* Alternative fix for HTML diff entity split This commit both reverts PR #13357 and uses the exiting implementation alredy used for spans to fix the same issue. That PR duplicates most of logic that is already present elsewhere and still was failing for some cases. This should be simpler as it uses the existing logic that already works for <span>s being split apart. Added both test cases as well. * Update gitdiff_test.go * fmt * entity can have uppercase letter, also add detailed comment per @zeripath
Showing the changes to services/gitdiff/gitdiff.go (1 of 2 files). Show all
Unified · Split
| @@ -182,6 +182,7 @@ var ( · 181 unmodified lines | |||
| 182 | codeTagSuffix = []byte(`</span>`) | 182 | codeTagSuffix = []byte(`</span>`) |
| 183 | ) | 183 | ) |
| 184 | var trailingSpanRegex = regexp.MustCompile(`<span\s*[[:alpha:]="]*?[>]?$`) | 184 | var trailingSpanRegex = regexp.MustCompile(`<span\s*[[:alpha:]="]*?[>]?$`) |
| 185 | var entityRegex = regexp.MustCompile(`&[#]*?[0-9[:alpha:]]*$`) | ||
| 185 | 186 | ||
| 186 | // shouldWriteInline represents combinations where we manually write inline changes | 187 | // shouldWriteInline represents combinations where we manually write inline changes |
| 187 | func shouldWriteInline(diff diffmatchpatch.Diff, lineType DiffLineType) bool { | 188 | func shouldWriteInline(diff diffmatchpatch.Diff, lineType DiffLineType) bool { |
| @@ -205,14 +206,40 @@ func diffToHTML(fileName string, diffs []diffmatchpatch.Diff, lineType DiffLineT · 17 unmodified lines | |||
| 205 | match = "" | 206 | match = "" |
| 206 | } | 207 | } |
| 207 | // Chroma HTML syntax highlighting is done before diffing individual lines in order to maintain consistency. | 208 | // Chroma HTML syntax highlighting is done before diffing individual lines in order to maintain consistency. |
| 208 | // Since inline changes might split in the middle of a chroma span tag, make we manually put it back together | 209 | // Since inline changes might split in the middle of a chroma span tag or HTML entity, make we manually put it back together |
| 209 | // before writing so we don't try insert added/removed code spans in the middle of an existing chroma span | 210 | // before writing so we don't try insert added/removed code spans in the middle of one of those |
| 210 | // and create broken HTML. | 211 | // and create broken HTML. This is done by moving incomplete HTML forward until it no longer matches our pattern of |
| 212 | // a line ending with an incomplete HTML entity or partial/opening <span>. | ||
| 213 | |||
| 214 | // EX: | ||
| 215 | // diffs[{Type: dmp.DiffDelete, Text: "language</span><span "}, | ||
| 216 | // {Type: dmp.DiffEqual, Text: "c"}, | ||
| 217 | // {Type: dmp.DiffDelete, Text: "lass="p">}] | ||
| 218 | |||
| 219 | // After first iteration | ||
| 220 | // diffs[{Type: dmp.DiffDelete, Text: "language</span>"}, //write out | ||
| 221 | // {Type: dmp.DiffEqual, Text: "<span c"}, | ||
| 222 | // {Type: dmp.DiffDelete, Text: "lass="p">,</span>}] | ||
| 223 | |||
| 224 | // After second iteration | ||
| 225 | // {Type: dmp.DiffEqual, Text: ""}, // write out | ||
| 226 | // {Type: dmp.DiffDelete, Text: "<span class="p">,</span>}] | ||
| 227 | |||
| 228 | // Final | ||
| 229 | // {Type: dmp.DiffDelete, Text: "<span class="p">,</span>}] | ||
| 230 | // end up writing <span class="removed-code"><span class="p">,</span></span> | ||
