fix(scripts): preserve uv extras across make dev restarts (#2754) (#2767)

`make dev` ran `uv sync` unconditionally on every restart, wiping any
optional extras the user had installed manually with
`uv sync --all-packages --extra postgres`. The Docker image-build path
already solved this via the `UV_EXTRAS` build-arg in backend/Dockerfile;
the local serve.sh path and the docker-compose-dev startup command
were the remaining outliers.

`scripts/serve.sh` now resolves extras before `uv sync`:
  1. honors `UV_EXTRAS` (parity with backend/Dockerfile and
     docker/docker-compose.yaml — no new convention introduced);
  2. falls back to parsing config.yaml — `database.backend: postgres`
     or legacy `checkpointer.type: postgres` auto-pins
     `--extra postgres`, so the common case needs zero extra config.
  3. detector stderr is no longer suppressed, so whitelist warnings or
     crashes surface to the dev terminal (review feedback).

Detection lives in `scripts/detect_uv_extras.py` (stdlib-only — has to
run before the venv exists). Extra names are validated against
`^[A-Za-z][A-Za-z0-9_-]*$` so a stray shell metacharacter in `.env`
cannot reach `uv sync` downstream (defense in depth).

`docker/docker-compose-dev.yaml`'s startup command is now extracted to
`docker/dev-entrypoint.sh` (review feedback — the inline command had
grown to a ~350-char one-liner). The script:
  - parses comma/whitespace-separated UV_EXTRAS, applying the same
    `^[A-Za-z][A-Za-z0-9_-]*$` whitelist as the local detector;
  - emits one `--extra X` flag per token, so `UV_EXTRAS=postgres,ollama`
    works in Docker dev too (harmonized with local — review feedback);
  - calls `uv sync --all-packages` (PR #2584) so workspace member
    extras (deerflow-harness's postgres extra) are installed;
  - keeps the existing self-heal `(uv sync || (recreate venv && retry))`
    branch;
  - exposes `--print-extras` for dry-run testing.

The compose file mounts the script read-only at runtime, so script
edits take effect on `make docker-restart` without an image rebuild.

The `--no-sync` alternative (a separate suggestion in the issue thread)
was considered but rejected for dev paths because it would drop the
self-heal branch and the auto-pickup of new pyproject deps. `--no-sync`
is already in use for the production CMD (`backend/Dockerfile:101`)
where it's appropriate.

Updates the asyncpg-missing error message to include the
`--all-packages` flag (matching #2584) plus the persistent install flow,
and expands `config.example.yaml` so all three install paths
(local / docker dev / docker image build) are documented with their
multi-extra capabilities.

Tests:
  - `tests/test_detect_uv_extras.py` (21 tests) — local-path env parsing,
    YAML edge cases, env-vs-config precedence, whitelist rejection of
    shell metacharacters.
  - `tests/test_dev_entrypoint.py` (15 tests) — docker-path validation
    via `--print-extras`, multi-extra parsing, metacharacter abort.
  - `tests/test_persistence_scaffold.py` (22 tests, unchanged) — passes
    with the merged `--all-packages --extra postgres` error message.

Co-authored-by: Willem Jiang <willem.jiang@gmail.com>
This commit is contained in:
Xinmin Zeng 2026-05-10 22:28:29 +08:00 committed by GitHub
parent 5127f08e1a
commit 94da8f67d7
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
8 changed files with 636 additions and 6 deletions

View File

@ -82,7 +82,14 @@ async def init_engine(
import asyncpg # noqa: F401
except ImportError:
raise ImportError(
"database.backend is set to 'postgres' but asyncpg is not installed.\nInstall it with:\n uv sync --all-packages --extra postgres\nOr switch to backend: sqlite in config.yaml for single-node deployment."
"database.backend is set to 'postgres' but asyncpg is not installed.\n"
"Install it with:\n"
" cd backend && uv sync --all-packages --extra postgres\n"
"On the next `make dev` the postgres extra is auto-detected from\n"
"config.yaml (database.backend: postgres) and reinstalled, so it\n"
"will not be wiped again. Set UV_EXTRAS=postgres in .env to opt in\n"
"explicitly. Or switch to backend: sqlite in config.yaml for\n"
"single-node deployment."
) from None
if backend == "sqlite":

