personal memory agent
Merge branch 'hopper-6hwyzh6f-transcripts-resilience'
+39
-22
+26
-19
apps/transcripts/routes.py
+26
-19
apps/transcripts/routes.py
···
6
6
from __future__ import annotations
7
7
8
8
import json
9
+
import logging
9
10
import os
10
11
import re
11
12
import shutil
···
35
36
from think.utils import day_dirs, day_path, segment_path
36
37
from think.utils import segment_key as validate_segment_key
37
38
38
-
# Regex for HHMMSS time format validation
39
-
TIME_RE = re.compile(r"\d{6}")
39
+
logger = logging.getLogger(__name__)
40
+
41
+
# Regex for YYYYMM month format validation
42
+
MONTH_RE = re.compile(r"\d{6}")
40
43
41
44
transcripts_bp = Blueprint(
42
45
"app:transcripts",
···
49
52
def index() -> Any:
50
53
"""Redirect to the most recent day with segments, falling back to today."""
51
54
today = date.today().strftime("%Y%m%d")
52
-
if cluster_segments(today):
53
-
return redirect(url_for("app:transcripts.transcripts_day", day=today))
54
55
for day in sorted(day_dirs().keys(), reverse=True):
55
56
if cluster_segments(day):
56
57
return redirect(url_for("app:transcripts.transcripts_day", day=day))
···
61
62
def transcripts_day(day: str) -> str:
62
63
"""Render transcript viewer for a specific day."""
63
64
if not DATE_RE.fullmatch(day):
64
-
return "", 404
65
+
return error_response("Day not found", 404)
65
66
66
67
title = format_date(day)
67
68
···
72
73
def transcript_ranges(day: str) -> Any:
73
74
"""Return available transcript ranges for a day."""
74
75
if not DATE_RE.fullmatch(day):
75
-
return "", 404
76
+
return error_response("Day not found", 404)
76
77
77
78
audio_ranges, screen_ranges = cluster_scan(day)
78
79
return jsonify({"audio": audio_ranges, "screen": screen_ranges})
···
85
86
Returns list of segments with their content types for the segment selector UI.
86
87
"""
87
88
if not DATE_RE.fullmatch(day):
88
-
return "", 404
89
+
return error_response("Day not found", 404)
89
90
90
91
segments = cluster_segments(day)
91
92
return jsonify({"segments": segments})
···
95
96
def transcript_day_data(day: str) -> Any:
96
97
"""Return combined ranges and segments for a day in a single response."""
97
98
if not DATE_RE.fullmatch(day):
98
-
return "", 404
99
+
return error_response("Day not found", 404)
99
100
100
101
audio_ranges, screen_ranges, segments = scan_day(day)
101
102
return jsonify({"audio": audio_ranges, "screen": screen_ranges, "segments": segments})
···
105
106
def serve_file(day: str, encoded_path: str) -> Any:
106
107
"""Serve actual media files for embedding."""
107
108
if not DATE_RE.fullmatch(day):
108
-
return "", 404
109
+
return error_response("Day not found", 404)
109
110
110
111
try:
111
112
rel_path = encoded_path.replace("__", "/")
···
113
114
114
115
day_dir = str(day_path(day, create=False))
115
116
if not os.path.commonpath([full_path, day_dir]) == day_dir:
116
-
return "", 403
117
+
return error_response("Invalid file path", 403)
117
118
118
119
if not os.path.isfile(full_path):
119
-
return "", 404
120
+
return error_response("File not found", 404)
120
121
121
122
return send_file(full_path)
122
123
123
124
except Exception:
124
-
return "", 404
125
+
return error_response("Failed to serve file", 404)
125
126
126
127
127
128
@transcripts_bp.route("/api/stats/<month>")
···
135
136
JSON dict mapping day (YYYYMMDD) to transcript range count.
136
137
Transcripts app is not facet-aware, so returns simple {day: count} mapping.
