kevin/life-echo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(dev): idempotent Alembic chain for squashed 0001 + clearer dev scripts

Browse Source 
- Make migrations 0002–0008 skip schema changes already applied when
  0001_initial creates current ORM (rename segments column, timeline FK,
  memoir phase flags, drop content_tsv, eval_* tables).
- development.sh / internal-eval.sh: surface Alembic stderr, warn on
  docker-style DB hosts, TCP port checks without lsof, verify Uvicorn
  listens before claiming started.
This commit is contained in:
Kevin
2026-04-07 10:34:09 +08:00
parent 29dec8fe32
commit ea97427767
7 changed files with 661 additions and 274 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
api/alembic/versions/0002_segments_user_input_text.py
View File

@@ -8,6 +8,7 @@ Revises: 0001_initial
from typing import Sequence, Union from typing import Sequence, Union
import sqlalchemy as sa
from alembic import op from alembic import op
revision: str = "0002_segments_user_input_text" revision: str = "0002_segments_user_input_text"
@@ -16,9 +17,19 @@ branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None depends_on: Union[str, Sequence[str], None] = None
def _column_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {column["name"] for column in inspector.get_columns(table_name)}
def upgrade() -> None: def upgrade() -> None:
op.execute("ALTER TABLE segments RENAME COLUMN transcript_text TO user_input_text") columns = _column_names("segments")
if "transcript_text" in columns and "user_input_text" not in columns:
op.execute("ALTER TABLE segments RENAME COLUMN transcript_text TO user_input_text")
def downgrade() -> None: def downgrade() -> None:
op.execute("ALTER TABLE segments RENAME COLUMN user_input_text TO transcript_text") columns = _column_names("segments")
if "user_input_text" in columns and "transcript_text" not in columns:
op.execute("ALTER TABLE segments RENAME COLUMN user_input_text TO transcript_text")

84
api/alembic/versions/0003_timeline_events_memory_source_id.py
View File

@@ -15,31 +15,67 @@ branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None depends_on: Union[str, Sequence[str], None] = None
def _column_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {column["name"] for column in inspector.get_columns(table_name)}
def _index_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {index["name"] for index in inspector.get_indexes(table_name)}
def _foreign_key_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {
fk["name"] for fk in inspector.get_foreign_keys(table_name) if fk.get("name")
}
def upgrade() -> None: def upgrade() -> None:
op.add_column( columns = _column_names("timeline_events")
"timeline_events", if "memory_source_id" not in columns:
sa.Column("memory_source_id", sa.String(), nullable=True), op.add_column(
) "timeline_events",
op.create_index( sa.Column("memory_source_id", sa.String(), nullable=True),
"ix_timeline_events_memory_source_id", )
"timeline_events",
["memory_source_id"], indexes = _index_names("timeline_events")
) if "ix_timeline_events_memory_source_id" not in indexes:
op.create_foreign_key( op.create_index(
"fk_timeline_events_memory_source_id_memory_sources", "ix_timeline_events_memory_source_id",
"timeline_events", "timeline_events",
"memory_sources", ["memory_source_id"],
["memory_source_id"], )
["id"],
ondelete="SET NULL", foreign_keys = _foreign_key_names("timeline_events")
) if "fk_timeline_events_memory_source_id_memory_sources" not in foreign_keys:
op.create_foreign_key(
"fk_timeline_events_memory_source_id_memory_sources",
"timeline_events",
"memory_sources",
["memory_source_id"],
["id"],
ondelete="SET NULL",
)
def downgrade() -> None: def downgrade() -> None:
op.drop_constraint( foreign_keys = _foreign_key_names("timeline_events")
"fk_timeline_events_memory_source_id_memory_sources", if "fk_timeline_events_memory_source_id_memory_sources" in foreign_keys:
"timeline_events", op.drop_constraint(
type_="foreignkey", "fk_timeline_events_memory_source_id_memory_sources",
) "timeline_events",
op.drop_index("ix_timeline_events_memory_source_id", table_name="timeline_events") type_="foreignkey",
op.drop_column("timeline_events", "memory_source_id") )
indexes = _index_names("timeline_events")
