fidc-backtest-engine/scripts/verify-no-legacy-data-source.sh

#!/usr/bin/env bash
set -euo pipefail

ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
cd "$ROOT_DIR"

fail() {
  local message="$1"
  local details="${2:-}"
  printf '[FAIL] %s\n' "$message" >&2
  if [[ -n "$details" ]]; then
    printf '%s\n' "$details" >&2
  fi
  exit 1
}

runtime_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'fidatacenter|FIDATACENTER|/v1/backtest/data|/v1/xuntou|ClickHouse|clickhouse|CLICKHOUSE|FIDC_BT_INSTRUMENT_METADATA_CSV|FIDC_BT_WRITE_SNAPSHOTS|FIDC_BT_WRITE_CSV_SNAPSHOTS|FIDC_BT_WRITE_COMBINED_SOURCE_ROW_CACHE|write_csv_snapshot_files|write_csv\(|FIDC_RISK_RUNTIME_FILE|FIDC_RISK_RUNTIME_URL|FIDC_FIRISK_RUNTIME_FILE|FiRisk runtime snapshot' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$runtime_hits" ]]; then
  fail "legacy fidatacenter/ClickHouse/CSV snapshot/FiRisk runtime data source references are not allowed in fidc-backtest-engine" "$runtime_hits"
fi

manifest_hits="$(
  rg -n --glob '!scripts/verify-no-legacy-data-source.sh' '\b(mysql|mariadb|clickhouse|clickhouse-rs|mysql_async|sqlx-mysql)\b' Cargo.toml crates 2>/dev/null || true
)"

if [[ -n "$manifest_hits" ]]; then
  fail "legacy database dependencies are not allowed in fidc-backtest-engine manifests" "$manifest_hits"
fi

local_only_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'FICLAW_DATA_AGENT_URL|ficlaw_data\.source_rows_v1|strategy-factory-source-lake://local|FIDC_BT_WRITE_COMBINED_SOURCE_ROW_CACHE' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$local_only_hits" ]]; then
  fail "legacy or local-only data source references are not allowed in fidc-backtest-engine runtime" "$local_only_hits"
fi

json_query_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    '/v1/query/(source-rows|daily-execution-prices|minute-execution-prices|instruments|corporate-actions)\.json' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$json_query_hits" ]]; then
  fail "JSON Source Lake high-throughput endpoints are not allowed in fidc-backtest-engine runtime; use Arrow endpoints" "$json_query_hits"
fi

fused_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'exported_fused|fidc_fused|fusion|fused|wide_table|wide table|source_rows_export|exported_daily|merged_daily|daily_merged|merged_source|materialized[_ -]source|materialized_source_rows|source_rows_materialized|融合表|融合宽表' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$fused_hits" ]]; then
  fail "exported fused tables are not allowed in fidc-backtest-engine runtime; use Strategy Factory Source Lake source rows directly" "$fused_hits"
fi

feature_store_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'research_feature_store|feature_store|FEATURE_STORE|daily_minute_current|FIDC_STRATEGY_FACTORY_ENABLE_FEATURE_STORE_CACHE|ALPHA_FACTORY_ENABLE_FEATURE_STORE_CACHE|ENABLE_FEATURE_STORE_CACHE' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$feature_store_hits" ]]; then
  fail "historical feature-store paths are not allowed in fidc-backtest-engine runtime; use Strategy Factory Source Lake raw/indicator/artifact partitions and discardable caches" "$feature_store_hits"
fi

truth_csv_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!crates/fidc-core/src/strategy_ai.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'FIDC_BT_TRUTH_STOCK_LIST_CSV|OMNI_BT_TRUTH_STOCK_LIST_CSV|OMNI_BACKTEST_TRUTH_STOCK_LIST_CSV|selection_source=truth_csv|truth_stock_list|truth_csv' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$truth_csv_hits" ]]; then
  fail "CSV truth stock-list overrides are not allowed in fidc-backtest-engine runtime; use Source Lake runtime spec selection only" "$truth_csv_hits"
fi

csv_snapshot_loader_hits="$(
  rg -n \
    --glob '!**/.git/**' \
    --glob '!**/target/**' \
    --glob '!**/docs/**' \
    --glob '!**/*.md' \
    --glob '!**/tests/**' \
    --glob '!**/*.test.rs' \
    --glob '!**/*_test.rs' \
    --glob '!scripts/verify-no-legacy-data-source.sh' \
    'from_csv_dir|from_partitioned_dir|instruments\.csv|candidate_flags\.csv|market\.csv|benchmark\.csv' \
    crates Cargo.toml \
    2>/dev/null || true
)"

if [[ -n "$csv_snapshot_loader_hits" ]]; then
  fail "CSV snapshot loaders are not allowed in fidc-backtest-engine runtime; construct DataSet from Source Lake components" "$csv_snapshot_loader_hits"
fi

printf '[OK] fidc-backtest-engine has no legacy runtime data-source references\n'