|
| 1 | +## Copy-on-Write Sequence Container |
| 2 | +## |
| 3 | +## Provides a seq-like container with transparent Copy-on-Write semantics. |
| 4 | +## Memory is shared until a mutation occurs, at which point a copy is made. |
| 5 | +## |
| 6 | +## Key features: |
| 7 | +## - Transparent CoW via =copy hook |
| 8 | +## - seq-compatible API |
| 9 | +## - O(1) copy operations |
| 10 | +## - O(n) mutation operations (only when shared) |
| 11 | +## - Value type semantics (can be copied, no GC pressure) |
| 12 | +## |
| 13 | +## Example: |
| 14 | +## ```nim |
| 15 | +## var original = @[1, 2, 3].toCowSeq() |
| 16 | +## var copy = original # O(1) - shares memory |
| 17 | +## copy.add(4) # Copy-on-Write triggered |
| 18 | +## # original: [1, 2, 3] |
| 19 | +## # copy: [1, 2, 3, 4] |
| 20 | +## ``` |
| 21 | + |
| 22 | +import std/[hashes] |
| 23 | + |
| 24 | +type |
| 25 | + CowSeqData*[T] = ref object |
| 26 | + ## Internal data container with reference counting |
| 27 | + data: seq[T] |
| 28 | + refCount: int |
| 29 | + |
| 30 | + CowSeq*[T] = object |
| 31 | + ## Copy-on-Write sequence container |
| 32 | + ## Behaves like seq[T] but shares memory until mutation |
| 33 | + dataRef: CowSeqData[T] |
| 34 | + |
| 35 | +# |
| 36 | +# Constructors |
| 37 | +# |
| 38 | + |
| 39 | +proc newCowSeq*[T](initialData: seq[T] = @[]): CowSeq[T] = |
| 40 | + ## Create a new CowSeq from a regular seq |
| 41 | + ## Time: O(1) - just wraps the seq |
| 42 | + result.dataRef = CowSeqData[T](data: initialData, refCount: 1) |
| 43 | + |
| 44 | +proc newCowSeq*[T](size: int): CowSeq[T] = |
| 45 | + ## Create a new CowSeq with pre-allocated size |
| 46 | + ## Time: O(n) |
| 47 | + result.dataRef = CowSeqData[T](data: newSeq[T](size), refCount: 1) |
| 48 | + |
| 49 | +proc toCowSeq*[T](s: seq[T]): CowSeq[T] {.inline.} = |
| 50 | + ## Convert a regular seq to CowSeq |
| 51 | + ## Time: O(1) |
| 52 | + newCowSeq(s) |
| 53 | + |
| 54 | +# |
| 55 | +# Lifecycle hooks (transparent CoW!) |
| 56 | +# |
| 57 | + |
| 58 | +proc `=copy`*[T](dest: var CowSeq[T], src: CowSeq[T]) = |
| 59 | + ## Copy hook - implements transparent Copy-on-Write |
| 60 | + ## This makes copies O(1) by sharing the reference |
| 61 | + if dest.dataRef == src.dataRef: |
| 62 | + return # Self-assignment |
| 63 | + |
| 64 | + # Release old reference |
| 65 | + if not dest.dataRef.isNil: |
| 66 | + dest.dataRef.refCount.dec |
| 67 | + if dest.dataRef.refCount <= 0: |
| 68 | + dest.dataRef = nil |
| 69 | + |
| 70 | + # Share new reference |
| 71 | + dest.dataRef = src.dataRef |
| 72 | + if not dest.dataRef.isNil: |
| 73 | + dest.dataRef.refCount.inc |
| 74 | + |
| 75 | +proc `=destroy`*[T](x: var CowSeq[T]) = |
| 76 | + ## Destructor - decrements reference count |
| 77 | + if not x.dataRef.isNil: |
| 78 | + x.dataRef.refCount.dec |
| 79 | + if x.dataRef.refCount <= 0: |
| 80 | + # Last reference, clean up |
| 81 | + x.dataRef = nil |
| 82 | + |
| 83 | +proc `=sink`*[T](dest: var CowSeq[T], src: CowSeq[T]) = |
| 84 | + ## Sink hook - transfers ownership without incrementing refcount |
| 85 | + if dest.dataRef == src.dataRef: |
| 86 | + return |
| 87 | + |
| 88 | + if not dest.dataRef.isNil: |
| 89 | + dest.dataRef.refCount.dec |
| 90 | + if dest.dataRef.refCount <= 0: |
| 91 | + dest.dataRef = nil |
| 92 | + |
| 93 | + dest.dataRef = src.dataRef |
| 94 | + |
| 95 | +# |
| 96 | +# Internal: Copy-on-Write trigger |
| 97 | +# |
| 98 | + |
| 99 | +proc ensureUnique[T](self: var CowSeq[T]) = |
| 100 | + ## Ensure this CowSeq has exclusive ownership of its data |
| 101 | + ## Triggers Copy-on-Write if data is shared |
| 102 | + ## Time: O(1) if not shared, O(n) if shared |
| 103 | + if self.dataRef.isNil: |
| 104 | + self.dataRef = CowSeqData[T](data: @[], refCount: 1) |
| 105 | + elif self.dataRef.refCount > 1: |
