codebee 0.1.68 → 0.1.70

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (133) hide show
  1. package/CHANGELOG.md +580 -543
  2. package/LICENSE +21 -21
  3. package/README.md +11 -0
  4. package/app/browser-final-1440.png +0 -0
  5. package/app/browser-final-390.png +0 -0
  6. package/app/browser-google-fallback.png +0 -0
  7. package/app/browser-google-test.png +0 -0
  8. package/app/browser-workbench-1440.png +0 -0
  9. package/app/browser-workbench-390.png +0 -0
  10. package/app/browser-workspace-google.png +0 -0
  11. package/app/browser-workspace-smoke.png +0 -0
  12. package/app/browser-workspace.png +0 -0
  13. package/app/core/acceptance.py +100 -100
  14. package/app/core/aiflavor.py +197 -197
  15. package/app/core/artifacts.py +148 -148
  16. package/app/core/assets.py +110 -110
  17. package/app/core/attachments.py +623 -623
  18. package/app/core/audit.py +30 -0
  19. package/app/core/automation.py +853 -758
  20. package/app/core/backup.py +470 -470
  21. package/app/core/beekeeper.py +183 -183
  22. package/app/core/benchstore.py +95 -0
  23. package/app/core/board.py +209 -209
  24. package/app/core/bookmeta.py +935 -930
  25. package/app/core/bookmeta_catalog.py +180 -180
  26. package/app/core/branching.py +137 -137
  27. package/app/core/builtin_agent.py +1673 -1234
  28. package/app/core/capability.py +144 -144
  29. package/app/core/catalog.py +7 -1
  30. package/app/core/chaptersafe.py +32 -32
  31. package/app/core/cleanup.py +303 -303
  32. package/app/core/compaction.py +201 -201
  33. package/app/core/covergen.py +356 -356
  34. package/app/core/defectretro.py +43 -43
  35. package/app/core/diagnostics.py +114 -114
  36. package/app/core/dispatch.py +190 -173
  37. package/app/core/dispatch_log.py +128 -128
  38. package/app/core/env_scrub.py +85 -85
  39. package/app/core/error_codes.py +33 -24
  40. package/app/core/errorlog.py +137 -137
  41. package/app/core/evalbench.py +720 -0
  42. package/app/core/evaluation.py +174 -174
  43. package/app/core/failure_notify.py +290 -290
  44. package/app/core/flows.py +228 -3
  45. package/app/core/goal_service.py +167 -167
  46. package/app/core/health.py +374 -374
  47. package/app/core/history.py +38 -38
  48. package/app/core/hooks.py +162 -162
  49. package/app/core/jobs.py +51 -30
  50. package/app/core/manager.py +2143 -2122
  51. package/app/core/market.py +362 -362
  52. package/app/core/market_remote.py +10 -0
  53. package/app/core/mcp_client.py +287 -0
  54. package/app/core/mocks.py +64 -64
  55. package/app/core/modelhub.py +193 -10
  56. package/app/core/notify.py +201 -106
  57. package/app/core/novel_quality.py +189 -0
  58. package/app/core/operations.py +228 -228
  59. package/app/core/paihang.py +138 -138
  60. package/app/core/paths.py +6 -2
  61. package/app/core/pipeline.py +426 -28
  62. package/app/core/planner.py +243 -22
  63. package/app/core/plugins.py +525 -0
  64. package/app/core/portguard.py +111 -111
  65. package/app/core/preview.py +293 -293
  66. package/app/core/publish/__init__.py +7 -7
  67. package/app/core/publish/auto.py +367 -367
  68. package/app/core/publish/browser.py +655 -539
  69. package/app/core/publish/fanqie.py +165 -159
  70. package/app/core/publish/flow.py +589 -440
  71. package/app/core/publish/flows-fanqie-calibrated.json +336 -332
  72. package/app/core/publish/ledger.py +229 -203
  73. package/app/core/publish/manager.py +824 -692
  74. package/app/core/publish/qimao.py +146 -141
  75. package/app/core/publish/ws.py +139 -139
  76. package/app/core/redact.py +38 -38
  77. package/app/core/remotesync.py +121 -0
  78. package/app/core/repeat_guard.py +129 -129
  79. package/app/core/router.py +398 -398
