cryo is the easiest way to extract blockchain data to parquet, csv, json, or a python dataframe.
cryo is also extremely flexible, with many different options to control how data is extracted + filtered + formatted
cryo is an early WIP, please report bugs + feedback to the issue tracker
note that cryo's default settings will slam a node too hard for use with 3rd party RPC providers. Instead, --requests-per-second and --max-concurrent-requests should be used to impose ratelimits. Such settings will be handled automatically in a future release.
to discuss cryo, check out the telegram group
use as cryo <dataset> [OPTIONS]
| Example | Command |
|---|---|
| Extract all logs from block 16,000,000 to block 17,000,000 | cryo logs -b 16M:17M |
| Extract blocks, logs, or traces missing from current directory | cryo blocks txs traces |
| Extract to csv instead of parquet | cryo blocks txs traces --csv |
| Extract only certain columns | cryo blocks --include number timestamp |
| Dry run to view output schemas or expected work | cryo storage_diffs --dry |
| Extract all USDC events | cryo logs --contract 0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48 |
For a more complex example, see the Uniswap Example.
cryo uses ETH_RPC_URL env var as the data source unless --rpc <url> is given
The simplest way to use cryo is as a cli tool:
git clone https://github.com/paradigmxyz/cryo
cd cryo
cargo install --path ./crates/cliThis method requires having rust installed. See rustup for instructions.
cargo install cryo_cliThis method requires having rust installed. See rustup for instructions.
Make sure that ~/.cargo/bin is on your PATH. One way to do this is by adding the line export PATH="$HOME/.cargo/bin:$PATH" to your ~/.bashrc or ~/.profile.
cryo can also be installed as a python package:
(make sure rust is installed first, see rustup)
pip install maturin
pip install cryopip install maturin
git clone https://github.com/paradigmxyz/cryo
cd cryo/crates/python
maturin build --release
pip install --force-reinstall <OUTPUT_OF_MATURIN_BUILD>.whlMany cryo cli options will affect output schemas by adding/removing columns or changing column datatypes.
cryo will always print out data schemas before collecting any data. To view these schemas without collecting data, use --dry to perform a dry run.
An attempt is made to ensure that the dataset schemas conform to a common set of design guidelines:
- By default, rows should contain enough information in their columns to be order-able (unless the rows do not have an intrinsic order).
- Columns should usually be named by their JSON-RPC or ethers.rs defaults, except in cases where a much more explicit name is available.
- To make joins across tables easier, a given piece of information should use the same datatype and column name across tables when possible.
- Large ints such as
u256should allow multiple conversions. Avaluecolumn of typeu256should allow:value_binary,value_string,value_f32,value_f64,value_u32,value_u64, andvalue_d128. These types can be specified at runtime using the--u256-typesargument. - By default, columns related to non-identifying cryptographic signatures are omitted by default. For example,
state_rootof a block orv/r/sof a transaction. - Integer values that can never be negative should be stored as unsigned integers.
- Every table should allow a
chain_idcolumn so that data from multiple chains can be easily stored in the same table.
Standard types across tables:
block_number:u32transaction_index:u32nonce:u32gas_used:u64gas_limit:u64chain_id:u64timestamp:u32
cryo currently obtains all of its data using the JSON-RPC protocol standard.
| dataset | blocks per request | results per block | method |
|---|---|---|---|
| Blocks | 1 | 1 | eth_getBlockByNumber |
| Transactions | 1 | multiple | eth_getBlockByNumber, eth_getBlockReceipts, eth_getTransactionReceipt |
| Logs | multiple | multiple | eth_getLogs |
| Contracts | 1 | multiple | trace_block |
| Traces | 1 | multiple | trace_block |
| State Diffs | 1 | multiple | trace_replayBlockTransactions |
| Vm Traces | 1 | multiple | trace_replayBlockTransactions |
cryo use ethers.rs to perform JSON-RPC requests, so it can be used any chain that ethers-rs is compatible with. This includes Ethereum, Optimism, Arbitrum, Polygon, BNB, and Avalanche.
