@furongjun1999/dsh-memory 0.9.6 → 0.9.8
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.
- package/README.md +150 -662
- package/codebuddy/CODEBUDDY.md +333 -204
- package/data/policy.json +38 -38
- package/docs/README.md +18 -0
- package/docs/discipline/harnesses.yaml +11 -3
- package/docs/discipline/templates/compact.txt.tmpl +20 -18
- package/docs/discipline/templates/full.md.tmpl +63 -61
- package/docs/discipline/templates/rules.mdc.tmpl +5 -3
- package/docs/discipline/templates/skill.md.tmpl +25 -23
- package/docs/discipline/templates/spec.txt.tmpl +2 -2
- package/docs/discipline/templates/zcode-user.md.tmpl +1 -1
- package/dsh/cordis.yml.example +139 -139
- package/lib/hooks.d.ts +19 -0
- package/lib/hooks.js +30 -7
- package/lib/index.js +18 -3
- package/lib/init.js +19 -4
- package/lib/lib/hook_audit.js +64 -2
- package/lib/lib/prompt_safety.js +14 -3
- package/lib/tools.d.ts +7 -0
- package/lib/tools.js +22 -1
- package/md_cg/_guard_kit.py +121 -121
- package/md_cg/anchor.py +114 -114
- package/md_cg/audit.py +15 -2
- package/md_cg/chain.py +3 -2
- package/md_cg/config_registry_guard.py +487 -487
- package/md_cg/conformance.py +747 -747
- package/md_cg/consolidate.py +1793 -1793
- package/md_cg/crypto.py +794 -794
- package/md_cg/hyperedge.py +262 -262
- package/md_cg/interface_schema.py +146 -146
- package/md_cg/lifecycle.py +426 -426
- package/md_cg/mdcos.py +12 -8
- package/md_cg/predict.py +1103 -1103
- package/md_cg/protect.py +654 -654
- package/md_cg/provenance.py +606 -606
- package/md_cg/queue_health.py +130 -130
- package/md_cg/refine.py +613 -613
- package/md_cg/retrieve.py +128 -128
- package/md_cg/revise.py +140 -140
- package/md_cg/rollback.py +657 -657
- package/md_cg/rollback_cli.py +124 -124
- package/md_cg/rotate.py +346 -346
- package/md_cg/routing.py +490 -490
- package/md_cg/self_surface.py +50 -50
- package/md_cg/semantic/canonical.py +144 -144
- package/md_cg/semantic/zh_en_atoms.py +141 -141
- package/md_cg/sources.py +1091 -1091
- package/md_cg/stg.py +3 -2
- package/md_cg/sustain.py +10 -1
- package/md_cg/test_audit_rotate_parity.py +279 -279
- package/md_cg/test_autonomy_modes.py +2013 -2013
- package/md_cg/test_cn_recall_grams.py +206 -206
- package/md_cg/test_code_kind_forbidden.py +101 -0
- package/md_cg/test_condaccept.py +156 -156
- package/md_cg/test_condgate.py +141 -141
- package/md_cg/test_heartbeat_ledger.py +9 -3
- package/md_cg/test_i50e_readside_protection.py +658 -658
- package/md_cg/test_issue111_gate_order.py +76 -76
- package/md_cg/test_issue111_scoring_contract.py +86 -86
- package/md_cg/test_issue121_calibration_bounds.py +293 -293
- package/md_cg/test_issue145_sleep_no_silent_rollback.py +167 -167
- package/md_cg/test_issue147_pack_id_escape.py +378 -378
- package/md_cg/test_issue148_merge_downgrade_leak.py +192 -192
- package/md_cg/test_issue149_accel_default_face.py +248 -248
- package/md_cg/test_issue150_empty_query_terms.py +328 -328
- package/md_cg/test_issue151_153_157_write_cache_index.py +151 -151
- package/md_cg/test_issue152_sealed_postings.py +309 -309
- package/md_cg/test_issue153b_index_generation_invalidation.py +142 -142
- package/md_cg/test_issue157b_index_remainder.py +344 -344
- package/md_cg/test_issue189_conclusion_bidirectional.py +223 -223
- package/md_cg/test_issue191_import_trust.py +382 -382
- package/md_cg/test_issue193_write_scope.py +378 -378
- package/md_cg/test_issue212_writeface_gaps.py +349 -349
- package/md_cg/test_issue214_reject_sensitivity.py +266 -266
- package/md_cg/test_issue215_condition_source_gap.py +272 -272
- package/md_cg/test_issue216_clash_routes.py +214 -214
- package/md_cg/test_issue217_slot_routes.py +248 -248
- package/md_cg/test_issue218_postcondition_value_blind.py +222 -222
- package/md_cg/test_issue258_pack_sensitivity.py +468 -468
- package/md_cg/test_issue265_coldverify_multiproc.py +438 -438
- package/md_cg/test_issue274_read_node_id_none.py +341 -341
- package/md_cg/test_issue275_layer_param.py +250 -250
- package/md_cg/test_issue277_node_id_traversal.py +384 -384
- package/md_cg/test_issue278_neg_separated_tension.py +271 -271
- package/md_cg/test_issue278_neg_window.py +294 -294
- package/md_cg/test_issue278_negative_layering.py +83 -83
- package/md_cg/test_issue284_skip_not_pass.py +291 -291
- package/md_cg/test_issue294_flywheel_sensitivity.py +573 -573
- package/md_cg/test_issue301_baseline_corpus_gate.py +245 -245
- package/md_cg/test_issue303_design_boundaries.py +262 -262
- package/md_cg/test_issue315_merge_branch_sensitivity.py +383 -383
- package/md_cg/test_issue319_tenant_unregistered_bypass.py +256 -256
- package/md_cg/test_issue321_mcp_deep_json.py +304 -304
- package/md_cg/test_issue322_node_write_lock.py +727 -727
- package/md_cg/test_issue324_write_size_fields.py +214 -214
- package/md_cg/test_issue325_audit_poison.py +229 -229
- package/md_cg/test_issue326_ingest_auto_whitelist.py +331 -331
- package/md_cg/test_issue326_key_perms.py +355 -355
- package/md_cg/test_issue326_twin_parser_parity.py +323 -323
- package/md_cg/test_issue326_vm_call_frame.py +569 -569
- package/md_cg/test_issue345_writelimit_effective_sens.py +233 -233
- package/md_cg/test_issue346_overwrite_sensitivity_gate.py +424 -424
- package/md_cg/test_issue348_insight_verify_op.py +199 -199
- package/md_cg/test_issue348_merge_rmw_lock.py +345 -345
- package/md_cg/test_issue350_casefold_collision.py +235 -235
- package/md_cg/test_issue351_ingest_root_fallback.py +201 -201
- package/md_cg/test_issue352_imgaudit.py +418 -418
- package/md_cg/test_issue352_pathsjson.py +537 -537
- package/md_cg/test_issue352_refindex.py +412 -412
- package/md_cg/test_issue352_rollback.py +604 -604
- package/md_cg/test_issue354_p2_writelimit_toctou.py +327 -327
- package/md_cg/test_issue354_rust_index_containment.py +571 -571
- package/md_cg/test_issue354_rust_wal_containment.py +606 -606
- package/md_cg/test_issue358_truncation_observable.py +247 -247
- package/md_cg/test_issue360_tidy_private_excerpt.py +195 -195
- package/md_cg/test_issue361_blindspot_not_silent_drop.py +494 -494
- package/md_cg/test_issue362_neg_condition.py +468 -468
- package/md_cg/test_issue363_lexical_idf.py +275 -275
- package/md_cg/test_issue370_readcache_assembly.py +488 -488
- package/md_cg/test_issue376_heartbeat_failed.py +293 -293
- package/md_cg/test_issue377_layer_protect_stamp.py +324 -324
- package/md_cg/test_issue378_protected_floor.py +310 -310
- package/md_cg/test_issue383_state_event_write_gate.py +462 -462
- package/md_cg/test_issue397_sens_eff_wiring.py +435 -435
- package/md_cg/test_issue398_toolface_op_parity.py +511 -511
- package/md_cg/test_issue399_hive_json_shape.py +317 -317
- package/md_cg/test_issue401_wisdom_codegen.py +295 -295
- package/md_cg/test_issue404_heal_receipt.py +389 -389
- package/md_cg/test_issue409_compact_index_completeness.py +509 -509
- package/md_cg/test_issue410_verify_demote_write.py +366 -366
- package/md_cg/test_issue411_scrub_idempotence.py +575 -575
- package/md_cg/test_issue412_discipline_producer_parity.py +311 -311
- package/md_cg/test_issue413_preimage_failclosed.py +239 -239
- package/md_cg/test_issue413_preimage_twin.py +339 -339
- package/md_cg/test_issue415_deps_gate_probe.py +357 -357