  80. package/app/core/runner.py +2142 -2067
  81. package/app/core/selfupdate.py +384 -384
  82. package/app/core/session_log.py +161 -161
  83. package/app/core/sessions.py +312 -312
  84. package/app/core/settings.py +186 -149
  85. package/app/core/settings_schema.py +270 -262
  86. package/app/core/share_page.py +114 -114
  87. package/app/core/skill_scan.py +122 -122
  88. package/app/core/skillpacks/fanqie-novel.md +80 -80
  89. package/app/core/skillpacks/market/character-bible.md +66 -66
  90. package/app/core/skillpacks/market/code-risk-checklist.md +58 -58
  91. package/app/core/skillpacks/market/git-workflow.md +57 -57
  92. package/app/core/skillpacks/market/release-notes.md +72 -72
  93. package/app/core/skillpacks/market/weekly-report.md +71 -71
  94. package/app/core/skillpacks/market/worldview-consistency.md +70 -70
  95. package/app/core/step_runner.py +132 -132
  96. package/app/core/store.py +104 -82
  97. package/app/core/task_compile.py +199 -199
  98. package/app/core/telemetry.py +292 -292
  99. package/app/core/tlsctx.py +64 -64
  100. package/app/core/token_meter.py +147 -147
  101. package/app/core/tracing.py +144 -144
  102. package/app/core/upstream.py +41 -41
  103. package/app/core/usage.py +1035 -862
  104. package/app/core/volumes.py +389 -389
  105. package/app/core/wxdigest.py +858 -858
  106. package/app/core/zentao.py +2614 -2465
  107. package/app/current-browser.png +0 -0
  108. package/app/main.py +3245 -2981
  109. package/app/mcp_server.py +175 -0
  110. package/app/pet.py +82 -21
  111. package/app/pick_dialog.py +110 -110
  112. package/app/plugins/code-review-helper/.codex-plugin/plugin.json +18 -0
  113. package/app/plugins/code-review-helper/skills/review.md +16 -0
  114. package/app/ui/app.js +15706 -14399
  115. package/app/ui/artdiff.js +74 -74
  116. package/app/ui/board.html +162 -162
  117. package/app/ui/board.js +305 -305
  118. package/app/ui/i18n.js +3012 -2770
  119. package/app/ui/icons/bee.svg +79 -79
  120. package/app/ui/index.html +277 -13
  121. package/app/ui/manifest.json +16 -16
  122. package/app/ui/qrcode.js +2297 -2297
  123. package/app/ui/style.css +802 -16
  124. package/bin/tutti.js +147 -147
  125. package/package.json +47 -46
  126. package/tools/wx_reader.py +261 -261
  127. package/app/.playwright-cli/console-2026-09-23T08-26-03-635Z.log +0 -88
  128. package/app/.playwright-cli/page-2026-09-23T08-26-03-845Z.yml +0 -126
  129. package/app/.playwright-cli/page-2026-09-23T08-29-44-203Z.png +0 -0
  130. package/app/.playwright-cli/page-2026-09-23T08-35-42-046Z.yml +0 -210
  131. package/app/.playwright-cli/page-2026-09-23T08-36-52-858Z.png +0 -0
  132. package/app/.playwright-cli/page-2026-09-23T08-37-33-938Z.png +0 -0
  133. package/app/.playwright-cli/run-detail-after.png +0 -0
@@ -0,0 +1,189 @@
1
+ # -*- coding: utf-8 -*-
2
+ """Deterministic quality checks for signing-oriented Chinese web fiction.
3
+
4
+ The LLM remains responsible for literary judgment. This module supplies
5
+ stable rubric defaults and small text signals so reviewers can cite evidence
6
+ for the common rejection reasons instead of relying on vague impressions.
7
+ """
8
+ from __future__ import annotations
9
+
10
+ import re
11
+
12
+
13
+ SIGNING_RUBRIC = [
14
+ "开篇吸引力",
15
+ "情节推进",
16
+ "文风统一",
17
+ "情感细腻度",
18
+ "节奏控制",
19
+ ]
20
+
21
+ OPENING_DIM = "开篇吸引力"
22
+ SIGNING_CHECKPOINT_WORDS = 20000 # 平台口径:前 2 万字用于签约评估
23
+
24
+ _SENTENCE_RE = re.compile(r"[^。!?!?;;\n]+[。!?!?;;]?")
25
+ _DIALOGUE_RE = re.compile(r"[“‘\"「『](.*?)[”’\"」』]", re.S)
26
+ _EMOTION_RE = re.compile(r"哭|泪|笑|怒|怕|颤|抖|咬|攥|握紧|心跳|呼吸|喉咙|指甲|沉默")
27
+
28
+
29
+ def rubric_for(task):
30
+ """Return the configured rubric, or the signing rubric for serial novels."""
