From 33855f1b30f470ee2ba9bd8c28196296d3a667bb Mon Sep 17 00:00:00 2001
From: kshitij <82637225+kshitijk4poor@users.noreply.github.com>
Date: Wed, 12 Aug 2026 13:54:08 +0530
Subject: [PATCH] perf(tools): linear-time masking rebuild + last-opener early
 exit

Efficiency review (measured with timeit probes) found two unbounded
costs on adversarial inputs:

- The masked-range rebuild copied the whole string once per range
  (O(n*k)): 50k tiny heredocs took 1.7s. Replaced with a single-pass
  segment join over the (sorted, non-overlapping) ranges: 152ms, and
  newlines are now counted on the original command instead of
  re-slicing.
- After the last '<<' occurrence no opener can start, but the scanner
  still walked the remaining text per-char: one heredoc followed by a
  1MB tail cost ~150ms. An rfind bound breaks out of the unit loop
  once the scan passes it: 0.3ms.

Typical commands are unaffected (the '<<' fast path already returns
first). 30/30 guard tests pass; mutation check re-run on the final
stack (no-op mutation -> 11 tests fail, restore -> green).
---
 tools/shell_heredoc.py | 23 ++++++++++++++++++-----
 1 file changed, 18 insertions(+), 5 deletions(-)

diff --git a/tools/shell_heredoc.py b/tools/shell_heredoc.py
index 53568783cf..eef9a8322e 100644
--- a/tools/shell_heredoc.py
+++ b/tools/shell_heredoc.py
@@ -296,8 +296,13 @@ def strip_inert_heredoc_bodies(command: str) -> str:
     # machine entirely. This function runs on every terminal tool call.
     if "<<" not in command:
         return command
+    # No heredoc opener can start after the last '<<' occurrence; once the
+    # scan passes it, the rest of the command needs no per-char walk.
+    last_opener_index = command.rfind("<<")
 
     while command_start < len(command):
+        if command_start > last_opener_index:
+            break
         command_end, specs, unknown_operator, has_list_operator = (
             _scan_heredoc_command_unit(command, command_start)
         )
@@ -339,8 +344,16 @@ def strip_inert_heredoc_bodies(command: str) -> str:
                 ranges.extend(body_ranges)
         command_start = body_cursor
 
-    result = command
-    for start, end in reversed(ranges):
-        replacement = "\n" * result[start:end].count("\n")
-        result = result[:start] + replacement + result[end:]
-    return result
+    if not ranges:
+        return command
+    # Single-pass rebuild: ranges are sorted and non-overlapping, so join the
+    # kept segments with newline-preserving replacements (avoids a quadratic
+    # full-string copy per masked range on heredoc-heavy commands).
+    parts: list[str] = []
+    previous = 0
+    for start, end in ranges:
+        parts.append(command[previous:start])
+        parts.append("\n" * command.count("\n", start, end))
+        previous = end
+    parts.append(command[previous:])
+    return "".join(parts)