ucvl-home-vision/genealogy.py

"""A private family register, independent from login accounts and camera roles."""
import datetime as dt
import json
import secrets
from urllib.parse import urlsplit


class Family:
    def __init__(self, app):
        self.a = app

    def initialize(self):
        with self.a.LOCK:
            self.a.DB.executescript('''
            CREATE TABLE IF NOT EXISTS people (id TEXT PRIMARY KEY, body TEXT NOT NULL);
            CREATE TABLE IF NOT EXISTS family_links (id TEXT PRIMARY KEY, body TEXT NOT NULL);
            CREATE TABLE IF NOT EXISTS family_events (id TEXT PRIMARY KEY, body TEXT NOT NULL);
            CREATE TABLE IF NOT EXISTS family_changes (
                revision INTEGER PRIMARY KEY, kind TEXT NOT NULL, target_id TEXT NOT NULL,
                before_json TEXT, after_json TEXT NOT NULL, actor TEXT NOT NULL,
                at TEXT NOT NULL, restored_from INTEGER);
            CREATE INDEX IF NOT EXISTS family_changes_target ON family_changes(kind,target_id,revision);
            ''')
            self.a.DB.commit()

    def authorize(self, user, edit=False):
        if user['role'] != 'admin' and user.get('familyAccess', 'none') not in (('edit',) if edit else ('read', 'edit')):
            raise self.a.Problem('没有家谱' + ('编辑' if edit else '查看') + '权限', 403)

    def snapshot(self, user):
        self.authorize(user)
        with self.a.LOCK:
            can_edit = user['role'] == 'admin' or user.get('familyAccess') == 'edit'
            return {'revision': self.a.setting('familyRevision', 0), 'people': self.a.objects('people'),
                    'links': self.a.objects('family_links'),
                    'events': [e for e in self.a.objects('family_events')
                               if can_edit or (e.get('status') == 'published' and not e.get('archived'))],
                    'canEdit': can_edit,
                    'selfId': user.get('personId', '')}

    def save(self, kind, data, user, restored_from=None):
        self.authorize(user, True)
        a = self.a
        with a.LOCK:
            revision = a.setting('familyRevision', 0)
            if type(data.get('revision')) is not int or data['revision'] != revision:
                raise a.Problem('家谱已被更新，请重新载入后再编辑', 409)
            if kind not in ('person', 'link', 'event'):
                raise a.Problem('家谱记录类型不存在', 404)
            table = {'person': 'people', 'link': 'family_links', 'event': 'family_events'}[kind]
            old = a.get_object(table, data.get('id'))
            if data.get('id') and not old:
                raise a.Problem('家谱记录不存在', 404)
            item = {'person': self.person, 'link': self.link, 'event': self.event}[kind](data, old)
            item.update(updatedAt=dt.datetime.now(dt.timezone.utc).isoformat(), updatedBy=user['username'])
            if kind == 'event':
                item.update(createdAt=old.get('createdAt', item['updatedAt']) if old else item['updatedAt'],
                            createdBy=old.get('createdBy', user['username']) if old else user['username'])
                # Withdrawal and restoration must retain the first publication time.
                if old and old.get('publishedAt'):
                    item['publishedAt'] = old['publishedAt']
                elif item['status'] == 'published':
                    item['publishedAt'] = item['updatedAt']
            # Revision and record are committed together, so concurrent forms cannot overwrite silently.
            with a.DB:
                a.DB.execute('INSERT OR REPLACE INTO ' + table + ' VALUES (?,?)', (item['id'], json.dumps(item)))
                a.DB.execute('INSERT OR REPLACE INTO settings VALUES (?,?)', ('familyRevision', json.dumps(revision + 1)))
                a.DB.execute('INSERT INTO family_changes VALUES (?,?,?,?,?,?,?,?)',
                             (revision + 1, kind, item['id'], json.dumps(old) if old else None,
                              json.dumps(item), user['username'], item['updatedAt'], restored_from))
            a.audit({'person': '更新家谱人物', 'link': '更新家谱关系', 'event': '更新家族记事'}[kind], user['username'])
            return {'item': item, 'revision': revision + 1}

