oxedyne/fe2o3/fe2o3_o3db_sync/src/file/stored.rs
9.0 KiB, 30 runs
created by r1870400018:801, which is this file's identity for as long as the history lasts, whatever it is later renamed to
download · who wrote it · its history
| 1 | use crate::{ |
| 2 | prelude::*, |
| 3 | data::{ |
| 4 | core::Key, |
| 5 | }, |
| 6 | file::floc::{ |
| 7 | FileLocation, |
| 8 | FileNum, |
| 9 | StoredFileLocation, |
| 10 | }, |
| 11 | }; |
| 12 | |
| 13 | use oxedyne_fe2o3_core::{ |
| 14 | byte::ToBytes, |
| 15 | }; |
| 16 | use oxedyne_fe2o3_jdat::{ |
| 17 | prelude::*, |
| 18 | id::NumIdDat, |
| 19 | }; |
| 20 | use oxedyne_fe2o3_hash::csum::ChecksumScheme; |
| 21 | use oxedyne_fe2o3_iop_db::api::Meta; |
| 22 | use oxedyne_fe2o3_iop_hash::csum::Checksummer; |
| 23 | |
| 24 | use std::{ |
| 25 | io::{ |
| 26 | self, |
| 27 | SeekFrom, |
| 28 | }, |
| 29 | }; |
| 30 | |
| 31 | /// Names one record, by a digest of its key and the stamp it was written with. An offset names a |
| 32 | /// record only within one generation of its file: a collection moves the records it keeps, and |
| 33 | /// with records of one size a new offset can equal an old one still in use elsewhere, so a |
| 34 | /// location matched by offset alone can be taken for another record (2026-09-23). |
| 35 | #[derive(Clone, Copy, Debug, Default, Eq, Ord, PartialEq, PartialOrd)] |
| 36 | pub struct RecordDigest(u64); |
| 37 | |
| 38 | impl RecordDigest { |
| 39 | |
| 40 | pub fn new< |
| 41 | const UIDL: usize, |
| 42 | UID: NumIdDat<UIDL>, |
| 43 | >( |
| 44 | key: &[u8], |
| 45 | meta: &Meta<UIDL, UID>, |
| 46 | ) |
| 47 | -> Outcome<Self> |
| 48 | { |
| 49 | // The key bytes are a daticle, which ends itself, so the stamp cannot run into it. |
| 50 | let mut buf = Vec::with_capacity(key.len() + Meta::<UIDL, UID>::BYTE_LEN); |
| 51 | buf.extend_from_slice(key); |
| 52 | buf = res!(meta.to_bytes(buf)); |
| 53 | Ok(Self(seahash::hash(&buf))) |
| 54 | } |
| 55 | } |
| 56 | |
| 57 | #[derive(Debug)] |
| 58 | pub struct StoredKey< |
| 59 | const UIDL: usize, |
| 60 | UID: NumIdDat<UIDL>, |
| 61 | > { |
| 62 | chash: alias::ChooseHash, // Hash used to determine cache pathway. |
| 63 | key: Key, |
| 64 | meta: Meta<UIDL, UID>, |
| 65 | csum: Vec<u8>, |
| 66 | } |
| 67 | |
| 68 | impl< |
| 69 | const UIDL: usize, |
| 70 | UID: NumIdDat<UIDL>, |
| 71 | > |
| 72 | StoredKey<UIDL, UID> |
| 73 | { |
| 74 | pub fn ref_chash(&self) -> &alias::ChooseHash { &self.chash } |
| 75 | pub fn mut_chash(&mut self) -> &mut alias::ChooseHash { &mut self.chash } |
| 76 | pub fn key(&self) -> &Key { &self.key } |
| 77 | pub fn into_key(self) -> Key { self.key } |
| 78 | pub fn meta(&self) -> &Meta<UIDL, UID> { &self.meta } |
| 79 | |
| 80 | /// This associated function tries to avoid unnecessary copying during the encoding of a |
| 81 | /// `StoredKey` by accepting a mutable reference to a vector of key bytes and appending a |
| 82 | /// timestamp and a checksum. |
| 83 | pub fn build_bytes<C: Checksummer>( |
| 84 | chash: alias::ChooseHash, |
| 85 | mut buf: Vec<u8>, |
| 86 | cind: Option<usize>, |
| 87 | meta: &Meta<UIDL, UID>, |
| 88 | csummer: C, |
| 89 | ) |
| 90 | -> Outcome<Vec<u8>> |
| 91 | { |
| 92 | let mut cbuf = chash.to_vec(); |
| 93 | cbuf.append(&mut buf); |
| 94 | buf = cbuf; |
| 95 | // 0. Append chunk index information. |
| 96 | let cind = match cind { |
| 97 | Some(uint) => Some(try_into!(u64, uint)), |
