faster population of channel names

This commit is contained in:
Jack Robison 2019-02-08 19:51:13 -05:00
parent 8abfdc82b2
commit aeb2891baa
No known key found for this signature in database
GPG key ID: DF25C68FE0239BB2

View file

@ -105,44 +105,53 @@ def get_content_claim_from_outpoint(transaction: sqlite3.Connection,
return StoredStreamClaim(*claim_fields)
def _batched_select(transaction, query, parameters):
for start_index in range(0, len(parameters), 900):
current_batch = parameters[start_index:start_index+900]
def _batched_select(transaction, query, parameters, batch_size=900):
for start_index in range(0, len(parameters), batch_size):
current_batch = parameters[start_index:start_index+batch_size]
bind = "({})".format(','.join(['?'] * len(current_batch)))
for result in transaction.execute(query.format(bind), current_batch):
yield result
def get_all_lbry_files(transaction: sqlite3.Connection) -> typing.List[typing.Dict]:
return [
{
"row_id": rowid,
"stream_hash": stream_hash,
"file_name": file_name, # hex
"download_directory": download_dir, # hex
"blob_data_rate": data_rate,
"status": status,
"sd_hash": sd_hash,
"key": stream_key,
"stream_name": stream_name, # hex
"suggested_file_name": suggested_file_name, # hex
"claim": StoredStreamClaim(stream_hash, *claim_args)
} for (rowid, stream_hash, file_name, download_dir, data_rate, status, _, sd_hash, stream_key,
stream_name, suggested_file_name, *claim_args) in _batched_select(
transaction, "select file.rowid, file.*, stream.*, 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 "
files = []
signed_claims = {}
for (rowid, stream_hash, file_name, download_dir, data_rate, status, _, sd_hash, stream_key,
stream_name, suggested_file_name, *claim_args) in _batched_select(
transaction, "select file.rowid, file.*, stream.*, c.* "
"from file inner join stream on file.stream_hash=stream.stream_hash "
"inner join content_claim cc on file.stream_hash=cc.stream_hash "
"inner join claim c on cc.claim_outpoint=c.claim_outpoint "
"where file.stream_hash in {} "
"order by c.rowid desc",
[
stream_hash
for (stream_hash,) in transaction.execute("select stream_hash from file")
]
"order by c.rowid desc", [
stream_hash for (stream_hash,) in transaction.execute("select stream_hash from file")]):
claim = StoredStreamClaim(stream_hash, *claim_args)
if claim.channel_claim_id:
if claim.channel_claim_id not in signed_claims:
signed_claims[claim.channel_claim_id] = []
signed_claims[claim.channel_claim_id].append(claim)
files.append(
{
"row_id": rowid,
"stream_hash": stream_hash,
"file_name": file_name, # hex
"download_directory": download_dir, # hex
"blob_data_rate": data_rate,
"status": status,
"sd_hash": sd_hash,
"key": stream_key,
"stream_name": stream_name, # hex
"suggested_file_name": suggested_file_name, # hex
"claim": claim
}
)
]
for claim_id in signed_claims.keys():
channel_name = transaction.execute("select claim_name from claim where claim_id=?", (claim_id, )).fetchone()
if channel_name:
for claim in signed_claims[claim_id]:
claim.channel_name = channel_name[0]
return files
class SQLiteStorage(SQLiteMixin):
@ -441,7 +450,7 @@ class SQLiteStorage(SQLiteMixin):
binascii.hexlify(download_directory.encode()).decode(), data_payment_rate, status)
)
def get_all_lbry_files(self) -> typing.List[typing.Dict]:
def get_all_lbry_files(self) -> typing.Awaitable[typing.List[typing.Dict]]:
return self.db.run(get_all_lbry_files)
def change_file_status(self, stream_hash: str, new_status: str):