    def history(self, user, kind, target_id, before=None):
        # Previous drafts and withdrawn text are visible only to family editors.
        self.authorize(user, True)
        a = self.a
        table = {'person': 'people', 'link': 'family_links', 'event': 'family_events'}.get(kind)
        with a.LOCK:
            if not table or not a.get_object(table, target_id):
                raise a.Problem('家谱记录不存在', 404)
            if before is not None:
                a.integer(before, 1, 2**53 - 1)
            rows = a.DB.execute('SELECT * FROM family_changes WHERE kind=? AND target_id=? AND revision<? '
                                'ORDER BY revision DESC LIMIT 21', (kind, target_id, before or 2**53 - 1)).fetchall()
            items = []
            for row in rows[:20]:
                old, new = json.loads(row['before_json'] or '{}'), json.loads(row['after_json'])
                fields = [k for k in new if k not in ('id', 'updatedAt', 'updatedBy', 'createdAt', 'createdBy', 'publishedAt')
                          and old.get(k) != new[k]]
                items.append(dict(revision=row['revision'], at=row['at'], actor=row['actor'], fields=fields,
                                  created=row['before_json'] is None, restoredFrom=row['restored_from']))
            return dict(items=items, nextBefore=items[-1]['revision'] if len(rows) > 20 else None,
                        revision=a.setting('familyRevision', 0))

    def change(self, user, change_revision):
        self.authorize(user, True)
        self.a.integer(change_revision, 1, 2**53 - 1)
        with self.a.LOCK:
            row = self.a.DB.execute('SELECT * FROM family_changes WHERE revision=?', (change_revision,)).fetchone()
            if not row:
                raise self.a.Problem('修改记录不存在', 404)
            return dict(revision=row['revision'], kind=row['kind'], targetId=row['target_id'],
                        before=json.loads(row['before_json']) if row['before_json'] else None,
                        after=json.loads(row['after_json']), at=row['at'], actor=row['actor'])

    def restore(self, data, user):
        self.authorize(user, True)
        with self.a.LOCK:
            change = self.change(user, data.get('changeRevision'))
            side = data.get('side')
            if side not in ('before', 'after') or change[side] is None:
                raise self.a.Problem('请选择有效的历史版本；创建前没有可恢复的内容')
            # Reuse all current relationship/date/publication validation. Restore is a new edit,
            # not an overwrite of history or a rollback of other people's records.
            return self.save(change['kind'], dict(change[side], revision=data.get('revision')),
                             user, restored_from=change['revision'])

    def event(self, data, old):
        a = self.a
        item = {'id': old['id'] if old else secrets.token_hex(8)}
        for key, maximum, required in [('title', 160, True), ('eventDate', 10, True),
                                       ('sourceName', 160, False),
                                       ('sourceUrl', 2000, False)]:
            item[key] = a.clean_text(data.get(key, ''), maximum, required)
        try:
            if dt.date.fromisoformat(item['eventDate']).isoformat() != item['eventDate']:
                raise ValueError()
        except ValueError:
            raise a.Problem('记事日期请填写有效日期，格式为 YYYY-MM-DD')
        for key, maximum, label in [('summary', 400, '摘要'), ('body', 12000, '正文')]:
            value = data.get(key, '')
            if not isinstance(value, str) or len(value) > maximum or any(ord(c) < 32 and c not in '\r\n\t' for c in value):
                raise a.Problem(f'记事{label}格式不正确，最多 {maximum} 字')
            item[key] = value.strip()
        item['personIds'] = data.get('personIds', [])
        if (not isinstance(item['personIds'], list) or len(item['personIds']) > 200
                or any(not isinstance(p, str) or not a.get_object('people', p) for p in item['personIds'])):
            raise a.Problem('请选择有效的关联人物，最多 200 位')
        item['personIds'] = list(dict.fromkeys(item['personIds']))
        item['status'] = data.get('status', old.get('status', 'draft') if old else 'draft')
        item['archived'] = data.get('archived', old.get('archived', False) if old else False)
        if item['status'] not in ('draft', 'published') or not isinstance(item['archived'], bool):
            raise a.Problem('记事状态不正确')
        if item['sourceUrl']:
            try:
                parsed = urlsplit(item['sourceUrl'])
                if parsed.scheme not in ('https', 'http') or not parsed.hostname or parsed.username or parsed.password:
                    raise ValueError()
                parsed.port
            except ValueError:
                raise a.Problem('报道链接请填写完整的 http 或 https 地址，不可包含账号密码')
        if item['status'] == 'published' and not (item['body'] or item['summary'] or item['sourceUrl']):
            raise a.Problem('发布前请填写记事正文、摘要或报道链接')
        return item