| 98 | None => None::<u64>, |
| 99 | }; |
| 100 | let dat_cind = dat!(cind); |
| 101 | buf = res!(dat_cind.to_bytes(buf)); |
| 102 | // 1. Append meta to key. |
| 103 | buf = res!(meta.to_bytes(buf)); |
| 104 | // 2. Calculate checksum and append. |
| 105 | (buf, _) = res!(csummer.append(buf)); |
| 106 | Ok(buf) |
| 107 | } |
| 108 | |
| 109 | /// Do these stored key bytes, as they lie in a data file, hold this key with this stamp? |
| 110 | /// Read in place, since this is on every read from a file: the key follows the cache hash, |
| 111 | /// and the stamp comes just before the checksum. Key bytes are a daticle, which ends itself, |
| 112 | /// so equal bytes at the head of the stored key are the whole of its key. |
| 113 | pub fn holds( |
| 114 | stored: &[u8], |
| 115 | key: &[u8], |
| 116 | meta: &Meta<UIDL, UID>, |
| 117 | csum_len: usize, |
| 118 | ) |
| 119 | -> Outcome<bool> |
| 120 | { |
| 121 | let key_end = constant::CACHE_HASH_BYTES + key.len(); |
| 122 | let meta_end = match stored.len().checked_sub(csum_len) { |
| 123 | Some(n) => n, |
| 124 | None => return Ok(false), |
| 125 | }; |
| 126 | let meta_start = match meta_end.checked_sub(Meta::<UIDL, UID>::BYTE_LEN) { |
| 127 | Some(n) => n, |
| 128 | None => return Ok(false), |
| 129 | }; |
| 130 | if key_end > meta_start || &stored[constant::CACHE_HASH_BYTES..key_end] != key { |
| 131 | return Ok(false); |
| 132 | } |
| 133 | let (stamp, _) = res!(Meta::<UIDL, UID>::from_bytes(&stored[meta_start..meta_end])); |
| 134 | Ok(stamp == *meta) |
| 135 | } |
| 136 | |
| 137 | /// Returns the key bytes, the complete `StoredKey` bytes and the length. |
| 138 | pub fn load< |
| 139 | C: Checksummer, |
| 140 | R: io::Read, |
| 141 | >( |
| 142 | mut r: &mut R, |
| 143 | csummer: C, |
| 144 | ) |
| 145 | -> Outcome<Option<(Self, Vec<u8>, usize)>> |
| 146 | { |
| 147 | // Load the chash bytes. |
| 148 | let mut chash = alias::ChooseHash::default(); |
| 149 | match r.read_exact(&mut chash) { |
| 150 | Err(e) => match e.kind() { |
| 151 | std::io::ErrorKind::UnexpectedEof => return Ok(None), |
| 152 | _ => return Err(err!(e, |
| 153 | "While trying to read cache hash from the start of a new key."; |
| 154 | Decode, Bytes)), |
| 155 | } |
| 156 | _ => (), |
| 157 | } |
| 158 | let mut skey = chash.to_vec(); |
| 159 | |
| 160 | // Load key Daticle. |
| 161 | let keybyts = res!(Dat::load_bytes(&mut r), Decode, Bytes); |
| 162 | skey.extend_from_slice(&keybyts[..]); |
| 163 | |
| 164 | // Load chunk index information. |
| 165 | let cind_byts = res!(Dat::load_bytes(&mut r), Decode, Bytes); |
| 166 | let (dat_cind, _) = res!(Dat::from_bytes(&cind_byts)); |
| 167 | let cind = match &dat_cind { |
| 168 | Dat::Opt(boxoptd) => { |
| 169 | match **boxoptd { |
| 170 | Some(Dat::U64(i)) => Some(try_into!(usize, i)), |
| 171 | None => None, |
| 172 | _ => return Err(err!( |
| 173 | "Expected Dat::Opt(Dat::U64), decoded {:?}.", dat_cind; |
| 174 | Invalid, Input)), |
| 175 | } |
| 176 | }, |
| 177 | _ => { |
| 178 | //debug!(sync_log::stream(), "chash={:02x?} keybyts={:02x?} cind_byts={:02x?}",chash,keybyts,cind_byts); |
| 179 | return Err(err!( |
| 180 | "Expected Dat::Opt(Dat::U64), decoded {:?}.", dat_cind; |
| 181 | Invalid, Input)); |
| 182 | }, |
| 183 | }; |
| 184 | skey.extend_from_slice(&cind_byts); |
| 185 | |
| 186 | // Load rest of data in one go, but we're forced to use a vec because of the potentially |
