Merge pull request #681 from luoliwoshang/llcppsymg/refine

llcppsymg:improve parse symbol

chore/_xtool/llcppsymg/llcppsymg.go

@@ -22,7 +22,6 @@ import (
 	"io"
 	"os"
 	"path/filepath"
-	"strconv"
 	"strings"
 	"unsafe"
 
@@ -61,11 +60,11 @@ func main() {
 
 	check(err)
 
-	filepaths := generateHeaderFilePath(conf.CFlags, conf.Include)
+	filepaths := genHeaderFilePath(conf.CFlags, conf.Include)
-	astInfos, err := parse.ParseHeaderFile(filepaths)
+	headerInfos, err := parse.ParseHeaderFile(filepaths, conf.TrimPrefixes)
 	check(err)
 
-	symbolInfo := getCommonSymbols(symbols, astInfos, conf.TrimPrefixes)
+	symbolInfo := getCommonSymbols(symbols, headerInfos, conf.TrimPrefixes)
 
 	err = genSymbolTableFile(symbolInfo)
 	check(err)
@@ -77,8 +76,8 @@ func check(err error) {
 	}
 }
 
-func parseDylibSymbols(lib string) ([]types.CPPSymbol, error) {
+func parseDylibSymbols(lib string) ([]*nm.Symbol, error) {
-	dylibPath, err := generateDylibPath(lib)
+	dylibPath, err := genDylibPath(lib)
 	if err != nil {
 		return nil, errors.New("failed to generate dylib path")
 	}
@@ -88,22 +87,15 @@ func parseDylibSymbols(lib string) ([]types.CPPSymbol, error) {
 		return nil, errors.New("failed to list symbols in dylib")
 	}
 
-	var symbols []types.CPPSymbol
+	var symbols []*nm.Symbol
-
 	for _, file := range files {
-		for _, sym := range file.Symbols {
+		symbols = append(symbols, file.Symbols...)
-			demangleName := decodeSymbolName(sym.Name)
-			symbols = append(symbols, types.CPPSymbol{
-				Symbol:       sym,
-				DemangleName: demangleName,
-			})
-		}
 	}
 
 	return symbols, nil
 }
 
-func generateDylibPath(lib string) (string, error) {
+func genDylibPath(lib string) (string, error) {
 	output := lib
 	libPath := ""
 	libName := ""
@@ -123,31 +115,20 @@ func generateDylibPath(lib string) (string, error) {
 	return dylibPath, nil
 }
 
-func decodeSymbolName(symbolName string) string {
+func decodeSymbol(symbolName string) string {
 	if symbolName == "" {
 		return ""
 	}
-
 	demangled := llvm.ItaniumDemangle(symbolName, true)
 	if demangled == nil {
 		return symbolName
 	}
 	defer c.Free(unsafe.Pointer(demangled))
-
 	demangleName := c.GoString(demangled)
-	if demangleName == "" {
+	return strings.TrimSpace(demangleName)
-		return symbolName
-	}
-
-	decodedName := strings.TrimSpace(demangleName)
-	decodedName = strings.ReplaceAll(decodedName,
-		"std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const",
-		"std::string")
-
-	return decodedName
 }
 
-func generateHeaderFilePath(cflags string, files []string) []string {
+func genHeaderFilePath(cflags string, files []string) []string {
 	prefixPath := cflags
 	prefixPath = strings.TrimPrefix(prefixPath, "-I")
 	var includePaths []string
@@ -157,77 +138,23 @@ func generateHeaderFilePath(cflags string, files []string) []string {
 	return includePaths
 }
 
