| 💾 Saved. d722711 k33g 4h ago | 1 | package fileedit |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "strings" |
| 6 | ) |
| 7 | |
| 8 | // op is what happened to a line. The values are the characters of the unified |
| 9 | // format, which saves a conversion table. |
| 10 | type op byte |
| 11 | |
| 12 | const ( |
| 13 | keep op = ' ' |
| 14 | del op = '-' |
| 15 | add op = '+' |
| 16 | ) |
| 17 | |
| 18 | // line is one line of the diff. oldNo and newNo are 1-based, 0 when the line |
| 19 | // does not exist on that side. |
| 20 | type line struct { |
| 21 | op op |
| 22 | text string |
| 23 | oldNo int |
| 24 | newNo int |
| 25 | } |
| 26 | |
| 27 | // comparison is a diff ready to render: the lines, and what splitting into |
| 28 | // lines cannot say — whether each side ended with a newline. Two files that |
| 29 | // differ only there are two different files, and `git apply` refuses a patch |
| 30 | // that ignores it. |
| 31 | type comparison struct { |
| 32 | lines []line |
| 33 | oldNL, newNL bool |
| 34 | } |
| 35 | |
| 36 | // compare splits, compares, and returns what is needed to display and to patch. |
| 37 | func compare(old, new string) comparison { |
| 38 | oldLines, oldNL := splitLines(old) |
| 39 | newLines, newNL := splitLines(new) |
| 40 | r := comparison{lines: diffLines(oldLines, newLines), oldNL: oldNL, newNL: newNL} |
| 41 | |
| 42 | // Edge case: identical line by line, only the final newline moved. Nothing |
| 43 | // is marked changed and the patch would be empty while the file changes. |
| 44 | // The last line is replayed — deleted then added — which is exactly what |
| 45 | // `git diff` prints. |
| 46 | if oldNL != newNL && !changed(r.lines) && len(r.lines) > 0 { |
| 47 | last := r.lines[len(r.lines)-1] |
| 48 | r.lines = append(r.lines[:len(r.lines)-1], |
| 49 | line{del, last.text, last.oldNo, 0}, |
| 50 | line{add, last.text, 0, last.newNo}, |
| 51 | ) |
| 52 | } |
| 53 | return r |
| 54 | } |
| 55 | |
| 56 | func (r comparison) changed() bool { return changed(r.lines) } |
| 57 | |
| 58 | // firstChanged is the number, in the NEW version, of the first touched line — |
| 59 | // 0 when nothing changed. Enough to open the file at the right place, or to |
| 60 | // say "it starts at line 42" rather than "done". |
| 61 | func (r comparison) firstChanged() int { |
| 62 | for _, l := range r.lines { |
| 63 | switch l.op { |
| 64 | case add: |
| 65 | return l.newNo |
| 66 | case del: |
| 67 | // A deletion has no line in the new version: point at where it |
| 68 | // happened. |
| 69 | return max(1, newNoBefore(r.lines, l)) |
| 70 | } |
| 71 | } |
| 72 | return 0 |
| 73 | } |
| 74 | |
| 75 | func (r comparison) stat() (added, deleted int) { |
| 76 | for _, l := range r.lines { |
| 77 | switch l.op { |
| 78 | case add: |
| 79 | added++ |
| 80 | case del: |
| 81 | deleted++ |
| 82 | } |
| 83 | } |
| 84 | return added, deleted |
| 85 | } |
| 86 | |
| 87 | // splitLines splits a body into lines and says whether it ended with a |
| 88 | // newline. Both matter: a file without a final newline is a different file, |
| 89 | // and the unified patch has to say so. |
| 90 | func splitLines(body string) (lines []string, finalNewline bool) { |
| 91 | if body == "" { |
