scripts/test_oa_merge.py

import os
import sys

sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from plagdetect.websearch import dedupe_candidates

# Simulate: OpenAlex found a Nature paper but LOCKED (no pdf_url);
# CORE found the SAME paper (same DOI) with an OA repository PDF.
openalex_locked = {
    "title": "A Deep Learning Framework for Protein Folding",
    "doi": "10.1038/s41586-021-03819-2", "pdf_url": None, "fulltext": "",
    "abstract": "", "provider": "openalex",
}
core_oa = {
    "title": "A deep learning framework for protein folding",  # case differs
    "doi": "10.1038/s41586-021-03819-2",
    "pdf_url": "https://repository.example.edu/oa/protein.pdf",
    "abstract": "We present a deep learning system ...", "provider": "core",
}
crossref_dup = {
    "title": "A Deep Learning Framework for Protein Folding",
    "doi": "10.1038/s41586-021-03819-2", "pdf_url": None, "provider": "crossref",
}

merged = dedupe_candidates([openalex_locked, crossref_dup, core_oa])
print("candidates after dedupe:", len(merged), "(expect 1)")
m = merged[0]
print("kept provider:", m["provider"], "(the first seen = openalex)")
print("pdf_url merged in:", m.get("pdf_url"))
print("oa_via:", m.get("oa_via"))
print("abstract merged in:", bool(m.get("abstract")))
assert len(merged) == 1, "should dedupe to one"
assert m["pdf_url"] == core_oa["pdf_url"], "CORE pdf must merge into locked paper"
assert m["oa_via"] == "core"
assert m["abstract"], "abstract should be filled from CORE"
print("\nMERGE OK — locked Nature paper now deep-readable via CORE OA copy")