- package/md_cg/test_issue416_entry_key_subset.py +525 -525
- package/md_cg/test_issue417_twophase_merge_landed.py +423 -423
- package/md_cg/test_issue425_ingest_library_path.py +331 -331
- package/md_cg/test_issue426c_enqueue_not_silent.py +346 -346
- package/md_cg/test_issue426e_restore_overwrite_live.py +310 -310
- package/md_cg/test_issue426f_paths_gbk_failclosed.py +301 -301
- package/md_cg/test_issue427_serve_status_shape.py +191 -191
- package/md_cg/test_issue431_432_468_protect_single_point.py +322 -322
- package/md_cg/test_issue433_outlook_threshold_single_point.py +278 -278
- package/md_cg/test_issue434_435_float_gate.py +90 -90
- package/md_cg/test_issue439_sibling_sensitivity_inherit.py +448 -448
- package/md_cg/test_issue43_default_policy.py +874 -874
- package/md_cg/test_issue444_scrub_evidence_plaintext.py +281 -281
- package/md_cg/test_issue445_metacognition_ledger_gate.py +454 -454
- package/md_cg/test_issue446_453_403_verified_evidence.py +382 -382
- package/md_cg/test_issue450_sensitivity_fallback_reason.py +51 -51
- package/md_cg/test_issue450_writegate_unknown_sens.py +203 -203
- package/md_cg/test_issue451_audit_actor.py +313 -313
- package/md_cg/test_issue452_ingest_root_sensitivity.py +299 -299
- package/md_cg/test_issue463_ghost_prune_guard.py +259 -259
- package/md_cg/test_issue466_basis_registry.py +469 -469
- package/md_cg/test_issue467_identity_sensitivity_parity.py +360 -360
- package/md_cg/test_issue469_scrub_override_passthrough.py +126 -126
- package/md_cg/test_issue470_wal_event_durable.py +427 -427
- package/md_cg/test_issue471_identity_readable.py +136 -136
- package/md_cg/test_issue472_audit_chain.py +90 -90
- package/md_cg/test_issue472_prune_ledger.py +105 -105
- package/md_cg/test_issue472_rollback_degraded.py +81 -81
- package/md_cg/test_issue491_compose_sandbox.py +460 -460
- package/md_cg/test_issue498_merge_trace_stale_index.py +280 -280
- package/md_cg/test_issue511_scoreboard_bad_files.py +256 -256
- package/md_cg/test_issue521_rotate_lock_suffix.py +110 -110
- package/md_cg/test_issue524_orphan_noise_no_demote.py +88 -88
- package/md_cg/test_issue525_read_null_semantics.py +52 -52
- package/md_cg/test_issue547_append_jsonl_whitelist.py +114 -114
- package/md_cg/test_issue547_audit_durable.py +96 -96
- package/md_cg/test_issue557_r2_write_entries_size_gate.py +83 -83
- package/md_cg/test_issue559_refine_prefix.py +69 -69
- package/md_cg/test_issue562_retired_faces.py +150 -0
- package/md_cg/test_issue71_queue_backlog.py +376 -376
- package/md_cg/test_issue72_coldverify_controlled_drain.py +420 -420
- package/md_cg/test_issue82_credential_rules.py +271 -271
- package/md_cg/test_issue91_verifier_sandbox.py +314 -314
- package/md_cg/test_issue93_remember_redaction.py +282 -282
- package/md_cg/test_logref.py +1141 -1141
- package/md_cg/test_md_access_parity.py +330 -330
- package/md_cg/test_md_cg_no_hive_dep.py +139 -139
- package/md_cg/test_mutation_residue.py +119 -119
- package/md_cg/test_mutation_rollback.py +835 -835
- package/md_cg/test_n123_derive_expiry_chain.py +210 -210
- package/md_cg/test_n184_keys_concurrent_provision.py +1 -1
- package/md_cg/test_n209_verify_write_face_gates.py +466 -466
- package/md_cg/test_n62_tenant_bind_failclosed.py +200 -200
- package/md_cg/test_neg_tail_honesty.py +733 -733
- package/md_cg/test_p21_tokens.py +294 -294
- package/md_cg/test_p27_docindex.py +792 -792
- package/md_cg/test_p28_refcheck.py +1053 -1053
- package/md_cg/test_p2_six_elements.py +476 -476
- package/md_cg/test_p33_ccg_wiring.py +300 -300
- package/md_cg/test_p3_legacy_closure.py +438 -438
- package/md_cg/test_p40_refine_worklist.py +244 -244
- package/md_cg/test_p4_freshness.py +781 -781
- package/md_cg/test_p9c_dedup_hints.py +288 -288
- package/md_cg/test_policy_required_ccg.py +444 -444
- package/md_cg/test_read_scope_b27.py +2 -2
- package/md_cg/test_retr_s5.py +312 -312
- package/md_cg/test_s1_ast_judge.py +109 -109
- package/md_cg/test_security_audit_v21.py +249 -249
- package/md_cg/test_state_event_op.py +521 -521
- package/md_cg/test_stgidx_index_parity.py +1024 -1024
- package/md_cg/test_sustain_bad_stamp.py +178 -0
- package/md_cg/test_sustain_bounded.py +652 -652
- package/md_cg/test_trust_edge_reachability.py +405 -405
- package/md_cg/test_w7_redlines.py +659 -659
- package/md_cg/test_write_return_shape.py +919 -919
- package/md_cg/test_writepipe.py +228 -228
- package/md_cg/trust.py +1114 -1114
- package/md_cg/twophase.py +303 -303
- package/md_cg/whitebox.py +536 -536
- package/md_cg/whitebox_kb/wisdom/compose_engine.py +1634 -1634
- package/md_cg/whitebox_kb/wisdom/core/condition_vm.py +318 -318
- package/md_cg/whitebox_kb/wisdom/core/parser.py +1082 -1082
- package/md_cg/whitebox_kb/wisdom/md_access.py +490 -490
- package/md_cg/whitebox_kb/wisdom/migrate_wisdom.py +497 -497
- package/md_cg/whitebox_kb/wisdom/mini_python.py +995 -995
- package/md_cg/whitebox_kb/wisdom/net_units.py +2900 -2900
- package/md_cg/whitebox_kb/wisdom/semantic_translate.py +4060 -4060
- package/md_cg/writepipe.py +1337 -1337
- package/package.json +3 -2
- package/scripts/workspace_index.py +328 -0
- package/src/hooks.ts +932 -908
- package/src/index.ts +21 -6
- package/src/init.ts +16 -4
- package/src/lib/hook_audit.ts +48 -2
- package/src/lib/mdcg_client.ts +469 -469
- package/src/lib/prompt_safety.ts +17 -3
- package/src/tools.ts +23 -1
- package/zcode/AGENTS.md +333 -204
|
@@ -1,497 +1,497 @@
|
|
|
1
|
-
#!/usr/bin/env python3
|
|
2
|
-
# -*- coding: utf-8 -*-
|
|
3
|
-
"""
|
|
4
|
-
migrate_wisdom · 智慧之书 → 灵枢主库迁移器(v1.16 · 统一信息差减少任务)
|
|
5
|
-
|
|
6
|
-
设计者方向(2026-08-18):智慧之书库迁移进灵枢主库,灵枢成为完整工程实现
|
|
7
|
-
(灵魂=智能论3.2 协议层 + 神经系统功能类比),统一「信息差减少」任务。
|
|
8
|
-
|
|
9
|
-
知识真身分层(迁移数据源):
|
|
10
|
-
META 层:wisdom-book-cloud.db 30 元学科卡(存在论/条件论/智能论/学科映射等)
|
|
11
|
-
87 边(31 causal + 56 similar,79 verified)
|
|
12
|
-
SUBJECT 层:knowledge-base/*_knowledge.json 51 学科卡(E1-E4,23-81 知识点/卡)
|
|
13
|
-
+ programming_lang_cards.json 3 语言卡
|
|
14
|
-
CAUSAL 层:causal_edge_candidates.json 学段递进因果边(R1 依赖型因果)
|
|
15
|
-
+ 元层 87 边中的 causal 边
|
|
16
|
-
|
|
17
|
-
迁移原则:
|
|
18
|
-
- 幂等:按卡名(state_attributes.name)查重,已存在跳过
|
|
19
|
-
- 可逆:迁移前备份目标库
|
|
20
|
-
- 诚实:verified 标记如实保留(元层边已验证的保持 verified;
|
|
21
|
-
学段递进边 R1 规则推导 → verified=0 confidence 0.75)
|
|
22
|
-
- 报告:迁移统计输出 migrate_wisdom_report.json(可复现)
|
|
23
|
-
纯标准库 · 零外部依赖
|
|
24
|
-
"""
|
|
25
|
-
|
|
26
|
-
import json
|
|
27
|
-
import os
|
|
28
|
-
import shutil
|
|
29
|
-
import sqlite3
|
|
30
|
-
import sys
|
|
31
|
-
import time
|
|
32
|
-
from typing import Dict, List, Optional
|
|
33
|
-
|
|
34
|
-
HERE = os.path.dirname(os.path.abspath(__file__))
|
|
35
|
-
|
|
36
|
-
|
|
37
|
-
# 生效条件:包内相对导入成功时返回 md_cg.consistency.is_discipline_node;失败(无包上下文/导入面异常)
|
|
38
|
-
# 返回 None——**退化可见**:调用方必须显式判 None 并把结论写进 phase(不静默假装通过)。
|
|
39
|
-
def _discipline_predicate():
|
|
40
|
-
"""纪律判据单点(#412 负向自证用):与判据**同用**同一个谓词,本文件不复制词形表。"""
|
|
41
|
-
try:
|
|
42
|
-
from ...consistency import is_discipline_node # 包内路径:md_cg.consistency
|
|
43
|
-
return is_discipline_node
|
|
44
|
-
except Exception: # noqa: BLE001 —— 独立运行时无包上下文
|
|
45
|
-
return None
|
|
46
|
-
sys.path.insert(0, HERE)
|
|
47
|
-
|
|
48
|
-