| 92 | return nil, true |
| 93 | } |
| 94 | finalNewline = strings.HasSuffix(body, "\n") |
| 95 | if finalNewline { |
| 96 | body = body[:len(body)-1] |
| 97 | } |
| 98 | return strings.Split(body, "\n"), finalNewline |
| 99 | } |
| 100 | |
| 101 | // maxCells caps the LCS table. Past it, a line-by-line comparison would cost |
| 102 | // hundreds of megabytes for a diff nobody will read: the middle block is then |
| 103 | // rendered as a block (all deleted, all added), still exact — only coarser. |
| 104 | const maxCells = 4 << 20 |
| 105 | |
| 106 | // diffLines compares two sequences of lines. |
| 107 | // |
| 108 | // Common prefix and suffix are removed BEFORE the LCS: a targeted edit touches |
| 109 | // a few lines in the middle of a file, and without that cut the table would be |
| 110 | // the size of the file squared. |
| 111 | func diffLines(old, new []string) []line { |
| 112 | var out []line |
| 113 | |
| 114 | head := 0 |
| 115 | for head < len(old) && head < len(new) && old[head] == new[head] { |
| 116 | out = append(out, line{keep, old[head], head + 1, head + 1}) |
| 117 | head++ |
| 118 | } |
| 119 | tail := 0 |
| 120 | for tail < len(old)-head && tail < len(new)-head && |
| 121 | old[len(old)-1-tail] == new[len(new)-1-tail] { |
| 122 | tail++ |
| 123 | } |
| 124 | |
| 125 | midOld, midNew := old[head:len(old)-tail], new[head:len(new)-tail] |
| 126 | out = append(out, middle(midOld, midNew, head)...) |
| 127 | |
| 128 | for i := len(old) - tail; i < len(old); i++ { |
| 129 | out = append(out, line{keep, old[i], i + 1, i + 1 - len(old) + len(new)}) |
| 130 | } |
| 131 | return out |
| 132 | } |
| 133 | |
| 134 | // middle compares what remains after trimming the edges. `off` is the number |
| 135 | // of lines already consumed, to number correctly. |
| 136 | func middle(old, new []string, off int) []line { |
| 137 | if len(old) == 0 && len(new) == 0 { |
| 138 | return nil |
| 139 | } |
| 140 | if len(old)*len(new) > maxCells { |
| 141 | out := make([]line, 0, len(old)+len(new)) |
| 142 | for i, l := range old { |
| 143 | out = append(out, line{del, l, off + i + 1, 0}) |
| 144 | } |
| 145 | for j, l := range new { |
| 146 | out = append(out, line{add, l, 0, off + j + 1}) |
| 147 | } |
| 148 | return out |
| 149 | } |
| 150 | |
| 151 | // Classic LCS: length table, then walk. Files edited by an agent fit in it |
| 152 | // comfortably. |
| 153 | lcs := make([][]int, len(old)+1) |
| 154 | for i := range lcs { |
| 155 | lcs[i] = make([]int, len(new)+1) |
| 156 | } |
| 157 | for i := len(old) - 1; i >= 0; i-- { |
| 158 | for j := len(new) - 1; j >= 0; j-- { |
| 159 | if old[i] == new[j] { |
| 160 | lcs[i][j] = lcs[i+1][j+1] + 1 |
| 161 | } else { |
| 162 | lcs[i][j] = max(lcs[i+1][j], lcs[i][j+1]) |
| 163 | } |
| 164 | } |
| 165 | } |
| 166 | |
| 167 | var out []line |
| 168 | i, j := 0, 0 |
| 169 | for i < len(old) && j < len(new) { |
| 170 | switch { |
| 171 | case old[i] == new[j]: |
| 172 | out = append(out, line{keep, old[i], off + i + 1, off + j + 1}) |
| 173 | i, j = i+1, j+1 |
| 174 | case lcs[i+1][j] >= lcs[i][j+1]: |
| 175 | // Deletion first on a tie: a replacement then reads "- old" then |
| 176 | // "+ new", the order one expects. |
