Skip to content

Commit b8159dc

Browse files
committed
Add final ContextBench lane readiness workflow
1 parent 2263513 commit b8159dc

1 file changed

Lines changed: 94 additions & 0 deletions

File tree

Lines changed: 94 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,94 @@
1+
name: ContextBench Final Lane Readiness
2+
3+
on:
4+
push:
5+
branches: [master]
6+
paths:
7+
- .github/workflows/contextbench-final-lane-readiness.yml
8+
- scripts/contextbench-final-lane-candidate-pack.mjs
9+
workflow_dispatch:
10+
11+
permissions:
12+
contents: read
13+
14+
jobs:
15+
final-lane-readiness:
16+
runs-on: ubuntu-latest
17+
timeout-minutes: 35
18+
strategy:
19+
fail-fast: false
20+
max-parallel: 2
21+
matrix:
22+
lane:
23+
- grepai
24+
- codegraphcontext
25+
env:
26+
ROOT: /tmp/contextbench-final-lane/${{ matrix.lane }}
27+
TASK_PAYLOADS: /tmp/contextbench-final-lane/${{ matrix.lane }}/task-payloads.json
28+
CHECKOUT_ROOT: /tmp/contextbench-final-checkouts-${{ matrix.lane }}
29+
TARGET_TASK_ID: SWE-Bench-Pro__go__maintenance__bugfix__4df06349
30+
LANE_ID: ${{ matrix.lane }}
31+
OLLAMA_HOST: http://127.0.0.1:11434
32+
steps:
33+
- uses: actions/checkout@v4
34+
- uses: pnpm/action-setup@v2
35+
with:
36+
version: 10
37+
- uses: actions/setup-node@v4
38+
with:
39+
node-version: '24'
40+
cache: pnpm
41+
- uses: actions/setup-python@v5
42+
with:
43+
python-version: '3.11'
44+
- name: Install shared setup and materialize selected Go task
45+
shell: bash
46+
run: |
47+
set -euo pipefail
48+
mkdir -p "$ROOT" "$CHECKOUT_ROOT" "$ROOT/tool" "$ROOT/logs"
49+
pnpm install --frozen-lockfile > "$ROOT/logs/pnpm-install.log" 2>&1
50+
python -m pip install "tree-sitter==0.20.4" "tree-sitter-languages==1.10.2" datasets pyarrow > "$ROOT/logs/pip-shared.log" 2>&1
51+
node scripts/contextbench-runner.mjs --validate-fixtures > "$ROOT/logs/validate-fixtures.log" 2>&1
52+
node scripts/contextbench-select-slice.mjs --write-task-payloads --out "$TASK_PAYLOADS.all" --checkout-root "$CHECKOUT_ROOT" > "$ROOT/logs/write-payloads.log" 2>&1
53+
node - <<'NODE'
54+
const fs = require('node:fs');
55+
const payloadPath = process.env.TASK_PAYLOADS;
56+
const target = process.env.TARGET_TASK_ID;
57+
const payload = JSON.parse(fs.readFileSync(`${payloadPath}.all`, 'utf8'));
58+
const task = payload.tasks.find((candidate) => candidate.instance_id === target);
59+
if (!task) throw new Error(`target task ${target} not found`);
60+
fs.writeFileSync(payloadPath, `${JSON.stringify({ ...payload, task_count: 1, tasks: [task] }, null, 2)}\n`);
61+
NODE
62+
node scripts/contextbench-select-slice.mjs --materialize-checkouts --payloads "$TASK_PAYLOADS" --max-tasks 1 > "$ROOT/logs/materialize.log" 2>&1
63+
- name: Install grepai with local Ollama embeddings
64+
if: matrix.lane == 'grepai'
65+
shell: bash
66+
run: |
67+
set -euo pipefail
68+
curl -sSL https://raw.githubusercontent.com/yoanbernabeu/grepai/main/install.sh | sh > "$ROOT/logs/grepai-install.log" 2>&1
69+
echo "$HOME/.local/bin" >> "$GITHUB_PATH"
70+
echo "$HOME/bin" >> "$GITHUB_PATH"
71+
curl -fsSL https://ollama.com/install.sh | sh > "$ROOT/logs/ollama-install.log" 2>&1
72+
nohup ollama serve > "$ROOT/logs/ollama-serve.log" 2>&1 &
73+
for i in {1..45}; do
74+
if curl -fsS "$OLLAMA_HOST/api/version" > "$ROOT/logs/ollama-version.log" 2>&1; then
75+
break
76+
fi
77+
sleep 2
78+
done
79+
curl -fsS "$OLLAMA_HOST/api/version" >> "$ROOT/logs/ollama-version.log" 2>&1
80+
ollama pull nomic-embed-text > "$ROOT/logs/ollama-pull-nomic-embed-text.log" 2>&1
81+
- name: Install codegraphcontext lane
82+
if: matrix.lane == 'codegraphcontext'
83+
shell: bash
84+
run: python -m pip install codegraphcontext kuzu real_ladybug > "$ROOT/logs/pip-codegraphcontext.log" 2>&1
85+
- name: Generate final lane readiness candidate pack
86+
shell: bash
87+
run: node scripts/contextbench-final-lane-candidate-pack.mjs
88+
- name: Upload final lane readiness artifacts
89+
if: always()
90+
uses: actions/upload-artifact@v4
91+
with:
92+
name: contextbench-final-lane-readiness-${{ matrix.lane }}
93+
path: /tmp/contextbench-final-lane/${{ matrix.lane }}
94+
retention-days: 14

0 commit comments

Comments
 (0)