codehere 1.0.1 → 1.0.2

package/dist/infrastructure/storage/project-thread-reader.js

@@ -1,409 +0,0 @@
-/**
- * Project thread reader — v0.6.0 M1 (UI-prep slice).
- *
- * Reads ~/.codehere/projects/<slug>/thread.jsonl files and reconstructs
- * the per-project thread state for API consumers (the new UI rendering
- * the persistent project thread shape).
- *
- * Reconstruction rules:
- *   - One file per project, identified by slug (sha1(cwd)[:12])
- *   - First line is `{_meta: 'project', ...}` — project header
- *   - Subsequent lines are `{_meta: 'project-update', ...}` (latest wins per
- *     field — Title agent uses this to set the title) OR `{type: 'task', ...}`
- *     (per-session entries; later entries with same sessionId override
- *     earlier — covers spawn entry → completion entry update)
- *
- * NOT in scope of this module:
- *   - Writing (handled by project-thread-mirror + project-thread-migration)
- *   - UI rendering (the new thread view will consume this reader's output)
- *   - Live SSE updates (polling is fine for the alpha; SSE in v0.6.x patch)
- */
-import { existsSync, readFileSync, readdirSync, statSync } from 'fs';
-import { join } from 'path';
-import { homedir } from 'os';
-import { estimateCostUsd } from '../cost/spend-aggregator.js';
-const DEFAULT_PROJECTS_DIR = join(process.env.CODEHERE_HOME || homedir(), '.codehere', 'projects');
-/**
- * Normalize a git remote URL to a stable identity key `host/org/repo`:
- *   - strips scheme (https/http/ssh/git) + any user@ info
- *   - handles scp-like `git@host:org/repo`
- *   - drops a trailing `.git` and trailing slashes
- *   - lowercases the host (path case preserved)
- * Returns null for empty/garbage input. Shared by the reader (identity grouping)
- * and the forward-mirror (write-time detection) so both compute the same id.
- */
-export function normalizeRemote(url) {
-    if (!url)
-        return null;
-    let s = String(url).trim();
-    if (!s)
-        return null;
-    const scp = s.match(/^[^@/]+@([^:/]+):(.+)$/); // git@host:org/repo(.git)
-    if (scp) {
-        s = scp[1] + '/' + scp[2];
-    }
-    else {
-        s = s.replace(/^[a-zA-Z][a-zA-Z0-9+.-]*:\/\//, ''); // scheme://
-        s = s.replace(/^[^@/]+@/, ''); // user@ (ssh://user@host/…)
-    }
-    s = s.replace(/\.git$/i, '').replace(/\/+$/, '');
-    if (!s)
-        return null;
-    const slash = s.indexOf('/');
-    if (slash === -1)
-        return s.toLowerCase();
-    return s.slice(0, slash).toLowerCase() + s.slice(slash);
-}
-/** True for hosts a clone must never reach — loopback, link-local (incl. cloud
- *  metadata 169.254.169.254), and RFC-1918 private ranges. `host` is the
- *  `host[:port]` segment from a normalized remote. */
-function isBlockedCloneHost(host) {
-    const h = String(host ?? '')
-        .split(':')[0]
-        .toLowerCase();
-    if (!h)
-        return true;
-    if (h === 'localhost' || h === '::1' || h === '0.0.0.0' || h.endsWith('.localhost'))
-        return true;
-    if (/^127\./.test(h))
-        return true; // loopback
-    if (/^169\.254\./.test(h))
-        return true; // link-local + cloud metadata
-    if (/^10\./.test(h))
-        return true; // RFC-1918
-    if (/^192\.168\./.test(h))
-        return true; // RFC-1918
-    if (/^172\.(1[6-9]|2\d|3[01])\./.test(h))
-        return true; // RFC-1918
-    if (/^0\./.test(h))
-        return true; // "this" network
-    return false;
-}
-/**
- * Parse a user-pasted GitHub (or any git) URL into a cloneable URL + a
- * predictable local target directory. Pure (no IO) so the clone route can be
- * tested without a network or filesystem. Accepts the forms a developer
- * actually pastes: full https, ssh (git@host:org/repo), bare host/org/repo,
- * and the `org/repo` GitHub shorthand. Rejects anything without a real
- * host/org/repo shape (or with whitespace — a git URL never has any).