View File

@ -0,0 +1,201 @@
"""Unit tests for scripts/detect_uv_extras.py.
The detector resolves uv extras for `make dev` so that postgres (and any
future opt-in extras) are not wiped on every restart see Issue #2754.
"""
from __future__ import annotations
import importlib.util
from pathlib import Path
import pytest
REPO_ROOT = Path(__file__).resolve().parents[2]
DETECT_SCRIPT_PATH = REPO_ROOT / "scripts" / "detect_uv_extras.py"
spec = importlib.util.spec_from_file_location("deerflow_detect_uv_extras", DETECT_SCRIPT_PATH)
assert spec is not None and spec.loader is not None
detect = importlib.util.module_from_spec(spec)
spec.loader.exec_module(detect)
@pytest.fixture
def isolated_cwd(tmp_path, monkeypatch):
"""Isolate `find_config_file()` from the real repo by chdir + clearing env."""
monkeypatch.chdir(tmp_path)
monkeypatch.delenv("UV_EXTRAS", raising=False)
monkeypatch.delenv("DEER_FLOW_CONFIG_PATH", raising=False)
return tmp_path
def test_parse_env_extras_supports_comma_and_whitespace():
assert detect.parse_env_extras("postgres") == ["postgres"]
assert detect.parse_env_extras("postgres,ollama") == ["postgres", "ollama"]
assert detect.parse_env_extras("postgres ollama") == ["postgres", "ollama"]
assert detect.parse_env_extras(" postgres , ollama ,") == ["postgres", "ollama"]
assert detect.parse_env_extras("") == []
def test_parse_env_extras_drops_shell_metacharacters(capsys):
"""A `.env` value containing shell injection bait must not pass through.
The whitelist guarantees the *bytes* that reach `uv sync` cannot include
shell metacharacters. Any name that looks identifier-like still survives
(uv itself will reject unknown extras with its own error), but `;`, `&`,
backticks, parentheses, slashes, etc. are stripped.
"""
# Pure-metacharacter inputs collapse to empty.
assert detect.parse_env_extras(";") == []
assert detect.parse_env_extras("$(whoami)") == []
assert detect.parse_env_extras("`echo bad`") == []
assert detect.parse_env_extras("postgres;evil") == [] # single token, contains `;`
# Splitting on whitespace yields ['rm'] which is identifier-shaped, but the
# destructive bits (`;`, `-rf`, `/`) are dropped.
assert detect.parse_env_extras("; rm -rf /") == ["rm"]
err = capsys.readouterr().err
assert "ignoring invalid UV_EXTRAS entry" in err
assert "';'" in err # confirms the dangerous token was reported and dropped
def test_parse_env_extras_rejects_leading_digits_and_punctuation():
"""Names must start with a letter — pyproject extras follow this shape."""
assert detect.parse_env_extras("1postgres") == []
assert detect.parse_env_extras("-postgres") == []
# Hyphens and underscores inside the name are fine.
assert detect.parse_env_extras("post_gres") == ["post_gres"]
assert detect.parse_env_extras("post-gres") == ["post-gres"]
def test_format_flags_emits_one_flag_per_extra():
assert detect.format_flags([]) == ""
assert detect.format_flags(["postgres"]) == "--extra postgres"
assert detect.format_flags(["postgres", "ollama"]) == "--extra postgres --extra ollama"
def test_strip_comment_preserves_quoted_hash():
assert detect._strip_comment("backend: postgres # trailing") == "backend: postgres"
assert detect._strip_comment('name: "value#with-hash"') == 'name: "value#with-hash"'
assert detect._strip_comment("# whole line comment") == ""
def test_section_value_finds_nested_key():
yaml_lines = [
"database:",
" backend: postgres",
" postgres_url: $DATABASE_URL",
"",
"checkpointer:",
" type: sqlite",
]
assert detect.section_value(yaml_lines, "database", "backend") == "postgres"
assert detect.section_value(yaml_lines, "checkpointer", "type") == "sqlite"