if "ix_timeline_events_memory_source_id" in indexes:
op.drop_index("ix_timeline_events_memory_source_id", table_name="timeline_events")
columns = _column_names("timeline_events")
if "memory_source_id" in columns:
op.drop_column("timeline_events", "memory_source_id")

52
api/alembic/versions/0006_segment_memoir_phase_flags.py
View File

@@ -16,27 +16,39 @@ branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None depends_on: Union[str, Sequence[str], None] = None
def _column_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {column["name"] for column in inspector.get_columns(table_name)}
def upgrade() -> None: def upgrade() -> None:
op.add_column( columns = _column_names("segments")
"segments", if "narrated" not in columns:
sa.Column( op.add_column(
"narrated", "segments",
sa.Boolean(), sa.Column(
nullable=False, "narrated",
server_default=sa.text("false"), sa.Boolean(),
), nullable=False,
) server_default=sa.text("false"),
op.add_column( ),
"segments", )
sa.Column( if "skip_narrative" not in columns:
"skip_narrative", op.add_column(
sa.Boolean(), "segments",
nullable=False, sa.Column(
server_default=sa.text("false"), "skip_narrative",
), sa.Boolean(),
) nullable=False,
server_default=sa.text("false"),
),
)
def downgrade() -> None: def downgrade() -> None:
op.drop_column("segments", "skip_narrative") columns = _column_names("segments")
op.drop_column("segments", "narrated") if "skip_narrative" in columns:
op.drop_column("segments", "skip_narrative")
if "narrated" in columns:
op.drop_column("segments", "narrated")

20
api/alembic/versions/0007_drop_chunk_content_tsv.py
View File

@@ -17,12 +17,22 @@ branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None depends_on: Union[str, Sequence[str], None] = None
def _column_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {column["name"] for column in inspector.get_columns(table_name)}
def upgrade() -> None: def upgrade() -> None:
op.drop_column("memory_chunks", "content_tsv") columns = _column_names("memory_chunks")
if "content_tsv" in columns:
op.drop_column("memory_chunks", "content_tsv")
def downgrade() -> None: def downgrade() -> None:
op.add_column( columns = _column_names("memory_chunks")
"memory_chunks", if "content_tsv" not in columns:
sa.Column("content_tsv", postgresql.TSVECTOR(), nullable=True), op.add_column(
) "memory_chunks",
sa.Column("content_tsv", postgresql.TSVECTOR(), nullable=True),
)

446
api/alembic/versions/0008_eval_regression_platform.py
View File

@@ -17,214 +17,246 @@ branch_labels: Union[str, Sequence[str], None] = None
depends_on: Union[str, Sequence[str], None] = None depends_on: Union[str, Sequence[str], None] = None
def _has_table(table_name: str) -> bool:
bind = op.get_bind()
inspector = sa.inspect(bind)
return inspector.has_table(table_name)
def _index_names(table_name: str) -> set[str]:
bind = op.get_bind()
inspector = sa.inspect(bind)
return {index["name"] for index in inspector.get_indexes(table_name)}
def upgrade() -> None: def upgrade() -> None:
op.create_table( if not _has_table("eval_regression_sets"):
"eval_regression_sets", op.create_table(
sa.Column("id", sa.String(), nullable=False), "eval_regression_sets",
sa.Column("name", sa.String(), nullable=False), sa.Column("id", sa.String(), nullable=False),
sa.Column("description", sa.Text(), nullable=True), sa.Column("name", sa.String(), nullable=False),
sa.Column( sa.Column("description", sa.Text(), nullable=True),
"created_at", sa.Column(
sa.DateTime(timezone=True), "created_at",
server_default=sa.text("now()"), sa.DateTime(timezone=True),
nullable=False, server_default=sa.text("now()"),
), nullable=False,
sa.PrimaryKeyConstraint("id"), ),
) sa.PrimaryKeyConstraint("id"),
op.create_table( )
"eval_versions", if not _has_table("eval_versions"):
sa.Column("id", sa.String(), nullable=False), op.create_table(
sa.Column("name", sa.String(), nullable=False), "eval_versions",
sa.Column("runner_kind", sa.String(), nullable=False), sa.Column("id", sa.String(), nullable=False),
sa.Column( sa.Column("name", sa.String(), nullable=False),
"config_json", postgresql.JSONB(astext_type=sa.Text()), nullable=True sa.Column("runner_kind", sa.String(), nullable=False),
), sa.Column(
sa.Column( "config_json", postgresql.JSONB(astext_type=sa.Text()), nullable=True