31
+ task = task if isinstance(task, dict) else {}
32
+ raw = task.get("rubric")
33
+ if isinstance(raw, str):
34
+ raw = [x.strip() for x in raw.replace(",", ",").split(",") if x.strip()]
35
+ if isinstance(raw, (list, tuple)):
36
+ values = [str(x).strip() for x in raw if str(x).strip()]
37
+ if values:
38
+ return values[:8]
39
+ if str(task.get("type") or "").lower() == "serial_novel":
40
+ return list(SIGNING_RUBRIC)
41
+ return []
42
+
43
+
44
+ def opening_requirements(chapter, target_words=None):
45
+ """Return concise, chapter-aware acceptance requirements for prompts."""
46
+ try:
47
+ chapter = int(chapter)
48
+ except (TypeError, ValueError):
49
+ chapter = 1
50
+ if chapter == 1:
51
+ try:
52
+ short = int(target_words or 0) < 1200
53
+ except (TypeError, ValueError):
54
+ short = False
55
+ core = ("前 1200 字" if short else "前 2000 字")
56
+ return ("黄金一章门禁:前 100 字让读者知道主角是谁;前 200 字出现异常、欲望或冲突;"
57
+ "前 600 字把困境具体化;%s 展示核心卖点;结尾必须留下可兑现的钩子。" % core
58
+ + "禁止用世界观、环境或人物关系长铺垫开场。")
59
+ return "本章门禁:开头尽快承接上一章并制造新变化;结尾必须留下下一步行动或可兑现的章末钩子。"
60
+
61
+
62
+ def signal_summary(text, chapter=None):
63
+ """Extract explainable text signals; no claim is made that they are scores."""
64
+ text = str(text or "").replace("\r\n", "\n").strip()
65
+ paragraphs = [p.strip() for p in re.split(r"\n\s*\n", text) if p.strip()]
66
+ sentences = [s.strip() for s in _SENTENCE_RE.findall(text) if s.strip()]
67
+ sentence_lengths = [len(re.sub(r"\s+", "", s)) for s in sentences]
68
+ dialogue_chars = sum(len(x) for x in _DIALOGUE_RE.findall(text))
69
+ emotion_hits = len(_EMOTION_RE.findall(text))
70
+ chars = len(re.sub(r"\s+", "", text))
71
+ avg = round(sum(sentence_lengths) / len(sentence_lengths), 1) if sentence_lengths else 0.0
72
+ long_sentences = sum(1 for n in sentence_lengths if n >= 55)
73
+ hints = []
74
+ if chapter == 1 and chars and chars < 600:
75
+ hints.append("首章正文不足 600 字,无法证明前 600 字已进入具体困境")
76
+ if avg >= 35 or long_sentences:
77
+ hints.append("存在偏长句,复查长定语、重复说明与动作拆分")
78
+ if len(paragraphs) >= 4 and chars / len(paragraphs) > 220:
79
+ hints.append("段落偏长,复查手机阅读下的节奏与信息密度")
80
+ if text and dialogue_chars == 0:
81
+ hints.append("缺少对白证据,复查是否用场景/行动推进冲突")
82
+ if text and emotion_hits == 0:
83
+ hints.append("缺少可见的情绪动作或生理反应,复查复杂情感是否写得过于概括")
84
+ return {
85
+ "chapter": chapter,
86
+ "chars": chars,
87
+ "paragraphs": len(paragraphs),
88
+ "sentences": len(sentences),
89
+ "avg_sentence_chars": avg,
90
+ "long_sentences": long_sentences,
91
+ "dialogue_chars": dialogue_chars,
92
+ "emotion_action_hits": emotion_hits,
93
+ "hints": hints[:6],
94
+ }
95
+
96
+
97
+ def format_signal_summary(summary):
98
+ """Render signal_summary output as a compact prompt block."""
99
+ if not isinstance(summary, dict):
100
+ return ""
101
+ parts = [
102
+ "文本信号(仅作证据,不替代文学判断):约 %s 字;%s 段;%s 句;平均句长 %s 字;长句 %s;对白 %s 字;情绪动作线索 %s 处。"
103
+ % (summary.get("chars", 0), summary.get("paragraphs", 0),
104
+ summary.get("sentences", 0), summary.get("avg_sentence_chars", 0),
105
+ summary.get("long_sentences", 0), summary.get("dialogue_chars", 0),
106
+ summary.get("emotion_action_hits", 0))]
107
+ hints = summary.get("hints") or []
108
+ if hints:
109
+ parts.append("确定性复查提示:" + ";".join(str(x) for x in hints))
110
+ return "\n".join(parts)
111
+
112
+
113
+ def style_anchor_block(anchor):
114
+ """Render the book-level style anchor as a prompt block; empty when absent."""