137
138
"""
138
-
if not TIME_RE.fullmatch(month):
139
-
return jsonify({"error": "Invalid month format, expected YYYYMM"}), 400
139
+
if not MONTH_RE.fullmatch(month):
140
+
return error_response("Invalid month format", 400)
140
141
141
142
stats: dict[str, int] = {}
142
143
···
203
204
- media_sizes: dict with audio/screen byte counts for raw media files
204
205
"""
205
206
if not DATE_RE.fullmatch(day):
206
-
return "", 404
207
+
return error_response("Invalid day format", 404)
207
208
208
209
if not validate_segment_key(segment_key):
209
-
return "", 404
210
+
return error_response("Invalid segment key format", 404)
210
211
211
212
segment_dir = str(segment_path(day, segment_key, stream))
212
213
if not os.path.isdir(segment_dir):
213
-
return "", 404
214
+
return error_response("Segment directory not found", 404)
214
215
215
216
chunks: list[dict] = []
216
217
audio_file_url = None
···
218
219
media_sizes: dict[str, int] = {"audio": 0, "screen": 0}
219
220
has_raw_reference = False
220
221
has_raw_file = False
222
+
warnings = 0
221
223
222
224
# Load speaker labels if available.
223
225
speaker_labels_path = Path(segment_dir) / "agents" / "speaker_labels.json"
···
306
308
if speaker_label:
307
309
chunk_data["speaker_label"] = speaker_label
308
310
chunks.append(chunk_data)
309
-
except Exception:
311
+
except Exception as e:
312
+
logger.warning("Failed to parse audio segment %s: %s", audio_path, e)
313
+
warnings += 1
310
314
continue
311
315
312
316
# Process screen files and collect video URLs for client-side decoding
···
396
400
"basic": is_basic,
397
401
}
398
402
)
399
-
except Exception:
403
+
except Exception as e:
404
+
logger.warning("Failed to parse screen segment %s: %s", screen_path, e)
405
+
warnings += 1
400
406
continue
401
407
402
408
# Sort all chunks by timestamp
···
427
433
"cost": cost_data["cost"],
428
434
"media_sizes": media_sizes,
429
435
"media_purged": media_purged,
436
+
"warnings": warnings,
430
437
}
431
438
)
432
439
-2
convey/root.py
-2
convey/root.py
···
117
117
def app_today() -> Any:
118
118
"""Redirect /app/today to the most recent day with journal data."""
119
119
today = date.today().strftime("%Y%m%d")
120
-
if cluster_segments(today):
121
-
return redirect(url_for("app:transcripts.transcripts_day", day=today))
122
120
for day in sorted(day_dirs().keys(), reverse=True):
123
121
if cluster_segments(day):
124
122
return redirect(url_for("app:transcripts.transcripts_day", day=day))
+1
tests/baselines/api/transcripts/segment-detail.json
+1
tests/baselines/api/transcripts/segment-detail.json
tests/fixtures/journal/indexer/journal.sqlite
tests/fixtures/journal/indexer/journal.sqlite
This is a binary file and will not be displayed.
+8
tests/test_think_utils.py
+8
tests/test_think_utils.py
···
8
8
import os
9
9
import sys
10
10
import tempfile
11
+
from datetime import time
11
12
from pathlib import Path
12
13
13
14
import pytest
···
17
18
DEFAULT_STREAM,
18
19
day_from_path,
19
20
iter_segments,
21
+
segment_parse,
20
22
segment_key,
21
23
setup_cli,
22
24
)
···
641
643
assert segment_key("prefix 143022_300 suffix") == "143022_300"
642
644
# Multiple potential matches (should match first)
643
645
assert segment_key("143022_300 and 150000_600") == "143022_300"
646
+
647
+
648
+
def test_segment_parse_clamps_midnight_crossing():
649
+
"""Test segment_parse clamps end time when a segment crosses midnight."""
650
+
assert segment_parse("235900_300") == (time(23, 59, 0), time(23, 59, 59))
651
+
assert segment_parse("143022_300") == (time(14, 30, 22), time(14, 35, 22))
644
652
645
653
646
654
class TestSetupCliConfigEnv:
+4
-1
think/utils.py
+4
-1
think/utils.py
···
413
413
# Compute end time by adding duration
414
414
start_dt = datetime.combine(datetime.today(), start_time)
415
415
end_dt = start_dt + timedelta(seconds=length_seconds)
416
-
end_time = end_dt.time()
416
+
if end_dt.date() > start_dt.date():
417
+
end_time = time(23, 59, 59)
418
+
else:
419
+
end_time = end_dt.time()
417
420
return (start_time, end_time)
418
421
except ValueError:
419
422
return (None, None)