mirror
/
moonstream
kopia lustrzana https://github.com/bugout-dev/moonstream
1
0
Forkuj 0
Kod Zgłoszenia Projekty Wydania Wiki Aktywność

Merge branch 'main' into crawl-smart-contract-addresses-from-transaction-receipts

pull/27/head
Neeraj Kashyap 2021-07-29 09:42:55 -07:00
rodzic 9388bbe6a9 5b7cdf6832
commit d2c30256b3
6 zmienionych plików z 166 dodań i 91 usunięć
Pokaż statystyki Ściągnij plik aktualizacji Ściągnij plik porównania Expand all files Collapse all files

3
backend/deploy/deploy.bash
Wyświetl plik

@ -4,6 +4,7 @@
# Main
APP_DIR="${APP_DIR:-/home/ubuntu/moonstream}"
APP_BACKEND_DIR="${APP_DIR}/backend"
AWS_DEFAULT_REGION="${AWS_DEFAULT_REGION:-us-east-1}"
PYTHON_ENV_DIR="${PYTHON_ENV_DIR:-/home/ubuntu/moonstream-env}"
PYTHON="${PYTHON_ENV_DIR}/bin/python"
@ -25,7 +26,7 @@ echo "Updating pip and setuptools"
echo
echo
echo "Updating Python dependencies"
"${PIP}" install -r "${APP_DIR}/requirements.txt"
"${PIP}" install -r "${APP_BACKEND_DIR}/requirements.txt"
echo
echo

2
backend/deploy/moonstream.service
Wyświetl plik

@ -5,7 +5,7 @@ After=network.target
[Service]
User=ubuntu
Group=www-data
WorkingDirectory=/home/ubuntu/moonstream
WorkingDirectory=/home/ubuntu/moonstream/backend
EnvironmentFile=/home/ubuntu/moonstream-secrets/app.env
ExecStart=/home/ubuntu/moonstream-env/bin/uvicorn --host 0.0.0.0 --port 7481 --workers 8 moonstream.api:app
SyslogIdentifier=moonstream

2
backend/requirements.txt
Wyświetl plik

@ -11,7 +11,7 @@ fastapi==0.66.0
h11==0.12.0
idna==3.2
jmespath==0.10.0
-e git+ssh://git@github.com/bugout-dev/moonstream.git@ad37dbd464fb9bb81907ab0b16b0e810e9da738f#egg=moonstreamdb&subdirectory=db
-e git+https://git@github.com/bugout-dev/moonstream.git@b9c828fc7f811af88a9f3a45dd7f5c4053433366#egg=moonstreamdb&subdirectory=db
mypy==0.910
mypy-extensions==0.4.3
pathspec==0.9.0