| 177 | out = append(out, line{del, old[i], off + i + 1, 0}) |
| 178 | i++ |
| 179 | default: |
| 180 | out = append(out, line{add, new[j], 0, off + j + 1}) |
| 181 | j++ |
| 182 | } |
| 183 | } |
| 184 | for ; i < len(old); i++ { |
| 185 | out = append(out, line{del, old[i], off + i + 1, 0}) |
| 186 | } |
| 187 | for ; j < len(new); j++ { |
| 188 | out = append(out, line{add, new[j], 0, off + j + 1}) |
| 189 | } |
| 190 | return out |
| 191 | } |
| 192 | |
| 193 | // newNoBefore finds the number of the kept line that precedes `target`. |
| 194 | func newNoBefore(lines []line, target line) int { |
| 195 | last := 0 |
| 196 | for _, l := range lines { |
| 197 | if l == target { |
| 198 | break |
| 199 | } |
| 200 | if l.newNo > 0 { |
| 201 | last = l.newNo |
| 202 | } |
| 203 | } |
| 204 | return last + 1 |
| 205 | } |
| 206 | |
| 207 | func changed(lines []line) bool { |
| 208 | for _, l := range lines { |
| 209 | if l.op != keep { |
| 210 | return true |
| 211 | } |
| 212 | } |
| 213 | return false |
| 214 | } |
| 215 | |
| 216 | // noNewline is the line the unified format requires when a file does not end |
| 217 | // with a newline. `git apply` reads it; without it, it adds a newline nobody |
| 218 | // asked for. |
| 219 | const noNewline = `\ No newline at end of file` |
| 220 | |
| 221 | // contextLines is how many unchanged lines are kept around each change. Three |
| 222 | // is the value of `diff -u` and of git. |
| 223 | const contextLines = 3 |
| 224 | |
| 225 | // unified renders a unified patch, applicable as-is by `patch -p1` or |
| 226 | // `git apply`. Paths are prefixed a/ and b/ like git does. |
| 227 | func (r comparison) unified(path string) string { |
| 228 | if !changed(r.lines) { |
| 229 | return "" |
| 230 | } |
| 231 | var b strings.Builder |
| 232 | fmt.Fprintf(&b, "--- a/%s\n+++ b/%s\n", path, path) |
| 233 | |
| 234 | lastOld, lastNew := 0, 0 |
| 235 | for _, l := range r.lines { |
| 236 | lastOld, lastNew = max(lastOld, l.oldNo), max(lastNew, l.newNo) |
| 237 | } |
| 238 | for _, h := range hunks(r.lines) { |
| 239 | oldStart, oldCount, newStart, newCount := h.ranges() |
| 240 | fmt.Fprintf(&b, "@@ -%d,%d +%d,%d @@\n", oldStart, oldCount, newStart, newCount) |
| 241 | for _, l := range h { |
| 242 | b.WriteByte(byte(l.op)) |
| 243 | b.WriteString(l.text) |
| 244 | b.WriteByte('\n') |
| 245 | // The marker follows the LAST line of each version, and a kept |
| 246 | // line belongs to both: it may carry it for both at once. |
| 247 | if (!r.oldNL && l.oldNo == lastOld && l.op != add) || |
| 248 | (!r.newNL && l.newNo == lastNew && l.op != del) { |
| 249 | b.WriteString(noNewline + "\n") |
| 250 | } |
| 251 | } |
| 252 | } |
| 253 | return b.String() |
| 254 | } |
| 255 | |
| 256 | // hunk is a group of contiguous lines shown together. |
| 257 | type hunk []line |
| 258 | |
| 259 | // hunks cuts the diff into pieces: each change plus `contextLines` on both |
| 260 | // sides. Touching pieces are merged — otherwise two neighbouring changes would |
| 261 | // produce overlapping @@ headers, and the patch would be refused. |
| 262 | func hunks(lines []line) []hunk { |
| 263 | var out []hunk |
| 264 | i := 0 |
| 265 | for i < len(lines) { |
| 266 | if lines[i].op == keep { |
