@tea-agent/loop-agent 0.33.4 → 0.33.5
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 +28 -0
- package/dist/executors/dag-pi-executor.js +34 -12
- package/dist/executors/shell-executor.js +26 -66
- package/dist/task/config-types.js +2 -2
- package/dist/worker/console/chat/chat-event-store.js +57 -18
- package/dist/worker/console/chat/routes.js +850 -170
- package/dist/worker/console/static/assets/{index-PzYzcuFG.js → index-CteJFFL2.js} +17 -17
- package/dist/worker/console/static/index.html +1 -1
- package/dist/worker/console/static-src/operator-chat/chat-sse-events.js +33 -8
- package/dist/worker/console/static-src/operator-chat/refs.js +3 -0
- package/dist/worker/console/static-src/operator-chat/useChatSessions.js +3 -0
- package/dist/worker/console/static-src/operator-chat/useChatStream.js +80 -2
- package/dist/workflows/dag/backend-test-markdown-workflow.js +17 -0
- package/dist/workflows/dag/backend-test-module-stem.js +21 -0
- package/dist/workflows/dag/backend-test-pytest-collection.js +44 -0
- package/dist/workflows/dag/backend-test-writer-completeness.js +130 -45
- package/dist/workflows/dag/init-hybrid.js +29 -160
- package/docs/templates/backend-test-dag.json +3 -3
- package/docs/templates/frontend-test-dag.generate-cases.prompt.md +2 -2
- package/docs/templates/frontend-test-dag.json +9 -9
- package/docs/templates/frontend-test-dag.retrieve-context.prompt.md +7 -7
- package/harness.json +1 -1
- package/package.json +1 -1
- package/skills/playwright-cli/SKILL.md +2 -3
|
@@ -6,7 +6,7 @@
|
|
|
6
6
|
<link rel="icon" type="image/svg+xml" href="/favicon.svg" />
|
|
7
7
|
<link rel="stylesheet" href="/inspect/operator-chrome.css" />
|
|
8
8
|
<title>Loop 操作台 · Operator Console</title>
|
|
9
|
-
<script type="module" crossorigin src="/assets/index-
|
|
9
|
+
<script type="module" crossorigin src="/assets/index-CteJFFL2.js"></script>
|
|
10
10
|
<link rel="stylesheet" crossorigin href="/assets/index-CnUXAqxG.css">
|
|
11
11
|
</head>
|
|
12
12
|
<body>
|
|
@@ -37,9 +37,7 @@ export function createChatSseEventApplier(deps) {
|
|
|
37
37
|
interviewSessionId: typeof payload.interviewSessionId === "string"
|
|
38
38
|
? payload.interviewSessionId
|
|
39
39
|
: current?.interviewSessionId,
|
|
40
|
-
taskId: typeof payload.taskId === "string"
|
|
41
|
-
? payload.taskId
|
|
42
|
-
: current?.taskId,
|
|
40
|
+
taskId: typeof payload.taskId === "string" ? payload.taskId : current?.taskId,
|
|
43
41
|
state: typeof payload.state === "string" ? payload.state : current?.state,
|
|
44
42
|
question: payload.question && typeof payload.question === "object"
|
|
45
43
|
? payload.question
|
|
@@ -50,9 +48,7 @@ export function createChatSseEventApplier(deps) {
|
|
|
50
48
|
if (eventName === "draft") {
|
|
51
49
|
setInterview((current) => ({
|
|
52
50
|
...(current ?? {}),
|
|
53
|
-
taskId: typeof payload.taskId === "string"
|
|
54
|
-
? payload.taskId
|
|
55
|
-
: current?.taskId,
|
|
51
|
+
taskId: typeof payload.taskId === "string" ? payload.taskId : current?.taskId,
|
|
56
52
|
draft: payload.draft && typeof payload.draft === "object"
|
|
57
53
|
? payload.draft
|
|
58
54
|
: current?.draft,
|
|
@@ -95,15 +91,44 @@ export function createChatSseEventApplier(deps) {
|
|
|
95
91
|
return;
|
|
96
92
|
}
|
|
97
93
|
if (eventName === "agent_start") {
|
|
94
|
+
// P2: a replayed agent_start must not resurrect an already-finished
|
|
95
|
+
// turn (aborted turns have no agent_settled in the ring, so replay
|
|
96
|
+
// would otherwise leave the UI streaming forever). Only the server's
|
|
97
|
+
// active turn may flip streaming on during replay; every other
|
|
98
|
+
// agent_start (including all of them when the server reports no
|
|
99
|
+
// active turn) is stale and ignored.
|
|
100
|
+
if (replay &&
|
|
101
|
+
turnId !== undefined &&
|
|
102
|
+
turnId !== refs.activeTurnIdRef.current) {
|
|
103
|
+
return;
|
|
104
|
+
}
|
|
105
|
+
if (turnId)
|
|
106
|
+
refs.activeTurnIdRef.current = turnId;
|
|
98
107
|
setStreaming(true);
|
|
99
108
|
setStreamingAssistantId(resolvedAssistantId);
|
|
100
109
|
return;
|
|
101
110
|
}
|
|
102
|
-
if (eventName === "agent_settled") {
|
|
111
|
+
if (eventName === "agent_settled" || eventName === "agent_end") {
|
|
112
|
+
// Both are terminal: agent_end = SDK run finished, agent_settled =
|
|
113
|
+
// post-turn continuation idle. Clear streaming and drop an empty
|
|
114
|
+
// placeholder bubble so a settled turn never shows a stuck "…" row.
