Merge PR #148 from aakhter/cod-164-scrollback-replay-crlf: replay full tmux scrollback on terminal reload + CRLF normalization

Includes review fixes: explicit ?full=1 trigger wired from initial page load, capture maxBuffer sized from config with -S line bound, capture returned alone (no byte-buffer duplication), early byte-cap before normalization.

src/mux-interface.ts

@@ -105,6 +105,20 @@ export interface RespawnPaneOptions {
   remote?: SessionRemote;
 }
 
+/** Options for pane buffer capture (COD-47 full-history mode). */
+export interface PaneCaptureOptions {
+  /** Capture the entire tmux scrollback instead of just the visible frame. */
+  fullHistory?: boolean;
+  /** Bound the full-history capture to this many scrollback lines (`-S -<N>`). */
+  historyLimitLines?: number;
+  /**
+   * Byte cap the consumer will keep from the capture. Sizes the child-process
+   * stdout buffer (with slack) so multi-MB scrollback dumps aren't killed by
+   * the 1MB execSync default (ENOBUFS).
+   */
+  maxCaptureBytes?: number;
+}
+
 /**
  * Terminal multiplexer interface.
  *
@@ -232,9 +246,16 @@ export interface TerminalMultiplexer extends EventEmitter {
   /** Respawn a dead pane with a fresh command. Returns the new PID or null on failure. */
   respawnPane(options: RespawnPaneOptions): Promise<number | null>;
 
-  /** Capture a pane's current tmux buffer with ANSI escape codes preserved. */
-  capturePaneBuffer?(muxName: string, paneTarget: string): string | null;
+  /**
+   * Capture a pane's current tmux buffer with ANSI escape codes preserved.
+   * Pass `{ fullHistory: true }` to capture the entire scrollback as linear
+   * text instead of just the visible single-screen frame (COD-47).
+   */
+  capturePaneBuffer?(muxName: string, paneTarget?: string, opts?: PaneCaptureOptions): string | null;
 
-  /** Capture the active pane's current tmux buffer with ANSI escape codes preserved. */
-  captureActivePaneBuffer?(muxName: string): string | null;
+  /**
+   * Capture the active pane's current tmux buffer with ANSI escape codes preserved.
+   * Pass `{ fullHistory: true }` to capture the entire scrollback (COD-47).
+   */
+  captureActivePaneBuffer?(muxName: string, opts?: PaneCaptureOptions): string | null;
 }

src/tmux-manager.ts

@@ -60,6 +60,7 @@ import type {
   MuxSessionWithStats,
   CreateSessionOptions,
   RespawnPaneOptions,
+  PaneCaptureOptions,
 } from './mux-interface.js';
 
 // ============================================================================
@@ -67,7 +68,15 @@ import type {
 // ============================================================================
 
 import { EXEC_TIMEOUT_MS } from './config/exec-timeout.js';
-import { DEFAULT_TMUX_HISTORY_LIMIT } from './config/terminal-history.js';
+import { DEFAULT_TMUX_HISTORY_LIMIT, DEFAULT_TERMINAL_BUFFER_MAX_BYTES } from './config/terminal-history.js';
+
+/**
+ * Extra stdout headroom for the full-history `capture-pane` child process on
+ * top of the consumer's byte cap: raw scrollback carries per-line SGR/ANSI
+ * overhead that the route pipeline strips before applying its cap, so the
+ * capture must be allowed to exceed the final payload size.
+ */
+const FULL_HISTORY_CAPTURE_SLACK_BYTES = 8 * 1024 * 1024;
 
 /** Delay after tmux session creation — enough for detached tmux to be queryable */
 const TMUX_CREATION_WAIT_MS = 100;
@@ -432,6 +441,26 @@ function truncatePaneLineByVisibleColumns(line: string, maxColumns: number): str
   return result;
 }
 
+/**
+ * Normalize scrollback line endings to `\r\n` so a fresh xterm replays each line
+ * at column 0 (COD-138).
+ *
+ * `capture-pane -p -e -S -` (full-history capture) joins scrollback rows with a
+ * BARE `\n`. The browser xterm is created with the default `convertEol: false`
+ * (correct for the live PTY stream, which already carries real `\r\n`), so a bare
+ * `\n` drops a row without returning the cursor to column 0. Replaying that raw
+ * buffer on a full page reload makes every line start one column further right —
+ * the diagonal "staircase". The visible/tab-switch path avoids this by repainting
+ * each row with an absolute cursor CSI (`formatPaneSnapshot`); the full-history
+ * path returns raw scrollback, so it must be CRLF-normalized here.
+ *
+ * `\r?\n → \r\n` is idempotent on already-CRLF input and leaves a lone `\r` (an
+ * intentional in-line column reset / overwrite) untouched.
+ */
+export function normalizeScrollbackEol(buffer: string): string {
+  return buffer.replace(/\r?\n/g, '\r\n');
+}
+
 export function formatPaneSnapshot(
   lines: string[],
   geometry: { cols: number; rows: number; cursorX: number; cursorY: number }
@@ -2327,30 +2356,65 @@ export class TmuxManager extends EventEmitter implements TerminalMultiplexer {
   }
 
