This is the reference codebase for measuring the performance of Universal-ctags parsers.
We assume you may have enough storage space on your PC.
Get the input code for a parser for the language you are interested in with following command line:
$ ./codebase clone <LANGUAGE>
The following command lists available languages:
$ ./codebase list-languages # LANGUAGE CODE C linux C++ qtbase rocksdb Go buildah kubernetes HTML cockpit JavaScript cockpit LdScript linux ObjectiveC gnustep-libs-base
An example preparing C source code:
$ ./codebae clone C ...
Run Universal-ctags for the cloned code with following command line:
$ [CTAGS_EXE=${where your ctags executable is}] ./codebase ctags <LANGUAGE> [<PROFILE>]
codebase refers CTAGS_EXE environment variable to run ctags. --ctags ${where your ctags executable is} is an alternative way to specify the ctags executable you want to test.
You can run ctags with different option combination. We call such option combination PROFILE. The following command is for listing predefined profiles:
$ ./codebase list-profiles PROFILE DESCRIPTION maximum0 Enables all extras, fields, and kinds minimum0 Disables all fields and extras.
Results are displayed to your terminal. Tee'ed output goes to a file under results/ directory.
An example command line for running C parser:
$ cd /home/yamato/hacking/ctags-github; ./autogen.sh; ./configure; make $ cd /home/yamato/hacking/codebase $ CTAGS_EXE=/home/yamato/hacking/ctags-github/ctags ./codebase ctags C version: 2c46f6d4 features: +wildcards +regex +iconv +option-directory +xpath +json +interactive +sandbox +yaml +aspell +packcc log: results/2c46f6d4,C...................,..........,time......,default...,2019-03-18-12:47:20.log tagsoutput: /dev/null cmdline: + /home/yamato/var/ctags-github/ctags --quiet --options=NONE --sort=no --options=profile.d/maps --totals=yes --languages=C -o - -R code/linux code/php-src code/ruby 27886 files, 19784832 lines (555082 kB) scanned in 15.1 seconds (36686 kB/s) 1158391 tags added to tag file real 0m15.172s user 0m14.735s sys 0m0.399s + set +x
In the above output, "36686 kB/s" represents the speed of parsing of C parser. "1158391 tags" represents the number of tags captured by C parser.
codebase stores the cloned source code to code directory. The directory is repository base. codebase makes another directory, lang, for your convince. As the name shows, lang is language bases. You can run ctags directly on a directory for the language you are interested in:
[codebase]$ cd lang
[codebase/lang]$ ls
Asm C++ GDScript Java Kconfg Lua Markdown Perl R SQL VHDL
Basic CMake Go JavaScript Kotlin Make Meson PHP RSpec SystemTap XSLT
C Fortran HTML Julia LdScript Man OpenAPI PuppetManifest Ruby Vala YACC
[codebase/lang]$ ~/var/ctags-developing/ctags -R JavaScript
ctags: Warning: ignoring null tag in JavaScript/mediawiki/.svgo.config.js(line: 6)
ctags: main/entry.c:1454: registerEntry: Assertion `corkIndex != CORK_NIL' failed.
ctags: main/entry.c:1454: parsing JavaScript/mediawiki/.svgo.config.js:6 as JavaScript
Aborted (core dumped)
You have to write a .lcopy file and put it to lcopy.d directory. See lcopy.d/linux.lcopy as an example:
REPO=https://github.com/torvalds/linux ALIGNMENT=v4.20 LANGUAGES=C,LdScript,Asm,Kconfig,DTS
REPO specifies a git repository. ALIGNMENT is a tag put on the git repository. ALIGNMENT allows users of codebase to get the same source tree. LANGUAGES is a comma separated language list.
You have to write a .ctags file and put it to profile.d directory. A line started from "# @" is used as a description for the profile. You may wan to use --options-maybe to extend profile without modifying existing .ctags files.
Let's optimize our parsers! Masatake YAMATO <[email protected]>