fix(backend): stop Sprites prepare truncating git history and leaking the clone token (#3411)
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com> Co-authored-by: Carlos Florencio <carlosmflorencio@gmail.com>
This commit is contained in:
Jonathan Marbutt
@@ -1,7 +1,17 @@
|
||||
package lifecycle
|
||||
|
||||
import (
|
||||
"crypto/sha256"
|
||||
"encoding/hex"
|
||||
)
|
||||
|
||||
const executorTypeSSH = "ssh"
|
||||
|
||||
// This hash identifies the built-in Sprites script that shipped before the
|
||||
// in-place, full-history materialization flow. Upgrade only that exact value;
|
||||
// user-authored prepare scripts must remain unchanged.
|
||||
const legacySpritesPrepareScriptSHA256 = "e656f9e51496e1bb1e5cee205058bbeaa7fe1ab37073f600e6c94310ead7be4d"
|
||||
|
||||
// DefaultPrepareScript returns the default prepare script for a given executor type string.
|
||||
func DefaultPrepareScript(executorType string) string {
|
||||
switch executorType {
|
||||
@@ -22,6 +32,11 @@ func DefaultPrepareScript(executorType string) string {
|
||||
}
|
||||
}
|
||||
|
||||
func isLegacySpritesPrepareScript(script string) bool {
|
||||
digest := sha256.Sum256([]byte(script))
|
||||
return hex.EncodeToString(digest[:]) == legacySpritesPrepareScriptSHA256
|
||||
}
|
||||
|
||||
// KandevBranchCheckoutPostlude returns a kandev-managed shell snippet that
|
||||
// guarantees the session's feature branch is checked out inside the
|
||||
// workspace, no matter what the user's stored prepare_script does.
|
||||
@@ -57,17 +72,25 @@ func KandevBranchCheckoutPostlude() string {
|
||||
# quoted literal that cannot inject commands. Do NOT wrap these in double
|
||||
# quotes — double quotes would re-expose $(...) command substitution. Do NOT
|
||||
# assume they are unquoted data either; the value carries its own quoting.
|
||||
#
|
||||
# Keep the complete commit graph so diff and rebase operations have a common
|
||||
# ancestor.
|
||||
worktree_branch={{worktree.branch}}
|
||||
repository_branch={{repository.branch}}
|
||||
(
|
||||
if [ -d {{workspace.path}}/.git ] \
|
||||
&& [ -n {{worktree.branch}} ] \
|
||||
&& [ {{worktree.branch}} != {{repository.branch}} ]; then
|
||||
&& [ -n "$worktree_branch" ] \
|
||||
&& [ "$worktree_branch" != "$repository_branch" ]; then
|
||||
cd {{workspace.path}} || exit 0
|
||||
if git rev-parse --verify {{worktree.branch}} >/dev/null 2>&1; then
|
||||
git checkout {{worktree.branch}}
|
||||
elif git fetch --depth=1 origin {{worktree.branch}} 2>/dev/null; then
|
||||
git checkout -b {{worktree.branch}} origin/{{worktree.branch}}
|
||||
if [ "$(git rev-parse --is-shallow-repository 2>/dev/null || echo false)" = true ]; then
|
||||
git fetch --unshallow --no-tags origin
|
||||
fi
|
||||
if git rev-parse --verify "$worktree_branch" >/dev/null 2>&1; then
|
||||
git checkout "$worktree_branch"
|
||||
elif git fetch --no-tags origin "+refs/heads/${worktree_branch}:refs/remotes/origin/${worktree_branch}" 2>/dev/null; then
|
||||
git checkout -b "$worktree_branch" "origin/$worktree_branch"
|
||||
else
|
||||
git checkout -b {{worktree.branch}}
|
||||
git checkout -b "$worktree_branch"
|
||||
fi
|
||||
fi
|
||||
) || true
|
||||
@@ -204,40 +227,106 @@ const defaultSpritesPrepareScript = `#!/bin/bash
|
||||
|
||||
set -euo pipefail
|
||||
|
||||
# ---- Add SSH host keys (prevent "Host key verification failed") ----
|
||||
mkdir -p ~/.ssh
|
||||
ssh-keyscan -t ed25519 github.com gitlab.com bitbucket.org >> ~/.ssh/known_hosts 2>/dev/null
|
||||
# SECURITY: the providers substitute a fully single-quoted, self-contained
|
||||
# shell token (scriptengine.shellQuote) for every DATA placeholder. Each one is
|
||||
# dereferenced BARE exactly once, into a shell variable, and used as "$var"
|
||||
# afterwards. Never write "{{...}}" — double quotes re-expose $(...) inside a
|
||||
# hostile branch name or clone URL, which is the injection the quoting exists
|
||||
# to prevent.
|
||||
workspace={{workspace.path}}
|
||||
repository_url={{repository.clone_url}}
|
||||
repository_branch={{repository.branch}}
|
||||
|
||||
# ---- Configure git/gh for HTTPS auth (token-based, no SSH keys needed) ----
|
||||
# Rewrite SSH URLs to HTTPS so git clone git@github.com:... works via token auth
|
||||
git config --global url."https://github.com/".insteadOf "git@github.com:"
|
||||
git config --global url."https://github.com/".insteadOf "ssh://git@github.com/"
