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
|
use cid::Cid;
use rust_unixfs::dir::{resolve, LookupError, ResolveError};
use std::convert::TryFrom;
use std::fmt;
use std::io::{Error as IoError, Read};
use std::path::PathBuf;
fn main() {
let path = match std::env::args()
.nth(1)
.map(|s| IpfsPath::try_from(s.as_str()))
{
Some(Ok(path)) => path,
Some(Err(e)) => {
eprintln!("Invalid path given as argument: {e}");
std::process::exit(1);
}
None => {
eprintln!("USAGE: {} IPFSPATH\n", std::env::args().next().unwrap());
eprintln!(
"Will resolve the given IPFSPATH to a CID through any UnixFS \
directories or HAMT shards from default go-ipfs 0.5 \
configuration flatfs blockstore and write the final CID into \
stdout"
);
std::process::exit(0);
}
};
let ipfs_path = match std::env::var("IPFS_PATH") {
Ok(s) => s,
Err(e) => {
eprintln!("IPFS_PATH is not set or could not be read: {e}");
std::process::exit(1);
}
};
let mut blocks = PathBuf::from(ipfs_path);
blocks.push("blocks");
let blockstore = ShardedBlockStore { root: blocks };
match walk(blockstore, path) {
Ok(Some(cid)) => {
println!("{cid}");
}
Ok(None) => {
eprintln!("not found");
}
Err(Error::OpeningFailed(e)) => {
eprintln!("{e}\n");
eprintln!("This is likely caused by either:");
eprintln!(" - ipfs does not have the block");
eprintln!(" - ipfs is configured to use non-flatfs storage");
eprintln!(" - ipfs is configured to use flatfs with different sharding");
std::process::exit(1);
}
Err(e) => {
eprintln!("Failed to walk the merkle tree: {e}");
std::process::exit(1);
}
}
}
#[derive(Debug)]
pub enum PathError {
InvalidCid(cid::Error),
InvalidPath,
}
impl fmt::Display for PathError {
fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
match self {
PathError::InvalidCid(e) => write!(fmt, "{e}"),
PathError::InvalidPath => write!(fmt, "invalid path"),
}
}
}
impl std::error::Error for PathError {}
/// Ipfs path following https://github.com/ipfs/go-path/
#[derive(Debug)]
pub struct IpfsPath {
/// Option to support moving the cid
root: Option<Cid>,
path: std::vec::IntoIter<String>,
}
impl From<Cid> for IpfsPath {
/// Creates a new `IpfsPath` from just the `Cid`, which is the same as parsing from a string
/// representation of a `Cid`, but cannot fail.
fn from(root: Cid) -> IpfsPath {
IpfsPath {
root: Some(root),
path: Vec::new().into_iter(),
}
}
}
impl TryFrom<&str> for IpfsPath {
type Error = PathError;
fn try_from(path: &str) -> Result<Self, Self::Error> {
let mut split = path.splitn(2, "/ipfs/");
let first = split.next();
let (_root, path) = match first {
Some("") => {
/* started with /ipfs/ */
if let Some(x) = split.next() {
// was /ipfs/x
("ipfs", x)
} else {
// just the /ipfs/
return Err(PathError::InvalidPath);
}
}
Some(x) => {
/* maybe didn't start with /ipfs/, need to check second */
if split.next().is_some() {
// x/ipfs/_
return Err(PathError::InvalidPath);
}
("", x)
}
None => return Err(PathError::InvalidPath),
};
let mut split = path.splitn(2, '/');
let root = split
.next()
.expect("first value from splitn(2, _) must exist");
let path = split
.next()
.iter()
.flat_map(|s| s.split('/').filter(|s| !s.is_empty()).map(String::from))
.collect::<Vec<_>>()
.into_iter();
let root = Some(Cid::try_from(root).map_err(PathError::InvalidCid)?);
Ok(IpfsPath { root, path })
}
}
impl IpfsPath {
pub fn take_root(&mut self) -> Option<Cid> {
self.root.take()
}
}
#[allow(clippy::result_large_err)]
fn walk(blocks: ShardedBlockStore, mut path: IpfsPath) -> Result<Option<Cid>, Error> {
use rust_unixfs::dir::MaybeResolved::*;
let mut buf = Vec::new();
let mut root = path.take_root().unwrap();
let mut cache = None;
for segment in path.path {
println!("cache {cache:?}");
buf.clear();
eprintln!("reading {root} to resolve {segment:?}");
blocks.as_file(&root.to_bytes())?.read_to_end(&mut buf)?;
let mut walker = match resolve(&buf, segment.as_str(), &mut cache)? {
Found(cid) => {
// either root was a Directory or we got lucky with a HAMT directory.
