feat: vendor gitea 1.16.2
This commit is contained in:
@@ -5,221 +5,159 @@
|
||||
package highlight
|
||||
|
||||
import (
|
||||
"bufio"
|
||||
"bytes"
|
||||
"fmt"
|
||||
gohtml "html"
|
||||
"html/template"
|
||||
"io"
|
||||
"path"
|
||||
"path/filepath"
|
||||
"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"
|
||||
chromahtml "github.com/alecthomas/chroma/v2/formatters/html"
|
||||
"github.com/alecthomas/chroma/v2/styles"
|
||||
lru "github.com/hashicorp/golang-lru/v2"
|
||||
)
|
||||
|
||||
// don't index files larger than this many bytes for performance purposes
|
||||
// don't highlight files larger than this many bytes for performance purposes
|
||||
const sizeLimit = 1024 * 1024
|
||||
|
||||
var (
|
||||
// For custom user mapping
|
||||
highlightMapping = map[string]string{}
|
||||
|
||||
once sync.Once
|
||||
|
||||
cache *lru.TwoQueueCache[string, any]
|
||||
|
||||
githubStyles = styles.Get("github")
|
||||
)
|
||||
|
||||
// NewContext loads custom highlight map from local config
|
||||
func NewContext() {
|
||||
once.Do(func() {
|
||||
highlightMapping = setting.GetHighlightMapping()
|
||||
|
||||
// The size 512 is simply a conservative rule of thumb
|
||||
c, err := lru.New2Q[string, any](512)
|
||||
if err != nil {
|
||||
panic(fmt.Sprintf("failed to initialize LRU cache for highlighter: %s", err))
|
||||
}
|
||||
cache = c
|
||||
})
|
||||
type globalVarsType struct {
|
||||
highlightMapping map[string]string
|
||||
githubStyles *chroma.Style
|
||||
}
|
||||
|
||||
// Code returns a HTML version of code string with chroma syntax highlighting classes and the matched lexer name
|
||||
func Code(fileName, language, code string) (output template.HTML, lexerName string) {
|
||||
NewContext()
|
||||
var (
|
||||
globalVarsMu sync.Mutex
|
||||
globalVarsPtr *globalVarsType
|
||||
)
|
||||
|
||||
func globalVars() *globalVarsType {
|
||||
// in the future, the globalVars might need to be re-initialized when settings change, so don't use sync.Once here
|
||||
globalVarsMu.Lock()
|
||||
defer globalVarsMu.Unlock()
|
||||
if globalVarsPtr == nil {
|
||||
globalVarsPtr = &globalVarsType{}
|
||||
globalVarsPtr.githubStyles = styles.Get("github")
|
||||
globalVarsPtr.highlightMapping = setting.GetHighlightMapping()
|
||||
}
|
||||
return globalVarsPtr
|
||||
}
|
||||
|
||||
// UnsafeSplitHighlightedLines splits highlighted code into lines preserving HTML tags
|
||||
// It always includes '\n', '\n' can appear at the end of each line or in the middle of HTML tags
|
||||
// The '\n' is necessary for copying code from web UI to preserve original code lines
|
||||
// ATTENTION: It uses the unsafe conversion between string and []byte for performance reason
|
||||
// DO NOT make any modification to the returned [][]byte slice items
|
||||
func UnsafeSplitHighlightedLines(code template.HTML) (ret [][]byte) {
|
||||
buf := util.UnsafeStringToBytes(string(code))
|
||||
lineCount := bytes.Count(buf, []byte("\n")) + 1
|
||||
ret = make([][]byte, 0, lineCount)
|
||||
nlTagClose := []byte("\n</")
|
||||
for {
|
||||
pos := bytes.IndexByte(buf, '\n')
|
||||
if pos == -1 {
|
||||
if len(buf) > 0 {
|
||||
ret = append(ret, buf)
|
||||
}
|
||||
return ret
|
||||
}
|
||||
// Chroma highlighting output sometimes have "</span>" right after \n, sometimes before.
|
||||
// * "<span>text\n</span>"
|
||||
// * "<span>text</span>\n"
|
||||
if bytes.HasPrefix(buf[pos:], nlTagClose) {
|
||||
pos1 := bytes.IndexByte(buf[pos:], '>')
|
||||
if pos1 != -1 {
|
||||
pos += pos1
|
||||
}
|
||||
}
|
||||
ret = append(ret, buf[:pos+1])
|
||||
buf = buf[pos+1:]
|
||||
}
|
||||
}
|
||||
|
||||
func htmlEscape(code string) template.HTML {
|
||||
return template.HTML(gohtml.EscapeString(code))
|
||||
}
|
||||
|
||||
// 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 htmlEscape(code), nil, ""
|
||||
}
|
||||
|
||||
var lexer chroma.Lexer
|
||||
|
||||
if len(language) > 0 {
|
||||
lexer = lexers.Get(language)
|
||||
|
||||
if lexer == nil {
|
||||
// Attempt stripping off the '?'
