Keep long /continue sessions when the first user prompt is past 32KB

frontends/continue_cmd.py

@@ -261,6 +261,49 @@ def _preview_from_file(path):
     lu = _last_user(tail_s) or _last_user(head.decode('utf-8', errors='replace'))
     if lu:
         return ' '.join(lu.split())[:120]
+    scanned = _preview_from_scan(path, sz)
+    if scanned:
+        return scanned
+    return ''
+
+
+def _first_real_user(text):
+    """First typed user prompt in `text`, skipping tool_result continuations."""
+    for label, body in _BLOCK_RE.findall(text or ''):
+        if label == 'Prompt':
+            t = _user_text(body)
+            if t:
+                return t
+    return ''
+
+
+def _preview_from_scan(path, sz):
+    """Find a user prompt that sits between the 32KB head and tail windows.
+
+    Long tool-heavy sessions often park the first real user text past the head
+    window, with no <summary> in the tail either. list_sessions() then drops
+    the log. Scan extra windows up to _GREP_WIN only when head+tail missed.
+    """
+    if sz <= _PREVIEW_WIN * 2:
+        return ''
+    limit = min(sz, _GREP_WIN)
+    try:
+        with open(path, 'rb') as fh:
+            offset = 0
+            carry = b''
+            while offset < limit:
+                fh.seek(offset)
+                chunk = fh.read(_PREVIEW_WIN)
+                if not chunk:
+                    break
+                text = (carry + chunk).decode('utf-8', errors='replace')
+                t = _first_real_user(text)
+                if t:
+                    return ' '.join(t.split())[:120]
+                carry = chunk[-4096:]
+                offset += len(chunk)
+    except OSError:
+        return ''
     return ''
 
 

frontends/tests/test_continue_preview.py

@@ -0,0 +1,44 @@
+"""Preview extraction for /continue when the first user prompt sits past the head window."""
+import json
+import sys
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parent.parent.parent
+sys.path.insert(0, str(ROOT / "frontends"))
+
+from continue_cmd import _preview_from_file, list_sessions  # noqa: E402
+
+
+def _block(prompt_obj, response_text="ok"):
+    prompt = json.dumps(prompt_obj, ensure_ascii=False)
+    return (
+        f"=== Prompt === extra\n{prompt}\n"
+        f"=== Response === extra\n[{{'type': 'text', 'text': {response_text!r}}}]\n"
+    )
+
+
+def _tool_block():
+    return _block(
+        {"role": "user", "content": [{"type": "tool_result", "content": "x" * 400}]}
+    )
+
+
+def test_preview_finds_user_prompt_past_32kb_head(tmp_path, monkeypatch):
+    log_dir = tmp_path / "temp" / "model_responses"
+    log_dir.mkdir(parents=True)
+    path = log_dir / "model_responses_4242.txt"
+
+    user_text = "Index remaining PDFs using focr house style and skip Jira keys."
+    prefix = _tool_block() * 80  # well over 32KB of tool_result rounds
+    suffix = _tool_block() * 80
+    middle = _block({"role": "user", "content": [{"type": "text", "text": user_text}]})
+    path.write_text(prefix + middle + suffix, encoding="utf-8")
+    assert path.stat().st_size > 64 * 1024
+
+    preview = _preview_from_file(str(path))
+    assert "Index remaining PDFs" in preview
+
+    monkeypatch.setattr("continue_cmd._LOG_GLOB", str(log_dir / "model_responses_*.txt"))
+    monkeypatch.setattr("continue_cmd._ROUNDS_CACHE_PATH", str(tmp_path / "continue_rounds_cache.json"))
+    sessions = list_sessions()
+    assert any("Index remaining PDFs" in item[2] for item in sessions)