1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
use std::result::Result as StdResult;
use nmt_rs::NamespaceMerkleHasher;
use serde::{Deserialize, Deserializer, Serialize};
use crate::namespaced_data::{NamespacedData, NamespacedDataId};
use crate::nmt::{Namespace, NamespacedSha2Hasher, Nmt, NS_SIZE};
use crate::row::RowId;
use crate::{DataAvailabilityHeader, Error, Result};
/// Represents either column or row of the [`ExtendedDataSquare`].
///
/// [`ExtendedDataSquare`]: crate::rsmt2d::ExtendedDataSquare
#[derive(Copy, Clone, Debug, PartialEq, Eq)]
#[repr(u8)]
pub enum AxisType {
/// A row of the data square.
Row = 0,
/// A column of the data square.
Col,
}
impl TryFrom<u8> for AxisType {
type Error = Error;
fn try_from(value: u8) -> StdResult<Self, Self::Error> {
match value {
0 => Ok(AxisType::Row),
1 => Ok(AxisType::Col),
n => Err(Error::InvalidAxis(n.into())),
}
}
}
/// The data matrix in Celestia blocks extended with parity data.
///
/// It is created by a fixed size chunks of data, called [`Share`]s.
/// Each share is a cell of the [`ExtendedDataSquare`].
///
/// # Structure
///
/// The [`ExtendedDataSquare`] consists of four quadrants. The first
/// quadrant (upper-left) is the original data submitted to the network,
/// referred to as `OriginalDataSquare`. The other three quadrants are
/// the parity data encoded row-wise or column-wise using Reed-Solomon
/// `codec` specified in `EDS`.
///
/// The below diagram shows how the `EDS` is constructed. First, the 2nd
/// and 3rd quadrants are created by computing Reed-Solomon parity data
/// of the original data square, row-wise for 2nd and column-wise for
/// 3rd quadrant. Then, the 4th quadrant is computed either row-wise
/// from 3rd or column-wise from 2nd quadrant.
///
/// ```text
/// ---------------------------
/// | | |
/// | --|-> |
/// | 1 --|-> 2 |
/// | --|-> |
/// | | | | | |
/// -------------+-------------
/// | v v v | |
/// | --|-> |
/// | 3 --|-> 4 |
/// | --|-> |
/// | | |
/// ---------------------------
/// ```
///
/// # Data availability
///
/// The [`DataAvailabilityHeader`] is created by computing [`Nmt`] merkle
/// roots of each row and column of [`ExtendedDataSquare`].
/// By putting those together there are some key
/// properties those have in terms of data availability.
///
/// Thanks to the parity data, to make original data unrecoverable, a malicious
/// actor would need to hide more than a half of the data from each row and column.
/// If we take `k` as the width of the `OriginalDataSquare`, then the attacker
/// needs to hide more than `(k + 1)^2` shares from the [`ExtendedDataSquare`].
/// For the `EDS` with a width of 4, the attacker needs to hide more than 50% of
/// all the shares and that value approaches 25% as the square grows.
///
/// This allows for really efficient data sampling, as the sampling node can reach
/// very high confidence that whole data is available by taking only a few samples.
///
/// # Example
///
/// This example shows rebuilding the merkle trees for each row of the EDS and compares
/// them with the root hashes stored in data availability header.
///
/// ```no_run
/// use celestia_types::nmt::{Namespace, NamespacedSha2Hasher, Nmt};
/// use celestia_types::Share;
/// use nmt_rs::NamespaceMerkleHasher;
/// # use celestia_types::{ExtendedDataSquare, ExtendedHeader};
/// # fn get_header(_: usize) -> ExtendedHeader {
/// # unimplemented!()
/// # }
/// # fn get_eds(_: usize) -> ExtendedDataSquare {
/// # unimplemented!()
/// # }
///
/// let block_height = 15;
/// let header = get_header(block_height);
/// let eds = get_eds(block_height);
/// let width = header.dah.square_len();
///
/// // for each row of the data square, build an nmt
/// for (y, row) in eds.data_square.chunks(width).enumerate() {
/// let mut nmt = Nmt::with_hasher(NamespacedSha2Hasher::with_ignore_max_ns(true));
///
/// for (x, leaf) in row.iter().enumerate() {
/// if x < width / 2 && y < width / 2 {
/// // the `OriginalDataSquare` part of the `EDS`
/// let share = Share::from_raw(leaf).unwrap();
/// let ns = share.namespace();
/// nmt.push_leaf(share.as_ref(), *ns).unwrap();
/// } else {
/// // the parity data computed using `eds.codec`
/// nmt.push_leaf(leaf, *Namespace::PARITY_SHARE).unwrap();
/// }
/// }
///
/// // check if the root corresponds to the one from the dah
/// let root = nmt.root();
/// assert_eq!(root, header.dah.row_root(y).unwrap());
/// }
/// ```
///
/// [`Nmt`]: crate::nmt::Nmt
/// [`Share`]: crate::share::Share
/// [`DataAvailabilityHeader`]: crate::DataAvailabilityHeader
#[derive(Debug, Clone, PartialEq, Eq, Serialize)]
pub struct ExtendedDataSquare {
/// The raw data of the EDS.