|
||||
if idx := strings.IndexByte(language, '?'); idx > 0 {
|
||||
lexer = lexers.Get(language[:idx])
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if lexer == nil {
|
||||
if val, ok := highlightMapping[path.Ext(fileName)]; ok {
|
||||
// use mapped value to find lexer
|
||||
lexer = lexers.Get(val)
|
||||
}
|
||||
}
|
||||
|
||||
if lexer == nil {
|
||||
if l, ok := cache.Get(fileName); ok {
|
||||
lexer = l.(chroma.Lexer)
|
||||
}
|
||||
}
|
||||
|
||||
if lexer == nil {
|
||||
lexer = lexers.Match(fileName)
|
||||
if lexer == nil {
|
||||
lexer = lexers.Fallback
|
||||
}
|
||||
cache.Add(fileName, lexer)
|
||||
}
|
||||
|
||||
return CodeFromLexer(lexer, code), formatLexerName(lexer.Config().Name)
|
||||
lexer = detectChromaLexerWithAnalyze(fileName, language, util.UnsafeStringToBytes(code)) // it is also slow
|
||||
return RenderCodeByLexer(lexer, code), lexer, formatLexerName(lexer.Config().Name)
|
||||
}
|
||||
|
||||
// CodeFromLexer returns a HTML version of code string with chroma syntax highlighting classes
|
||||
func CodeFromLexer(lexer chroma.Lexer, code string) template.HTML {
|
||||
formatter := html.New(html.WithClasses(true),
|
||||
html.WithLineNumbers(false),
|
||||
html.PreventSurroundingPre(true),
|
||||
// RenderCodeByLexer returns a HTML version of code string with chroma syntax highlighting classes
|
||||
func RenderCodeByLexer(lexer chroma.Lexer, code string) template.HTML {
|
||||
formatter := chromahtml.New(chromahtml.WithClasses(true),
|
||||
chromahtml.WithLineNumbers(false),
|
||||
chromahtml.PreventSurroundingPre(true),
|
||||
)
|
||||
|
||||
htmlbuf := bytes.Buffer{}
|
||||
htmlw := bufio.NewWriter(&htmlbuf)
|
||||
|
||||
iterator, err := lexer.Tokenise(nil, code)
|
||||
if err != nil {
|
||||
log.Error("Can't tokenize code: %v", err)
|
||||
return template.HTML(template.HTMLEscapeString(code))
|
||||
return htmlEscape(code)
|
||||
}
|
||||
|
||||
htmlBuf := &bytes.Buffer{}
|
||||
// style not used for live site but need to pass something
|
||||
err = formatter.Format(htmlw, githubStyles, iterator)
|
||||
err = formatter.Format(htmlBuf, globalVars().githubStyles, iterator)
|
||||
if err != nil {
|
||||
log.Error("Can't format code: %v", err)
|
||||
return template.HTML(template.HTMLEscapeString(code))
|
||||
return htmlEscape(code)
|
||||
}
|
||||
|
||||
_ = htmlw.Flush()
|
||||
// Chroma will add newlines for certain lexers in order to highlight them properly
|
||||
// Once highlighted, strip them here, so they don't cause copy/paste trouble in HTML output
|
||||
return template.HTML(strings.TrimSuffix(htmlbuf.String(), "\n"))
|
||||
return template.HTML(util.UnsafeBytesToString(htmlBuf.Bytes()))
|
||||
}
|
||||
|
||||
// File returns a slice of chroma syntax highlighted HTML lines of code and the matched lexer name
|
||||
func File(fileName, language string, code []byte) ([]template.HTML, string, error) {
|
||||
NewContext()
|
||||
|
||||
if len(code) > sizeLimit {
|
||||
return PlainText(code), "", nil
|
||||
// RenderFullFile returns a slice of chroma syntax highlighted HTML lines of code and the matched lexer name
|
||||
func RenderFullFile(fileName, language string, code []byte) ([]template.HTML, string) {
|
||||
if language == LanguagePlaintext || len(code) > sizeLimit {
|
||||
return renderPlainText(code), formatLexerName(LanguagePlaintext)
|
||||
}
|
||||
|
||||
formatter := html.New(html.WithClasses(true),
