用 libclang 解析；无法解析某些标记（Python in Windows）

Question

我有一些代码（从here and 中获取和改编），它使用libclang解析Python（Widnows）中的C++源文件并获取所有其 声明语句，如下所示：

import clang.cindex

def parse_decl(node):
    reference_node = node.get_definition()
    if node.kind.is_declaration():
        print(node.kind, node.kind.name, 
              node.location.line, ',', node.location.column, 
              reference_node.displayname)

    for ch in node.get_children():
        parse_decl(ch)

# configure path
clang.cindex.Config.set_library_file('C:/Program Files (x86)/LLVM/bin/libclang.dll')

index = clang.cindex.Index.create()
trans_unit = index.parse(r'C:\path\to\sourcefile\test.cpp', args=['-std=c++11'])

parse_decl(trans_unit.cursor)

对于以下 C++ 源文件 (test_ok.cpp):

/* test_ok.cpp
 */

#include <iostream>
#include <fstream>
#include <string>
#include <algorithm>
#include <cmath>
#include <iomanip>
using namespace std;

int main (int argc, char *argv[]) {
  int linecount = 0;
  double array[1000], sum=0, median=0, add=0;
  string filename;
  if (argc <= 1)
      {
          cout << "Error: no filename specified" << endl;
          return 0;
      }
//program checks if a filename is specified
  filename = argv[1];
  ifstream myfile (filename.c_str());
  if (myfile.is_open())
  {
    myfile >> array[linecount];
    while ( myfile.good() )
    {
        linecount++;
        myfile >> array[linecount];
    }
    myfile.close();
  }

parse 方法按应有的方式解析并输出：

CursorKind.USING_DIRECTIVE USING_DIRECTIVE 10 , 17 std
CursorKind.FUNCTION_DECL FUNCTION_DECL 12 , 5 main(int, char **)
CursorKind.PARM_DECL PARM_DECL 12 , 15 argc
CursorKind.PARM_DECL PARM_DECL 12 , 27 argv
CursorKind.VAR_DECL VAR_DECL 13 , 7 linecount
CursorKind.VAR_DECL VAR_DECL 14 , 10 array
CursorKind.VAR_DECL VAR_DECL 14 , 23 sum
CursorKind.VAR_DECL VAR_DECL 14 , 30 median
CursorKind.VAR_DECL VAR_DECL 14 , 40 add
CursorKind.VAR_DECL VAR_DECL 15 , 10 filename
CursorKind.VAR_DECL VAR_DECL 23 , 12 myfile

Process finished with exit code 0

然而,

对于以下 C++ 源文件 (test.cpp):

/* test.cpp
*/
#include <iostream>
#include <vector>
#include <fstream>
#include <cmath>
#include <algorithm>
#include <iomanip>

using namespace std;

void readfunction(vector<double>& numbers, ifstream& myfile)
{

  double number;
  while (myfile >> number) {
  numbers.push_back(number);}

}

double meanfunction(vector<double>& numbers)
{

  double total=0;
  vector<double>::const_iterator i;
  for (i=numbers.begin(); i!=numbers.end(); ++i) {
  total +=*i; }
  return total/numbers.size();

}

解析不完整:

CursorKind.USING_DIRECTIVE USING_DIRECTIVE 8 , 17 std
CursorKind.VAR_DECL VAR_DECL 10 , 6 readfunction

Process finished with exit code 0

解析无法处理 vector<double>& numbers 等行，并停止解析该部分代码。

我认为该问题与该问题的另一个 SO question. I have tried to explicitly use the std=c++11 parse argument with no success. In an answer 中描述的问题相似（即使它没有解决问题）也建议使用 -x c++ 但 我不知道如何在上面的代码中添加它。

任何人都可以指出 libclang 的解决方案来解析像 test.cpp 中那样的 C++ 语句吗？

此外，我能否让它继续解析，即使它遇到无法解析的标记？

Answer 1

默认情况下，libclang 不添加编译系统包含路径。

始终确保您已经检查了诊断信息 - 如编译器错误消息，它们往往会指示如何解决任何问题。在这种情况下，很明显存在包含问题：

<Diagnostic severity 4, location <SourceLocation file 'test.cpp', line 3, column 10>, spelling "'iostream' file not found">

如果您确定 libclang 添加了这些路径，它应该会开始工作。

This question includes an approach to solving this problem. This seems to be a recurring theme on Whosebug, so I wrote ccsyspath 以帮助找到 OSX、Linux 和 Windows 上的那些路径。稍微简化您的代码：

import clang.cindex
clang.cindex.Config.set_library_file('C:/Program Files (x86)/LLVM/bin/libclang.dll')
import ccsyspath

index = clang.cindex.Index.create()

args    = '-x c++ --std=c++11'.split()
syspath = ccsyspath.system_include_paths('clang++')
incargs = [ b'-I' + inc for inc in syspath ]
args    = args + incargs

trans_unit = index.parse('test.cpp', args=args)
for node in trans_unit.cursor.walk_preorder():
    if node.location.file is None:
        continue
    if node.location.file.name != 'test.cpp':
        continue
    if node.kind.is_declaration():
        print(node.kind, node.location)

我的 args 最终成为：

['-x',
 'c++',
 '--std=c++11',
 '-IC:\Program Files (x86)\LLVM\bin\..\lib\clang\3.8.0\include',
 '-IC:\Program Files (x86)\Microsoft Visual Studio 12.0\VC\include',
 '-IC:\Program Files (x86)\Windows Kits\8.1\include\shared',
 '-IC:\Program Files (x86)\Windows Kits\8.1\include\um',
 '-IC:\Program Files (x86)\Windows Kits\8.1\include\winrt']

输出为：

(CursorKind.USING_DIRECTIVE, <SourceLocation file 'test.cpp', line 10, column 17>)
(CursorKind.FUNCTION_DECL, <SourceLocation file 'test.cpp', line 12, column 6>)
(CursorKind.PARM_DECL, <SourceLocation file 'test.cpp', line 12, column 35>)
(CursorKind.PARM_DECL, <SourceLocation file 'test.cpp', line 12, column 54>)
(CursorKind.VAR_DECL, <SourceLocation file 'test.cpp', line 15, column 14>)
(CursorKind.FUNCTION_DECL, <SourceLocation file 'test.cpp', line 21, column 8>)
(CursorKind.PARM_DECL, <SourceLocation file 'test.cpp', line 21, column 37>)
(CursorKind.VAR_DECL, <SourceLocation file 'test.cpp', line 24, column 14>)
(CursorKind.VAR_DECL, <SourceLocation file 'test.cpp', line 25, column 40>)

用 libclang 解析；无法解析某些标记（Python in Windows）

Parsing with libclang; unable to parse certain tokens (Python in Windows)

python

parsing

c++11

clang

libclang