# 默认路径(可参数覆盖);目标库走 LINGSHU_DB 环境变量,否则家目录 .dsh 布局
|
|
49
|
-
DEFAULT_META_SRC = os.path.join(os.path.dirname(os.path.abspath(__file__)), "wisdom-book-cloud.db")
|
|
50
|
-
DEFAULT_DST = os.environ.get("LINGSHU_DB") or os.path.join(
|
|
51
|
-
os.path.expanduser("~"), ".dsh", "profiles", "web", "data", "lingshu.db")
|
|
52
|
-
DEFAULT_CAUSAL = os.path.join(HERE, "causal_edge_candidates.json")
|
|
53
|
-
|
|
54
|
-
|
|
55
|
-
class WisdomMigrator:
|
|
56
|
-
"""智慧之书 → 灵枢主库迁移器"""
|
|
57
|
-
|
|
58
|
-
def __init__(self, meta_src: str = DEFAULT_META_SRC,
|
|
59
|
-
dst: str = DEFAULT_DST,
|
|
60
|
-
subject_dir: Optional[str] = None,
|
|
61
|
-
causal_src: str = DEFAULT_CAUSAL,
|
|
62
|
-
backup: bool = True):
|
|
63
|
-
"""迁移会话初始化:源库·目标库路径与游标准备。"""
|
|
64
|
-
self.meta_src = meta_src
|
|
65
|
-
self.dst = dst
|
|
66
|
-
self.subject_dir = subject_dir or HERE
|
|
67
|
-
self.causal_src = causal_src
|
|
68
|
-
self.report: Dict = {"phases": {}}
|
|
69
|
-
if backup and os.path.exists(dst):
|
|
70
|
-
bak = dst + f".bak_migrate_{time.strftime('%Y%m%d_%H%M%S')}"
|
|
71
|
-
shutil.copy2(dst, bak)
|
|
72
|
-
self.report["backup"] = bak
|
|
73
|
-
|
|
74
|
-
# ---------------- META 层:30 元卡 + 87 边 ----------------
|
|
75
|
-
def migrate_meta(self) -> Dict:
|
|
76
|
-
phase = {"nodes": 0, "edges": 0, "skipped_nodes": 0, "skipped_edges": 0}
|
|
77
|
-
if not os.path.exists(self.meta_src):
|
|
78
|
-
phase["error"] = f"源库不存在: {self.meta_src}"
|
|
79
|
-
self.report["phases"]["meta"] = phase
|
|
80
|
-
return phase
|
|
81
|
-
src = sqlite3.connect(self.meta_src)
|
|
82
|
-
dst = sqlite3.connect(self.dst)
|
|
83
|
-
sc, dc = src.cursor(), dst.cursor()
|
|
84
|
-
# 已存在的节点 id 集合(幂等)
|
|
85
|
-
existing = {r[0] for r in dc.execute("SELECT id FROM nodes").fetchall()}
|
|
86
|
-
# 搬节点
|
|
87
|
-
rows = sc.execute("SELECT * FROM nodes").fetchall()
|
|
88
|
-
cols = [r[1] for r in sc.execute("PRAGMA table_info(nodes)").fetchall()]
|
|
89
|
-
for row in rows:
|
|
90
|
-
nid = row[0]
|
|
91
|
-
if nid in existing:
|
|
92
|
-
phase["skipped_nodes"] += 1
|
|
93
|
-
continue
|
|
94
|
-
dc.execute(
|
|
95
|
-
"INSERT INTO nodes (%s) VALUES (%s)" %
|
|
96
|
-
(",".join(cols), ",".join("?" * len(cols))), row)
|
|
97
|
-
phase["nodes"] += 1
|
|
98
|
-
# 搬边(仅两端都存在于目标库的边)
|
|
99
|
-
erows = sc.execute("SELECT * FROM edges").fetchall()
|
|
100
|
-
ecols = [r[1] for r in sc.execute("PRAGMA table_info(edges)").fetchall()]
|
|
101
|
-
eexisting = {r[0] for r in dc.execute("SELECT id FROM edges").fetchall()}
|
|
102
|
-
for row in erows:
|
|
103
|
-
eid, sid, tid = row[0], row[1], row[2]
|
|
104
|
-
if eid in eexisting:
|
|
105
|
-
phase["skipped_edges"] += 1
|
|
106
|
-
continue
|
|
107
|
-
if not dc.execute("SELECT id FROM nodes WHERE id=?", (sid,)).fetchone():
|
|
108
|
-
phase["skipped_edges"] += 1
|
|
109
|
-
continue
|
|
110
|
-
if not dc.execute("SELECT id FROM nodes WHERE id=?", (tid,)).fetchone():
|
|
111
|
-
phase["skipped_edges"] += 1
|
|
112
|
-
continue
|
|
113
|
-
dc.execute(
|
|
114
|
-
"INSERT INTO edges (%s) VALUES (%s)" %
|
|
115
|
-
(",".join(ecols), ",".join("?" * len(ecols))), row)
|
|
116
|
-
phase["edges"] += 1
|
|
117
|
-
dst.commit()
|
|
118
|
-
# verified 统计
|
|
119
|
-
phase["verified_edges"] = dc.execute(
|
|
120
|
-
"SELECT COUNT(*) FROM edges WHERE verified=1").fetchone()[0]
|
|
121
|
-
src.close()
|
|
122
|
-
dst.close()
|
|
123
|
-
self.report["phases"]["meta"] = phase
|
|
124
|
-
return phase
|
|
125
|
-
|
|
126
|
-
# ---------------- SUBJECT 层:51 学科卡 + 语言卡 ----------------
|
|
127
|
-
def migrate_subject_cards(self) -> Dict:
|
|
128
|
-
"""学科卡批量迁移:逐识别卡解析→按名去重→写入目标 nodes 表,返回 nodes·skipped·files 迁移统计。"""
|
|
129
|
-
phase = {"nodes": 0, "skipped": 0, "files": 0, "sources": []}
|
|
130
|
-
dst = sqlite3.connect(self.dst)
|
|
131
|
-
dc = dst.cursor()
|
|
132
|
-
# 已存在的卡名(幂等,从 state_attributes.name 解析)
|
|
133
|
-
existing_names = set()
|
|
134
|
-
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
135
|
-
if not sa:
|
|
136
|
-
continue
|
|
137
|
-
try:
|
|
138
|
-
existing_names.add(json.loads(sa).get("name", ""))
|
|
139
|
-
except Exception:
|
|
140
|
-
pass
|
|
141
|
-
edu_level = {"E1": 1, "E2": 2, "E3": 3, "E4": 4, "E5": 5}
|
|
142
|
-
files = sorted(f for f in os.listdir(self.subject_dir)
|
|
143
|
-
if f.endswith("_knowledge.json") or f == "programming_lang_cards.json")
|
|
144
|
-
import re
|
|
145
|
-
for fn in files:
|
|
146
|
-
path = os.path.join(self.subject_dir, fn)
|
|
147
|
-
try:
|
|
148
|
-
d = json.load(open(path, encoding="utf-8"))
|
|
149
|
-
except Exception as e:
|
|
150
|
-
phase["sources"].append({"file": fn, "error": str(e)[:40]})
|
|
151
|
-
continue
|
|
152
|
-
phase["files"] += 1
|
|
153
|
-
# dict(学科卡)或 list(语言卡)统一为卡列表
|
|
154
|
-
cards = d if isinstance(d, list) else [d]
|
|
155
|
-
for card in cards:
|
|
156
|
-
content_map = card.get("content", {}) if isinstance(card, dict) else {}
|
|
157
|
-
meta = card.get("meta", {}) if isinstance(card, dict) else {}
|
|
158
|
-
name = meta.get("name") or card.get("name") or fn.replace("_knowledge.json", "")
|
|
159
|
-
if not name or name in existing_names:
|
|
160
|
-
phase["skipped"] += 1
|
|
161
|
-
continue
|
|
162
|
-
style = meta.get("style", "")
|
|
163
|
-
m = re.search(r"(E\d)", style)
|
|
164
|
-
edu = m.group(1) if m else card.get("edu_level")
|
|
165
|
-
level = card.get("level") or {"E1": 1, "E2": 2, "E3": 3, "E4": 4, "E5": 5}.get(edu, 2)
|
|
166
|
-
domain = (meta.get("domain") or card.get("domain")
|
|
167
|
-
or name.split(" ")[0])
|
|
168
|
-
if content_map and isinstance(content_map, dict):
|
|
169
|
-
kps = list(content_map.items())
|
|
170
|
-
content_lines = [f"{name}(学科知识卡 · {len(kps)} 知识点)"]
|
|
171
|
-
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps)]
|
|
172
|
-
else:
|
|
173
|
-
# 语言卡:response 驱动(trigger/action/counters)
|
|
174
|
-
resp = card.get("response", {})
|
|
175
|
-
content_lines = [f"{name}(语言卡 · {card.get('status', 'pending')})"]
|
|
176
|
-
if isinstance(content_map, str) and content_map:
|
|
177
|
-
content_lines.append(content_map[:300])
|
|
178
|
-
content_lines.append("触发: " + str(resp.get("trigger", ""))[:200])
|
|
179
|
-
content_lines.append("动作: " + str(resp.get("action", ""))[:200])
|
|
180
|
-
content_lines.append("护栏: " + str(resp.get("counters", ""))[:200])
|
|
181
|
-
nid = f"node_subject_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
182
|
-
tags = [f"domain:{domain}", f"level:L{level}",
|
|
183
|
-
f"status:verified", "subject_card",
|
|
184
|
-
f"edu:{edu}" if edu else "edu:unknown"]
|
|
185
|
-
sa = json.dumps({"name": name, "domain": domain, "level": level,
|
|
186
|
-
"edu_level": edu, "kind": "subject_card",
|
|
187
|
-
"source": fn,
|
|
188
|
-
"kp_count": len(content_map) if isinstance(content_map, dict) else 0},
|
|
189
|
-
ensure_ascii=False)
|
|
190
|
-
cs = json.dumps({"observation_position": f"{name} 外部观测位",
|
|
191
|
-
"observation_tool": "学科知识卡(知识综述)",
|
|
192
|
-
"time_window": [0.0, 9999999999.0],
|
|
193
|
-
"existence_constraint": "通用规律/知识(开源非盈利知识库)"},
|
|
194
|
-
ensure_ascii=False)
|
|
195
|
-
now = time.time()
|
|
196
|
-
dc.execute(
|
|
197
|
-
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
198
|
-
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
