
2273
|
+
assert_equal(top_help_res.exitcode, 0)
|
|
2274
|
+
assert_equal(contains(top_help_res.stdout, "Usage:"), 1)
|
|
2275
|
+
|
|
2276
|
+
pack_help_res := run_scent(["pack", "--help"])
|
|
2277
|
+
assert_equal(pack_help_res.exitcode, 0)
|
|
2278
|
+
assert_equal(contains(pack_help_res.stdout, "Usage:"), 1)
|
|
2279
|
+
assert_equal(contains(pack_help_res.stdout, "Unknown argument"), 0)
|
|
2280
|
+
|
|
2281
|
+
pack_version_res := run_scent(["pack", "--version"])
|
|
2282
|
+
assert_equal(pack_version_res.exitcode, 0)
|
|
2283
|
+
assert_equal(contains(pack_version_res.stdout, "Scent "), 1)
|
|
2284
|
+
|
|
2285
|
+
version_res := run_scent(["version"])
|
|
2286
|
+
assert_equal(version_res.exitcode, 0)
|
|
2287
|
+
assert_equal(contains(version_res.stdout, "Scent "), 1)
|
|
2288
|
+
|
|
2289
|
+
flag_version_res := run_scent(["--version"])
|
|
2290
|
+
assert_equal(flag_version_res.exitcode, 0)
|
|
2291
|
+
assert_equal(contains(flag_version_res.stdout, "Scent "), 1)
|
|
2292
|
+
}
|
|
2293
|
+
|
|
2294
|
+
test "pack and dry-run smoke" {
|
|
2295
|
+
cwd := os_getcwd()
|
|
2296
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2297
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2298
|
+
smoke_root := "/tmp/scent_phase8_fix"
|
|
2299
|
+
smoke_repo := smoke_root + "/repo"
|
|
2300
|
+
smoke_out := smoke_root + "/out"
|
|
2301
|
+
smoke_dry_out := smoke_root + "/dry-run-out"
|
|
2302
|
+
smoke_human_out := smoke_root + "/dry-run-human-out"
|
|
2303
|
+
|
|
2304
|
+
func shq(value) {
|
|
2305
|
+
text := to_string(value)
|
|
2306
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2307
|
+
}
|
|
2308
|
+
|
|
2309
|
+
func run_cmd(command) {
|
|
2310
|
+
res := execute_status(command, {"timeout_ms": 120000})
|
|
2311
|
+
if type(res) == "struct" {
|
|
2312
|
+
return res
|
|
2313
|
+
}
|
|
2314
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2315
|
+
}
|
|
2316
|
+
|
|
2317
|
+
func run_scent_in_repo(args) {
|
|
2318
|
+
command := "cd " + shq(smoke_repo) + " && " + shq(kujo_bin) + " run " + shq(scent_script)
|
|
2319
|
+
i := 0
|
|
2320
|
+
while i < len(args) {
|
|
2321
|
+
command = command + " " + shq(args[i])
|
|
2322
|
+
i = i + 1
|
|
2323
|
+
}
|
|
2324
|
+
return run_cmd(command)
|
|
2325
|
+
}
|
|
2326
|
+
|
|
2327
|
+
func parse_json_safe(text) {
|
|
2328
|
+
try {
|
|
2329
|
+
return {"ok": true, "value": parse_json(text)}
|
|
2330
|
+
} except err {
|
|
2331
|
+
return {"ok": false, "error": to_string(err), "value": {}}
|
|
2332
|
+
}
|
|
2333
|
+
}
|
|
2334
|
+
|
|
2335
|
+
func require_truthy(value, label) {
|
|
2336
|
+
if value == false || value == 0 {
|
|
2337
|
+
print("Assertion failed: " + label)
|
|
2338
|
+
exit(1)
|
|
2339
|
+
}
|
|
2340
|
+
}
|
|
2341
|
+
|
|
2342
|
+
func require_files_exist(root, files) {
|
|
2343
|
+
i := 0
|
|
2344
|
+
while i < len(files) {
|
|
2345
|
+
file := files[i]
|
|
2346
|
+
require_truthy(file_exists(root + "/" + file), file + " exists")
|
|
2347
|
+
i = i + 1
|
|
2348
|
+
}
|
|
2349
|
+
return null
|
|
2350
|
+
}
|
|
2351
|
+
|
|
2352
|
+
func smoke_pack_args(task, out_dir, json_stdout, dry_run) {
|
|
2353
|
+
values := [
|
|
2354
|
+
"pack",
|
|
2355
|
+
"--task", task,
|
|
2356
|
+
"--out", out_dir,
|
|
2357
|
+
"--changed",
|
|
2358
|
+
"--staged",
|
|
2359
|
+
"--unstaged",
|
|
2360
|
+
"--include", "docs",
|
|
2361
|
+
"--include", "src",
|
|
2362
|
+
"--include", "tests",
|
|
2363
|
+
"--include", "logs",
|
|
2364
|
+
"--include", "reports",
|
|
2365
|
+
"--include", "notes",
|
|
2366
|
+
"--max-files", "10",
|
|
2367
|
+
"--max-file-bytes", "2000",
|
|
2368
|
+
"--format", "both"
|
|
2369
|
+
]
|
|
2370
|
+
|
|
2371
|
+
if json_stdout {
|
|
2372
|
+
values = push(values, "--json")
|
|
2373
|
+
}
|
|
2374
|
+
if dry_run {
|
|
2375
|
+
values = push(values, "--dry-run")
|
|
2376
|
+
}
|
|
2377
|
+
return values
|
|
2378
|
+
}
|
|
2379
|
+
|
|
2380
|
+
_ = run_cmd("rm -rf " + shq(smoke_root))
|
|
2381
|
+
_ = run_cmd("mkdir -p " + shq(smoke_repo + "/docs") + " " + shq(smoke_repo + "/src") + " " + shq(smoke_repo + "/tests") + " " + shq(smoke_repo + "/logs") + " " + shq(smoke_repo + "/reports") + " " + shq(smoke_repo + "/notes"))
|
|
2382
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git init -q")
|
|
2383
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git config user.name " + shq("Scent Test"))
|
|
2384
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git config user.email " + shq("scent@example.com"))
|
|
2385
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' '# Smoke repo' 'Initial content' > README.md")
|
|
2386
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'Initial instructions' 'Follow local cues only.' > AGENTS.md")
|
|
2387
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' '# Architecture' 'Deterministic paths only.' > docs/architecture.md")
|
|
2388
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'Potential risk' 'API_KEY=abc123' > docs/security.md")
|
|
2389
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'print(\"hello\")' > src/main.kujo")
|
|
2390
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'workflow notes' > src/workflow.kujo")
|
|
2391
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'assert true' > tests/sample_test.kujo")
|
|
2392
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'sample log' > logs/sample.log")
|
|
2393
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' '{\"score\": 7}' > reports/eval.json")
|
|
2394
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'context notes' > notes/context.md")
|
|
2395
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git add . && git commit -q -m " + shq("seed"))
|
|
2396
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n' 'Unstaged task context.' >> README.md")
|
|
2397