"created_at", ),
sa.DateTime(timezone=True), sa.Column(
server_default=sa.text("now()"), "created_at",
nullable=False, sa.DateTime(timezone=True),
), server_default=sa.text("now()"),
sa.PrimaryKeyConstraint("id"), nullable=False,
) ),
op.create_table( sa.PrimaryKeyConstraint("id"),
"eval_cases", )
sa.Column("id", sa.String(), nullable=False), if not _has_table("eval_cases"):
sa.Column("regression_set_id", sa.String(), nullable=False), op.create_table(
sa.Column("source_conversation_id", sa.String(), nullable=True), "eval_cases",
sa.Column("source_user_id", sa.String(), nullable=True), sa.Column("id", sa.String(), nullable=False),
sa.Column("title", sa.String(), nullable=True), sa.Column("regression_set_id", sa.String(), nullable=False),
sa.Column( sa.Column("source_conversation_id", sa.String(), nullable=True),
"user_utterances", sa.Column("source_user_id", sa.String(), nullable=True),
postgresql.JSONB(astext_type=sa.Text()), sa.Column("title", sa.String(), nullable=True),
nullable=False, sa.Column(
), "user_utterances",
sa.Column("reference_memoir_markdown", sa.Text(), nullable=True), postgresql.JSONB(astext_type=sa.Text()),
sa.Column( nullable=False,
"is_protected", ),
sa.Boolean(), sa.Column("reference_memoir_markdown", sa.Text(), nullable=True),
server_default=sa.text("false"), sa.Column(
nullable=False, "is_protected",
), sa.Boolean(),
sa.Column("meta", postgresql.JSONB(astext_type=sa.Text()), nullable=True), server_default=sa.text("false"),
sa.Column( nullable=False,
"created_at", ),
sa.DateTime(timezone=True), sa.Column("meta", postgresql.JSONB(astext_type=sa.Text()), nullable=True),
server_default=sa.text("now()"), sa.Column(
nullable=False, "created_at",
), sa.DateTime(timezone=True),
sa.ForeignKeyConstraint( server_default=sa.text("now()"),
["regression_set_id"], nullable=False,
["eval_regression_sets.id"], ),
ondelete="CASCADE", sa.ForeignKeyConstraint(
), ["regression_set_id"],
sa.PrimaryKeyConstraint("id"), ["eval_regression_sets.id"],
) ondelete="CASCADE",
op.create_index( ),
"ix_eval_cases_source_conversation_id", sa.PrimaryKeyConstraint("id"),
"eval_cases", )
["source_conversation_id"], indexes = _index_names("eval_cases")
unique=False, if "ix_eval_cases_source_conversation_id" not in indexes:
) op.create_index(
op.create_index( "ix_eval_cases_source_conversation_id",
"ix_eval_cases_source_user_id", "eval_cases",
"eval_cases", ["source_conversation_id"],
["source_user_id"], unique=False,
unique=False, )
) if "ix_eval_cases_source_user_id" not in indexes:
op.create_table( op.create_index(
"eval_experiments", "ix_eval_cases_source_user_id",
sa.Column("id", sa.String(), nullable=False), "eval_cases",
sa.Column("name", sa.String(), nullable=False), ["source_user_id"],
sa.Column("regression_set_id", sa.String(), nullable=False), unique=False,
sa.Column("baseline_version_id", sa.String(), nullable=False), )
sa.Column("candidate_version_id", sa.String(), nullable=False), if not _has_table("eval_experiments"):
sa.Column("rubric_pack", sa.String(), nullable=False), op.create_table(
sa.Column( "eval_experiments",
"composite_weights_json", sa.Column("id", sa.String(), nullable=False),
postgresql.JSONB(astext_type=sa.Text()), sa.Column("name", sa.String(), nullable=False),
nullable=True, sa.Column("regression_set_id", sa.String(), nullable=False),
), sa.Column("baseline_version_id", sa.String(), nullable=False),
sa.Column("status", sa.String(), nullable=False), sa.Column("candidate_version_id", sa.String(), nullable=False),
sa.Column("error_message", sa.Text(), nullable=True), sa.Column("rubric_pack", sa.String(), nullable=False),
sa.Column( sa.Column(
"created_at", "composite_weights_json",
sa.DateTime(timezone=True), postgresql.JSONB(astext_type=sa.Text()),
server_default=sa.text("now()"), nullable=True,
nullable=False, ),
), sa.Column("status", sa.String(), nullable=False),
sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True), sa.Column("error_message", sa.Text(), nullable=True),
sa.ForeignKeyConstraint( sa.Column(
["baseline_version_id"], "created_at",
["eval_versions.id"], sa.DateTime(timezone=True),
), server_default=sa.text("now()"),
sa.ForeignKeyConstraint( nullable=False,
["candidate_version_id"], ),