|
|
103
115
|
setStreaming(false);
|
|
104
116
|
setStreamingAssistantId(null);
|
|
105
|
-
if (
|
|
117
|
+
if (eventName === "agent_settled" &&
|
|
118
|
+
!replay &&
|
|
119
|
+
typeof document !== "undefined" &&
|
|
120
|
+
document.hidden) {
|
|
106
121
|
setUnread((current) => current + 1);
|
|
122
|
+
}
|
|
123
|
+
setMessages((m) => m.filter((msg) => !(msg.id === resolvedAssistantId &&
|
|
124
|
+
msg.role === "assistant" &&
|
|
125
|
+
!msg.text.trim())));
|
|
126
|
+
// Close out tool calls that never got a tool_result (abort/fail):
|
|
127
|
+
// without this the process rail keeps showing "运行中" forever even
|
|
128
|
+
// though the turn is over.
|
|
129
|
+
setToolCalls((prev) => prev.map((t) => t.status === "running" && t.assistantMessageId === resolvedAssistantId
|
|
130
|
+
? { ...t, status: "error", finishedAt: Date.now() }
|
|
131
|
+
: t));
|
|
107
132
|
return;
|
|
108
133
|
}
|
|
109
134
|
if (eventName === "operation-ref" || eventName === "operation-status") {
|
|
@@ -10,6 +10,7 @@ export function useChatRefs() {
|
|
|
10
10
|
const seenEventIdsRef = useRef(new Set());
|
|
11
11
|
const turnAssistantIdsRef = useRef(new Map());
|
|
12
12
|
const reconnectAbortRef = useRef(null);
|
|
13
|
+
const activeTurnIdRef = useRef(null);
|
|
13
14
|
const usageResponseKeysRef = useRef(new Set());
|
|
14
15
|
const isComposingRef = useRef(false);
|
|
15
16
|
const compositionEndedAtRef = useRef(0);
|
|
@@ -26,6 +27,7 @@ export function useChatRefs() {
|
|
|
26
27
|
seenEventIdsRef,
|
|
27
28
|
turnAssistantIdsRef,
|
|
28
29
|
reconnectAbortRef,
|
|
30
|
+
activeTurnIdRef,
|
|
29
31
|
usageResponseKeysRef,
|
|
30
32
|
isComposingRef,
|
|
31
33
|
compositionEndedAtRef,
|
|
@@ -40,6 +42,7 @@ export function useChatRefs() {
|
|
|
40
42
|
seenEventIdsRef,
|
|
41
43
|
turnAssistantIdsRef,
|
|
42
44
|
reconnectAbortRef,
|
|
45
|
+
activeTurnIdRef,
|
|
43
46
|
usageResponseKeysRef,
|
|
44
47
|
isComposingRef,
|
|
45
48
|
compositionEndedAtRef,
|
|
@@ -23,6 +23,7 @@ export function useChatSessions(params) {
|
|
|
23
23
|
refs.lastEventIdRef.current = null;
|
|
24
24
|
refs.seenEventIdsRef.current = new Set();
|
|
25
25
|
refs.turnAssistantIdsRef.current = new Map();
|
|
26
|
+
refs.activeTurnIdRef.current = null;
|
|
26
27
|
resetForNewSession();
|
|
27
28
|
setInput("");
|
|
28
29
|
setPendingImages([]);
|
|
@@ -125,6 +126,7 @@ export function useChatSessions(params) {
|
|
|
125
126
|
};
|
|
126
127
|
refs.sessionRef.current = next;
|
|
127
128
|
refs.streamingRef.current = Boolean(body.session.activeTurnId);
|
|
129
|
+
refs.activeTurnIdRef.current = body.session.activeTurnId ?? null;
|
|
128
130
|
refs.seenEventIdsRef.current = new Set();
|
|
129
131
|
refs.turnAssistantIdsRef.current = new Map();
|
|
130
132
|
refs.lastEventIdRef.current = null;
|
|
@@ -192,6 +194,7 @@ export function useChatSessions(params) {
|
|
|
192
194
|
// from the durable store, but tool/gate/operation projections live
|
|
193
195
|
// in the event ring and must replay once. applySseBlock dedups by
|
|
194
196
|
// eventId and marks replay so usage/unread/sound do not double-count.
|
|
197
|
+
refs.activeTurnIdRef.current = body.session.activeTurnId ?? null;
|
|
195
198
|
refs.seenEventIdsRef.current = new Set();
|
|
196
199
|
refs.turnAssistantIdsRef.current = new Map();
|
|
197
200
|
refs.lastEventIdRef.current = null;
|
|
@@ -123,7 +123,84 @@ export function useChatStream(params) {
|
|
|
123
123
|
refs.streamingRef.current = false;
|
|
124
124
|
setStreaming(false);
|
|
125
125
|
setStreamingAssistantId(null);
|
|
126
|
-
|
|
126
|
+
// User-initiated stop: immediately drop the still-empty placeholder
|
|
127
|
+
// bubble (no text yet) instead of leaving a forever "…" row (pi-web
|
|
128
|
+
// hides empty assistant rows on abort).
