# Recruiter-safe sample from Synapse, an independently owned project.
# This is the real test logic for public NASA caption data.

from nasa_ingest import (
    NASA_WEBB_CAPTION_URL,
    NASA_WEBB_MEDIA_URL,
    build_nasa_worker_result,
    link_topics,
    parse_srt,
)


SAMPLE_SRT = """1
00:00:00,000 --> 00:00:03,000
The James Webb Space Telescope is NASA's next great observatory.

2
00:00:03,500 --> 00:00:08,250
Webb studies infrared astronomy, exoplanets, and galaxies after Hubble.

3
00:00:09,000 --> 00:00:11,000
NASA works with ESA and CSA on the mission.
"""


def test_parse_srt_converts_captions_to_ordered_transcript_chunks():
    chunks = parse_srt(SAMPLE_SRT)

    assert [chunk.chunk_index for chunk in chunks] == [0, 1, 2]
    assert chunks[0].start_ms == 0
    assert chunks[0].end_ms == 3000
    assert chunks[1].start_ms == 3500
    assert chunks[1].end_ms == 8250
    assert chunks[0].text == (
        "The James Webb Space Telescope is NASA's next great observatory."
    )


def test_link_topics_detects_known_terms_without_duplicate_links_per_chunk():
    chunks = parse_srt(
        """1
00:00:00,000 --> 00:00:04,000
NASA says NASA and Webb build on Hubble.
"""
    )

    links = link_topics(chunks)

    assert [link.label for link in links] == [
        "James Webb Space Telescope",
        "Hubble Space Telescope",
        "NASA",
    ]
    assert [link.transcript_chunk_index for link in links] == [0, 0, 0]
    assert all(0 <= link.confidence <= 1 for link in links)
    assert all(link.review_status in {"needs_review", "accepted"} for link in links)


def test_build_nasa_worker_result_includes_caption_provenance_and_learning_links():
    result = build_nasa_worker_result(
        srt_text=SAMPLE_SRT,
        job_id="job_nasa_demo",
        workspace_id="workspace_nasa_demo",
    )

    media = result["media"][0]
    first_chunk = media["transcriptChunks"][0]

    assert result["jobId"] == "job_nasa_demo"
    assert result["workspaceId"] == "workspace_nasa_demo"
    assert media["title"] == "NASA Learning Demo - James Webb Space Telescope"
    assert media["sourceMetadata"]["nasaId"] == "James_Webb_Space_Telescope__An_Overview_1080"
    assert media["sourceMetadata"]["mediaUrl"] == NASA_WEBB_MEDIA_URL
    assert media["sourceMetadata"]["captionUrl"] == NASA_WEBB_CAPTION_URL
    assert media["sourceMetadata"]["transcriptProvenance"] == "official_caption"
    assert first_chunk["searchMetadata"]["embeddingStatus"] == "pending"
    assert any(
        link["label"] == "James Webb Space Telescope"
        for chunk in media["transcriptChunks"]
        for link in chunk["learningLinks"]
    )
    assert any(
        "Wikipedia" in link["attribution"]
        for chunk in media["transcriptChunks"]
        for link in chunk["learningLinks"]
    )