161
crawlers/moonstreamcrawlers/cli.py
Wyświetl plik

@ -6,81 +6,95 @@ from distutils.util import strtobool
import json
import sys
import time
from typing import List
from .ethereum import crawl, check_missing_blocks, process_contract_deployments
from .ethereum import (
crawl_blocks_executor,
crawl_blocks,
check_missing_blocks,
get_latest_blocks,
process_contract_deployments,
)
from .settings import MOONSTREAM_CRAWL_WORKERS
def get_blocks_numbers_lists(
bottom_block_number: int, top_block_number: int
) -> List[List[int]]:
def yield_blocks_numbers_lists(blocks_range_str: str) -> None:
"""
Generate list of blocks.
Block steps used to prevent long executor tasks and data loss possibility.
"""
block_step = 1000
blocks_numbers_list_raw = list(range(top_block_number, bottom_block_number - 1, -1))
blocks_numbers_list_raw_len = len(blocks_numbers_list_raw)
# Block steps used to prevent long executor tasks and data loss possibility
# Block step 2 convert [1,2,3] -> [[1,2],[3]]
if blocks_numbers_list_raw_len / block_step > 1:
blocks_numbers_lists = [
blocks_numbers_list_raw[i : i + block_step]
for i in range(0, blocks_numbers_list_raw_len, block_step)
]
else:
blocks_numbers_lists = [blocks_numbers_list_raw]
return blocks_numbers_lists, blocks_numbers_list_raw_len
try:
blocks_start_end = blocks_range_str.split("-")
bottom_block_number = int(blocks_start_end[0])
top_block_number = int(blocks_start_end[1])
required_blocks_len = top_block_number - bottom_block_number + 1
except Exception:
print(
"Wrong format provided, expected {bottom_block}-{top_block}, as ex. 105-340"
)
return
print(f"Required {required_blocks_len} blocks to process")
while not top_block_number < bottom_block_number:
temp_bottom_block_number = top_block_number - block_step
if temp_bottom_block_number < bottom_block_number:
temp_bottom_block_number = bottom_block_number - 1
blocks_numbers_list = list(
range(top_block_number, temp_bottom_block_number, -1)
)
yield blocks_numbers_list
top_block_number -= block_step
def ethcrawler_blocks_sync_handler(args: argparse.Namespace) -> None:
"""
Synchronize latest Ethereum blocks with database.
"""
while True:
bottom_block_number, top_block_number = get_latest_blocks(
bool(strtobool(args.transactions))
)
if bottom_block_number >= top_block_number:
print(
f"Synchronization is unnecessary for blocks {bottom_block_number}-{top_block_number}"
)
break
for blocks_numbers_list in yield_blocks_numbers_lists(
f"{bottom_block_number}-{top_block_number}"
):
print(f"Adding blocks {blocks_numbers_list[0]}-{blocks_numbers_list[-1]}")
crawl_blocks_executor(
block_numbers_list=blocks_numbers_list,
with_transactions=bool(strtobool(args.transactions)),
)
print(f"Synchronized blocks from {bottom_block_number} to {top_block_number}")
time.sleep(10)
def ethcrawler_blocks_add_handler(args: argparse.Namespace) -> None:
"""
Add blocks to moonstream database.
"""
try:
blocks_start_end = args.blocks.split("-")
bottom_block_number = int(blocks_start_end[0])
top_block_number = int(blocks_start_end[1])
except Exception:
print(
"Wrong format provided, expected {bottom_block}-{top_block}, as ex. 105-340"
)
return
blocks_numbers_lists, blocks_numbers_list_raw_len = get_blocks_numbers_lists(
bottom_block_number, top_block_number
)
startTime = time.time()
for blocks_numbers_list in blocks_numbers_lists:
crawl(
for blocks_numbers_list in yield_blocks_numbers_lists(args.blocks):
print(f"Adding blocks {blocks_numbers_list[0]}-{blocks_numbers_list[-1]}")
crawl_blocks_executor(
block_numbers_list=blocks_numbers_list,
with_transactions=bool(strtobool(args.transactions)),
)
print(
f"Required time: {time.time() - startTime} for: {blocks_numbers_list_raw_len} "
f"blocks with {MOONSTREAM_CRAWL_WORKERS} workers"
)
print(f"Required {time.time() - startTime} with {MOONSTREAM_CRAWL_WORKERS} workers")
def ethcrawler_blocks_missing_handler(args: argparse.Namespace) -> None:
try:
blocks_start_end = args.blocks.split("-")
bottom_block_number = int(blocks_start_end[0])
top_block_number = int(blocks_start_end[1])
except Exception:
print(
"Wrong format provided, expected {bottom_block}-{top_block}, as ex. 105-340"
)
return
blocks_numbers_lists, blocks_numbers_list_raw_len = get_blocks_numbers_lists(
bottom_block_number, top_block_number
)
startTime = time.time()
missing_blocks_numbers_total = []
for blocks_numbers_list in blocks_numbers_lists:
for blocks_numbers_list in yield_blocks_numbers_lists(args.blocks):
print(
f"Check missing blocks from {blocks_numbers_list[0]} to {blocks_numbers_list[-1]}"
)
@ -93,14 +107,22 @@ def ethcrawler_blocks_missing_handler(args: argparse.Namespace) -> None:
time.sleep(5)
if (len(missing_blocks_numbers_total)) > 0:
crawl(
missing_blocks_numbers_total,
with_transactions=bool(strtobool(args.transactions)),
)
if bool(strtobool(args.lazy)):
print("Executed lazy block crawler")
crawl_blocks(
missing_blocks_numbers_total,
with_transactions=bool(strtobool(args.transactions)),
verbose=True,
)
else:
crawl_blocks_executor(
missing_blocks_numbers_total,
with_transactions=bool(strtobool(args.transactions)),
verbose=True,
)
print(
f"Required time: {time.time() - startTime} for: {blocks_numbers_list_raw_len} "
f"blocks with {MOONSTREAM_CRAWL_WORKERS} workers"
f" with {len(missing_blocks_numbers_total)} missing blocks"
f"Required {time.time() - startTime} with {MOONSTREAM_CRAWL_WORKERS} workers "
f"for {len(missing_blocks_numbers_total)} missing blocks"
)
@ -134,6 +156,18 @@ def main() -> None:
description="Ethereum blocks commands"
)
parser_ethcrawler_blocks_sync = subcommands_ethcrawler_blocks.add_parser(
"synchronize", description="Synchronize to latest ethereum block commands"
)
parser_ethcrawler_blocks_sync.add_argument(
"-t",
"--transactions",
choices=["True", "False"],
default="True",
help="Add or not block transactions",
)
parser_ethcrawler_blocks_sync.set_defaults(func=ethcrawler_blocks_sync_handler)
parser_ethcrawler_blocks_add = subcommands_ethcrawler_blocks.add_parser(
"add", description="Add ethereum blocks commands"
)
@ -147,7 +181,7 @@ def main() -> None:
"-t",
"--transactions",
choices=["True", "False"],
default="False",
default="True",
help="Add or not block transactions",
)
parser_ethcrawler_blocks_add.set_defaults(func=ethcrawler_blocks_add_handler)
@ -165,9 +199,16 @@ def main() -> None:
"-t",
"--transactions",
choices=["True", "False"],
default="False",
default="True",
help="Add or not block transactions",
)
parser_ethcrawler_blocks_missing.add_argument(
"-l",
"--lazy",
choices=["True", "False"],
default="False",
help="Lazy block adding one by one",
)
parser_ethcrawler_blocks_missing.set_defaults(
func=ethcrawler_blocks_missing_handler
)