|
||||
|
||||
# Kandev can write into the workspace before this script runs, so the checkout
|
||||
# is not guaranteed to be owned by the user running the prepare script.
|
||||
git config --global --add safe.directory '*'
|
||||
|
||||
# Configure GitHub token for gh CLI and git operations
|
||||
# GH_TOKEN is the primary env var for gh CLI authentication
|
||||
{{github.auth_setup}}
|
||||
|
||||
# ---- Install pnpm globally ----
|
||||
curl -fsSL https://github.com/pnpm/pnpm/releases/download/v10.32.1/pnpm-linux-x64 -o /usr/local/bin/pnpm
|
||||
chmod +x /usr/local/bin/pnpm
|
||||
# ---- Install pnpm (best effort, never fatal) ----
|
||||
# Use corepack so the repository packageManager value controls the pnpm version.
|
||||
# Package-manager setup errors must not stop sandbox preparation.
|
||||
if ! command -v pnpm >/dev/null 2>&1; then
|
||||
corepack enable pnpm >/dev/null 2>&1 || true
|
||||
fi
|
||||
if ! pnpm --version >/dev/null 2>&1; then
|
||||
npm install -g pnpm >/dev/null 2>&1 || true
|
||||
fi
|
||||
|
||||
# ---- Git identity ----
|
||||
{{git.identity_setup}}
|
||||
|
||||
# ---- Clone repository ----
|
||||
# ---- Materialize the primary repository ----
|
||||
# Initialize in place because Office files can exist before preparation starts.
|
||||
# Fetch a blobless commit graph so later diff and rebase operations retain history.
|
||||
# The kandev-managed feature-branch checkout is appended as an invariant
|
||||
# postlude (see KandevBranchCheckoutPostlude) — keep it out of the default
|
||||
# so old profiles snapshotting this script and the postlude never disagree.
|
||||
# SECURITY: the providers substitute fully single-quoted tokens (shellQuote) for
|
||||
# these placeholders. printf takes them as separate arguments, NOT inside a
|
||||
# double-quoted string (which would re-expand a hostile URL/branch). Do not
|
||||
# reintroduce an echo of these placeholders inside a double-quoted string.
|
||||
printf 'Cloning %s (branch: %s)...\n' {{repository.clone_url}} {{repository.branch}}
|
||||
git clone --depth=1 --quiet --branch {{repository.branch}} {{repository.clone_url}} {{workspace.path}}
|
||||
cd {{workspace.path}}
|
||||
# Never print $repository_url because it can contain a credential.
|
||||
mkdir -p "$workspace"
|
||||
if [ -n "$repository_url" ]; then
|
||||
scrub_origin() {
|
||||
origin_url=$(git -C "$workspace" remote get-url origin 2>/dev/null || true)
|
||||
if [ -n "$origin_url" ]; then
|
||||
clean_origin_url=$(printf '%s' "$origin_url" | sed 's|https://[^@]*@github.com/|https://github.com/|')
|
||||
git -C "$workspace" remote set-url origin "$clean_origin_url" 2>/dev/null || true
|
||||
fi
|
||||
}
|
||||
trap scrub_origin EXIT HUP INT TERM
|
||||
if git -C "$workspace" rev-parse --git-dir >/dev/null 2>&1; then
|
||||
configured_url=$(git -C "$workspace" config --get remote.origin.url 2>/dev/null || true)
|
||||
if [ -z "$configured_url" ]; then
|
||||
git -C "$workspace" remote add origin "$repository_url"
|
||||
else
|
||||
configured_id=$(printf '%s' "$configured_url" | sed 's|://[^/@]*@|://|')
|
||||
expected_id=$(printf '%s' "$repository_url" | sed 's|://[^/@]*@|://|')
|
||||
if [ "$configured_id" != "$expected_id" ]; then
|
||||
echo 'kandev: target origin identity conflict' >&2
|
||||
exit 1
|
||||
fi
|
||||
git -C "$workspace" remote set-url origin "$repository_url"
|
||||
fi
|
||||
else
|
||||
git init -q "$workspace"
|
||||
git -C "$workspace" remote add origin "$repository_url"
|
||||
fi
|
||||
git_dir=$(git -C "$workspace" rev-parse --absolute-git-dir)
|
||||
exclude_file="$git_dir/info/exclude"
|
||||
mkdir -p "$(dirname "$exclude_file")"
|
||||
touch "$exclude_file"
|
||||
grep -Fqx '/.kandev/' "$exclude_file" || printf '%s\n' '/.kandev/' >>"$exclude_file"
|
||||
|
||||
# Strip embedded token from remote URL to avoid persisting credentials in .git/config
|
||||
git remote set-url origin "$(git remote get-url origin | sed 's|https://[^@]*@github.com/|https://github.com/|')" 2>/dev/null || true
|
||||
if [ -z "$repository_branch" ]; then
|
||||
echo 'kandev: target repository has no base branch' >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
git -C "$workspace" config remote.origin.promisor true
|
||||
git -C "$workspace" config remote.origin.partialclonefilter blob:none
|
||||
echo "Fetching $repository_branch..."
|
||||
if ! git -C "$workspace" fetch --filter=blob:none --no-tags origin \
|
||||
"+refs/heads/${repository_branch}:refs/remotes/origin/${repository_branch}"; then
|
||||
echo 'kandev: target base branch is unavailable' >&2
|
||||
exit 1
|
||||
fi
|
||||
# A reused checkout keeps its current branch, local commits, and untracked
|
||||
# files. Only an empty repository needs its initial branch created.
|
||||
if ! git -C "$workspace" rev-parse --verify HEAD >/dev/null 2>&1; then
|
||||
git -C "$workspace" checkout -q -b "$repository_branch" "refs/remotes/origin/$repository_branch"
|
||||
fi
|
||||
|
||||
scrub_origin
|
||||
trap - EXIT HUP INT TERM
|
||||
fi
|
||||
|
||||
cd "$workspace"
|
||||
|
||||
# ---- Repository setup (if configured) ----
|
||||
{{repository.setup_script}}
|
||||
|
||||
Reference in New Issue
Block a user