Model/tests/scripts/test_rename_sharepoint_files.py
Daniel Roth 787f7731b2 UPRN-prefixed property folders rename to a single-UPRN filename 🟥
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-03 12:09:02 +00:00

304 lines
9.6 KiB
Python

from pathlib import Path
from typing import Any
from unittest.mock import MagicMock
import pytest
from backend.pashub_fetcher.sharepoint_subfolders import SharepointSubfolders
from orchestration.sharepoint_renamer_orchestrator import (
ASSESSMENT_SUBFOLDER,
BASE_PATH,
SharepointRenamerOrchestrator,
build_canonical_filename,
)
def _assessment_path(root: str, address: str, postcode: str) -> str:
return (
f"{root}/{address}, {postcode}"
f"/{SharepointSubfolders.ASSESSMENT.value}/{ASSESSMENT_SUBFOLDER}"
)
def _write_csv(tmp_path: Path, rows: list[tuple[str, str, str]]) -> str:
lines = ["UPRN,Address,Postcode"] + [",".join(row) for row in rows]
csv_file = tmp_path / "addresses.csv"
csv_file.write_text("\n".join(lines) + "\n", encoding="utf-8")
return str(csv_file)
def _make_file(name: str, item_id: str = "id-1") -> dict[str, Any]:
return {"name": name, "id": item_id, "file": {}}
def _make_folder(name: str) -> dict[str, Any]:
return {"name": name, "folder": {}}
def _make_package(name: str) -> dict[str, Any]:
return {"name": name, "package": {}}
def _make_orchestrator(sp: MagicMock, dry_run: bool = False) -> SharepointRenamerOrchestrator:
orchestrator = SharepointRenamerOrchestrator.__new__(SharepointRenamerOrchestrator)
orchestrator._sp_client = sp
orchestrator._dry_run = dry_run
return orchestrator
# ---------------------------------------------------------------------------
# build_canonical_filename
# ---------------------------------------------------------------------------
def test_already_canonical_returns_none() -> None:
assert build_canonical_filename("100", "1 High St", "AB1 2CD", "100_High St AB1 2CD_Report.pdf") is None
def test_strips_address_prefix_and_adds_uprn() -> None:
result = build_canonical_filename("100", "1 High St", "AB1 2CD", "1 High St AB1 2CD - Survey.pdf")
assert result == "100_1 High St AB1 2CD_Survey.pdf"
def test_no_prefix_still_canonical() -> None:
result = build_canonical_filename("100", "1 High St", "AB1 2CD", "Survey.pdf")
assert result == "100_1 High St AB1 2CD_Survey.pdf"
# ---------------------------------------------------------------------------
# _discover_roots — batch subfolders become extra roots
# ---------------------------------------------------------------------------
def test_discover_roots_includes_sero_batch_folders() -> None:
# Arrange
sp = MagicMock()
sp.get_folders_in_path.return_value = {
"value": [
_make_folder("1 High St, AB1 2CD"),
_make_folder("_Sero Batch 3"),
_make_folder("_Sero Batch 2"),
_make_file("stray.pdf"),
]
}
# Act
roots = _make_orchestrator(sp)._discover_roots()
# Assert
assert roots == [
BASE_PATH,
f"{BASE_PATH}/_Sero Batch 2",
f"{BASE_PATH}/_Sero Batch 3",
]
sp.get_folders_in_path.assert_called_once_with(BASE_PATH)
# ---------------------------------------------------------------------------
# run — properties under a batch root are still found and renamed
# ---------------------------------------------------------------------------
def test_run_renames_files_for_property_under_batch_root(tmp_path: Path) -> None:
# Arrange
csv_path = _write_csv(tmp_path, [("100", "1 High St", "AB1 2CD")])
batch_assessment = _assessment_path(
f"{BASE_PATH}/_Sero Batch 2", "1 High St", "AB1 2CD"
)
sp = MagicMock()
def fake_get(path: str) -> dict[str, Any]:
if path == BASE_PATH:
return {"value": [_make_folder("_Sero Batch 2")]}
if path == batch_assessment:
return {"value": [_make_file("Survey.pdf", "id-1")]}
raise ValueError("not found")
sp.get_folders_in_path.side_effect = fake_get
# Act
SharepointRenamerOrchestrator(sp, csv_path).run()
# Assert
sp.rename_file.assert_called_once_with("id-1", "100_1 High St AB1 2CD_Survey.pdf")
def test_run_uprn_prefixed_address_yields_single_uprn_filename(
tmp_path: Path,
) -> None:
# Arrange
csv_path = _write_csv(tmp_path, [("100", "100_1 High St", "AB1 2CD")])
batch_assessment = _assessment_path(
f"{BASE_PATH}/_Sero Batch 2", "100_1 High St", "AB1 2CD"
)
sp = MagicMock()
def fake_get(path: str) -> dict[str, Any]:
if path == BASE_PATH:
return {"value": [_make_folder("_Sero Batch 2")]}
if path == batch_assessment:
return {"value": [_make_file("Survey.pdf", "id-1")]}
raise ValueError("not found")
sp.get_folders_in_path.side_effect = fake_get
# Act
SharepointRenamerOrchestrator(sp, csv_path).run()
# Assert
sp.rename_file.assert_called_once_with("id-1", "100_1 High St AB1 2CD_Survey.pdf")
