|
1
|
"""Tests for navegador.ingestion.wiki — WikiIngester.""" |
|
2
|
|
|
3
|
import tempfile |
|
4
|
from pathlib import Path |
|
5
|
from unittest.mock import MagicMock, patch |
|
6
|
|
|
7
|
import pytest |
|
8
|
|
|
9
|
from navegador.graph.schema import NodeLabel |
|
10
|
from navegador.ingestion.wiki import WikiIngester, _extract_terms |
|
11
|
|
|
12
|
# ── Unit: _extract_terms ────────────────────────────────────────────────────── |
|
13
|
|
|
14
|
class TestExtractTerms: |
|
15
|
def test_extracts_headings(self): |
|
16
|
md = "# Introduction\n## Getting Started\n### Deep Dive\n" |
|
17
|
terms = _extract_terms(md) |
|
18
|
assert "Introduction" in terms |
|
19
|
assert "Getting Started" in terms |
|
20
|
assert "Deep Dive" in terms |
|
21
|
|
|
22
|
def test_extracts_bold_asterisk(self): |
|
23
|
md = "Use **GraphStore** for all persistence." |
|
24
|
terms = _extract_terms(md) |
|
25
|
assert "GraphStore" in terms |
|
26
|
|
|
27
|
def test_extracts_bold_underscore(self): |
|
28
|
md = "The __FalkorDB__ module is required." |
|
29
|
terms = _extract_terms(md) |
|
30
|
assert "FalkorDB" in terms |
|
31
|
|
|
32
|
def test_deduplicates(self): |
|
33
|
md = "# GraphStore\nUse **GraphStore** here too." |
|
34
|
terms = _extract_terms(md) |
|
35
|
assert terms.count("GraphStore") == 1 |
|
36
|
|
|
37
|
def test_empty_markdown(self): |
|
38
|
assert _extract_terms("") == [] |
|
39
|
|
|
40
|
def test_no_headings_no_bold(self): |
|
41
|
terms = _extract_terms("plain text with no markup") |
|
42
|
assert terms == [] |
|
43
|
|
|
44
|
def test_preserves_order(self): |
|
45
|
md = "# Alpha\n# Beta\n**Gamma**" |
|
46
|
terms = _extract_terms(md) |
|
47
|
assert terms == ["Alpha", "Beta", "Gamma"] |
|
48
|
|
|
49
|
|
|
50
|
# ── Unit: ingest_local ──────────────────────────────────────────────────────── |
|
51
|
|
|
52
|
class TestIngestLocal: |
|
53
|
def _make_store(self): |
|
54
|
store = MagicMock() |
|
55
|
store.query.return_value = MagicMock(result_set=[]) |
|
56
|
return store |
|
57
|
|
|
58
|
def test_ingests_markdown_files(self): |
|
59
|
store = self._make_store() |
|
60
|
ingester = WikiIngester(store) |
|
61
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
62
|
(Path(tmpdir) / "home.md").write_text("# Welcome\nThis is home.") |
|
63
|
(Path(tmpdir) / "guide.md").write_text("## Usage\nSome guide.") |
|
64
|
stats = ingester.ingest_local(tmpdir) |
|
65
|
assert stats["pages"] == 2 |
|
66
|
|
|
67
|
def test_skips_non_markdown(self): |
|
68
|
store = self._make_store() |
|
69
|
ingester = WikiIngester(store) |
|
70
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
71
|
(Path(tmpdir) / "readme.md").write_text("# Readme") |
|
72
|
(Path(tmpdir) / "image.png").write_bytes(b"\x89PNG") |
|
73
|
stats = ingester.ingest_local(tmpdir) |
|
74
|
assert stats["pages"] == 1 |
|
75
|
|
|
76
|
def test_raises_if_dir_missing(self): |
|
77
|
store = self._make_store() |
|
78
|
ingester = WikiIngester(store) |
|
79
|
with pytest.raises(FileNotFoundError): |
|
80
|
ingester.ingest_local("/nonexistent/path") |
|
81
|
|
|
82
|
def test_creates_wiki_page_node(self): |
|
83
|
store = self._make_store() |
|
84
|
ingester = WikiIngester(store) |
|
85
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
86
|
(Path(tmpdir) / "arch.md").write_text("# Architecture") |
|
87
|
ingester.ingest_local(tmpdir) |
|
88
|
store.create_node.assert_called_once() |
|
89
|
call_args = store.create_node.call_args |
|
90
|
assert call_args[0][0] == NodeLabel.WikiPage |
|
91
|
props = call_args[0][1] |
|
92
|
assert props["name"] == "arch" |
|
93
|
assert props["source"] == "local" |
|
94
|
|
|
95
|
def test_page_name_normalisation(self): |
|
96
|
store = self._make_store() |
|
97
|
ingester = WikiIngester(store) |
|
98
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
99
|
(Path(tmpdir) / "getting-started.md").write_text("# Hi") |
|
100
|
ingester.ingest_local(tmpdir) |
|
101
|
props = store.create_node.call_args[0][1] |
|
102
|
assert props["name"] == "getting started" |
|
103
|
|
