lbry-sdk/lbrynet/extras/daemon/storage.py

894 lines
38 KiB
Python

import logging
import os
import sqlite3
import traceback
from binascii import hexlify, unhexlify
from decimal import Decimal
from twisted.internet import defer, task, threads
from twisted.enterprise import adbapi
from lbrynet.schema.claim import ClaimDict
from lbrynet.schema.decode import smart_decode
from lbrynet.extras.daemon import conf
from lbrynet.blob.CryptBlob import CryptBlobInfo
from lbrynet.dht.constants import dataExpireTimeout
from torba.client.constants import COIN
log = logging.getLogger(__name__)
def _get_next_available_file_name(download_directory, file_name):
base_name, ext = os.path.splitext(file_name)
i = 0
while os.path.isfile(os.path.join(download_directory, file_name)):
i += 1
file_name = "%s_%i%s" % (base_name, i, ext)
return os.path.join(download_directory, file_name)
def _open_file_for_writing(download_directory, suggested_file_name):
file_path = _get_next_available_file_name(download_directory, suggested_file_name)
try:
file_handle = open(file_path, 'wb')
file_handle.close()
except IOError:
log.error(traceback.format_exc())
raise ValueError(
"Failed to open %s. Make sure you have permission to save files to that location." % file_path
)
return os.path.basename(file_path)
def open_file_for_writing(download_directory, suggested_file_name):
"""
Used to touch the path of a file to be downloaded
:param download_directory: (str)
:param suggested_file_name: (str)
:return: (str) basename
"""
return threads.deferToThread(_open_file_for_writing, download_directory, suggested_file_name)
def rerun_if_locked(f):
max_attempts = 5
def rerun(err, rerun_count, *args, **kwargs):
connection = args[0]
reactor = connection.reactor
log.debug("Failed to execute (%s): %s", err, args)
if err.check(sqlite3.OperationalError) and "database is locked" in str(err.value):
log.warning("database was locked. rerunning %s with args %s, kwargs %s",
str(f), str(args), str(kwargs))
if rerun_count < max_attempts:
delay = 2**rerun_count
return task.deferLater(reactor, delay, inner_wrapper, rerun_count + 1, *args, **kwargs)
raise err
def check_needed_rerun(result, rerun_count):
if rerun_count:
log.info("successfully reran database query")
return result
def inner_wrapper(rerun_count, *args, **kwargs):
d = f(*args, **kwargs)
d.addCallback(check_needed_rerun, rerun_count)
d.addErrback(rerun, rerun_count, *args, **kwargs)
return d
def wrapper(*args, **kwargs):
return inner_wrapper(0, *args, **kwargs)
return wrapper
class SqliteConnection(adbapi.ConnectionPool):
def __init__(self, db_path):
super().__init__('sqlite3', db_path, check_same_thread=False)
@rerun_if_locked
def runInteraction(self, interaction, *args, **kw):
return super().runInteraction(interaction, *args, **kw)
@classmethod
def set_reactor(cls, reactor):
cls.reactor = reactor
class SQLiteStorage:
CREATE_TABLES_QUERY = """
pragma foreign_keys=on;
pragma journal_mode=WAL;
create table if not exists blob (
blob_hash char(96) primary key not null,
blob_length integer not null,
next_announce_time integer not null,
should_announce integer not null default 0,
status text not null,
last_announced_time integer,
single_announce integer
);
create table if not exists stream (
stream_hash char(96) not null primary key,
sd_hash char(96) not null references blob,
stream_key text not null,
stream_name text not null,
suggested_filename text not null
);
create table if not exists stream_blob (
stream_hash char(96) not null references stream,
blob_hash char(96) references blob,
position integer not null,
iv char(32) not null,
primary key (stream_hash, blob_hash)
);
create table if not exists claim (
claim_outpoint text not null primary key,
claim_id char(40) not null,
claim_name text not null,
amount integer not null,