assert detect.section_value(yaml_lines, "database", "missing") is None
assert detect.section_value(yaml_lines, "absent_section", "anything") is None
def test_section_value_ignores_commented_lines():
yaml_lines = [
"# database:",
"# backend: postgres",
"database:",
" backend: sqlite",
]
assert detect.section_value(yaml_lines, "database", "backend") == "sqlite"
def test_section_value_strips_quotes():
yaml_lines = [
"database:",
' backend: "postgres"',
]
assert detect.section_value(yaml_lines, "database", "backend") == "postgres"
def test_section_value_does_not_descend_into_grandchildren():
yaml_lines = [
"database:",
" backend: sqlite",
" nested:",
" backend: postgres",
]
# Only the immediate child level counts — keeps the parser predictable.
assert detect.section_value(yaml_lines, "database", "backend") == "sqlite"
def test_detect_from_config_postgres_via_database(tmp_path):
cfg = tmp_path / "config.yaml"
cfg.write_text("database:\n backend: postgres\n postgres_url: $DATABASE_URL\n")
assert detect.detect_from_config(cfg) == ["postgres"]
def test_detect_from_config_postgres_via_checkpointer(tmp_path):
cfg = tmp_path / "config.yaml"
cfg.write_text("checkpointer:\n type: postgres\n connection_string: postgresql://localhost/db\n")
assert detect.detect_from_config(cfg) == ["postgres"]
def test_detect_from_config_sqlite_returns_no_extras(tmp_path):
cfg = tmp_path / "config.yaml"
cfg.write_text("database:\n backend: sqlite\n sqlite_dir: .deer-flow/data\n")
assert detect.detect_from_config(cfg) == []
def test_detect_from_config_dedupes_when_both_present(tmp_path):
cfg = tmp_path / "config.yaml"
cfg.write_text("checkpointer:\n type: postgres\ndatabase:\n backend: postgres\n")
# Sorted unique extras, no double-counting.
assert detect.detect_from_config(cfg) == ["postgres"]
def test_detect_from_config_missing_file_returns_empty(tmp_path):
assert detect.detect_from_config(tmp_path / "does-not-exist.yaml") == []
def test_resolve_extras_env_overrides_config(isolated_cwd, monkeypatch):
cfg = isolated_cwd / "config.yaml"
cfg.write_text("database:\n backend: sqlite\n")
monkeypatch.setenv("UV_EXTRAS", "postgres")
assert detect.resolve_extras() == ["postgres"]
def test_resolve_extras_env_supports_multiple(isolated_cwd, monkeypatch):
monkeypatch.setenv("UV_EXTRAS", "postgres,ollama")
assert detect.resolve_extras() == ["postgres", "ollama"]
def test_resolve_extras_falls_back_to_config(isolated_cwd):
(isolated_cwd / "config.yaml").write_text("database:\n backend: postgres\n")
assert detect.resolve_extras() == ["postgres"]
def test_resolve_extras_respects_explicit_config_path(tmp_path, monkeypatch):
monkeypatch.delenv("UV_EXTRAS", raising=False)
elsewhere = tmp_path / "elsewhere.yaml"
elsewhere.write_text("database:\n backend: postgres\n")
monkeypatch.chdir(tmp_path)
monkeypatch.setenv("DEER_FLOW_CONFIG_PATH", str(elsewhere))
assert detect.resolve_extras() == ["postgres"]
def test_resolve_extras_no_config_no_env(isolated_cwd):
assert detect.resolve_extras() == []
def test_resolve_extras_finds_backend_subdir_config(isolated_cwd):
sub = isolated_cwd / "backend"
sub.mkdir()
(sub / "config.yaml").write_text("database:\n backend: postgres\n")
assert detect.resolve_extras() == ["postgres"]
def test_resolve_extras_root_config_takes_precedence(isolated_cwd):
(isolated_cwd / "config.yaml").write_text("database:\n backend: sqlite\n")
sub = isolated_cwd / "backend"
sub.mkdir()
(sub / "config.yaml").write_text("database:\n backend: postgres\n")
# Root config.yaml is checked first, matching the precedence in serve.sh.
assert detect.resolve_extras() == []

