Files
caiwuzongzhang/src/bank_importer/subjects.py
T

399 lines
15 KiB
Python

"""Statutory subject suggestions, mirror mapping and confirmation (B-44).
Subjects are stored from one participating company's perspective and the
other side is the fixed mirror (应收<->应付, 其他应收<->其他应付), so the two
companies can never record conflicting subjects. Bank summary/purpose text
only ever produces a *suggestion*; nothing here confirms a subject
automatically. Confirmation is an explicit administrator decision carrying
``expected_revision`` and an idempotency key.
"""
from __future__ import annotations
import json
import re
import sqlite3
from . import matching
SUBJECTS = ("receivable", "payable", "other_receivable", "other_payable")
SUBJECT_RULE_VERSION = "subject-suggest-draft-v1"
MIRROR = {
"receivable": "payable",
"payable": "receivable",
"other_receivable": "other_payable",
"other_payable": "other_receivable",
}
SUBJECT_LABELS = {
"receivable": "应收",
"payable": "应付",
"other_receivable": "其他应收",
"other_payable": "其他应付",
}
_FULL_WIDTH = str.maketrans(
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz0123456789",
"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
"abcdefghijklmnopqrstuvwxyz0123456789",
)
# Draft v1 dictionary. Exact-keyword matching only; every hit is a suggestion
# and never an automatic posting. Trade-type keywords are deliberately absent
# until the group supplies an approved dictionary (they always go to review).
_LOAN_LIKE = ("借款", "往来款", "资金往来", "临时借款", "资金调拨", "代垫", "垫付")
_REPAY_LIKE = ("还款", "归还借款", "归还往来款")
class SubjectConflictError(ValueError):
"""A stale revision or idempotency conflict (mapped to HTTP 409)."""
class SubjectInputError(ValueError):
"""Invalid input for a subject decision (mapped to HTTP 400/422)."""
def mirror_subject(subject_code: str) -> str:
if subject_code not in MIRROR:
raise SubjectInputError("科目必须是应收/应付/其他应收/其他应付之一。")
return MIRROR[subject_code]
def subject_label(subject_code: str) -> str:
return SUBJECT_LABELS.get(subject_code, subject_code)
def _normalize(text: object) -> str:
return re.sub(r"[\s\ufeff]+", "", str(text or "").translate(_FULL_WIDTH))
def _bank_evidence_texts(connection: sqlite3.Connection, ledger_event_id: int) -> dict[str, str]:
"""Purpose/summary text of the B-43 source rows behind a bank event."""
row = connection.execute(
"""
SELECT bs.bank_event_id FROM ledger_event_bank_sources bs
WHERE bs.ledger_event_id = ?
""",
(ledger_event_id,),
).fetchone()
if row is None:
return {"purpose": "", "summary": ""}
decision = matching._current_decision_for_event(connection, row["bank_event_id"])
if decision is None:
return {"purpose": "", "summary": ""}
observations = matching._decision_observations(connection, decision["id"])
texts: dict[str, list[str]] = {"purpose": [], "summary": []}
for observation in observations:
source = connection.execute(
"SELECT purpose, summary FROM source_rows WHERE id = ?",
(observation["source_row_id"],),
).fetchone()
if source is None:
continue
for key in ("purpose", "summary"):
value = str(source[key] or "").strip()
if value:
texts[key].append(value)
return {
"purpose": " ".join(texts["purpose"]),
"summary": " ".join(texts["summary"]),
}
def compute_suggestions(
connection: sqlite3.Connection, ledger_event_id: int
) -> list[dict[str, object]]:
""" Deterministic draft suggestions for a pending event, never confirmation.
Purpose rules take precedence over summary rules. When both loan-like and
repay-like keywords match, both candidates are returned as a conflict for
the reviewer; no priority breaks the tie.