|
104
|
def test_creates_documents_edge_when_term_matches(self): |
|
105
|
store = MagicMock() |
|
106
|
store.query.return_value = MagicMock(result_set=[["Concept", "GraphStore"]]) |
|
107
|
ingester = WikiIngester(store) |
|
108
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
109
|
(Path(tmpdir) / "page.md").write_text("# GraphStore\nSome text.") |
|
110
|
stats = ingester.ingest_local(tmpdir) |
|
111
|
assert stats["links"] >= 1 |
|
112
|
store.create_edge.assert_called() |
|
113
|
|
|
114
|
def test_no_links_when_no_term_match(self): |
|
115
|
store = self._make_store() # query returns [] |
|
116
|
ingester = WikiIngester(store) |
|
117
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
118
|
(Path(tmpdir) / "page.md").write_text("# UnknownTerm\nText.") |
|
119
|
stats = ingester.ingest_local(tmpdir) |
|
120
|
assert stats["links"] == 0 |
|
121
|
store.create_edge.assert_not_called() |
|
122
|
|
|
123
|
def test_content_capped_at_4000_chars(self): |
|
124
|
store = self._make_store() |
|
125
|
ingester = WikiIngester(store) |
|
126
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
127
|
(Path(tmpdir) / "long.md").write_text("x" * 10000) |
|
128
|
ingester.ingest_local(tmpdir) |
|
129
|
props = store.create_node.call_args[0][1] |
|
130
|
assert len(props["content"]) <= 4000 |
|
131
|
|
|
132
|
def test_returns_stats_dict(self): |
|
133
|
store = self._make_store() |
|
134
|
ingester = WikiIngester(store) |
|
135
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
136
|
stats = ingester.ingest_local(tmpdir) |
|
137
|
assert "pages" in stats |
|
138
|
assert "links" in stats |
|
139
|
assert stats["pages"] == 0 |
|
140
|
|
|
141
|
def test_recursive_glob(self): |
|
142
|
store = self._make_store() |
|
143
|
ingester = WikiIngester(store) |
|
144
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
145
|
subdir = Path(tmpdir) / "sub" |
|
146
|
subdir.mkdir() |
|
147
|
(subdir / "nested.md").write_text("# Nested") |
|
148
|
stats = ingester.ingest_local(tmpdir) |
|
149
|
assert stats["pages"] == 1 |
|
150
|
|
|
151
|
|
|
152
|
# ── Unit: _try_link edge-type handling ──────────────────────────────────────── |
|
153
|
|
|
154
|
class TestTryLink: |
|
155
|
def test_handles_invalid_label_gracefully(self): |
|
156
|
store = MagicMock() |
|
157
|
store.query.return_value = MagicMock(result_set=[["InvalidLabel", "foo"]]) |
|
158
|
ingester = WikiIngester(store) |
|
159
|
result = ingester._try_link("page", "foo") |
|
160
|
assert result == 0 |
|
161
|
|
|
162
|
def test_creates_edge_for_valid_label(self): |
|
163
|
store = MagicMock() |
|
164
|
store.query.return_value = MagicMock(result_set=[["Concept", "MyService"]]) |
|
165
|
ingester = WikiIngester(store) |
|
166
|
result = ingester._try_link("wiki page", "MyService") |
|
167
|
assert result == 1 |
|
168
|
store.create_edge.assert_called_once() |
|
169
|
|
|
170
|
def test_returns_zero_on_unknown_label(self): |
|
171
|
store = MagicMock() |
|
172
|
store.query.return_value = MagicMock(result_set=[["UnknownLabel", "node"]]) |
|
173
|
ingester = WikiIngester(store) |
|
174
|
result = ingester._try_link("page", "node") |
|
175
|
assert result == 0 |
|
176
|
|
|
177
|
def test_propagates_store_error(self): |
|
178
|
store = MagicMock() |
|
179
|
store.query.return_value = MagicMock(result_set=[["Concept", "node"]]) |
|
180
|
store.create_edge.side_effect = Exception("DB error") |
|
181
|
ingester = WikiIngester(store) |
|
182
|
with pytest.raises(Exception, match="DB error"): |
|
183
|
ingester._try_link("page", "node") |
|
184
|
|
|
185
|
|
|
186
|
# ── GitHub clone (ingest_github) ────────────────────────────────────────────── |
|
187
|
|
|
188
|
class TestIngestGithub: |
|
189
|
def _make_store(self): |
|
190
|
store = MagicMock() |
|
191
|
store.query.return_value = MagicMock(result_set=[]) |
|
192
|
return store |
|
193
|
|
|
194
|
def test_clones_wiki_and_ingests_local(self): |
|
195
|
store = self._make_store() |
|
196
|
ingester = WikiIngester(store) |
|
197
|
|
|
198
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