# ---------------------------------------------------------------------------
# _process_folder — files only at root level
# ---------------------------------------------------------------------------
def test_renames_top_level_files(caplog: pytest.LogCaptureFixture) -> None:
sp = MagicMock()
sp.get_folders_in_path.return_value = {
"value": [
_make_file("Survey.pdf", "id-1"),
_make_file("Report.docx", "id-2"),
]
}
_make_orchestrator(sp)._process_folder("some/path", "100", "1 High St", "AB1 2CD")
assert sp.rename_file.call_count == 2
sp.rename_file.assert_any_call("id-1", "100_1 High St AB1 2CD_Survey.pdf")
sp.rename_file.assert_any_call("id-2", "100_1 High St AB1 2CD_Report.docx")
# ---------------------------------------------------------------------------
# _process_folder — recursive two-level hierarchy
# ---------------------------------------------------------------------------
def test_recurses_into_subfolders_and_renames_all_files() -> None:
sp = MagicMock()
root_contents: dict[str, Any] = {
"value": [
_make_file("Root.pdf", "root-file"),
_make_folder("SubA"),
]
}
suba_contents: dict[str, Any] = {
"value": [
_make_file("Sub.pdf", "sub-file"),
]
}
sp.get_folders_in_path.side_effect = lambda path: (
root_contents if path == "base/path" else suba_contents
)
_make_orchestrator(sp)._process_folder("base/path", "200", "2 Main Rd", "XY9 8ZW")
assert sp.rename_file.call_count == 2
sp.rename_file.assert_any_call("root-file", "200_2 Main Rd XY9 8ZW_Root.pdf")
sp.rename_file.assert_any_call("sub-file", "200_2 Main Rd XY9 8ZW_Sub.pdf")
sp.get_folders_in_path.assert_any_call("base/path/SubA")
# ---------------------------------------------------------------------------
# _process_folder — non-file, non-folder items are skipped
# ---------------------------------------------------------------------------
def test_ignores_package_items() -> None:
sp = MagicMock()
sp.get_folders_in_path.return_value = {"value": [_make_package("Notebook")]}
_make_orchestrator(sp)._process_folder("some/path", "300", "3 Oak Ave", "ZZ1 1ZZ")
sp.rename_file.assert_not_called()
assert sp.get_folders_in_path.call_count == 1
# ---------------------------------------------------------------------------
# _process_folder — missing folder
# ---------------------------------------------------------------------------
def test_missing_folder_returns_false_without_warning(
caplog: pytest.LogCaptureFixture,
) -> None:
# Arrange
sp = MagicMock()
sp.get_folders_in_path.side_effect = ValueError("not found")
# Act
found = _make_orchestrator(sp)._process_folder(
"missing/path", "400", "4 Elm St", "AA2 2BB"
)
# Assert
assert found is False
sp.rename_file.assert_not_called()
assert not any("Missing folder" in r.message for r in caplog.records)
def test_run_warns_once_when_property_missing_from_all_roots(
tmp_path: Path, caplog: pytest.LogCaptureFixture
) -> None:
# Arrange
csv_path = _write_csv(tmp_path, [("400", "4 Elm St", "AA2 2BB")])
sp = MagicMock()
def fake_get(path: str) -> dict[str, Any]:
if path == BASE_PATH:
return {"value": [_make_folder("_Sero Batch 2")]}
raise ValueError("not found")
sp.get_folders_in_path.side_effect = fake_get
# Act
SharepointRenamerOrchestrator(sp, csv_path).run()
# Assert
sp.rename_file.assert_not_called()
warnings = [r for r in caplog.records if "Missing folder" in r.message]
assert len(warnings) == 1
assert "400" in warnings[0].message
# ---------------------------------------------------------------------------
# _process_folder — already-canonical files are skipped
# ---------------------------------------------------------------------------
def test_skips_already_canonical_files() -> None:
sp = MagicMock()
sp.get_folders_in_path.return_value = {
"value": [_make_file("500_Pine Ln BB3 3CC_Doc.pdf", "id-y")]
}
_make_orchestrator(sp)._process_folder("some/path", "500", "5 Pine Ln", "BB3 3CC")
sp.rename_file.assert_not_called()
# ---------------------------------------------------------------------------
# _process_folder — dry_run=True logs intent but never calls rename_file
# ---------------------------------------------------------------------------
def test_dry_run_logs_would_rename_without_calling_api(
caplog: pytest.LogCaptureFixture,
) -> None:
sp = MagicMock()
sp.get_folders_in_path.return_value = {
"value": [_make_file("Survey.pdf", "id-1")]
}
_make_orchestrator(sp, dry_run=True)._process_folder(
"some/path", "100", "1 High St", "AB1 2CD"
)
sp.rename_file.assert_not_called()
assert any("Would rename" in r.message for r in caplog.records)