height integer not null,
serialized_metadata blob not null,
channel_claim_id text,
address text not null,
claim_sequence integer not null
);
create table if not exists file (
stream_hash text primary key not null references stream,
file_name text not null,
download_directory text not null,
blob_data_rate real not null,
status text not null
);
create table if not exists content_claim (
stream_hash text unique not null references file,
claim_outpoint text not null references claim,
primary key (stream_hash, claim_outpoint)
);
create table if not exists support (
support_outpoint text not null primary key,
claim_id text not null,
amount integer not null,
address text not null
);
create table if not exists reflected_stream (
sd_hash text not null,
reflector_address text not null,
timestamp integer,
primary key (sd_hash, reflector_address)
);
"""
def __init__(self, db_dir, reactor=None):
if not reactor:
from twisted.internet import reactor
self.db_dir = db_dir
self._db_path = os.path.join(db_dir, "lbrynet.sqlite")
log.info("connecting to database: %s", self._db_path)
self.db = SqliteConnection(self._db_path)
self.db.set_reactor(reactor)
self.clock = reactor
# used to refresh the claim attributes on a ManagedEncryptedFileDownloader when a
# change to the associated content claim occurs. these are added by the file manager
# when it loads each file
self.content_claim_callbacks = {} # {<stream_hash>: <callable returning a deferred>}
self.check_should_announce_lc = None
if 'reflector' not in conf.settings['components_to_skip']:
self.check_should_announce_lc = task.LoopingCall(self.verify_will_announce_all_head_and_sd_blobs)
@defer.inlineCallbacks
def setup(self):
def _create_tables(transaction):
transaction.executescript(self.CREATE_TABLES_QUERY)
yield self.db.runInteraction(_create_tables)
if self.check_should_announce_lc and not self.check_should_announce_lc.running:
self.check_should_announce_lc.start(600)
defer.returnValue(None)
@defer.inlineCallbacks
def run_and_return_one_or_none(self, query, *args):
result = yield self.db.runQuery(query, args)
if result:
defer.returnValue(result[0][0])
else:
defer.returnValue(None)
@defer.inlineCallbacks
def run_and_return_list(self, query, *args):
result = yield self.db.runQuery(query, args)
if result:
defer.returnValue([i[0] for i in result])
else:
defer.returnValue([])
def run_and_return_id(self, query, *args):
def do_save(t):
t.execute(query, args)
return t.lastrowid
return self.db.runInteraction(do_save)
def stop(self):
if self.check_should_announce_lc and self.check_should_announce_lc.running:
self.check_should_announce_lc.stop()
self.db.close()
return defer.succeed(True)
# # # # # # # # # blob functions # # # # # # # # #
def add_completed_blob(self, blob_hash, length, next_announce_time, should_announce, status="finished"):
log.debug("Adding a completed blob. blob_hash=%s, length=%i", blob_hash, length)
values = (blob_hash, length, next_announce_time or 0, int(bool(should_announce)), status, 0, 0)
return self.db.runOperation("insert or replace into blob values (?, ?, ?, ?, ?, ?, ?)", values)
def set_should_announce(self, blob_hash, next_announce_time, should_announce):
return self.db.runOperation(
"update blob set next_announce_time=?, should_announce=? where blob_hash=?",
(next_announce_time or 0, int(bool(should_announce)), blob_hash)
)
def get_blob_status(self, blob_hash):
return self.run_and_return_one_or_none(
"select status from blob where blob_hash=?", blob_hash
)
@defer.inlineCallbacks
def add_known_blob(self, blob_hash, length):
yield self.db.runOperation(
"insert or ignore into blob values (?, ?, ?, ?, ?, ?, ?)", (blob_hash, length, 0, 0, "pending", 0, 0)
)
def should_announce(self, blob_hash):
return self.run_and_return_one_or_none(
"select should_announce from blob where blob_hash=?", blob_hash
