personal memory agent
feat: entity enrichment observations from Kindle importer
Book entities get "By {author}" and engagement count observations,
Person (author) entities get "Author of {title}" observations.
Per-book earliest clip date used in observation text. Engagement
counts exclude bookmarks (count clip_type == "highlight" explicitly).
+246
-2
+212
tests/test_kindle_importer.py
+212
tests/test_kindle_importer.py
···
7
7
import tempfile
8
8
from pathlib import Path
9
9
10
+
from think.entities.observations import load_observations
10
11
from think.importers.kindle import KindleImporter, _parse_block, _parse_date
11
12
12
13
importer = KindleImporter()
···
203
204
finally:
204
205
os.unlink(f.name)
205
206
os.environ.pop("JOURNAL_PATH", None)
207
+
208
+
209
+
def test_observations_author_of(tmp_path, monkeypatch):
210
+
content = _make_clippings_file([_make_clipping()])
211
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
212
+
f.write(content)
213
+
f.flush()
214
+
try:
215
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
216
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
217
+
author_obs = load_observations("test.kindle", "Author Name")
218
+
author_contents = [o["content"] for o in author_obs]
219
+
assert "Author of Test Book (via Kindle, 2025-03-15)" in author_contents
220
+
finally:
221
+
os.unlink(f.name)
222
+
223
+
224
+
def test_observations_by_author(tmp_path, monkeypatch):
225
+
content = _make_clippings_file([_make_clipping()])
226
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
227
+
f.write(content)
228
+
f.flush()
229
+
try:
230
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
231
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
232
+
book_obs = load_observations("test.kindle", "Test Book")
233
+
book_contents = [o["content"] for o in book_obs]
234
+
assert "By Author Name (via Kindle, 2025-03-15)" in book_contents
235
+
finally:
236
+
os.unlink(f.name)
237
+
238
+
239
+
def test_observations_engagement(tmp_path, monkeypatch):
240
+
content = _make_clippings_file(
241
+
[
242
+
_make_clipping(
243
+
meta="- Your Highlight on page 42 | location 100-101 | Added on Saturday, March 15, 2025 10:30:00 AM",
244
+
),
245
+
_make_clipping(
246
+
meta="- Your Highlight on page 43 | location 102-103 | Added on Saturday, March 15, 2025 10:31:00 AM",
247
+
content="Second highlight.",
248
+
),
249
+
_make_clipping(
250
+
meta="- Your Note on page 44 | location 104 | Added on Saturday, March 15, 2025 10:32:00 AM",
251
+
content="A note.",
252
+
),
253
+
]
254
+
)
255
+
highlights_only = _make_clippings_file(
256
+
[
257
+
_make_clipping(
258
+
meta="- Your Highlight on page 42 | location 100-101 | Added on Saturday, March 15, 2025 10:30:00 AM",
259
+
),
260
+
_make_clipping(
261
+
meta="- Your Highlight on page 43 | location 102-103 | Added on Saturday, March 15, 2025 10:31:00 AM",
262
+
content="Second highlight.",
263
+
),
264
+
]
265
+
)
266
+
267
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
268
+
f.write(content)
269
+
f.flush()
270
+
try:
271
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
272
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
273
+
book_obs = load_observations("test.kindle", "Test Book")
274
+
book_contents = [o["content"] for o in book_obs]
275
+
assert "2 highlights, 1 notes (via Kindle, 2025-03-15)" in book_contents
276
+
finally:
277
+
os.unlink(f.name)
278
+
279
+
second_tmp_path = tmp_path / "highlights_only"
280
+
second_tmp_path.mkdir()
281
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
282
+
f.write(highlights_only)
283
+
f.flush()
284
+
try:
285
+
monkeypatch.setenv("JOURNAL_PATH", str(second_tmp_path))
286
+
importer.process(Path(f.name), second_tmp_path, facet="test.kindle")
287
+
book_obs = load_observations("test.kindle", "Test Book")
288
+
book_contents = [o["content"] for o in book_obs]
289
+
assert "2 highlights (via Kindle, 2025-03-15)" in book_contents
290
+
finally:
291
+
os.unlink(f.name)
292
+
293
+
294
+
def test_observations_multi_book_author(tmp_path, monkeypatch):
295
+
content = _make_clippings_file(
296
+
[
297
+
_make_clipping(title="Test Book (Author Name)"),
298
+
_make_clipping(title="Second Book (Author Name)"),
299
+
]
300
+
)
301
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
302
+
f.write(content)
303
+
f.flush()
304
+
try:
305
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
306
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
307
+
author_obs = load_observations("test.kindle", "Author Name")
308
+
author_contents = [o["content"] for o in author_obs]
309
+
assert "Author of Test Book (via Kindle, 2025-03-15)" in author_contents
310
+
assert "Author of Second Book (via Kindle, 2025-03-15)" in author_contents
311
+
finally:
312
+
os.unlink(f.name)
313
+
314
+
315
+
def test_observations_no_author(tmp_path, monkeypatch):
316
+
content = _make_clippings_file([_make_clipping(title="Title Without Author")])
317
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
318
+
f.write(content)
319
+
f.flush()
320
+
try:
321
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
322
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
323
+
book_obs = load_observations("test.kindle", "Title Without Author")
324
+
book_contents = [o["content"] for o in book_obs]
325
+
assert not any(c.startswith("By ") for c in book_contents)
326
+
author_entities_dir = tmp_path / "facets" / "test.kindle" / "entities"
327
+
if author_entities_dir.exists():
328
+
entity_names = {
329
+
entity_dir.name
330
+
for entity_dir in author_entities_dir.iterdir()
331
+
if entity_dir.is_dir()
332
+
}
333
+
assert "author_name" not in entity_names
334
+
finally:
335
+
os.unlink(f.name)
336
+
337
+
338
+
def test_observations_dedup_on_reimport(tmp_path, monkeypatch):
339
+
content = _make_clippings_file([_make_clipping()])
340
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
341
+
f.write(content)
342
+
f.flush()
343
+
try:
344
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
345
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
346
+
first = load_observations("test.kindle", "Test Book")
347
+
first_by_author = [
348
+
o
349
+
for o in first
350
+
if o["content"] == "By Author Name (via Kindle, 2025-03-15)"
351
+
]
352
+
assert len(first_by_author) == 1
353
+
354
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
355
+
second = load_observations("test.kindle", "Test Book")
356
+
second_by_author = [
357
+
o
358
+
for o in second
359
+
if o["content"] == "By Author Name (via Kindle, 2025-03-15)"
360
+
]
361
+
assert len(second_by_author) == 1
362
+
finally:
363
+
os.unlink(f.name)
364
+
365
+
366
+
def test_observations_engagement_notes_only(tmp_path, monkeypatch):
367
+
"""Notes-only book gets notes count without highlights."""