199
|
-
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
200
|
-
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
201
|
-
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.8, 0.7,
|
|
202
|
-
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
203
|
-
"{}", sa))
|
|
204
|
-
phase["nodes"] += 1
|
|
205
|
-
existing_names.add(name)
|
|
206
|
-
dst.commit()
|
|
207
|
-
dst.close()
|
|
208
|
-
self.report["phases"]["subject"] = phase
|
|
209
|
-
return phase
|
|
210
|
-
|
|
211
|
-
# ---------------- CAUSAL 层:学段递进因果边 ----------------
|
|
212
|
-
def migrate_causal_edges(self) -> Dict:
|
|
213
|
-
phase = {"edges": 0, "skipped": 0}
|
|
214
|
-
if not os.path.exists(self.causal_src):
|
|
215
|
-
phase["error"] = f"因果边源不存在: {self.causal_src}"
|
|
216
|
-
self.report["phases"]["causal"] = phase
|
|
217
|
-
return phase
|
|
218
|
-
cands = json.load(open(self.causal_src, encoding="utf-8"))
|
|
219
|
-
dst = sqlite3.connect(self.dst)
|
|
220
|
-
dc = dst.cursor()
|
|
221
|
-
# name → id 映射(从 state_attributes;支持子串匹配——卡名长,candidate 名短)
|
|
222
|
-
name2id = {}
|
|
223
|
-
name_pool = []
|
|
224
|
-
for (nid, sa) in dc.execute("SELECT id, state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
225
|
-
if not sa:
|
|
226
|
-
continue
|
|
227
|
-
try:
|
|
228
|
-
nm = json.loads(sa).get("name", "")
|
|
229
|
-
except Exception:
|
|
230
|
-
nm = ""
|
|
231
|
-
if nm:
|
|
232
|
-
name2id[nm] = nid
|
|
233
|
-
name_pool.append((nm, nid))
|
|
234
|
-
def resolve(name: str) -> Optional[str]:
|
|
235
|
-
if not name:
|
|
236
|
-
return None
|
|
237
|
-
if name in name2id:
|
|
238
|
-
return name2id[name]
|
|
239
|
-
# 子串匹配:candidate 短名 ⊂ 卡长名
|
|
240
|
-
for nm, nid in name_pool:
|
|
241
|
-
if name in nm or nm in name:
|
|
242
|
-
return nid
|
|
243
|
-
return None
|
|
244
|
-
# 元层卡的 name 在 state_attributes 里
|
|
245
|
-
for cand in cands:
|
|
246
|
-
a, b = cand.get("a", ""), cand.get("b", "")
|
|
247
|
-
sid, tid = resolve(a), resolve(b)
|
|
248
|
-
if not sid or not tid:
|
|
249
|
-
phase["skipped"] += 1
|
|
250
|
-
continue
|
|
251
|
-
if dc.execute("SELECT id FROM edges WHERE source_id=? AND target_id=? "
|
|
252
|
-
"AND relation_type=?", (sid, tid, cand.get("kind", "causal"))).fetchone():
|
|
253
|
-
phase["skipped"] += 1
|
|
254
|
-
continue
|
|
255
|
-
eid = f"edge_causal_{abs(hash(a+b)) % 10**10}_{int(time.time()*1000)}"
|
|
256
|
-
cs = json.dumps({"observation_position": "教育体系学段递进/学科归属",
|
|
257
|
-
"observation_tool": f"{cand.get('rule', 'R1')} 规则",
|
|
258
|
-
"time_window": [0.0, 9999999999.0],
|
|
259
|
-
"existence_constraint": cand.get("note", "")},
|
|
260
|
-
ensure_ascii=False)
|
|
261
|
-
now = time.time()
|
|
262
|
-
dc.execute(
|
|
263
|
-
"INSERT INTO edges (id, source_id, target_id, relation_type, condition_space, "
|
|
264
|
-
"confidence, weight, verified, created_at, last_verified, source_evidence) "
|
|
265
|
-
"VALUES (?,?,?,?,?,?,?,?,?,?,?)",
|
|
266
|
-
(eid, sid, tid, cand.get("kind", "causal"), cs,
|
|
267
|
-
cand.get("confidence", 0.75), 0.75,
|
|
268
|
-
0, now, None, "inferred"))
|
|
269
|
-
phase["edges"] += 1
|
|
270
|
-
dst.commit()
|
|
271
|
-
dst.close()
|
|
272
|
-
self.report["phases"]["causal"] = phase
|
|
273
|
-
return phase
|
|
274
|
-
|
|
275
|
-
# ---------------- 元学科聚合卡:7 张元学科卡 ----------------
|
|
276
|
-
def migrate_meta_discipline_cards(self,
|
|
277
|
-
skeleton_src: Optional[str] = None) -> Dict:
|
|
278
|
-
"""历史学/地理学/政治学/语言学/计算机科学/工程学/模拟电子技术 聚合卡。
|
|
279
|
-
数据源:knowledge_skeleton 子学科知识点并集(骨架锚点,内容待填充——
|
|
280
|
-
诚实标注,不伪造具体知识内容)。"""
|
|
281
|
-
phase = {"nodes": 0, "skipped": 0, "sources": []}
|
|
282
|
-
skeleton = json.load(open(skeleton_src or os.path.join(HERE, "knowledge_skeleton.json"),
|
|
283
|
-
encoding="utf-8"))
|
|
284
|
-
meta_disciplines = {
|
|
285
|
-
"计算机科学": ["程序设计", "数据结构", "算法设计与分析", "操作系统",
|
|
286
|
-
"计算机网络", "数据库系统", "编译原理", "软件工程",
|
|
287
|
-
"计算机组成原理", "人工智能"],
|
|
288
|
-
"工程学": ["材料力学", "电路原理", "模拟电子技术", "数字电路",
|
|
289
|
-
"工程制图", "土木工程基础"],
|
|
290
|
-
"语言学": ["古代汉语", "现代汉语", "文学理论"],
|
|
291
|
-
"历史学": ["历史"],
|
|
292
|
-
"地理学": ["地理"],
|
|
293
|
-
"政治学": ["政治"],
|
|
294
|
-
"模拟电子技术": ["模拟电子技术"],
|
|
295
|
-
}
|
|
296
|
-
dst = sqlite3.connect(self.dst)
|
|
297
|
-
dc = dst.cursor()
|
|
298
|
-
existing_names = set()
|
|
299
|
-
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
300
|
-
if not sa:
|
|
301
|
-
continue
|
|
302
|
-
try:
|
|
303
|
-
existing_names.add(json.loads(sa).get("name", ""))
|
|
304
|
-
except Exception:
|
|
305
|
-
pass
|
|
306
|
-
for name, subs in meta_disciplines.items():
|
|
307
|
-
if name in existing_names:
|
|
308
|
-
phase["skipped"] += 1
|
|
309
|
-
continue
|
|
310
|
-
kps = {}
|
|
311
|
-
for sub in subs:
|
|
312
|
-
sub_data = skeleton.get(sub)
|
|
313
|
-
if not sub_data:
|
|
314
|
-
continue
|
|
315
|
-
if isinstance(sub_data, dict):
|
|
316
|
-
for stage, units in sub_data.items():
|
|
317
|
-
for unit in units:
|
|
318
|
-
for kp in unit.get("知识点", []):
|
|
319
|
-
kps.setdefault(kp, f"[{sub}·{stage}] 骨架锚点,内容待填充")
|
|
320
|
-
elif isinstance(sub_data, list):
|
|
321
|
-
for unit in sub_data:
|
|
322
|
-
for kp in unit.get("知识点", []):
|
|
323
|
-
kps.setdefault(kp, f"[{sub}] 骨架锚点,内容待填充")
|
|
324
|
-
if not kps:
|
|
325
|
-
phase["sources"].append({"name": name, "error": "骨架无子学科知识点"})
|
|
326
|
-
continue
|
|
327
|
-
content_lines = [f"{name}(元学科聚合卡 · {len(kps)} 知识点,骨架锚点待填充)"]
|
|
328
|
-
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps.items())]
|
|
329
|
-
nid = f"node_meta_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
330
|
-
tags = [f"domain:{name}", "level:L4", "status:pending", "subject_card",
|
|
331
|
-
"meta_discipline"]
|
|
332
|
-
# #412(2026-10-16):**学科面 ≠ 纪律面**——元学科卡与纪律判据**同用**同一谓词做
|
|
333
|
-
# **负向自证**:这些卡**不得**被判为纪律节点(否则「学科卡被引用」会走硬冲突
|
|
334
|
-
# REJECT 路径,是明显误伤;#412 的语义面裁定即此)。将来若有人在此给卡片加纪律
|
|
335
|
-
# tag,本断言 fail-loud,而不是静默改变冲突语义。
|
|
336
|
-
_disc = _discipline_predicate()
|
|
337
|
-
phase["discipline_parity"] = ("checked" if _disc is not None
|
|
338
|
-
else "skipped_no_predicate")
|
|
339
|
-
if _disc is not None and _disc({"tags": tags}, nid):
|
|
340
|
-
raise RuntimeError(
|
|
341
|
-
"元学科卡被误判为纪律节点(#412:学科 ≠ 纪律):%r / %s" % (tags, nid))
|
|
342
|
-
sa = json.dumps({"name": name, "domain": name, "level": 4,
|
|
343
|
-
"kind": "meta_discipline", "source": "skeleton 聚合",
|
|
344
|
-
"kp_count": len(kps)}, ensure_ascii=False)
|
|
345
|
-
cs = json.dumps({"observation_position": f"{name} 学科体系观测位",
|
|
346
|
-
"observation_tool": "课标骨架聚合",
|
|
347
|
-
"time_window": [0.0, 9999999999.0],
|
|
348
|
-
"existence_constraint": "学科归属/结构关系(骨架锚点,内容待填充)"},
|
|
349
|
-
ensure_ascii=False)
|
|
350
|
-
now = time.time()
|
|
351
|
-
dc.execute(
|
|
352
|
-
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
353
|
-
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
354
|
-
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
355
|
-
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
356
|
-
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.7, 0.5,
|
|
357
|
-
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
358
|
-
"{}", sa))
|
|
359
|
-
phase["nodes"] += 1
|
|
360
|
-
existing_names.add(name)
|
|
361
|
-