115
+ anchor = str(anchor or "").strip()
116
+ if not anchor:
117
+ return ""
118
+ return ("## 全书文风锚(大纲裁定,所有章节起草/修订/评审必须遵守;"
119
+ "与其冲突处以本锚为准)\n%s\n" % anchor[:400])
120
+
121
+
122
+ def fingerprint(summary):
123
+ """Normalize a signal_summary into per-1000-char style metrics."""
124
+ if not isinstance(summary, dict):
125
+ return {}
126
+ chars = max(1, int(summary.get("chars") or 0))
127
+ return {
128
+ "avg_sentence_chars": float(summary.get("avg_sentence_chars") or 0.0),
129
+ "dialogue_per_k": round(1000.0 * float(summary.get("dialogue_chars") or 0) / chars, 1),
130
+ "emotion_per_k": round(1000.0 * float(summary.get("emotion_action_hits") or 0) / chars, 1),
131
+ }
132
+
133
+
134
+ def _median(values):
135
+ xs = sorted(float(v) for v in values if isinstance(v, (int, float)))
136
+ if not xs:
137
+ return 0.0
138
+ mid = len(xs) // 2
139
+ return xs[mid] if len(xs) % 2 else round((xs[mid - 1] + xs[mid]) / 2.0, 1)
140
+
141
+
142
+ def style_deviation_notes(baselines, summary, min_baseline=2):
143
+ """Cross-chapter style drift warnings from deterministic fingerprints.
144
+
145
+ baselines: signal_summary dicts of the book's previous chapters. Needs at
146
+ least ``min_baseline`` chapters before speaking (开篇几章没有稳定基线);
147
+ all notes are evidence hints for reviewers/revisers, never scores.
148
+ """
149
+ notes = []
150
+ base = [fingerprint(b) for b in (baselines or []) if isinstance(b, dict)]
151
+ base = [b for b in base if b]
152
+ if len(base) < min_baseline:
153
+ return notes
154
+ cur = fingerprint(summary)
155
+ if not cur:
156
+ return notes
157
+ avg_base = _median([b["avg_sentence_chars"] for b in base])
158
+ if avg_base >= 12 and cur["avg_sentence_chars"] > avg_base * 1.45:
159
+ notes.append("本章平均句长 %s 字,此前各章中位 %s 字——句式明显变长,"
160
+ "复查长定语、说明性插叙与节奏拖沓"
161
+ % (cur["avg_sentence_chars"], avg_base))
162
+ elif avg_base >= 12 and cur["avg_sentence_chars"] < avg_base * 0.62:
163
+ notes.append("本章平均句长 %s 字,此前各章中位 %s 字——句式明显变碎,"
164
+ "复查是否为凑字数拆散动作与因果"
165
+ % (cur["avg_sentence_chars"], avg_base))
166
+ dial_base = _median([b["dialogue_per_k"] for b in base])
167
+ if dial_base >= 40 and cur["dialogue_per_k"] < dial_base * 0.45:
168
+ notes.append("对白占比骤降(本章 %s/千字,此前中位 %s/千字)——"
169
+ "复查是否换成了大段叙述,文风与推进方式是否漂移"
170
+ % (cur["dialogue_per_k"], dial_base))
171
+ emo_base = _median([b["emotion_per_k"] for b in base])
172
+ if emo_base >= 1.5 and cur["emotion_per_k"] < emo_base * 0.4:
173
+ notes.append("情绪动作密度骤降(本章 %s/千字,此前中位 %s/千字)——"
174
+ "复查复杂情感是否退回「很伤心/很愤怒」式概括"
175
+ % (cur["emotion_per_k"], emo_base))
176
+ return notes
177
+
178
+
179
+ def opening_below(means, threshold):
180
+ """True when the signing opening dimension is scored below the line."""
181
+ try:
182
+ value = float((means or {}).get(OPENING_DIM))
183
+ except (TypeError, ValueError):
184
+ return False
185
+ try:
186
+ line = float(threshold)
187
+ except (TypeError, ValueError):
188
+ return False
189
+ return value < line
@@ -1,228 +1,228 @@
1
- # -*- coding: utf-8 -*-
2
- """Append-only external operation ledger.
3
-
4
- External writes cannot promise exactly-once delivery. Each write therefore
5
- has a stable operation id and an explicit state: pending, confirmed, failed or
6
- unknown. Unknown means the caller must reconcile with the remote system before
7
- retrying; it is never treated as success.
