foreword
Among the many auto-completion plug-ins of the Vim editor, YouCompleteMe (YCM) is definitely one of the best plug-ins to use, but its configuration process is often daunting for beginners. After many years of tossing by the author, it has basically reached the level of daily use.
If readers are not familiar with the YouCompleteMe plugin, it is recommended to refer to the author's blog Vim: Configure Python and C++ common plugins first .
In order to better understand this article, readers are suggested to read my previous blog Vim: Extending YCM's semantic analysis function for third-party libraries of C-family languages . In this blog, we solved the problem of YCM's semantic analysis error, but the improvement of autocompletion function is limited.
When using Vim+YCM, YCM usually cannot automatically complete functions and definitions in third-party software libraries (such as OpenCV in c++ mode). For this reason, this article mainly introduces how to use ctags to generate tags files, and further improves the automatic completion kinetic energy of YCM in third-party software libraries on the basis of the previous blog.
configuration process
This article takes opencv as an example to introduce how to expand the automatic completion function of YCM by generating tags files.
Generation of Tags file
1 install ctags
sudo apt-get install ctags
Note that according to the requirements of YCM, exuberant-ctags needs to be installed [ 1 ] ^{[1]}[ 1 ] . Fortunately, the above command will install exuberant-ctags by default. You can verify the version of ctags with the following command:
ctags --versionMy output here is as follows:
Exuberant Ctags 5.9~svn20110310, Copyright (C) 1996-2009 Darren Hiebert Addresses: <[email protected]>, http://ctags.sourceforge.net Optional compiled features: +wildcards, +regex
2 Generate Tags file
This article takes opencv as an example, first download the opencv source code, and then execute the following command in the moudles path:
# 注意=+l中是小写的L，而不是1。
ctags -R --fields=+l *
Note that the parameters required by '–fields=+l' c++, if you need to use c language, you also need to add
--langmap=c:.c.hparameters
Then a tags file will appear under the current folder. Let's observe some of the contents of the Tags file, as follows:
imread imgcodecs/src/loadsave.cpp /^Mat imread( const String& filename, int flags )$/;" f language:C++ namespace:cv
imread_ imgcodecs/src/loadsave.cpp /^imread_( const String& filename, int flags, Mat& mat )$/;" f language:C++ namespace:cv
imreadmulti imgcodecs/src/loadsave.cpp /^bool imreadmulti(const String& filename, std::vector<Mat>& mats, int flags)$/;" f language:C++ namespace:cv
imreadmulti imgcodecs/src/loadsave.cpp /^bool imreadmulti(const String& filename, std::vector<Mat>& mats, int start, int count, int flags)$/;" f language:C++ namespace:cv
imreadmulti_ imgcodecs/src/loadsave.cpp /^imreadmulti_(const String& filename, int flags, std::vector<Mat>& mats, int start, int count)$/;" f language:C++ namespace:cv
imshow highgui/misc/java/src/java/highgui+HighGui.java /^ public static void imshow(String winname, Mat img) {$/;" m language:Java class:HighGui
imshow highgui/src/window.cpp /^void cv::imshow( const String& winname, InputArray _img )$/;" f language:C++ class:cv
imshow highgui/src/window.cpp /^void cv::imshow(const String& winname, const ogl::Texture2D& _tex)$/;" f language:C++ class:cv
imshow videoio/src/cap_winrt_bridge.cpp /^void VideoioBridge::imshow()$/;" f language:C++ class:VideoioBridge
imwrite imgcodecs/src/loadsave.cpp /^bool imwrite( const String& filename, InputArray _img,$/;" f language:C++ namespace:cv
imwrite_ imgcodecs/src/loadsave.cpp /^static bool imwrite_( const String& filename, const std::vector<Mat>& img_vec,$/;" f language:C++ namespace:cv
imwritemulti imgcodecs/include/opencv2/imgcodecs.hpp /^bool imwritemulti(const String& filename, InputArrayOfArrays img,$/;" f language:C++ namespace:cv
It can be seen that the tags file contains the function name and basic definition, and which file it is in, and YCM can parse them very well.
3. Configure the tags file path
For this step, you only need to:
-
Move the tags file to a custom location
-
point to it in the vimrc file
For example, I changed the name of the tags file I just generated and put it in the following location: ~/.vim/tagfiles/opencv.tags, and then added the following settings in vimrc:
# 以下二选一即可
set tags=opencv.tags #相对位置，需要将tags文件放在当前工程下
set tags+=~/.vim/tagfiles/opencv.tags #绝对位置,tags文件就放在`~/.vim/tagfiles`内
YCM configuration
configuration
Add the following configuration under vimrc:
let g:ycm_collect_identifiers_from_tags_files = 1
In this way, YCM will automatically find the tags file defined in vimrc and parse it. You can run the following command in vim to see which tags file YCM has loaded in the current buffer:
:echo tagfiles()
The results of my computer running are as follows:
['/home/xuyangcao/.vim/tagfiles/opencv.tags']
Precautions
According to the documentation of YCM, it is found that using clangd completer in the compilation process is smarter than libclang completer, and parsing tags files is more comprehensive. The advantages of Clangd completer can be found in reference [3], so I won’t introduce too much here.
Therefore, in the process of compiling YCM, it is recommended to use --clanged-completerparameters.
Summarize
This article takes opencv as an example to introduce how to expand YCM's auto-completion and syntax highlighting functions by generating tags files.
References
1 YCM does not read identifiers from my tags files