199
|
wiki_dir = Path(tmpdir) |
|
200
|
(wiki_dir / "home.md").write_text("# Home\nWelcome.") |
|
201
|
|
|
202
|
mock_result = MagicMock() |
|
203
|
mock_result.returncode = 0 |
|
204
|
|
|
205
|
with patch("subprocess.run", return_value=mock_result) as mock_run, \ |
|
206
|
patch("tempfile.mkdtemp", return_value=str(tmpdir)): |
|
207
|
stats = ingester.ingest_github("owner/repo") |
|
208
|
mock_run.assert_called_once() |
|
209
|
cmd = mock_run.call_args[0][0] |
|
210
|
assert "git" in cmd |
|
211
|
assert "clone" in cmd |
|
212
|
assert "https://github.com/owner/repo.wiki.git" in cmd |
|
213
|
assert stats["pages"] == 1 |
|
214
|
|
|
215
|
def test_returns_empty_on_clone_failure(self): |
|
216
|
store = self._make_store() |
|
217
|
ingester = WikiIngester(store) |
|
218
|
|
|
219
|
mock_result = MagicMock() |
|
220
|
mock_result.returncode = 1 |
|
221
|
mock_result.stderr = "fatal: repository not found" |
|
222
|
|
|
223
|
with patch("subprocess.run", return_value=mock_result): |
|
224
|
stats = ingester.ingest_github("owner/empty-repo") |
|
225
|
assert stats == {"pages": 0, "links": 0} |
|
226
|
|
|
227
|
def test_uses_token_in_url(self): |
|
228
|
store = self._make_store() |
|
229
|
ingester = WikiIngester(store) |
|
230
|
|
|
231
|
mock_result = MagicMock() |
|
232
|
mock_result.returncode = 1 |
|
233
|
mock_result.stderr = "auth error" |
|
234
|
|
|
235
|
with patch("subprocess.run", return_value=mock_result) as mock_run: |
|
236
|
ingester.ingest_github("owner/repo", token="mytoken") |
|
237
|
cmd = mock_run.call_args[0][0] |
|
238
|
assert "[email protected]" in cmd[3] |
|
239
|
|
|
240
|
def test_uses_explicit_clone_dir(self): |
|
241
|
store = self._make_store() |
|
242
|
ingester = WikiIngester(store) |
|
243
|
|
|
244
|
with tempfile.TemporaryDirectory() as tmpdir: |
|
245
|
mock_result = MagicMock() |
|
246
|
mock_result.returncode = 0 |
|
247
|
|
|
248
|
with patch("subprocess.run", return_value=mock_result): |
|
249
|
ingester.ingest_github("owner/repo", clone_dir=tmpdir) |
|
250
|
# Should not crash |
|
251
|
|
|
252
|
|
|
253
|
# ── GitHub API (ingest_github_api) ──────────────────────────────────────────── |
|
254
|
|
|
255
|
class TestIngestGithubApi: |
|
256
|
def _make_store(self): |
|
257
|
store = MagicMock() |
|
258
|
store.query.return_value = MagicMock(result_set=[]) |
|
259
|
return store |
|
260
|
|
|
261
|
def test_fetches_readme_and_ingests(self): |
|
262
|
store = self._make_store() |
|
263
|
ingester = WikiIngester(store) |
|
264
|
|
|
265
|
import base64 |
|
266
|
import json as _json |
|
267
|
readme_content = base64.b64encode(b"# README\nSome content").decode() |
|
268
|
mock_response_data = { |
|
269
|
"content": readme_content, |
|
270
|
"html_url": "https://github.com/owner/repo/blob/main/README.md", |
|
271
|
} |
|
272
|
|
|
273
|
mock_resp = MagicMock() |
|
274
|
mock_resp.read.return_value = _json.dumps(mock_response_data).encode() |
|
275
|
mock_resp.__enter__ = lambda s: s |
|
276
|
mock_resp.__exit__ = MagicMock(return_value=False) |
|
277
|
|
|
278
|
with patch("urllib.request.urlopen", return_value=mock_resp): |
|
279
|
stats = ingester.ingest_github_api("owner/repo") |
|
280
|
assert stats["pages"] >= 1 |
|
281
|
|
|
282
|
def test_skips_missing_files_gracefully(self): |
|
283
|
store = self._make_store() |
|
284
|
ingester = WikiIngester(store) |
|
285
|
|
|
286
|
with patch("urllib.request.urlopen", side_effect=Exception("404")): |
|
287
|
stats = ingester.ingest_github_api("owner/repo") |
|
288
|
assert stats == {"pages": 0, "links": 0} |
|
289
|
|
|
290
|
def test_uses_auth_header_with_token(self): |
|
291
|
store = self._make_store() |
|
292
|
ingester = WikiIngester(store) |
|
293
|
|
|
294
|
with patch("urllib.request.urlopen", side_effect=Exception("skip")), \ |
|
295
|
patch("urllib.request.Request") as mock_req: |
|
296
|
ingester.ingest_github_api("owner/repo", token="mytoken") |
|
297
|
# Just verify no crash and token path was exercised |
|
298
|
assert mock_req.called |
|
299
|
|