|
|
129
|
+
setMessages((m) => m.filter((msg) => !(msg.id === refs.assistantIdRef.current &&
|
|
130
|
+
msg.role === "assistant" &&
|
|
131
|
+
!msg.text.trim())));
|
|
132
|
+
}, [refs, setStreaming, setStreamingAssistantId, setMessages]);
|
|
133
|
+
// Reconcile client streaming state with the server (pi-web get_state loop).
|
|
134
|
+
// When SSE events are missed (network drop, backgrounded tab, half-open
|
|
135
|
+
// connection) agent_end/agent_settled never arrive and the UI stays
|
|
136
|
+
// streaming forever. If the server reports no active turn while we still
|
|
137
|
+
// think a turn is streaming, force the same finish a terminal event would
|
|
138
|
+
// have produced. Also refreshes activeTurnIdRef so replayed agent_start
|
|
139
|
+
// frames are judged against current server truth.
|
|
140
|
+
useEffect(() => {
|
|
141
|
+
if (!session || !streaming)
|
|
142
|
+
return;
|
|
143
|
+
const sid = session.sessionId;
|
|
144
|
+
let cancelled = false;
|
|
145
|
+
const sync = async () => {
|
|
146
|
+
if (cancelled)
|
|
147
|
+
return;
|
|
148
|
+
try {
|
|
149
|
+
const res = await fetch(`${origin}/api/operator/v1/chat/sessions/${encodeURIComponent(sid)}/state`, { credentials: "include" });
|
|
150
|
+
if (!res.ok)
|
|
151
|
+
return;
|
|
152
|
+
const body = (await res.json());
|
|
153
|
+
const activeTurnId = body.state?.activeTurnId ?? null;
|
|
154
|
+
refs.activeTurnIdRef.current = activeTurnId;
|
|
155
|
+
if (activeTurnId !== null || !refs.streamingRef.current)
|
|
156
|
+
return;
|
|
157
|
+
// Server idle while the client still thinks a turn is streaming.
|
|
158
|
+
// Skip when an inline prompt request is still in flight — the
|
|
159
|
+
// server may not have created its turn lease yet (POST prompt
|
|
160
|
+
// creates the turn server-side after the request lands), and
|
|
161
|
+
// force-finishing then would kill a just-sent prompt.
|
|
162
|
+
if (refs.inlineStreamSessionIdsRef.current.has(sid))
|
|
163
|
+
return;
|
|
164
|
+
const pendingAssistantId = refs.assistantIdRef.current;
|
|
165
|
+
refs.streamingRef.current = false;
|
|
166
|
+
setStreaming(false);
|
|
167
|
+
setStreamingAssistantId(null);
|
|
168
|
+
refs.abortRef.current?.abort();
|
|
169
|
+
refs.abortRef.current = null;
|
|
170
|
+
refs.assistantIdRef.current = null;
|
|
171
|
+
setMessages((m) => m.filter((msg) => !(msg.id === pendingAssistantId &&
|
|
172
|
+
msg.role === "assistant" &&
|
|
173
|
+
!msg.text.trim())));
|
|
174
|
+
}
|
|
175
|
+
catch {
|
|
176
|
+
// Network still down — the next poll / visibility / online tick retries.
|
|
177
|
+
}
|
|
178
|
+
};
|
|
179
|
+
const onVisible = () => {
|
|
180
|
+
if (document.visibilityState === "visible")
|
|
181
|
+
void sync();
|
|
182
|
+
};
|
|
183
|
+
const onOnline = () => {
|
|
184
|
+
void sync();
|
|
185
|
+
};
|
|
186
|
+
const interval = setInterval(() => void sync(), 15_000);
|
|
187
|
+
document.addEventListener("visibilitychange", onVisible);
|
|
188
|
+
window.addEventListener("online", onOnline);
|
|
189
|
+
return () => {
|
|
190
|
+
cancelled = true;
|
|
191
|
+
clearInterval(interval);
|
|
192
|
+
document.removeEventListener("visibilitychange", onVisible);
|
|
193
|
+
window.removeEventListener("online", onOnline);
|
|
194
|
+
};
|
|
195
|
+
}, [
|
|
196
|
+
origin,
|
|
197
|
+
refs,
|
|
198
|
+
session,
|
|
199
|
+
streaming,
|
|
200
|
+
setStreaming,
|
|
201
|
+
setStreamingAssistantId,
|
|
202
|
+
setMessages,
|
|
203
|
+
]);
|
|
127
204
|
/** Open the reconnectable GET /events stream for a session and apply events
|
|
128
205
|
* until it closes, then reopen with the latest Last-Event-ID (M1-T04/T05).