View File

@ -0,0 +1,102 @@
"""Unit tests for docker/dev-entrypoint.sh (UV_EXTRAS validation + parsing).
Exercises the script via its `--print-extras` dry-run hook so we don't actually
launch uvicorn or hit /app/logs. Together with test_detect_uv_extras.py these
cover both the local make-dev path and the docker-compose-dev path with the
same shape see PR #2767 / Issue #2754.
"""
from __future__ import annotations
import os
import subprocess
from pathlib import Path
import pytest
REPO_ROOT = Path(__file__).resolve().parents[2]
ENTRYPOINT = REPO_ROOT / "docker" / "dev-entrypoint.sh"
def _run(uv_extras: str | None) -> subprocess.CompletedProcess[str]:
"""Invoke `dev-entrypoint.sh --print-extras` with UV_EXTRAS set."""
env = os.environ.copy()
env.pop("UV_EXTRAS", None)
if uv_extras is not None:
env["UV_EXTRAS"] = uv_extras
return subprocess.run(
["sh", str(ENTRYPOINT), "--print-extras"],
env=env,
capture_output=True,
text=True,
check=False,
)
def test_entrypoint_script_exists_and_is_posix_sh():
assert ENTRYPOINT.is_file()
# Catch syntax errors before runtime — `sh -n` is a parse-only check.
proc = subprocess.run(["sh", "-n", str(ENTRYPOINT)], capture_output=True, text=True, check=False)
assert proc.returncode == 0, proc.stderr
def test_no_uv_extras_yields_empty_flags():
proc = _run(None)
assert proc.returncode == 0
assert proc.stdout.strip() == ""
def test_single_extra():
proc = _run("postgres")
assert proc.returncode == 0
assert proc.stdout.strip() == "--extra postgres"
def test_multi_extra_comma_separated():
proc = _run("postgres,ollama")
assert proc.returncode == 0
assert proc.stdout.strip() == "--extra postgres --extra ollama"
def test_multi_extra_whitespace_separated():
proc = _run("postgres ollama")
assert proc.returncode == 0
assert proc.stdout.strip() == "--extra postgres --extra ollama"
def test_multi_extra_mixed_separators():
proc = _run(" postgres , ollama ,")
assert proc.returncode == 0
assert proc.stdout.strip() == "--extra postgres --extra ollama"
def test_empty_string_yields_empty_flags():
proc = _run("")
assert proc.returncode == 0
assert proc.stdout.strip() == ""
@pytest.mark.parametrize(
"bad_value",
[
"; rm -rf /", # the canonical injection attempt
"$(whoami)", # command substitution
"`echo bad`", # backticks
"postgres;evil", # mixed legal+illegal in a single token
"1postgres", # leading digit
"-postgres", # leading hyphen
"post gres extra/path", # contains slash
],
)
def test_metacharacters_abort_with_nonzero_exit(bad_value):
proc = _run(bad_value)
assert proc.returncode != 0, f"expected abort for {bad_value!r}, got 0"
assert "is invalid" in proc.stderr
assert proc.stdout.strip() == ""
def test_underscores_and_hyphens_in_name_are_allowed():
"""Mirrors uv's accepted shape for `[project.optional-dependencies]` keys."""
proc = _run("post_gres,post-gres")
assert proc.returncode == 0
assert proc.stdout.strip() == "--extra post_gres --extra post-gres"

View File

@ -871,9 +871,25 @@ skill_evolution:
#
# Postgres mode: put your connection URL in .env as DATABASE_URL,
# then reference it here with $DATABASE_URL.
# Install the driver first:
# Local: uv sync --extra postgres
# Docker: UV_EXTRAS=postgres docker compose build
#
# Install the driver — Issue #2754 fix lands `UV_EXTRAS` in every code path:
# Local `make dev` auto-detects from `database.backend: postgres` below
# and passes `--extra postgres` to `uv sync` on every restart, so
# the extra is no longer wiped. To opt in explicitly (or layer
# extras like `postgres,ollama`), set in project-root .env:
# UV_EXTRAS=postgres
# Docker dev `make docker-start` reads `UV_EXTRAS` from project-root .env via
# `env_file`. Set:
# UV_EXTRAS=postgres
# Multiple extras (`postgres,ollama`) supported here too — see
# docker/dev-entrypoint.sh.
# Docker img build-arg `UV_EXTRAS=postgres docker compose build` — single
# extra only at build time (backend/Dockerfile passes the value
# as one token to `--extra`).
#
# First-time bootstrap (before `make dev`):
# cd backend && uv sync --all-packages --extra postgres
# (--all-packages propagates the extra into workspace members — see PR #2584)
#
# NOTE: When both `checkpointer` and `database` are configured,
# `checkpointer` takes precedence for LangGraph state persistence.