["eval_versions.id"], sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True),
), sa.ForeignKeyConstraint(
sa.ForeignKeyConstraint( ["baseline_version_id"],
["regression_set_id"], ["eval_versions.id"],
["eval_regression_sets.id"], ),
), sa.ForeignKeyConstraint(
sa.PrimaryKeyConstraint("id"), ["candidate_version_id"],
) ["eval_versions.id"],
op.create_table( ),
"eval_runs", sa.ForeignKeyConstraint(
sa.Column("id", sa.String(), nullable=False), ["regression_set_id"],
sa.Column("experiment_id", sa.String(), nullable=False), ["eval_regression_sets.id"],
sa.Column("case_id", sa.String(), nullable=False), ),
sa.Column("side", sa.String(), nullable=False), sa.PrimaryKeyConstraint("id"),
sa.Column("status", sa.String(), nullable=False), )
sa.Column("error_message", sa.Text(), nullable=True), if not _has_table("eval_runs"):
sa.Column("memoir_markdown", sa.Text(), nullable=True), op.create_table(
sa.Column("conversation_score_total", sa.Float(), nullable=True), "eval_runs",
sa.Column("memoir_score_total", sa.Float(), nullable=True), sa.Column("id", sa.String(), nullable=False),
sa.Column("composite_score", sa.Float(), nullable=True), sa.Column("experiment_id", sa.String(), nullable=False),
sa.Column( sa.Column("case_id", sa.String(), nullable=False),
"judge_bundle_json", sa.Column("side", sa.String(), nullable=False),
postgresql.JSONB(astext_type=sa.Text()), sa.Column("status", sa.String(), nullable=False),
nullable=True, sa.Column("error_message", sa.Text(), nullable=True),
), sa.Column("memoir_markdown", sa.Text(), nullable=True),
sa.Column("started_at", sa.DateTime(timezone=True), nullable=True), sa.Column("conversation_score_total", sa.Float(), nullable=True),
sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True), sa.Column("memoir_score_total", sa.Float(), nullable=True),
sa.ForeignKeyConstraint( sa.Column("composite_score", sa.Float(), nullable=True),
["case_id"], sa.Column(
["eval_cases.id"], "judge_bundle_json",
), postgresql.JSONB(astext_type=sa.Text()),
sa.ForeignKeyConstraint( nullable=True,
["experiment_id"], ),
["eval_experiments.id"], sa.Column("started_at", sa.DateTime(timezone=True), nullable=True),
ondelete="CASCADE", sa.Column("completed_at", sa.DateTime(timezone=True), nullable=True),
), sa.ForeignKeyConstraint(
sa.PrimaryKeyConstraint("id"), ["case_id"],
sa.UniqueConstraint( ["eval_cases.id"],
"experiment_id", ),
"case_id", sa.ForeignKeyConstraint(
"side", ["experiment_id"],
name="uq_eval_run_experiment_case_side", ["eval_experiments.id"],
), ondelete="CASCADE",
) ),
op.create_table( sa.PrimaryKeyConstraint("id"),
"eval_run_turns", sa.UniqueConstraint(
sa.Column("id", sa.String(), nullable=False), "experiment_id",
sa.Column("run_id", sa.String(), nullable=False), "case_id",
sa.Column("turn_index", sa.Integer(), nullable=False), "side",
sa.Column("user_utterance", sa.Text(), nullable=False), name="uq_eval_run_experiment_case_side",
sa.Column("assistant_reply", sa.Text(), nullable=True), ),
sa.Column("duration_ms", sa.Integer(), nullable=True), )
sa.Column( if not _has_table("eval_run_turns"):
"judge_scores_json", op.create_table(
postgresql.JSONB(astext_type=sa.Text()), "eval_run_turns",
nullable=True, sa.Column("id", sa.String(), nullable=False),
), sa.Column("run_id", sa.String(), nullable=False),
sa.Column("judge_rationale", sa.Text(), nullable=True), sa.Column("turn_index", sa.Integer(), nullable=False),
sa.ForeignKeyConstraint( sa.Column("user_utterance", sa.Text(), nullable=False),
["run_id"], sa.Column("assistant_reply", sa.Text(), nullable=True),
["eval_runs.id"], sa.Column("duration_ms", sa.Integer(), nullable=True),
ondelete="CASCADE", sa.Column(
), "judge_scores_json",
sa.PrimaryKeyConstraint("id"), postgresql.JSONB(astext_type=sa.Text()),
sa.UniqueConstraint("run_id", "turn_index", name="uq_eval_run_turn_index"), nullable=True,
) ),
op.create_table( sa.Column("judge_rationale", sa.Text(), nullable=True),
"eval_gate_verdicts", sa.ForeignKeyConstraint(
sa.Column("id", sa.String(), nullable=False), ["run_id"],
sa.Column("experiment_id", sa.String(), nullable=False), ["eval_runs.id"],
sa.Column("passed", sa.Boolean(), nullable=False), ondelete="CASCADE",
sa.Column("mean_composite_delta", sa.Float(), nullable=True), ),
sa.Column( sa.PrimaryKeyConstraint("id"),
