test: add containerized real-codex E2E (docker) (#686)

.dockerignore

@@ -0,0 +1,8 @@
+node_modules
+dist
+tmp
+artifacts
+*.log
+.env
+*.tgz
+.git

.github/workflows/ci-e2e-docker.yml

@@ -0,0 +1,35 @@
+name: E2E (real codex, Docker)
+
+on:
+  workflow_dispatch:
+    inputs:
+      forge:
+        description: "run forge phase (needs BILI_CODEX_COMPACT support in master)"
+        type: boolean
+        default: false
+
+jobs:
+  e2e-docker:
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    steps:
+      - uses: actions/checkout@v4
+      - name: Preflight (in container)
+        run: bash scripts/e2e-docker/run.sh preflight
+        env:
+          E2E_UPSTREAM_URL: ${{ secrets.E2E_UPSTREAM_URL }}
+          E2E_UPSTREAM_KEY: ${{ secrets.E2E_UPSTREAM_KEY }}
+      - name: E2E suite (in container)
+        run: bash scripts/e2e-docker/run.sh full
+        env:
+          E2E_UPSTREAM_URL: ${{ secrets.E2E_UPSTREAM_URL }}
+          E2E_UPSTREAM_KEY: ${{ secrets.E2E_UPSTREAM_KEY }}
+          E2E_FORGE: ${{ inputs.forge && '1' || '' }}
+      - name: Upload logs
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: e2e-docker-logs
+          path: tmp/docker-e2e/
+          retention-days: 7
+          if-no-files-found: ignore

package.json

@@ -15,7 +15,9 @@
     "typecheck": "tsc --noEmit --project tsconfig.build.json",
     "test": "node --import tsx --test tests/*.test.ts",
     "start": "node dist/index.js",
-    "test:e2e": "node --import tsx --test tests/e2e/e2e-codex.test.ts"
+    "test:e2e": "node --import tsx --test tests/e2e/e2e-codex.test.ts",
+    "e2e:docker": "bash scripts/e2e-docker/run.sh full",
+    "e2e:docker:preflight": "bash scripts/e2e-docker/run.sh preflight"
   },
   "keywords": [
     "context",

scripts/e2e-docker/Dockerfile

@@ -0,0 +1,17 @@
+FROM node:22-slim
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends ca-certificates \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /app
+
+COPY package.json package-lock.json ./
+RUN npm ci --ignore-scripts --no-fund --no-audit
+
+COPY . .
+RUN npm run build
+
+RUN npm install -g @openai/codex@latest
+
+CMD ["bash", "-lc", "E2E_CHECK=1 node --import tsx --test tests/e2e/e2e-codex.test.ts"]

scripts/e2e-docker/run.sh

@@ -0,0 +1,46 @@
+#!/usr/bin/env bash
+set -eo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
+cd "$REPO_ROOT"
+
+IMAGE="${BILI_E2E_IMAGE:-billion-context-e2e:local}"
+NETWORK="${E2E_DOCKER_NETWORK:-bridge}"
+LOG_DIR="${BILI_E2E_LOG_DIR:-$REPO_ROOT/tmp/docker-e2e}"
+MODE="${1:-full}"
+
+case "$MODE" in
+  full | preflight) ;;
+  *) echo "usage: $0 [full|preflight]" >&2; exit 64 ;;
+esac
+
+if ! command -v docker >/dev/null 2>&1; then
+  echo "error: docker not found on PATH" >&2
+  exit 3
+fi
+
+echo ">> building image $IMAGE"
+docker build -f scripts/e2e-docker/Dockerfile -t "$IMAGE" .
+
+FORWARD=()
+for v in E2E_UPSTREAM_URL E2E_UPSTREAM_KEY E2E_MODEL E2E_FORGE E2E_TMO E2E_CODEX_BIN; do
+  if [ -n "${!v:-}" ]; then FORWARD+=(-e "$v=${!v}"); fi
+done
+
+mkdir -p "$LOG_DIR"
+
+# Default upstream is host loopback; in a bridge container that's the container
+# itself. Use E2E_DOCKER_NETWORK=host, or set E2E_UPSTREAM_URL reachable from the container.
+if [ "$MODE" = "preflight" ]; then
+  CMD="E2E_CHECK=1 node --import tsx --test tests/e2e/e2e-codex.test.ts"
+else
+  CMD="ACP_TEST_E2E=1 node --import tsx --test tests/e2e/e2e-codex.test.ts"
+fi
+
+echo ">> running e2e [$MODE] in container (network=$NETWORK, logs=$LOG_DIR)"
+exec docker run --rm \
+  --network "$NETWORK" \
+  -v "$LOG_DIR:/app/tmp" \
+  "${FORWARD[@]}" \
+  "$IMAGE" bash -lc "$CMD"