|
|
129
206
|
*
|
|
@@ -194,7 +271,8 @@ export function useChatStream(params) {
|
|
|
194
271
|
// leak a dangling connection. It resumes after settle/stop to catch the tail.
|
|
195
272
|
useEffect(() => {
|
|
196
273
|
if (!session ||
|
|
197
|
-
(streaming &&
|
|
274
|
+
(streaming &&
|
|
275
|
+
refs.inlineStreamSessionIdsRef.current.has(session.sessionId)))
|
|
198
276
|
return;
|
|
199
277
|
const ac = new AbortController();
|
|
200
278
|
refs.reconnectAbortRef.current = ac;
|
|
@@ -5,6 +5,7 @@ import { access, mkdir, readFile, readdir, writeFile } from "node:fs/promises";
|
|
|
5
5
|
import { createWriteStream } from "node:fs";
|
|
6
6
|
import path from "node:path";
|
|
7
7
|
import { promisify } from "node:util";
|
|
8
|
+
import { isPriorityOnlyBackendTestModuleStem } from "./backend-test-module-stem.js";
|
|
8
9
|
import { parseJacocoXml, } from "./backend-test-coverage-contract.js";
|
|
9
10
|
import { backendTestCaseManifestSchema, computeCaseManifestCoverageSummary, } from "./backend-test-case-manifest.js";
|
|
10
11
|
const execFileAsync = promisify(execFile);
|
|
@@ -398,9 +399,13 @@ function markdownSecretFindings(relativeFile, markdown) {
|
|
|
398
399
|
export function hasBlockingBackendMarkdownSafetyFindings(report) {
|
|
399
400
|
return /^## Safety Status\s*\r?\n\s*BLOCKED\s*$/m.test(report);
|
|
400
401
|
}
|
|
402
|
+
export function hasBlockingBackendMarkdownModuleStemFindings(report) {
|
|
403
|
+
return /^## Module Stem Status\s*\r?\n\s*BLOCKED\s*$/m.test(report);
|
|
404
|
+
}
|
|
401
405
|
export async function validateBackendMarkdownCases(input) {
|
|
402
406
|
const findings = [];
|
|
403
407
|
const safetyFindings = [];
|
|
408
|
+
const moduleStemFindings = [];
|
|
404
409
|
let files = [];
|
|
405
410
|
try {
|
|
406
411
|
files = await markdownFiles(input.workspaceRoot);
|
|
@@ -419,6 +424,9 @@ export async function validateBackendMarkdownCases(input) {
|
|
|
419
424
|
let caseCount = 0;
|
|
420
425
|
for (const file of moduleFiles) {
|
|
421
426
|
const relativeFile = path.relative(input.workspaceRoot, file).replaceAll(path.sep, "/");
|
|
427
|
+
if (isPriorityOnlyBackendTestModuleStem(path.basename(file, path.extname(file)))) {
|
|
428
|
+
moduleStemFindings.push(`priority-only-module-stem: ${relativeFile}; regenerate the README Module Index with a stable business resource/domain stem`);
|
|
429
|
+
}
|
|
422
430
|
const markdown = await readFile(file, "utf8");
|
|
423
431
|
safetyFindings.push(...markdownSecretFindings(relativeFile, markdown));
|
|
424
432
|
if (!markdown.trim()) {
|
|
@@ -479,6 +487,7 @@ export async function validateBackendMarkdownCases(input) {
|
|
|
479
487
|
if (missing.length > 0) {
|
|
480
488
|
findings.push(`required AC IDs are not covered by final Markdown cases: ${missing.join(", ")}`);
|
|
481
489
|
}
|
|
490
|
+
findings.push(...moduleStemFindings);
|
|
482
491
|
const status = findings.length === 0 ? "PASS" : "FAIL";
|
|
483
492
|
const report = [
|
|
484
493
|
"# Backend Markdown Case Validation",
|
|
@@ -497,6 +506,14 @@ export async function validateBackendMarkdownCases(input) {
|
|
|
497
506
|
"- Source-reference validity check: not applied",
|
|
498
507
|
"- Markdown quality sensitive-information finding: not advisory; see Safety Status",
|
|
499
508
|
"",
|
|
509
|
+
"## Module Stem Status",
|
|
510
|
+
"",
|
|
511
|
+
moduleStemFindings.length > 0 ? "BLOCKED" : "PASS",
|
|
512
|
+
"",
|
|
513
|
+
...(moduleStemFindings.length > 0
|
|
514
|
+
? moduleStemFindings.map((finding) => `- ${finding}`)
|
|
515
|
+
: ["- All Markdown modules use non-priority business stems."]),
|
|
516
|
+
"",
|
|
500
517
|
"## Safety Status",
|
|
501
518
|
"",
|
|
502
519
|
safetyFindings.length > 0 ? "BLOCKED" : "PASS",
|
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
import path from "node:path";
|
|
2
|
+
const PRIORITY_ONLY_MODULE_STEMS = new Set(["p0", "p1", "p2"]);
|
|
3
|
+
export function normalizeBackendTestModuleStemCandidate(raw) {
|
|
4
|
+