8
- """
9
- from __future__ import annotations
10
-
11
- import hashlib
12
- import json
13
- import secrets
14
- import threading
15
- import time
16
- import urllib.error
17
- from pathlib import Path
18
-
19
- from . import paths
20
-
21
- LOCK = threading.RLock()
22
- STATUSES = ("pending", "confirmed", "failed", "unknown")
23
- PENDING_TTL_SECONDS = 15 * 60
24
-
25
-
26
- def _file(day=None) -> Path:
27
- day = day or time.strftime("%Y-%m-%d")
28
- return paths.DATA_DIR / "operations" / ("operations-%s.jsonl" % day[:7].replace("-", ""))
29
-
30
-
31
- def _read_all():
32
- rows = []
33
- root = paths.DATA_DIR / "operations"
34
- try:
35
- files = sorted(root.glob("operations-*.jsonl"))
36
- except OSError:
37
- files = []
38
- for fp in files:
39
- try:
40
- for line in fp.read_text(encoding="utf-8").splitlines():
41
- try:
42
- row = json.loads(line)
43
- except (TypeError, ValueError):
44
- continue
45
- if isinstance(row, dict) and row.get("operation_id"):
46
- rows.append(row)
47
- except OSError:
48
- continue
49
- return rows
50
-
51
-
52
- def _write_all(rows):
53
- """Test/support migration helper; production writes use append-only _append."""
54
- root = paths.DATA_DIR / "operations"
55
- root.mkdir(parents=True, exist_ok=True)
56
- fp = _file()
57
- fp.write_text("".join(json.dumps(r, ensure_ascii=False) + "\n" for r in rows),
58
- encoding="utf-8")
59
-
60
-
61
- def _append(row):
62
- try:
63
- fp = _file(row.get("submitted_at") and
64
- time.strftime("%Y-%m-%d", time.localtime(float(row["submitted_at"]))))
65
- fp.parent.mkdir(parents=True, exist_ok=True)
66
- with fp.open("a", encoding="utf-8") as handle:
67
- handle.write(json.dumps(row, ensure_ascii=False, sort_keys=True) + "\n")
68
- handle.flush()
69
- except Exception:
70
- # Observability must not turn a remote write into a local failure.
71
- pass
72
-
73
-
74
- def _request_hash(request) -> str:
75
- raw = json.dumps(request or {}, ensure_ascii=False, sort_keys=True,
76
- separators=(",", ":"), default=str)
77
- return hashlib.sha256(raw.encode("utf-8")).hexdigest()
78
-
79
-
80
- def _operation_id(target: str) -> str:
81
- stamp = time.strftime("%Y%m%d-%H%M%S")
82
- return "op-%s-%s-%s" % (stamp, secrets.token_hex(4),
83
- hashlib.sha256(str(target).encode()).hexdigest()[:6])
84
-
85
-
86
- def _safe_meta(metadata):
87
- if not isinstance(metadata, dict):
88
- return {}
89
- out = {}
90
- for key, value in metadata.items():
91
- name = str(key)[:40]
92
- if any(token in name.lower() for token in ("key", "token", "password", "secret")):
93
- continue
94
- if isinstance(value, (str, int, float, bool)) or value is None:
95
- out[name] = str(value)[:300] if isinstance(value, str) else value
96
- return out
97
-
98
-
99
- def begin(target, request=None, *, task_id="", run_id="", metadata=None):
100
- operation_id = _operation_id(target)
101
- now = time.time()
102
- row = {
103
- "operation_id": operation_id,
104
- "target": str(target or "")[:160],
105
- "request_hash": _request_hash(request),
106
- "submitted_at": now,
107
- "status": "pending",
108
- "remote_receipt": "",
109
- "error": "",
110
- "last_reconciled_at": None,
111
- "task_id": str(task_id or "")[:80],
112
- "run_id": str(run_id or "")[:80],
113
- "metadata": _safe_meta(metadata),
114
- }
115
- with LOCK:
116
- _append(row)
117
- return operation_id
118
-
119
-
120
- def update(operation_id, status, *, remote_receipt="", error="",
121
- metadata=None, reconciled=False):
122
- operation_id = str(operation_id or "").strip()
123
- status = str(status or "").strip().lower()
124
- if not operation_id or status not in STATUSES:
125
- return False
126
- with LOCK:
127
- # Read and transition-check under the same lock as append. Two
128
- # concurrent callbacks must not race a pending operation into
129
- # conflicting terminal states.