   /**
-   * Capture the current visible text and SGR styles of a specific pane.
+   * Capture a pane's text and SGR styles.
    *
-   * `capture-pane -e` is sanitized by `formatPaneSnapshot`: SGR color/style
-   * codes are preserved, while cursor/erase/scroll-region controls are stripped
-   * before rows are repainted at absolute positions in browser xterm.
+   * Two modes:
+   * - Visible (default): `capture-pane -p -e` grabs only the on-screen frame,
+   *   then `formatPaneSnapshot` repaints each row at its absolute position so
+   *   the browser xterm reproduces the live frame. Used for fast tab switches.
+   * - Full history (`opts.fullHistory`): `capture-pane -p -e -J -S -<N>` grabs
+   *   the tmux scrollback (COD-47, bounded to the configured history limit),
+   *   returned as linear scrollback text with SGR codes preserved (NOT
+   *   repositioned — a multi-screen history can't be painted into a single
+   *   visible frame, so the snapshot repaint is skipped). `-J` re-joins lines
+   *   hard-wrapped at the pane width so they reflow in the browser xterm.
+   *   Used for full page reloads so the user gets back their scroll history.
+   *   Caveat: lines tmux has already evicted past its history-limit are gone.
    */
-  capturePaneBuffer(muxName: string, paneTarget: string): string | null {
+  capturePaneBuffer(muxName: string, paneTarget?: string, opts?: PaneCaptureOptions): string | null {
     if (IS_TEST_MODE) return '';
-    if (!isValidMuxName(muxName)) {
-      console.error('[TmuxManager] Invalid session name in capturePaneBuffer:', muxName);
-      return null;
-    }
-    if (!SAFE_PANE_TARGET_PATTERN.test(paneTarget)) {
-      console.error('[TmuxManager] Invalid pane target:', paneTarget);
+    const target = resolveTmuxPaneTarget(muxName, paneTarget);
+    if (!target) {
+      console.error('[TmuxManager] Invalid pane target in capturePaneBuffer:', { muxName, paneTarget });
       return null;
     }
 
-    const target = paneTarget.startsWith('%') ? `${muxName}.${paneTarget}` : `${muxName}.%${paneTarget}`;
+    const fullHistory = opts?.fullHistory === true;
 