)
def count_should_announce_blobs(self):
return self.run_and_return_one_or_none(
"select count(*) from blob where should_announce=1 and status='finished'"
)
def get_all_should_announce_blobs(self):
return self.run_and_return_list(
"select blob_hash from blob where should_announce=1 and status='finished'"
)
@defer.inlineCallbacks
def get_all_finished_blobs(self):
blob_hashes = yield self.run_and_return_list(
"select blob_hash from blob where status='finished'"
)
defer.returnValue([unhexlify(blob_hash) for blob_hash in blob_hashes])
def count_finished_blobs(self):
return self.run_and_return_one_or_none(
"select count(*) from blob where status='finished'"
)
def update_last_announced_blob(self, blob_hash, last_announced):
return self.db.runOperation(
"update blob set next_announce_time=?, last_announced_time=?, single_announce=0 where blob_hash=?",
(int(last_announced + (dataExpireTimeout / 2)), int(last_announced), blob_hash)
)
def should_single_announce_blobs(self, blob_hashes, immediate=False):
def set_single_announce(transaction):
now = self.clock.seconds()
for blob_hash in blob_hashes:
if immediate:
transaction.execute(
"update blob set single_announce=1, next_announce_time=? "
"where blob_hash=? and status='finished'", (int(now), blob_hash)
)
else:
transaction.execute(
"update blob set single_announce=1 where blob_hash=? and status='finished'", (blob_hash, )
)
return self.db.runInteraction(set_single_announce)
def get_blobs_to_announce(self):
def get_and_update(transaction):
timestamp = self.clock.seconds()
if conf.settings['announce_head_blobs_only']:
r = transaction.execute(
"select blob_hash from blob "
"where blob_hash is not null and "
"(should_announce=1 or single_announce=1) and next_announce_time<? and status='finished'",
(timestamp,)
)
else:
r = transaction.execute(
"select blob_hash from blob where blob_hash is not null "
"and next_announce_time<? and status='finished'", (timestamp,)
)
blobs = [b[0] for b in r.fetchall()]
return blobs
return self.db.runInteraction(get_and_update)
def delete_blobs_from_db(self, blob_hashes):
def delete_blobs(transaction):
for blob_hash in blob_hashes:
transaction.execute("delete from blob where blob_hash=?;", (blob_hash,))
return self.db.runInteraction(delete_blobs)
def get_all_blob_hashes(self):
return self.run_and_return_list("select blob_hash from blob")
# # # # # # # # # stream blob functions # # # # # # # # #
def add_blobs_to_stream(self, stream_hash, blob_infos):
def _add_stream_blobs(transaction):
for blob_info in blob_infos:
transaction.execute("insert into stream_blob values (?, ?, ?, ?)",
(stream_hash, blob_info.get('blob_hash', None),
blob_info['blob_num'], blob_info['iv']))
return self.db.runInteraction(_add_stream_blobs)
@defer.inlineCallbacks
def add_known_blobs(self, blob_infos):
for blob_info in blob_infos:
if blob_info.get('blob_hash') and blob_info['length']:
yield self.add_known_blob(blob_info['blob_hash'], blob_info['length'])
def verify_will_announce_head_and_sd_blobs(self, stream_hash):
# fix should_announce for imported head and sd blobs
return self.db.runOperation(
"update blob set should_announce=1 "
"where should_announce=0 and "
"blob.blob_hash in "
" (select b.blob_hash from blob b inner join stream s on b.blob_hash=s.sd_hash and s.stream_hash=?) "
"or blob.blob_hash in "
" (select b.blob_hash from blob b "
" inner join stream_blob s2 on b.blob_hash=s2.blob_hash and s2.position=0 and s2.stream_hash=?)",
(stream_hash, stream_hash)
)
def verify_will_announce_all_head_and_sd_blobs(self):
return self.db.runOperation(
"update blob set should_announce=1 "
"where should_announce=0 and "
"blob.blob_hash in "
" (select b.blob_hash from blob b inner join stream s on b.blob_hash=s.sd_hash) "
"or blob.blob_hash in "