- *
- * The target dir is `<baseDir>/<repo>` — visible + predictable, NOT buried in
- * a hidden dir, so the user can `cd` to it. The caller decides what to do when
- * the dir already exists (we open it rather than re-clone — see the route).
- */
-export function parseGitHubCloneTarget(rawUrl, baseDir) {
-    const raw = String(rawUrl ?? '').trim();
-    if (!raw)
-        return { ok: false, error: 'url_required' };
-    if (/\s/.test(raw))
-        return { ok: false, error: 'invalid_url' };
-    let cloneUrl;
-    if (/^[^@/]+@[^:/]+:.+/.test(raw)) {
-        cloneUrl = raw; // git@host:org/repo(.git)
-    }
-    else if (/^[a-zA-Z][a-zA-Z0-9+.-]*:\/\//.test(raw)) {
-        cloneUrl = raw; // already has a scheme (https://, ssh://, …)
-    }
-    else if (/^[\w.-]+\.[\w.-]+\/.+/.test(raw)) {
-        cloneUrl = 'https://' + raw; // host.tld/org/repo (no scheme)
-    }
-    else if (/^[\w.-]+\/[\w.-]+$/.test(raw)) {
-        cloneUrl = 'https://github.com/' + raw; // org/repo GitHub shorthand
-    }
-    else {
-        return { ok: false, error: 'invalid_url' };
-    }
-    // Scheme allowlist — only https + ssh reach `git clone`. Rejects file:// (local
-    // file read), http:// (plaintext + the cloud-metadata / localhost SSRF vector),
-    // and every other transport. The scp form (git@host:…) carries no scheme and is
-    // allowed. This is the SSRF/file:// surface a pre-push review flagged.
-    const scheme = cloneUrl.match(/^([a-zA-Z][a-zA-Z0-9+.-]*):\/\//);
-    if (scheme && !['https', 'ssh'].includes(scheme[1].toLowerCase()))
-        return { ok: false, error: 'invalid_url' };
-    const norm = normalizeRemote(cloneUrl); // host/org/repo, lowercased host
-    if (!norm)
-        return { ok: false, error: 'invalid_url' };
-    const parts = norm.split('/');
-    if (parts.length < 3)
-        return { ok: false, error: 'invalid_url' }; // need host + org + repo
-    // Internal-host denylist — don't let a clone target loopback / link-local /
-    // RFC-1918 (the rest of the SSRF surface: metadata endpoints, internal probing).
-    if (isBlockedCloneHost(parts[0]))
-        return { ok: false, error: 'invalid_url' };
-    const repo = (parts[parts.length - 1] || '').replace(/[^\w.-]/g, '');
-    // Reject '.'/'..'/empty — '.' and '..' survive the char-class strip and would
-    // traverse out of baseDir (join(baseDir, '..') = its parent). The clone dir
-    // must be a fresh leaf inside baseDir, nothing else.
-    if (!repo || repo === '.' || repo === '..')
-        return { ok: false, error: 'invalid_url' };
-    return { ok: true, cloneUrl, repo, dir: join(baseDir, repo) };
-}
-/** List all known projects with summary stats. Cheap; no full task content. */
-export function listProjects(options) {
-    const root = options?.projectsDir ?? DEFAULT_PROJECTS_DIR;
-    if (!existsSync(root))
-        return [];
-    let stat;
-    try {
-        stat = statSync(root);
-    }
-    catch {
-        return [];
-    }
-    if (!stat.isDirectory())
-        return [];
-    const slugs = readdirSync(root).filter((entry) => {
-        const p = join(root, entry);
-        try {
-            return statSync(p).isDirectory();
-        }
-        catch {
-            return false;
-        }
-    });
-    const perSlug = [];
-    for (const slug of slugs) {
-        // Pattern 4 — sidebar presence dots need agent attribution per project.
-        // Pull the full task list so we can compute agents + per-project cost.
