@haystackeditor/cli 0.18.0 → 0.20.0

package/dist/commands/verify-precompute.js

@@ -1,6 +1,6 @@
 import { spawn, execFileSync } from 'node:child_process';
 import { createHash, randomUUID } from 'node:crypto';
-import { closeSync, constants, cpSync, existsSync, fsyncSync, lstatSync, mkdirSync, mkdtempSync, openSync, renameSync, rmSync, writeFileSync } from 'node:fs';
+import { closeSync, constants, cpSync, existsSync, fsyncSync, lstatSync, mkdirSync, mkdtempSync, openSync, readFileSync, renameSync, rmSync, writeFileSync } from 'node:fs';
 import { createGzip } from 'node:zlib';
 import { isAbsolute, join, resolve } from 'node:path';
 import { tmpdir } from 'node:os';
@@ -12,7 +12,8 @@ import { HaystackApiError, haystackJson, REPO_NOT_ENTITLED_CODE, SUBSCRIPTION_RE
 import { trackPrecomputeEvent } from '../utils/telemetry.js';
 import { findSecretsInPatch, isSecretBearingPath } from '../utils/secret-paths.js';
 import { startPrecomputeDelivery } from './precompute-delivery.js';
-import { CRAWL_TITLE_MAX_CHARS } from './crawl-contract.js';
+import { CRAWL_BUDGET_MAX_MS, CRAWL_BUDGET_MIN_MS, CRAWL_TITLE_MAX_CHARS } from './crawl-contract.js';
+import { blockLines, parseOnboardingBlock } from './verify-onboarding.js';
 // --hook has five seconds to capture the exact source and publish local custody.
 // Its detached sender owns the bounded network delivery and acknowledgement.
 // An explicit invocation still waits synchronously for the server response.
@@ -66,7 +67,7 @@ function gitOutput(args, gitRoot, deadline, options = {}) {
         throw new PrecomputeFailure('Could not derive the repository snapshot.');
     }
 }
-function resolveOriginRepository(gitRoot) {
+export function resolveOriginRepository(gitRoot) {
     try {
         const remote = parseRemoteUrl('origin');
         if (!isRepositoryComponent(remote.owner)
@@ -80,8 +81,8 @@ function resolveOriginRepository(gitRoot) {
         throw new PrecomputeFailure('No supported origin repository was found.');
     }
 }
-/** The shared base rule (utils/verify-base.ts), bounded by the hook's deadline. */
-function deriveBaseSha(gitRoot, deadline) {
+/** The shared base rule (utils/verify-base.ts), bounded by the caller's deadline. */
+export function deriveBaseSha(gitRoot, deadline) {
     try {
         return resolveVerifyBaseSha('HEAD', {
             run: (args, options) => gitOutput(args, gitRoot, deadline, options?.remote
@@ -259,7 +260,9 @@ async function buildCompressedPatch(gitRoot, baseSha, workCommit, deadline) {
         child.stdout.pipe(gzip);
     });
 }
-export async function derivePrecomputeRequest(gitRoot, owner, repository, deadline = Date.now() + HOOK_WALL_MS) {
+export async function derivePrecomputeRequest(gitRoot, owner, repository, deadline = Date.now() + HOOK_WALL_MS, 
+/** How long the crawl may take (CRAWL-V1 amendment 8); undefined: the service's default. */
+budgetMs) {
     // Capture this before git and compression work so delivery latency cannot
     // make an older stop look newer than a later one.
     const observedAt = Date.now();
@@ -287,7 +290,7 @@ export async function derivePrecomputeRequest(gitRoot, owner, repository, deadli
             cacheKey,
             patchGzBase64: patchGzip.toString('base64'),
             patchSha256: createHash('sha256').update(patchGzip).digest('hex'),
-            ...(workCommit === baseSha ? {} : { crawl: { treeSha, changeTitle } }),
+            ...(workCommit === baseSha ? {} : { crawl: { treeSha, changeTitle, ...(budgetMs === undefined ? {} : { budgetMs }) } }),
         },
     };
 }
@@ -330,7 +333,8 @@ function readWalkerStatus(value) {
     }
     throw new PrecomputeFailure('The precompute service returned an invalid walker status.');
 }
-const CRAWL_UNAVAILABLE_REASONS = ['not-configured', 'runner-unavailable', 'profile-ambiguous', 'superseded', 'enqueue-failed'];
+const CRAWL_UNAVAILABLE_REASONS = ['runner-unavailable', 'superseded', 'enqueue-failed'];
+const isOnboardRunId = (value) => typeof value === 'string' && /^ob_[0-9a-f]{48}$/.test(value);
 function readCrawlHookStatus(value) {
     if (value && typeof value === 'object' && !Array.isArray(value)) {
         const record = value;
@@ -341,6 +345,18 @@ function readCrawlHookStatus(value) {
             const reason = CRAWL_UNAVAILABLE_REASONS.find(known => known === record.reason);
             if (record.status === 'unavailable' && reason !== undefined)
                 return { status: record.status, reason };
+            // Crawl amendment 8: the base has no published onboarding record yet.