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n' 'Staged security note.' >> docs/security.md")
|
|
2398
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git add docs/security.md")
|
|
2399
|
+
|
|
2400
|
+
task := "review workflow cues and risks"
|
|
2401
|
+
pack_res := run_scent_in_repo(smoke_pack_args(task, smoke_out, true, false))
|
|
2402
|
+
assert_equal(pack_res.exitcode, 0)
|
|
2403
|
+
|
|
2404
|
+
pack_out := parse_json_safe(to_string(pack_res.stdout))
|
|
2405
|
+
require_truthy(pack_out["ok"], "pack output ok")
|
|
2406
|
+
pack_data := pack_out["value"]
|
|
2407
|
+
assert_equal(pack_data["estimated_tokens"], 85)
|
|
2408
|
+
assert_equal(pack_data["included_files"], 10)
|
|
2409
|
+
assert_equal(len(pack_data["warnings"]), 0)
|
|
2410
|
+
require_files_exist(smoke_out, ["context.json", "context.md", "files.json", "manifest.json", "metadata.json", "redactions.json"])
|
|
2411
|
+
|
|
2412
|
+
context := read_file(smoke_out + "/context.json")
|
|
2413
|
+
context_json := parse_json_safe(context)
|
|
2414
|
+
require_truthy(context_json["ok"], "context json parses")
|
|
2415
|
+
context_data := context_json["value"]
|
|
2416
|
+
assert_equal(context_data["git"]["dirty"], true)
|
|
2417
|
+
require_truthy(len(context_data["changed_files"]) > 0, "changed files captured")
|
|
2418
|
+
require_truthy(contains(to_json(context_data["changed_files"]), "README.md"), "README.md captured as changed")
|
|
2419
|
+
require_truthy(contains(to_json(context_data["changed_files"]), "docs/security.md"), "docs/security.md captured as changed")
|
|
2420
|
+
require_truthy(contains(to_json(context_data["git"]["staged_files"]), "docs/security.md"), "staged metadata captured")
|
|
2421
|
+
require_truthy(contains(to_json(context_data["git"]["unstaged_files"]), "README.md"), "unstaged metadata captured")
|
|
2422
|
+
assert_equal(contains(context, "No changed files detected."), 0)
|
|
2423
|
+
require_truthy(contains(context, "[REDACTED:API_KEY]"), "redaction marker present")
|
|
2424
|
+
require_truthy(contains(context, "\"schema_version\""), "schema version present")
|
|
2425
|
+
files_json := parse_json_safe(read_file(smoke_out + "/files.json"))
|
|
2426
|
+
require_truthy(files_json["ok"], "files json parses")
|
|
2427
|
+
files_data := files_json["value"]
|
|
2428
|
+
files_by_path := {}
|
|
2429
|
+
idx := 0
|
|
2430
|
+
while idx < len(files_data) {
|
|
2431
|
+
item := files_data[idx]
|
|
2432
|
+
files_by_path[item["path"]] = item
|
|
2433
|
+
idx = idx + 1
|
|
2434
|
+
}
|
|
2435
|
+
require_truthy(files_by_path["README.md"]["is_changed"], "README.md marked changed")
|
|
2436
|
+
require_truthy(files_by_path["README.md"]["is_unstaged"], "README.md marked unstaged")
|
|
2437
|
+
require_truthy(files_by_path["docs/security.md"]["is_changed"], "docs/security.md marked changed")
|
|
2438
|
+
require_truthy(files_by_path["docs/security.md"]["is_staged"], "docs/security.md marked staged")
|
|
2439
|
+
|
|
2440
|
+
dry_res := run_scent_in_repo(smoke_pack_args(task, smoke_dry_out, true, true))
|
|
2441
|
+
assert_equal(dry_res.exitcode, 0)
|
|
2442
|
+
|
|
2443
|
+
dry_out := parse_json_safe(to_string(dry_res.stdout))
|
|
2444
|
+
require_truthy(dry_out["ok"], "dry-run output ok")
|
|
2445
|
+
dry_data := dry_out["value"]
|
|
2446
|
+
assert_equal(type(dry_data["context_json"]), "null")
|
|
2447
|
+
assert_equal(type(dry_data["context_md"]), "null")
|
|
2448
|
+
assert_equal(dry_data["estimated_tokens"], 85)
|
|
2449
|
+
assert_equal(dry_data["included_files"], 10)
|
|
2450
|
+
assert_equal(len(dry_data["warnings"]), 0)
|
|
2451
|
+
assert_equal(file_exists(smoke_dry_out + "/context.json"), false)
|
|
2452
|
+
assert_equal(file_exists(smoke_dry_out + "/context.md"), false)
|
|
2453
|
+
|
|
2454
|
+
human_res := run_scent_in_repo(smoke_pack_args(task, smoke_human_out, false, true))
|
|
2455
|
+
assert_equal(human_res.exitcode, 0)
|
|
2456
|
+
expected_human := "scent pack complete\noutput_dir: " + smoke_human_out + "\nincluded_files: 10\nestimated_tokens: 85 / 12000\n"
|
|
2457
|
+
assert_equal(to_string(human_res.stdout), expected_human)
|
|
2458
|
+
assert_equal(file_exists(smoke_human_out + "/context.json"), false)
|
|
2459
|
+
assert_equal(file_exists(smoke_human_out + "/context.md"), false)
|
|
2460
|
+
}
|
|
2461
|
+
|
|
2462
|
+
test "redaction and scoped include hardening" {
|
|
2463
|
+
cwd := os_getcwd()
|
|
2464
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2465
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2466
|
+
smoke_root := "/tmp/scent_hardening"
|
|
2467
|
+
smoke_repo := smoke_root + "/repo"
|
|
2468
|
+
smoke_out := smoke_root + "/out"
|
|
2469
|
+
|
|
2470
|
+
func shq(value) {
|
|
2471
|
+
text := to_string(value)
|
|
2472
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2473
|
+
}
|
|
2474
|
+
|
|
2475
|
+
func run_cmd(command) {
|
|
2476
|
+
res := execute_status(command, {"timeout_ms": 120000})
|
|
2477
|
+
if type(res) == "struct" {
|
|
2478
|
+
return res
|
|
2479
|
+
}
|
|
2480
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2481
|
+
}
|
|
2482
|
+
|
|
2483
|
+
func run_scent_in_repo(args) {
|
|
2484
|
+
command := "cd " + shq(smoke_repo) + " && " + shq(kujo_bin) + " run " + shq(scent_script)
|
|
2485
|
+
i := 0
|
|
2486
|
+
while i < len(args) {
|
|
2487
|
+
command = command + " " + shq(args[i])
|
|
2488
|
+
i = i + 1
|
|
2489
|
+
}
|
|
2490
|
+
return run_cmd(command)
|
|
2491
|
+
}
|
|
2492
|
+
|
|
2493
|
+
func parse_json_safe(text) {
|
|
2494
|
+
try {
|
|
2495
|
+
return {"ok": true, "value": parse_json(text)}
|
|
2496
|
+
} except err {
|
|
2497
|
+
return {"ok": false, "error": to_string(err), "value": {}}
|
|
2498
|
+
}
|
|
2499
|
+
}
|
|
2500
|
+
|
|
2501
|
+
func require_truthy(value, label) {
|
|
2502
|
+
if value == false || value == 0 {
|
|
2503
|
+
print("Assertion failed: " + label)
|
|
2504