130
- current = get(operation_id)
131
- if current is None:
132
- return False
133
- current_status = str(current.get("status") or "pending")
134
- # External effects are monotonic: a confirmed write must never be
135
- # downgraded by cleanup/error handling. Unknown/failed can only be
136
- # promoted by an explicit remote reconciliation.
137
- if current_status == "confirmed" and status != "confirmed":
138
- return False
139
- if current_status in ("unknown", "failed") and status == "confirmed" and not reconciled:
140
- return False
141
- if current_status in ("unknown", "failed") and status != current_status and not reconciled:
142
- return False
143
- if status == "confirmed" and not remote_receipt:
144
- remote_receipt = current.get("remote_receipt") or ""
145
- now = time.time()
146
- row = {
147
- "operation_id": operation_id,
148
- "status": status,
149
- "remote_receipt": str(remote_receipt or "")[:300],
150
- "error": str(error or "")[:500],
151
- "last_reconciled_at": now if reconciled else None,
152
- "updated_at": now,
153
- }
154
- if metadata:
155
- row["metadata"] = _safe_meta(metadata)
156
- _append(row)
157
- return True
158
-
159
-
160
- def confirm(operation_id, remote_receipt="", *, metadata=None):
161
- return update(operation_id, "confirmed", remote_receipt=remote_receipt,
162
- metadata=metadata)
163
-
164
-
165
- def fail(operation_id, error="", *, metadata=None):
166
- return update(operation_id, "failed", error=error, metadata=metadata)
167
-
168
-
169
- def mark_unknown(operation_id, error="", *, metadata=None):
170
- return update(operation_id, "unknown", error=error, metadata=metadata)
171
-
172
-
173
- def reconcile(operation_id, status, remote_receipt="", error="", metadata=None):
174
- """Record a remote lookup result for an unknown operation."""
175
- return update(operation_id, status, remote_receipt=remote_receipt, error=error,
176
- metadata=metadata, reconciled=True)
177
-
178
-
179
- def get(operation_id):
180
- operation_id = str(operation_id or "")
181
- latest = None
182
- with LOCK:
183
- for row in _read_all():
184
- if row.get("operation_id") == operation_id:
185
- latest = dict(latest or {}, **row)
186
- return latest
187
-
188
-
189
- def recent(*, target="", status="", limit=100):
190
- latest = {}
191
- with LOCK:
192
- for row in _read_all():
193
- if target and str(row.get("target") or "") != str(target):
194
- continue
195
- latest[row["operation_id"]] = dict(latest.get(row["operation_id"]) or {}, **row)
196
- rows = list(latest.values())
197
- if status:
198
- rows = [r for r in rows if r.get("status") == status]
199
- rows.sort(key=lambda r: float(r.get("updated_at") or r.get("submitted_at") or 0), reverse=True)
200
- return rows[:max(1, min(1000, int(limit or 100)))]
201
-
202
-
203
- def recover_pending(*, now=None, ttl_seconds=PENDING_TTL_SECONDS):
204
- """Turn stale pending writes into unknown so restart never replays blindly."""
205
- now = time.time() if now is None else float(now)
206
- changed = 0
207
- for row in recent(status="pending", limit=1000):
208
- submitted = float(row.get("submitted_at") or 0)
209
- if submitted and now - submitted > max(1, float(ttl_seconds)):
210
- if mark_unknown(row["operation_id"], "进程重启后未收到远端回执,请先对账"):
211
- changed += 1
212
- return changed
213
-
214
-
215
- def status_for_exception(error) -> str:
216
- text = str(error or "").lower()
217
- if isinstance(error, (TimeoutError, ConnectionError, urllib.error.URLError)) or any(x in text for x in
218
- ("timeout", "timed out", "超时", "connection", "connect", "connection reset",
219
- "断开", "断连", "连不上", "网络", "network", "unreachable", "disconnect")):
220
- return "unknown"
221
- return "failed"
222
-
223
-
224
- def finish_exception(operation_id, error):
225
- status = status_for_exception(error)
226
- if status == "unknown":
227
- return mark_unknown(operation_id, str(error))
228
- return fail(operation_id, str(error))
1
+ # -*- coding: utf-8 -*-
2
+ """Append-only external operation ledger.
3
+
4
+ External writes cannot promise exactly-once delivery. Each write therefore
5
+ has a stable operation id and an explicit state: pending, confirmed, failed or
6
+ unknown. Unknown means the caller must reconcile with the remote system before
7
+ retrying; it is never treated as success.