-func getCommonSymbols(dylibSymbols []types.CPPSymbol, astInfoList []types.ASTInformation, prefix []string) []types.SymbolInfo {
+func getCommonSymbols(dylibSymbols []*nm.Symbol, symbolMap map[string]string, prefix []string) []*types.SymbolInfo {
-	var commonSymbols []types.SymbolInfo
+	var commonSymbols []*types.SymbolInfo
-	functionNameMap := make(map[string]int)
+	for _, dylibSym := range dylibSymbols {
-
+		symName := strings.TrimPrefix(dylibSym.Name, "_")
-	for _, astInfo := range astInfoList {
+		if goName, ok := symbolMap[symName]; ok {
-		for _, dylibSym := range dylibSymbols {
+			symbolInfo := &types.SymbolInfo{
-			if strings.TrimPrefix(dylibSym.Name, "_") == astInfo.Symbol {
+				Mangle: symName,
-				cppName := generateCPPName(astInfo)
+				CPP:    decodeSymbol(dylibSym.Name),
-				functionNameMap[cppName]++
+				Go:     goName,
-				symbolInfo := types.SymbolInfo{
-					Mangle: strings.TrimPrefix(dylibSym.Name, "_"),
-					CPP:    cppName,
-					Go:     generateMangle(astInfo, functionNameMap[cppName], prefix),
-				}
-				commonSymbols = append(commonSymbols, symbolInfo)
-				break
 			}
+			commonSymbols = append(commonSymbols, symbolInfo)
 		}
 	}
-
 	return commonSymbols
 }
 
-func generateCPPName(astInfo types.ASTInformation) string {
+func genSymbolTableFile(symbolInfos []*types.SymbolInfo) error {
-	cppName := astInfo.Name
-	if astInfo.Class != "" {
-		cppName = astInfo.Class + "::" + astInfo.Name
-	}
-	return cppName
-}
-
-func generateMangle(astInfo types.ASTInformation, count int, prefixes []string) string {
-	astInfo.Class = removePrefix(astInfo.Class, prefixes)
-	astInfo.Name = removePrefix(astInfo.Name, prefixes)
-	res := ""
-	if astInfo.Class != "" {
-		if astInfo.Class == astInfo.Name {
-			res = "(*" + astInfo.Class + ")." + "Init"
-			if count > 1 {
-				res += "__" + strconv.Itoa(count-1)
-			}
-		} else if astInfo.Name == "~"+astInfo.Class {
-			res = "(*" + astInfo.Class + ")." + "Dispose"
-			if count > 1 {
-				res += "__" + strconv.Itoa(count-1)
-			}
-		} else {
-			res = "(*" + astInfo.Class + ")." + astInfo.Name
-			if count > 1 {
-				res += "__" + strconv.Itoa(count-1)
-			}
-		}
-	} else {
-		res = astInfo.Name
-		if count > 1 {
-			res += "__" + strconv.Itoa(count-1)
-		}
-	}
-	return res
-}
-
-func removePrefix(str string, prefixes []string) string {
-	for _, prefix := range prefixes {
-		if strings.HasPrefix(str, prefix) {
-			return strings.TrimPrefix(str, prefix)
-		}
-	}
-	return str
-}
-
-func genSymbolTableFile(symbolInfos []types.SymbolInfo) error {
 	// keep open follow code block can run successfully
 	for i := range symbolInfos {
 		println("symbol", symbolInfos[i].Go)
@@ -269,6 +196,7 @@ func genSymbolTableFile(symbolInfos []types.SymbolInfo) error {
 	}
 	return nil
 }
+
 func readExistingSymbolTable(fileName string) (map[string]types.SymbolInfo, error) {
 	existingSymbols := make(map[string]types.SymbolInfo)
 

chore/_xtool/llcppsymg/parse/parse.go

@@ -3,23 +3,27 @@ package parse
 import (
 	"errors"
 	"strconv"
+	"strings"
 	"unsafe"
 
 	"github.com/goplus/llgo/c"
 	"github.com/goplus/llgo/c/clang"
-	"github.com/goplus/llgo/chore/llcppg/types"
 )
 
 type Context struct {
 	namespaceName string
 	className     string
-	astInfo       []types.ASTInformation
+	prefixes      []string
+	symbolMap     map[string]string
 	currentFile   string
+	nameCounts    map[string]int
 }
 
