stonks-oracle/tests/integration/seed_minio.py

"""Seed MinIO buckets with sample normalized text files for integration tests.

Uploads synthetic normalized text corresponding to documents seeded by
seed_sandbox.py.  Each file is keyed by content_hash so the query API and
other services can locate them in the stonks-normalized bucket.

Usage:
    python -m tests.integration.seed_minio

Environment variables:
    MINIO_ENDPOINT   (default: minio:9000)
    MINIO_ACCESS_KEY (default: minioadmin)
    MINIO_SECRET_KEY (default: minioadmin)
    MINIO_SECURE     (default: false)
"""

import os
from io import BytesIO

from minio import Minio

BUCKET = "stonks-normalized"

# ---------------------------------------------------------------------------
# Sample normalized text content keyed by content_hash.
# These hashes match the documents inserted by seed_sandbox.py (DOC_01–DOC_10).
# We seed at least 5 files covering a mix of news, filings, and macro events.
# ---------------------------------------------------------------------------

NORMALIZED_TEXTS: dict[str, str] = {
    "hash_doc_01": (
        "Apple Inc reported fourth-quarter earnings that exceeded Wall Street "
        "expectations, driven by stronger-than-anticipated iPhone sales across "
        "all major markets. Revenue for the quarter came in at $89.5 billion, "
        "up 6 percent year over year, while earnings per share reached $1.46.\n\n"
        "Services revenue continued its upward trajectory, hitting a new record "
        "of $22.3 billion. Management highlighted growth in Apple TV+ subscribers "
        "and the expanding installed base of over 2.2 billion active devices.\n\n"
        "Greater China revenue declined 2 percent amid a competitive smartphone "
        "landscape, though management expressed confidence in the region's "
        "long-term trajectory. Gross margin expanded to 46.2 percent, reflecting "
        "favorable product mix and supply chain efficiencies."
    ),
    "hash_doc_02": (
        "Microsoft Corporation reported a 29 percent year-over-year increase in "
        "Azure cloud revenue, surpassing analyst estimates and reinforcing the "
        "company's position as a leading cloud infrastructure provider.\n\n"
        "Total Intelligent Cloud segment revenue reached $25.9 billion for the "
        "quarter. CEO Satya Nadella attributed the acceleration to enterprise "
        "adoption of AI workloads running on Azure, including OpenAI-powered "
        "services integrated into Microsoft 365 Copilot.\n\n"
        "Operating income for the segment grew 23 percent, with margins "
        "expanding despite increased capital expenditure on data center "
        "capacity. The company guided for continued double-digit Azure growth "
        "in the coming quarter."
    ),
    "hash_doc_03": (
        "JPMorgan Chase & Co filed its annual 10-K report with the Securities "
        "and Exchange Commission, disclosing record full-year net income of "
        "$49.6 billion. The filing detailed strong performance across all major "
        "business lines.\n\n"
        "Investment banking fees rose 18 percent for the year, driven by a "
        "rebound in equity and debt underwriting activity. The consumer banking "
        "division reported net interest income of $89.3 billion, benefiting "
        "from the elevated rate environment.\n\n"
        "The filing noted credit provisions of $9.8 billion, reflecting a "
        "cautious outlook on consumer credit quality. Total assets stood at "
        "$3.9 trillion, with a Common Equity Tier 1 ratio of 15.0 percent, "
        "well above regulatory minimums."
    ),
    "hash_doc_05": (
        "Exxon Mobil Corporation announced a 15 percent increase in Permian "
        "Basin production output, reaching 620,000 barrels of oil equivalent "
        "per day. The expansion was attributed to improved drilling efficiency "
        "and the integration of Pioneer Natural Resources assets.\n\n"
        "Total upstream production for the quarter averaged 3.7 million barrels "
        "of oil equivalent per day. Management reiterated its target of "
        "achieving 4.0 million barrels per day by year-end through organic "
        "growth and operational optimization.\n\n"
        "Downstream margins remained under pressure due to elevated refining "
        "costs and softer demand in European markets. The company maintained "
        "its quarterly dividend at $0.95 per share."
    ),
    "hash_doc_08": (
        "The Federal Reserve held its benchmark interest rate steady at the "
        "5.25 to 5.50 percent range following its January 2025 policy meeting, "
        "in line with market expectations. The decision was unanimous among "
        "voting members of the Federal Open Market Committee.\n\n"
        "In the accompanying statement, the committee acknowledged continued "
        "progress on inflation, with the core Personal Consumption Expenditures "
        "index declining to 2.6 percent. Chair Jerome Powell signaled that rate "
        "cuts could begin as early as the second quarter if disinflation trends "
        "persist.\n\n"
        "Treasury yields fell modestly following the announcement, with the "
        "10-year note declining 5 basis points to 4.12 percent. Equity markets "
        "rallied on the dovish forward guidance, with the S&P 500 gaining "
        "0.8 percent in after-hours trading."
    ),
    "hash_doc_09": (
        "Trade tensions between the United States and China escalated after "
        "the White House proposed a new round of tariffs targeting advanced "
        "semiconductor equipment and AI-related technology exports. The "
        "proposed measures would expand existing restrictions on chip "
        "manufacturing tools.\n\n"
        "Beijing responded with a statement warning of retaliatory measures "
        "on US agricultural and energy exports. Analysts noted that the "
        "escalation could disrupt supply chains for major technology companies "
        "with significant manufacturing operations in China.\n\n"
        "Shares of semiconductor equipment makers declined 3 to 5 percent in "
        "pre-market trading. Apple, which assembles the majority of its "
        "iPhones in China, saw its stock dip 1.2 percent on concerns about "
        "potential supply chain disruptions."
    ),
    "hash_doc_10": (
        "JPMorgan Chase reported a 20 percent increase in investment banking "
        "fees for the fourth quarter, driven by a surge in mergers and "
        "acquisitions advisory revenue and a recovery in initial public "
        "offering activity.\n\n"
        "The bank advised on several high-profile transactions during the "
        "quarter, including three deals valued at over $10 billion each. "
        "Equity capital markets revenue doubled compared to the prior year "
        "period as IPO volumes returned to pre-pandemic levels.\n\n"
        "Fixed income trading revenue rose 8 percent, supported by elevated "
        "volatility in interest rate and credit markets. The bank's total "
        "markets revenue reached $7.1 billion for the quarter, exceeding "
        "consensus estimates by approximately 5 percent."
    ),
}


def seed_minio() -> None:
    """Upload sample normalized text files to the stonks-normalized bucket."""
    client = Minio(
        os.environ.get("MINIO_ENDPOINT", "minio:9000"),
        access_key=os.environ.get("MINIO_ACCESS_KEY", "minioadmin"),
        secret_key=os.environ.get("MINIO_SECRET_KEY", "minioadmin"),
        secure=os.environ.get("MINIO_SECURE", "false").lower() == "true",
    )

    # Ensure bucket exists (should already be created by minio-bucket-init Job)
    if not client.bucket_exists(BUCKET):
        client.make_bucket(BUCKET)

    uploaded = 0
    for content_hash, text in NORMALIZED_TEXTS.items():
        key = f"{content_hash}.txt"
        data = text.encode("utf-8")
        client.put_object(
            BUCKET,
            key,
            BytesIO(data),
            length=len(data),
            content_type="text/plain",
        )
        uploaded += 1
        print(f"  uploaded {BUCKET}/{key} ({len(data)} bytes)")

    print(f"Seeded {uploaded} normalized text files into {BUCKET}.")


if __name__ == "__main__":
    seed_minio()