" (select b.blob_hash from blob b "
" inner join stream_blob s2 on b.blob_hash=s2.blob_hash and s2.position=0)"
)
# # # # # # # # # stream functions # # # # # # # # #
def store_stream(self, stream_hash, sd_hash, stream_name, stream_key, suggested_file_name,
stream_blob_infos):
"""
Add a stream to the stream table
:param stream_hash: hash of the assembled stream
:param sd_hash: hash of the sd blob
:param stream_key: blob decryption key
:param stream_name: the name of the file the stream was generated from
:param suggested_file_name: (str) suggested file name for stream
:param stream_blob_infos: (list) of blob info dictionaries
:return: (defer.Deferred)
"""
def _store_stream(transaction):
transaction.execute("insert into stream values (?, ?, ?, ?, ?);",
(stream_hash, sd_hash, stream_key, stream_name,
suggested_file_name))
for blob_info in stream_blob_infos:
transaction.execute("insert into stream_blob values (?, ?, ?, ?)",
(stream_hash, blob_info.get('blob_hash', None),
blob_info['blob_num'], blob_info['iv']))
return self.db.runInteraction(_store_stream)
@defer.inlineCallbacks
def delete_stream(self, stream_hash):
sd_hash = yield self.get_sd_blob_hash_for_stream(stream_hash)
stream_blobs = yield self.get_blobs_for_stream(stream_hash)
blob_hashes = [b.blob_hash for b in stream_blobs if b.blob_hash is not None]
def _delete_stream(transaction):
transaction.execute("delete from content_claim where stream_hash=? ", (stream_hash,))
transaction.execute("delete from file where stream_hash=? ", (stream_hash, ))
transaction.execute("delete from stream_blob where stream_hash=?", (stream_hash, ))
transaction.execute("delete from stream where stream_hash=? ", (stream_hash, ))
transaction.execute("delete from blob where blob_hash=?", (sd_hash, ))
for blob_hash in blob_hashes:
transaction.execute("delete from blob where blob_hash=?;", (blob_hash, ))
yield self.db.runInteraction(_delete_stream)
def get_all_streams(self):
return self.run_and_return_list("select stream_hash from stream")
def get_stream_info(self, stream_hash):
d = self.db.runQuery("select stream_name, stream_key, suggested_filename, sd_hash from stream "
"where stream_hash=?", (stream_hash, ))
d.addCallback(lambda r: None if not r else r[0])
return d
def check_if_stream_exists(self, stream_hash):
d = self.db.runQuery("select stream_hash from stream where stream_hash=?", (stream_hash, ))
d.addCallback(lambda r: True if len(r) else False)
return d
def get_blob_num_by_hash(self, stream_hash, blob_hash):
return self.run_and_return_one_or_none(
"select position from stream_blob where stream_hash=? and blob_hash=?",
stream_hash, blob_hash
)
def get_stream_blob_by_position(self, stream_hash, blob_num):
return self.run_and_return_one_or_none(
"select blob_hash from stream_blob where stream_hash=? and position=?",
stream_hash, blob_num
)
def get_blobs_for_stream(self, stream_hash, only_completed=False):
def _get_blobs_for_stream(transaction):
crypt_blob_infos = []
stream_blobs = transaction.execute(
"select blob_hash, position, iv from stream_blob where stream_hash=?", (stream_hash, )
).fetchall()
if only_completed:
lengths = transaction.execute(
"select b.blob_hash, b.blob_length from blob b "
"inner join stream_blob s ON b.blob_hash=s.blob_hash and b.status='finished' and s.stream_hash=?",
(stream_hash, )
).fetchall()
else:
lengths = transaction.execute(
"select b.blob_hash, b.blob_length from blob b "
"inner join stream_blob s ON b.blob_hash=s.blob_hash and s.stream_hash=?",
(stream_hash, )
).fetchall()
blob_length_dict = {}
for blob_hash, length in lengths:
blob_length_dict[blob_hash] = length
for blob_hash, position, iv in stream_blobs:
blob_length = blob_length_dict.get(blob_hash, 0)
crypt_blob_infos.append(CryptBlobInfo(blob_hash, position, blob_length, iv))