-func newContext() *Context {
+func newContext(prefixes []string) *Context {
 	return &Context{
-		astInfo: make([]types.ASTInformation, 0),
+		prefixes:   prefixes,
+		symbolMap:  make(map[string]string),
+		nameCounts: make(map[string]int),
 	}
 }
 
@@ -35,70 +39,83 @@ func (c *Context) setCurrentFile(filename string) {
 	c.currentFile = filename
 }
 
-var context = newContext()
+func (c *Context) removePrefix(str string) string {
+	for _, prefix := range c.prefixes {
+		if strings.HasPrefix(str, prefix) {
+			return strings.TrimPrefix(str, prefix)
+		}
+	}
+	return str
+}
 
-func collectFuncInfo(cursor clang.Cursor) types.ASTInformation {
+func (c *Context) genGoName(name string) string {
+	class := c.removePrefix(c.className)
+	name = c.removePrefix(name)
 
-	info := types.ASTInformation{
+	var baseName string
-		Namespace: context.namespaceName,
+	if class == "" {
-		Class:     context.className,
+		baseName = name
+	} else {
+		baseName = c.genMethodName(class, name)
 	}
 
+	return c.addSuffix(baseName)
+}
+
+func (c *Context) genMethodName(class, name string) string {
+	prefix := "(*" + class + ")."
+	if class == name {
+		return prefix + "Init"
+	}
+	if name == "~"+class {
+		return prefix + "Dispose"
+	}
+	return prefix + name
+}
+
+func (c *Context) addSuffix(name string) string {
+	c.nameCounts[name]++
+	count := c.nameCounts[name]
+	if count > 1 {
+		return name + "__" + strconv.Itoa(count-1)
+	}
+	return name
+}
+
+var context = newContext([]string{})
+
+func collectFuncInfo(cursor clang.Cursor) {
 	cursorStr := cursor.String()
 	symbol := cursor.Mangling()
 
-	info.Name = c.GoString(cursorStr.CStr())
+	name := c.GoString(cursorStr.CStr())
-
+	symbolName := c.GoString(symbol.CStr())
-	info.Symbol = c.GoString(symbol.CStr())
+	if len(symbolName) >= 1 && symbolName[0] == '_' {
-	if len(info.Symbol) >= 1 {
+		symbolName = symbolName[1:]
-		if info.Symbol[0] == '_' {
-			info.Symbol = info.Symbol[1:]
-		}
 	}
-
 	defer symbol.Dispose()
 	defer cursorStr.Dispose()
 
-	if context.namespaceName != "" {
+	goName := context.genGoName(name)
-		info.Namespace = context.namespaceName
+	context.symbolMap[symbolName] = goName
-	}
-	if context.className != "" {
-		info.Class = context.className
-	}
-
-	typeStr := cursor.ResultType().String()
-	defer typeStr.Dispose()
-	info.ReturnType = c.GoString(typeStr.CStr())
-
-	info.Parameters = make([]types.Parameter, cursor.NumArguments())
-	for i := 0; i < int(cursor.NumArguments()); i++ {
-		argCurSor := cursor.Argument(c.Uint(i))
-		argType := argCurSor.Type().String()
-		argName := argCurSor.String()
-		info.Parameters[i] = types.Parameter{
-			Name: c.GoString(argName.CStr()),
-			Type: c.GoString(argType.CStr()),
-		}
-
-		argType.Dispose()
-		argName.Dispose()
-	}
-
-	return info
 }
 
