openai-streaming-response
Model to use for streaming completion
import sys
import os
import subprocess
import time
import urllib.request
import json
# ─────────────────────────────────────────
# PRE_EXECUTION
# ─────────────────────────────────────────
for attempt in range(2):
try:
req = urllib.request.Request(
"https://checklist.day/api/registry/openai",
headers={"User-Agent": "checklist-agent/1.0"}
)
with urllib.request.urlopen(req, timeout=10) as resp:
registry = json.loads(resp.read())
break
except Exception as e:
if attempt == 1:
print(f"ABORT: registry unreachable — {e}")
sys.exit(1)
time.sleep(2)
warnings = registry.get("warnings", [])
if warnings:
print("[openai] WARNINGS:")
for w in warnings if isinstance(warnings, list) else [warnings]:
print(f" ⚠ {w}")
# ─────────────────────────────────────────
# EXECUTION
# ─────────────────────────────────────────
subprocess.check_call([sys.executable, "-m", "pip", "install", "-q", "openai>=1.0.0"])
from openai import OpenAI
OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
OPENAI_MODEL = os.environ.get("OPENAI_MODEL", "gpt-4o-mini")
if not OPENAI_API_KEY:
print("ABORT: OPENAI_API_KEY env var not set")
sys.exit(1)
client = OpenAI(api_key=OPENAI_API_KEY)
chunks_received = 0
accumulated = []
# FOOTGUN: stream=True returns a generator, not a response object
# FOOTGUN: chunk.choices[0].delta.content is None on first and last chunks — must guard with `or ""`
# FOOTGUN: usage is None on all chunks except the last when stream_options={"include_usage": True}
with client.chat.completions.create(
model=OPENAI_MODEL,
messages=[
{"role": "user", "content": "Count from 1 to 5, one number per line, nothing else."},
],
max_tokens=32,
temperature=0,
stream=True,
) as stream:
for chunk in stream:
chunks_received += 1
# FOOTGUN: always guard — delta.content is None on role/finish chunks
delta = chunk.choices[0].delta.content if chunk.choices else None
if delta:
accumulated.append(delta)
print(delta, end="", flush=True)
print() # newline after stream
content = "".join(accumulated)
stream_ok = chunks_received > 1 and len(content) > 0
print(f" chunks received: {chunks_received}")
print(f" content length: {len(content)} chars")
# ─────────────────────────────────────────
# POST_EXECUTION
# ─────────────────────────────────────────
assert stream_ok, f"FAIL: streaming failed — chunks={chunks_received}, content={content!r}"
assert chunks_received > 1, "FAIL: expected multiple chunks, got 1 (not streaming?)"
# Verify numbers 1-5 are present in output
for n in ["1", "2", "3", "4", "5"]:
assert n in content, f"FAIL: expected '{n}' in streamed content, got: {content!r}"
result = {
"chunks_received": chunks_received,
"content": content.strip(),
"stream_ok": stream_ok,
}
print(json.dumps(result, indent=2))
print("PASS")