"protected_regressions_json", sa.UniqueConstraint("run_id", "turn_index", name="uq_eval_run_turn_index"),
postgresql.JSONB(astext_type=sa.Text()), )
nullable=True, if not _has_table("eval_gate_verdicts"):
), op.create_table(
sa.Column( "eval_gate_verdicts",
"details_json", sa.Column("id", sa.String(), nullable=False),
postgresql.JSONB(astext_type=sa.Text()), sa.Column("experiment_id", sa.String(), nullable=False),
nullable=True, sa.Column("passed", sa.Boolean(), nullable=False),
), sa.Column("mean_composite_delta", sa.Float(), nullable=True),
sa.Column( sa.Column(
"computed_at", "protected_regressions_json",
sa.DateTime(timezone=True), postgresql.JSONB(astext_type=sa.Text()),
server_default=sa.text("now()"), nullable=True,
nullable=False, ),
), sa.Column(
sa.ForeignKeyConstraint( "details_json",
["experiment_id"], postgresql.JSONB(astext_type=sa.Text()),
["eval_experiments.id"], nullable=True,
ondelete="CASCADE", ),
), sa.Column(
sa.PrimaryKeyConstraint("id"), "computed_at",
sa.UniqueConstraint("experiment_id"), sa.DateTime(timezone=True),
) server_default=sa.text("now()"),
nullable=False,
),
sa.ForeignKeyConstraint(
["experiment_id"],
["eval_experiments.id"],
ondelete="CASCADE",
),
sa.PrimaryKeyConstraint("id"),
sa.UniqueConstraint("experiment_id"),
)
def downgrade() -> None: def downgrade() -> None:
op.drop_table("eval_gate_verdicts") if _has_table("eval_gate_verdicts"):
op.drop_table("eval_run_turns") op.drop_table("eval_gate_verdicts")
op.drop_table("eval_runs") if _has_table("eval_run_turns"):
op.drop_table("eval_experiments") op.drop_table("eval_run_turns")
op.drop_index("ix_eval_cases_source_user_id", table_name="eval_cases") if _has_table("eval_runs"):
op.drop_index("ix_eval_cases_source_conversation_id", table_name="eval_cases") op.drop_table("eval_runs")
op.drop_table("eval_cases") if _has_table("eval_experiments"):
op.drop_table("eval_versions") op.drop_table("eval_experiments")
op.drop_table("eval_regression_sets") if _has_table("eval_cases"):
indexes = _index_names("eval_cases")
if "ix_eval_cases_source_user_id" in indexes:
op.drop_index("ix_eval_cases_source_user_id", table_name="eval_cases")
if "ix_eval_cases_source_conversation_id" in indexes:
op.drop_index("ix_eval_cases_source_conversation_id", table_name="eval_cases")
op.drop_table("eval_cases")
if _has_table("eval_versions"):
op.drop_table("eval_versions")
if _has_table("eval_regression_sets"):
op.drop_table("eval_regression_sets")

159
api/development.sh
View File

@@ -163,6 +163,122 @@ wait_postgres_ready() {
return 1 return 1
} }
get_effective_database_url() {
if [[ -n "${DATABASE_URL:-}" ]]; then
printf '%s\n' "${DATABASE_URL}"
return 0
fi
if [[ -f "${ROOT_DIR}/.env" ]]; then
local line
line="$(sed -n 's/^DATABASE_URL=//p' "${ROOT_DIR}/.env" | sed -n '1p')"
line="${line%\"}"
line="${line#\"}"
line="${line%\'}"
line="${line#\'}"
if [[ -n "${line}" ]]; then
printf '%s\n' "${line}"
return 0
fi
fi
return 1
}
warn_database_url_host_pitfall() {
local database_url
local host
if ! database_url="$(get_effective_database_url)"; then
return 0
fi
if [[ "${database_url}" =~ @([^:/?#]+) ]]; then
host="${BASH_REMATCH[1]}"
case "${host}" in
postgres|db|postgres-dev|postgresql)
print_warn "检测到 DATABASE_URL 主机为 ${host}；在宿主机执行 Alembic/uvicorn 时通常应使用 localhost"
;;
esac
fi
}
print_alembic_failure_hint() {
local log_file="$1"
local log_output
log_output="$(sed -n '1,200p' "${log_file}")"
if [[ "${log_output}" == *'could not translate host name "postgres"'* ]] || [[ "${log_output}" == *"Name or service not known"* ]]; then
print_warn "看起来 DATABASE_URL 指向了容器内主机名；在宿主机运行时请改用 localhost:5432"
elif [[ "${log_output}" == *"Connection refused"* ]] || [[ "${log_output}" == *"could not connect to server"* ]]; then
print_warn "PostgreSQL 连接被拒绝；请确认容器已启动且 DATABASE_URL 与 docker-compose.dev.yml 暴露端口一致"
elif [[ "${log_output}" == *"password authentication failed"* ]]; then
print_warn "PostgreSQL 用户名或密码不匹配；请核对 .env.development 中的 DATABASE_URL"
elif [[ "${log_output}" == *"No such file or directory"* ]] || [[ "${log_output}" == *"can't open file"* ]]; then
print_warn "Alembic 依赖的文件或工作目录可能不正确；请确认在 api/ 目录运行脚本"
fi
}
is_port_listening() {
local port="$1"
if command -v lsof >/dev/null 2>&1; then
lsof -nP -iTCP:"${port}" -sTCP:LISTEN >/dev/null 2>&1
return $?