crypt_blob_infos = sorted(crypt_blob_infos, key=lambda info: info.blob_num)
return crypt_blob_infos
return self.db.runInteraction(_get_blobs_for_stream)
def get_pending_blobs_for_stream(self, stream_hash):
return self.run_and_return_list(
"select s.blob_hash from stream_blob s "
"inner join blob b on b.blob_hash=s.blob_hash and b.status='pending' "
"where stream_hash=?",
stream_hash
)
def get_stream_of_blob(self, blob_hash):
return self.run_and_return_one_or_none(
"select stream_hash from stream_blob where blob_hash=?", blob_hash
)
def get_sd_blob_hash_for_stream(self, stream_hash):
return self.run_and_return_one_or_none(
"select sd_hash from stream where stream_hash=?", stream_hash
)
def get_stream_hash_for_sd_hash(self, sd_blob_hash):
return self.run_and_return_one_or_none(
"select stream_hash from stream where sd_hash = ?", sd_blob_hash
)
# # # # # # # # # file stuff # # # # # # # # #
@defer.inlineCallbacks
def save_downloaded_file(self, stream_hash, file_name, download_directory, data_payment_rate):
# touch the closest available file to the file name
file_name = yield open_file_for_writing(unhexlify(download_directory).decode(), unhexlify(file_name).decode())
result = yield self.save_published_file(
stream_hash, hexlify(file_name.encode()), download_directory, data_payment_rate
)
defer.returnValue(result)
def save_published_file(self, stream_hash, file_name, download_directory, data_payment_rate, status="stopped"):
return self.run_and_return_id(
"insert into file values (?, ?, ?, ?, ?)",
stream_hash, file_name, download_directory, data_payment_rate, status
)
def get_filename_for_rowid(self, rowid):
return self.run_and_return_one_or_none("select file_name from file where rowid=?", rowid)
def get_all_lbry_files(self):
def _lbry_file_dict(rowid, stream_hash, file_name, download_dir, data_rate, status, _, sd_hash, stream_key,
stream_name, suggested_file_name):
return {
"row_id": rowid,
"stream_hash": stream_hash,
"file_name": file_name,
"download_directory": download_dir,
"blob_data_rate": data_rate,
"status": status,
"sd_hash": sd_hash,
"key": stream_key,
"stream_name": stream_name,
"suggested_file_name": suggested_file_name
}
def _get_all_files(transaction):
return [
_lbry_file_dict(*file_info) for file_info in transaction.execute(
"select file.rowid, file.*, stream.* "
"from file inner join stream on file.stream_hash=stream.stream_hash"
).fetchall()
]
d = self.db.runInteraction(_get_all_files)
return d
def change_file_status(self, rowid, new_status):
d = self.db.runQuery("update file set status=? where rowid=?", (new_status, rowid))
d.addCallback(lambda _: new_status)
return d
def get_lbry_file_status(self, rowid):
return self.run_and_return_one_or_none(
"select status from file where rowid = ?", rowid
)
def get_rowid_for_stream_hash(self, stream_hash):
return self.run_and_return_one_or_none(
"select rowid from file where stream_hash=?", stream_hash
)
# # # # # # # # # support functions # # # # # # # # #
def save_supports(self, claim_id, supports):
# TODO: add 'address' to support items returned for a claim from lbrycrdd and lbryum-server
def _save_support(transaction):
transaction.execute("delete from support where claim_id=?", (claim_id, ))
for support in supports:
transaction.execute(
"insert into support values (?, ?, ?, ?)",
("%s:%i" % (support['txid'], support['nout']), claim_id, int(support['amount'] * COIN),
support.get('address', ""))
)
return self.db.runInteraction(_save_support)
def get_supports(self, *claim_ids):
def _format_support(outpoint, supported_id, amount, address):
return {
"txid": outpoint.split(":")[0],
"nout": int(outpoint.split(":")[1]),
"claim_id": supported_id,
"amount": float(Decimal(amount) / Decimal(COIN)),
"address": address,
}
def _get_supports(transaction):
if len(claim_ids) == 1:
bind = "=?"
