Fine tune diff highlighting

services/gitdiff/gitdiff.go

@@ -16,6 +16,7 @@ import (
 	"path"
 	"sort"
 	"strings"
+	"time"
 
 	"code.gitea.io/gitea/models/db"
 	git_model "code.gitea.io/gitea/models/git"
@@ -123,8 +124,14 @@ type DiffHTMLOperation struct {
 // BlobExcerptChunkSize represent max lines of excerpt
 const BlobExcerptChunkSize = 20
 
-// MaxDiffHighlightEntireFileSize is the maximum file size that will be highlighted with "entire file diff"
-const MaxDiffHighlightEntireFileSize = 1 * 1024 * 1024
+// Chroma seems extremely slow when highlighting large files, it might take dozens or hundreds of milliseconds.
+// When fully highlighting a diff with a lot of large files, it would take many seconds or even dozens of seconds.
+// So, don't highlight the entire file if it's too large, or highlighting takes too long.
+// When there is no full-file highlighting, the legacy "line-by-line" highlighting is still applied as the fallback.
+const (
+	MaxFullFileHighlightSizeLimit = 256 * 1024
+	MaxFullFileHighlightTimeLimit = 2 * time.Second
+)
 
 // GetType returns the type of DiffLine.
 func (d *DiffLine) GetType() int {
@@ -564,7 +571,7 @@ func getCommitFileLineCountAndLimitedContent(commit *git.Commit, filePath string
 	if err != nil {
 		return 0, nil
 	}
-	w := &limitByteWriter{limit: MaxDiffHighlightEntireFileSize + 1}
+	w := &limitByteWriter{limit: MaxFullFileHighlightSizeLimit + 1}
 	lineCount, err = blob.GetBlobLineCount(w)
 	if err != nil {
 		return 0, nil
@@ -1317,6 +1324,8 @@ func GetDiffForRender(ctx context.Context, repoLink string, gitRepo *git.Reposit
 		return nil, err
 	}
 
+	startTime := time.Now()
+
 	checker, err := attribute.NewBatchChecker(gitRepo, opts.AfterCommitID, []string{attribute.LinguistVendored, attribute.LinguistGenerated, attribute.LinguistLanguage, attribute.GitlabLanguage, attribute.Diff})
 	if err != nil {
 		return nil, err
@@ -1356,7 +1365,8 @@ func GetDiffForRender(ctx context.Context, repoLink string, gitRepo *git.Reposit
 			diffFile.Sections = append(diffFile.Sections, tailSection)
 		}
 
-		shouldFullFileHighlight := !setting.Git.DisableDiffHighlight && attrDiff.Value() == ""
+		shouldFullFileHighlight := attrDiff.Value() == "" // only do highlight if no custom diff command
+		shouldFullFileHighlight = shouldFullFileHighlight && time.Since(startTime) < MaxFullFileHighlightTimeLimit
 		if shouldFullFileHighlight {
 			if limitedContent.LeftContent != nil {
 				diffFile.highlightedLeftLines.value = highlightCodeLinesForDiffFile(diffFile, true /* left */, limitedContent.LeftContent.buf.Bytes())
@@ -1380,7 +1390,7 @@ func highlightCodeLinesForDiffFile(diffFile *DiffFile, isLeft bool, rawContent [
 }
 
 func highlightCodeLines(name, lang string, sections []*DiffSection, isLeft bool, rawContent []byte) map[int]template.HTML {
-	if setting.Git.DisableDiffHighlight || len(rawContent) > MaxDiffHighlightEntireFileSize {
+	if setting.Git.DisableDiffHighlight || len(rawContent) > MaxFullFileHighlightSizeLimit {
 		return nil
 	}
 

services/gitdiff/highlightdiff.go

@@ -298,15 +298,21 @@ func (hcd *highlightCodeDiff) convertToPlaceholders(htmlContent template.HTML) s
 	return res.String()
 }
 
-func (hcd *highlightCodeDiff) extractNextPlaceholder(buf []byte, lastIdx int) (idx int, placeholder rune, runeLen int, token string) {
-	for idx = lastIdx; idx < len(buf); {
-		placeholder, runeLen = utf8.DecodeRune(buf[idx:])
-		if token = hcd.placeholderTokenMap[placeholder]; token != "" {
-			break
+// recoverOneRune tries to recover one rune
+// * if the rune is a placeholder, it will be recovered to the corresponding content
+// * otherwise it will be returned as is
+func (hcd *highlightCodeDiff) recoverOneRune(buf []byte) (r rune, runeLen int, isSingleTag bool, recovered string) {
+	r, runeLen = utf8.DecodeRune(buf)
+	token := hcd.placeholderTokenMap[r]
+	if token == "" {
+		return r, runeLen, false, "" // rune itself, not a placeholder
+	} else if token[0] == '<' {
+		if token[1] == '<' {
+			return 0, runeLen, false, token[1 : len(token)-1] // full tag `<<span>content</span>>`, recover to `<span>content</span>`
 		}
-		idx += runeLen
+		return r, runeLen, true, token // single tag
 	}
-	return idx, placeholder, runeLen, token
+	return 0, runeLen, false, token // HTML entity
 }
 