| 231 | // Instead of <span class="removed-code">lass="p",</span></span> | ||
| 232 | |||
| 211 | m := trailingSpanRegex.FindStringSubmatchIndex(diff.Text) | 233 | m := trailingSpanRegex.FindStringSubmatchIndex(diff.Text) |
| 212 | if m != nil { | 234 | if m != nil { |
| 213 | match = diff.Text[m[0]:m[1]] | 235 | match = diff.Text[m[0]:m[1]] |
| 214 | diff.Text = strings.TrimSuffix(diff.Text, match) | 236 | diff.Text = strings.TrimSuffix(diff.Text, match) |
| 215 | } | 237 | } |
| 238 | m = entityRegex.FindStringSubmatchIndex(diff.Text) | ||
| 239 | if m != nil { | ||
| 240 | match = diff.Text[m[0]:m[1]] | ||
| 241 | diff.Text = strings.TrimSuffix(diff.Text, match) | ||
| 242 | } | ||
| 216 | // Print an existing closing span first before opening added/remove-code span so it doesn't unintentionally close it | 243 | // Print an existing closing span first before opening added/remove-code span so it doesn't unintentionally close it |
| 217 | if strings.HasPrefix(diff.Text, "</span>") { | 244 | if strings.HasPrefix(diff.Text, "</span>") { |
| 218 | buf.WriteString("</span>") | 245 | buf.WriteString("</span>") |
| @@ -290,9 +317,6 @@ func init() { · 71 unmodified lines | |||
| 290 | diffMatchPatch.DiffEditCost = 100 | 317 | diffMatchPatch.DiffEditCost = 100 |
| 291 | } | 318 | } |
| 292 | 319 | ||
| 293 | var unterminatedEntityRE = regexp.MustCompile(`&[^ ;]*$`) | ||
| 294 | var unstartedEntiyRE = regexp.MustCompile(`^[^ ;]*;`) | ||
| 295 | |||
| 296 | // GetComputedInlineDiffFor computes inline diff for the given line. | 320 | // GetComputedInlineDiffFor computes inline diff for the given line. |
| 297 | func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) template.HTML { | 321 | func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) template.HTML { |
| 298 | if setting.Git.DisableDiffHighlight { | 322 | if setting.Git.DisableDiffHighlight { |
| @@ -333,89 +357,11 @@ func (diffSection *DiffSection) GetComputedInlineDiffFor(diffLine *DiffLine) tem · 34 unmodified lines | |||
| 333 | diffRecord := diffMatchPatch.DiffMain(highlight.Code(diffSection.FileName, diff1[1:]), highlight.Code(diffSection.FileName, diff2[1:]), true) | 357 | diffRecord := diffMatchPatch.DiffMain(highlight.Code(diffSection.FileName, diff1[1:]), highlight.Code(diffSection.FileName, diff2[1:]), true) |
| 334 | diffRecord = diffMatchPatch.DiffCleanupEfficiency(diffRecord) | 358 | diffRecord = diffMatchPatch.DiffCleanupEfficiency(diffRecord) |
| 335 | 359 | ||
| 336 | // Now we need to clean up the split entities | ||
| 337 | diffRecord = unsplitEntities(diffRecord) | ||
| 338 | diffRecord = diffMatchPatch.DiffCleanupEfficiency(diffRecord) | 360 | diffRecord = diffMatchPatch.DiffCleanupEfficiency(diffRecord) |
| 339 | 361 | ||
| 340 | return diffToHTML(diffSection.FileName, diffRecord, diffLine.Type) | 362 | return diffToHTML(diffSection.FileName, diffRecord, diffLine.Type) |
| 341 | } | 363 | } |
| 342 | 364 | ||
| 343 | // unsplitEntities looks for broken up html entities. It relies on records being presimplified and the data being passed in being valid html | ||
| 344 | func unsplitEntities(records []diffmatchpatch.Diff) []diffmatchpatch.Diff { | ||
| 345 | // Unsplitting entities is simple... | ||
| 346 | // | ||
| 347 | // Iterate through all be the last records because if we're the last record then there's nothing we can do | ||