87
crawlers/moonstreamcrawlers/ethereum.py
Wyświetl plik

@ -19,7 +19,7 @@ def connect(ipc_path: str = MOONSTREAM_IPC_PATH):
return web3_client
def add_block(db_session, block: BlockData, block_number: int) -> None:
def add_block(db_session, block: BlockData) -> None:
"""
Add block if doesn't presented in database.
"""
@ -43,29 +43,46 @@ def add_block(db_session, block: BlockData, block_number: int) -> None:
transactions_root=block.transactionsRoot.hex(),
)
db_session.add(block_obj)
print(f"Added new block: {block_number}")
def add_block_transaction(db_session, block_number: int, tx) -> None:
def add_block_transactions(db_session, block: BlockData) -> None:
"""
Add block transaction if doesn't presented in database.
Add block transactions.
"""
tx_obj = EthereumTransaction(
hash=tx.hash.hex(),
block_number=block_number,
from_address=tx["from"],
to_address=tx.to,
gas=tx.gas,
gas_price=tx.gasPrice,
input=tx.input,
nonce=tx.nonce,
transaction_index=tx.transactionIndex,
value=tx.value,
for tx in block.transactions:
tx_obj = EthereumTransaction(
hash=tx.hash.hex(),
block_number=block.number,
from_address=tx["from"],
to_address=tx.to,
gas=tx.gas,
gas_price=tx.gasPrice,
input=tx.input,
nonce=tx.nonce,
transaction_index=tx.transactionIndex,
value=tx.value,
)
db_session.add(tx_obj)
def get_latest_blocks(with_transactions: bool = False) -> None:
web3_client = connect()
block_latest: BlockData = web3_client.eth.get_block(
"latest", full_transactions=with_transactions
)
db_session.add(tx_obj)
with yield_db_session_ctx() as db_session:
block_latest_exist = (
db_session.query(EthereumBlock)
.order_by(EthereumBlock.block_number.desc())
.first()
)
return block_latest_exist.block_number, block_latest.number
def process_blocks(blocks_numbers: List[int], with_transactions: bool = False):
def crawl_blocks(
blocks_numbers: List[int], with_transactions: bool = False, verbose: bool = False
) -> None:
"""
Open database and geth sessions and fetch block data from blockchain.
"""
@ -75,14 +92,22 @@ def process_blocks(blocks_numbers: List[int], with_transactions: bool = False):
block: BlockData = web3_client.eth.get_block(
block_number, full_transactions=with_transactions
)
add_block(db_session, block, block_number)
add_block(db_session, block)
if with_transactions:
for tx in block.transactions:
add_block_transaction(db_session, block.number, tx)
add_block_transactions(db_session, block)
db_session.commit()
if verbose:
print(f"Added {block_number} block")
def check_missing_blocks(blocks_numbers: List[int]):
def check_missing_blocks(blocks_numbers: List[int]) -> List[int]:
"""
Query block from postgres. If block does not presented in database,
add to missing blocks numbers list.
"""
missing_blocks_numbers = []
for block_number in blocks_numbers:
with yield_db_session_ctx() as db_session:
@ -96,18 +121,24 @@ def check_missing_blocks(blocks_numbers: List[int]):
return missing_blocks_numbers
def crawl(block_numbers_list: List[int], with_transactions: bool = False) -> None:
def crawl_blocks_executor(
block_numbers_list: List[int],
with_transactions: bool = False,
verbose: bool = False,
) -> None:
"""
Execute crawler.
Execute crawler in processes.
"""
with ProcessPoolExecutor(max_workers=MOONSTREAM_CRAWL_WORKERS) as executor:
for worker in range(1, MOONSTREAM_CRAWL_WORKERS + 1):
print(
f"Added executor for list of blocks with len: {len(block_numbers_list[worker-1::MOONSTREAM_CRAWL_WORKERS])}"
)
worker_block_numbers_list = block_numbers_list[
worker - 1 :: MOONSTREAM_CRAWL_WORKERS
]
if verbose:
print(f"Spawned process for {len(worker_block_numbers_list)} blocks")
executor.submit(
process_blocks,
block_numbers_list[worker - 1 :: MOONSTREAM_CRAWL_WORKERS],
crawl_blocks,
worker_block_numbers_list,
with_transactions,
)

2
db/.gitignore vendored
Wyświetl plik

@ -163,6 +163,8 @@ cython_debug/
# Custom
dev.env
prod.env
dev.env.ps1
prod.env.ps1
alembic.dev.ini
alembic.prod.ini
.db/