|
+
exit(1)
|
|
2505
|
+
}
|
|
2506
|
+
}
|
|
2507
|
+
|
|
2508
|
+
_ = run_cmd("rm -rf " + shq(smoke_root))
|
|
2509
|
+
_ = run_cmd("mkdir -p " + shq(smoke_repo + "/docs") + " " + shq(smoke_repo + "/src"))
|
|
2510
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && git init -q")
|
|
2511
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n' '# Hardening repo' > README.md")
|
|
2512
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n' 'openai leak sk-12345678901234567890' > docs/security.md")
|
|
2513
|
+
_ = run_cmd("cd " + shq(smoke_repo) + " && printf '%s\\n%s\\n' 'github ghp_123456789012345678901234567890123456' 'aws AKIA1234567890ABCDEF' > src/app.kujo")
|
|
2514
|
+
|
|
2515
|
+
pack_res := run_scent_in_repo([
|
|
2516
|
+
"pack",
|
|
2517
|
+
"--task", "review hardening",
|
|
2518
|
+
"--out", smoke_out,
|
|
2519
|
+
"--include", "docs",
|
|
2520
|
+
"--include", "src",
|
|
2521
|
+
"--max-files", "5",
|
|
2522
|
+
"--format", "both",
|
|
2523
|
+
"--json"
|
|
2524
|
+
])
|
|
2525
|
+
assert_equal(pack_res.exitcode, 0)
|
|
2526
|
+
|
|
2527
|
+
pack_out := parse_json_safe(to_string(pack_res.stdout))
|
|
2528
|
+
require_truthy(pack_out["ok"], "hardening pack output parses")
|
|
2529
|
+
context := read_file(smoke_out + "/context.json")
|
|
2530
|
+
require_truthy(contains(context, "[REDACTED:TOKEN]"), "token placeholder present")
|
|
2531
|
+
require_truthy(contains(context, "[REDACTED:ACCESS_KEY]"), "access key placeholder present")
|
|
2532
|
+
assert_equal(contains(context, "sk-12345678901234567890"), 0)
|
|
2533
|
+
assert_equal(contains(context, "ghp_123456789012345678901234567890123456"), 0)
|
|
2534
|
+
assert_equal(contains(context, "AKIA1234567890ABCDEF"), 0)
|
|
2535
|
+
require_truthy(contains(context, "\"docs/security.md\""), "first repeated include selected")
|
|
2536
|
+
require_truthy(contains(context, "\"src/app.kujo\""), "second repeated include selected")
|
|
2537
|
+
|
|
2538
|
+
escape_res := run_scent_in_repo([
|
|
2539
|
+
"pack",
|
|
2540
|
+
"--task", "escape",
|
|
2541
|
+
"--include", "../outside",
|
|
2542
|
+
"--dry-run",
|
|
2543
|
+
"--json"
|
|
2544
|
+
])
|
|
2545
|
+
assert_equal(escape_res.exitcode, 2)
|
|
2546
|
+
require_truthy(contains(escape_res.stdout, "--include path cannot contain '..'"), "parent path rejected")
|
|
2547
|
+
|
|
2548
|
+
absolute_res := run_scent_in_repo([
|
|
2549
|
+
"pack",
|
|
2550
|
+
"--task", "absolute escape",
|
|
2551
|
+
"--include", "/tmp",
|
|
2552
|
+
"--dry-run",
|
|
2553
|
+
"--json"
|
|
2554
|
+
])
|
|
2555
|
+
assert_equal(absolute_res.exitcode, 2)
|
|
2556
|
+
require_truthy(contains(absolute_res.stdout, "--include path must stay inside the repository"), "absolute outside path rejected")
|
|
2557
|
+
}
|
|
2558
|
+
|
|
2559
|
+
test "ten bug regressions" {
|
|
2560
|
+
cwd := os_getcwd()
|
|
2561
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2562
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2563
|
+
bug_root := "/tmp/scent_bug_regressions"
|
|
2564
|
+
bug_repo := bug_root + "/repo"
|
|
2565
|
+
bug_out := bug_repo + "/pack"
|
|
2566
|
+
budget_out := bug_root + "/budget-out"
|
|
2567
|
+
late_out := bug_root + "/late-out"
|
|
2568
|
+
|
|
2569
|
+
func shq(value) {
|
|
2570
|
+
text := to_string(value)
|
|
2571
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2572
|
+
}
|
|
2573
|
+
|
|
2574
|
+
func run_cmd(command) {
|
|
2575
|
+
res := execute_status(command, {"timeout_ms": 120000})
|
|
2576
|
+
if type(res) == "struct" {
|
|
2577
|
+
return res
|
|
2578
|
+
}
|
|
2579
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2580
|
+
}
|
|
2581
|
+
|
|
2582
|
+
func run_scent_in_repo(args) {
|
|
2583
|
+
command := "cd " + shq(bug_repo) + " && " + shq(kujo_bin) + " run " + shq(scent_script)
|
|
2584
|
+
i := 0
|
|
2585
|
+
while i < len(args) {
|
|
2586
|
+
command = command + " " + shq(args[i])
|
|
2587
|
+
i = i + 1
|
|
2588
|
+
}
|
|
2589
|
+
return run_cmd(command)
|
|
2590
|
+
}
|
|
2591
|
+
|
|
2592
|
+
func require_truthy(value, label) {
|
|
2593
|
+
if value == false || value == 0 {
|
|
2594
|
+
print("Assertion failed: " + label)
|
|
2595
|
+
exit(1)
|
|
2596
|
+
}
|
|
2597
|
+
}
|
|
2598
|
+
|
|
2599
|
+
_ = run_cmd("rm -rf " + shq(bug_root))
|
|
2600
|
+
_ = run_cmd("mkdir -p " + shq(bug_repo + "/docs") + " " + shq(bug_repo + "/src") + " " + shq(bug_repo + "/focus") + " " + shq(bug_out))
|
|
2601
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && git init -q")
|
|
2602
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf '%s\\n' '# Regression repo' > README.md")
|
|
2603
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf '%s\\n' 'EXCLUDED_DOC_MARKER' > docs/excluded.md")
|
|
2604
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf '%s\\n%s\\n%s\\n%s\\n%s\\n%s\\n%s\\n' 'password=supersecret # not redacted' 'print(\"token\", value == 3)' '\"estimated_tokens\": 5' '\"password\": \"abc=def\"' '```' '# injected heading' '```' > src/sample.txt")
|
|
2605
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf 'safe\\000BINARY_SECRET_MARKER' > src/blob")
|
|
2606
|
+
_ = run_cmd("printf '%s\\n' 'SYMLINK_SECRET_MARKER' > " + shq(bug_root + "/outside.txt"))
|
|
2607
|
+
_ = run_cmd("ln -s " + shq(bug_root + "/outside.txt") + " " + shq(bug_repo + "/src/outside-link"))
|
|
2608
|
+
_ = run_cmd("printf '%s\\n' 'STALE_OUTPUT_MARKER' > " + shq(bug_out + "/context.md"))
|
|
2609
|
+
|
|
2610
|
+
pack_res := run_scent_in_repo([
|
|
2611
|
+
"pack", "--task", "review regression markers", "--out", bug_out,
|
|
2612
|
+
"--include", ".", "--exclude", "docs/", "--max-files", "20",
|
|
2613
|
+
"--format", "both", "--json"
|
|
2614
|
+
])
|
|
2615
|
+
assert_equal(pack_res.exitcode, 0)
|
|
2616
|
+
context := read_file(bug_out + "/context.json")
|
|
2617
|
+
context_md := read_file(bug_out + "/context.md")