-        const thread = readProjectThread(slug, { projectsDir: root });
-        if (!thread)
-            continue;
-        const agentChrono = thread.tasks
-            .filter((t) => typeof t.agent === 'string' && t.agent)
-            .sort((a, b) => b.startedAt - a.startedAt) // most-recent first
-            .map((t) => t.agent);
-        // Per-project spend — same estimator the per-task /api/tasks cost uses.
-        let totalCostCents = 0;
-        for (const t of thread.tasks) {
-            const u = t.artifacts?.usage;
-            if (!u)
-                continue;
-            totalCostCents += estimateCostUsd(t.agent, u) * 100;
-        }
-        perSlug.push({
-            slug: thread.slug,
-            id: normalizeRemote(thread.remote) ?? 'local:' + thread.slug,
-            remote: thread.remote,
-            cwd: thread.cwd,
-            title: thread.title,
-            taskCount: thread.taskCount,
-            totalCostCents,
-            lastActivityAt: thread.lastActivityAt,
-            agentChrono,
-        });
-    }
-    // Second pass — merge slugs that resolve to the same project id (clones of
-    // one git remote). Local-only projects have a unique `local:<slug>` id and
-    // never merge. The most-recently-active member is canonical (its slug + title
-    // win, so rename / room links keep working).
-    const byId = new Map();
-    for (const s of perSlug) {
-        const g = byId.get(s.id);
-        if (g)
-            g.push(s);
-        else
-            byId.set(s.id, [s]);
-    }
-    const summaries = [];
-    for (const [id, members] of byId) {
-        members.sort((a, b) => (b.lastActivityAt ?? 0) - (a.lastActivityAt ?? 0));
-        const head = members[0];
-        const localPaths = [...new Set(members.map((m) => m.cwd).filter((c) => !!c))];
-        // Recency-ordered agent stack across all members (head first).
-        const recentAgents = [];
-        const seen = new Set();
-        for (const m of members) {
-            for (const a of m.agentChrono) {
-                if (seen.has(a))
-                    continue;
-                seen.add(a);
-                recentAgents.push(a);
-                if (recentAgents.length >= 3)
-                    break;
-            }
-            if (recentAgents.length >= 3)
-                break;
-        }
-        summaries.push({
-            id,
-            remote: head.remote,
-            slug: head.slug,
-            cwd: head.cwd,
-            localPaths,
-            title: head.title,
-            taskCount: members.reduce((n, m) => n + m.taskCount, 0),
-            totalCostCents: Math.round(members.reduce((n, m) => n + m.totalCostCents, 0) * 100) / 100,
-            lastActivityAt: members.reduce((acc, m) => (m.lastActivityAt && (acc === null || m.lastActivityAt > acc) ? m.lastActivityAt : acc), null),
-            lastAgent: recentAgents[0] ?? null,
-            recentAgents,
-        });
-    }
-    // Sort by recency (most recent first)
-    summaries.sort((a, b) => (b.lastActivityAt ?? 0) - (a.lastActivityAt ?? 0));
-    return summaries;
-}
-/** Read a single project thread by slug. Returns null if not found. */
-export function readProjectThread(slug, options) {
-    const root = options?.projectsDir ?? DEFAULT_PROJECTS_DIR;
-    const file = join(root, slug, 'thread.jsonl');
-    if (!existsSync(file))
-        return null;
-    let content;
-    try {
-        content = readFileSync(file, 'utf-8');
-    }
-    catch {
-        return null;
-    }
-    let cwd = null;
-    let title = null;
-    let remote = null;
-    const tasksById = new Map();
-    const archivedState = new Map(); // sessionId -> archived (latest event wins)
-    for (const line of content.split('\n')) {
-        const trimmed = line.trim();
-        if (!trimmed)
-            continue;
-        let rec;
-        try {
-            rec = JSON.parse(trimmed);
-        }
-        catch {
-            continue;
-        }
-        if (rec._meta === 'project') {
-            if (typeof rec.cwd === 'string')
-                cwd = rec.cwd;
-            if (typeof rec.title === 'string')
-                title = rec.title;
-            if (typeof rec.remote === 'string')
-                remote = rec.remote;
-        }
-        else if (rec._meta === 'project-update') {
-            if (typeof rec.title === 'string')
-                title = rec.title;
-            if (typeof rec.cwd === 'string' && cwd === null)
-                cwd = rec.cwd;
-            // Latest non-empty remote wins; an explicit null (checked, none found)
-            // leaves the prior value alone — we only ever upgrade toward a real remote.
