功能：完善里程边界与氢耗流式统计

2026-09-02 14:05:04 +08:00
parent 4201878c35
commit 8759668d39
23 changed files with 1638 additions and 159 deletions
@@ -0,0 +1,126 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+BINARY="${BINARY:-/opt/lingniu-go-native/current/stats-backfill}"
+DATE_FROM="${DATE_FROM:-}"
+DATE_TO="${DATE_TO:-}"
+PARALLEL="${PARALLEL:-4}"
+APPLY=false
+WORK_DIR=""
+
+usage() {
+  cat <<'USAGE'
+Usage: replay-pure-hydrogen-mileage.sh --date-from YYYY-MM-DD --date-to YYYY-MM-DD [options]
+
+Options:
+  --apply             Write the rebuilt daily mileage rows. Default is dry-run.
+  --parallel N        Concurrent VIN workers. Default: 4.
+  --binary PATH       stats-backfill binary path.
+  --work-dir PATH     Keep VIN lists and per-VIN logs in this directory.
+USAGE
+}
+
+if [[ "${1:-}" == "--worker" ]]; then
+  vin="$2"
+  dry_run="$3"
+  log_dir="$4"
+  env \
+    BACKFILL_METHOD=scan \
+    BACKFILL_DRY_RUN="$dry_run" \
+    BACKFILL_DATE_FROM="$DATE_FROM" \
+    BACKFILL_DATE_TO="$DATE_TO" \
+    BACKFILL_PROTOCOLS=GB32960 \
+    BACKFILL_VINS="$vin" \
+    BACKFILL_PROGRESS_EVERY=100000 \
+    "$BINARY" >"$log_dir/$vin.log" 2>&1
+  printf 'completed vin=%s\n' "$vin"
+  exit 0
+fi
+
+while (($#)); do
+  case "$1" in
+    --date-from) DATE_FROM="$2"; shift 2 ;;
+    --date-to) DATE_TO="$2"; shift 2 ;;
+    --parallel) PARALLEL="$2"; shift 2 ;;
+    --binary) BINARY="$2"; shift 2 ;;
+    --work-dir) WORK_DIR="$2"; shift 2 ;;
+    --apply) APPLY=true; shift ;;
+    -h|--help) usage; exit 0 ;;
+    *) echo "unknown argument: $1" >&2; usage >&2; exit 2 ;;
+  esac
+done
+
+if [[ ! "$DATE_FROM" =~ ^[0-9]{4}-[0-9]{2}-[0-9]{2}$ || ! "$DATE_TO" =~ ^[0-9]{4}-[0-9]{2}-[0-9]{2}$ ]]; then
+  echo "--date-from and --date-to are required" >&2
+  exit 2
+fi
+if [[ ! "$PARALLEL" =~ ^[1-9][0-9]*$ ]]; then
+  echo "--parallel must be a positive integer" >&2
+  exit 2
+fi
+if [[ ! -x "$BINARY" ]]; then
+  echo "stats-backfill is not executable: $BINARY" >&2
+  exit 2
+fi
+
+if [[ -z "$WORK_DIR" ]]; then
+  WORK_DIR="$(mktemp -d /tmp/pure-hydrogen-replay.XXXXXX)"
+else
+  mkdir -p "$WORK_DIR"
+fi
+LOG_DIR="$WORK_DIR/logs"
+mkdir -p "$LOG_DIR"
+
+hydrogen_vins="$WORK_DIR/hydrogen-vins.txt"
+active_vins="$WORK_DIR/active-gb32960-vins.txt"
+target_vins="$WORK_DIR/target-vins.txt"
+pending_vins="$WORK_DIR/pending-vins.txt"
+
+BACKFILL_LIST_HYDROGEN_VINS=true "$BINARY" | LC_ALL=C sort -u >"$hydrogen_vins"
+env \
+  BACKFILL_LIST_ACTIVE_VINS=true \
+  BACKFILL_DATE_FROM="$DATE_FROM" \
+  BACKFILL_DATE_TO="$DATE_TO" \
+  BACKFILL_PROTOCOLS=GB32960 \
+  "$BINARY" | LC_ALL=C sort -u >"$active_vins"
+comm -12 "$hydrogen_vins" "$active_vins" >"$target_vins"
+
+target_count="$(wc -l <"$target_vins" | tr -d ' ')"
+: >"$pending_vins"
+while IFS= read -r vin; do
+  if [[ -f "$LOG_DIR/$vin.log" ]] && grep -q 'stats backfill complete' "$LOG_DIR/$vin.log"; then
+    continue
+  fi
+  echo "$vin" >>"$pending_vins"
+done <"$target_vins"
+pending_count="$(wc -l <"$pending_vins" | tr -d ' ')"
+dry_run=true
+if [[ "$APPLY" == true ]]; then
+  dry_run=false
+fi
+printf 'pure-hydrogen replay: date=%s..%s vins=%s pending=%s parallel=%s dry_run=%s work_dir=%s\n' \
+  "$DATE_FROM" "$DATE_TO" "$target_count" "$pending_count" "$PARALLEL" "$dry_run" "$WORK_DIR"
+
+if [[ "$pending_count" == "0" ]]; then
+  printf 'pure-hydrogen replay complete: vins=%s logs=%s\n' "$target_count" "$LOG_DIR"
+  exit 0
+fi
+
+export BINARY DATE_FROM DATE_TO
+set +e
+xargs -I '{}' -P "$PARALLEL" "$0" --worker '{}' "$dry_run" "$LOG_DIR" <"$pending_vins"
+xargs_status=$?
+set -e
+
+failed_vins="$WORK_DIR/failed-vins.txt"
+: >"$failed_vins"
+while IFS= read -r vin; do
+  if [[ ! -f "$LOG_DIR/$vin.log" ]] || ! grep -q 'stats backfill complete' "$LOG_DIR/$vin.log"; then
+    echo "$vin" >>"$failed_vins"
+  fi
+done <"$target_vins"
+if [[ "$xargs_status" != "0" || -s "$failed_vins" ]]; then
+  echo "replay incomplete; inspect $failed_vins and $LOG_DIR" >&2
+  exit 1
+fi
+printf 'pure-hydrogen replay complete: vins=%s logs=%s\n' "$target_count" "$LOG_DIR"