| 106 | + # Copy-on-Write happens here! |
| 107 | + let newData = self.dataRef.data # Deep copy of seq |
| 108 | + self.dataRef.refCount.dec |
| 109 | + self.dataRef = CowSeqData[T](data: newData, refCount: 1) |
| 110 | + |
| 111 | +# |
| 112 | +# Read-only operations (O(1), no CoW) |
| 113 | +# |
| 114 | + |
| 115 | +proc len*[T](self: CowSeq[T]): int {.inline.} = |
| 116 | + ## Return the number of elements |
| 117 | + ## Time: O(1) |
| 118 | + if self.dataRef.isNil: 0 |
| 119 | + else: self.dataRef.data.len |
| 120 | + |
| 121 | +proc high*[T](self: CowSeq[T]): int {.inline.} = |
| 122 | + ## Return the highest valid index |
| 123 | + ## Time: O(1) |
| 124 | + self.len - 1 |
| 125 | + |
| 126 | +proc low*[T](self: CowSeq[T]): int {.inline.} = |
| 127 | + ## Return the lowest valid index (always 0) |
| 128 | + ## Time: O(1) |
| 129 | + 0 |
| 130 | + |
| 131 | +proc `[]`*[T](self: CowSeq[T], idx: int): lent T {.inline.} = |
| 132 | + ## Access element at index (read-only) |
| 133 | + ## Time: O(1) |
| 134 | + self.dataRef.data[idx] |
| 135 | + |
| 136 | +proc `[]`*[T](self: CowSeq[T], slice: HSlice[int, int]): seq[T] = |
| 137 | + ## Return a slice as a regular seq |
| 138 | + ## Time: O(k) where k is slice size |
| 139 | + if self.dataRef.isNil: |
| 140 | + return @[] |
| 141 | + self.dataRef.data[slice] |
| 142 | + |
| 143 | +# |
| 144 | +# Mutable operations (trigger CoW if shared) |
| 145 | +# |
| 146 | + |
| 147 | +proc `[]=`*[T](self: var CowSeq[T], idx: int, val: T) = |
| 148 | + ## Set element at index |
| 149 | + ## Time: O(1) if not shared, O(n) if shared (CoW) |
| 150 | + self.ensureUnique() |
| 151 | + self.dataRef.data[idx] = val |
| 152 | + |
| 153 | +proc add*[T](self: var CowSeq[T], val: T) = |
| 154 | + ## Add element to end |
| 155 | + ## Time: O(1) amortized if not shared, O(n) if shared (CoW) |
| 156 | + self.ensureUnique() |
| 157 | + self.dataRef.data.add(val) |
| 158 | + |
| 159 | +proc add*[T](self: var CowSeq[T], other: CowSeq[T]) = |
| 160 | + ## Add all elements from another CowSeq |
| 161 | + ## Time: O(k) where k is other.len |
| 162 | + if other.len == 0: |
| 163 | + return |
| 164 | + self.ensureUnique() |
| 165 | + for item in other: |
| 166 | + self.dataRef.data.add(item) |
| 167 | + |
| 168 | +proc add*[T](self: var CowSeq[T], other: seq[T]) = |
| 169 | + ## Add all elements from a regular seq |
| 170 | + ## Time: O(k) where k is other.len |
| 171 | + if other.len == 0: |
| 172 | + return |
| 173 | + self.ensureUnique() |
| 174 | + self.dataRef.data.add(other) |
| 175 | + |
| 176 | +proc delete*[T](self: var CowSeq[T], idx: int) = |
| 177 | + ## Delete element at index |
| 178 | + ## Time: O(n) |
| 179 | + self.ensureUnique() |
| 180 | + self.dataRef.data.delete(idx) |
| 181 | + |
| 182 | +proc delete*[T](self: var CowSeq[T], first: int, last: int) = |
| 183 | + ## Delete range of elements [first..last] |
| 184 | + ## Time: O(n) |
| 185 | + self.ensureUnique() |
| 186 | + # Nim's seq.delete doesn't have range, do it manually |
| 187 | + for i in countdown(last, first): |
| 188 | + self.dataRef.data.delete(i) |
| 189 | + |
| 190 | +proc insert*[T](self: var CowSeq[T], val: T, idx: int = 0) = |
| 191 | + ## Insert element at index |
| 192 | + ## Time: O(n) |
| 193 | + self.ensureUnique() |
| 194 | + self.dataRef.data.insert(val, idx) |
| 195 | + |
| 196 | +proc setLen*[T](self: var CowSeq[T], newLen: int) = |
| 197 | + ## Set length (grows or shrinks) |
| 198 | + ## Time: O(1) if shrinking, O(k) if growing |
| 199 | + self.ensureUnique() |
| 200 | + self.dataRef.data.setLen(newLen) |
| 201 | + |
| 202 | +# |
| 203 | +# Iteration |
| 204 | +# |
| 205 | + |
| 206 | +iterator items*[T](self: CowSeq[T]): lent T = |
| 207 | + ## Iterate over elements (read-only) |