return path.basename(raw)
|
|
5
|
+
.replace(/\.(?:md|py)$/i, "")
|
|
6
|
+
.toLowerCase()
|
|
7
|
+
.replace(/[^a-z0-9]+/g, "_")
|
|
8
|
+
.replace(/^_+|_+$/g, "")
|
|
9
|
+
.replace(/_+/g, "_");
|
|
10
|
+
}
|
|
11
|
+
export function isPriorityOnlyBackendTestModuleStem(raw) {
|
|
12
|
+
return PRIORITY_ONLY_MODULE_STEMS.has(normalizeBackendTestModuleStemCandidate(raw));
|
|
13
|
+
}
|
|
14
|
+
export function isPriorityOnlyBackendPytestScript(script) {
|
|
15
|
+
const basename = path.posix.basename(script.replaceAll("\\", "/"));
|
|
16
|
+
const match = /^test_(.+)\.py$/i.exec(basename);
|
|
17
|
+
return match ? isPriorityOnlyBackendTestModuleStem(match[1]) : false;
|
|
18
|
+
}
|
|
19
|
+
export function priorityOnlyBackendPytestScripts(paths) {
|
|
20
|
+
return [...new Set(paths.map((item) => item.replaceAll("\\", "/")).filter(isPriorityOnlyBackendPytestScript))].sort();
|
|
21
|
+
}
|
|
@@ -2,6 +2,7 @@ import { createHash } from "node:crypto";
|
|
|
2
2
|
import { readdir, readFile, writeFile, mkdir } from "node:fs/promises";
|
|
3
3
|
import path from "node:path";
|
|
4
4
|
import { z } from "zod";
|
|
5
|
+
import { priorityOnlyBackendPytestScripts } from "./backend-test-module-stem.js";
|
|
5
6
|
const SHA256 = /^[a-f0-9]{64}$/;
|
|
6
7
|
const MAX_DIAGNOSTIC_CHARS = 12_000;
|
|
7
8
|
export const backendPytestCollectionFindingSchema = z.object({
|
|
@@ -273,6 +274,41 @@ export function assessBackendPytestCollection(input) {
|
|
|
273
274
|
fixtureStderrExcerpt: "",
|
|
274
275
|
});
|
|
275
276
|
}
|
|
277
|
+
export function assessPriorityOnlyBackendPytestModules(input) {
|
|
278
|
+
const invalidScripts = priorityOnlyBackendPytestScripts(input.inventory.assetFiles);
|
|
279
|
+
if (invalidScripts.length === 0)
|
|
280
|
+
return undefined;
|
|
281
|
+
return backendPytestCollectionFactsSchema.parse({
|
|
282
|
+
schemaId: "backend-test-pytest-collection-v3",
|
|
283
|
+
phase: input.phase,
|
|
284
|
+
status: "BLOCKED",
|
|
285
|
+
repairEligible: false,
|
|
286
|
+
repairAttempt: input.phase === "final" ? 1 : 0,
|
|
287
|
+
collectionAttempted: false,
|
|
288
|
+
fixtureResolutionAttempted: false,
|
|
289
|
+
fixtureResolutionStatus: "NOT_RUN",
|
|
290
|
+
fixtureResolutionExitCode: null,
|
|
291
|
+
repairPaths: [],
|
|
292
|
+
mappedScripts: [...input.mappedScripts],
|
|
293
|
+
existingMappedScripts: input.inventory.mappedScripts,
|
|
294
|
+
missingMappedScripts: [],
|
|
295
|
+
assetFiles: input.inventory.assetFiles,
|
|
296
|
+
inputHashes: input.inventory.inputHashes,
|
|
297
|
+
pytestExitCode: null,
|
|
298
|
+
collectedItemCount: 0,
|
|
299
|
+
collectedItemIds: [],
|
|
300
|
+
findings: invalidScripts.map((script) => ({
|
|
301
|
+
kind: "priority-only-pytest-module",
|
|
302
|
+
classification: "test-asset-defect",
|
|
303
|
+
repairability: "blocked",
|
|
304
|
+
detail: `priority-only pytest module is forbidden and cannot be auto-renamed: ${script}`,
|
|
305
|
+
})),
|
|
306
|
+
stdoutExcerpt: "",
|
|
307
|
+
stderrExcerpt: "",
|
|
308
|
+
fixtureStdoutExcerpt: "",
|
|
309
|
+
fixtureStderrExcerpt: "",
|
|
310
|
+
});
|
|
311
|
+
}
|
|
276
312
|
export function assessMissingBackendPytestScripts(input) {
|
|
277
313
|
return backendPytestCollectionFactsSchema.parse({
|
|
278
314
|
schemaId: "backend-test-pytest-collection-v3",
|
|
@@ -408,6 +444,10 @@ export async function materializeBackendTestExecutionReadiness(input) {
|
|
|
408
444
|
throw new Error("backend-test execution readiness requires effective collection and fixture-resolution PASS");
|
|
409
445
|
}
|
|
410
446
|
const current = await buildBackendPytestAssetInventory(input.workspaceRoot, input.effective.mappedScripts);
|
|
447
|
+
const priorityOnly = priorityOnlyBackendPytestScripts(current.assetFiles);
|
|
448
|
+
if (priorityOnly.length > 0) {
|
|
449
|
+
throw new Error(`priority-only-pytest-module: ${priorityOnly.join(", ")}`);
|
|
450
|
+
}
|
|
411
451
|
assertSameInventory(input.effective, current);
|
|
412
452
|