85
docker/dev-entrypoint.sh Executable file
View File

@ -0,0 +1,85 @@
#!/usr/bin/env sh
#
# DeerFlow gateway dev entrypoint — runs inside the docker-compose-dev gateway
# container. Extracted from docker/docker-compose-dev.yaml's inline `command:`
# (PR #2767, addressing review on Issue #2754).
#
# Responsibilities:
# 1. Resolve `--extra X` flags from UV_EXTRAS (comma- or whitespace-separated,
# mirroring scripts/detect_uv_extras.py for parity with local `make dev`).
# 2. Validate each extra against [A-Za-z][A-Za-z0-9_-]* so a stray shell
# metacharacter in `.env` cannot reach `uv sync`.
# 3. `uv sync --all-packages` so workspace member extras (deerflow-harness's
# postgres extra in particular) are installed — see PR #2584.
# 4. Self-heal: if the first sync fails, recreate .venv and retry once.
# 5. Hand off to uvicorn with reload, replacing this shell so uvicorn becomes
# PID 1 inside the container.
#
# Anchored at /bin/sh (not bash) since alpine-based base images may not ship
# bash. Uses POSIX-only constructs throughout.
set -e
# `--print-extras` is a dry-run hook: parse + validate UV_EXTRAS, print the
# resulting `--extra X` flags to stdout, and exit. Used by the unit test in
# backend/tests/test_dev_entrypoint.py and useful for ad-hoc debugging.
PRINT_EXTRAS_ONLY=0
if [ "${1:-}" = "--print-extras" ]; then
PRINT_EXTRAS_ONLY=1
fi
# Mirror the legacy command's behavior: redirect both stdout and stderr to the
# host-mounted log file (../logs/gateway.log → /app/logs/gateway.log). Skip
# the redirect under --print-extras so the test runner can capture stdout.
if [ "$PRINT_EXTRAS_ONLY" = "0" ]; then
exec >/app/logs/gateway.log 2>&1
fi
# ── Resolve extras ──────────────────────────────────────────────────────────
EXTRAS_FLAGS=""
if [ -n "${UV_EXTRAS:-}" ]; then
# Normalize comma → space, then split on whitespace via the unquoted `for`.
for raw in $(printf '%s' "$UV_EXTRAS" | tr ',' ' '); do
[ -z "$raw" ] && continue
# Reject anything that does not look like an identifier.
# Two patterns: leading non-letter, or any non-[A-Za-z0-9_-] character.
case "$raw" in
[!A-Za-z]* | *[!A-Za-z0-9_-]*)
echo "[startup] UV_EXTRAS entry '$raw' is invalid (must match [A-Za-z][A-Za-z0-9_-]*) — aborting" >&2
exit 1
;;
esac
EXTRAS_FLAGS="$EXTRAS_FLAGS --extra $raw"
done
fi
if [ "$PRINT_EXTRAS_ONLY" = "1" ]; then
# Trim leading space for tidier output, then exit.
printf '%s\n' "${EXTRAS_FLAGS# }"
exit 0
fi
if [ -n "$EXTRAS_FLAGS" ]; then
echo "[startup] uv extras:$EXTRAS_FLAGS"
fi
# ── Sync dependencies (with self-heal) ──────────────────────────────────────
cd /app/backend
# `--all-packages` propagates extras into workspace members (PR #2584).
# `$EXTRAS_FLAGS` intentionally unquoted so each `--extra X` becomes its own arg.
# shellcheck disable=SC2086 # word-splitting is intentional here
if ! uv sync --all-packages $EXTRAS_FLAGS; then
echo "[startup] uv sync failed; recreating .venv and retrying once"
uv venv --allow-existing .venv
# shellcheck disable=SC2086
uv sync --all-packages $EXTRAS_FLAGS
fi
# ── Hand off to uvicorn ─────────────────────────────────────────────────────
PYTHONPATH=. exec uv run uvicorn app.gateway.app:app \
--host 0.0.0.0 --port 8001 \
--reload --reload-include='*.yaml .env'