-            if (typeof rec.remote === 'string' && rec.remote)
-                remote = rec.remote;
-        }
-        else if (rec._meta === 'archived' && typeof rec.sessionId === 'string') {
-            archivedState.set(rec.sessionId, rec.archived === true);
-        }
-        else if (rec.type === 'task' && typeof rec.sessionId === 'string') {
-            const existing = tasksById.get(rec.sessionId) ?? {
-                sessionId: rec.sessionId,
-                agent: '',
-                prompt: '',
-                cwd: '',
-                startedAt: 0,
-            };
-            // Merge fields — later entries override earlier (covers
-            // spawn → completion update pattern from forward-mirror)
-            for (const k of [
-                'agent',
-                'prompt',
-                'cwd',
-                'startedAt',
-                'status',
-                'durationMs',
-                'completedAt',
-                'steps',
-                'artifacts',
-                'parentSessionId',
-                'finalReport',
-            ]) {
-                if (rec[k] !== undefined) {
-                    existing[k] = rec[k];
-                }
-            }
-            tasksById.set(rec.sessionId, existing);
-        }
-    }
-    // Stamp the archived flag (soft-delete) from the latest archived event per task.
-    for (const t of tasksById.values()) {
-        const a = archivedState.get(t.sessionId);
-        if (a !== undefined)
-            t.archived = a;
-    }
-    const tasks = [...tasksById.values()].sort((a, b) => a.startedAt - b.startedAt);
-    const lastActivityAt = tasks.reduce((acc, t) => {
-        const ts = t.completedAt ?? t.startedAt;
-        if (ts && (acc === null || ts > acc))
-            return ts;
-        return acc;
-    }, null);
-    const limit = options?.taskLimit;
-    const cappedTasks = limit === undefined ? tasks : limit === 0 ? [] : tasks.slice(-limit);
-    return {
-        slug,
-        cwd,
-        title,
-        remote,
-        // Count only what the UI actually shows: the sidebar tree + room list both
-        // hide archived + demo/seed tasks, so counting them inflated the badge
-        // (a project read "420" while showing a handful).
-        taskCount: countVisibleTasks(tasks),
-        tasks: cappedTasks,
-        lastActivityAt,
-    };
-}
-/** The honest THREAD count: distinct top-level conversations a user started.
- *  Excludes soft-deleted (archived) and demo/seed tasks, and — the big one —
- *  tasks with a parentSessionId, which are reply turns + auto cross-vendor
- *  reviewers (continuations of a thread, not new threads). Counting those made a
- *  project read "400+" while showing a handful of real threads. */
-function countVisibleTasks(tasks) {
-    return tasks.filter((t) => {
-        if (t.archived || t.parentSessionId)
-            return false;
-        if (/^demo-/.test(t.sessionId || ''))
-            return false;
-        // Failover-lineage marker ("↻ Switched from X to Y (hit a rate limit)") — a
-        // continuation edge, not a thread the user started. Matches the sidebar tree.
-        const p = t.prompt || '';
-        if (p.includes('Switched from') && p.includes('rate limit'))
-            return false;
-        return true;
-    }).length;
-}
-/**
- * Like readProjectThread, but merges every clone of the same repo (slugs that share
- * a git remote) into ONE thread — so a project room shows the repo's FULL history,
- * not just the single slug the UI happened to open. A local-only project (no remote)
- * has no clones to merge and is returned unchanged. Tasks are deduped by sessionId.