8
+ """
9
+ from __future__ import annotations
10
+
11
+ import hashlib
12
+ import json
13
+ import secrets
14
+ import threading
15
+ import time
16
+ import urllib.error
17
+ from pathlib import Path
18
+
19
+ from . import paths
20
+
21
+ LOCK = threading.RLock()
22
+ STATUSES = ("pending", "confirmed", "failed", "unknown")
23
+ PENDING_TTL_SECONDS = 15 * 60
24
+
25
+
26
+ def _file(day=None) -> Path:
27
+ day = day or time.strftime("%Y-%m-%d")
28
+ return paths.DATA_DIR / "operations" / ("operations-%s.jsonl" % day[:7].replace("-", ""))
29
+
30
+
31
+ def _read_all():
32
+ rows = []
33
+ root = paths.DATA_DIR / "operations"
34
+ try:
35
+ files = sorted(root.glob("operations-*.jsonl"))
36
+ except OSError:
37
+ files = []
38
+ for fp in files:
39
+ try:
40
+ for line in fp.read_text(encoding="utf-8").splitlines():
41
+ try:
42
+ row = json.loads(line)
43
+ except (TypeError, ValueError):
44
+ continue
45
+ if isinstance(row, dict) and row.get("operation_id"):
46
+ rows.append(row)
47
+ except OSError:
48
+ continue
49
+ return rows
50
+
51
+
52
+ def _write_all(rows):
53
+ """Test/support migration helper; production writes use append-only _append."""
54
+ root = paths.DATA_DIR / "operations"
55
+ root.mkdir(parents=True, exist_ok=True)
56
+ fp = _file()
57
+ fp.write_text("".join(json.dumps(r, ensure_ascii=False) + "\n" for r in rows),
58
+ encoding="utf-8")
59
+
60
+
61
+ def _append(row):
62
+ try:
63
+ fp = _file(row.get("submitted_at") and
64
+ time.strftime("%Y-%m-%d", time.localtime(float(row["submitted_at"]))))
65
+ fp.parent.mkdir(parents=True, exist_ok=True)
66
+ with fp.open("a", encoding="utf-8") as handle:
67
+ handle.write(json.dumps(row, ensure_ascii=False, sort_keys=True) + "\n")
68
+ handle.flush()
69
+ except Exception:
70
+ # Observability must not turn a remote write into a local failure.
71
+ pass
72
+
73
+
74
+ def _request_hash(request) -> str:
75
+ raw = json.dumps(request or {}, ensure_ascii=False, sort_keys=True,
76
+ separators=(",", ":"), default=str)
77
+ return hashlib.sha256(raw.encode("utf-8")).hexdigest()
78
+
79
+
80
+ def _operation_id(target: str) -> str:
81
+ stamp = time.strftime("%Y%m%d-%H%M%S")
82
+ return "op-%s-%s-%s" % (stamp, secrets.token_hex(4),
83
+ hashlib.sha256(str(target).encode()).hexdigest()[:6])
84
+
85
+
86
+ def _safe_meta(metadata):
87
+ if not isinstance(metadata, dict):
88
+ return {}
89
+ out = {}
90
+ for key, value in metadata.items():
91
+ name = str(key)[:40]
92
+ if any(token in name.lower() for token in ("key", "token", "password", "secret")):
93
+ continue
94
+ if isinstance(value, (str, int, float, bool)) or value is None:
95
+ out[name] = str(value)[:300] if isinstance(value, str) else value
96
+ return out
97
+
98
+
99
+ def begin(target, request=None, *, task_id="", run_id="", metadata=None):
100
+ operation_id = _operation_id(target)
101
+ now = time.time()
102
+ row = {
103
+ "operation_id": operation_id,
104
+ "target": str(target or "")[:160],
105
+ "request_hash": _request_hash(request),
106
+ "submitted_at": now,
107
+ "status": "pending",
108
+ "remote_receipt": "",
109
+ "error": "",
110
+ "last_reconciled_at": None,
111
+ "task_id": str(task_id or "")[:80],
112
+ "run_id": str(run_id or "")[:80],
113
+ "metadata": _safe_meta(metadata),
114
+ }
115
+ with LOCK:
116
+ _append(row)
117
+ return operation_id
118
+
119
+
120
+ def update(operation_id, status, *, remote_receipt="", error="",
121
+ metadata=None, reconciled=False):
122
+ operation_id = str(operation_id or "").strip()
123
+ status = str(status or "").strip().lower()
124
+ if not operation_id or status not in STATUSES:
125
+ return False
126
+ with LOCK:
127
+ # Read and transition-check under the same lock as append. Two
128
+ # concurrent callbacks must not race a pending operation into
129
+ # conflicting terminal states.