A future version of cryo will be able to bypass JSON-RPC and query node data directly.
- Code is arranged into the following crates:
cryo_cli: convert textual data into cryo function callscryo_freeze: core cryo codecryo_python: cryo python adaptercryo_to_df: procedural macro for generating dataset definitions
- Do not use panics (including
panic!,todo!,unwrap(), andexpect()) except in the following circumstances: tests, build scripts, lazy static blocks, and procedural macros
(output of cryo help)
cryo extracts blockchain data to parquet, csv, or json
Usage: cryo [OPTIONS] [DATATYPE]...
Arguments:
[DATATYPE]... datatype(s) to collect, use cryo datasets to see all available
Options:
--remember Remember current command for future use
-v, --verbose Extra verbosity
--no-verbose Run quietly without printing information to stdout
-h, --help Print help
-V, --version Print version
Content Options:
-b, --blocks <BLOCKS>... Block numbers, see syntax below
--timestamps <TIMESTAMPS>... Timestamp numbers in unix, overriden by blocks
-t, --txs <TXS>... Transaction hashes, see syntax below
-a, --align Align chunk boundaries to regular intervals,
e.g. (1000 2000 3000), not (1106 2106 3106)
--reorg-buffer <N_BLOCKS> Reorg buffer, save blocks only when this old,
can be a number of blocks [default: 0]
-i, --include-columns [<COLS>...] Columns to include alongside the defaults,
use `all` to include all available columns
-e, --exclude-columns [<COLS>...] Columns to exclude from the defaults
--columns [<COLS>...] Columns to use instead of the defaults,
use `all` to use all available columns
--u256-types <U256_TYPES>... Set output datatype(s) of U256 integers
[default: binary, string, f64]
--hex Use hex string encoding for binary columns
-s, --sort [<SORT>...] Columns(s) to sort by, `none` for unordered
--exclude-failed Exclude items from failed transactions
Source Options:
-r, --rpc <RPC> RPC url [default: ETH_RPC_URL env var]
--network-name <NETWORK_NAME> Network name [default: name of eth_getChainId]
Acquisition Options:
-l, --requests-per-second <limit> Ratelimit on requests per second
--max-retries <R> Max retries for provider errors [default: 5]
--initial-backoff <B> Initial retry backoff time (ms) [default: 500]
--max-concurrent-requests <M> Global number of concurrent requests
--max-concurrent-chunks <M> Number of chunks processed concurrently
--chunk-order <CHUNK_ORDER> Chunk collection order (normal, reverse, or random)
-d, --dry Dry run, collect no data
Output Options:
-c, --chunk-size <CHUNK_SIZE> Number of blocks per file [default: 1000]
--n-chunks <N_CHUNKS> Number of files (alternative to --chunk-size)
--partition-by <PARTITION_BY> Dimensions to partition by
-o, --output-dir <OUTPUT_DIR> Directory for output files [default: .]