dst.commit()
|
|
362
|
-
dst.close()
|
|
363
|
-
self.report["phases"]["meta_discipline"] = phase
|
|
364
|
-
return phase
|
|
365
|
-
|
|
366
|
-
# ---------------- 学段拆分卡:骨架×综合卡 → 学段卡 ----------------
|
|
367
|
-
def migrate_stage_cards(self, skeleton_src: Optional[str] = None) -> Dict:
|
|
368
|
-
"""knowledge_skeleton.json 学科×学段 × 综合卡知识点 → 学段卡。
|
|
369
|
-
补历史/地理/语文/英语/政治的学段拆分(33 条因果边跳过的原因)。
|
|
370
|
-
卡名含学段(如 初中历史)→ causal_edge_candidates 子串匹配自动接通。"""
|
|
371
|
-
phase = {"nodes": 0, "skipped": 0, "files": 0, "sources": []}
|
|
372
|
-
skeleton = json.load(open(skeleton_src or os.path.join(HERE, "knowledge_skeleton.json"),
|
|
373
|
-
encoding="utf-8"))
|
|
374
|
-
# 学科 → (综合卡文件, 学段→卡名规则)
|
|
375
|
-
card_files = {
|
|
376
|
-
"语文": "chinese_knowledge.json",
|
|
377
|
-
"英语": "english_knowledge.json",
|
|
378
|
-
"历史": "history_knowledge.json",
|
|
379
|
-
"地理": "geography_knowledge.json",
|
|
380
|
-
"政治": "politics_knowledge.json",
|
|
381
|
-
}
|
|
382
|
-
name_rule = {
|
|
383
|
-
"政治": lambda s: "小学道德与法治" if s == "小学" else (
|
|
384
|
-
"初中道德与法治" if s == "初中" else "高中思想政治"),
|
|
385
|
-
}
|
|
386
|
-
dst = sqlite3.connect(self.dst)
|
|
387
|
-
dc = dst.cursor()
|
|
388
|
-
existing_names = set()
|
|
389
|
-
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
390
|
-
if not sa:
|
|
391
|
-
continue
|
|
392
|
-
try:
|
|
393
|
-
existing_names.add(json.loads(sa).get("name", ""))
|
|
394
|
-
except Exception:
|
|
395
|
-
pass
|
|
396
|
-
import re
|
|
397
|
-
for subject, fn in card_files.items():
|
|
398
|
-
path = os.path.join(self.subject_dir, fn)
|
|
399
|
-
if not os.path.exists(path):
|
|
400
|
-
phase["sources"].append({"subject": subject, "error": "综合卡缺失"})
|
|
401
|
-
continue
|
|
402
|
-
try:
|
|
403
|
-
content_map = json.load(open(path, encoding="utf-8")).get("content", {})
|
|
404
|
-
except Exception as e:
|
|
405
|
-
phase["sources"].append({"subject": subject, "error": str(e)[:40]})
|
|
406
|
-
continue
|
|
407
|
-
stages = skeleton.get(subject, {})
|
|
408
|
-
for stage, units in stages.items():
|
|
409
|
-
kp_names = []
|
|
410
|
-
for unit in units:
|
|
411
|
-
kp_names.extend(unit.get("知识点", []))
|
|
412
|
-
kps = {k: content_map[k] for k in kp_names if k in content_map}
|
|
413
|
-
if not kps:
|
|
414
|
-
phase["sources"].append({"subject": subject, "stage": stage,
|
|
415
|
-
"error": "综合卡无对应知识点"})
|
|
416
|
-
continue
|
|
417
|
-
name = name_rule.get(subject, lambda s: s + subject)(stage)
|
|
418
|
-
if name in existing_names:
|
|
419
|
-
phase["skipped"] += 1
|
|
420
|
-
continue
|
|
421
|
-
domain = subject
|
|
422
|
-
edu = {"小学": "E1", "初中": "E2", "高中": "E3"}.get(stage, "E2")
|
|
423
|
-
level = {"E1": 1, "E2": 2, "E3": 3}.get(edu, 2)
|
|
424
|
-
content_lines = [f"{name}(学段知识卡 · {len(kps)} 知识点,源自骨架 {subject}.{stage})"]
|
|
425
|
-
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps.items())]
|
|
426
|
-
nid = f"node_stage_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
427
|
-
tags = [f"domain:{domain}", f"level:L{level}", f"status:verified",
|
|
428
|
-
"subject_card", f"edu:{edu}", "stage_card"]
|
|
429
|
-
sa = json.dumps({"name": name, "domain": domain, "level": level,
|
|
430
|
-
"edu_level": edu, "kind": "stage_card",
|
|
431
|
-
"source": "skeleton+%s" % fn,
|
|
432
|
-
"kp_count": len(kps)}, ensure_ascii=False)
|
|
433
|
-
cs = json.dumps({"observation_position": f"{name} 外部观测位",
|
|
434
|
-
"observation_tool": "课标骨架×综合卡",
|
|
435
|
-
"time_window": [0.0, 9999999999.0],
|
|
436
|
-
"existence_constraint": "通用规律/知识(开源非盈利知识库)"},
|
|
437
|
-
ensure_ascii=False)
|
|
438
|
-
now = time.time()
|
|
439
|
-
dc.execute(
|
|
440
|
-
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
441
|
-
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
442
|
-
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
443
|
-
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
444
|
-
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.8, 0.7,
|
|
445
|
-
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
446
|
-
"{}", sa))
|
|
447
|
-
phase["nodes"] += 1
|
|
448
|
-
existing_names.add(name)
|
|
449
|
-
dst.commit()
|
|
450
|
-
dst.close()
|
|
451
|
-
self.report["phases"]["stage"] = phase
|
|
452
|
-
return phase
|
|
453
|
-
|
|
454
|
-
# ---------------- 总报告 ----------------
|
|
455
|
-
def run(self, meta: bool = True, subject: bool = True,
|
|
456
|
-
causal: bool = True, stage: bool = True,
|
|
457
|
-
meta_discipline: bool = True, persist: bool = True) -> Dict:
|
|
458
|
-
"""执行完整迁移流水线:meta→学科卡→因果→阶段→纪律→持久化(各相由布尔开关独立启停)。"""
|
|
459
|
-
if meta:
|
|
460
|
-
self.migrate_meta()
|
|
461
|
-
if subject:
|
|
462
|
-
self.migrate_subject_cards()
|
|
463
|
-
if stage:
|
|
464
|
-
self.migrate_stage_cards()
|
|
465
|
-
if meta_discipline:
|
|
466
|
-
self.migrate_meta_discipline_cards()
|
|
467
|
-
if causal:
|
|
468
|
-
self.migrate_causal_edges()
|
|
469
|
-
# 目标库终态统计
|
|
470
|
-
dst = sqlite3.connect(self.dst)
|
|
471
|
-
dc = dst.cursor()
|
|
472
|
-
self.report["dst_final"] = {
|
|
473
|
-
"nodes": dc.execute("SELECT COUNT(*) FROM nodes").fetchone()[0],
|
|
474
|
-
"knowledge_nodes": dc.execute(
|
|
475
|
-
"SELECT COUNT(*) FROM nodes WHERE layer='knowledge'").fetchone()[0],
|
|
476
|
-
"edges": dc.execute("SELECT COUNT(*) FROM edges").fetchone()[0],
|
|
477
|
-
"causal_edges": dc.execute(
|
|
478
|
-
"SELECT COUNT(*) FROM edges WHERE relation_type='causal'").fetchone()[0],
|
|
479
|
-
"verified_edges": dc.execute(
|
|
480
|
-
"SELECT COUNT(*) FROM edges WHERE verified=1").fetchone()[0],
|
|
481
|
-
}
|
|
482
|
-
dst.close()
|
|
483
|
-
self.report["generated_at"] = time.strftime("%Y-%m-%d %H:%M:%S")
|
|
484
|
-
self.report["source"] = "migrate_wisdom v1.16(智慧之书→灵枢主库)"
|
|
485
|
-
if persist:
|
|
486
|
-
path = os.path.join(HERE, "migrate_wisdom_report.json")
|
|
487
|
-
with open(path, "w", encoding="utf-8") as f:
|
|
488
|
-
json.dump(self.report, f, ensure_ascii=False, indent=2)
|
|
489
|
-
self.report["report_path"] = path
|
|
490
|
-
return self.report
|
|
491
|
-
|
|
492
|
-
|
|
493
|
-
if __name__ == "__main__":
|
|
494
|
-
sys.stdout.reconfigure(encoding="utf-8")
|
|
495
|
-
mig = WisdomMigrator()
|
|
496
|
-
rep = mig.run()
|
|
497
|
-
print(json.dumps(rep, ensure_ascii=False, indent=2))
|
|
1
|
+
#!/usr/bin/env python3
|
|
2
|
+
# -*- coding: utf-8 -*-
|
|
3
|
+
"""
|
|
4
|
+
migrate_wisdom · 智慧之书 → 灵枢主库迁移器(v1.16 · 统一信息差减少任务)
|
|
5
|
+
|
|
6
|
+
设计者方向(2026-08-18):智慧之书库迁移进灵枢主库,灵枢成为完整工程实现
|
|
7
|
+
(灵魂=智能论3.2 协议层 + 神经系统功能类比),统一「信息差减少」任务。
|
|
8
|
+
|
|
9
|
+
知识真身分层(迁移数据源):
|
|
10
|
+
META 层:wisdom-book-cloud.db 30 元学科卡(存在论/条件论/智能论/学科映射等)
|
|
11
|
+
87 边(31 causal + 56 similar,79 verified)
|
|
12
|
+
SUBJECT 层:knowledge-base/*_knowledge.json 51 学科卡(E1-E4,23-81 知识点/卡)
|
|
13
|
+
+ programming_lang_cards.json 3 语言卡
|
|
14
|
+
CAUSAL 层:causal_edge_candidates.json 学段递进因果边(R1 依赖型因果)
|
|
15
|
+
+ 元层 87 边中的 causal 边
|
|
16
|
+
|
|
17
|
+
迁移原则:
|
|
18
|
+
- 幂等:按卡名(state_attributes.name)查重,已存在跳过
|
|
19
|
+
- 可逆:迁移前备份目标库
|
|
20
|
+
- 诚实:verified 标记如实保留(元层边已验证的保持 verified;
|
|
21
|
+
学段递进边 R1 规则推导 → verified=0 confidence 0.75)
|
|
22
|
+
- 报告:迁移统计输出 migrate_wisdom_report.json(可复现)
|
|
23
|
+
纯标准库 · 零外部依赖
|
|
24
|
+
"""
|
|
25
|
+
|
|
26
|
+
import json
|