130
+ current = get(operation_id)
131
+ if current is None:
132
+ return False
133
+ current_status = str(current.get("status") or "pending")
134
+ # External effects are monotonic: a confirmed write must never be
135
+ # downgraded by cleanup/error handling. Unknown/failed can only be
136
+ # promoted by an explicit remote reconciliation.
137
+ if current_status == "confirmed" and status != "confirmed":
138
+ return False
139
+ if current_status in ("unknown", "failed") and status == "confirmed" and not reconciled:
140
+ return False
141
+ if current_status in ("unknown", "failed") and status != current_status and not reconciled:
142
+ return False
143
+ if status == "confirmed" and not remote_receipt:
144
+ remote_receipt = current.get("remote_receipt") or ""
145
+ now = time.time()
146
+ row = {
147
+ "operation_id": operation_id,
148
+ "status": status,
149
+ "remote_receipt": str(remote_receipt or "")[:300],
150
+ "error": str(error or "")[:500],
151
+ "last_reconciled_at": now if reconciled else None,
152
+ "updated_at": now,
153
+ }
154
+ if metadata:
155
+ row["metadata"] = _safe_meta(metadata)
156
+ _append(row)
157
+ return True
158
+
159
+
160
+ def confirm(operation_id, remote_receipt="", *, metadata=None):
161
+ return update(operation_id, "confirmed", remote_receipt=remote_receipt,
162
+ metadata=metadata)
163
+
164
+
165
+ def fail(operation_id, error="", *, metadata=None):
166
+ return update(operation_id, "failed", error=error, metadata=metadata)
167
+
168
+
169
+ def mark_unknown(operation_id, error="", *, metadata=None):
170
+ return update(operation_id, "unknown", error=error, metadata=metadata)
171
+
172
+
173
+ def reconcile(operation_id, status, remote_receipt="", error="", metadata=None):
174
+ """Record a remote lookup result for an unknown operation."""
175
+ return update(operation_id, status, remote_receipt=remote_receipt, error=error,
176
+ metadata=metadata, reconciled=True)
177
+
178
+
179
+ def get(operation_id):
180
+ operation_id = str(operation_id or "")
181
+ latest = None
182
+ with LOCK:
183
+ for row in _read_all():
184
+ if row.get("operation_id") == operation_id:
185
+ latest = dict(latest or {}, **row)
186
+ return latest
187
+
188
+
189
+ def recent(*, target="", status="", limit=100):
190
+ latest = {}
191
+ with LOCK:
192
+ for row in _read_all():
193
+ if target and str(row.get("target") or "") != str(target):
194
+ continue
195
+ latest[row["operation_id"]] = dict(latest.get(row["operation_id"]) or {}, **row)
196
+ rows = list(latest.values())
197
+ if status:
198
+ rows = [r for r in rows if r.get("status") == status]
199
+ rows.sort(key=lambda r: float(r.get("updated_at") or r.get("submitted_at") or 0), reverse=True)
200
+ return rows[:max(1, min(1000, int(limit or 100)))]
201
+
202
+
203
+ def recover_pending(*, now=None, ttl_seconds=PENDING_TTL_SECONDS):
204
+ """Turn stale pending writes into unknown so restart never replays blindly."""
205
+ now = time.time() if now is None else float(now)
206
+ changed = 0
207
+ for row in recent(status="pending", limit=1000):
208
+ submitted = float(row.get("submitted_at") or 0)
209
+ if submitted and now - submitted > max(1, float(ttl_seconds)):
210
+ if mark_unknown(row["operation_id"], "进程重启后未收到远端回执,请先对账"):
211
+ changed += 1
212
+ return changed
213
+
214
+
215
+ def status_for_exception(error) -> str:
216
+ text = str(error or "").lower()
217
+ if isinstance(error, (TimeoutError, ConnectionError, urllib.error.URLError)) or any(x in text for x in
218
+ ("timeout", "timed out", "超时", "connection", "connect", "connection reset",
219
+ "断开", "断连", "连不上", "网络", "network", "unreachable", "disconnect")):
220
+ return "unknown"
221
+ return "failed"
222
+
223
+
224
+ def finish_exception(operation_id, error):
225
+ status = status_for_exception(error)
226
+ if status == "unknown":
227
+ return mark_unknown(operation_id, str(error))
228
+ return fail(operation_id, str(error))