--subdirs <SUBDIRS>... Subdirectories for output files
can be `datatype`, `network`, or custom string
--label <LABEL> Label to add to each filename
--overwrite Overwrite existing files instead of skipping
--csv Save as csv instead of parquet
--json Save as json instead of parquet
--row-group-size <GROUP_SIZE> Number of rows per row group in parquet file
--n-row-groups <N_ROW_GROUPS> Number of rows groups in parquet file
--no-stats Do not write statistics to parquet files
--compression <NAME [#]>... Compression algorithm and level [default: lz4]
--report-dir <REPORT_DIR> Directory to save summary report
[default: {output_dir}/.cryo/reports]
--no-report Avoid saving a summary report
Dataset-specific Options:
--address <ADDRESS>... Address(es)
--to-address <address>... To Address(es)
--from-address <address>... From Address(es)
--call-data <CALL_DATA>... Call data(s) to use for eth_calls
--function <FUNCTION>... Function(s) to use for eth_calls
--inputs <INPUTS>... Input(s) to use for eth_calls
--slot <SLOT>... Slot(s)
--contract <CONTRACT>... Contract address(es)
--topic0 <TOPIC0>... Topic0(s) [aliases: event]
--topic1 <TOPIC1>... Topic1(s)
--topic2 <TOPIC2>... Topic2(s)
--topic3 <TOPIC3>... Topic3(s)
--event-signature <SIG>... Event signature for log decoding
--inner-request-size <BLOCKS> Blocks per request (eth_getLogs) [default: 1]
--js-tracer <tracer> Event signature for log decoding
Optional Subcommands:
cryo help display help message
cryo help syntax display block + tx specification syntax
cryo help datasets display list of all datasets
cryo help <DATASET(S)> display info about a dataset
(output of cryo help syntax)
Block specification syntax
- can use numbers --blocks 5000 6000 7000
- can use ranges --blocks 12M:13M 15M:16M
- can use a parquet file --blocks ./path/to/file.parquet[:COLUMN_NAME]
- can use multiple parquet files --blocks ./path/to/files/*.parquet[:COLUMN_NAME]
- numbers can contain { _ . K M B } 5_000 5K 15M 15.5M
- omitting range end means latest 15.5M: == 15.5M:latest
- omitting range start means 0 :700 == 0:700
- minus on start means minus end -1000:7000 == 6001:7001
- plus sign on end means plus start 15M:+1000 == 15M:15.001M
- can use every nth value 2000:5000:1000 == 2000 3000 4000
- can use n values total 100:200/5 == 100 124 149 174 199
Timestamp specification syntax
- can use numbers --timestamp 5000 6000 7000
- can use ranges --timestamp 12M:13M 15M:16M
- can use a parquet file --timestamp ./path/to/file.parquet[:COLUMN_NAME]
- can use multiple parquet files --timestamp ./path/to/files/*.parquet[:COLUMN_NAME]
- can contain { _ . m h d w M y } 31_536_000 525600m 8760h 365d 52.143w 12.17M 1y
- omitting range end means latest 15.5M: == 15.5M:latest
- omitting range start means 0 :700 == 0:700
- minus on start means minus end -1000:7000 == 6001:7001
- plus sign on end means plus start 15M:+1000 == 15M:15.001M
- can use n values total 100:200/5 == 100 124 149 174 199
Transaction specification syntax
- can use transaction hashes --txs TX_HASH1 TX_HASH2 TX_HASH3
- can use a parquet file --txs ./path/to/file.parquet[:COLUMN_NAME]
(default column name is transaction_hash)
- can use multiple parquet files --txs ./path/to/ethereum__logs*.parquet
(output of cryo help datasets)
cryo datasets
─────────────
- address_appearances
- balance_diffs
- balance_reads
- balances
- blocks
- code_diffs
- code_reads
- codes
- contracts
- erc20_balances
- erc20_metadata
- erc20_supplies
- erc20_transfers
- erc20_approvals
- erc721_metadata
- erc721_transfers
- eth_calls
- four_byte_counts (alias = 4byte_counts)
- geth_calls
- geth_code_diffs
- geth_balance_diffs
- geth_storage_diffs
- geth_nonce_diffs
- geth_opcodes
- javascript_traces (alias = js_traces)
- logs (alias = events)
- native_transfers
- nonce_diffs
- nonce_reads
- nonces
- slots (alias = storages)
- storage_diffs (alias = slot_diffs)
- storage_reads (alias = slot_reads)
- traces
- trace_calls
- transactions (alias = txs)
- vm_traces (alias = opcode_traces)
dataset group names
───────────────────
- blocks_and_transactions: blocks, transactions
- call_trace_derivatives: contracts, native_transfers, traces
- geth_state_diffs: geth_balance_diffs, geth_code_diffs, geth_nonce_diffs, geth_storage_diffs
- state_diffs: balance_diffs, code_diffs, nonce_diffs, storage_diffs
- state_reads: balance_reads, code_reads, nonce_reads, storage_reads
use cryo help <DATASET> to print info about a specific dataset