|
27
|
+
import os
|
|
28
|
+
import shutil
|
|
29
|
+
import sqlite3
|
|
30
|
+
import sys
|
|
31
|
+
import time
|
|
32
|
+
from typing import Dict, List, Optional
|
|
33
|
+
|
|
34
|
+
HERE = os.path.dirname(os.path.abspath(__file__))
|
|
35
|
+
|
|
36
|
+
|
|
37
|
+
# 生效条件:包内相对导入成功时返回 md_cg.consistency.is_discipline_node;失败(无包上下文/导入面异常)
|
|
38
|
+
# 返回 None——**退化可见**:调用方必须显式判 None 并把结论写进 phase(不静默假装通过)。
|
|
39
|
+
def _discipline_predicate():
|
|
40
|
+
"""纪律判据单点(#412 负向自证用):与判据**同用**同一个谓词,本文件不复制词形表。"""
|
|
41
|
+
try:
|
|
42
|
+
from ...consistency import is_discipline_node # 包内路径:md_cg.consistency
|
|
43
|
+
return is_discipline_node
|
|
44
|
+
except Exception: # noqa: BLE001 —— 独立运行时无包上下文
|
|
45
|
+
return None
|
|
46
|
+
sys.path.insert(0, HERE)
|
|
47
|
+
|
|
48
|
+
# 默认路径(可参数覆盖);目标库走 LINGSHU_DB 环境变量,否则家目录 .dsh 布局
|
|
49
|
+
DEFAULT_META_SRC = os.path.join(os.path.dirname(os.path.abspath(__file__)), "wisdom-book-cloud.db")
|
|
50
|
+
DEFAULT_DST = os.environ.get("LINGSHU_DB") or os.path.join(
|
|
51
|
+
os.path.expanduser("~"), ".dsh", "profiles", "web", "data", "lingshu.db")
|
|
52
|
+
DEFAULT_CAUSAL = os.path.join(HERE, "causal_edge_candidates.json")
|
|
53
|
+
|
|
54
|
+
|
|
55
|
+
class WisdomMigrator:
|
|
56
|
+
"""智慧之书 → 灵枢主库迁移器"""
|
|
57
|
+
|
|
58
|
+
def __init__(self, meta_src: str = DEFAULT_META_SRC,
|
|
59
|
+
dst: str = DEFAULT_DST,
|
|
60
|
+
subject_dir: Optional[str] = None,
|
|
61
|
+
causal_src: str = DEFAULT_CAUSAL,
|
|
62
|
+
backup: bool = True):
|
|
63
|
+
"""迁移会话初始化:源库·目标库路径与游标准备。"""
|
|
64
|
+
self.meta_src = meta_src
|
|
65
|
+
self.dst = dst
|
|
66
|
+
self.subject_dir = subject_dir or HERE
|
|
67
|
+
self.causal_src = causal_src
|
|
68
|
+
self.report: Dict = {"phases": {}}
|
|
69
|
+
if backup and os.path.exists(dst):
|
|
70
|
+
bak = dst + f".bak_migrate_{time.strftime('%Y%m%d_%H%M%S')}"
|
|
71
|
+
shutil.copy2(dst, bak)
|
|
72
|
+
self.report["backup"] = bak
|
|
73
|
+
|
|
74
|
+
# ---------------- META 层:30 元卡 + 87 边 ----------------
|
|
75
|
+
def migrate_meta(self) -> Dict:
|
|
76
|
+
phase = {"nodes": 0, "edges": 0, "skipped_nodes": 0, "skipped_edges": 0}
|
|
77
|
+
if not os.path.exists(self.meta_src):
|
|
78
|
+
phase["error"] = f"源库不存在: {self.meta_src}"
|
|
79
|
+
self.report["phases"]["meta"] = phase
|
|
80
|
+
return phase
|
|
81
|
+
src = sqlite3.connect(self.meta_src)
|
|
82
|
+
dst = sqlite3.connect(self.dst)
|
|
83
|
+
sc, dc = src.cursor(), dst.cursor()
|
|
84
|
+
# 已存在的节点 id 集合(幂等)
|
|
85
|
+
existing = {r[0] for r in dc.execute("SELECT id FROM nodes").fetchall()}
|
|
86
|
+
# 搬节点
|
|
87
|
+
rows = sc.execute("SELECT * FROM nodes").fetchall()
|
|
88
|
+
cols = [r[1] for r in sc.execute("PRAGMA table_info(nodes)").fetchall()]
|
|
89
|
+
for row in rows:
|
|
90
|
+
nid = row[0]
|
|
91
|
+
if nid in existing:
|
|
92
|
+
phase["skipped_nodes"] += 1
|
|
93
|
+
continue
|
|
94
|
+
dc.execute(
|
|
95
|
+
"INSERT INTO nodes (%s) VALUES (%s)" %
|
|
96
|
+
(",".join(cols), ",".join("?" * len(cols))), row)
|
|
97
|
+
phase["nodes"] += 1
|
|
98
|
+
# 搬边(仅两端都存在于目标库的边)
|
|
99
|
+
erows = sc.execute("SELECT * FROM edges").fetchall()
|
|
100
|
+
ecols = [r[1] for r in sc.execute("PRAGMA table_info(edges)").fetchall()]
|
|
101
|
+
eexisting = {r[0] for r in dc.execute("SELECT id FROM edges").fetchall()}
|
|
102
|
+
for row in erows:
|
|
103
|
+
eid, sid, tid = row[0], row[1], row[2]
|
|
104
|
+
if eid in eexisting:
|
|
105
|
+
phase["skipped_edges"] += 1
|
|
106
|
+
continue
|
|
107
|
+
if not dc.execute("SELECT id FROM nodes WHERE id=?", (sid,)).fetchone():
|
|
108
|
+
phase["skipped_edges"] += 1
|
|
109
|
+
continue
|
|
110
|
+
if not dc.execute("SELECT id FROM nodes WHERE id=?", (tid,)).fetchone():
|
|
111
|
+
phase["skipped_edges"] += 1
|
|
112
|
+
continue
|
|
113
|
+
dc.execute(
|
|
114
|
+
"INSERT INTO edges (%s) VALUES (%s)" %
|
|
115
|
+
(",".join(ecols), ",".join("?" * len(ecols))), row)
|
|
116
|
+
phase["edges"] += 1
|
|
117
|
+
dst.commit()
|
|
118
|
+
# verified 统计
|
|
119
|
+
phase["verified_edges"] = dc.execute(
|
|
120
|
+
"SELECT COUNT(*) FROM edges WHERE verified=1").fetchone()[0]
|
|
121
|
+
src.close()
|
|
122
|
+
dst.close()
|
|
123
|
+
self.report["phases"]["meta"] = phase
|
|
124
|
+
return phase
|
|
125
|
+
|
|
126
|
+
# ---------------- SUBJECT 层:51 学科卡 + 语言卡 ----------------
|
|
127
|
+
def migrate_subject_cards(self) -> Dict:
|
|
128
|
+
"""学科卡批量迁移:逐识别卡解析→按名去重→写入目标 nodes 表,返回 nodes·skipped·files 迁移统计。"""
|
|
129
|
+
phase = {"nodes": 0, "skipped": 0, "files": 0, "sources": []}
|
|
130
|
+
dst = sqlite3.connect(self.dst)
|
|
131
|
+
dc = dst.cursor()
|
|
132
|
+
# 已存在的卡名(幂等,从 state_attributes.name 解析)
|
|
133
|
+
existing_names = set()
|
|
134
|
+
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
135
|
+
if not sa:
|
|
136
|
+
continue
|
|
137
|
+
try:
|
|
138
|
+
existing_names.add(json.loads(sa).get("name", ""))
|
|
139
|
+
except Exception:
|
|
140
|
+
pass
|
|
141
|
+
edu_level = {"E1": 1, "E2": 2, "E3": 3, "E4": 4, "E5": 5}
|
|
142
|
+
files = sorted(f for f in os.listdir(self.subject_dir)
|
|
143
|
+
if f.endswith("_knowledge.json") or f == "programming_lang_cards.json")
|
|
144
|
+
import re
|
|
145
|
+
for fn in files:
|
|
146
|
+
path = os.path.join(self.subject_dir, fn)
|
|
147
|
+
try:
|
|
148
|
+
d = json.load(open(path, encoding="utf-8"))
|
|
149
|
+
except Exception as e:
|
|
150
|
+
phase["sources"].append({"file": fn, "error": str(e)[:40]})
|
|
151
|
+
continue
|
|
152
|
+
phase["files"] += 1
|
|
153
|
+
# dict(学科卡)或 list(语言卡)统一为卡列表
|
|
154
|
+
cards = d if isinstance(d, list) else [d]
|
|
155
|
+
for card in cards:
|
|
156
|
+
content_map = card.get("content", {}) if isinstance(card, dict) else {}
|
|
157
|
+
meta = card.get("meta", {}) if isinstance(card, dict) else {}
|
|
158
|
+
name = meta.get("name") or card.get("name") or fn.replace("_knowledge.json", "")
|
|
159
|
+
if not name or name in existing_names:
|
|
160
|
+
phase["skipped"] += 1
|
|
161
|
+
continue
|
|
162
|
+
style = meta.get("style", "")
|
|
163
|
+
m = re.search(r"(E\d)", style)
|
|
164
|
+
edu = m.group(1) if m else card.get("edu_level")
|
|
165
|
+
level = card.get("level") or {"E1": 1, "E2": 2, "E3": 3, "E4": 4, "E5": 5}.get(edu, 2)
|
|
166
|
+
domain = (meta.get("domain") or card.get("domain")
|
|
167
|
+
or name.split(" ")[0])
|
|
168
|
+
if content_map and isinstance(content_map, dict):
|
|
169
|
+
kps = list(content_map.items())
|
|
170
|
+
content_lines = [f"{name}(学科知识卡 · {len(kps)} 知识点)"]
|
|
171
|
+
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps)]
|
|
172
|
+
else:
|
|
173
|
+
# 语言卡:response 驱动(trigger/action/counters)
|
|
174
|
+
resp = card.get("response", {})
|
|
175
|
+
content_lines = [f"{name}(语言卡 · {card.get('status', 'pending')})"]
|
|
176
|
+
if isinstance(content_map, str) and content_map:
|
|
177
|
+
content_lines.append(content_map[:300])
|
|
178
|
+
content_lines.append("触发: " + str(resp.get("trigger", ""))[:200])
|
|
179
|
+
content_lines.append("动作: " + str(resp.get("action", ""))[:200])
|
|
180
|
+
content_lines.append("护栏: " + str(resp.get("counters", ""))[:200])
|
|
181
|
+
nid = f"node_subject_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
182
|
+
tags = [f"domain:{domain}", f"level:L{level}",