- */
-export function readMergedProjectThread(slug, options) {
-    const root = options?.projectsDir ?? DEFAULT_PROJECTS_DIR;
-    const head = readProjectThread(slug, { projectsDir: root });
-    if (!head || !head.remote)
-        return head;
-    const id = normalizeRemote(head.remote);
-    let entries;
-    try {
-        entries = readdirSync(root).filter((e) => /^[a-f0-9]{12}$/.test(e) && e !== slug);
-    }
-    catch {
-        return head;
-    }
-    const byId = new Map();
-    for (const t of head.tasks)
-        byId.set(t.sessionId, t);
-    for (const e of entries) {
-        const sib = readProjectThread(e, { projectsDir: root });
-        if (!sib || normalizeRemote(sib.remote) !== id)
-            continue;
-        for (const t of sib.tasks)
-            if (!byId.has(t.sessionId))
-                byId.set(t.sessionId, t);
-    }
-    const tasks = [...byId.values()].sort((a, b) => a.startedAt - b.startedAt);
-    const lastActivityAt = tasks.reduce((acc, t) => {
-        const ts = t.completedAt ?? t.startedAt;
-        return ts && (acc === null || ts > acc) ? ts : acc;
-    }, null);
-    return { ...head, tasks, taskCount: countVisibleTasks(tasks), lastActivityAt };
-}
-//# sourceMappingURL=project-thread-reader.js.map

package/dist/infrastructure/storage/project-thread-title-backfill.js

@@ -1,214 +0,0 @@
-/**
- * Title backfill — v0.6.0 follow-up to M9.
- *
- * The M9 hidden Title agent (generate-thread-title.ts) only fires on the FIRST
- * task entry written to a project's thread.jsonl. Projects seeded by the v0.6
- * migration (project-thread-migration.ts) bypass that path — they get a
- * project-meta header but no Title agent invocation. Result: 5/7 projects on
- * a typical dogfood machine show "(untitled)" in the /threads switcher even
- * though their first task carries enough signal to title from.
- *
- * This backfill closes that gap. For every project with title=null and at
- * least one task entry, it fires the Title agent on the first task's prompt +
- * cwd and appends a `{_meta: 'project-update', title: ...}` line. Reader
- * (project-thread-reader.ts) takes the latest project-update title.
- *
- * Idempotent: skips projects that already have a title (or that have NO tasks
- * to title from). Safe to run on every daemon boot. Best-effort: any LLM
- * failure (provider missing, timeout, rate limit) skips that project silently.
- *
- * Cost budget: ~1 LLM call per untitled project on first run after this lands.
- * Subsequent boots are no-ops since titles persist.
- */
-import { existsSync, readdirSync, readFileSync, appendFileSync, statSync } from 'fs';
-import { join } from 'path';
-import { homedir } from 'os';
-import { generateThreadTitle } from '../../application/use-cases/generate-thread-title.js';
-const DEFAULT_PROJECTS_DIR = join(process.env.CODEHERE_HOME || homedir(), '.codehere', 'projects');
-function scanThread(slug, file) {
-    let title = null;
-    let firstPrompt = null;
-    let firstCwd = null;
-    let firstStartedAt = Number.MAX_SAFE_INTEGER;
-    let content;
-    try {
-        content = readFileSync(file, 'utf-8');
-    }
-    catch {
-        return { slug, title, firstPrompt, firstCwd };
-    }
-    for (const line of content.split('\n')) {
-        const trimmed = line.trim();
-        if (!trimmed)
-            continue;
-        let rec;
-        try {
-            rec = JSON.parse(trimmed);
-        }
-        catch {
-            continue;
-        }
-        if (rec._meta === 'project') {
-            if (typeof rec.title === 'string')
-                title = rec.title;
-        }
-        else if (rec._meta === 'project-update') {
-            if (typeof rec.title === 'string')
-                title = rec.title;
-        }
-        else if (rec.type === 'task') {
-            const startedAt = typeof rec.startedAt === 'number' ? rec.startedAt : Number.MAX_SAFE_INTEGER;
-            if (startedAt < firstStartedAt) {
-                firstStartedAt = startedAt;
-                if (typeof rec.prompt === 'string')
-                    firstPrompt = rec.prompt;
-                if (typeof rec.cwd === 'string')
-                    firstCwd = rec.cwd;
-            }
-        }
-    }
-    return { slug, title, firstPrompt, firstCwd };
-}
-/**
- * Backfill missing titles. Returns counts; never throws (best-effort per project).
- *
- * Disable via CODEHERE_DISABLE_TITLE_AGENT=1 (same env var that gates M9 itself —
- * keeps the toggle surface minimal).