|
|
2618
|
+
|
|
2619
|
+
# 1: credential lines cannot bypass redaction by containing the word "redacted".
|
|
2620
|
+
assert_equal(contains(context, "supersecret"), 0)
|
|
2621
|
+
# 2: ordinary code mentioning a token is not corrupted.
|
|
2622
|
+
require_truthy(contains(context, "print(\\\"token\\\", value == 3)"), "benign token code preserved")
|
|
2623
|
+
require_truthy(contains(context, "\\\"estimated_tokens\\\": 5"), "non-secret token metadata preserved")
|
|
2624
|
+
# 3: separators in credential values do not corrupt the preserved key prefix.
|
|
2625
|
+
require_truthy(contains(context, "\\\"password\\\": [REDACTED:PASSWORD]"), "credential key prefix preserved")
|
|
2626
|
+
# 4: embedded Markdown fences receive a strictly longer outer fence.
|
|
2627
|
+
require_truthy(contains(context_md, "````text"), "longer Markdown fence emitted")
|
|
2628
|
+
# 5: extensionless files containing NUL bytes are not packed.
|
|
2629
|
+
assert_equal(contains(context, "BINARY_SECRET_MARKER"), 0)
|
|
2630
|
+
|
|
2631
|
+
# 6: root and trailing-slash selectors work as documented.
|
|
2632
|
+
assert_equal(contains(context, "EXCLUDED_DOC_MARKER"), 0)
|
|
2633
|
+
require_truthy(contains(context, "src/sample.txt"), "root include selects nested source")
|
|
2634
|
+
# 7: traversal never follows repository symlinks.
|
|
2635
|
+
assert_equal(contains(context, "SYMLINK_SECRET_MARKER"), 0)
|
|
2636
|
+
symlink_res := run_scent_in_repo(["pack", "--task", "symlink", "--include", "src/outside-link", "--dry-run", "--json"])
|
|
2637
|
+
assert_equal(symlink_res.exitcode, 2)
|
|
2638
|
+
require_truthy(contains(symlink_res.stdout, "cannot contain a symlink"), "explicit symlink selector rejected")
|
|
2639
|
+
# 8: an existing output directory is excluded from the next pack.
|
|
2640
|
+
assert_equal(contains(context, "STALE_OUTPUT_MARKER"), 0)
|
|
2641
|
+
# 9: pack is mandatory instead of silently accepting a flag-only invocation.
|
|
2642
|
+
missing_pack_res := run_scent_in_repo(["--task", "missing subcommand", "--dry-run", "--json"])
|
|
2643
|
+
assert_equal(missing_pack_res.exitcode, 2)
|
|
2644
|
+
require_truthy(contains(missing_pack_res.stdout, "expected subcommand: pack"), "missing subcommand rejected")
|
|
2645
|
+
|
|
2646
|
+
# 10: budget-excluded files have one manifest decision and no stale redaction audit.
|
|
2647
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf '%0600d' 0 | tr '0' 'a' > focus/a.txt")
|
|
2648
|
+
_ = run_cmd("cd " + shq(bug_repo) + " && printf '%s\\n' 'password=budget-secret' > focus/b.txt")
|
|
2649
|
+
budget_res := run_scent_in_repo([
|
|
2650
|
+
"pack", "--task", "budget audit", "--out", budget_out,
|
|
2651
|
+
"--include", "focus", "--budget", "1555", "--max-files", "3",
|
|
2652
|
+
"--format", "json", "--json"
|
|
2653
|
+
])
|
|
2654
|
+
assert_equal(budget_res.exitcode, 0)
|
|
2655
|
+
budget_manifest := read_file(budget_out + "/manifest.json")
|
|
2656
|
+
budget_redactions := read_file(budget_out + "/redactions.json")
|
|
2657
|
+
assert_equal(len(split(budget_manifest, "\"path\": \"focus/b.txt\"")) - 1, 1)
|
|
2658
|
+
assert_equal(contains(budget_redactions, "focus/b.txt"), 0)
|
|
2659
|
+
|
|
2660
|
+
# Explicit directories are rescanned even if the bounded baseline traversal saw them.
|
|
2661
|
+
_ = run_cmd("mkdir -p " + shq(bug_repo + "/a_bulk") + " " + shq(bug_repo + "/z_focus"))
|
|
2662
|
+
_ = run_cmd("printf '%s\\n' 'a_bulk/' > " + shq(bug_repo + "/.gitignore"))
|
|
2663
|
+
_ = run_cmd("i=1; while [ $i -le 2001 ]; do : > " + shq(bug_repo + "/a_bulk") + "/file-$i.txt; i=$((i + 1)); done")
|
|
2664
|
+
_ = run_cmd("printf '%s\\n' 'LATE_INCLUDE_MARKER' > " + shq(bug_repo + "/z_focus/wanted.txt"))
|
|
2665
|
+
late_res := run_scent_in_repo([
|
|
2666
|
+
"pack", "--task", "late include", "--out", late_out,
|
|
2667
|
+
"--include", "z_focus", "--exclude", "a_bulk", "--max-files", "1", "--format", "json", "--json"
|
|
2668
|
+
])
|
|
2669
|
+
assert_equal(late_res.exitcode, 0)
|
|
2670
|
+
late_context := read_file(late_out + "/context.json")
|
|
2671
|
+
require_truthy(contains(late_context, "LATE_INCLUDE_MARKER"), "explicit include survives baseline traversal cap")
|
|
2672
|
+
}
|
|
2673
|
+
|
|
2674
|
+
test "redaction runs before file-size clipping" {
|
|
2675
|
+
audit_scope := true
|
|
2676
|
+
func audit_shq(value) {
|
|
2677
|
+
text := to_string(value)
|
|
2678
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2679
|
+
}
|
|
2680
|
+
|
|
2681
|
+
func audit_run(command) {
|
|
2682
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2683
|
+
if type(result) == "struct" {
|
|
2684
|
+
return result
|
|
2685
|
+
}
|
|
2686
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2687
|
+
}
|
|
2688
|
+
|
|
2689
|
+
func audit_run_scent(repo, argv) {
|
|
2690
|
+
cwd := os_getcwd()
|
|
2691
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2692
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2693
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2694
|
+
i := 0
|
|
2695
|
+
while i < len(argv) {
|
|
2696
|
+
command = command + " " + audit_shq(argv[i])
|
|
2697
|
+
i = i + 1
|
|
2698
|
+
}
|
|
2699
|
+
return audit_run(command)
|
|
2700
|
+
}
|
|
2701
|
+
root := "/tmp/scent_regression_redact_clip"
|
|
2702
|
+
repo := root + "/repo"
|
|
2703
|
+
out := root + "/out"
|
|
2704
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2705
|
+
_ = write_file(repo + "/secret.txt", "sk-ABCDEFGHIJKLMNOPQRSTUVWXYZ123456\n", true)
|
|
2706
|
+
|
|
2707
|
+
result := audit_run_scent(repo, [
|
|
2708
|
+
"pack", "--task", "secret clipping", "--include", "secret.txt",
|
|
2709
|
+
"--out", out, "--max-files", "1", "--max-file-bytes", "10",
|
|
2710
|
+
"--format", "both", "--json"
|
|
2711
|
+
])
|
|
2712
|
+
assert_equal(result.exitcode, 0)
|
|