|
|
183
|
+
f"status:verified", "subject_card",
|
|
184
|
+
f"edu:{edu}" if edu else "edu:unknown"]
|
|
185
|
+
sa = json.dumps({"name": name, "domain": domain, "level": level,
|
|
186
|
+
"edu_level": edu, "kind": "subject_card",
|
|
187
|
+
"source": fn,
|
|
188
|
+
"kp_count": len(content_map) if isinstance(content_map, dict) else 0},
|
|
189
|
+
ensure_ascii=False)
|
|
190
|
+
cs = json.dumps({"observation_position": f"{name} 外部观测位",
|
|
191
|
+
"observation_tool": "学科知识卡(知识综述)",
|
|
192
|
+
"time_window": [0.0, 9999999999.0],
|
|
193
|
+
"existence_constraint": "通用规律/知识(开源非盈利知识库)"},
|
|
194
|
+
ensure_ascii=False)
|
|
195
|
+
now = time.time()
|
|
196
|
+
dc.execute(
|
|
197
|
+
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
198
|
+
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
199
|
+
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
200
|
+
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
201
|
+
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.8, 0.7,
|
|
202
|
+
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
203
|
+
"{}", sa))
|
|
204
|
+
phase["nodes"] += 1
|
|
205
|
+
existing_names.add(name)
|
|
206
|
+
dst.commit()
|
|
207
|
+
dst.close()
|
|
208
|
+
self.report["phases"]["subject"] = phase
|
|
209
|
+
return phase
|
|
210
|
+
|
|
211
|
+
# ---------------- CAUSAL 层:学段递进因果边 ----------------
|
|
212
|
+
def migrate_causal_edges(self) -> Dict:
|
|
213
|
+
phase = {"edges": 0, "skipped": 0}
|
|
214
|
+
if not os.path.exists(self.causal_src):
|
|
215
|
+
phase["error"] = f"因果边源不存在: {self.causal_src}"
|
|
216
|
+
self.report["phases"]["causal"] = phase
|
|
217
|
+
return phase
|
|
218
|
+
cands = json.load(open(self.causal_src, encoding="utf-8"))
|
|
219
|
+
dst = sqlite3.connect(self.dst)
|
|
220
|
+
dc = dst.cursor()
|
|
221
|
+
# name → id 映射(从 state_attributes;支持子串匹配——卡名长,candidate 名短)
|
|
222
|
+
name2id = {}
|
|
223
|
+
name_pool = []
|
|
224
|
+
for (nid, sa) in dc.execute("SELECT id, state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
225
|
+
if not sa:
|
|
226
|
+
continue
|
|
227
|
+
try:
|
|
228
|
+
nm = json.loads(sa).get("name", "")
|
|
229
|
+
except Exception:
|
|
230
|
+
nm = ""
|
|
231
|
+
if nm:
|
|
232
|
+
name2id[nm] = nid
|
|
233
|
+
name_pool.append((nm, nid))
|
|
234
|
+
def resolve(name: str) -> Optional[str]:
|
|
235
|
+
if not name:
|
|
236
|
+
return None
|
|
237
|
+
if name in name2id:
|
|
238
|
+
return name2id[name]
|
|
239
|
+
# 子串匹配:candidate 短名 ⊂ 卡长名
|
|
240
|
+
for nm, nid in name_pool:
|
|
241
|
+
if name in nm or nm in name:
|
|
242
|
+
return nid
|
|
243
|
+
return None
|
|
244
|
+
# 元层卡的 name 在 state_attributes 里
|
|
245
|
+
for cand in cands:
|
|
246
|
+
a, b = cand.get("a", ""), cand.get("b", "")
|
|
247
|
+
sid, tid = resolve(a), resolve(b)
|
|
248
|
+
if not sid or not tid:
|
|
249
|
+
phase["skipped"] += 1
|
|
250
|
+
continue
|
|
251
|
+
if dc.execute("SELECT id FROM edges WHERE source_id=? AND target_id=? "
|
|
252
|
+
"AND relation_type=?", (sid, tid, cand.get("kind", "causal"))).fetchone():
|
|
253
|
+
phase["skipped"] += 1
|
|
254
|
+
continue
|
|
255
|
+
eid = f"edge_causal_{abs(hash(a+b)) % 10**10}_{int(time.time()*1000)}"
|
|
256
|
+
cs = json.dumps({"observation_position": "教育体系学段递进/学科归属",
|
|
257
|
+
"observation_tool": f"{cand.get('rule', 'R1')} 规则",
|
|
258
|
+
"time_window": [0.0, 9999999999.0],
|
|
259
|
+
"existence_constraint": cand.get("note", "")},
|
|
260
|
+
ensure_ascii=False)
|
|
261
|
+
now = time.time()
|
|
262
|
+
dc.execute(
|
|
263
|
+
"INSERT INTO edges (id, source_id, target_id, relation_type, condition_space, "
|
|
264
|
+
"confidence, weight, verified, created_at, last_verified, source_evidence) "
|
|
265
|
+
"VALUES (?,?,?,?,?,?,?,?,?,?,?)",
|
|
266
|
+
(eid, sid, tid, cand.get("kind", "causal"), cs,
|
|
267
|
+
cand.get("confidence", 0.75), 0.75,
|
|
268
|
+
0, now, None, "inferred"))
|
|
269
|
+
phase["edges"] += 1
|
|
270
|
+
dst.commit()
|
|
271
|
+
dst.close()
|
|
272
|
+
self.report["phases"]["causal"] = phase
|
|
273
|
+
return phase
|
|
274
|
+
|
|
275
|
+
# ---------------- 元学科聚合卡:7 张元学科卡 ----------------
|
|
276
|
+
def migrate_meta_discipline_cards(self,
|
|
277
|
+
skeleton_src: Optional[str] = None) -> Dict:
|
|
278
|
+
"""历史学/地理学/政治学/语言学/计算机科学/工程学/模拟电子技术 聚合卡。
|
|
279
|
+
数据源:knowledge_skeleton 子学科知识点并集(骨架锚点,内容待填充——
|
|
280
|
+
诚实标注,不伪造具体知识内容)。"""
|
|
281
|
+
phase = {"nodes": 0, "skipped": 0, "sources": []}
|
|
282
|
+
skeleton = json.load(open(skeleton_src or os.path.join(HERE, "knowledge_skeleton.json"),
|
|
283
|
+
encoding="utf-8"))
|
|
284
|
+
meta_disciplines = {
|
|
285
|
+
"计算机科学": ["程序设计", "数据结构", "算法设计与分析", "操作系统",
|
|
286
|
+
"计算机网络", "数据库系统", "编译原理", "软件工程",
|
|
287
|
+
"计算机组成原理", "人工智能"],
|
|
288
|
+
"工程学": ["材料力学", "电路原理", "模拟电子技术", "数字电路",
|
|
289
|
+
"工程制图", "土木工程基础"],
|
|
290
|
+
"语言学": ["古代汉语", "现代汉语", "文学理论"],
|
|
291
|
+
"历史学": ["历史"],
|
|
292
|
+
"地理学": ["地理"],
|
|
293
|
+
"政治学": ["政治"],
|
|
294
|
+
"模拟电子技术": ["模拟电子技术"],
|
|
295
|
+
}
|
|
296
|
+
dst = sqlite3.connect(self.dst)
|
|
297
|
+
dc = dst.cursor()
|
|
298
|
+
existing_names = set()
|
|
299
|
+
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
300
|
+
if not sa:
|
|
301
|
+
continue
|
|
302
|
+
try:
|
|
303
|
+
existing_names.add(json.loads(sa).get("name", ""))
|
|
304
|
+
except Exception:
|
|
305
|
+
pass
|
|
306
|
+
for name, subs in meta_disciplines.items():
|
|
307
|
+
if name in existing_names:
|
|
308
|
+
phase["skipped"] += 1
|
|
309
|
+
continue
|
|
310
|
+
kps = {}
|
|
311
|
+
for sub in subs:
|
|
312
|
+
sub_data = skeleton.get(sub)
|
|
313
|
+
if not sub_data:
|
|
314
|
+
continue
|
|
315
|
+
if isinstance(sub_data, dict):
|
|
316
|
+
for stage, units in sub_data.items():
|
|
317
|
+
for unit in units:
|
|
318
|
+
for kp in unit.get("知识点", []):
|
|
319
|
+
kps.setdefault(kp, f"[{sub}·{stage}] 骨架锚点,内容待填充")
|
|
320
|
+
elif isinstance(sub_data, list):
|
|
321
|
+
for unit in sub_data:
|
|
322
|
+
for kp in unit.get("知识点", []):
|
|
323
|
+
kps.setdefault(kp, f"[{sub}] 骨架锚点,内容待填充")
|
|
324
|
+
if not kps:
|
|
325
|
+
phase["sources"].append({"name": name, "error": "骨架无子学科知识点"})
|
|
326
|
+
continue
|
|
327
|
+
content_lines = [f"{name}(元学科聚合卡 · {len(kps)} 知识点,骨架锚点待填充)"]
|
|
328
|
+
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps.items())]
|
|
329
|
+
nid = f"node_meta_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
330
|
+
tags = [f"domain:{name}", "level:L4", "status:pending", "subject_card",
|
|
331
|
+
"meta_discipline"]
|
|
332
|
+
# #412(2026-10-16):**学科面 ≠ 纪律面**——元学科卡与纪律判据**同用**同一谓词做
|
|
333
|
+
# **负向自证**:这些卡**不得**被判为纪律节点(否则「学科卡被引用」会走硬冲突
|
|
334
|
+
# REJECT 路径,是明显误伤;#412 的语义面裁定即此)。将来若有人在此给卡片加纪律
|
|
335
|
+
# tag,本断言 fail-loud,而不是静默改变冲突语义。
|
|
336
|
+
_disc = _discipline_predicate()
|
|
337
|
+
phase["discipline_parity"] = ("checked" if _disc is not None
|
|
338
|
+
else "skipped_no_predicate")
|
|
339
|
+
if _disc is not None and _disc({"tags": tags}, nid):
|
|
340
|
+
raise RuntimeError(
|
|
341
|
+
"元学科卡被误判为纪律节点(#412:学科 ≠ 纪律):%r / %s" % (tags, nid))
|
|
342
|
+
sa = json.dumps({"name": name, "domain": name, "level": 4,
|
|
343
|
+
"kind": "meta_discipline", "source": "skeleton 聚合",
|
|
344
|
+
"kp_count": len(kps)}, ensure_ascii=False)