View File

@ -125,8 +125,15 @@ services:
UV_IMAGE: ${UV_IMAGE:-ghcr.io/astral-sh/uv:0.7.20}
UV_INDEX_URL: ${UV_INDEX_URL:-https://pypi.org/simple}
container_name: deer-flow-gateway
command: sh -c "{ cd backend && (uv sync || (echo '[startup] uv sync failed; recreating .venv and retrying once' && uv venv --allow-existing .venv && uv sync)) && PYTHONPATH=. uv run uvicorn app.gateway.app:app --host 0.0.0.0 --port 8001 --reload --reload-include='*.yaml .env'; } > /app/logs/gateway.log 2>&1"
# Startup logic lives in docker/dev-entrypoint.sh — UV_EXTRAS validation,
# `uv sync --all-packages`, .venv self-heal, and uvicorn handoff. Keeps
# this file readable and lets the script be linted (shellcheck-clean).
# See PR #2767 / Issue #2754.
command: ["sh", "/usr/local/bin/dev-entrypoint.sh"]
volumes:
# Mount the dev entrypoint as a read-only file so edits to the script
# take effect on `make docker-restart` without requiring an image rebuild.
- ./dev-entrypoint.sh:/usr/local/bin/dev-entrypoint.sh:ro
- ../backend/:/app/backend/
# Preserve the .venv built during Docker image build — mounting the full backend/
# directory above would otherwise shadow it with the (empty) host directory.

180
scripts/detect_uv_extras.py Executable file
View File

@ -0,0 +1,180 @@
#!/usr/bin/env python3
"""Resolve uv extras for local `uv sync` based on environment + config.yaml.
Order of resolution:
1. `UV_EXTRAS` env var. Comma- or whitespace-separated names so multiple
extras can be layered (e.g. ``UV_EXTRAS=postgres,ollama``). The same
parsing semantics apply in the Docker dev container via
``docker/dev-entrypoint.sh``. The Docker image-build path
(``backend/Dockerfile``) still treats `UV_EXTRAS` as a single token, so
``UV_EXTRAS=postgres,ollama`` would only install ``postgres,ollama`` as
one (invalid) extra at build time author build-time values as a
single name.
2. Auto-detection from config.yaml currently maps:
- database.backend == postgres -> postgres
- checkpointer.type == postgres -> postgres
Each extra name is validated against ``^[A-Za-z][A-Za-z0-9_-]*$`` (the same
shape uv enforces for `[project.optional-dependencies]` keys). Anything else
is dropped with a stderr warning so a stray shell metacharacter in `.env`
cannot reach the `uv sync` invocation downstream.
Output: space-separated `--extra <name>` flags ready for splat into
`uv sync`, e.g. `--extra postgres`. Empty output means "no extras".
Intentionally implemented with the standard library only: this script must run
*before* `uv sync` has populated the venv, so it cannot depend on PyYAML.
"""
from __future__ import annotations
import os
import re
import sys
from pathlib import Path
# Mirrors uv's accepted shape for extra names — keeps the eventual
# `uv sync --extra <name>` invocation free of shell metacharacters even when
# `UV_EXTRAS` comes from `.env` or another semi-trusted source.
_EXTRA_NAME_RE = re.compile(r"^[A-Za-z][A-Za-z0-9_-]*$")
def _validate_extras(names: list[str]) -> list[str]:
valid: list[str] = []
for name in names:
if _EXTRA_NAME_RE.match(name):
valid.append(name)
else:
print(
f"detect_uv_extras: ignoring invalid UV_EXTRAS entry {name!r} (must match [A-Za-z][A-Za-z0-9_-]*)",
file=sys.stderr,
)
return valid
def parse_env_extras(value: str) -> list[str]:
"""Split UV_EXTRAS into a list, accepting comma or whitespace separators."""
parts = re.split(r"[\s,]+", value.strip())
return _validate_extras([p for p in parts if p])
def find_config_file() -> Path | None:
"""Locate config.yaml using the same precedence as serve.sh."""
explicit = os.environ.get("DEER_FLOW_CONFIG_PATH")
if explicit:
candidate = Path(explicit)
if candidate.is_file():
return candidate
for path in (Path("config.yaml"), Path("backend/config.yaml")):
if path.is_file():
return path
return None
_SECTION_RE = re.compile(r"^([A-Za-z_][\w-]*)\s*:\s*$")
_KEY_RE = re.compile(r"^\s+([A-Za-z_][\w-]*)\s*:\s*(\S.*?)\s*$")
def _strip_comment(line: str) -> str:
"""Drop trailing `#` comments while preserving `#` inside quoted strings."""
in_quote: str | None = None
out: list[str] = []
for ch in line:
if in_quote is not None:
out.append(ch)
if ch == in_quote:
in_quote = None
continue
if ch in ("'", '"'):
in_quote = ch
out.append(ch)
elif ch == "#":
break
else:
out.append(ch)
return "".join(out).rstrip()
def _unquote(value: str) -> str:
if len(value) >= 2 and value[0] == value[-1] and value[0] in ("'", '"'):
return value[1:-1]
return value
def section_value(lines: list[str], section: str, key: str) -> str | None:
"""Return the value of `section.key` from a flat-ish YAML, or None.
Only handles the shallow shape DeerFlow uses for these settings:
database:
backend: postgres
Nested mappings deeper than the immediate child level are ignored on
purpose that keeps this parser predictable without a full YAML stack.
"""
inside = False
child_indent: int | None = None
for raw in lines:
line = _strip_comment(raw)
if not line.strip():
continue
sect_match = _SECTION_RE.match(line)
if sect_match:
inside = sect_match.group(1) == section
child_indent = None
continue
if not inside:
continue
stripped = line.lstrip()
indent = len(line) - len(stripped)
if indent == 0:
inside = False
continue
if child_indent is None:
child_indent = indent
if indent < child_indent:
inside = False
continue
if indent != child_indent:
continue
key_match = _KEY_RE.match(line)
if key_match and key_match.group(1) == key:
return _unquote(key_match.group(2).strip())
return None
def detect_from_config(path: Path) -> list[str]:
try:
text = path.read_text(encoding="utf-8", errors="replace")
except OSError:
return []
lines = text.splitlines()
extras: set[str] = set()
if (section_value(lines, "database", "backend") or "").lower() == "postgres":
extras.add("postgres")
if (section_value(lines, "checkpointer", "type") or "").lower() == "postgres":
extras.add("postgres")
return sorted(extras)
def resolve_extras() -> list[str]:
env = os.environ.get("UV_EXTRAS", "")
if env.strip():
return parse_env_extras(env)
config = find_config_file()
if config is None:
return []
return detect_from_config(config)
def format_flags(extras: list[str]) -> str:
return " ".join(f"--extra {e}" for e in extras)
def main() -> int:
extras = resolve_extras()
if extras:
sys.stdout.write(format_flags(extras))
return 0
if __name__ == "__main__":
sys.exit(main())