 func visit(cursor, parent clang.Cursor, clientData c.Pointer) clang.ChildVisitResult {
-	if cursor.Kind == clang.Namespace {
+	if cursor.Kind == clang.CursorNamespace {
 		nameStr := cursor.String()
+		defer nameStr.Dispose()
+
 		context.setNamespaceName(c.GoString(nameStr.CStr()))
 		clang.VisitChildren(cursor, visit, nil)
 		context.setNamespaceName("")
-	} else if cursor.Kind == clang.ClassDecl {
+	} else if cursor.Kind == clang.CursorClassDecl {
 		nameStr := cursor.String()
+		defer nameStr.Dispose()
+
 		context.setClassName(c.GoString(nameStr.CStr()))
 		clang.VisitChildren(cursor, visit, nil)
 		context.setClassName("")
-	} else if cursor.Kind == clang.CXXMethod || cursor.Kind == clang.FunctionDecl || cursor.Kind == clang.Constructor || cursor.Kind == clang.Destructor {
+	} else if cursor.Kind == clang.CursorCXXMethod || cursor.Kind == clang.CursorFunctionDecl || cursor.Kind == clang.CursorConstructor || cursor.Kind == clang.CursorDestructor {
 		loc := cursor.Location()
 		var file clang.File
 		var line, column c.Uint
@@ -107,9 +124,7 @@ func visit(cursor, parent clang.Cursor, clientData c.Pointer) clang.ChildVisitRe
 		filename := file.FileName()
 
 		if c.Strcmp(filename.CStr(), c.AllocaCStr(context.currentFile)) == 0 {
-			info := collectFuncInfo(cursor)
+			collectFuncInfo(cursor)
-			info.Location = c.GoString(filename.CStr()) + ":" + strconv.Itoa(int(line)) + ":" + strconv.Itoa(int(column))
-			context.astInfo = append(context.astInfo, info)
 		}
 
 		defer filename.Dispose()
@@ -118,14 +133,13 @@ func visit(cursor, parent clang.Cursor, clientData c.Pointer) clang.ChildVisitRe
 	return clang.ChildVisit_Continue
 }
 
-func ParseHeaderFile(filepaths []string) ([]types.ASTInformation, error) {
+func ParseHeaderFile(filepaths []string, prefixes []string) (map[string]string, error) {
-
 	index := clang.CreateIndex(0, 0)
 	args := make([]*c.Char, 3)
 	args[0] = c.Str("-x")
 	args[1] = c.Str("c++")
 	args[2] = c.Str("-std=c++11")
-	context = newContext()
+	context = newContext(prefixes)
 
 	for _, filename := range filepaths {
 		unit := index.ParseTranslationUnit(
@@ -149,5 +163,5 @@ func ParseHeaderFile(filepaths []string) ([]types.ASTInformation, error) {
 
 	index.Dispose()
 
-	return context.astInfo, nil
+	return context.symbolMap, nil
 }

chore/llcppg/types/types.go

@@ -16,10 +16,6 @@
 
 package types
 
-import (
-	"github.com/goplus/llgo/xtool/nm"
-)
-
 // Config represents a configuration for the llcppg tool.
 type Config struct {
 	Name         string   `json:"name"`
@@ -29,27 +25,6 @@ type Config struct {
 	TrimPrefixes []string `json:"trimPrefixes"`
 }
 
-type CPPSymbol struct {
-	DemangleName string
-	*nm.Symbol
-}
-
-type ASTInformation struct {
-	Namespace   string      `json:"namespace"`
-	Class       string      `json:"class"`
-	Name        string      `json:"name"`
-	BaseClasses []string    `json:"baseClasses"`
-	ReturnType  string      `json:"returnType"`
-	Location    string      `json:"location"`
-	Parameters  []Parameter `json:"parameters"`
-	Symbol      string      `json:"symbol"`
-}
-
-type Parameter struct {
-	Name string `json:"name"`
-	Type string `json:"type"`
-}
-
 type SymbolInfo struct {
 	Mangle string `json:"mangle"` // C++ Symbol
 	CPP    string `json:"c++"`    // C++ function name