From 7db7714ff7c6859654584b90e1b9a188559fc45b Mon Sep 17 00:00:00 2001
From: Mingyang Wu <aprylewu@gmail.com>
Date: Mon, 21 Sep 2026 18:02:14 +0800
Subject: [PATCH] Fix rehash repair of individually corrupted shims (#3540)

---
 libexec/pyenv-rehash |  47 ++++++++-----
 test/rehash.bats     | 164 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 195 insertions(+), 16 deletions(-)

diff --git a/libexec/pyenv-rehash b/libexec/pyenv-rehash
index 19d94274..cbc26597 100755
--- a/libexec/pyenv-rehash
+++ b/libexec/pyenv-rehash
@@ -80,10 +80,8 @@ fi
 unset tested_for_other_write_errors
 
 # The prototype shim file is a script that re-execs itself, passing
-# its filename and any arguments to `pyenv exec`. This file is
-# hard-linked for every executable and then removed. The linking
-# technique is fast, uses less disk space than unique files, and also
-# serves as a locking mechanism.
+# its filename and any arguments to `pyenv exec`. This file is copied
+# for every executable and then removed. It also serves as a lock.
 create_prototype_shim() {
   cat > "$PROTOTYPE_SHIM_PATH" <<SH
 #!/usr/bin/env bash
@@ -100,18 +98,30 @@ fi
 exec "$(command -v pyenv)" exec "\$program" "\$@"
 SH
   chmod +x "$PROTOTYPE_SHIM_PATH"
+  local LC_ALL=C
+  IFS= read -r -d '' PROTOTYPE_SHIM_CONTENTS < "$PROTOTYPE_SHIM_PATH" || true
+  PROTOTYPE_SHIM_READ_SIZE=$((${#PROTOTYPE_SHIM_CONTENTS} + 1))
 }
 
-# If the contents of the prototype shim file differ from the contents
-# of the first shim in the shims directory, assume pyenv has been
-# upgraded and the existing shims need to be removed.
-remove_outdated_shims() {
-  local shim
-  for shim in "$SHIM_PATH"/*; do
-    if ! diff "$PROTOTYPE_SHIM_PATH" "$shim" >/dev/null 2>&1; then
-      rm -f "$SHIM_PATH"/*
+# Install a shim when it is first registered, replacing an existing copy
+# if it was overwritten. Hooks can then customize it without a later
+# registration overwriting their changes.
+install_shims() {
+  local shim file shim_contents
+  # Count bytes consistently across Bash versions, including multibyte paths.
+  local LC_ALL=C
+  for shim; do
+    file="${SHIM_PATH}/${shim}"
+    if [[ -f "$file" || ( -L "$file" && ! -e "$file" ) ]]; then
+      shim_contents=
+      # Read with a builtin to avoid a subprocess per shim. A successful
+      # read means a NUL or extra character was found. Limit the read in
+      # case a large file was accidentally written over a shim.
+      if IFS= read -r -d '' -n "$PROTOTYPE_SHIM_READ_SIZE" shim_contents 2>/dev/null < "$file" || [[ $shim_contents != "$PROTOTYPE_SHIM_CONTENTS" ]]; then
+        rm -f "$file"
+      fi
     fi
-    break
+    [ -e "$file" ] || cp "$PROTOTYPE_SHIM_PATH" "$file"
   done
 }
 
@@ -132,7 +142,10 @@ if ((${BASH_VERSINFO[0]} > 3)); then
 
   # Registers the name of a shim to be generated.
   register_shim() {
-    registered_shims["$1"]=1
+    if [[ ! ${registered_shims["$1"]} ]]; then
+      install_shims "$1"
+      registered_shims["$1"]=1
+    fi
   }
 
   # Install all shims registered via `make_shims` or `register_shim` directly.
@@ -162,7 +175,10 @@ else # Same for bash < 4.
     registered_shims=" "
 
     register_shim() {
-      registered_shims="${registered_shims}${1} "
+      if [[ "$registered_shims" != *" ${1} "* ]]; then
+        install_shims "$1"
+        registered_shims="${registered_shims}${1} "
+      fi
     }
 
     install_registered_shims() {
@@ -188,7 +204,6 @@ shopt -s nullglob
 # Create the prototype shim, then register shims for all known
 # executables.
 create_prototype_shim
-remove_outdated_shims
 # shellcheck disable=SC2046
 make_shims $(pyenv-versions --executables)
 
diff --git a/test/rehash.bats b/test/rehash.bats
index 3d31d5eb..56651250 100755
--- a/test/rehash.bats
+++ b/test/rehash.bats
@@ -88,6 +88,170 @@ OUT
   assert [ ! -e "${PYENV_ROOT}/shims/oldshim1" ]
 }
 