2713
|
+
context := read_file(out + "/context.json")
|
|
2714
|
+
assert_equal(contains(context, "sk-ABCDEFG"), 0)
|
|
2715
|
+
}
|
|
2716
|
+
|
|
2717
|
+
test "max-file-bytes is a UTF-8 byte limit" {
|
|
2718
|
+
audit_scope := true
|
|
2719
|
+
func audit_shq(value) {
|
|
2720
|
+
text := to_string(value)
|
|
2721
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2722
|
+
}
|
|
2723
|
+
|
|
2724
|
+
func audit_run(command) {
|
|
2725
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2726
|
+
if type(result) == "struct" {
|
|
2727
|
+
return result
|
|
2728
|
+
}
|
|
2729
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2730
|
+
}
|
|
2731
|
+
|
|
2732
|
+
func audit_run_scent(repo, argv) {
|
|
2733
|
+
cwd := os_getcwd()
|
|
2734
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2735
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2736
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2737
|
+
i := 0
|
|
2738
|
+
while i < len(argv) {
|
|
2739
|
+
command = command + " " + audit_shq(argv[i])
|
|
2740
|
+
i = i + 1
|
|
2741
|
+
}
|
|
2742
|
+
return audit_run(command)
|
|
2743
|
+
}
|
|
2744
|
+
root := "/tmp/scent_regression_utf8_bytes"
|
|
2745
|
+
repo := root + "/repo"
|
|
2746
|
+
out := root + "/out"
|
|
2747
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2748
|
+
_ = write_file(repo + "/unicode.txt", "12345678éZ", true)
|
|
2749
|
+
|
|
2750
|
+
result := audit_run_scent(repo, [
|
|
2751
|
+
"pack", "--task", "unicode bytes", "--include", "unicode.txt",
|
|
2752
|
+
"--out", out, "--max-files", "1", "--max-file-bytes", "10",
|
|
2753
|
+
"--format", "json", "--json"
|
|
2754
|
+
])
|
|
2755
|
+
assert_equal(result.exitcode, 0)
|
|
2756
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
2757
|
+
assert_equal(context["selected_files"][0]["content"], "12345678é")
|
|
2758
|
+
assert_equal(context["selected_files"][0]["truncated"], true)
|
|
2759
|
+
}
|
|
2760
|
+
|
|
2761
|
+
test "output reuse removes stale optional context artifacts" {
|
|
2762
|
+
audit_scope := true
|
|
2763
|
+
func audit_shq(value) {
|
|
2764
|
+
text := to_string(value)
|
|
2765
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2766
|
+
}
|
|
2767
|
+
|
|
2768
|
+
func audit_run(command) {
|
|
2769
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2770
|
+
if type(result) == "struct" {
|
|
2771
|
+
return result
|
|
2772
|
+
}
|
|
2773
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2774
|
+
}
|
|
2775
|
+
|
|
2776
|
+
func audit_run_scent(repo, argv) {
|
|
2777
|
+
cwd := os_getcwd()
|
|
2778
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2779
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2780
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2781
|
+
i := 0
|
|
2782
|
+
while i < len(argv) {
|
|
2783
|
+
command = command + " " + audit_shq(argv[i])
|
|
2784
|
+
i = i + 1
|
|
2785
|
+
}
|
|
2786
|
+
return audit_run(command)
|
|
2787
|
+
}
|
|
2788
|
+
root := "/tmp/scent_regression_output_reuse"
|
|
2789
|
+
repo := root + "/repo"
|
|
2790
|
+
out := root + "/out"
|
|
2791
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2792
|
+
_ = write_file(repo + "/README.md", "# fixture\n", true)
|
|
2793
|
+
|
|
2794
|
+
both_result := audit_run_scent(repo, [
|
|
2795
|
+
"pack", "--task", "reuse", "--include", "README.md", "--out", out,
|
|
2796
|
+
"--max-files", "1", "--format", "both", "--json"
|
|
2797
|
+
])
|
|
2798
|
+
assert_equal(both_result.exitcode, 0)
|
|
2799
|
+
assert_equal(file_exists(out + "/context.md"), true)
|
|
2800
|
+
assert_equal(file_exists(out + "/context.json"), true)
|
|
2801
|
+
|
|
2802
|
+
md_result := audit_run_scent(repo, [
|
|
2803
|
+
"pack", "--task", "reuse", "--include", "README.md", "--out", out,
|
|
2804
|
+
"--max-files", "1", "--format", "md", "--json"
|
|
2805
|
+
])
|
|
2806
|
+
assert_equal(md_result.exitcode, 0)
|
|
2807
|
+
assert_equal(file_exists(out + "/context.md"), true)
|
|
2808
|
+
assert_equal(file_exists(out + "/context.json"), false)
|
|
2809
|
+
|
|
2810
|
+
json_result := audit_run_scent(repo, [
|
|
2811
|
+
"pack", "--task", "reuse", "--include", "README.md", "--out", out,
|
|
2812
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
2813
|
+
])
|
|
2814
|
+
assert_equal(json_result.exitcode, 0)
|
|
2815
|
+
assert_equal(file_exists(out + "/context.md"), false)
|
|
2816
|
+
assert_equal(file_exists(out + "/context.json"), true)
|
|
2817
|
+
}
|
|
2818
|
+
|
|
2819
|
+
test "git metadata preserves unusual path names exactly" {
|
|
2820
|
+
audit_scope := true
|
|
2821
|
+
func audit_shq(value) {
|
|
2822
|
+
text := to_string(value)
|
|
2823
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2824
|
+
}
|
|
2825
|
+
|
|
2826
|
+
func audit_run(command) {
|
|
2827
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2828
|
+
if type(result) == "struct" {
|
|
2829
|
+
return result
|
|
2830
|
+
}
|
|
2831
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2832
|
+
}
|
|
2833
|
+
|
|
2834
|
+
func audit_run_scent(repo, argv) {
|
|
2835
|
+
cwd := os_getcwd()
|
|
2836
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2837
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2838
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2839
|
+
i := 0
|
|
2840
|
+
while i < len(argv) {
|
|
2841
|
+
command = command + " " + audit_shq(argv[i])
|
|
2842
|
+
i = i + 1
|
|
2843
|
+
}
|
|
2844
|
+
return audit_run(command)
|
|
2845
|
+
}
|
|
2846
|
+
root := "/tmp/scent_regression_git_names"
|
|
2847
|
+
repo := root + "/repo"
|
|
2848
|
+
out := root + "/out"
|
|
2849
|
+
newline_name := "odd\nname.txt"
|
|
2850
|
+
spaced_name := " leading and trailing .txt "
|
|
2851
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q && git config user.name Test && git config user.email test@example.com")
|
|
2852
|
+