#[serde(with = "celestia_tendermint_proto::serializers::bytes::vec_base64string")]
pub data_square: Vec<Vec<u8>>,
/// The codec used to encode parity shares.
pub codec: String,
/// pre-calculated square length
#[serde(skip)]
square_len: usize,
}
impl ExtendedDataSquare {
/// Create a new EDS out of the provided shares. Returns error if number of shares isn't
/// a square number
pub fn new(shares: Vec<Vec<u8>>, codec: String) -> Result<Self> {
let square_len = f64::sqrt(shares.len() as f64) as usize;
if square_len * square_len != shares.len() {
return Err(Error::EdsInvalidDimentions);
}
Ok(Self {
data_square: shares,
codec,
square_len,
})
}
/// Return row with index
pub fn row(&self, index: usize) -> Result<Vec<Vec<u8>>> {
Ok(self
.data_square
.get(index * self.square_len..(index + 1) * self.square_len)
.ok_or(Error::EdsIndexOutOfRange(index))?
.to_vec())
}
/// Return colum with index
pub fn column(&self, mut index: usize) -> Result<Vec<Vec<u8>>> {
let mut r = Vec::with_capacity(self.square_len);
while index < self.data_square.len() {
r.push(
self.data_square
.get(index)
.ok_or(Error::EdsIndexOutOfRange(index))?
.to_vec(),
);
index += self.square_len;
}
Ok(r)
}
/// Return column or row with the provided index
pub fn axis(&self, axis: AxisType, index: usize) -> Result<Vec<Vec<u8>>> {
match axis {
AxisType::Col => self.column(index),
AxisType::Row => self.row(index),
}
}
/// Get EDS square length
pub fn square_len(&self) -> usize {
self.square_len
}
/// Return all the shares that belong to the provided namespace in the EDS.
/// Results are returned as a list of rows of shares with the inclusion proof
pub fn get_namespaced_data(
&self,
namespace: Namespace,
dah: &DataAvailabilityHeader,
height: u64,
) -> Result<Vec<NamespacedData>> {
let mut data = Vec::new();
for i in 0u16..self.square_len as u16 {
let row_root = dah.row_root(i.into()).unwrap();
if !row_root.contains::<NamespacedSha2Hasher>(*namespace) {
continue;
}
let mut shares = Vec::with_capacity(self.square_len);
let mut tree = Nmt::with_hasher(NamespacedSha2Hasher::with_ignore_max_ns(true));
for (col, s) in self.row(i.into())?.iter().enumerate() {
let ns = if col < self.square_len / 2 {
Namespace::from_raw(&s[..NS_SIZE])?
} else {
Namespace::PARITY_SHARE
};
tree.push_leaf(s, *ns).map_err(Error::Nmt)?;
if ns == namespace {
shares.push(s.clone());
}
}
let row = RowId::new(i, height)?;
let proof = tree.get_namespace_proof(*namespace);
let namespaced_data_id = NamespacedDataId { row, namespace };
data.push(NamespacedData {
namespaced_data_id,
proof: proof.into(),
shares,
})
}
Ok(data)
}
}
#[derive(Deserialize)]
struct RawExtendedDataSquare {
#[serde(with = "celestia_tendermint_proto::serializers::bytes::vec_base64string")]
pub data_square: Vec<Vec<u8>>,
pub codec: String,
}
impl<'de> Deserialize<'de> for ExtendedDataSquare {
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: Deserializer<'de>,
{
let eds = RawExtendedDataSquare::deserialize(deserializer)?;
let share_number = eds.data_square.len();
ExtendedDataSquare::new(eds.data_square, eds.codec).map_err(|_| {
<D::Error as serde::de::Error>::invalid_length(
share_number,
&"number of shares must be a perfect square",
)
})
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn axis_type_serialization() {
assert_eq!(AxisType::Row as u8, 0);
assert_eq!(AxisType::Col as u8, 1);
}
#[test]
fn axis_type_deserialization() {
assert_eq!(AxisType::try_from(0).unwrap(), AxisType::Row);
assert_eq!(AxisType::try_from(1).unwrap(), AxisType::Col);
let axis_type_err = AxisType::try_from(2).unwrap_err();
assert!(matches!(axis_type_err, Error::InvalidAxis(2)));
let axis_type_err = AxisType::try_from(99).unwrap_err();
assert!(matches!(axis_type_err, Error::InvalidAxis(99)));
}
#[test]
fn get_namespaced_data() {
let eds_json = include_str!("../test_data/shwap_samples/eds.json");
let eds: ExtendedDataSquare = serde_json::from_str(eds_json).unwrap();
let dah_json = include_str!("../test_data/shwap_samples/dah.json");
let dah: DataAvailabilityHeader = serde_json::from_str(dah_json).unwrap();
let height = 45577;
let rows = eds
.get_namespaced_data(Namespace::new_v0(&[1, 170]).unwrap(), &dah, height)
.unwrap();
assert_eq!(rows.len(), 1);
let row = &rows[0];
row.validate(&dah).unwrap();
assert_eq!(row.shares.len(), 2);
let rows = eds
.get_namespaced_data(Namespace::new_v0(&[1, 187]).unwrap(), &dah, height)
.unwrap();
assert_eq!(rows.len(), 2);
assert_eq!(rows[0].shares.len(), 1);
assert_eq!(rows[1].shares.len(), 4);
for row in rows {
row.validate(&dah).unwrap();
}
}
}