const status = input.scenarioParamStatus === "FAIL"
|
|
413
453
|
? "BLOCKED"
|
|
@@ -442,6 +482,10 @@ export async function assertBackendTestExecutionReadinessFresh(workspaceRoot, re
|
|
|
442
482
|
throw new Error(`backend pytest execution readiness is ${readiness.status}`);
|
|
443
483
|
}
|
|
444
484
|
const current = await buildBackendPytestAssetInventory(workspaceRoot, readiness.mappedScripts);
|
|
485
|
+
const priorityOnly = priorityOnlyBackendPytestScripts(current.assetFiles);
|
|
486
|
+
if (priorityOnly.length > 0) {
|
|
487
|
+
throw new Error(`priority-only-pytest-module: ${priorityOnly.join(", ")}`);
|
|
488
|
+
}
|
|
445
489
|
assertSameInventory({ mappedScripts: readiness.mappedScripts, assetFiles: Object.keys(readiness.assetHashes), inputHashes: readiness.assetHashes }, current);
|
|
446
490
|
}
|
|
447
491
|
export async function assertBackendPytestCollectionFresh(workspaceRoot, effective) {
|
|
@@ -2,6 +2,7 @@ import { createHash } from "node:crypto";
|
|
|
2
2
|
import { access, mkdir, readFile, readdir, writeFile } from "node:fs/promises";
|
|
3
3
|
import path from "node:path";
|
|
4
4
|
import { z } from "zod";
|
|
5
|
+
import { isPriorityOnlyBackendTestModuleStem } from "./backend-test-module-stem.js";
|
|
5
6
|
/**
|
|
6
7
|
* Maps a backend-test generation writer task id to its writer-progress role.
|
|
7
8
|
* Returns undefined for nodes that are not completeness-gated generators.
|
|
@@ -96,31 +97,26 @@ function hasMarkdownTable(section, headerNeedle) {
|
|
|
96
97
|
* phantom missing modules when the model discusses forbidden filenames
|
|
97
98
|
* ("do not create 1.md / 9.md") inside the README body.
|
|
98
99
|
*/
|
|
99
|
-
function
|
|
100
|
+
function moduleStemInvalidReason(raw) {
|
|
100
101
|
if (!raw)
|
|
101
|
-
return
|
|
102
|
+
return "invalid-syntax";
|
|
102
103
|
const stem = normalizeBackendTestModuleStem(raw);
|
|
104
|
+
if (isPriorityOnlyBackendTestModuleStem(stem))
|
|
105
|
+
return "priority-only-module-stem";
|
|
103
106
|
if (stem.toLowerCase() === "readme")
|
|
104
|
-
return
|
|
107
|
+
return "reserved-module-stem";
|
|
105
108
|
if (!/^[a-z][a-z0-9_]*$/.test(stem))
|
|
106
|
-
return
|
|
109
|
+
return "invalid-syntax";
|
|
107
110
|
if (/^(?:be|tp|ac|req|br)[_-]/i.test(stem))
|
|
108
|
-
return
|
|
109
|
-
return
|
|
111
|
+
return "case-like-module-stem";
|
|
112
|
+
return undefined;
|
|
110
113
|
}
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
export function extractModuleStemsFromReadme(readme) {
|
|
118
|
-
const stems = [];
|
|
119
|
-
// Only trust testcase/md/<stem>.md mentions that appear inside markdown
|
|
120
|
-
// table rows (`| ... testcase/md/x.md ... |`) or as canonical relative
|
|
121
|
-
// links (`[label](./x.md)`). Free-form prose mentions such as a recovery
|
|
122
|
-
// note listing `testcase/md/1.md` must NOT be treated as authoritative
|
|
123
|
-
// module references, otherwise the gate invents phantom missing modules.
|
|
114
|
+
function looksLikeValidModuleStem(raw) {
|
|
115
|
+
return moduleStemInvalidReason(raw) === undefined;
|
|
116
|
+
}
|
|
117
|
+
/** Inspect only authoritative README module-index candidates. */
|
|
118
|
+
export function inspectModuleStemsFromReadme(readme) {
|
|
119
|
+
const candidates = [];
|
|
124
120
|
const tableRowLines = readme
|
|
125
121
|
.replaceAll("\r\n", "\n")
|
|
126
122
|
.replaceAll("\r", "\n")
|
|
@@ -128,19 +124,42 @@ export function extractModuleStemsFromReadme(readme) {
|
|
|
128
124
|
.filter((line) => line.includes("|"));
|
|
129
125
|
for (const line of tableRowLines) {
|
|
130
126
|
for (const match of line.matchAll(/`?testcase\/md\/([A-Za-z0-9_.-]+)\.md`?/g)) {
|
|
131
|
-
|
|
132
|
-
stems.push(match[1]);
|
|
127
|
+
candidates.push(match[1]);
|
|
133
128
|
}
|
|
134
129
|
for (const match of line.matchAll(/\|\s*`?([A-Za-z0-9_.-]+)`?\s*\|\s*`?testcase\/test_/g)) {
|
|
135
|
-
|
|
136
|
-
|
|
130
|
+
// This compatibility shape can also capture a numeric count column.