// With HAMTDirectories the top level can contain a direct link to the target, but
// it's more likely it will be found under some bucket, which would be the third
// case in this match.
println!("got lucky: found {cid} for {segment:?}");
println!("cache {cache:?}");
root = cid;
continue;
}
NotFound => return Ok(None),
// when we stumble upon a HAMT shard, we'll need to look up other blocks in order to
// find the final link. The current implementation cannot search for the directory by
// hashing the name and looking it up, but the implementation can be changed underneath
// without changes to the API.
//
// HAMTDirecotories or HAMT shards are multi-block directories where the entires are
// bucketed per their hash value.
NeedToLoadMore(walker) => walker,
};
eprintln!("walking {root} on {segment:?}");
let mut other_blocks = 1;
loop {
let (first, _) = walker.pending_links();
buf.clear();
eprintln!(" -> reading {first} while searching for {segment:?}");
blocks.as_file(&first.to_bytes())?.read_to_end(&mut buf)?;
match walker.continue_walk(&buf, &mut cache)? {
NotFound => {
println!("cache {cache:?}");
return Ok(None);
}
Found(cid) => {
eprintln!(
" resolved {segment} from {root} after {other_blocks} blocks to {cid}"
);
root = cid;
break;
}
NeedToLoadMore(next) => walker = next,
}
other_blocks += 1;
}
}
println!("cache {cache:?}");
Ok(Some(root))
}
enum Error {
OpeningFailed(IoError),
Other(IoError),
Traversal(ResolveError),
}
impl From<IoError> for Error {
fn from(e: IoError) -> Error {
Error::Other(e)
}
}
impl From<ResolveError> for Error {
fn from(e: ResolveError) -> Error {
Error::Traversal(e)
}
}
impl From<LookupError> for Error {
fn from(e: LookupError) -> Error {
Error::Traversal(e.into())
}
}
impl fmt::Display for Error {
fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
use Error::*;
match self {
OpeningFailed(e) => write!(fmt, "File opening failed: {e}"),
Other(e) => write!(fmt, "Other file related io error: {e}"),
Traversal(e) => write!(fmt, "Walking failed, please report this as a bug: {e:?}"),
}
}
}
struct ShardedBlockStore {
root: PathBuf,
}
impl ShardedBlockStore {
fn as_path(&self, key: &[u8]) -> PathBuf {
// assume that we have a block store with second-to-last/2 sharding
// files in Base32Upper
let encoded = multibase::Base::Base32Upper.encode(key);
let len = encoded.len();
// this is safe because base32 is ascii
let dir = &encoded[(len - 3)..(len - 1)];
assert_eq!(dir.len(), 2);
let mut path = self.root.clone();
path.push(dir);
path.push(encoded);
path.set_extension("data");
path
}
#[allow(clippy::result_large_err)]
fn as_file(&self, key: &[u8]) -> Result<std::fs::File, Error> {
let path = self.as_path(key);
std::fs::OpenOptions::new()
.read(true)
.open(path)
.map_err(Error::OpeningFailed)
}
}
|