     try {
-      const buffer = execSync(`${this.tmux()} capture-pane -p -e -t ${shellescape(target)}`, {
+      // `-S -<N>` starts the capture N lines above the visible frame (tmux
+      // clamps to the top of history), so tmux never serializes more scrollback
+      // than the configured history limit retains.
+      const requestedLines = opts?.historyLimitLines;
+      const historyLines =
+        typeof requestedLines === 'number' && Number.isFinite(requestedLines) && requestedLines > 0
+          ? Math.trunc(requestedLines)
+          : DEFAULT_TMUX_HISTORY_LIMIT;
+      const captureFlags = fullHistory ? `capture-pane -p -e -J -S -${historyLines}` : 'capture-pane -p -e';
+      // execSync's default maxBuffer (1MB) kills multi-MB scrollback dumps
+      // (ENOBUFS) and would silently degrade full-history capture to the byte
+      // buffer for exactly the long sessions it exists for — size it from the
+      // consumer's byte cap plus ANSI-overhead slack instead.
+      const execOpts: { encoding: 'utf-8'; timeout: number; maxBuffer?: number } = {
         encoding: 'utf-8',
         timeout: EXEC_TIMEOUT_MS,
-      }).replace(/\n+$/g, '');
+      };
+      if (fullHistory) {
+        execOpts.maxBuffer =
+          (opts?.maxCaptureBytes ?? DEFAULT_TERMINAL_BUFFER_MAX_BYTES) + FULL_HISTORY_CAPTURE_SLACK_BYTES;
+      }
+      const buffer = execSync(`${this.tmux()} ${captureFlags} -t ${shellescape(target)}`, execOpts).replace(
+        /\n+$/g,
+        ''
+      );
+      // Full-history spans many screens — return it as raw linear scrollback
+      // rather than repainting rows at single-screen absolute positions. tmux
+      // joins scrollback rows with a bare `\n`; normalize to `\r\n` so a fresh
+      // xterm (convertEol:false) starts each replayed line at column 0 instead
+      // of staircasing diagonally (COD-138).
+      if (fullHistory) {
+        return normalizeScrollbackEol(buffer);
+      }
       try {
         const cursor = execSync(
           `${this.tmux()} display-message -p -t ${shellescape(target)} '#{cursor_x} #{cursor_y} #{pane_width} #{pane_height}'`,
@@ -2375,9 +2439,19 @@ export class TmuxManager extends EventEmitter implements TerminalMultiplexer {
       } catch (cursorErr) {
         console.error('[TmuxManager] Failed to query pane cursor after capture:', cursorErr);
       }
-      return buffer;
+      // Cursor query failed or geometry was invalid, so we skip the absolute-
+      // positioned snapshot repaint and fall back to the raw capture. Normalize
+      // its bare `\n` line endings to `\r\n` so the replay doesn't staircase
+      // diagonally in a fresh xterm (COD-138, same reason as the fullHistory path).
+      return normalizeScrollbackEol(buffer);
     } catch (err) {
-      console.error('[TmuxManager] Failed to capture pane buffer:', err);
+      // ENOBUFS carries the truncated multi-MB stdout on the error object —
+      // log a concise line instead of dumping it into the journal.
+      if ((err as NodeJS.ErrnoException)?.code === 'ENOBUFS') {
+        console.error('[TmuxManager] Pane capture exceeded maxBuffer (ENOBUFS); falling back to byte history');
+      } else {
+        console.error('[TmuxManager] Failed to capture pane buffer:', err);
+      }
       return null;
     }
   }
@@ -2388,7 +2462,7 @@ export class TmuxManager extends EventEmitter implements TerminalMultiplexer {
    * Pane ids are not stable across respawns or restores, so callers should not
    * assume the first pane remains `%0`.
    */
-  captureActivePaneBuffer(muxName: string): string | null {
+  captureActivePaneBuffer(muxName: string, opts?: PaneCaptureOptions): string | null {
     if (IS_TEST_MODE) return '';
     if (!isValidMuxName(muxName)) {
       console.error('[TmuxManager] Invalid session name in captureActivePaneBuffer:', muxName);
@@ -2401,7 +2475,7 @@ export class TmuxManager extends EventEmitter implements TerminalMultiplexer {
         timeout: EXEC_TIMEOUT_MS,
       }).trim();
       const target = resolveActivePaneTarget(output);
-      return target ? this.capturePaneBuffer(muxName, target) : null;
+      return target ? this.capturePaneBuffer(muxName, target, opts) : null;
     } catch (err) {
       console.error('[TmuxManager] Failed to resolve active pane for capture:', err);
       return null;

src/web/public/app.js

@@ -477,6 +477,7 @@ class CodemanApp {
     this._initGeneration = 0;     // dedup concurrent handleInit calls
     this._initFallbackTimer = null; // fallback timer if SSE init doesn't arrive
     this._selectGeneration = 0;   // cancel stale selectSession loads
+    this._initialFullBufferLoad = true; // first buffer load after a page load fetches full tmux scrollback (COD-47)
     this.terminalLoadStates = new Map(); // Map<sessionId, { generation, phase }>
     this.respawnStatus = {};
     this.respawnTimers = {}; // Track timed respawn timers
@@ -3911,7 +3912,16 @@ class CodemanApp {
 
       this._setTerminalLoadState(sessionId, selectGen, 'fetching');
       _crashDiag.log('FETCH_START');
-      const res = await fetch(`/api/sessions/${sessionId}/terminal?tail=${TERMINAL_TAIL_SIZE}`);
+      // The FIRST buffer load after a page load requests the full tmux scrollback
+      // (?full=1, COD-47) so history that scrolled off the server's byte buffer
+      // comes back after a reload. Tab switches keep the fast ?tail= frame path.
+      const useFullHistory = this._initialFullBufferLoad === true;
+      this._initialFullBufferLoad = false;
+      const res = await fetch(
+        useFullHistory
+          ? `/api/sessions/${sessionId}/terminal?full=1`
+          : `/api/sessions/${sessionId}/terminal?tail=${TERMINAL_TAIL_SIZE}`
+      );
       if (this._isStaleSelect(selectGen)) {
         this._clearTerminalLoadState(sessionId, selectGen);
         return;

src/web/routes/session-routes.ts

@@ -1016,11 +1016,22 @@ export function registerSessionRoutes(
 
   // Query params:
   //   tail=<bytes> - Only return last N bytes (faster initial load)
+  //   full=1       - Full page reload: replay the entire tmux scrollback (COD-47)
   app.get('/api/sessions/:id/terminal', async (req) => {
     const { id } = req.params as { id: string };
-    const query = req.query as { tail?: string };
+    const query = req.query as { tail?: string; full?: string };
     const session = findSessionOrFail(ctx, id);
 