_ = write_file(repo + "/" + newline_name, "seed", true)
|
|
2853
|
+
_ = write_file(repo + "/" + spaced_name, "seed", true)
|
|
2854
|
+
commit_result := audit_run("cd " + audit_shq(repo) + " && git add . && git commit -q -m seed")
|
|
2855
|
+
assert_equal(commit_result.exitcode, 0)
|
|
2856
|
+
_ = write_file(repo + "/" + newline_name, "changed", true)
|
|
2857
|
+
_ = write_file(repo + "/" + spaced_name, "changed", true)
|
|
2858
|
+
|
|
2859
|
+
result := audit_run_scent(repo, [
|
|
2860
|
+
"pack", "--task", "git paths", "--changed", "--out", out,
|
|
2861
|
+
"--max-files", "3", "--format", "json", "--json"
|
|
2862
|
+
])
|
|
2863
|
+
assert_equal(result.exitcode, 0)
|
|
2864
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
2865
|
+
assert_equal(contains(context["changed_files"], newline_name), true)
|
|
2866
|
+
assert_equal(contains(context["changed_files"], spaced_name), true)
|
|
2867
|
+
files := parse_json(read_file(out + "/files.json"))
|
|
2868
|
+
assert_equal(files[0]["path"], spaced_name)
|
|
2869
|
+
assert_equal(files[1]["path"], newline_name)
|
|
2870
|
+
assert_equal(files[0]["is_changed"], 1)
|
|
2871
|
+
assert_equal(files[1]["is_changed"], 1)
|
|
2872
|
+
}
|
|
2873
|
+
|
|
2874
|
+
test "recognized secrets in task text are redacted from artifacts" {
|
|
2875
|
+
audit_scope := true
|
|
2876
|
+
func audit_shq(value) {
|
|
2877
|
+
text := to_string(value)
|
|
2878
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2879
|
+
}
|
|
2880
|
+
|
|
2881
|
+
func audit_run(command) {
|
|
2882
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2883
|
+
if type(result) == "struct" {
|
|
2884
|
+
return result
|
|
2885
|
+
}
|
|
2886
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2887
|
+
}
|
|
2888
|
+
|
|
2889
|
+
func audit_run_scent(repo, argv) {
|
|
2890
|
+
cwd := os_getcwd()
|
|
2891
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2892
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2893
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2894
|
+
i := 0
|
|
2895
|
+
while i < len(argv) {
|
|
2896
|
+
command = command + " " + audit_shq(argv[i])
|
|
2897
|
+
i = i + 1
|
|
2898
|
+
}
|
|
2899
|
+
return audit_run(command)
|
|
2900
|
+
}
|
|
2901
|
+
root := "/tmp/scent_regression_task_secret"
|
|
2902
|
+
repo := root + "/repo"
|
|
2903
|
+
out := root + "/out"
|
|
2904
|
+
secret := "sk-ABCDEFGHIJKLMNOPQRSTUVWXYZ123456"
|
|
2905
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2906
|
+
_ = write_file(repo + "/README.md", "# fixture\n", true)
|
|
2907
|
+
|
|
2908
|
+
result := audit_run_scent(repo, [
|
|
2909
|
+
"pack", "--task", "audit " + secret, "--include", "README.md", "--out", out,
|
|
2910
|
+
"--max-files", "1", "--format", "both", "--json"
|
|
2911
|
+
])
|
|
2912
|
+
assert_equal(result.exitcode, 0)
|
|
2913
|
+
artifacts := ["context.md", "context.json", "manifest.json", "files.json", "redactions.json", "metadata.json"]
|
|
2914
|
+
i := 0
|
|
2915
|
+
while i < len(artifacts) {
|
|
2916
|
+
assert_equal(contains(read_file(out + "/" + artifacts[i]), secret), 0)
|
|
2917
|
+
i = i + 1
|
|
2918
|
+
}
|
|
2919
|
+
assert_equal(contains(read_file(out + "/context.json"), "[REDACTED:TOKEN]"), 1)
|
|
2920
|
+
}
|
|
2921
|
+
|
|
2922
|
+
test "files inventory agrees with selected context files" {
|
|
2923
|
+
audit_scope := true
|
|
2924
|
+
func audit_shq(value) {
|
|
2925
|
+
text := to_string(value)
|
|
2926
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2927
|
+
}
|
|
2928
|
+
|
|
2929
|
+
func audit_run(command) {
|
|
2930
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2931
|
+
if type(result) == "struct" {
|
|
2932
|
+
return result
|
|
2933
|
+
}
|
|
2934
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2935
|
+
}
|
|
2936
|
+
|
|
2937
|
+
func audit_run_scent(repo, argv) {
|
|
2938
|
+
cwd := os_getcwd()
|
|
2939
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2940
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2941
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2942
|
+
i := 0
|
|
2943
|
+
while i < len(argv) {
|
|
2944
|
+
command = command + " " + audit_shq(argv[i])
|
|
2945
|
+
i = i + 1
|
|
2946
|
+
}
|
|
2947
|
+
return audit_run(command)
|
|
2948
|
+
}
|
|
2949
|
+
root := "/tmp/scent_regression_files_selected"
|
|
2950
|
+
repo := root + "/repo"
|
|
2951
|
+
out := root + "/out"
|
|
2952
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2953
|
+
_ = write_file(repo + "/app.txt", "selected content\n", true)
|
|
2954
|
+
|
|
2955
|
+
result := audit_run_scent(repo, [
|
|
2956
|
+
"pack", "--task", "app", "--include", "app.txt", "--out", out,
|
|
2957
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
2958
|
+
])
|
|
2959
|
+
assert_equal(result.exitcode, 0)
|
|
2960
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
2961
|
+
files := parse_json(read_file(out + "/files.json"))
|
|
2962
|
+
assert_equal(len(context["selected_files"]), 1)
|
|
2963
|
+
assert_equal(files[0]["path"], context["selected_files"][0]["path"])
|
|
2964
|
+
assert_equal(files[0]["selected"], true)
|
|
2965
|
+
assert_equal(files[0]["selection_reason"], context["selected_files"][0]["reason"])
|
|
2966
|
+
}
|
|
2967
|
+
|
|
2968
|
+
test "selected unreadable files receive an accurate exclusion reason" {
|
|
2969
|
+
audit_scope := true
|
|
2970
|
+
func audit_shq(value) {
|
|
2971
|
+
text := to_string(value)
|
|
2972
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
2973
|
+
}
|
|
2974
|
+
|
|
2975
|
+
func audit_run(command) {
|
|
2976
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
2977
|
+
if type(result) == "struct" {
|
|
2978
|
+
return result
|
|
2979
|
+
}
|
|
2980
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
2981
|
+
}
|
|
2982
|
+
|
|
2983
|
+
func audit_run_scent(repo, argv) {
|
|
2984
|
+