else:
bind = "in ({})".format(','.join('?' for _ in range(len(claim_ids))))
return [
_format_support(*support_info)
for support_info in transaction.execute(
f"select * from support where claim_id {bind}",
tuple(claim_ids)
).fetchall()
]
return self.db.runInteraction(_get_supports)
# # # # # # # # # claim functions # # # # # # # # #
@defer.inlineCallbacks
def save_claims(self, claim_infos):
def _save_claims(transaction):
content_claims_to_update = []
support_callbacks = []
for claim_info in claim_infos:
outpoint = "%s:%i" % (claim_info['txid'], claim_info['nout'])
claim_id = claim_info['claim_id']
name = claim_info['name']
amount = int(COIN * claim_info['amount'])
height = claim_info['height']
address = claim_info['address']
sequence = claim_info['claim_sequence']
try:
certificate_id = claim_info['value'].get('content_claims_to_update', {}).get('certificateId')
except AttributeError:
certificate_id = None
try:
if claim_info['value'].get('stream', {}).get('source', {}).get('sourceType') == "lbry_sd_hash":
source_hash = claim_info['value'].get('stream', {}).get('source', {}).get('source')
else:
source_hash = None
except AttributeError:
source_hash = None
serialized = claim_info.get('hex') or hexlify(smart_decode(claim_info['value']).serialized)
transaction.execute(
"insert or replace into claim values (?, ?, ?, ?, ?, ?, ?, ?, ?)",
(outpoint, claim_id, name, amount, height, serialized, certificate_id, address, sequence)
)
if 'supports' in claim_info: # if this response doesn't have support info don't overwrite the existing
# support info
support_callbacks.append(self.save_supports(claim_id, claim_info['supports']))
if not source_hash:
continue
stream_hash = transaction.execute(
"select file.stream_hash from stream "
"inner join file on file.stream_hash=stream.stream_hash where sd_hash=?", (source_hash, )
).fetchone()
if not stream_hash:
continue
stream_hash = stream_hash[0]
known_outpoint = transaction.execute(
"select claim_outpoint from content_claim where stream_hash=?", (stream_hash, )
)
known_claim_id = transaction.execute(
"select claim_id from claim "
"inner join content_claim c3 ON claim.claim_outpoint=c3.claim_outpoint "
"where c3.stream_hash=?", (stream_hash, )
)
if not known_claim_id:
content_claims_to_update.append((stream_hash, outpoint))
elif known_outpoint != outpoint:
content_claims_to_update.append((stream_hash, outpoint))
update_file_callbacks = []
for stream_hash, outpoint in content_claims_to_update:
self._save_content_claim(transaction, outpoint, stream_hash)
if stream_hash in self.content_claim_callbacks:
update_file_callbacks.append(self.content_claim_callbacks[stream_hash]())
return update_file_callbacks, support_callbacks
content_dl, support_dl = yield self.db.runInteraction(_save_claims)
if content_dl:
yield defer.DeferredList(content_dl)
if support_dl:
yield defer.DeferredList(support_dl)
def save_claims_for_resolve(self, claim_infos):
to_save = []
for info in claim_infos:
if 'value' in info:
if info['value']:
to_save.append(info)
else:
if 'certificate' in info and info['certificate']['value']:
to_save.append(info['certificate'])
if 'claim' in info and info['claim']['value']:
to_save.append(info['claim'])
return self.save_claims(to_save)
def get_old_stream_hashes_for_claim_id(self, claim_id, new_stream_hash):
return self.run_and_return_list(
"select f.stream_hash from file f "
"inner join content_claim cc on f.stream_hash=cc.stream_hash "
"inner join claim c on c.claim_outpoint=cc.claim_outpoint and c.claim_id=? "
"where f.stream_hash!=?", claim_id, new_stream_hash
)
@staticmethod
def _save_content_claim(transaction, claim_outpoint, stream_hash):
# get the claim id and serialized metadata
claim_info = transaction.execute(
"select claim_id, serialized_metadata from claim where claim_outpoint=?", (claim_outpoint,)