368
+
content = _make_clippings_file(
369
+
[
370
+
_make_clipping(
371
+
meta="- Your Note on page 10 | Added on Saturday, March 15, 2025 10:30:00 AM",
372
+
content="A note.",
373
+
),
374
+
_make_clipping(
375
+
meta="- Your Note on page 11 | Added on Saturday, March 15, 2025 10:31:00 AM",
376
+
content="Another note.",
377
+
),
378
+
]
379
+
)
380
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
381
+
f.write(content)
382
+
f.flush()
383
+
try:
384
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
385
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
386
+
book_obs = load_observations("test.kindle", "Test Book")
387
+
book_contents = [o["content"] for o in book_obs]
388
+
assert "2 notes (via Kindle, 2025-03-15)" in book_contents
389
+
# No highlights count should appear
390
+
assert not any("highlights" in c for c in book_contents)
391
+
finally:
392
+
os.unlink(f.name)
393
+
394
+
395
+
def test_observations_engagement_excludes_bookmarks(tmp_path, monkeypatch):
396
+
content = _make_clippings_file(
397
+
[
398
+
_make_clipping(
399
+
meta="- Your Highlight on page 42 | location 100-101 | Added on Saturday, March 15, 2025 10:30:00 AM",
400
+
),
401
+
_make_clipping(
402
+
meta="- Your Bookmark on page 43 | location 102-103 | Added on Saturday, March 15, 2025 10:31:00 AM",
403
+
content="Saved bookmark.",
404
+
),
405
+
]
406
+
)
407
+
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False) as f:
408
+
f.write(content)
409
+
f.flush()
410
+
try:
411
+
monkeypatch.setenv("JOURNAL_PATH", str(tmp_path))
412
+
importer.process(Path(f.name), tmp_path, facet="test.kindle")
413
+
book_obs = load_observations("test.kindle", "Test Book")
414
+
book_contents = [o["content"] for o in book_obs]
415
+
assert "1 highlights (via Kindle, 2025-03-15)" in book_contents
416
+
finally:
417
+
os.unlink(f.name)
206
418
207
419
208
420
# --- Registry test ---
+34
-2
think/importers/kindle.py
+34
-2
think/importers/kindle.py
···
331
331
)
332
332
entry_segment_map = {idx: segment for idx, segment in enumerate(item_segments)}
333
333
manifest_entries: list[dict] = []
334
+
book_obs: dict[str, list[str]] = {}
335
+
author_obs: dict[str, list[str]] = {}
334
336
for book_idx, (book_title, indices) in enumerate(sorted(books_map.items())):
335
337
book_entries = [entries[i] for i in indices]
336
338
author = book_entries[0].get("author", "")
···
359
361
],
360
362
}
361
363
)
364
+
obs_date = first_dt.strftime("%Y-%m-%d")
365
+
highlight_only = sum(
366
+
1 for entry in book_entries if entry.get("clip_type") == "highlight"
367
+
)
368
+
observations: list[str] = []
369
+
if author:
370
+
observations.append(f"By {author} (via Kindle, {obs_date})")
371
+
if highlight_only > 0 and note_count > 0:
372
+
observations.append(
373
+
f"{highlight_only} highlights, {note_count} notes "
374
+
f"(via Kindle, {obs_date})"
375
+
)
376
+
elif highlight_only > 0:
377
+
observations.append(
378
+
f"{highlight_only} highlights (via Kindle, {obs_date})"
379
+
)
380
+
elif note_count > 0:
381
+
observations.append(f"{note_count} notes (via Kindle, {obs_date})")
382
+
if observations:
383
+
book_obs[book_title] = observations
384
+
if author:
385
+
author_obs.setdefault(author, []).append(
386
+
f"Author of {book_title} (via Kindle, {obs_date})"
387
+
)
362
388
write_content_manifest(import_id, manifest_entries)
363
389
364
390
segment_days = {day for day, _ in segments}
···
371
397
)
372
398
entity_defs: list[dict] = []
373
399
for book in sorted(books):
374
-
entity_defs.append({"name": book, "type": "Book"})
400
+
d: dict = {"name": book, "type": "Book"}
401
+
if book in book_obs:
402
+
d["observations"] = book_obs[book]
403
+
entity_defs.append(d)
375
404
for author in sorted(authors):
376
405
if author:
377
-
entity_defs.append({"name": author, "type": "Person"})
406
+
d = {"name": author, "type": "Person"}
407
+
if author in author_obs:
408
+
d["observations"] = author_obs[author]
409
+
entity_defs.append(d)
378
410
379
411
resolved = seed_entities(facet, earliest_day, entity_defs)
380
412
entities_seeded = len(resolved)