+@test "repairs an overwritten shim after the first shim" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  printf '2\n' > "${PYENV_ROOT}/shims/python"
+
+  run pyenv-rehash
+  assert_success ""
+  assert cmp "${PYENV_ROOT}/shims/aaa" "${PYENV_ROOT}/shims/python"
+
+  PYENV_VERSION=3.4 run "${PYENV_ROOT}/shims/python"
+  assert_success
+}
+
+@test "repairs truncated shim contents and preserves trailing newlines" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+
+  for contents in empty newline nul; do
+    case "$contents" in
+      empty) : > "${PYENV_ROOT}/shims/python" ;;
+      newline) printf '\n' >> "${PYENV_ROOT}/shims/python" ;;
+      nul) printf '\0' >> "${PYENV_ROOT}/shims/python" ;;
+    esac
+    run pyenv-rehash
+    assert_success ""
+    assert cmp "${PYENV_ROOT}/shims/aaa" "${PYENV_ROOT}/shims/python"
+  done
+}
+
+@test "leaves valid shims unchanged when repairing another shim" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  touch -t 200001010000.00 "${PYENV_ROOT}/shims/aaa"
+  touch -t 200101010000.00 "${PYENV_TEST_DIR}/newer"
+  printf '2\n' > "${PYENV_ROOT}/shims/python"
+
+  run pyenv-rehash
+  assert_success ""
+  assert [ "${PYENV_ROOT}/shims/aaa" -ot "${PYENV_TEST_DIR}/newer" ]
+  assert cmp "${PYENV_ROOT}/shims/aaa" "${PYENV_ROOT}/shims/python"
+}
+
+@test "repairs an unreadable shim without additional output" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  chmod -r "${PYENV_ROOT}/shims/python"
+  skip_if_nonposix_security -r "${PYENV_ROOT}/shims/python"
+
+  run pyenv-rehash
+  assert_success ""
+  assert cmp "${PYENV_ROOT}/shims/aaa" "${PYENV_ROOT}/shims/python"
+}
+
+@test "does not read registered directories as shims" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  rm "${PYENV_ROOT}/shims/python"
+  mkdir "${PYENV_ROOT}/shims/python"
+
+  run pyenv-rehash
+  assert_success ""
+  assert [ -d "${PYENV_ROOT}/shims/python" ]
+}
+
+@test "leaves valid shims unchanged with a multibyte root path" {
+  [[ $(LC_ALL=en_US.UTF-8 locale charmap 2>/dev/null) == UTF-8 ]] || skip "-- UTF-8 locale not installed"
+  export LC_ALL=en_US.UTF-8
+  export PYENV_ROOT="${PYENV_ROOT}/pyenv 路径"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  touch -t 200001010000.00 "${PYENV_ROOT}/shims/python"
+  touch -t 200101010000.00 "${PYENV_TEST_DIR}/newer"
+
+  run pyenv-rehash
+  assert_success ""
+  assert [ "${PYENV_ROOT}/shims/python" -ot "${PYENV_TEST_DIR}/newer" ]
+}
+
+@test "repairs explicitly registered hidden shims and preserves other dotfiles" {
+  create_hook rehash hidden.bash <<'SH'
+register_shim .foo
+register_shim python
+SH
+  pyenv-rehash
+  printf 'keep\n' > "${PYENV_ROOT}/shims/.keep"
+  printf '2\n' > "${PYENV_ROOT}/shims/.foo"
+
+  run pyenv-rehash
+  assert_success ""
+  assert cmp "${PYENV_ROOT}/shims/python" "${PYENV_ROOT}/shims/.foo"
+  run cat "${PYENV_ROOT}/shims/.keep"
+  assert_success "keep"
+  assert [ ! -e "${PYENV_ROOT}/shims/.pyenv-shim" ]
+}
+
+@test "replaces a dangling registered symlink without writing through it" {
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "python"
+  pyenv-rehash
+  rm "${PYENV_ROOT}/shims/python"
+  ln -s "${PYENV_TEST_DIR}/missing/python" "${PYENV_ROOT}/shims/python"
+
+  run pyenv-rehash
+  assert_success ""
+  assert [ ! -L "${PYENV_ROOT}/shims/python" ]
+  assert [ ! -e "${PYENV_TEST_DIR}/missing/python" ]
+  assert cmp "${PYENV_ROOT}/shims/aaa" "${PYENV_ROOT}/shims/python"
+}
+
+@test "preserves sourceable shims from the built-in rehash hook" {
+  export PYENV_HOOK_PATH="${_PYENV_INSTALL_PREFIX}/pyenv.d"
+  export PYENV_VERSION=3.4
+  create_alt_executable_in_version "3.4" "aaa"
+  create_alt_executable_in_version "3.4" "activate" <<'SH'
+export PYENV_ACTIVATED=yes
+SH
+  create_alt_executable_in_version "3.4" "activate.fish"
+  create_alt_executable_in_version "3.4" "gettext.sh"
+
+  pyenv-rehash
+  run pyenv-rehash
+  assert_success ""
+  assert cmp "${PYENV_ROOT}/shims/activate" "${PYENV_ROOT}/shims/activate.fish"
+  assert cmp "${PYENV_ROOT}/shims/activate" "${PYENV_ROOT}/shims/gettext.sh"
+
+  run bash -c '. "$PYENV_ROOT/shims/activate"; echo "$PYENV_ACTIVATED"'
+  assert_success "yes"
+}
+
+@test "does not overwrite hook customizations when a shim is registered again" {
+  create_alt_executable_in_version "3.4" "python"
+  create_hook rehash custom.bash <<'SH'
+register_shim .foo
+printf 'custom\n' > "$SHIM_PATH/.foo"
+printf 'custom\n' > "$SHIM_PATH/python"
+register_shim python
+register_shim .foo
+SH
+
+  run pyenv-rehash
+  assert_success ""
+  run cat "${PYENV_ROOT}/shims/python"
+  assert_success "custom"
+  run cat "${PYENV_ROOT}/shims/.foo"
+  assert_success "custom"
+}
+
+@test "repairs existing shims before rehash hooks invoke them" {
+  create_alt_executable_in_version "3.4" "python" "echo works"
+  pyenv-rehash
+  printf '2\n' > "${PYENV_ROOT}/shims/python"
+  create_hook rehash invoke.bash <<'SH'
+PYENV_VERSION=3.4 "$SHIM_PATH/python"
+SH
+
+  run pyenv-rehash
+  assert_success "works"
+}
+
 @test "binary install locations containing spaces" {
   create_alt_executable_in_version "dirname1 p247" "python"
   create_alt_executable_in_version "dirname2 preview1" "py.test"