| 348 | for i := 0; i+1 < len(records); i++ { | ||
| 349 | record := &records[i] | ||
| 350 | |||
| 351 | // Look for an unterminated entity at the end of the line | ||
| 352 | unterminated := unterminatedEntityRE.FindString(record.Text) | ||
| 353 | if len(unterminated) == 0 { | ||
| 354 | continue | ||
| 355 | } | ||
| 356 | |||
| 357 | switch record.Type { | ||
| 358 | case diffmatchpatch.DiffEqual: | ||
| 359 | // If we're an diff equal we want to give this unterminated entity to our next delete and insert | ||
| 360 | record.Text = record.Text[0 : len(record.Text)-len(unterminated)] | ||
| 361 | records[i+1].Text = unterminated + records[i+1].Text | ||
| 362 | |||
| 363 | nextType := records[i+1].Type | ||
| 364 | |||
| 365 | if nextType == diffmatchpatch.DiffEqual { | ||
| 366 | continue | ||
| 367 | } | ||
| 368 | |||
| 369 | // if the next in line is a delete then we will want the thing after that to be an insert and so on. | ||
| 370 | oneAfterType := diffmatchpatch.DiffInsert | ||
| 371 | if nextType == diffmatchpatch.DiffInsert { | ||
| 372 | oneAfterType = diffmatchpatch.DiffDelete | ||
| 373 | } | ||
| 374 | |||
| 375 | if i+2 < len(records) && records[i+2].Type == oneAfterType { | ||
| 376 | records[i+2].Text = unterminated + records[i+2].Text | ||
| 377 | } else { | ||
| 378 | records = append(records[:i+2], append([]diffmatchpatch.Diff{ | ||
| 379 | { | ||
| 380 | Type: oneAfterType, | ||
| 381 | Text: unterminated, | ||
| 382 | }}, records[i+2:]...)...) | ||
| 383 | } | ||
| 384 | case diffmatchpatch.DiffDelete: | ||
| 385 | fallthrough | ||
| 386 | case diffmatchpatch.DiffInsert: | ||
| 387 | // if we're an insert or delete we want to claim the terminal bit of the entity from the next equal in line | ||
| 388 | targetType := diffmatchpatch.DiffInsert | ||
| 389 | if record.Type == diffmatchpatch.DiffInsert { | ||
| 390 | targetType = diffmatchpatch.DiffDelete | ||
| 391 | } | ||
| 392 | next := &records[i+1] | ||
| 393 | if next.Type == diffmatchpatch.DiffEqual { | ||
| 394 | // if the next is an equal we need to snaffle the entity end off the start and add an delete/insert | ||
| 395 | if terminal := unstartedEntiyRE.FindString(next.Text); len(terminal) > 0 { | ||
| 396 | record.Text += terminal | ||
| 397 | next.Text = next.Text[len(terminal):] | ||
| 398 | records = append(records[:i+2], append([]diffmatchpatch.Diff{ | ||
| 399 | { | ||
| 400 | Type: targetType, | ||
| 401 | Text: unterminated, | ||
| 402 | }}, records[i+2:]...)...) | ||
| 403 | } | ||
| 404 | } else if next.Type == targetType { | ||
| 405 | // if the next is an insert we need to snaffle the entity end off the one after that and add it to both. | ||
| 406 | if i+2 < len(records) && records[i+2].Type == diffmatchpatch.DiffEqual { | ||
| 407 | if terminal := unstartedEntiyRE.FindString(records[i+2].Text); len(terminal) > 0 { | ||
| 408 | record.Text += terminal | ||
| 409 | next.Text += terminal | ||
| 410 | records[i+2].Text = records[i+2].Text[len(terminal):] | ||
| 411 | } | ||
| 412 | } | ||
| 413 | } | ||
| 414 | } | ||
| 415 | } | ||
| 416 | return records | ||
| 417 | } | ||
| 418 | |||
| 419 | // DiffFile represents a file diff. | 365 | // DiffFile represents a file diff. |
| 420 | type DiffFile struct { | 366 | type DiffFile struct { |
| 421 | Name string | 367 | Name string |