"""
from .ledger_events import current_revision
revision = current_revision(connection, ledger_event_id)
if revision is None or revision["state"] != "pending_subject":
return []
texts = _bank_evidence_texts(connection, ledger_event_id)
purpose = _normalize(texts["purpose"])
summary = _normalize(texts["summary"])
search = purpose or summary
payer = revision["payer_company_id"]
payee = revision["payee_company_id"]
loan_hit = next((word for word in _LOAN_LIKE if word in search), None)
repay_hit = next((word for word in _REPAY_LIKE if word in search), None)
suggestions: list[dict[str, object]] = []
if loan_hit:
suggestions.append(
{
"suggested_perspective_company_id": payer,
"suggested_subject_code": "other_receivable",
"reason": f"匹配建议词典「{loan_hit}」,建议付款方其他应收",
"rule_version": SUBJECT_RULE_VERSION,
"evidence": {
"keyword": loan_hit,
"matched_text": search,
"approved": False,
},
}
)
if repay_hit:
suggestions.append(
{
"suggested_perspective_company_id": payee,
"suggested_subject_code": "other_receivable",
"reason": f"匹配建议词典「{repay_hit}」,建议收款方其他应收",
"rule_version": SUBJECT_RULE_VERSION,
"evidence": {
"keyword": repay_hit,
"matched_text": search,
"approved": False,
},
}
)
return suggestions
def store_suggestions(connection: sqlite3.Connection, ledger_event_id: int) -> int:
"""Compute and append suggestions for a pending event. Returns count stored."""
from .ledger_events import current_revision
revision = current_revision(connection, ledger_event_id)
if revision is None or revision["state"] != "pending_subject":
return 0
stored = 0
for suggestion in compute_suggestions(connection, ledger_event_id):
from .db import utc_now
connection.execute(
"""
INSERT INTO ledger_subject_suggestions (
ledger_event_id, source_revision_id,
suggested_perspective_company_id, suggested_subject_code,
rule_version, evidence_json, created_at
) VALUES (?, ?, ?, ?, ?, ?, ?)
""",
(
ledger_event_id, revision["id"],
suggestion["suggested_perspective_company_id"],
suggestion["suggested_subject_code"],
suggestion["rule_version"],
json.dumps(suggestion.get("evidence", {}), ensure_ascii=False),
utc_now(),
),
)
stored += 1
return stored
def confirm_subject(
connection: sqlite3.Connection,
ledger_event_id: int,
*,
perspective_company_id: int,
subject_code: str,
reason: str,
expected_revision: int | None,
request_key: str | None,
actor: sqlite3.Row,
) -> dict[str, object]:
"""Confirm a subject, turning a pending event into a confirmed revision."""
from .ledger_events import append_revision, current_revision
reason = (reason or "").strip()
if not reason:
raise SubjectInputError("必须填写科目确认依据。")
if subject_code not in SUBJECTS:
raise SubjectInputError("科目必须是应收/应付/其他应收/其他应付之一。")
began = False
if not connection.in_transaction:
connection.execute("BEGIN IMMEDIATE")
began = True
try:
if request_key:
existing = connection.execute(
"""
SELECT * FROM ledger_event_revisions
WHERE ledger_event_id = ? AND idempotency_key = ?
ORDER BY id LIMIT 1
""",
(ledger_event_id, request_key),
).fetchone()
if existing is not None:
if began:
connection.commit()
return _revision_payload(connection, existing)
current = current_revision(connection, ledger_event_id)
if current is None:
raise SubjectConflictError("该事件不存在或没有当前修订。")
if current["state"] != "pending_subject":
raise SubjectConflictError("只有待确认科目的事件可以确认科目。")
# ``expected_revision`` may be the revision row id (what the API/UI
# sends as ``ledger_revision_id``) or the per-event sequence number;
# both identify the exact revision the client saw.
if expected_revision is not None and int(expected_revision) not in (
current["id"], current["revision"],
):
raise SubjectConflictError("事件已发生变更,请刷新后重试。")
participants = {current["payer_company_id"], current["payee_company_id"]}
if perspective_company_id not in participants:
raise SubjectInputError("视角公司必须是事件参与方。")
revision_id = append_revision(
connection,
ledger_event_id,
state="confirmed",
effective_at=current["effective_at"],
amount=current["amount"],
currency=current["currency"],
payer_company_id=current["payer_company_id"],
payee_company_id=current["payee_company_id"],
perspective_company_id=perspective_company_id,
subject_code=subject_code,
source_kind=current["source_kind"],
source_revision_token=current["source_revision_token"],
posting_kind=current["posting_kind"],
reverses_ledger_event_id=current["reverses_ledger_event_id"],
adjusts_ledger_event_id=current["adjusts_ledger_event_id"],
rule_version=current["rule_version"] or SUBJECT_RULE_VERSION,
evidence_json=current["evidence_json"],
idempotency_key=request_key,
actor=actor,
reason=reason,
supersedes_revision_id=current["id"],
)
row = connection.execute(
"SELECT * FROM ledger_event_revisions WHERE id = ?", (revision_id,)
).fetchone()
except Exception:
if began:
connection.rollback()
raise
else:
if began:
connection.commit()
return _revision_payload(connection, row)
def park_subject(
connection: sqlite3.Connection,
ledger_event_id: int,
*,
disposition: str,
reason: str,
expected_revision: int | None,
request_key: str | None,
actor: sqlite3.Row,
) -> dict[str, object]:
"""Record 退回/转异常 without confirming a statutory subject.