+            if (record.status === 'onboarding' && isOnboardRunId(record.onboardRunId))
+                return { status: 'onboarding', onboardRunId: record.onboardRunId };
+        }
+        if (Object.keys(record).length === 3 && record.status === 'onboarding-blocked'
+            && (record.onboardRunId === null || isOnboardRunId(record.onboardRunId))) {
+            try {
+                return { status: 'onboarding-blocked', onboardRunId: record.onboardRunId, block: parseOnboardingBlock(record.block) };
+            }
+            catch (error) {
+                throw new PrecomputeFailure(`The precompute service returned an invalid onboarding block (${error instanceof Error ? error.message : String(error)}).`);
+            }
         }
     }
     throw new PrecomputeFailure('The precompute service returned an invalid crawl status.');
@@ -348,9 +364,7 @@ function readCrawlHookStatus(value) {
 /** Why the service could not start a crawl, in plain words. */
 export function crawlUnavailableText(reason) {
     switch (reason) {
-        case 'not-configured': return 'this repository is not set up for crawling yet';
         case 'runner-unavailable': return 'no machine that can crawl this repository is available right now';
-        case 'profile-ambiguous': return 'the machines that crawl this repository disagree about how to run it';
         case 'superseded': return 'a newer stop in this repository came first';
         case 'enqueue-failed': return 'the service could not queue it';
     }
@@ -447,19 +461,41 @@ function printResult(result) {
         console.log(`A matching crawl already exists (${crawl.runId}).`);
     else if (crawl?.status === 'unavailable')
         console.log(`Crawl unavailable: ${crawlUnavailableText(crawl.reason)}.`);
+    else if (crawl?.status === 'onboarding') {
+        console.log(`The app is being onboarded for this base (${crawl.onboardRunId}); the crawl runs when it is ready. Run \`haystack verify\` to follow it.`);
+    }
+    else if (crawl?.status === 'onboarding-blocked')
+        console.log(blockLines(crawl.block).join('\n'));
     else
         console.log('Crawl status was not reported by the service.');
 }
 /** The hook, an explicit precompute and `haystack verify` all capture here.
  * The repository defaults to origin's, as the hook's does. */
-export async function captureCheckout(deadline, repository) {
+/** The repository's crawl time: .haystack.json crawl.minutes at the git root, whole minutes from 1 to 30 (CRAWL-V1 amendment 8),
+ * or undefined when the file or the field is absent. A value out of range fails the capture, never silently becomes the default. */
+function configuredCrawlMinutes(gitRoot) {
+    const path = join(gitRoot, '.haystack.json');
+    if (!existsSync(path))
+        return undefined;
+    const minutes = JSON.parse(readFileSync(path, 'utf8')).crawl?.minutes;
+    if (minutes === undefined)
+        return undefined;
+    if (typeof minutes !== 'number' || !Number.isInteger(minutes) || minutes * 60_000 < CRAWL_BUDGET_MIN_MS || minutes * 60_000 > CRAWL_BUDGET_MAX_MS) {
+        throw new PrecomputeFailure(`.haystack.json crawl.minutes must be a whole number from ${CRAWL_BUDGET_MIN_MS / 60_000} to ${CRAWL_BUDGET_MAX_MS / 60_000}.`);
+    }
+    return minutes;
+}
+export async function captureCheckout(deadline, repository, 
+/** `haystack verify --minutes`; undefined: the repository's .haystack.json crawl.minutes, else the service's default. */
+minutes) {
     const gitRoot = findGitRoot();
     if (!gitRoot) {
         throw new PrecomputeFailure('Run this command from a git repository.');
     }
     const metadataDirectory = walkerMetadataDirectory(gitRoot, deadline);
     const target = repository ?? resolveOriginRepository(gitRoot);
-    const derivation = await derivePrecomputeRequest(gitRoot, target.owner, target.repository, deadline);
+    const chosen = minutes ?? configuredCrawlMinutes(gitRoot);
+    const derivation = await derivePrecomputeRequest(gitRoot, target.owner, target.repository, deadline, chosen === undefined ? undefined : chosen * 60_000);
     return { metadataDirectory, derivation };
 }
 /** The explicit (non-hook) submission of a capture: one synchronous