|
|
131
|
+
// Keep valid stems and the priority-only defect we diagnose, while
|
|
132
|
+
// ignoring non-stem incidental cells such as `1`.
|
|
133
|
+
if (looksLikeValidModuleStem(match[1]) || isPriorityOnlyBackendTestModuleStem(match[1])) {
|
|
134
|
+
candidates.push(match[1]);
|
|
135
|
+
}
|
|
137
136
|
}
|
|
138
137
|
}
|
|
139
138
|
for (const match of readme.matchAll(/\[[^\]]+\]\(\.\/([A-Za-z0-9_.-]+)\.md\)/g)) {
|
|
140
|
-
|
|
141
|
-
stems.push(match[1]);
|
|
139
|
+
candidates.push(match[1]);
|
|
142
140
|
}
|
|
143
|
-
|
|
141
|
+
const uniqueCandidates = orderedUnique(candidates);
|
|
142
|
+
const invalid = uniqueCandidates.flatMap((raw) => {
|
|
143
|
+
const reasonCode = moduleStemInvalidReason(raw);
|
|
144
|
+
return reasonCode
|
|
145
|
+
? [{ raw, normalized: normalizeBackendTestModuleStem(raw), reasonCode }]
|
|
146
|
+
: [];
|
|
147
|
+
});
|
|
148
|
+
return {
|
|
149
|
+
candidates: uniqueCandidates,
|
|
150
|
+
validStems: orderedUnique(uniqueCandidates
|
|
151
|
+
.filter(looksLikeValidModuleStem)
|
|
152
|
+
.map((stem) => normalizeBackendTestModuleStem(stem))),
|
|
153
|
+
invalid,
|
|
154
|
+
};
|
|
155
|
+
}
|
|
156
|
+
/**
|
|
157
|
+
* Exported for reuse by callers that only need the trusted shard set.
|
|
158
|
+
* Completeness gates must use inspectModuleStemsFromReadme so invalid
|
|
159
|
+
* authoritative candidates cannot be silently dropped.
|
|
160
|
+
*/
|
|
161
|
+
export function extractModuleStemsFromReadme(readme) {
|
|
162
|
+
return inspectModuleStemsFromReadme(readme).validStems;
|
|
144
163
|
}
|
|
145
164
|
async function listMarkdownModules(workspaceRoot) {
|
|
146
165
|
const root = path.join(workspaceRoot, "testcase", "md");
|
|
@@ -242,12 +261,80 @@ function moduleStructuralDetail(result) {
|
|
|
242
261
|
return "module file is structurally complete";
|
|
243
262
|
return `module file is empty or has structural issues: ${result.reasons.join("; ")}`;
|
|
244
263
|
}
|
|
264
|
+
function pythonStructureOutsideStrings(source) {
|
|
265
|
+
const input = source.replace(/\r\n/g, "\n");
|
|
266
|
+
let output = "";
|
|
267
|
+
let index = 0;
|
|
268
|
+
let quote;
|
|
269
|
+
let triple = false;
|
|
270
|
+
let comment = false;
|
|
271
|
+
while (index < input.length) {
|
|
272
|
+
const char = input[index];
|
|
273
|
+
if (comment) {
|
|
274
|
+
if (char === "\n") {
|
|
275
|
+
comment = false;
|
|
276
|
+
output += "\n";
|
|
277
|
+
}
|
|
278
|
+
else
|
|
279
|
+
output += " ";
|
|
280
|
+
index += 1;
|
|
281
|
+
continue;
|
|
282
|
+
}
|
|
283
|
+
if (quote) {
|
|
284
|
+
if (char === "\\") {
|
|
285
|
+
output += " ";
|
|
286
|
+
if (index + 1 < input.length)
|
|
287
|
+
output += input[index + 1] === "\n" ? "\n" : " ";
|
|
288
|
+
index += 2;
|
|
289
|
+
continue;
|
|
290
|
+
}
|
|
291
|
+
if (triple && input.slice(index, index + 3) === quote.repeat(3)) {
|
|
292
|
+
output += " ";
|
|
293
|
+
index += 3;
|
|
294
|
+
quote = undefined;
|
|
295
|
+
triple = false;
|
|
296
|
+
continue;
|
|
297
|
+
}
|
|
298
|
+
if (!triple && char === quote) {
|
|
299
|
+
output += " ";
|
|
300
|
+
index += 1;
|
|
301
|
+
quote = undefined;
|
|
302
|
+
continue;
|
|
303
|
+
}
|
|
304
|
+
if (!triple && char === "\n")
|
|
305
|
+
return { text: output, stringsClosed: false };
|
|
306
|
+
output += char === "\n" ? "\n" : " ";
|
|
307
|
+
index += 1;
|
|
308
|
+
continue;
|
|
309
|
+
}
|
|
310
|
+
if (char === "#") {
|
|
311
|
+
comment = true;
|
|
312
|
+
output += " ";
|
|
313
|
+
index += 1;
|
|
314
|
+
continue;
|
|
315
|
+
}
|
|
316
|
+
if (char === "'" || char === '"') {
|
|
317
|
+
quote = char;
|
|
318
|
+
triple = input.slice(index, index + 3) === char.repeat(3);
|
|
319
|
+
output += triple ? " " : " ";
|
|
320
|
+
index += triple ? 3 : 1;
|
|
321
|
+
continue;
|
|
322
|
+
}
|
|
323
|
+
output += char;
|
|
324
|
+
index += 1;
|
|
325
|
+
}
|
|
326
|
+
return { text: output, stringsClosed: quote === undefined };
|
|
327
|
+
}
|
|
245
328
|
function pythonParseable(source) {
|
|
246
|
-
const
|
|
247
|
-
if (!