|
||||
html.WithLineNumbers(false),
|
||||
html.PreventSurroundingPre(true),
|
||||
)
|
||||
|
||||
var lexer chroma.Lexer
|
||||
|
||||
// provided language overrides everything
|
||||
if language != "" {
|
||||
lexer = lexers.Get(language)
|
||||
}
|
||||
|
||||
if lexer == nil {
|
||||
if val, ok := highlightMapping[filepath.Ext(fileName)]; ok {
|
||||
lexer = lexers.Get(val)
|
||||
}
|
||||
}
|
||||
|
||||
if lexer == nil {
|
||||
guessLanguage := analyze.GetCodeLanguage(fileName, code)
|
||||
|
||||
lexer = lexers.Get(guessLanguage)
|
||||
if lexer == nil {
|
||||
lexer = lexers.Match(fileName)
|
||||
if lexer == nil {
|
||||
lexer = lexers.Fallback
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
lexer := detectChromaLexerWithAnalyze(fileName, language, code)
|
||||
lexerName := formatLexerName(lexer.Config().Name)
|
||||
|
||||
iterator, err := lexer.Tokenise(nil, string(code))
|
||||
if err != nil {
|
||||
return nil, "", fmt.Errorf("can't tokenize code: %w", err)
|
||||
rendered := RenderCodeByLexer(lexer, util.UnsafeBytesToString(code))
|
||||
unsafeLines := UnsafeSplitHighlightedLines(rendered)
|
||||
lines := make([]template.HTML, len(unsafeLines))
|
||||
for idx, lineBytes := range unsafeLines {
|
||||
lines[idx] = template.HTML(util.UnsafeBytesToString(lineBytes))
|
||||
}
|
||||
|
||||
tokensLines := chroma.SplitTokensIntoLines(iterator.Tokens())
|
||||
htmlBuf := &bytes.Buffer{}
|
||||
|
||||
lines := make([]template.HTML, 0, len(tokensLines))
|
||||
for _, tokens := range tokensLines {
|
||||
iterator = chroma.Literator(tokens...)
|
||||
err = formatter.Format(htmlBuf, githubStyles, iterator)
|
||||
if err != nil {
|
||||
return nil, "", fmt.Errorf("can't format code: %w", err)
|
||||
}
|
||||
lines = append(lines, template.HTML(htmlBuf.String()))
|
||||
htmlBuf.Reset()
|
||||
}
|
||||
|
||||
return lines, lexerName, nil
|
||||
return lines, lexerName
|
||||
}
|
||||
|
||||
// PlainText returns non-highlighted HTML for code
|
||||
func PlainText(code []byte) []template.HTML {
|
||||
r := bufio.NewReader(bytes.NewReader(code))
|
||||
m := make([]template.HTML, 0, bytes.Count(code, []byte{'\n'})+1)
|
||||
for {
|
||||
content, err := r.ReadString('\n')
|
||||
if err != nil && err != io.EOF {
|
||||
log.Error("failed to read string from buffer: %v", err)
|
||||
break
|
||||
// renderPlainText returns non-highlighted HTML for code
|
||||
func renderPlainText(code []byte) []template.HTML {
|
||||
lines := make([]template.HTML, 0, bytes.Count(code, []byte{'\n'})+1)
|
||||
pos := 0
|
||||
for pos < len(code) {
|
||||
var content []byte
|
||||
nextPos := bytes.IndexByte(code[pos:], '\n')
|
||||
if nextPos == -1 {
|
||||
content = code[pos:]
|
||||
pos = len(code)
|
||||
} else {
|
||||
content = code[pos : pos+nextPos+1]
|
||||
pos += nextPos + 1
|
||||
}
|
||||
if content == "" && err == io.EOF {
|
||||
break
|
||||
}
|
||||
s := template.HTML(gohtml.EscapeString(content))
|
||||
m = append(m, s)
|
||||
lines = append(lines, htmlEscape(util.UnsafeBytesToString(content)))
|
||||
}
|
||||
return m
|
||||
return lines
|
||||
}
|
||||
|
||||
func formatLexerName(name string) string {
|
||||
if name == "fallback" {
|
||||
if name == LanguagePlaintext || name == chromaLexerFallback {
|
||||
return "Plaintext"
|
||||
}
|
||||
|
||||
return util.ToTitleCaseNoLower(name)
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user