View File

@ -157,9 +157,41 @@ fi
# ── Install dependencies ────────────────────────────────────────────────────
# Pick a Python for the extras detector. Falls back to plain `python` for
# Windows/Git Bash where only `python` is on PATH.
if command -v python3 >/dev/null 2>&1; then
DETECT_PYTHON="python3"
elif command -v python >/dev/null 2>&1; then
DETECT_PYTHON="python"
else
DETECT_PYTHON=""
fi
# Resolve uv extras (postgres, etc.) from UV_EXTRAS or config.yaml so that
# `uv sync` does not wipe out optional dependencies on every restart. See
# scripts/detect_uv_extras.py and Issue #2754 for context. The detector
# whitelists extra names against `^[A-Za-z][A-Za-z0-9_-]*$`, so the unquoted
# splat below only sees valid uv argument tokens.
#
# Stderr is intentionally NOT redirected so the user sees:
# - whitelist warnings (e.g. "ignoring invalid UV_EXTRAS entry ';'");
# - detector crashes (e.g. unexpected Python error).
# `|| true` keeps `set -e` from killing dev startup on a detector failure;
# the result is just an empty UV_EXTRAS_FLAGS, which means "no extras".
UV_EXTRAS_FLAGS=""
if [ -n "$DETECT_PYTHON" ]; then
UV_EXTRAS_FLAGS=$("$DETECT_PYTHON" "$REPO_ROOT/scripts/detect_uv_extras.py" || { echo "[serve.sh] detect_uv_extras.py failed (exit $?) — proceeding without extras" >&2; echo ""; })
fi
if ! $SKIP_INSTALL; then
echo "Syncing dependencies..."
(cd backend && uv sync --quiet) || { echo "✗ Backend dependency install failed"; exit 1; }
if [ -n "$UV_EXTRAS_FLAGS" ]; then
echo " • uv extras: $UV_EXTRAS_FLAGS"
fi
# `--all-packages` propagates extras into workspace members (deerflow-harness
# in particular). Required for postgres extras — see PR #2584.
# Intentionally unquoted to splat multiple `--extra X` pairs.
(cd backend && uv sync --quiet --all-packages $UV_EXTRAS_FLAGS) || { echo "✗ Backend dependency install failed"; exit 1; }
(cd frontend && pnpm install --silent) || { echo "✗ Frontend dependency install failed"; exit 1; }
echo "✓ Dependencies synced"
else