ibnezzoubayr/Atay-Makhzan
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Refactor highlight and diff (#36599)

Browse Source 
1. fix a performance regression when using line-by-line highlighting
* the root cause is that chroma's `lexers.Get` is slow and a lexer cache
is missing during recent changes
2. clarify the chroma lexer detection behavior
* now we fully manage our logic to detect lexer, and handle overriding
problems, everything is fully under control
3. clarify "code analyze" behavior, now only 2 usages:
* only use file name and language to detect lexer (very fast), mainly
for "diff" page which contains a lot of files
* if no lexer is detected by file name and language, use code content to
detect again (slow), mainly for "view file" or "blame" page, which can
get best result
4. fix git diff bug, it caused "broken pipe" error for large diff files
This commit is contained in:
wxiaoguang
2026-02-13 08:15:46 +08:00
committed by GitHub
parent d69b786097
commit 0d8bd7720d
12 changed files with 427 additions and 155 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/highlight/highlight.go
+8 -76
View File
@@ -11,20 +11,16 @@ import (
gohtml "html"
"html/template"
"io"
"path"
"strings"
"sync"
"code.gitea.io/gitea/modules/analyze"
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/setting"
"code.gitea.io/gitea/modules/util"
"github.com/alecthomas/chroma/v2"
"github.com/alecthomas/chroma/v2/formatters/html"
"github.com/alecthomas/chroma/v2/lexers"
"github.com/alecthomas/chroma/v2/styles"
"github.com/go-enry/go-enry/v2"
)
// don't index files larger than this many bytes for performance purposes
@@ -84,85 +80,21 @@ func UnsafeSplitHighlightedLines(code template.HTML) (ret [][]byte) {
}
}
func getChromaLexerByLanguage(fileName, lang string) chroma.Lexer {
lang, _, _ = strings.Cut(lang, "?") // maybe, the value from gitattributes might contain `?` parameters?
ext := path.Ext(fileName)
// the "lang" might come from enry, it has different naming for some languages
switch lang {
case "F#":
lang = "FSharp"
case "Pascal":
lang = "ObjectPascal"
case "C":
if ext == ".C" || ext == ".H" {
lang = "C++"
}
}
if lang == "" && util.AsciiEqualFold(ext, ".sql") {
// there is a bug when using MySQL lexer: "--\nSELECT", the second line will be rendered as comment incorrectly
lang = "SQL"
}
// lexers.Get is slow if the language name can't be matched directly: it does extra "Match" call to iterate all lexers
return lexers.Get(lang)
}
// GetChromaLexerWithFallback returns a chroma lexer by given file name, language and code content. All parameters can be optional.
// When code content is provided, it will be slow if no lexer is found by file name or language.
// If no lexer is found, it will return the fallback lexer.
func GetChromaLexerWithFallback(fileName, lang string, code []byte) (lexer chroma.Lexer) {
if lang != "" {
lexer = getChromaLexerByLanguage(fileName, lang)
}
if lexer == nil {
fileExt := path.Ext(fileName)
if val, ok := globalVars().highlightMapping[fileExt]; ok {
lexer = getChromaLexerByLanguage(fileName, val) // use mapped value to find lexer
}
}
if lexer == nil {
// when using "code" to detect, analyze.GetCodeLanguage is slower, it iterates many rules to detect language from content
// this is the old logic: use enry to detect language, and use chroma to render, but their naming is different for some languages
enryLanguage := analyze.GetCodeLanguage(fileName, code)
lexer = getChromaLexerByLanguage(fileName, enryLanguage)
if lexer == nil {
if enryLanguage != enry.OtherLanguage {
log.Warn("No chroma lexer found for enry detected language: %s (file: %s), need to fix the language mapping between enry and chroma.", enryLanguage, fileName)
}
lexer = lexers.Match(fileName) // lexers.Match will search by its basename and extname
}
}
return util.IfZero(lexer, lexers.Fallback)
}
func renderCode(fileName, language, code string, slowGuess bool) (output template.HTML, lexerName string) {
// RenderCodeSlowGuess tries to get a lexer by file name and language first,
// if not found, it will try to guess the lexer by code content, which is slow (more than several hundreds of milliseconds).
func RenderCodeSlowGuess(fileName, language, code string) (output template.HTML, lexer chroma.Lexer, lexerDisplayName string) {
// diff view newline will be passed as empty, change to literal '\n' so it can be copied
// preserve literal newline in blame view
if code == "" || code == "\n" {
return "\n", ""
return "\n", nil, ""
}
if len(code) > sizeLimit {
return template.HTML(template.HTMLEscapeString(code)), ""
return template.HTML(template.HTMLEscapeString(code)), nil, ""
}
var codeForGuessLexer []byte
if slowGuess {
// it is slower to guess lexer by code content, so only do it when necessary
codeForGuessLexer = util.UnsafeStringToBytes(code)
}
lexer := GetChromaLexerWithFallback(fileName, language, codeForGuessLexer)
return RenderCodeByLexer(lexer, code), formatLexerName(lexer.Config().Name)
}
func RenderCodeFast(fileName, language, code string) (output template.HTML, lexerName string) {
return renderCode(fileName, language, code, false)
}
func RenderCodeSlowGuess(fileName, language, code string) (output template.HTML, lexerName string) {
return renderCode(fileName, language, code, true)
lexer = detectChromaLexerWithAnalyze(fileName, language, util.UnsafeStringToBytes(code)) // it is also slow
return RenderCodeByLexer(lexer, code), lexer, formatLexerName(lexer.Config().Name)
}
// RenderCodeByLexer returns a HTML version of code string with chroma syntax highlighting classes
@@ -204,7 +136,7 @@ func RenderFullFile(fileName, language string, code []byte) ([]template.HTML, st
html.PreventSurroundingPre(true),
)
lexer := GetChromaLexerWithFallback(fileName, language, code)
lexer := detectChromaLexerWithAnalyze(fileName, language, code)
lexerName := formatLexerName(lexer.Config().Name)
iterator, err := lexer.Tokenise(nil, string(code))