Mercurial > libervia-backend
view sat/memory/sqla_mapping.py @ 3751:073f386a191a
core (memory/disco): fix hard encoded value in `findServiceEntity`
author | Goffi <goffi@goffi.org> |
---|---|
date | Fri, 13 May 2022 17:56:40 +0200 |
parents | 658ddbabaf36 |
children | 24c1c06c865b |
line wrap: on
line source
#!/usr/bin/env python3 # Libervia: an XMPP client # Copyright (C) 2009-2021 Jérôme Poisson (goffi@goffi.org) # This program is free software: you can redistribute it and/or modify # it under the terms of the GNU Affero General Public License as published by # the Free Software Foundation, either version 3 of the License, or # (at your option) any later version. # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU Affero General Public License for more details. # You should have received a copy of the GNU Affero General Public License # along with this program. If not, see <http://www.gnu.org/licenses/>. import pickle import json from datetime import datetime import time import enum from sqlalchemy import ( MetaData, Column, Integer, Text, Float, Boolean, DateTime, Enum, JSON, ForeignKey, UniqueConstraint, Index, DDL, event ) from sqlalchemy.orm import declarative_base, relationship from sqlalchemy.types import TypeDecorator from sqlalchemy.sql.functions import now from twisted.words.protocols.jabber import jid from wokkel import generic Base = declarative_base( metadata=MetaData( naming_convention={ "ix": 'ix_%(column_0_label)s', "uq": "uq_%(table_name)s_%(column_0_name)s", "ck": "ck_%(table_name)s_%(constraint_name)s", "fk": "fk_%(table_name)s_%(column_0_name)s_%(referred_table_name)s", "pk": "pk_%(table_name)s" } ) ) # keys which are in message data extra but not stored in extra field this is # because those values are stored in separate fields NOT_IN_EXTRA = ('stanza_id', 'received_timestamp', 'update_uid') class SyncState(enum.Enum): #: synchronisation is currently in progress IN_PROGRESS = 1 #: synchronisation is done COMPLETED = 2 #: something wrong happened during synchronisation, won't sync ERROR = 3 #: synchronisation won't be done even if a syncing analyser matches NO_SYNC = 4 class SubscriptionState(enum.Enum): SUBSCRIBED = 1 PENDING = 2 class LegacyPickle(TypeDecorator): """Handle troubles with data pickled by former version of SàT This type is temporary until we do migration to a proper data type """ # Blob is used on SQLite but gives errors when used here, while Text works fine impl = Text cache_ok = True def process_bind_param(self, value, dialect): if value is None: return None return pickle.dumps(value, 0) def process_result_value(self, value, dialect): if value is None: return None # value types are inconsistent (probably a consequence of Python 2/3 port # and/or SQLite dynamic typing) try: value = value.encode() except AttributeError: pass # "utf-8" encoding is needed to handle Python 2 pickled data return pickle.loads(value, encoding="utf-8") class Json(TypeDecorator): """Handle JSON field in DB independant way""" # Blob is used on SQLite but gives errors when used here, while Text works fine impl = Text cache_ok = True def process_bind_param(self, value, dialect): if value is None: return None return json.dumps(value) def process_result_value(self, value, dialect): if value is None: return None return json.loads(value) class JsonDefaultDict(Json): """Json type which convert NULL to empty dict instead of None""" def process_result_value(self, value, dialect): if value is None: return {} return json.loads(value) class Xml(TypeDecorator): impl = Text cache_ok = True def process_bind_param(self, value, dialect): if value is None: return None return value.toXml() def process_result_value(self, value, dialect): if value is None: return None return generic.parseXml(value.encode()) class JID(TypeDecorator): """Store twisted JID in text fields""" impl = Text cache_ok = True def process_bind_param(self, value, dialect): if value is None: return None return value.full() def process_result_value(self, value, dialect): if value is None: return None return jid.JID(value) class Profile(Base): __tablename__ = "profiles" id = Column( Integer, primary_key=True, nullable=True, ) name = Column(Text, unique=True) params = relationship("ParamInd", back_populates="profile", passive_deletes=True) private_data = relationship( "PrivateInd", back_populates="profile", passive_deletes=True ) private_bin_data = relationship( "PrivateIndBin", back_populates="profile", passive_deletes=True ) class Component(Base): __tablename__ = "components" profile_id = Column( ForeignKey("profiles.id", ondelete="CASCADE"), nullable=True, primary_key=True ) entry_point = Column(Text, nullable=False) profile = relationship("Profile") class History(Base): __tablename__ = "history" __table_args__ = ( UniqueConstraint("profile_id", "stanza_id", "source", "dest"), Index("history__profile_id_timestamp", "profile_id", "timestamp"), Index( "history__profile_id_received_timestamp", "profile_id", "received_timestamp" ) ) uid = Column(Text, primary_key=True) stanza_id = Column(Text) update_uid = Column(Text) profile_id = Column(ForeignKey("profiles.id", ondelete="CASCADE")) source = Column(Text) dest = Column(Text) source_res = Column(Text) dest_res = Column(Text) timestamp = Column(Float, nullable=False) received_timestamp = Column(Float) type = Column( Enum( "chat", "error", "groupchat", "headline", "normal", # info is not XMPP standard, but used to keep track of info like join/leave # in a MUC "info", name="message_type", create_constraint=True, ), nullable=False, ) extra = Column(LegacyPickle) profile = relationship("Profile") messages = relationship("Message", backref="history", passive_deletes=True) subjects = relationship("Subject", backref="history", passive_deletes=True) thread = relationship( "Thread", uselist=False, back_populates="history", passive_deletes=True ) def __init__(self, *args, **kwargs): source_jid = kwargs.pop("source_jid", None) if source_jid is not None: kwargs["source"] = source_jid.userhost() kwargs["source_res"] = source_jid.resource dest_jid = kwargs.pop("dest_jid", None) if dest_jid is not None: kwargs["dest"] = dest_jid.userhost() kwargs["dest_res"] = dest_jid.resource super().__init__(*args, **kwargs) @property def source_jid(self) -> jid.JID: return jid.JID(f"{self.source}/{self.source_res or ''}") @source_jid.setter def source_jid(self, source_jid: jid.JID) -> None: self.source = source_jid.userhost self.source_res = source_jid.resource @property def dest_jid(self): return jid.JID(f"{self.dest}/{self.dest_res or ''}") @dest_jid.setter def dest_jid(self, dest_jid: jid.JID) -> None: self.dest = dest_jid.userhost self.dest_res = dest_jid.resource def __repr__(self): dt = datetime.fromtimestamp(self.timestamp) return f"History<{self.source_jid.full()}->{self.dest_jid.full()} [{dt}]>" def serialise(self): extra = self.extra if self.stanza_id is not None: extra["stanza_id"] = self.stanza_id if self.update_uid is not None: extra["update_uid"] = self.update_uid if self.received_timestamp is not None: extra["received_timestamp"] = self.received_timestamp if self.thread is not None: extra["thread"] = self.thread.thread_id if self.thread.parent_id is not None: extra["thread_parent"] = self.thread.parent_id return { "from": f"{self.source}/{self.source_res}" if self.source_res else self.source, "to": f"{self.dest}/{self.dest_res}" if self.dest_res else self.dest, "uid": self.uid, "message": {m.language or '': m.message for m in self.messages}, "subject": {m.language or '': m.subject for m in self.subjects}, "type": self.type, "extra": extra, "timestamp": self.timestamp, } def as_tuple(self): d = self.serialise() return ( d['uid'], d['timestamp'], d['from'], d['to'], d['message'], d['subject'], d['type'], d['extra'] ) @staticmethod