Files
bot-bottle/bot_bottle/git_gate_host_key.py
T
didericis-claude d496e30681 refactor(git-gate): replace regex frontmatter edit with parse→mutate→serialize
The regex-based _insert_host_key_in_frontmatter had a bug: child_indent
was overwritten on every line with indent > repo_indent (including
grandchildren), so host_key landed inside the key: block instead of at
the repo level.

Replace with a clean parse → mutate → re-serialize approach:
- Add serialize_yaml_subset() to yaml_subset.py (block-style, 2-space
  indent, round-trips through parse_yaml_subset)
- Replace _insert_host_key_in_frontmatter + _update_frontmatter_in_file
  with _add_host_key_to_frontmatter that parses the frontmatter dict,
  sets fm["git-gate"]["repos"][name]["host_key"], and re-serializes
- _find_and_update_bottle_file simplified to read file → delegate to
  _add_host_key_to_frontmatter → write if changed
2026-07-09 17:21:57 +00:00

242 lines
8.4 KiB
Python

"""Preflight host-key population for git-gate upstreams (issue #333).
When a git-gate repo entry lacks a `host_key`, this module either:
- headless: dies with a clear config error.
- interactive: fetches the key via ssh-keyscan, prompts the operator to
confirm, and optionally persists it to the bottle config file on disk.
Public entry point: `preflight_host_keys(manifest, headless=..., home_md=...)`.
"""
from __future__ import annotations
import dataclasses
import subprocess
import sys
from pathlib import Path
from typing import cast
from .log import die, info
from .manifest import Manifest
from .yaml_subset import YamlSubsetError, parse_frontmatter, serialize_yaml_subset
# Preferred key types, most secure first.
_KEY_TYPE_PREFERENCE = (
"ssh-ed25519",
"ecdsa-sha2-nistp256",
"ecdsa-sha2-nistp384",
"ecdsa-sha2-nistp521",
"ssh-rsa",
)
def fetch_host_key(host: str, port: str) -> str:
"""Return an SSH public key for `host`:`port` via ssh-keyscan.
Returns the key in `<type> <base64-data>` format (the host prefix is
stripped so the result can be stored in `host_key` and later formatted
into a known_hosts line by `git_gate_known_hosts_line`).
Prefers ed25519 > ecdsa > rsa; falls back to the first key type
returned if none of the preferred types are present.
Raises `RuntimeError` on subprocess failure, timeout, or no result.
Uses only the Python stdlib (subprocess)."""
args = ["ssh-keyscan"]
if port and port != "22":
args += ["-p", port]
args.append(host)
try:
result = subprocess.run(
args, capture_output=True, text=True, timeout=15, check=False,
)
except OSError as e:
raise RuntimeError(
f"ssh-keyscan: could not launch for {host}:{port}: {e}"
) from e
except subprocess.TimeoutExpired as e:
raise RuntimeError(f"ssh-keyscan timed out for {host}:{port}") from e
# known_hosts format: "[host]:port type data" or "host type data"
# Strip the host/port prefix; collect "type -> type data" by type.
found: dict[str, str] = {}
for line in result.stdout.splitlines():
line = line.strip()
if not line or line.startswith("#"):
continue
parts = line.split(None, 2)
if len(parts) == 3 and parts[1] not in found:
found[parts[1]] = f"{parts[1]} {parts[2]}"
for preferred in _KEY_TYPE_PREFERENCE:
if preferred in found:
return found[preferred]
if found:
return next(iter(found.values()))
raise RuntimeError(
f"ssh-keyscan returned no host key for {host}:{port}."
+ (f" stderr: {result.stderr.strip()!r}" if result.stderr.strip() else "")
)
# ---------------------------------------------------------------------------
# Frontmatter editing
# ---------------------------------------------------------------------------
def _add_host_key_to_frontmatter(file_text: str, repo_name: str, host_key: str) -> str:
"""Return an updated copy of `file_text` with `host_key` set on the
named repo entry in the YAML frontmatter.
Parses the frontmatter into a dict, sets the key, and re-serializes.
Returns the original text unchanged when: the file has no frontmatter,
the git-gate.repos.<repo_name> entry is absent or already has a
host_key, or the frontmatter cannot be parsed."""
try:
fm, body = parse_frontmatter(file_text)
except YamlSubsetError:
return file_text
git_gate = fm.get("git-gate")
if not isinstance(git_gate, dict):
return file_text
repos = git_gate.get("repos")