|
|
329
|
+
const normalized = source.replace(/\r\n/g, "\n");
|
|
330
|
+
if (!normalized.trim())
|
|
248
331
|
return false;
|
|
249
|
-
if (/^\s*(?:def|class|import|from)\b/m.test(
|
|
332
|
+
if (/^\s*(?:def|class|import|from)\b/m.test(normalized) === false)
|
|
333
|
+
return false;
|
|
334
|
+
const structural = pythonStructureOutsideStrings(normalized);
|
|
335
|
+
if (!structural.stringsClosed)
|
|
250
336
|
return false;
|
|
337
|
+
const text = structural.text;
|
|
251
338
|
const openParens = (text.match(/\(/g) ?? []).length;
|
|
252
339
|
const closeParens = (text.match(/\)/g) ?? []).length;
|
|
253
340
|
const openBrackets = (text.match(/\[/g) ?? []).length;
|
|
@@ -260,15 +347,6 @@ function pythonParseable(source) {
|
|
|
260
347
|
return false;
|
|
261
348
|
if (openBraces !== closeBraces)
|
|
262
349
|
return false;
|
|
263
|
-
if (/("""|''')[\s\S]*$/.test(text)) {
|
|
264
|
-
const triples = text.match(/("""|''')/g) ?? [];
|
|
265
|
-
if (triples.length % 2 !== 0)
|
|
266
|
-
return false;
|
|
267
|
-
}
|
|
268
|
-
// The balanced-bracket counts above already catch a genuinely truncated
|
|
269
|
-
// file (an unclosed def/call leaves unbalanced parens). The previous
|
|
270
|
-
// per-line "def ... ( ... $" heuristic was a false-positive source for
|
|
271
|
-
// legal multi-line definitions such as `def f(\n x,\n):` — removed.
|
|
272
350
|
return true;
|
|
273
351
|
}
|
|
274
352
|
/**
|
|
@@ -318,12 +396,7 @@ function looksLikeValidPythonModule(source) {
|
|
|
318
396
|
return false;
|
|
319
397
|
if (/^\s*(?:def|class|import|from)\b/m.test(text) === false)
|
|
320
398
|
return false;
|
|
321
|
-
|
|
322
|
-
const triples = text.match(/("""|''')/g) ?? [];
|
|
323
|
-
if (triples.length % 2 !== 0)
|
|
324
|
-
return false;
|
|
325
|
-
}
|
|
326
|
-
return true;
|
|
399
|
+
return pythonStructureOutsideStrings(text).stringsClosed;
|
|
327
400
|
}
|
|
328
401
|
export function buildOutputLimitRecoveryPrompt(input) {
|
|
329
402
|
return buildOutputLimitRecoverySection(input);
|
|
@@ -842,7 +915,19 @@ export async function assessBackendTestMdPlanCompleteness(workspaceRoot) {
|
|
|
842
915
|
// The README must advertise a non-empty module index so the downstream
|
|
843
916
|
// manifest shell has stems to shard; the indexed module files themselves
|
|
844
917
|
// are produced by map_agent children and are NOT required here.
|
|
845
|
-
const
|
|
918
|
+
const stemInspection = inspectModuleStemsFromReadme(readme);
|
|
919
|
+
const indexedStems = stemInspection.validStems;
|
|
920
|
+
for (const invalid of stemInspection.invalid) {
|
|
921
|
+
if (!brokenPaths.includes("testcase/md/README.md")) {
|
|
922
|
+
brokenPaths.push("testcase/md/README.md");
|
|
923
|
+
}
|
|
924
|
+
issues.push({
|
|
925
|
+
code: "T5",
|
|
926
|
+
path: "testcase/md/README.md",
|
|
927
|
+
detail: `${invalid.reasonCode}: ${invalid.normalized}; use a stable lowercase business resource/domain stem`,
|
|
928
|
+
recoverable: true,
|
|
929
|
+
});
|
|
930
|
+
}
|
|
846
931
|
if (indexedStems.length === 0) {
|
|
847
932
|
if (!brokenPaths.includes("testcase/md/README.md")) {
|
|
848
933
|
brokenPaths.push("testcase/md/README.md");
|