def debug_collection(history_collection): for idx, history in enumerate(history_collection): history.debug_msg(idx) def debug_msg(self, idx=None): """Print messages""" dt = datetime.fromtimestamp(self.timestamp) if idx is not None: dt = f"({idx}) {dt}" parts = [] parts.append(f"[{dt}]<{self.source_jid.full()}->{self.dest_jid.full()}> ") for message in self.messages: if message.language: parts.append(f"[{message.language}] ") parts.append(f"{message.message}\n") print("".join(parts)) class Message(Base): __tablename__ = "message" __table_args__ = ( Index("message__history_uid", "history_uid"), ) id = Column( Integer, primary_key=True, ) history_uid = Column(ForeignKey("history.uid", ondelete="CASCADE")) message = Column(Text) language = Column(Text) def __repr__(self): lang_str = f"[{self.language}]" if self.language else "" msg = f"{self.message[:20]}…" if len(self.message)>20 else self.message content = f"{lang_str}{msg}" return f"Message<{content}>" class Subject(Base): __tablename__ = "subject" __table_args__ = ( Index("subject__history_uid", "history_uid"), ) id = Column( Integer, primary_key=True, ) history_uid = Column(ForeignKey("history.uid", ondelete="CASCADE")) subject = Column(Text) language = Column(Text) def __repr__(self): lang_str = f"[{self.language}]" if self.language else "" msg = f"{self.subject[:20]}…" if len(self.subject)>20 else self.subject content = f"{lang_str}{msg}" return f"Subject<{content}>" class Thread(Base): __tablename__ = "thread" __table_args__ = ( Index("thread__history_uid", "history_uid"), ) id = Column( Integer, primary_key=True, ) history_uid = Column(ForeignKey("history.uid", ondelete="CASCADE")) thread_id = Column(Text) parent_id = Column(Text) history = relationship("History", uselist=False, back_populates="thread") def __repr__(self): return f"Thread<{self.thread_id} [parent: {self.parent_id}]>" class ParamGen(Base): __tablename__ = "param_gen" category = Column(Text, primary_key=True) name = Column(Text, primary_key=True) value = Column(Text) class ParamInd(Base): __tablename__ = "param_ind" category = Column(Text, primary_key=True) name = Column(Text, primary_key=True) profile_id = Column( ForeignKey("profiles.id", ondelete="CASCADE"), primary_key=True ) value = Column(Text) profile = relationship("Profile", back_populates="params") class PrivateGen(Base): __tablename__ = "private_gen" namespace = Column(Text, primary_key=True) key = Column(Text, primary_key=True) value = Column(Text) class PrivateInd(Base): __tablename__ = "private_ind" namespace = Column(Text, primary_key=True) key = Column(Text, primary_key=True) profile_id = Column( ForeignKey("profiles.id", ondelete="CASCADE"), primary_key=True ) value = Column(Text) profile = relationship("Profile", back_populates="private_data") class PrivateGenBin(Base): __tablename__ = "private_gen_bin" namespace = Column(Text, primary_key=True) key = Column(Text, primary_key=True) value = Column(LegacyPickle) class PrivateIndBin(Base): __tablename__ = "private_ind_bin" namespace = Column(Text, primary_key=True) key = Column(Text, primary_key=True) profile_id = Column( ForeignKey("profiles.id", ondelete="CASCADE"), primary_key=True ) value = Column(LegacyPickle) profile = relationship("Profile", back_populates="private_bin_data") class File(Base): __tablename__ = "files" __table_args__ = ( Index("files__profile_id_owner_parent", "profile_id", "owner", "parent"), Index( "files__profile_id_owner_media_type_media_subtype", "profile_id", "owner", "media_type", "media_subtype" ) ) id = Column(Text, primary_key=True) public_id = Column(Text, unique=True) version = Column(Text, primary_key=True) parent = Column(Text, nullable=False) type = Column( Enum( "file", "directory", name="file_type", create_constraint=True ), nullable=False, server_default="file", ) file_hash = Column(Text) hash_algo = Column(Text) name = Column(Text, nullable=False) size = Column(Integer) namespace = Column(Text) media_type = Column(Text) media_subtype = Column(Text) created = Column(Float, nullable=False) modified = Column(Float) owner = Column(JID) access = Column(JsonDefaultDict) extra = Column(JsonDefaultDict) profile_id = Column(ForeignKey("profiles.id", ondelete="CASCADE")) profile = relationship("Profile") class PubsubNode(Base): __tablename__ = "pubsub_nodes" __table_args__ = ( UniqueConstraint("profile_id", "service", "name"), ) id = Column(Integer, primary_key=True) profile_id = Column( ForeignKey("profiles.id", ondelete="CASCADE") ) service = Column(JID) name = Column(Text, nullable=False) subscribed = Column( Boolean(create_constraint=True, name="subscribed_bool"), nullable=False ) analyser = Column(Text) sync_state = Column( Enum( SyncState, name="sync_state", create_constraint=True, ), nullable=True ) sync_state_updated = Column( Float, nullable=False, default=time.time() ) type_ = Column( Text, name="type", nullable=True ) subtype = Column( Text, nullable=True ) extra = Column(JSON) items = relationship("PubsubItem", back_populates="node", passive_deletes=True) subscriptions = relationship("PubsubSub", back_populates="node", passive_deletes=True) def __str__(self): return f"Pubsub node {self.name!r} at {self.service}" class PubsubSub(Base): """Subscriptions to pubsub nodes Used by components managing a pubsub service """ __tablename__ = "pubsub_subs" __table_args__ = ( UniqueConstraint("node_id", "subscriber"), ) id = Column(Integer, primary_key=True) node_id = Column(ForeignKey("pubsub_nodes.id", ondelete="CASCADE"), nullable=False) subscriber = Column(JID) state = Column( Enum( SubscriptionState, name="state", create_constraint=True, ), nullable=True ) node = relationship("PubsubNode", back_populates="subscriptions") class PubsubItem(Base): __tablename__ = "pubsub_items" __table_args__ = ( UniqueConstraint("node_id", "name"), ) id = Column(Integer, primary_key=True) node_id = Column(ForeignKey("pubsub_nodes.id", ondelete="CASCADE"), nullable=False) name = Column(Text, nullable=False) data = Column(Xml, nullable=False) created = Column(DateTime, nullable=False, server_default=now()) updated = Column(DateTime, nullable=False, server_default=now(), onupdate=now()) parsed = Column(JSON) node = relationship("PubsubNode", back_populates="items") ## Full-Text Search # create @event.listens_for(PubsubItem.__table__, "after_create") def fts_create(target, connection, **kw): """Full-Text Search table creation""" if connection.engine.name == "sqlite": # Using SQLite FTS5 queries = [ "CREATE VIRTUAL TABLE pubsub_items_fts " "USING fts5(data, content=pubsub_items, content_rowid=id)", "CREATE TRIGGER pubsub_items_fts_sync_ins AFTER INSERT ON pubsub_items BEGIN" " INSERT INTO pubsub_items_fts(rowid, data) VALUES (new.id, new.data);" "END", "CREATE TRIGGER pubsub_items_fts_sync_del AFTER DELETE ON pubsub_items BEGIN" " INSERT INTO pubsub_items_fts(pubsub_items_fts, rowid, data) " "VALUES('delete', old.id, old.data);" "END", "CREATE TRIGGER pubsub_items_fts_sync_upd AFTER UPDATE ON pubsub_items BEGIN" " INSERT INTO pubsub_items_fts(pubsub_items_fts, rowid, data) VALUES" "('delete', old.id, old.data);" " INSERT INTO pubsub_items_fts(rowid, data) VALUES(new.id, new.data);" "END" ] for q in queries: connection.execute(DDL(q)) # drop @event.listens_for(PubsubItem.__table__, "before_drop") def fts_drop(target, connection, **kw): "Full-Text Search table drop" "" if connection.engine.name == "sqlite": # Using SQLite FTS5 queries = [ "DROP TRIGGER IF EXISTS pubsub_items_fts_sync_ins", "DROP TRIGGER IF EXISTS pubsub_items_fts_sync_del", "DROP TRIGGER IF EXISTS pubsub_items_fts_sync_upd", "DROP TABLE IF EXISTS pubsub_items_fts", ] for q in queries: connection.execute(DDL(q))