fi
if [[ -x "${PYTHON_BIN}" ]]; then
"${PYTHON_BIN}" - "${port}" <<'PY' >/dev/null 2>&1
import socket
import sys
sock = socket.socket()
sock.settimeout(0.2)
try:
sock.connect(("127.0.0.1", int(sys.argv[1])))
except OSError:
raise SystemExit(1)
finally:
sock.close()
raise SystemExit(0)
PY
return $?
fi
return 1
}
wait_for_tcp_listener() {
local pid="$1"
local port="$2"
local timeout="${3:-8}"
local waited=0
while (( waited < timeout )); do
if is_port_listening "${port}"; then
return 0
fi
if ! is_pid_alive "${pid}"; then
return 1
fi
sleep 1
waited=$((waited + 1))
done
return 2
}
ensure_background_process_alive() {
local name="$1"
local pid="$2"
sleep 1
if ! is_pid_alive "${pid}"; then
print_err "${name} 启动后立即退出，请查看上方日志"
exit 1
fi
}
ensure_venv() { ensure_venv() {
print_header "检查 Python 虚拟环境" print_header "检查 Python 虚拟环境"
@@ -199,16 +315,25 @@ check_env_file() {
print_warn "未找到 .env，应用可能因缺少配置启动失败" print_warn "未找到 .env，应用可能因缺少配置启动失败"
else else
print_ok "检测到 .env" print_ok "检测到 .env"
warn_database_url_host_pitfall
fi fi
} }
run_migrations() { run_migrations() {
print_header "执行数据库迁移" print_header "执行数据库迁移"
cd "${ROOT_DIR}" cd "${ROOT_DIR}"
if uv run alembic upgrade head 2>/dev/null; then local log_file
log_file="$(mktemp -t life-echo-alembic.XXXXXX.log)"
if uv run alembic upgrade head >"${log_file}" 2>&1; then
print_ok "Alembic 迁移已就绪" print_ok "Alembic 迁移已就绪"
rm -f "${log_file}"
else else
print_warn "Alembic 迁移失败（可能数据库未启动或 DATABASE_URL 未配置），应用启动可能失败" print_warn "Alembic 迁移失败（可能数据库未启动或 DATABASE_URL 未配置），应用启动可能失败"
print_alembic_failure_hint "${log_file}"
print_warn "Alembic 输出（最近 40 行）："
tail -n 40 "${log_file}"
rm -f "${log_file}"
fi fi
} }
@@ -216,12 +341,10 @@ start_services() {
print_header "启动 FastAPI 和 Celery" print_header "启动 FastAPI 和 Celery"
cd "${ROOT_DIR}" cd "${ROOT_DIR}"
if command -v lsof >/dev/null 2>&1; then if is_port_listening "${API_PORT}"; then
if lsof -nP -iTCP:"${API_PORT}" -sTCP:LISTEN >/dev/null 2>&1; then print_err "端口 ${API_PORT} 已被占用，无法启动新的 Uvicorn。"
print_err "端口 ${API_PORT} 已被占用，无法启动新的 Uvicorn。" print_err "请先结束占用进程，例如: lsof -nP -iTCP:${API_PORT} -sTCP:LISTEN"
print_err "请先结束占用进程，例如: lsof -nP -iTCP:${API_PORT} -sTCP:LISTEN" exit 1
exit 1
fi
fi fi
# 迁移由 main.py 在启动时执行；排除 alembic 目录与 alembic.ini，避免编辑迁移时触发整进程重载 # 迁移由 main.py 在启动时执行；排除 alembic 目录与 alembic.ini，避免编辑迁移时触发整进程重载
@@ -230,10 +353,30 @@ start_services() {
--reload-exclude 'alembic.ini' \ --reload-exclude 'alembic.ini' \
--host "${API_HOST}" --port "${API_PORT}" & --host "${API_HOST}" --port "${API_PORT}" &
API_PID=$! API_PID=$!