+    // `full=1` is the EXPLICIT full-reload signal (COD-47): the browser reloaded
+    // the page and wants the whole scroll history back, so we capture the ENTIRE
+    // tmux scrollback and the user gets back history that scrolled off Codeman's
+    // byte buffer. Requests WITHOUT it — tab switches (`tail=`) and the legacy
+    // no-param callers (response-viewer fallback, clearTerminal refresh) — keep
+    // the fast visible-frame capture.
+    const tailBytes = query.tail ? parseInt(query.tail, 10) : 0;
+    const isFullReload = query.full === '1' || query.full === 'true';
+    const { tmuxHistoryLimit, terminalBufferMaxBytes } = await ctx.getTerminalHistoryConfig();
+
     // Prepend the live tmux pane buffer so tab-switch replay shows the current
     // on-screen frame, not just the accumulated byte history. This matters for
     // TUI modes (codex/opencode) that repaint only their latest frame: the
@@ -1031,24 +1042,57 @@ export function registerSessionRoutes(
     const muxName = session.muxName;
     const liveMuxBuffer =
       muxName && typeof ctx.mux.captureActivePaneBuffer === 'function'
-        ? ctx.mux.captureActivePaneBuffer(muxName)
+        ? ctx.mux.captureActivePaneBuffer(
+            muxName,
+            isFullReload
+              ? { fullHistory: true, historyLimitLines: tmuxHistoryLimit, maxCaptureBytes: terminalBufferMaxBytes }
+              : undefined
+          )
         : null;
-    const rawBuffer =
-      liveMuxBuffer !== null && liveMuxBuffer.length > 0
-        ? session.terminalBufferLength > 0
+    const hasLiveMuxBuffer = liveMuxBuffer !== null && liveMuxBuffer.length > 0;
+    const source: 'history' | 'mux-visible' | 'mux-full-history' = hasLiveMuxBuffer
+      ? isFullReload
+        ? 'mux-full-history'
+        : 'mux-visible'
+      : 'history';
+    let rawBuffer: string;
+    if (liveMuxBuffer !== null && liveMuxBuffer.length > 0) {
+      // Full-history capture is the RENDERED form of everything already in the
+      // byte buffer (up to tmux eviction) — return it alone. Prepending the byte
+      // history would replay the whole conversation twice: `\x1b[2J` clears only
+      // the viewport, not xterm scrollback. The history+clear+frame concat stays
+      // for the visible-frame path, where the single pane frame lacks history.
+      rawBuffer = isFullReload
+        ? liveMuxBuffer
+        : session.terminalBufferLength > 0
           ? `${session.terminalBuffer}\x1b[H\x1b[2J${liveMuxBuffer}`
-          : liveMuxBuffer
-        : session.terminalBuffer;
-    const tailBytes = query.tail ? parseInt(query.tail, 10) : 0;
+          : liveMuxBuffer;
+    } else {
+      rawBuffer = session.terminalBuffer;
+    }
     const fullSize = rawBuffer.length;
     let truncated = false;
     let cleanBuffer: string;
 
+    // Cap the payload EARLY — before the regex normalization passes below run
+    // over it. A full-history tmux capture can be tens of MB of scrollback;
+    // normalizing all of it would stall the event loop only to discard most
+    // bytes anyway. Keep the most RECENT bytes (slice from the end) and align
+    // to a line boundary so we never start mid-ANSI-escape.
+    if (terminalBufferMaxBytes > 0 && rawBuffer.length > terminalBufferMaxBytes) {
+      rawBuffer = rawBuffer.slice(-terminalBufferMaxBytes);
+      truncated = true;
+      const capNewline = rawBuffer.indexOf('\n');
+      if (capNewline > 0 && capNewline < 4096) {
+        rawBuffer = rawBuffer.slice(capNewline + 1);
+      }
+    }
+
     // Strip redundant Ink spinner/status redraws BEFORE tailing.
     // During long thinking phases, Ink rewrites the same rows thousands of times
     // (500KB+). Without stripping, tail mode returns only spinner frames and
     // the terminal appears empty when switching tabs.
-    let strippedBuffer = stripInkRedrawBloat(rawBuffer);
+    let strippedBuffer = session.mode === 'shell' ? rawBuffer : stripInkRedrawBloat(rawBuffer);
 
     // Strip alt-screen toggles and scrollback-erase from Codex/Claude byte
     // streams. xterm.js obeys them by switching to its scrollback-less alt
@@ -1096,6 +1140,7 @@ export function registerSessionRoutes(
       status: session.status,
       fullSize,
       truncated,
+      source,
     };
   });