cwd := os_getcwd()
|
|
2985
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
2986
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
2987
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
2988
|
+
i := 0
|
|
2989
|
+
while i < len(argv) {
|
|
2990
|
+
command = command + " " + audit_shq(argv[i])
|
|
2991
|
+
i = i + 1
|
|
2992
|
+
}
|
|
2993
|
+
return audit_run(command)
|
|
2994
|
+
}
|
|
2995
|
+
root := "/tmp/scent_regression_unreadable"
|
|
2996
|
+
repo := root + "/repo"
|
|
2997
|
+
out := root + "/out"
|
|
2998
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
2999
|
+
_ = write_file(repo + "/locked.txt", "private", true)
|
|
3000
|
+
_ = audit_run("chmod 000 " + audit_shq(repo + "/locked.txt"))
|
|
3001
|
+
|
|
3002
|
+
result := audit_run_scent(repo, [
|
|
3003
|
+
"pack", "--task", "locked", "--include", "locked.txt", "--out", out,
|
|
3004
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
3005
|
+
])
|
|
3006
|
+
_ = audit_run("chmod 600 " + audit_shq(repo + "/locked.txt"))
|
|
3007
|
+
metadata := parse_json(read_file(out + "/metadata.json"))
|
|
3008
|
+
assert_equal(metadata["exit_code"], 1)
|
|
3009
|
+
manifest := parse_json(read_file(out + "/manifest.json"))
|
|
3010
|
+
assert_equal(manifest[0]["path"], "locked.txt")
|
|
3011
|
+
assert_equal(manifest[0]["decision"], "exclude")
|
|
3012
|
+
assert_equal(contains(manifest[0]["reason"], "could not be read"), 1)
|
|
3013
|
+
}
|
|
3014
|
+
|
|
3015
|
+
test "package scripts produce justified validation commands" {
|
|
3016
|
+
audit_scope := true
|
|
3017
|
+
func audit_shq(value) {
|
|
3018
|
+
text := to_string(value)
|
|
3019
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
3020
|
+
}
|
|
3021
|
+
|
|
3022
|
+
func audit_run(command) {
|
|
3023
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
3024
|
+
if type(result) == "struct" {
|
|
3025
|
+
return result
|
|
3026
|
+
}
|
|
3027
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
3028
|
+
}
|
|
3029
|
+
|
|
3030
|
+
func audit_run_scent(repo, argv) {
|
|
3031
|
+
cwd := os_getcwd()
|
|
3032
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
3033
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
3034
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
3035
|
+
i := 0
|
|
3036
|
+
while i < len(argv) {
|
|
3037
|
+
command = command + " " + audit_shq(argv[i])
|
|
3038
|
+
i = i + 1
|
|
3039
|
+
}
|
|
3040
|
+
return audit_run(command)
|
|
3041
|
+
}
|
|
3042
|
+
|
|
3043
|
+
root := "/tmp/scent_regression_package_commands"
|
|
3044
|
+
repo := root + "/repo"
|
|
3045
|
+
out := root + "/out"
|
|
3046
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
3047
|
+
_ = write_file(repo + "/package.json", "{\"scripts\":{\"test\":\"run-tests\",\"lint\":\"run-lint\",\"build\":\"run-build\"}}", true)
|
|
3048
|
+
|
|
3049
|
+
result := audit_run_scent(repo, [
|
|
3050
|
+
"pack", "--task", "commands", "--include", "package.json", "--out", out,
|
|
3051
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
3052
|
+
])
|
|
3053
|
+
assert_equal(result.exitcode, 0)
|
|
3054
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
3055
|
+
assert_equal(context["commands"], ["npm run build", "npm run lint", "npm test"])
|
|
3056
|
+
}
|
|
3057
|
+
|
|
3058
|
+
test "changed-file union deduplicates staged and unstaged paths" {
|
|
3059
|
+
audit_scope := true
|
|
3060
|
+
func audit_shq(value) {
|
|
3061
|
+
text := to_string(value)
|
|
3062
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
3063
|
+
}
|
|
3064
|
+
|
|
3065
|
+
func audit_run(command) {
|
|
3066
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
3067
|
+
if type(result) == "struct" {
|
|
3068
|
+
return result
|
|
3069
|
+
}
|
|
3070
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
3071
|
+
}
|
|
3072
|
+
|
|
3073
|
+
func audit_run_scent(repo, argv) {
|
|
3074
|
+
cwd := os_getcwd()
|
|
3075
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
3076
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
3077
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
3078
|
+
i := 0
|
|
3079
|
+
while i < len(argv) {
|
|
3080
|
+
command = command + " " + audit_shq(argv[i])
|
|
3081
|
+
i = i + 1
|
|
3082
|
+
}
|
|
3083
|
+
return audit_run(command)
|
|
3084
|
+
}
|
|
3085
|
+
|
|
3086
|
+
root := "/tmp/scent_regression_changed_union"
|
|
3087
|
+
repo := root + "/repo"
|
|
3088
|
+
out := root + "/out"
|
|
3089
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q && git config user.name Test && git config user.email test@example.com")
|
|
3090
|
+
_ = write_file(repo + "/both.txt", "seed", true)
|
|
3091
|
+
_ = audit_run("cd " + audit_shq(repo) + " && git add both.txt && git commit -q -m seed")
|
|
3092
|
+
_ = write_file(repo + "/both.txt", "staged", true)
|
|
3093
|
+
_ = audit_run("cd " + audit_shq(repo) + " && git add both.txt")
|
|
3094
|
+
_ = write_file(repo + "/both.txt", "unstaged", true)
|
|
3095
|
+
|
|
3096
|
+
result := audit_run_scent(repo, [
|
|
3097
|
+
"pack", "--task", "git", "--changed", "--out", out,
|
|
3098
|
+
"--max-files", "2", "--format", "json", "--json"
|
|
3099
|
+
])
|
|
3100
|
+
assert_equal(result.exitcode, 0)
|
|
3101
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
3102
|
+
assert_equal(context["git"]["staged_files"], ["both.txt"])
|
|
3103
|
+
assert_equal(context["git"]["unstaged_files"], ["both.txt"])
|
|
3104
|
+
assert_equal(context["changed_files"], ["both.txt"])
|
|
3105
|
+
}
|
|
3106
|
+
|
|
3107
|
+
test "artifact writes replace symlinks instead of following them" {
|
|
3108
|
+
audit_scope := true
|
|
3109
|
+
func audit_shq(value) {
|
|
3110
|
+
text := to_string(value)
|
|
3111
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
3112
|
+
}
|
|
3113
|
+
|
|
3114
|
+
func audit_run(command) {
|
|
3115
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