The event stays ``pending_subject`` so it never enters confirmed balances.
``exception`` is hidden from the active review queue; ``return`` remains
visible so the company can supplement materials.
"""
from .ledger_events import append_revision, current_revision
if disposition not in ("return", "exception"):
raise SubjectInputError("科目处理只能是退回或转异常。")
reason = (reason or "").strip()
if not reason:
raise SubjectInputError("必须填写处理依据。")
began = False
if not connection.in_transaction:
connection.execute("BEGIN IMMEDIATE")
began = True
try:
if request_key:
existing = connection.execute(
"""
SELECT * FROM ledger_event_revisions
WHERE ledger_event_id = ? AND idempotency_key = ?
ORDER BY id LIMIT 1
""",
(ledger_event_id, request_key),
).fetchone()
if existing is not None:
if began:
connection.commit()
return _revision_payload(connection, existing)
current = current_revision(connection, ledger_event_id)
if current is None:
raise SubjectConflictError("该事件不存在或没有当前修订。")
if current["state"] != "pending_subject":
raise SubjectConflictError("只有待确认科目的事件可以退回或转异常。")
if expected_revision is not None and int(expected_revision) not in (
current["id"], current["revision"],
):
raise SubjectConflictError("事件已发生变更,请刷新后重试。")
evidence = json.loads(current["evidence_json"] or "{}") if current["evidence_json"] else {}
evidence["admin_disposition"] = disposition
revision_id = append_revision(
connection,
ledger_event_id,
state="pending_subject",
effective_at=current["effective_at"],
amount=current["amount"],
currency=current["currency"],
payer_company_id=current["payer_company_id"],
payee_company_id=current["payee_company_id"],
perspective_company_id=None,
subject_code=None,
source_kind=current["source_kind"],
source_revision_token=current["source_revision_token"],
posting_kind=current["posting_kind"],
reverses_ledger_event_id=current["reverses_ledger_event_id"],
adjusts_ledger_event_id=current["adjusts_ledger_event_id"],
rule_version=current["rule_version"] or SUBJECT_RULE_VERSION,
evidence_json=json.dumps(evidence, ensure_ascii=False),
idempotency_key=request_key,
actor=actor,
reason=reason,
supersedes_revision_id=current["id"],
)
row = connection.execute(
"SELECT * FROM ledger_event_revisions WHERE id = ?", (revision_id,)
).fetchone()
except Exception:
if began:
connection.rollback()
raise
else:
if began:
connection.commit()
return _revision_payload(connection, row)
def _revision_payload(connection: sqlite3.Connection, revision: sqlite3.Row) -> dict[str, object]:
company = connection.execute(
"SELECT name FROM companies WHERE id = ?", (revision["perspective_company_id"],)
).fetchone()
return {
"ledger_event_id": revision["ledger_event_id"],
"revision_id": revision["id"],
"revision": revision["revision"],
"state": revision["state"],
"effective_at": revision["effective_at"],
"amount": revision["amount"],
"currency": revision["currency"],
"payer_company_id": revision["payer_company_id"],
"payee_company_id": revision["payee_company_id"],
"perspective_company_id": revision["perspective_company_id"],
"perspective_company_name": company["name"] if company else None,
"subject_code": revision["subject_code"],
"subject_label": subject_label(revision["subject_code"])
if revision["subject_code"]
else None,
"posting_kind": revision["posting_kind"],
"source_kind": revision["source_kind"],
"reason": revision["reason"],
"created_at": revision["created_at"],
}