 func (hcd *highlightCodeDiff) recoverOneDiff(str string) template.HTML {
@@ -315,49 +321,65 @@ func (hcd *highlightCodeDiff) recoverOneDiff(str string) template.HTML {
 	var diffCodeOpenTag string
 	diffCodeCloseTag := hcd.placeholderTokenMap[hcd.diffCodeClose]
 	strBytes := util.UnsafeStringToBytes(str)
+
+	// this loop is slightly longer than expected, for performance consideration
 	for idx := 0; idx < len(strBytes); {
-		newIdx, placeholder, lastRuneLen, token := hcd.extractNextPlaceholder(strBytes, idx)
-		if newIdx != idx {
+		// take a look at the next rune
+		r, runeLen, isSingleTag, recovered := hcd.recoverOneRune(strBytes[idx:])
+		idx += runeLen
+
+		// loop section 1: if it isn't a single tag, then try to find the following runes until the next single tag, and recover them together
+		if !isSingleTag {
 			if diffCodeOpenTag != "" {
+				// start the "added/removed diff tag" if the current token is in the diff part
 				sb.WriteString(diffCodeOpenTag)
-				sb.Write(strBytes[idx:newIdx])
-				sb.WriteString(diffCodeCloseTag)
+			}
+			if recovered != "" {
+				sb.WriteString(recovered)
 			} else {
-				sb.Write(strBytes[idx:newIdx])
+				sb.WriteRune(r)
+			}
+			// inner loop to recover following runes until the next single tag
+			for idx < len(strBytes) {
+				r, runeLen, isSingleTag, recovered = hcd.recoverOneRune(strBytes[idx:])
+				idx += runeLen
+				if isSingleTag {
+					break
+				}
+				if recovered != "" {
+					sb.WriteString(recovered)
+				} else {
+					sb.WriteRune(r)
+				}
+			}
+			if diffCodeOpenTag != "" {
+				// end the "added/removed diff tag" if the current token is in the diff part
+				sb.WriteString(diffCodeCloseTag)
 			}
-		} // else: no text content before, the current token is a placeholder
-		if token == "" {
-			break // reaches the string end, no more placeholder
 		}
-		idx = newIdx + lastRuneLen
 
-		// for HTML entity
-		if token[0] == '&' {
-			sb.WriteString(token)
-			continue
+		if !isSingleTag {
+			break // the inner loop has already consumed all remaining runes, no more single tag found
 		}
 
-		// for various HTML tags
-		var recovered string
-		if token[1] == '<' { // full tag `<<span>content</span>>`, recover to `<span>content</span>`
-			recovered = token[1 : len(token)-1]
-			if diffCodeOpenTag != "" {
-				recovered = diffCodeOpenTag + recovered + diffCodeCloseTag
-			} // else: just use the recovered content
-		} else if token[1] != '/' { // opening tag
+		// loop section 2: for opening/closing HTML tags
+		placeholder := r
+		if recovered[1] != '/' { // opening tag
 			if placeholder == hcd.diffCodeAddedOpen || placeholder == hcd.diffCodeRemovedOpen {
-				diffCodeOpenTag = token
+				diffCodeOpenTag = recovered
+				recovered = ""
 			} else {
-				recovered = token
 				tagStack = append(tagStack, recovered)
 			}
 		} else { // closing tag
 			if placeholder == hcd.diffCodeClose {
 				diffCodeOpenTag = "" // the highlighted diff is closed, no more diff
+				recovered = ""
 			} else if len(tagStack) != 0 {
-				recovered = token
 				tagStack = tagStack[:len(tagStack)-1]
-			} // else: if no opening tag in stack yet, skip the closing tag
+			} else {
+				recovered = ""
+			}
 		}
 		sb.WriteString(recovered)
 	}

services/gitdiff/highlightdiff_test.go

@@ -77,15 +77,13 @@ func TestDiffWithHighlight(t *testing.T) {
 
 	t.Run("ComplexDiff1", func(t *testing.T) {
 		oldCode, _ := highlight.RenderCodeFast("a.go", "Go", `xxx || yyy`)
-		newCode, _ := highlight.RenderCodeFast("a.go", "Go", `bot.xxx || bot.yyy`)
+		newCode, _ := highlight.RenderCodeFast("a.go", "Go", `bot&xxx || bot&yyy`)
 		hcd := newHighlightCodeDiff()
 		out := hcd.diffLineWithHighlight(DiffLineAdd, oldCode, newCode)
 		assert.Equal(t, strings.ReplaceAll(`
-<span class="added-code"><span class="nx">bot</span></span>
-<span class="added-code"><span class="p">.</span></span>
+<span class="added-code"><span class="nx">bot</span></span><span class="o"><span class="added-code">&amp;</span></span>
 <span class="nx">xxx</span><span class="w"> </span><span class="o">||</span><span class="w"> </span>
-<span class="added-code"><span class="nx">bot</span></span>
-<span class="added-code"><span class="p">.</span></span>
+<span class="added-code"><span class="nx">bot</span></span><span class="o"><span class="added-code">&amp;</span></span>
 <span class="nx">yyy</span>`, "\n", ""), string(out))
 	})