| 267 | i++ |
| 268 | continue |
| 269 | } |
| 270 | start := max(0, i-contextLines) |
| 271 | if n := len(out); n > 0 { |
| 272 | if prevEnd := indexAfter(lines, out[n-1]); prevEnd >= start { |
| 273 | out[n-1] = append(out[n-1], lines[prevEnd:hunkEnd(lines, i)]...) |
| 274 | i = hunkEnd(lines, i) |
| 275 | continue |
| 276 | } |
| 277 | } |
| 278 | out = append(out, append(hunk(nil), lines[start:hunkEnd(lines, i)]...)) |
| 279 | i = hunkEnd(lines, i) |
| 280 | } |
| 281 | return out |
| 282 | } |
| 283 | |
| 284 | // hunkEnd finds the end of the piece opened at `i`: advance while a change |
| 285 | // occurs within `contextLines` lines, then keep the trailing context. |
| 286 | func hunkEnd(lines []line, i int) int { |
| 287 | last := i |
| 288 | for j := i; j < len(lines); j++ { |
| 289 | if lines[j].op != keep { |
| 290 | last = j |
| 291 | continue |
| 292 | } |
| 293 | if j-last > contextLines { |
| 294 | break |
| 295 | } |
| 296 | } |
| 297 | return min(len(lines), last+contextLines+1) |
| 298 | } |
| 299 | |
| 300 | // indexAfter returns the position, in `lines`, of the line following the last |
| 301 | // element of `h`. |
| 302 | func indexAfter(lines []line, h hunk) int { |
| 303 | if len(h) == 0 { |
| 304 | return 0 |
| 305 | } |
| 306 | target := h[len(h)-1] |
| 307 | for i, l := range lines { |
| 308 | if l == target { |
| 309 | return i + 1 |
| 310 | } |
| 311 | } |
| 312 | return 0 |
| 313 | } |
| 314 | |
| 315 | // ranges computes the @@ header of the piece. |
| 316 | func (h hunk) ranges() (oldStart, oldCount, newStart, newCount int) { |
| 317 | for _, l := range h { |
| 318 | if l.op != add { |
| 319 | if oldStart == 0 { |
| 320 | oldStart = l.oldNo |
| 321 | } |
| 322 | oldCount++ |
| 323 | } |
| 324 | if l.op != del { |
| 325 | if newStart == 0 { |
| 326 | newStart = l.newNo |
| 327 | } |
| 328 | newCount++ |
| 329 | } |
| 330 | } |
| 331 | if oldCount == 0 { |
| 332 | oldStart = 0 |
| 333 | } |
| 334 | if newCount == 0 { |
| 335 | newStart = 0 |
| 336 | } |
| 337 | return oldStart, oldCount, newStart, newCount |
| 338 | } |
| 339 | |
| 340 | // render formats the diff for the screen and for the model: a line number, a |
| 341 | // sign, the text. No colours here at all — the same text goes back to the |
| 342 | // model, and ANSI codes in a tool result only get in its way; the screen gets |
| 343 | // its grey from the caller's preview mechanism. |
| 344 | // |
| 345 | // The number shown is ALWAYS the new version's: the one to look at after the |
| 346 | // edit. A deleted line no longer exists there, so it gets the number of the |
| 347 | // place where it vanished — a replacement shows "-" and "+" on the SAME number |
| 348 | // and the column never goes backwards. |
| 349 | func (r comparison) render() string { |
| 350 | if !changed(r.lines) { |
| 351 | return "" |
| 352 | } |
| 353 | var b strings.Builder |
| 354 | for _, h := range hunks(r.lines) { |
| 355 | current := 0 |
| 356 | for _, l := range h { |
| 357 | no := l.newNo |
| 358 | if no > 0 { |
| 359 | current = no |
| 360 | } else { |
| 361 | no = current + 1 |
| 362 | } |
| 363 | fmt.Fprintf(&b, "%4d %c %s\n", no, l.op, l.text) |
| 364 | } |
| 365 | } |
| 366 | return b.String() |
| 367 | } |