print_ok "FastAPI 已启动 (PID: ${API_PID})" local api_start_status=0
if wait_for_tcp_listener "${API_PID}" "${API_PORT}" 8; then
api_start_status=0
else
api_start_status=$?
fi
case "${api_start_status}" in
0)
print_ok "FastAPI 已启动 (PID: ${API_PID})"
;;
1)
print_err "FastAPI 启动失败，进程已退出；请查看上方 Uvicorn 日志"
exit 1
;;
*)
print_err "FastAPI 进程仍存活，但端口 ${API_PORT} 未在预期时间内开始监听"
exit 1
;;
esac
"${CELERY_BIN}" -A app.tasks.celery_app worker --loglevel=info --pool="${CELERY_POOL}" & "${CELERY_BIN}" -A app.tasks.celery_app worker --loglevel=info --pool="${CELERY_POOL}" &
CELERY_PID=$! CELERY_PID=$!
ensure_background_process_alive "Celery" "${CELERY_PID}"
print_ok "Celery 已启动 (PID: ${CELERY_PID})" print_ok "Celery 已启动 (PID: ${CELERY_PID})"
echo echo

159
api/internal-eval.sh
View File

@@ -194,6 +194,122 @@ wait_postgres_ready() {
return 1 return 1
} }
get_effective_database_url() {
if [[ -n "${DATABASE_URL:-}" ]]; then
printf '%s\n' "${DATABASE_URL}"
return 0
fi
if [[ -f "${ROOT_DIR}/.env" ]]; then
local line
line="$(sed -n 's/^DATABASE_URL=//p' "${ROOT_DIR}/.env" | sed -n '1p')"
line="${line%\"}"
line="${line#\"}"
line="${line%\'}"
line="${line#\'}"
if [[ -n "${line}" ]]; then
printf '%s\n' "${line}"
return 0
fi
fi
return 1
}
warn_database_url_host_pitfall() {
local database_url
local host
if ! database_url="$(get_effective_database_url)"; then
return 0
fi
if [[ "${database_url}" =~ @([^:/?#]+) ]]; then
host="${BASH_REMATCH[1]}"
case "${host}" in
postgres|db|postgres-dev|postgresql)
print_warn "检测到 DATABASE_URL 主机为 ${host}；在宿主机执行 Alembic/uvicorn 时通常应使用 localhost"
;;
esac
fi
}
print_alembic_failure_hint() {
local log_file="$1"
local log_output
log_output="$(sed -n '1,200p' "${log_file}")"
if [[ "${log_output}" == *'could not translate host name "postgres"'* ]] || [[ "${log_output}" == *"Name or service not known"* ]]; then
print_warn "看起来 DATABASE_URL 指向了容器内主机名；在宿主机运行时请改用 localhost:5432"
elif [[ "${log_output}" == *"Connection refused"* ]] || [[ "${log_output}" == *"could not connect to server"* ]]; then
print_warn "PostgreSQL 连接被拒绝；请确认容器已启动且 DATABASE_URL 与 docker-compose.dev.yml 暴露端口一致"
elif [[ "${log_output}" == *"password authentication failed"* ]]; then
print_warn "PostgreSQL 用户名或密码不匹配；请核对 .env.development 中的 DATABASE_URL"
elif [[ "${log_output}" == *"No such file or directory"* ]] || [[ "${log_output}" == *"can't open file"* ]]; then
print_warn "Alembic 依赖的文件或工作目录可能不正确；请确认在 api/ 目录运行脚本"
fi
}
is_port_listening() {
local port="$1"
if command -v lsof >/dev/null 2>&1; then
lsof -nP -iTCP:"${port}" -sTCP:LISTEN >/dev/null 2>&1
return $?
fi
if [[ -x "${VENV_DIR}/bin/python" ]]; then
"${VENV_DIR}/bin/python" - "${port}" <<'PY' >/dev/null 2>&1
import socket
import sys
sock = socket.socket()
sock.settimeout(0.2)
try:
sock.connect(("127.0.0.1", int(sys.argv[1])))
except OSError:
raise SystemExit(1)
finally:
sock.close()
raise SystemExit(0)
PY
return $?