    def person(self, data, old):
        a = self.a
        item = {'id': old['id'] if old else secrets.token_hex(8)}
        for key, maximum in [('name', 80), ('alias', 120), ('birthDate', 10), ('deathDate', 10),
                             ('birthplace', 160), ('biography', 2000), ('note', 500)]:
            value = data.get(key, '')
            if key in ('biography', 'note'):
                if not isinstance(value, str) or len(value) > maximum or any(ord(c) < 32 and c not in '\r\n\t' for c in value):
                    raise a.Problem('生平或备注格式不正确')
                item[key] = value.strip()
            else:
                item[key] = a.clean_text(value, maximum, key == 'name')
        for key in ('birthDate', 'deathDate'):
            if item[key]:
                try:
                    if dt.date.fromisoformat(item[key]).isoformat() != item[key]:
                        raise ValueError()
                except ValueError:
                    raise a.Problem('日期请使用公历 YYYY-MM-DD；不确定可留空，在备注记录')
        if item['birthDate'] and item['deathDate'] and item['birthDate'] > item['deathDate']:
            raise a.Problem('逝世日期不能早于出生日期')
        item['gender'] = data.get('gender', 'unknown')
        item['lifeStatus'] = data.get('lifeStatus', 'alive')
        if item['gender'] not in ('male', 'female', 'unknown') or item['lifeStatus'] not in ('alive', 'deceased', 'unknown'):
            raise a.Problem('人物属性不正确')
        if item['deathDate'] and item['lifeStatus'] != 'deceased':
            raise a.Problem('填写逝世日期时，请选择已故')
        item['archived'] = data.get('archived', False)
        if not isinstance(item['archived'], bool):
            raise a.Problem('归档状态不正确')
        if not old or item['birthDate'] != old.get('birthDate', ''):
            self.check_birth_order([e for e in a.objects('family_links')
                                    if item['id'] in (e['fromId'], e['toId'])], item)
        return item

    def check_birth_order(self, links, proposed_person=None):
        people = {p['id']: p for p in self.a.objects('people')}
        if proposed_person:
            people[proposed_person['id']] = proposed_person
        for edge in links:
            if not edge.get('active') or edge['kind'] != 'parent' or edge.get('lineage') != 'biological':
                continue
            parent, child = people.get(edge['fromId']), people.get(edge['toId'])
            if parent and child and parent.get('birthDate') and child.get('birthDate') and parent['birthDate'] >= child['birthDate']:
                raise self.a.Problem('亲生父母的出生日期必须早于子女，请核对日期或亲子属性')

    def link(self, data, old):
        a = self.a
        left, right = data.get('fromId'), data.get('toId')
        if not isinstance(left, str) or not isinstance(right, str) or not a.get_object('people', left) or not a.get_object('people', right):
            raise a.Problem('请先录入关系双方的人物')
        if left == right:
            raise a.Problem('不能把人物与自己建立亲属关系')
        kind = data.get('kind')
        if kind not in ('parent', 'spouse'):
            raise a.Problem('关系类型不正确')
        lineage = data.get('lineage', 'unspecified')
        if lineage not in ('biological', 'adoptive', 'step', 'unspecified'):
            raise a.Problem('亲子关系属性不正确')
        active = data.get('active', True)
        if not isinstance(active, bool):
            raise a.Problem('关系状态不正确')
        item = dict(id=old['id'] if old else secrets.token_hex(8), fromId=left, toId=right, kind=kind,
                    lineage=lineage if kind == 'parent' else 'unspecified', active=active,
                    note=a.clean_text(data.get('note', ''), 300))
        self.check_birth_order([item])
        links = [e for e in a.objects('family_links') if e.get('active') and e['id'] != item['id']]
        if active:
            for e in links:
                same = (e['fromId'], e['toId']) == (left, right)
                if kind == 'spouse':
                    same = {e['fromId'], e['toId']} == {left, right}
                if e['kind'] == kind and same:
                    raise a.Problem('这条关系已经存在', 409)
            graph = {}
            for e in links + [item]:
                if e['kind'] == 'parent':
                    graph.setdefault(e['fromId'], []).append(e['toId'])
            # Iterative walk supports deep ancestry without Python recursion limits.
            def reaches(start, target):
                pending, seen = [start], set()
                while pending:
                    p = pending.pop()
                    if p in seen:
                        continue
                    seen.add(p)
                    for child in graph.get(p, []):
                        if child == target:
                            return True
                        pending.append(child)
                return False
            if kind == 'parent' and reaches(right, left):
                raise a.Problem('这条关系会形成祖先与后代循环，请检查方向')
            for e in links + [item]:
                if e['kind'] == 'spouse' and (reaches(e['fromId'], e['toId']) or reaches(e['toId'], e['fromId'])):
                    raise a.Problem('配偶不能同时是直系祖先或后代，请检查已有关系')
        return item