| 187 | // variable checksummer byte length. |
| 188 | let mut buf = vec![0; Meta::<UIDL, UID>::BYTE_LEN + res!(csummer.len())]; |
| 189 | res!(r.read_exact(&mut buf), Decode, Bytes); |
| 190 | |
| 191 | // Fuse with the key. |
| 192 | skey.extend_from_slice(&buf); |
| 193 | let skey_len = skey.len(); |
| 194 | |
| 195 | // Verify checksum. |
| 196 | let csum = res!(csummer.verify(&skey)); |
| 197 | |
| 198 | // Remove the cache hash from the front of the key, now that the overall checksum has been |
| 199 | // verified. |
| 200 | skey.drain(..constant::CACHE_HASH_BYTES); |
| 201 | |
| 202 | // Read Meta data. |
| 203 | let (meta, _) = res!(Meta::from_bytes(&buf)); |
| 204 | |
| 205 | Ok(Some(( |
| 206 | Self { |
| 207 | chash, |
| 208 | key: match cind { |
| 209 | Some(i) => Key::Chunk(keybyts, i), |
| 210 | None => Key::Complete(keybyts), |
| 211 | }, |
| 212 | meta, |
| 213 | csum, |
| 214 | }, |
| 215 | skey, |
| 216 | skey_len, |
| 217 | ))) |
| 218 | } |
| 219 | |
| 220 | } |
| 221 | |
| 222 | pub struct StoredValue {} |
| 223 | |
| 224 | impl StoredValue { |
| 225 | |
| 226 | /// This associated function is intended to avoid unnecessary copying during the encoding of a |
| 227 | /// `StoredValue` by accepting a mutable reference to a vector of value bytes and appending a |
| 228 | /// checksum. |
| 229 | pub fn build_bytes<C: Checksummer>( |
| 230 | mut buf: Vec<u8>, |
| 231 | csummer: C, |
| 232 | ) |
| 233 | -> Outcome<Vec<u8>> |
| 234 | { |
| 235 | (buf, _) = res!(csummer.append(buf)); |
| 236 | Ok(buf) |
| 237 | } |
| 238 | |
| 239 | pub fn count<RS: io::Read + io::Seek>( |
| 240 | mut rs: &mut RS, |
| 241 | csumlen: usize, |
| 242 | ) |
| 243 | -> Outcome<usize> |
| 244 | { |
| 245 | // 1. Count value Daticle bytes. |
| 246 | let n = res!(Dat::count_bytes(&mut rs), Decode, Bytes); |
| 247 | if n == 0 { |
| 248 | return Ok(0); |
| 249 | } |
| 250 | let csumlen_i64 = try_into!(i64, csumlen); |
| 251 | res!(rs.seek(SeekFrom::Current(csumlen_i64))); |
| 252 | Ok(n + csumlen) |
| 253 | } |
| 254 | |
| 255 | } |
| 256 | |
| 257 | #[derive(Debug)] |
| 258 | pub struct StoredIndex { |
| 259 | sfloc: StoredFileLocation, |
| 260 | csum: Vec<u8>, |
| 261 | } |
| 262 | |
| 263 | impl StoredIndex { |
| 264 | |
| 265 | pub fn keyval_len(&self) -> u64 { |
| 266 | self.sfloc.floc.keyval().len |
| 267 | } |
| 268 | |
| 269 | pub fn ref_file_location(&self) -> &FileLocation { &self.sfloc.floc } |
| 270 | pub fn own_file_location(self) -> FileLocation { self.sfloc.floc } |
| 271 | pub fn ref_stored_file_location(&self) -> &StoredFileLocation { &self.sfloc } |
| 272 | pub fn own_stored_file_location(self) -> StoredFileLocation { self.sfloc } |
| 273 | |
| 274 | pub fn read< |
| 275 | C: Checksummer, |
| 276 | R: io::Read, |
| 277 | >( |
| 278 | r: &mut R, |
| 279 | fnum: FileNum, |
| 280 | csummer: C, |
| 281 | ) |
| 282 | -> Outcome<(Option<Self>, usize)> |
| 283 | { |
| 284 | // Read a StoredFileLocation, minus the checksum (but calculating a checksum). |
| 285 | let (sfloc, n, csum1) = res!(StoredFileLocation::read_bytes(r, fnum, csummer.clone())); |
| 286 | let (csum2, n2) = res!(csummer.read_bytes(r)); |
| 287 | res!(ChecksumScheme::compare(&csum1, &csum2)); |
| 288 | Ok(( |
| 289 | Some(Self { |
| 290 | sfloc: sfloc, |
| 291 | csum: csum2, |
| 292 | }), |
| 293 | n + n2, |
| 294 | )) |
| 295 | } |
| 296 | |
| 297 | } |