fi
return 1
}
wait_for_tcp_listener() {
local pid="$1"
local port="$2"
local timeout="${3:-8}"
local waited=0
while (( waited < timeout )); do
if is_port_listening "${port}"; then
return 0
fi
if ! is_pid_alive "${pid}"; then
return 1
fi
sleep 1
waited=$((waited + 1))
done
return 2
}
ensure_background_process_alive() {
local name="$1"
local pid="$2"
sleep 1
if ! is_pid_alive "${pid}"; then
print_err "${name} 启动后立即退出，请查看上方日志"
exit 1
fi
}
ensure_venv() { ensure_venv() {
print_header "检查 Python 虚拟环境" print_header "检查 Python 虚拟环境"
@@ -243,16 +359,25 @@ check_env_file() {
print_warn "未找到 .env，应用可能因缺少配置启动失败" print_warn "未找到 .env，应用可能因缺少配置启动失败"
else else
print_ok "检测到 .env" print_ok "检测到 .env"
warn_database_url_host_pitfall
fi fi
} }
run_migrations() { run_migrations() {
print_header "执行数据库迁移" print_header "执行数据库迁移"
cd "${ROOT_DIR}" cd "${ROOT_DIR}"
if uv run alembic upgrade head 2>/dev/null; then local log_file
log_file="$(mktemp -t life-echo-alembic.XXXXXX.log)"
if uv run alembic upgrade head >"${log_file}" 2>&1; then
print_ok "Alembic 迁移已就绪" print_ok "Alembic 迁移已就绪"
rm -f "${log_file}"
else else
print_warn "Alembic 迁移失败（可能数据库未启动或 DATABASE_URL 未配置），应用启动可能失败" print_warn "Alembic 迁移失败（可能数据库未启动或 DATABASE_URL 未配置），应用启动可能失败"
print_alembic_failure_hint "${log_file}"
print_warn "Alembic 输出（最近 40 行）："
tail -n 40 "${log_file}"
rm -f "${log_file}"
fi fi
} }
@@ -297,12 +422,10 @@ start_services() {
print_header "启动 Internal Eval API 与 Celery" print_header "启动 Internal Eval API 与 Celery"
cd "${ROOT_DIR}" cd "${ROOT_DIR}"
if command -v lsof >/dev/null 2>&1; then if is_port_listening "${INTERNAL_EVAL_PORT}"; then
if lsof -nP -iTCP:"${INTERNAL_EVAL_PORT}" -sTCP:LISTEN >/dev/null 2>&1; then print_err "端口 ${INTERNAL_EVAL_PORT} 已被占用，无法启动内部评测 Uvicorn。"
print_err "端口 ${INTERNAL_EVAL_PORT} 已被占用，无法启动内部评测 Uvicorn。" print_err "请先结束占用进程，或设置 INTERNAL_EVAL_PORT 为其他端口"
print_err "请先结束占用进程，或设置 INTERNAL_EVAL_PORT 为其他端口" exit 1
exit 1
fi
fi fi
# 与主开发脚本一致：评审/生产 LLM 等从 .env 读取；文档默认关闭，本地可 export INTERNAL_EVAL_ENABLE_DOCS=1 # 与主开发脚本一致：评审/生产 LLM 等从 .env 读取；文档默认关闭，本地可 export INTERNAL_EVAL_ENABLE_DOCS=1
@@ -311,11 +434,31 @@ start_services() {
--reload-exclude 'alembic.ini' \ --reload-exclude 'alembic.ini' \
--host "${INTERNAL_EVAL_HOST}" --port "${INTERNAL_EVAL_PORT}" & --host "${INTERNAL_EVAL_HOST}" --port "${INTERNAL_EVAL_PORT}" &
API_PID=$! API_PID=$!
print_ok "Internal Eval API 已启动 (PID: ${API_PID})" local api_start_status=0
if wait_for_tcp_listener "${API_PID}" "${INTERNAL_EVAL_PORT}" 8; then
api_start_status=0
else
api_start_status=$?
fi
case "${api_start_status}" in
0)
print_ok "Internal Eval API 已启动 (PID: ${API_PID})"
;;
1)
print_err "Internal Eval API 启动失败，进程已退出；请查看上方 Uvicorn 日志"
exit 1
;;
*)
print_err "Internal Eval API 进程仍存活，但端口 ${INTERNAL_EVAL_PORT} 未在预期时间内开始监听"
exit 1
;;
esac
if [[ "${SKIP_CELERY}" != "1" ]]; then if [[ "${SKIP_CELERY}" != "1" ]]; then
"${CELERY_BIN}" -A app.tasks.celery_app worker --loglevel=info --pool="${CELERY_POOL}" & "${CELERY_BIN}" -A app.tasks.celery_app worker --loglevel=info --pool="${CELERY_POOL}" &
CELERY_PID=$! CELERY_PID=$!
ensure_background_process_alive "Celery" "${CELERY_PID}"
print_ok "Celery 已启动 (PID: ${CELERY_PID})" print_ok "Celery 已启动 (PID: ${CELERY_PID})"
else else
print_warn "已跳过 Celery (SKIP_CELERY=1)；实验 run 接口需要 worker 才能执行" print_warn "已跳过 Celery (SKIP_CELERY=1)；实验 run 接口需要 worker 才能执行"