).fetchone()
if not claim_info:
raise Exception("claim not found")
new_claim_id, claim = claim_info[0], ClaimDict.deserialize(unhexlify(claim_info[1]))
# certificate claims should not be in the content_claim table
if not claim.is_stream:
raise Exception("claim does not contain a stream")
# get the known sd hash for this stream
known_sd_hash = transaction.execute(
"select sd_hash from stream where stream_hash=?", (stream_hash,)
).fetchone()
if not known_sd_hash:
raise Exception("stream not found")
# check the claim contains the same sd hash
if known_sd_hash[0].encode() != claim.source_hash:
raise Exception("stream mismatch")
# if there is a current claim associated to the file, check that the new claim is an update to it
current_associated_content = transaction.execute(
"select claim_outpoint from content_claim where stream_hash=?", (stream_hash,)
).fetchone()
if current_associated_content:
current_associated_claim_id = transaction.execute(
"select claim_id from claim where claim_outpoint=?", current_associated_content
).fetchone()[0]
if current_associated_claim_id != new_claim_id:
raise Exception(
f"mismatching claim ids when updating stream {current_associated_claim_id} vs {new_claim_id}"
)
# update the claim associated to the file
transaction.execute("insert or replace into content_claim values (?, ?)", (stream_hash, claim_outpoint))
@defer.inlineCallbacks
def save_content_claim(self, stream_hash, claim_outpoint):
yield self.db.runInteraction(self._save_content_claim, claim_outpoint, stream_hash)
# update corresponding ManagedEncryptedFileDownloader object
if stream_hash in self.content_claim_callbacks:
file_callback = self.content_claim_callbacks[stream_hash]
yield file_callback()
@defer.inlineCallbacks
def get_content_claim(self, stream_hash, include_supports=True):
def _get_claim_from_stream_hash(transaction):
claim_info = transaction.execute(
"select c.*, "
"case when c.channel_claim_id is not null then "
"(select claim_name from claim where claim_id==c.channel_claim_id) "
"else null end as channel_name from content_claim "
"inner join claim c on c.claim_outpoint=content_claim.claim_outpoint "
"and content_claim.stream_hash=? order by c.rowid desc", (stream_hash,)
).fetchone()
if not claim_info:
return None
channel_name = claim_info[-1]
result = _format_claim_response(*claim_info[:-1])
if channel_name:
result['channel_name'] = channel_name
return result
result = yield self.db.runInteraction(_get_claim_from_stream_hash)
if result and include_supports:
supports = yield self.get_supports(result['claim_id'])
result['supports'] = supports
result['effective_amount'] = float(
sum([support['amount'] for support in supports]) + result['amount']
)
defer.returnValue(result)
@defer.inlineCallbacks
def get_claims_from_stream_hashes(self, stream_hashes, include_supports=True):
def _batch_get_claim(transaction):
results = {}
bind = "({})".format(','.join('?' for _ in range(len(stream_hashes))))
claim_infos = transaction.execute(
"select content_claim.stream_hash, c.* from content_claim "
"inner join claim c on c.claim_outpoint=content_claim.claim_outpoint "
"and content_claim.stream_hash in {} order by c.rowid desc".format(bind),
tuple(stream_hashes)
).fetchall()
channel_id_infos = {}
for claim_info in claim_infos:
if claim_info[7]:
streams = channel_id_infos.get(claim_info[7], [])
streams.append(claim_info[0])
channel_id_infos[claim_info[7]] = streams
for claim_info in claim_infos:
stream_hash = claim_info[0]
result = _format_claim_response(*claim_info[1:])
results[stream_hash] = result
bind = "({})".format(','.join('?' for _ in range(len(channel_id_infos))))
for claim_id, channel_name in transaction.execute(
f"select claim_id, claim_name from claim where claim_id in {bind}",
tuple(channel_id_infos.keys())
).fetchall():
for stream_hash in channel_id_infos[claim_id]:
results[stream_hash]['channel_name'] = channel_name
return results
claims = yield self.db.runInteraction(_batch_get_claim)
if include_supports:
all_supports = {}
for support in (yield self.get_supports(*[claim['claim_id'] for claim in claims.values()])):
all_supports.setdefault(support['claim_id'], []).append(support)
for stream_hash in claims.keys():
claim = claims[stream_hash]
supports = all_supports.get(claim['claim_id'], [])
claim['supports'] = supports
claim['effective_amount'] = float(
sum([support['amount'] for support in supports]) + claim['amount']
)
claims[stream_hash] = claim
defer.returnValue(claims)
@defer.inlineCallbacks
def get_claim(self, claim_outpoint, include_supports=True):
def _get_claim(transaction):
claim_info = transaction.execute("select c.*, "
"case when c.channel_claim_id is not null then "
"(select claim_name from claim where claim_id==c.channel_claim_id) "
"else null end as channel_name from claim c where claim_outpoint = ?",
(claim_outpoint,)).fetchone()
channel_name = claim_info[-1]
result = _format_claim_response(*claim_info[:-1])
if channel_name:
result['channel_name'] = channel_name
return result
result = yield self.db.runInteraction(_get_claim)
if include_supports:
supports = yield self.get_supports(result['claim_id'])
result['supports'] = supports
result['effective_amount'] = float(
sum([support['amount'] for support in supports]) + result['amount']
)
defer.returnValue(result)
def get_unknown_certificate_ids(self):
def _get_unknown_certificate_claim_ids(transaction):
return [
claim_id for (claim_id,) in transaction.execute(
"select distinct c1.channel_claim_id from claim as c1 "
"where c1.channel_claim_id!='' "
"and c1.channel_claim_id not in "
"(select c2.claim_id from claim as c2)"
).fetchall()
]
return self.db.runInteraction(_get_unknown_certificate_claim_ids)
@defer.inlineCallbacks
def get_pending_claim_outpoints(self):
claim_outpoints = yield self.run_and_return_list("select claim_outpoint from claim where height=-1")
results = {} # {txid: [nout, ...]}
for outpoint_str in claim_outpoints:
txid, nout = outpoint_str.split(":")
outputs = results.get(txid, [])
outputs.append(int(nout))
results[txid] = outputs
if results:
log.debug("missing transaction heights for %i claims", len(results))
defer.returnValue(results)
def save_claim_tx_heights(self, claim_tx_heights):
def _save_claim_heights(transaction):
for outpoint, height in claim_tx_heights.items():
transaction.execute(
"update claim set height=? where claim_outpoint=? and height=-1",
(height, outpoint)
)
return self.db.runInteraction(_save_claim_heights)
# # # # # # # # # reflector functions # # # # # # # # #
def update_reflected_stream(self, sd_hash, reflector_address, success=True):
if success:
return self.db.runOperation(
"insert or replace into reflected_stream values (?, ?, ?)",
(sd_hash, reflector_address, self.clock.seconds())
)
return self.db.runOperation(
"delete from reflected_stream where sd_hash=? and reflector_address=?",
(sd_hash, reflector_address)
)
def get_streams_to_re_reflect(self):
return self.run_and_return_list(
"select s.sd_hash from stream s "
"left outer join reflected_stream r on s.sd_hash=r.sd_hash "
"where r.timestamp is null or r.timestamp < ?",
self.clock.seconds() - conf.settings['auto_re_reflect_interval']
)
# Helper functions
def _format_claim_response(outpoint, claim_id, name, amount, height, serialized, channel_id, address, claim_sequence):
r = {
"name": name,
"claim_id": claim_id,
"address": address,
"claim_sequence": claim_sequence,
"value": ClaimDict.deserialize(unhexlify(serialized)).claim_dict,
"height": height,
"amount": float(Decimal(amount) / Decimal(COIN)),
"nout": int(outpoint.split(":")[1]),
"txid": outpoint.split(":")[0],
"channel_claim_id": channel_id,
"channel_name": None
}
return r