| 208 | + if not self.dataRef.isNil: |
| 209 | + for item in self.dataRef.data: |
| 210 | + yield item |
| 211 | + |
| 212 | +iterator mitems*[T](self: var CowSeq[T]): var T = |
| 213 | + ## Iterate over elements (mutable) |
| 214 | + ## Triggers CoW if shared |
| 215 | + self.ensureUnique() |
| 216 | + for item in self.dataRef.data.mitems: |
| 217 | + yield item |
| 218 | + |
| 219 | +iterator pairs*[T](self: CowSeq[T]): tuple[key: int, val: lent T] = |
| 220 | + ## Iterate over (index, element) pairs |
| 221 | + if not self.dataRef.isNil: |
| 222 | + for i, item in self.dataRef.data.pairs: |
| 223 | + yield (i, item) |
| 224 | + |
| 225 | +# |
| 226 | +# Conversion |
| 227 | +# |
| 228 | + |
| 229 | +proc asSeq*[T](self: CowSeq[T]): seq[T] = |
| 230 | + ## Convert to regular seq (creates a copy) |
| 231 | + ## Named asSeq to avoid collision with sequtils.toSeq |
| 232 | + ## Time: O(1) if just reading, O(n) if copying |
| 233 | + if self.dataRef.isNil: @[] |
| 234 | + else: self.dataRef.data |
| 235 | + |
| 236 | +proc toOpenArray*[T](self: CowSeq[T], first, last: int): seq[T] = |
| 237 | + ## Return a slice as a seq |
| 238 | + ## Time: O(k) where k is slice size |
| 239 | + if self.dataRef.isNil or first > last or first < 0: |
| 240 | + return @[] |
| 241 | + |
| 242 | + let lastIdx = min(last, self.high) |
| 243 | + result = newSeq[T](lastIdx - first + 1) |
| 244 | + for i in first..lastIdx: |
| 245 | + result[i - first] = self.dataRef.data[i] |
| 246 | + |
| 247 | +# |
| 248 | +# Comparison |
| 249 | +# |
| 250 | + |
| 251 | +proc `==`*[T](a, b: CowSeq[T]): bool = |
| 252 | + ## Equality comparison |
| 253 | + ## Time: O(1) if same reference, O(n) otherwise |
| 254 | + if a.dataRef == b.dataRef: |
| 255 | + return true # Same reference, definitely equal |
| 256 | + |
| 257 | + if a.len != b.len: |
| 258 | + return false |
| 259 | + |
| 260 | + for i in 0..<a.len: |
| 261 | + if a[i] != b[i]: |
| 262 | + return false |
| 263 | + |
| 264 | + return true |
| 265 | + |
| 266 | +proc hash*[T](self: CowSeq[T]): Hash = |
| 267 | + ## Hash function for use in tables/sets |
| 268 | + ## Time: O(n) |
| 269 | + var h: Hash = 0 |
| 270 | + for item in self: |
| 271 | + h = h !& hash(item) |
| 272 | + result = !$h |
| 273 | + |
| 274 | +# |
| 275 | +# Utility |
| 276 | +# |
| 277 | + |
| 278 | +proc contains*[T](self: CowSeq[T], val: T): bool = |
| 279 | + ## Check if value exists in sequence |
| 280 | + ## Time: O(n) |
| 281 | + if self.dataRef.isNil: |
| 282 | + return false |
| 283 | + for item in self.dataRef.data: |
| 284 | + if item == val: |
| 285 | + return true |
| 286 | + return false |
| 287 | + |
| 288 | +proc find*[T](self: CowSeq[T], val: T): int = |
| 289 | + ## Find index of value, returns -1 if not found |
| 290 | + ## Time: O(n) |
| 291 | + if self.dataRef.isNil: |
| 292 | + return -1 |
| 293 | + for i, item in self.dataRef.data: |
| 294 | + if item == val: |
| 295 | + return i |
| 296 | + return -1 |
| 297 | + |
| 298 | +proc `$`*[T](self: CowSeq[T]): string = |
| 299 | + ## String representation |
| 300 | + if self.dataRef.isNil: |
| 301 | + return "@[]" |
| 302 | + result = "@[" |
| 303 | + for i, item in self: |
| 304 | + if i > 0: |
| 305 | + result.add(", ") |
| 306 | + result.add($item) |
| 307 | + result.add("]") |
| 308 | + |
| 309 | +# |
| 310 | +# Debug helpers |
| 311 | +# |
| 312 | + |
| 313 | +proc getRefCount*[T](self: CowSeq[T]): int = |
| 314 | + ## Get current reference count (for debugging/testing) |
| 315 | + if self.dataRef.isNil: 0 |
| 316 | + else: self.dataRef.refCount |
| 317 | + |
| 318 | +proc isShared*[T](self: CowSeq[T]): bool = |
| 319 | + ## Check if this CowSeq shares data with others |
| 320 | + if self.dataRef.isNil: false |
| 321 | + else: self.dataRef.refCount > 1 |
| 322 | + |
0 commit comments