|
|
345
|
+
cs = json.dumps({"observation_position": f"{name} 学科体系观测位",
|
|
346
|
+
"observation_tool": "课标骨架聚合",
|
|
347
|
+
"time_window": [0.0, 9999999999.0],
|
|
348
|
+
"existence_constraint": "学科归属/结构关系(骨架锚点,内容待填充)"},
|
|
349
|
+
ensure_ascii=False)
|
|
350
|
+
now = time.time()
|
|
351
|
+
dc.execute(
|
|
352
|
+
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
353
|
+
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
354
|
+
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
355
|
+
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
356
|
+
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.7, 0.5,
|
|
357
|
+
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
358
|
+
"{}", sa))
|
|
359
|
+
phase["nodes"] += 1
|
|
360
|
+
existing_names.add(name)
|
|
361
|
+
dst.commit()
|
|
362
|
+
dst.close()
|
|
363
|
+
self.report["phases"]["meta_discipline"] = phase
|
|
364
|
+
return phase
|
|
365
|
+
|
|
366
|
+
# ---------------- 学段拆分卡:骨架×综合卡 → 学段卡 ----------------
|
|
367
|
+
def migrate_stage_cards(self, skeleton_src: Optional[str] = None) -> Dict:
|
|
368
|
+
"""knowledge_skeleton.json 学科×学段 × 综合卡知识点 → 学段卡。
|
|
369
|
+
补历史/地理/语文/英语/政治的学段拆分(33 条因果边跳过的原因)。
|
|
370
|
+
卡名含学段(如 初中历史)→ causal_edge_candidates 子串匹配自动接通。"""
|
|
371
|
+
phase = {"nodes": 0, "skipped": 0, "files": 0, "sources": []}
|
|
372
|
+
skeleton = json.load(open(skeleton_src or os.path.join(HERE, "knowledge_skeleton.json"),
|
|
373
|
+
encoding="utf-8"))
|
|
374
|
+
# 学科 → (综合卡文件, 学段→卡名规则)
|
|
375
|
+
card_files = {
|
|
376
|
+
"语文": "chinese_knowledge.json",
|
|
377
|
+
"英语": "english_knowledge.json",
|
|
378
|
+
"历史": "history_knowledge.json",
|
|
379
|
+
"地理": "geography_knowledge.json",
|
|
380
|
+
"政治": "politics_knowledge.json",
|
|
381
|
+
}
|
|
382
|
+
name_rule = {
|
|
383
|
+
"政治": lambda s: "小学道德与法治" if s == "小学" else (
|
|
384
|
+
"初中道德与法治" if s == "初中" else "高中思想政治"),
|
|
385
|
+
}
|
|
386
|
+
dst = sqlite3.connect(self.dst)
|
|
387
|
+
dc = dst.cursor()
|
|
388
|
+
existing_names = set()
|
|
389
|
+
for (sa,) in dc.execute("SELECT state_attributes FROM nodes WHERE layer='knowledge'").fetchall():
|
|
390
|
+
if not sa:
|
|
391
|
+
continue
|
|
392
|
+
try:
|
|
393
|
+
existing_names.add(json.loads(sa).get("name", ""))
|
|
394
|
+
except Exception:
|
|
395
|
+
pass
|
|
396
|
+
import re
|
|
397
|
+
for subject, fn in card_files.items():
|
|
398
|
+
path = os.path.join(self.subject_dir, fn)
|
|
399
|
+
if not os.path.exists(path):
|
|
400
|
+
phase["sources"].append({"subject": subject, "error": "综合卡缺失"})
|
|
401
|
+
continue
|
|
402
|
+
try:
|
|
403
|
+
content_map = json.load(open(path, encoding="utf-8")).get("content", {})
|
|
404
|
+
except Exception as e:
|
|
405
|
+
phase["sources"].append({"subject": subject, "error": str(e)[:40]})
|
|
406
|
+
continue
|
|
407
|
+
stages = skeleton.get(subject, {})
|
|
408
|
+
for stage, units in stages.items():
|
|
409
|
+
kp_names = []
|
|
410
|
+
for unit in units:
|
|
411
|
+
kp_names.extend(unit.get("知识点", []))
|
|
412
|
+
kps = {k: content_map[k] for k in kp_names if k in content_map}
|
|
413
|
+
if not kps:
|
|
414
|
+
phase["sources"].append({"subject": subject, "stage": stage,
|
|
415
|
+
"error": "综合卡无对应知识点"})
|
|
416
|
+
continue
|
|
417
|
+
name = name_rule.get(subject, lambda s: s + subject)(stage)
|
|
418
|
+
if name in existing_names:
|
|
419
|
+
phase["skipped"] += 1
|
|
420
|
+
continue
|
|
421
|
+
domain = subject
|
|
422
|
+
edu = {"小学": "E1", "初中": "E2", "高中": "E3"}.get(stage, "E2")
|
|
423
|
+
level = {"E1": 1, "E2": 2, "E3": 3}.get(edu, 2)
|
|
424
|
+
content_lines = [f"{name}(学段知识卡 · {len(kps)} 知识点,源自骨架 {subject}.{stage})"]
|
|
425
|
+
content_lines += [f"{i+1}. {k}: {v}" for i, (k, v) in enumerate(kps.items())]
|
|
426
|
+
nid = f"node_stage_{abs(hash(name)) % 10**10}_{int(time.time()*1000)}"
|
|
427
|
+
tags = [f"domain:{domain}", f"level:L{level}", f"status:verified",
|
|
428
|
+
"subject_card", f"edu:{edu}", "stage_card"]
|
|
429
|
+
sa = json.dumps({"name": name, "domain": domain, "level": level,
|
|
430
|
+
"edu_level": edu, "kind": "stage_card",
|
|
431
|
+
"source": "skeleton+%s" % fn,
|
|
432
|
+
"kp_count": len(kps)}, ensure_ascii=False)
|
|
433
|
+
cs = json.dumps({"observation_position": f"{name} 外部观测位",
|
|
434
|
+
"observation_tool": "课标骨架×综合卡",
|
|
435
|
+
"time_window": [0.0, 9999999999.0],
|
|
436
|
+
"existence_constraint": "通用规律/知识(开源非盈利知识库)"},
|
|
437
|
+
ensure_ascii=False)
|
|
438
|
+
now = time.time()
|
|
439
|
+
dc.execute(
|
|
440
|
+
"INSERT INTO nodes (id, content, modality, spatial_coordinates, "
|
|
441
|
+
"temporal_coordinate, condition_space, importance, confidence, layer, "
|
|
442
|
+
"access_count, last_access, created_at, tags, semantic_coordinates, "
|
|
443
|
+
"state_attributes) VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)",
|
|
444
|
+
(nid, "\n".join(content_lines), "text", "{}", now, cs, 0.8, 0.7,
|
|
445
|
+
"knowledge", 0, None, now, json.dumps(tags, ensure_ascii=False),
|
|
446
|
+
"{}", sa))
|
|
447
|
+
phase["nodes"] += 1
|
|
448
|
+
existing_names.add(name)
|
|
449
|
+
dst.commit()
|
|
450
|
+
dst.close()
|
|
451
|
+
self.report["phases"]["stage"] = phase
|
|
452
|
+
return phase
|
|
453
|
+
|
|
454
|
+
# ---------------- 总报告 ----------------
|
|
455
|
+
def run(self, meta: bool = True, subject: bool = True,
|
|
456
|
+
causal: bool = True, stage: bool = True,
|
|
457
|
+
meta_discipline: bool = True, persist: bool = True) -> Dict:
|
|
458
|
+
"""执行完整迁移流水线:meta→学科卡→因果→阶段→纪律→持久化(各相由布尔开关独立启停)。"""
|
|
459
|
+
if meta:
|
|
460
|
+
self.migrate_meta()
|
|
461
|
+
if subject:
|
|
462
|
+
self.migrate_subject_cards()
|
|
463
|
+
if stage:
|
|
464
|
+
self.migrate_stage_cards()
|
|
465
|
+
if meta_discipline:
|
|
466
|
+
self.migrate_meta_discipline_cards()
|
|
467
|
+
if causal:
|
|
468
|
+
self.migrate_causal_edges()
|
|
469
|
+
# 目标库终态统计
|
|
470
|
+
dst = sqlite3.connect(self.dst)
|
|
471
|
+
dc = dst.cursor()
|
|
472
|
+
self.report["dst_final"] = {
|
|
473
|
+
"nodes": dc.execute("SELECT COUNT(*) FROM nodes").fetchone()[0],
|
|
474
|
+
"knowledge_nodes": dc.execute(
|
|
475
|
+
"SELECT COUNT(*) FROM nodes WHERE layer='knowledge'").fetchone()[0],
|
|
476
|
+
"edges": dc.execute("SELECT COUNT(*) FROM edges").fetchone()[0],
|
|
477
|
+
"causal_edges": dc.execute(
|
|
478
|
+
"SELECT COUNT(*) FROM edges WHERE relation_type='causal'").fetchone()[0],
|
|
479
|
+
"verified_edges": dc.execute(
|
|
480
|
+
"SELECT COUNT(*) FROM edges WHERE verified=1").fetchone()[0],
|
|
481
|
+
}
|
|
482
|
+
dst.close()
|
|
483
|
+
self.report["generated_at"] = time.strftime("%Y-%m-%d %H:%M:%S")
|
|
484
|
+
self.report["source"] = "migrate_wisdom v1.16(智慧之书→灵枢主库)"
|
|
485
|
+
if persist:
|
|
486
|
+
path = os.path.join(HERE, "migrate_wisdom_report.json")
|
|
487
|
+
with open(path, "w", encoding="utf-8") as f:
|
|
488
|
+
json.dump(self.report, f, ensure_ascii=False, indent=2)
|
|
489
|
+
self.report["report_path"] = path
|
|
490
|
+
return self.report
|
|
491
|
+
|
|
492
|
+
|
|
493
|
+
if __name__ == "__main__":
|
|
494
|
+
sys.stdout.reconfigure(encoding="utf-8")
|
|
495
|
+
mig = WisdomMigrator()
|
|
496
|
+
rep = mig.run()
|
|
497
|
+
print(json.dumps(rep, ensure_ascii=False, indent=2))
|