3116
|
+
if type(result) == "struct" {
|
|
3117
|
+
return result
|
|
3118
|
+
}
|
|
3119
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
3120
|
+
}
|
|
3121
|
+
|
|
3122
|
+
func audit_run_scent(repo, argv) {
|
|
3123
|
+
cwd := os_getcwd()
|
|
3124
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
3125
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
3126
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
3127
|
+
i := 0
|
|
3128
|
+
while i < len(argv) {
|
|
3129
|
+
command = command + " " + audit_shq(argv[i])
|
|
3130
|
+
i = i + 1
|
|
3131
|
+
}
|
|
3132
|
+
return audit_run(command)
|
|
3133
|
+
}
|
|
3134
|
+
|
|
3135
|
+
root := "/tmp/scent_regression_artifact_symlink"
|
|
3136
|
+
repo := root + "/repo"
|
|
3137
|
+
out := root + "/out"
|
|
3138
|
+
victim := root + "/victim.json"
|
|
3139
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " " + audit_shq(out) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
3140
|
+
_ = write_file(repo + "/app.txt", "safe", true)
|
|
3141
|
+
_ = write_file(victim, "VICTIM_SENTINEL", true)
|
|
3142
|
+
link_result := audit_run("ln -s " + audit_shq(victim) + " " + audit_shq(out + "/context.json"))
|
|
3143
|
+
assert_equal(link_result.exitcode, 0)
|
|
3144
|
+
|
|
3145
|
+
result := audit_run_scent(repo, [
|
|
3146
|
+
"pack", "--task", "safe", "--include", "app.txt", "--out", out,
|
|
3147
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
3148
|
+
])
|
|
3149
|
+
assert_equal(result.exitcode, 0)
|
|
3150
|
+
assert_equal(read_file(victim), "VICTIM_SENTINEL")
|
|
3151
|
+
assert_equal(path_is_symlink(out + "/context.json"), false)
|
|
3152
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
3153
|
+
assert_equal(context["selected_files"][0]["path"], "app.txt")
|
|
3154
|
+
}
|
|
3155
|
+
|
|
3156
|
+
test "fallback traversal uses alphabetical directory order" {
|
|
3157
|
+
audit_scope := true
|
|
3158
|
+
func audit_shq(value) {
|
|
3159
|
+
text := to_string(value)
|
|
3160
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
3161
|
+
}
|
|
3162
|
+
|
|
3163
|
+
func audit_run(command) {
|
|
3164
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
3165
|
+
if type(result) == "struct" {
|
|
3166
|
+
return result
|
|
3167
|
+
}
|
|
3168
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
3169
|
+
}
|
|
3170
|
+
|
|
3171
|
+
func audit_run_scent(repo, argv) {
|
|
3172
|
+
cwd := os_getcwd()
|
|
3173
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
3174
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
3175
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
3176
|
+
i := 0
|
|
3177
|
+
while i < len(argv) {
|
|
3178
|
+
command = command + " " + audit_shq(argv[i])
|
|
3179
|
+
i = i + 1
|
|
3180
|
+
}
|
|
3181
|
+
return audit_run(command)
|
|
3182
|
+
}
|
|
3183
|
+
|
|
3184
|
+
root := "/tmp/scent_regression_fallback_sort"
|
|
3185
|
+
repo := root + "/repo"
|
|
3186
|
+
out := root + "/out"
|
|
3187
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
3188
|
+
_ = write_file(repo + "/z.txt", "z", true)
|
|
3189
|
+
_ = write_file(repo + "/a.txt", "a", true)
|
|
3190
|
+
_ = write_file(repo + "/m.txt", "m", true)
|
|
3191
|
+
|
|
3192
|
+
result := audit_run_scent(repo, [
|
|
3193
|
+
"pack", "--task", "unrelated", "--out", out,
|
|
3194
|
+
"--max-files", "3", "--format", "json", "--json"
|
|
3195
|
+
])
|
|
3196
|
+
assert_equal(result.exitcode, 0)
|
|
3197
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
3198
|
+
selected_paths := []
|
|
3199
|
+
i := 0
|
|
3200
|
+
while i < len(context["selected_files"]) {
|
|
3201
|
+
selected_paths = push(selected_paths, context["selected_files"][i]["path"])
|
|
3202
|
+
i = i + 1
|
|
3203
|
+
}
|
|
3204
|
+
assert_equal(selected_paths, ["a.txt", "m.txt", "z.txt"])
|
|
3205
|
+
}
|
|
3206
|
+
|
|
3207
|
+
test "malformed Cargo metadata does not suggest Cargo commands" {
|
|
3208
|
+
audit_scope := true
|
|
3209
|
+
func audit_shq(value) {
|
|
3210
|
+
text := to_string(value)
|
|
3211
|
+
return "'" + replace(text, "'", "'\"'\"'") + "'"
|
|
3212
|
+
}
|
|
3213
|
+
|
|
3214
|
+
func audit_run(command) {
|
|
3215
|
+
result := execute_status(command, {"timeout_ms": 120000})
|
|
3216
|
+
if type(result) == "struct" {
|
|
3217
|
+
return result
|
|
3218
|
+
}
|
|
3219
|
+
return {"success": false, "exitcode": 1, "stdout": "", "stderr": "invalid execute_status result"}
|
|
3220
|
+
}
|
|
3221
|
+
|
|
3222
|
+
func audit_run_scent(repo, argv) {
|
|
3223
|
+
cwd := os_getcwd()
|
|
3224
|
+
kujo_bin := env_or("KUJO_BIN", path_join(cwd, "../kujo/target/release/kujo"))
|
|
3225
|
+
scent_script := env_or("SCENT_SCRIPT", path_join(cwd, "scent.kujo"))
|
|
3226
|
+
command := "cd " + audit_shq(repo) + " && " + audit_shq(kujo_bin) + " run " + audit_shq(scent_script)
|
|
3227
|
+
i := 0
|
|
3228
|
+
while i < len(argv) {
|
|
3229
|
+
command = command + " " + audit_shq(argv[i])
|
|
3230
|
+
i = i + 1
|
|
3231
|
+
}
|
|
3232
|
+
return audit_run(command)
|
|
3233
|
+
}
|
|
3234
|
+
|
|
3235
|
+
root := "/tmp/scent_regression_malformed_cargo"
|
|
3236
|
+
repo := root + "/repo"
|
|
3237
|
+
out := root + "/out"
|
|
3238
|
+
_ = audit_run("rm -rf " + audit_shq(root) + " && mkdir -p " + audit_shq(repo) + " && cd " + audit_shq(repo) + " && git init -q")
|
|
3239
|
+
_ = write_file(repo + "/Cargo.toml", "not = [valid", true)
|
|
3240
|
+
|
|
3241
|
+
result := audit_run_scent(repo, [
|
|
3242
|
+
"pack", "--task", "commands", "--include", "Cargo.toml", "--out", out,
|
|
3243
|
+
"--max-files", "1", "--format", "json", "--json"
|
|
3244
|
+
])
|
|
3245
|
+
assert_equal(result.exitcode, 0)
|
|
3246
|
+
context := parse_json(read_file(out + "/context.json"))
|
|
3247
|
+
assert_equal(context["commands"], [])
|
|
3248
|
+
}
|
|
3249
|
+
|
|
3250
|
+
main()
|