if not isinstance(repos, dict):
return file_text
repo = repos.get(repo_name)
if not isinstance(repo, dict):
return file_text
if repo.get("host_key"):
return file_text
cast(dict[str, object], repo)["host_key"] = host_key
return f"---\n{serialize_yaml_subset(fm)}---\n{body}"
def _find_and_update_bottle_file(
bottles_dir: Path, repo_name: str, host_key: str,
) -> bool:
"""Scan `bottles_dir/*.md` for the first file that directly declares
`repo_name` without a `host_key`, update it, and return True.
Walks files in sorted order so the result is deterministic. Does NOT
follow `extends:` chains — only files that explicitly declare the repo
entry are candidates. Returns False when no such file is found."""
if not bottles_dir.is_dir():
return False
for path in sorted(bottles_dir.glob("*.md")):
try:
text = path.read_text(encoding="utf-8")
except (OSError, UnicodeDecodeError):
continue
updated = _add_host_key_to_frontmatter(text, repo_name, host_key)
if updated == text:
continue
path.write_text(updated, encoding="utf-8")
info(f"wrote host_key for {repo_name!r} to {path}")
return True
return False
# ---------------------------------------------------------------------------
# Interactive prompt helper
# ---------------------------------------------------------------------------
def _prompt_tty(message: str) -> str:
"""Write `message` to stderr and read a line from /dev/tty (or stdin)."""
sys.stderr.write(message)
sys.stderr.flush()
try:
with open("/dev/tty", "r", encoding="utf-8") as tty:
return tty.readline().rstrip("\n")
except OSError:
return sys.stdin.readline().rstrip("\n")
# ---------------------------------------------------------------------------
# Public API
# ---------------------------------------------------------------------------
def preflight_host_keys(
manifest: Manifest,
*,
headless: bool,
home_md: Path | None,
) -> Manifest:
"""Ensure every git-gate repo entry has a `host_key` configured.
For entries whose `KnownHostKey` is empty:
- headless: calls `die()` with a clear message naming the repos.
- interactive: fetches the key via ssh-keyscan, shows it to the
operator, and requests confirmation. If accepted, optionally
persists it to the bottle config file on disk; the key is always
applied in memory for this launch regardless of the persistence
choice. Aborted confirmation calls `die()`.
Returns a (possibly updated) Manifest. If all entries already have
host keys the original manifest is returned unchanged."""
bottle = manifest.bottle
missing = [e for e in bottle.git if not e.KnownHostKey]
if not missing:
return manifest
if headless:
names = ", ".join(repr(e.Name) for e in missing)
die(
f"git-gate: no host_key configured for repo(s) {names}. "
f"Add host_key to each bottle git-gate.repos entry, or run "
f"interactively once to have it fetched and saved automatically."
)
bottles_dir = (home_md / "bottles") if home_md is not None else None
updated_entries = list(bottle.git)
for entry in missing:
host = entry.UpstreamHost
port = entry.UpstreamPort
label = f"git-gate.repos[{entry.Name!r}]"
info(f"{label}: no host_key configured; fetching from {host}:{port}")
try:
key = fetch_host_key(host, port)
except RuntimeError as e:
die(f"git-gate: {label}: {e}")
sys.stderr.write(f"\ngit-gate: host key for {label}:\n {key}\n\n")
confirm = _prompt_tty("Is this host key correct? [y/N] ")
if confirm.strip().lower() not in ("y", "yes"):
die(f"git-gate: {label}: host key not confirmed; aborting launch")
if bottles_dir is not None:
save = _prompt_tty(
f"Save host_key for {entry.Name!r} to the bottle config file? [y/N] "
)
if save.strip().lower() in ("y", "yes"):
ok = _find_and_update_bottle_file(bottles_dir, entry.Name, key)
if not ok:
sys.stderr.write(
f"git-gate: {label}: bottle file not found or not writable; "
f"host_key kept in memory for this session only\n"
)
idx = next(i for i, e in enumerate(updated_entries) if e.Name == entry.Name)
updated_entries[idx] = dataclasses.replace(entry, KnownHostKey=key)
updated_bottle = dataclasses.replace(bottle, git=tuple(updated_entries))
return dataclasses.replace(manifest, bottle=updated_bottle)
__all__ = [
"fetch_host_key",
"preflight_host_keys",
"_add_host_key_to_frontmatter",
"_find_and_update_bottle_file",
]