- */
-export async function backfillThreadTitles(options) {
-    const start = Date.now();
-    const result = {
-        scanned: 0,
-        alreadyTitled: 0,
-        noTasks: 0,
-        titlesGenerated: 0,
-        titleAgentFailed: 0,
-        durationMs: 0,
-    };
-    if (process.env.CODEHERE_DISABLE_TITLE_AGENT === '1') {
-        result.durationMs = Date.now() - start;
-        return result;
-    }
-    const root = options?.projectsDir ?? DEFAULT_PROJECTS_DIR;
-    if (!existsSync(root)) {
-        result.durationMs = Date.now() - start;
-        return result;
-    }
-    let stat;
-    try {
-        stat = statSync(root);
-    }
-    catch {
-        result.durationMs = Date.now() - start;
-        return result;
-    }
-    if (!stat.isDirectory()) {
-        result.durationMs = Date.now() - start;
-        return result;
-    }
-    let slugs;
-    try {
-        slugs = readdirSync(root);
-    }
-    catch {
-        result.durationMs = Date.now() - start;
-        return result;
-    }
-    const cap = options?.maxProjects ?? Number.MAX_SAFE_INTEGER;
-    let backfilledCount = 0;
-    for (const slug of slugs) {
-        if (backfilledCount >= cap)
-            break;
-        const projectDir = join(root, slug);
-        let s;
-        try {
-            s = statSync(projectDir);
-        }
-        catch {
-            continue;
-        }
-        if (!s.isDirectory())
-            continue;
-        const file = join(projectDir, 'thread.jsonl');
-        if (!existsSync(file))
-            continue;
-        result.scanned++;
-        const scanned = scanThread(slug, file);
-        if (scanned.title) {
-            result.alreadyTitled++;
-            continue;
-        }
-        if (!scanned.firstPrompt || !scanned.firstCwd) {
-            // Nothing to title from — skip. (Migration-seeded headers without task
-            // entries fall here, as do projects where the first task lost its prompt
-            // somehow.)
-            result.noTasks++;
-            continue;
-        }
-        let title;
-        try {
-            title = await generateThreadTitle(scanned.firstPrompt, scanned.firstCwd);
-        }
-        catch {
-            // Title agent should never throw (it returns null on failure), but guard
-            // anyway so one bad call doesn't sink the whole backfill.
-            title = null;
-        }
-        if (!title) {
-            result.titleAgentFailed++;
-            continue;
-        }
-        // Race guard: backfill scanned BEFORE the LLM call, but the LLM blocks
-        // for seconds. A user rename in that window leaves a userOverride:true
-        // marker the pre-scan never saw. Re-read the same file just before the
-        // write and skip if the user got there first. (Inlined rather than calling
-        // hasManualOverride() because that helper hardcodes the production
-        // projects dir; backfill supports a custom projectsDir for tests.)
-        let userBeatUs = false;
-        try {
-            const recheck = readFileSync(file, 'utf-8');
-            for (const line of recheck.split('\n')) {
-                const t = line.trim();
-                if (!t)
-                    continue;
-                try {
-                    const rec = JSON.parse(t);
-                    if (rec._meta === 'project-update' && rec.userOverride === true) {
-                        userBeatUs = true;
-                        break;
-                    }
-                }
-                catch {
-                    /* skip malformed */
-                }
-            }
-        }
-        catch {
-            /* if we can't read, fall through and let the append fail */
-        }
-        if (userBeatUs) {
-            result.alreadyTitled++;
-            continue;
-        }
-        try {
-            const updateLine = JSON.stringify({
-                _meta: 'project-update',
-                slug,
-                title,
-                cwd: scanned.firstCwd,
-                backfilledAt: Date.now(),
-                updatedAt: Date.now(),
-            });
-            appendFileSync(file, updateLine + '\n', 'utf-8');
-            result.titlesGenerated++;
-            backfilledCount++;
-        }
-        catch {
-            // Append failure is logged via counter but doesn't sink the loop.
-            result.titleAgentFailed++;
-        }
-    }
-    result.durationMs = Date.now() - start;
-    return result;
-}
-//# sourceMappingURL=project-thread-title-backfill.js.map